VARIATION IN MINERAL LEVELS AND IMMUNE RESPONSES RELATIVE TO ENVIRONMENTAL AND INDIVIDUAL CONDITIONS IN ADULT FEMALE MOOSE IN CENTRAL BRITISH COLUMBIA by Carlie Rae O’Brien B.Sc., Trent University, 2021 THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN NATURAL RESOURCES AND ENVIRONMENTAL STUDIES UNIVERSITY OF NORTHERN BRITISH COLUMBIA August 2025 © Carlie O’Brien, 2025 Abstract Environmental change can compromise the health and fitness of individual wildlife, leading to negative consequences for populations. Understanding how environmental change relates to wildlife health and fitness is therefore crucial for informing effective conservation and management strategies. Mineral status and immune function are key components of animal health that are sensitive to changes in habitat, climate, and disturbance regimes, and may therefore serve as useful biomarkers for examining how environmental variation corresponds with health and population resilience in wildlife. Moose (Alces alces) are one species whose health may be affected by environmental change. Over the past two decades, moose populations in central British Columbia (BC) have declined dramatically following a severe mountain pine beetle epidemic and subsequent timber salvage logging, which resulted in a heavily altered landscape. In response to these declines, the Province of BC initiated a long-term research project on adult female moose. This research documented cases of starvation and health-related mortalities, along with suboptimal pregnancy rates, which suggests that bottom-up factors may have contributed to the observed declines. My thesis draws on and supplements information collected as part of the BC Provincial Moose Research Project to investigate associations between bottom-up factors and moose health. Specifically, I examined environmental and individual correlates of essential mineral concentrations and immune responses in female moose to better characterize patterns linking environmental variation and moose health. First, I examined whether mineral concentrations in the hair of adult female moose were associated with environmental factors in their summer–autumn habitat. I used hair samples collected during winter captures to quantify the concentrations of 15 macro and trace minerals. ii Using generalized linear mixed-effects models, I tested whether variation in mineral concentrations may have reflected differences in habitat composition, landscape disturbance, and climatic conditions. I found that precipitation was an important predictor of selenium and zinc concentrations, suggesting that mineral uptake could be influenced by climate-driven effects on vegetation. Moose that spent more time in deciduous forests had greater concentrations of potassium and magnesium, possibly reflecting the nutritional value of these forest stands. Furthermore, moose with access to recent wildfire burns had greater zinc levels, suggesting that fire could enhance forage quality or availability. Collectively, these findings reveal patterns in moose nutritional health in relation to environmental conditions. Second, I measured concentrations of multiple immune biomarkers in the serum of female moose and investigated how these markers related to individual condition and parasite exposure. Moose with greater fat reserves had higher concentrations of interleukin-12, suggesting that individuals in better condition may be able to allocate more resources toward immune function. Total globulin concentrations were elevated in moose exposed to both microand macro-parasites, reflecting immune activation in response to parasitic challenges. I also found correlations between zinc levels and both IL-12 and total globulin, whereas copper concentrations were associated with haptoglobin, indicating a potential role of trace minerals in modulating immune responses. Combined, my results highlight connections between nutrition, immune function, and parasite exposure in moose. Collectively, my findings offer novel insights into patterns of variation in moose health in relation to environmental conditions. Moreover, my findings provide baseline data on a range of health biomarkers in female moose and highlight the importance of future monitoring to assess the effects of environmental change on wildlife health. iii Table of Contents Abstract........................................................................................................................................................ ii Table of Contents ....................................................................................................................................... iv List of Tables .............................................................................................................................................. vi List of Figures............................................................................................................................................. ix Acknowledgements .................................................................................................................................... xi CHAPTER 1: Introduction ........................................................................................................................ 1 Background ............................................................................................................................................. 1 Moose ecology in central British Columbia .......................................................................................... 5 Thesis objectives and format.................................................................................................................. 7 CHAPTER 2: Habitat composition, landscape disturbance, and climatic conditions are associated with hair mineral concentrations in moose............................................................................................... 9 Introduction ............................................................................................................................................. 9 Methods.................................................................................................................................................. 13 Study areas .......................................................................................................................................... 13 Animal capture, radio-collaring, and sample collection .................................................................... 16 Analysis of hair mineral concentrations ............................................................................................. 17 Environmental variables ..................................................................................................................... 18 Statistical analysis .............................................................................................................................. 23 Results .................................................................................................................................................... 26 Essential macro minerals .................................................................................................................... 27 Essential trace minerals ...................................................................................................................... 30 Discussion .............................................................................................................................................. 32 CHAPTER 3: Immune biomarkers vary in relation to body fat, trace mineral status, and parasite exposure in moose ..................................................................................................................................... 41 Introduction ........................................................................................................................................... 41 Methods.................................................................................................................................................. 46 Ethics statement .................................................................................................................................. 46 Study system ........................................................................................................................................ 46 Sample collection ................................................................................................................................ 47 Immune biomarkers ............................................................................................................................ 48 Physiological and parasitic variables................................................................................................. 49 Statistical analysis .............................................................................................................................. 51 Results .................................................................................................................................................... 54 Discussion .............................................................................................................................................. 63 iv CHAPTER 4: Conclusions ....................................................................................................................... 71 Research summary and implications .................................................................................................. 71 Limitations and future directions ........................................................................................................ 74 References .................................................................................................................................................. 78 APPENDIX A: Supplemental Information for Chapter 2 .................................................................... 94 APPENDIX B: Supplemental Information for Chapter 3 .................................................................. 102 v List of Tables Table 2.1. Variables considered to explain patterns of essential mineral concentrations in the hair (n = 60) of adult female moose (Alces alces) in central British Columbia, Canada, from 2020 to 2022……………………………………………………………………………………………....22 Table 2.2. A-priori mixed effects models used to explain concentrations of minerals in hair (n = 60) collected from female moose (Alces alces) during the winters of 2020–2022 in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model………...25 Table 2.3. Mean, standard deviation (SD), median, and range mineral concentrations (µg/g, dry weight) in hair (n = 60) from female moose (Alces alces) collected in the winters of 2020–2022 from populations in Bonaparte (BP; n = 31) and Prince George South (PGS; n = 29).…………26 Table 2.4. Candidate models and model selection statistics used to explain the hair mineral concentrations (n = 60) of adult female moose (Alces alces) in two study areas in central British Columbia, Canada, from 2020–2022. Models that were within the top model set (< 2 ΔAICc) and ranked higher than the null model, as well as null models, are included. The full list of candidate models and model selection statistics can be found in Appendix A. Variables in bold font indicate an influential relationship with the dependent variable (85% CI does not overlap 0). Predictor variables did not explain substantially more variation in the data than the null model for minerals Cu, Fe, Mn, and Mo, and therefore, these results are not presented. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model.……………………………………………..29 Table 3.1. Serum immune biomarker concentrations (n = 60) from adult female moose (Alces alces) sampled in winter (2020–2022) from two populations in central British Columbia, Canada: the Bonaparte Plateau (BP; n = 31) and Prince George South (PGS; n = 29). Cytokine concentrations are reported in pg/mL, and globulin and haptoglobin in g/L. Cytokine concentration values that fell below the assay detection range (out of range; OOR) and could not be extrapolated were excluded from descriptive statistics. Fluorescence intensity values, which do not have a defined limit of detection, are reported for the full sample size. …………………56 Table 3.2. Summary of physiological and parasitic variables (n = 60) used to predict immune biomarker concentrations in female moose (Alces alces) serum sampled during the winters of 2020–2022 in two populations: the Bonaparte Plateau (BP; n = 31) and Prince George South (PGS; n = 29), central British Columbia, Canada. Continuous variables (body fat percentage and serum trace minerals Cu, Se, and Zn) are presented as mean ± standard deviation. Categorical variables (pregnancy status, winter tick presence, and alphaherpesvirus serostatus) are presented as counts with corresponding percentages in parentheses. These values represent averages and proportions calculated across both study areas, all three sampling years, and include repeated measurements from individuals………………………………………………………………….57 vi Table 3.3. Model selection statistics used to predict serum immune biomarker concentrations (n = 60 for most biomarkers; haptoglobin: n = 58 due to removal of two outliers) in adult female moose (Alces alces) in two study areas in central British Columbia, Canada, from 2020–2022. Models that were within the top model set (< 2 ΔAICc) and ranked higher than the null model, as well as null models, are included. Variables in bold font indicate an influential relationship with the dependent variable (85% CI does not overlap 0). Predictor variables did not explain substantially more variation in the data than the null model for immune biomarkers IL-1β, IL-10, and IL-4, and therefore, these results are not presented. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as conditional fixed effects in every model…………………………………………………………59 Table A.1. Full candidate models and model selection statistics used to explain potassium (K) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model…….94 Table A.2. Full candidate models and model selection statistics used to explain magnesium (Mg) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model…….95 Table A.3. Full candidate models and model selection statistics used to explain copper (Cu) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model…….96 Table A.4. Full candidate models and model selection statistics used to explain iron (Fe) concentrations in the hair (n = 59) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model…….97 Table A.5. Full candidate models and model selection statistics used to explain manganese (Mn) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model…….98 Table A.6. Full candidate models and model selection statistics used to explain molybdenum (Mo) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model……………………………………………………………………………………………..99 Table A.7. Full candidate models and model selection statistics used to explain selenium (Se) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model…...100 vii Table A.8. Full candidate models and model selection statistics used to explain zinc (Zn) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model…...101 viii List of Figures Figure 2.1. Locations of Prince George South (top) and the Bonaparte Plateau (bottom), the two study areas for exploring connections between adult female moose (Alces alces) hair mineral concentrations and environmental conditions in central British Columbia, Canada, from 2020 to 2022………………………………………………………………………………………………15 Figure 2.2. Coefficient estimates and 85% confidence intervals (CI) for independent variables in the top-ranked generalized linear mixed effects models (< 2 ΔAICc) explaining (A) K concentrations and (B) Mg concentrations, two macro minerals in female moose (Alces alces) hair. Hair samples (n = 60) were collected during capture events in the winters of 2020–2022 from two study areas (Prince George South and the Bonaparte Plateau) in central British Columbia, Canada. An independent variable has an influential relationship with the mineral if the CI does not overlap 0. Continuous predictors deciduous forest and mixed forest were standardized to a mean of zero and a standard deviation of one prior to analysis. Coefficients were ordered from most positive to most negative based on standardized estimates……………28 Figure 2.3. Coefficient estimates and 85% confidence intervals (CI) for independent variables in the top-ranked generalized linear mixed effects models (< 2 ΔAICc) explaining (A) Se concentrations and (B) Zn concentrations, two trace minerals in female moose (Alces alces) hair. Hair samples (n = 60) were collected during capture events in the winters of 2020–2022 from two study areas (Prince George South and the Bonaparte Plateau) in central British Columbia, Canada. An independent variable has an influential relationship with the mineral if the CI does not overlap 0. Continuous predictors temperature and precipitation were standardized to a mean of zero and a standard deviation of one prior to analysis. Coefficients were ordered from most positive to most negative based on standardized estimates……………………………………...30 Figure 3.1. Coefficient estimates and 85% confidence intervals (CI) for independent variables in the top-ranked generalized linear mixed effects models (< 2 ΔAICc) explaining interleukin-12 concentrations in female moose (Alces alces) serum. Serum samples (n = 60) were collected during capture events in the winters of 2020–2022 from two study areas (Prince George South and the Bonaparte Plateau) in central British Columbia, Canada. An independent variable has an influential relationship with the mineral if the CI does not overlap 0. Continuous predictors serum Zn and body fat were standardized to a mean of zero and a standard deviation of one prior to analysis. Coefficients were ordered from most positive to most negative based on standardized estimates…………………………………………………...……………………………………..58 Figure 3.2. Coefficient estimates and 85% confidence intervals (CI) for independent variables in the top-ranked generalized linear mixed effects models (< 2 ΔAICc) explaining total globulin concentrations in female moose (Alces alces) serum. Serum samples (n = 60) were collected during capture events in the winters of 2020–2022 from two study areas (Prince George South and the Bonaparte Plateau) in central British Columbia, Canada. An independent variable has an influential relationship with the mineral if the CI does not overlap 0. The continuous predictor serum Zn was standardized to a mean of zero and a standard deviation of one prior to analysis.61 ix Figure 3.3. Coefficient estimates and 85% confidence intervals (CI) for independent variables in the top-ranked generalized linear mixed effects models (< 2 ΔAICc) explaining haptoglobin concentrations in female moose (Alces alces) serum. Serum samples (n = 58; two outliers removed) were collected during capture events in the winters of 2020–2022 from two study areas (Prince George South and the Bonaparte Plateau) in central British Columbia, Canada. An independent variable has an influential relationship with the mineral if the CI does not overlap 0. Continuous predictors serum Cu and serum Se were standardized to a mean of zero and a standard deviation of one prior to analysis. Coefficients were ordered from most positive to most negative based on standardized estimates. AHV = Alphaherpesvirus…………………………...62 Figure B.1. Spearman correlation matrix showing paired correlation coefficients between immune biomarkers concentrations measured in serum samples (n = 60) from adult female moose sampled in winter (2020–2022) from two populations in central British Columbia, Canada: the Bonaparte Plateau (BP; n = 31) and Prince George South (PGS; n = 29). Each coloured box represents the strength and direction of the paired correlation, with colours ranging from red (negative correlation) to blue (positive correlation). The correlation coefficient ranges from -1 (perfect negative correlation) to 1 (perfect positive correlation), with 0 indicating no correlation. Immune biomarkers GM-CSF, IFN-γ, IL-8, and TNF-α were excluded from this analysis due to having less than 80% detectable samples.…………………...…………………102 x Acknowledgements This thesis is the result of countless hours of collaboration and support from many people. First and foremost, I am deeply grateful to my supervisor, Dr. Heather Bryan, for her unwavering guidance, encouragement, and patience over the past three years. I have benefited immensely from Heather’s mentorship and enthusiasm, and I believe her support has been invaluable to my growth as an aspiring wildlife scientist. Heather embodies the kind of teacher and mentor I aspire to be in my own career. I also wish to thank my committee members, Drs. Roy Rea, Caeley Thacker, and Owen Slater, for their thoughtful feedback, insightful discussions, and the many ideas that deepened my understanding of my research and strengthened this thesis. This work would not have been possible without the collaboration and support of the many people involved in the Provincial Moose Research Project. I feel fortunate to have worked alongside a dedicated team of scientists, wildlife managers, and resource professionals committed to understanding and conserving moose populations in British Columbia. The expertise and insights shared by provincial wildlife experts offered valuable perspectives that greatly enriched my research experience. I am especially grateful to Morgan Anderson and Shari Willmott for their prompt assistance and help with data sharing, which made my analyses possible. Thank you to the many funders that made this research possible. This work was supported by the Habitat Conservation Trust Foundation (Project 7-588), the Forest Enhancement Society of BC, the National Sciences and Engineering Council of Canada (NSERC; RGPIN-2020-06845), the BC Graduate Scholarship Program (BCGSP), an Al Martin HCTF Conservation Fellowship, the Alces Journal, and EcoCanada. To the many members of the Wildlife and Ecosystem Bioindicators Lab—far too numerous to name individually—thank you for the friendships, venting sessions, turtle mocha breaks, and for making Prince George feel like a home away from home. I am incredibly fortunate to have worked alongside such a compassionate and enthusiastic group of people. In particular, I’d like to thank Caroline Lesage, Carl-Evan Jefferies, and our unofficial lab member Lisa Koetke for their invaluable assistance with laboratory and/or coding work. Finally, I know I could not have completed this project without the unwavering support of my family and friends. To Simon (and Linus), thank you for encouraging me to pursue my passion for wildlife. Your love and endless support over the past few years have allowed me to immerse myself in this work. I am forever grateful for your sacrifice in moving over 4,000 km so I could follow my dreams and for standing by me through the most challenging parts of this journey with your encouragement, empathy, and understanding. xi CHAPTER 1: Introduction Background In the face of rapid environmental change, wildlife are increasingly subjected to pressures that may compromise individual fitness and population performance (Acevedo-Whitehouse and Duffus, 2009). Landscape disturbances caused by both natural and human activities, along with climate change, altered predator-prey dynamics, pollution, pathogen spill-over, overexploitation, and the complex interactions among these pressures, have contributed to the declines of many species (Ceballos et al., 2015; Newbold et al., 2015). Understanding the relative importance and mechanisms by which these pressures affect different species is essential for identifying and prioritizing stewardship strategies that promote the long-term sustainability of wildlife populations. Monitoring biomarkers of health in wildlife provides valuable insight into the physiological impacts of environmental change on wildlife populations. Historically, wildlife health was viewed narrowly as the mere presence or absence of disease (Stephen, 2014). However, in recent decades, the concept of wildlife health has evolved to encompass a dynamic interplay of biological, social, and environmental factors that collectively influence the resilience and sustainability of wildlife populations through effects on individual fitness (Stephen, 2014; Thacker et al., 2019; Wittrock et al., 2019; Aleuy et al., 2023). Health biomarkers are inherently dynamic, varying across time and space, and can reflect both intrinsic biological characteristics (e.g., immunity, nutrient acquisition, stress response, genetics) and extrinsic factors (e.g., climate and weather, habitat features) (Thacker et al., 2019; Aleuy et al., 2023). Assessing biomarkers of wildlife health at both the individual and population level allows for early detection of deviations 1 from baseline health, thereby informing targeted and effective monitoring strategies (Wikelski and Cooke, 2006; Cooke et al., 2013). Climate and landscape disturbances can affect wildlife health and fitness by altering habitat conditions and resource availability, ultimately shaping population dynamics through complex interactions (Acevedo-Whitehouse and Duffus, 2009). These disturbances can directly influence the quality and quantity of forage within habitats and the energetic balance (i.e., intake versus expenditure) of some wildlife species (Schmidt et al., 2002; van Beest et al., 2012; Oster et al., 2018; van Beest et al., 2023). Understanding the relative influences of climate and landscape disturbances on the nutritional value of habitat is particularly important in large herbivores, as nutrition is a key determinant of reproduction and survival for many species (Parker et al., 2009; Stephenson et al., 2020). Among the various indicators of wildlife health, biomarkers representing mineral status and immune function offer valuable insight into the multiple, interacting environmental factors affecting wild animals through changes in nutritional status and physiological responses. Since mineral status and immune competence are closely linked to fitness traits (Ohmer et al., 2021; Mosbacher et al., 2022; Rioux et al., 2022), including reproduction and survival, measuring these biomarkers in response to environmental variation can help uncover the mechanisms connecting anthropogenic disturbances to individual health, condition, and ultimately population dynamics. Macro and trace minerals, including essential elements and toxic heavy metals, are important components of an animal’s diet that are associated with individual health. Mammals require up to 29 minerals to grow, reproduce, survive, and maintain proper physiological function (Kincaid, 2000; Suttle, 2010; Oster et al., 2024). Macro minerals, such as calcium, magnesium, and potassium, are required in relatively large amounts to facilitate various 2 metabolic and physiological processes, such as maintaining homeostasis, regulating metabolism, ensuring organ function, and promoting optimal growth and development (Kincaid, 2000; Underwood, 2012). In contrast, trace minerals, such as copper, selenium, and zinc, are required in much smaller amounts but are crucial for supporting enzymatic reactions, immune function, and antioxidant defenses (Kincaid, 2000; Underwood, 2012). In free-ranging ungulates, mineral deficiencies or toxicities have been associated with impaired growth (Pollock, 2005), reduced reproductive success (Flynn et al., 1977; Stephenson et al., 2001), and compromised immunity (Flueck, 2012). In some cases, inappropriate levels of minerals have been implicated in population declines (O’Hara et al., 2001; Frank et al., 2004; Murray et al., 2006). Thus, identifying factors associated with mineral concentrations in animal tissues may provide important insights into both population health and broader patterns of nutritional condition. The bioavailability of minerals to herbivores is likely to change with shifts in vegetation quality and availability caused by environmental change (Ohlson and Staaland, 2001; van Beest et al., 2023), which could have consequences for animal health. Climate and landscape alterations can influence mineral bioavailability by modifying soil chemistry, nutrient cycling, and plants’ ability to uptake minerals (Schmidt et al., 2002; Oster et al., 2018; van Beest et al., 2023). Additionally, reductions in overall forage abundance due to environmental change may limit herbivores’ capacity to consume sufficient quantities of mineral-rich plants. Understanding the effects of environmental conditions on mineral levels can therefore provide insights into herbivore responses to environmental change, thereby informing management strategies aimed at improving health. For example, identifying areas where herbivores exhibit lower essential mineral levels could help prioritize habitat protection or guide efforts such as mineral supplementation where appropriate. 3 Traditionally, mineral concentrations in animals have been measured using organ tissues (e.g., liver or kidney) or serum samples (McDowell, 1992; Herdt and Hoff, 2011). In recent years, hair analysis has emerged as a valuable alternative for assessing mineral status in wildlife, particularly in research and monitoring contexts (Cygan-Szczegielniak et al., 2018; Jutha et al., 2022; Mosbacher et al., 2022; Rioux et al., 2022). Hair offers a practical advantage over other tissue types as it can be easily collected during capture, is simple to store, and reflects long-term mineral accumulation, making it particularly suitable for field studies. Additionally, hair mineral concentrations may offer valuable insights into population-level trends, as they are often associated with individual health and demographic outcomes. For example, Rioux et al. (2022) observed that higher zinc and sodium concentrations, along with lower cesium and manganese levels, were linked with improved adult survival in caribou (Rangifer tarandus caribou). A similar study in muskoxen (Ovibos moschatus) revealed that variations in copper, selenium, and molybdenum levels were associated with changes in annual calf recruitment (Mosbacher et al., 2022). Together, these studies highlight the potential of hair as an informative tissue for measuring mineral concentrations and examining their relationships with environmental characteristics in free-ranging wildlife. Immune function is another physiological process associated with wildlife health and resilience, serving as a primary defense against pathogens and shaping patterns of disease occurrence and outcomes in the host (Schmid-Hempel, 2021). The immune system is complex and multifaceted, encompassing innate and adaptive responses that work together to protect animals from infections, parasites, and other challenges. Hosts differ markedly in their ability to mount immune responses, reflecting differences in genetic background, life-history characteristics, nutritional condition, prior exposure to pathogens, and environmental factors. 4 Understanding the dynamics of immunity is therefore useful in monitoring population health for conservation (Brock et al., 2012, Ohmer et al., 2021). Given that immune function is energetically costly and requires considerable nutrients, immune biomarkers can offer insight into fitness-related traits and energetic trade-offs. Animals allocate limited energy among immunity, growth, reproduction, and maintenance, and these investments may shift under nutrient restrictions (Stearns, 1989; Sheldon and Verhulst, 1996). For example, animals in poor body condition may exhibit reduced immune reactivity (GilotFromont et al., 2012). In roe deer (Capreolus capreolus), immune phenotypes varied with body condition, where individuals in better condition had higher levels of parameters related to innate immunity (Gilot-Fromont et al., 2012). In addition, direct trade-offs between reproductive effort and immune function have been documented in Soay sheep (Ovis aries), where high antibody responsiveness was associated with improved survival during harsh winters but reduced reproductive output (Graham et al., 2010). Certain nutrient restrictions, including imbalances in trace minerals, may also influence immunity independent of energy restriction (Jolles et al., 2015). Trace minerals play a crucial role in immune function by supporting enzymatic processes, antioxidant defenses, and cellular immunity, further linking nutritional status with immunocompetence (Kincaid, 2000; Underwood, 2012). Therefore, environmental change may exacerbate trade-offs in resource allocation by altering immunological, endocrinological, and physiological responses, which in turn could profoundly affect immunity and infectious disease dynamics. Moose ecology in central British Columbia A better understanding of moose (Alces alces) health can provide valuable insights into patterns associated with environmental change at both individual and population levels. As the 5 largest living members of the deer family (Cervidae), moose occur throughout the boreal and temperate forests of North America and Eurasia (Franzmann, 1981). Moose are an iconic species of northern ecosystems that play critical roles in predator-prey systems, nutrient cycling, and forest succession. Moreover, moose are culturally significant to many First Nations and economically important to hunters and guide outfitters. Although moose populations fluctuate naturally over time, there have been notable declines in parts of North America (Timmermann and Rodgers, 2017). In some regions of British Columbia (BC), Canada, moose populations declined by 50–70% between the early 2000s and 2010 (Kuzyk et al., 2018). This period coincided with a severe mountain pine beetle (Dendroctonus ponderosae) outbreak, which caused extensive pine tree mortality across much of the province. Subsequently, large-scale salvage logging, primarily through clearcutting, led to a sharp increase in the Allowable Annual Cut (Parfitt, 2007). Rapid forest disturbance has been hypothesized as a factor contributing to moose declines in the central interior of BC. In 2012, the Province of British Columbia and partners initiated a long-term research project to understand the effects of natural and anthropogenic disturbances on moose population declines and to identify restoration management options (Kuzyk and Heard, 2014). As part of this initiative, provincial biologists have monitored adult female and calf moose across five study areas representing a range of forest disturbance intensities and ecosystem types to examine patterns in moose distribution, health, and survival. Since the beginning of the project, adult females have been captured annually to maintain approximately 30 active GPS collars per study area, and biological samples have been collected from both live-captured animals and mortality sites. Although predation was the most frequently identified cause of mortality, investigations into adult female deaths revealed a higher-than-expected number of cases linked to health-related 6 factors, along with evidence of suboptimal pregnancy rates across the province (Thacker et al., 2019). Interestingly, an initial assessment of trace mineral concentrations in organ tissues and serum revealed variation across populations, with suboptimal levels of several trace minerals (e.g., iron, cobalt, copper, manganese, selenium, and zinc) in some individuals. Moreover, investigations of adult female moose health suggest that the occurrence and potential impacts of several additional health determinants, such as body fat, stress, and pathogen prevalence, may vary across study areas and years. Together, these findings highlight the need for further research to better understand how moose health relates to environmental variation across different landscapes. Given the complex interactions among natural and human disturbances and climate factors affecting moose habitat and population dynamics, a deeper understanding of moose health at individual and population levels is essential. Health biomarkers can serve as early indicators of broader ecological change, especially in regions undergoing rapid landscape transformation. Currently, no single factor fully explains the observed differences in overall health among BC moose populations. Although some populations in central BC have shown signs of stabilization in recent years, ongoing monitoring remains critical to determine whether these trends persist and to support effective management interventions. Therefore, ongoing monitoring that expands the range and types of health biomarkers, combined with long-term, multi-level studies, is critical to thoroughly assessing how biomarkers of individual health may relate to and potentially help to predict moose population dynamics in the region. Thesis objectives and format My thesis explores how mineral concentrations and immune responses vary in relation to environmental and individual conditions in moose from two of the provincial study areas. It is 7 structured into four chapters. In Chapter 1, “Introduction”, I provided an overview of the importance of assessing wildlife health in the context of environmental change. I also described the historical and ecological context of moose populations in British Columbia (BC), introduce the study system, and outline the landscape change hypothesis originally proposed by Kuzyk and Heard (2014). In Chapter 2, titled “Habitat composition, landscape disturbance, and climatic conditions are associated with hair mineral concentrations in moose”, I examined associations between hair mineral concentrations and environmental conditions using GPS collar data, spatial datasets, and hair samples collected during winter captures. I also established baseline hair mineral concentrations for this demographic. In Chapter 3, titled “Immune biomarkers vary in relation to body fat, trace mineral status, and parasite exposure in moose”, I investigated how aspects of host physiological condition and parasite exposure are associated with variation in immune function by measuring a suite of serum immune biomarkers that represent different components of the immune system. Chapters 2 and 3 are presented in manuscript format for journal publication and use first-person plural to recognize the contributions of my collaborators. In Chapter 4, “Conclusions”, I provide a final summation of the results and synthesize the findings of both chapters. I also address the general limitations of this study and offer recommendations for future research. Overall, this research advances our understanding of adult female moose health in changing environments. My work offers novel insights into potential ways by which moose respond to landscape disturbances and climate change and expands the suite of health biomarkers that can be used for longitudinal monitoring at individual and population levels. 8 CHAPTER 2: Habitat composition, landscape disturbance, and climatic conditions are associated with hair mineral concentrations in moose Introduction The current unprecedented rate of environmental change is imposing stressors on wildlife populations around the globe. Climate change and landscape disturbances shape wildlife population dynamics in part via bottom-up effects on wildlife health and fitness (AcevedoWhitehouse and Duffus, 2009). Climate, for example, affects forage quantity and quality for herbivores via effects on plant growth rate, species composition, and phenology (Post et al., 2008; Park et al., 2020; Brown et al., 2022). Moreover, anthropogenic and natural landscape disturbances, such as industrial forest harvesting and wildfire, can alter the quality and composition of vegetation through changes in forest structure and successional stages (Kasischke et al., 2006; Rocha-Santos et al., 2016; Whitman et al., 2019). These changes in vegetation affect large herbivores that rely on plant communities for nutrition and whose population growth may be limited by the nutritional quality of their habitat (Parker et al., 2009; Stephenson et al., 2020). Understanding how climate and landscape disturbances relate to the nutritional value of habitat is therefore particularly important in large herbivores. Mineral status is one nutritional currency that is associated with the health and fitness of large herbivores (Barboza et al., 2009). Essential minerals, including macro minerals (elements required in large amounts) and trace minerals (elements needed in smaller quantities), are required in the diet of all animals to support physiological and biochemical functions such as growth, immunity, reproduction, and survival (Kincaid, 2000; Underwood, 2012). In contrast, non-essential elements, such as toxic heavy metals, have no established biological functions but can interfere with the absorption or activity of essential minerals (Abdulla and Chmielnicka, 9 1989; Ali and Khan, 2018). Animals may experience adverse health effects when homeostatic mechanisms that regulate levels of specific minerals are disrupted and mineral levels in tissues fall above or below the normal physiological range (Baj et al., 2023). Although the importance of minerals for physiological functioning in domestic livestock is well established, little is known about their role in wild herbivores (Blakley et al., 2000; French et al., 2017; Bondo et al., 2019; Jutha et al., 2022). However, similar effects of mineral imbalances—whether deficiencies or toxicities—are presumed to occur in wild herbivores, and have been linked with poor health, reduced reproductive success, and compromised population performance (Flynn et al., 1977; Flueck, 1994; O’Hara et al., 2001; Flueck et al., 2012; Newby and DeCesare, 2020). Therefore, monitoring of minerals should be a key component of evaluating health trends in individuals and populations of free-ranging herbivores. Large herbivores derive the bulk of their minerals from forage but may supplement their intake with alternative nutrient-rich sources, such as mineral licks (Spears, 1994; Ayotte et al., 2006). The quality, abundance, and accessibility of minerals in forage vary across time and space (Ohlson and Staaland, 2001; van Beest et al., 2023). Despite this variation, individuals must obtain sufficient quantities of minerals, as demanded and constrained by their physiology. Mineral acquisition is further complicated by changes in mineral bioavailability in forage resulting from climate change and landscape disturbances. These factors can modify mineral bioavailability by altering soil chemistry, nutrient cycling, and the mineral uptake capacity of plants (Schmidt et al., 2002; Oster et al., 2018; van Beest et al., 2023). Moreover, reductions in overall forage abundance caused by environmental change may limit the ability of herbivores to ingest sufficient quantities of mineral-rich plants. Although recent studies have explored associations between minerals and health in free-ranging herbivores (Jutha et al., 2022; 10 Mosbacher et al., 2022), few have investigated how environmental conditions correspond to variation in mineral concentrations. Exploring these associations could offer insights into herbivore responses to environmental change and reinforce the use of mineral concentrations as biomarkers of health. Moose (Alces alces) in central British Columbia (BC), Canada, are ideal subjects to explore associations between environmental conditions and mineral concentrations. In this region, populations of moose declined by as much as 70% in the early 2000s (Kuzyk et al., 2018). This period coincided with a widespread outbreak of mountain pine beetle (Dendroctonous ponderosae) and subsequent salvaging logging of beetle-killed timber, resulting in a heavily altered landscape (Alfaro et al., 2015). The dramatic declines led to a coordinated research project between the Province of BC and partners to understand the effects of landscape disturbances on moose populations. Observations of apparent starvation and health-related mortalities, coupled with suboptimal pregnancy rates, suggest that bottom-up factors may be influencing the viability of populations (Thacker et al., 2019). Minerals play a particularly important role in the health of moose, where deficiencies in essential minerals such as cobalt, copper, iron, and selenium have been identified as contributing to moose mortalities, poor reproductive output, and ultimately poor performance in several populations in North America (Flynn et al., 1977; O’Hara et al., 2001; Frank et al., 2004; Murray et al., 2006). Despite these findings elsewhere, our understanding of mineral levels in moose in central BC remains somewhat limited, partly due to logistical challenges associated with assessing mineral concentrations in free-ranging wildlife. In animals, mineral concentrations are typically quantified using storage organs (i.e., liver or kidney) since these organs reflect the availability of minerals in the body (McDowell, 1992). 11 However, analyzing organs in wildlife requires either invasive biopsies or post-mortem sampling. In free-ranging wildlife, organs are often consumed by predators or scavengers before they can be collected as part of mortality investigations. Therefore, in live-captured animals, mineral concentrations are measured most commonly in blood or serum (Herdt and Hoff, 2011). Mineral concentrations in blood and serum, however, reflect only short-term mineral status that fluctuates with metabolic demands and homeostatic controls (Underwood, 2012) and often correlate poorly with those in storage organs (Blakley et al., 2000). As an alternative approach to evaluate mineral concentrations, hair sampling and analysis is increasingly being used to assess mineral concentrations in wildlife research and monitoring (e.g., Jutha et al., 2022; Mosbacher et al., 2022; Rioux et al., 2022; Herrada et al., 2024; Dickinson et al., 2025). Minerals are incorporated into the hair shaft during the period of hair growth from circulating blood that feeds the growing hair follicle (Combs, 1987). Hair growth occurs in a defined time period after which it becomes metabolically inactive; thus, hair mineral concentrations reflect an animal’s physiological status at the time of hair growth (Combs, 1987). Previous research has used hair to assess individual health in woodland caribou (Rangifer tarandus caribou; Jutha et al., 2022) and as an indicator of demographic rates and broader population trends in woodland caribou (Rioux et al., 2022) and muskoxen (Ovibos moschatus; Mosbacher et al., 2022). These studies highlight the potential of using hair as an indicator of mineral status, however, the extent to which hair concentrations reflect whole-body mineral levels remains unclear. Hair may not always provide an accurate representation of an animal’s overall mineral status, and factors such as age and sex can also influence mineral concentrations (Combs, 1987). Despite these limitations, hair sampling offers a promising alternative approach for assessing long-term concentrations of select 12 minerals in wildlife, particularly once validated and in situations where conventional monitoring methods are impractical (Jutha et al., 2022). Here, we studied the associations between environmental conditions, habitat composition, and mineral concentrations in moose sampled from two populations in central British Columbia, from 2020 to 2022. We used hair samples collected from adult female moose at capture in winter to measure a suite of 15 minerals known to be related to health and fitness outcomes. Then, we tested whether indices of habitat composition, landscape disturbance, and/or climatic conditions were correlated with differences in the uptake of minerals into hair. We predicted that environmental conditions that are known to influence the quality or quantity of forage would correspond with differences in mineral concentrations in moose hair. We also documented baseline hair mineral concentrations in female moose and demonstrated the potential of using hair mineral concentrations to reveal patterns in moose responses to environmental variation. Methods Study areas Our study took place in two areas in central British Columbia, Canada (Fig. 2.1). The Bonaparte Plateau (BP) is located on the traditional territory of the Secwépemc First Nation, north of Kamloops, BC. The BP study area encompasses 6,800 km2 and lies at 51°13′ N latitude and 120°81′ W longitude. The Prince George South (PGS) study area is located on the traditional territories of the Lheidli T’enneh and Saik’uz First Nations, southwest of Prince George, BC. The PGS study area covers an area of 11,000 km2 and lies at 53°56′ N latitude and 123°63′ W longitude. The climate of both study areas is humid continental, characterized by short and dry, warm summers and long, cold winters. On the BP, the long-term mean annual temperature is 13 9.5°C, with long-term average winter temperatures around 0.1°C (December to March) and summer temperatures around 20.5°C (June to August) (Environment and Climate Canada, 2024). The BP area receives an average of 215.9 mm of rainfall and 63.1 cm of snowfall annually (Environment and Climate Change Canada, 2024). In PGS, the mean annual temperature is 4.3°C, with winter conditions averaging -5.2°C and summer temperatures near 14.9°C (Environment and Climate Change Canada, 2024). The PGS area receives an average of 432.0 mm of rainfall annually, along with 203.9 cm of snowfall (Environment and Climate Change Canada, 2024). The BP area is characterized by three dominant Biogeoclimatic Ecosystem Classification Zones, including Interior Douglas‐fir (IDF), Sub‐Boreal Pine‐Spruce (SBPS), and Montane Spruce (MS), whereas PGS is primarily within the Sub-Boreal Spruce (SBS) Biogeoclimatic Ecosystem Classification (BEC) zone (Meidinger and Pojar, 1991). Vegetation in the study areas is diverse, consisting of coniferous and deciduous forests at various seral stages, along with nonforested habitats such as wetlands and lakes. Common coniferous tree species in both study areas include lodgepole pine (Pinus contorta var. latifolia), hybrid spruce (Picea glauca × engelmannii), subalpine fir (Abies lasiocarpa), and Douglas-fir (Pseudotsuga menziesii), while broadleaf deciduous species include black cottonwood (Populus balsamifera), trembling aspen (Populus tremuloides), and paper birch (Betula papyrifera; Meidinger and Pojar, 1991). Historically, wildfire was the primary disturbance regime in central BC but has largely been replaced by commercial forestry. Both study areas have recently undergone extensive salvage logging in response to large-scale mountain pine beetle outbreaks (Alfaro et al., 2015). In addition to moose, both study areas support populations of other large cervids, including mule deer (Odocoileus hemionus), white-tailed deer (Odocoileus virginianus), and elk 14 (Cervus canadensis). The primary predator of adult moose is wolves (Canis lupus), though black bears (Ursus americanus), grizzly bears (Ursus arctos), and cougar (Puma concolor) are also present. Moose density was estimated to be 254 ± 41/1000 km² in BP during the winter of 2017– 2018 and 400 ± 78/1000 km² in PGS during the winter of 2016–2017 (Kuzyk et al., 2018). Figure 2.1. Locations of Prince George South (top) and the Bonaparte Plateau (bottom), the two study areas for exploring connections between adult female moose (Alces alces) hair mineral concentrations and environmental conditions in central British Columbia, Canada, from 2020 to 2022. 15 Animal capture, radio-collaring, and sample collection We focused our study on 31 adult female moose (≥1.5 years old, nBP = 17, nPGS = 14) captured in the winters (December–February) of 2020–2022 by the Province of BC as part of a large-scale province-wide study investigating factors affecting moose population change (Kuzyk et al., 2018). For simplicity, we grouped moose captured between the consecutive months of December and February as part of a single study year (i.e., individuals captured between December 2019 and February 2020 are considered part of the 2020 study year). Moose were captured using either aerial net gunning and physical restraint or chemical immobilization by aerial darting. Capture personnel equipped female moose with GPS-telemetry collars (Vectronic Aerospace VERTEX Survey Globalstar or Survey Iridium radio collars, or Advanced Telemetry Systems G2110E radio collars) that were programmed to record one to six fixes per day (Procter et al., 2020). Hair samples were collected from each individual moose between the shoulders using needle-nosed pliers in an area that was dry and visually free of contaminants (Procter et al., 2020). The hair samples were dried and stored in envelopes at room temperature prior to sample analysis (Thacker et al., 2019). Moose were recaptured and sampled in successive years where possible to support longitudinal monitoring of the health status of individual moose (Thacker et al., 2019). The detailed methods used for animal capture, sampling, and monitoring have been previously described (Kuzyk et al., 2018, Procter et al., 2020). All captures were conducted in accordance with the British Columbia Wildlife Act under permit CB17-277227. In addition, analysis of data from this project was approved by the Animal Care and Use Committee at the University of Northern British Columbia (ACUC Protocol Number 2021-01). 16 Analysis of hair mineral concentrations The sample preparation and analysis of hair were conducted following methods previously described (Jutha et al., 2022; Rakic, 2022; Aguilar et al., 2023). We inspected hair samples under a dissecting microscope and manually removed hair follicles, bulbs, or undercoat if present. Using plastic forceps, we thoroughly washed hair samples three times in 95% ethanol (Greenfield Global Inc.) followed by ultrapure Type 1 reagent-grade water to eliminate all possible surface contamination and external element deposits from environmental exposure (Smith et al., 2007). We transferred washed samples into clean paper envelopes and oven-dried them at 50˚C for 24 hours. Once fully dried, we weighed 70 mg of dried hair and added it to a Teflon vial along with 2 mL of 70% nitric acid (HNO3 [TraceMetal™ Grade], Fisher Chemical™). The vials were digested using a high-pressure microwave reactor (ETHOS EZ Microwave Digestion System, Milestone, Shelton, CT, USA). The digester temperature was gradually increased from room temperature to a peak of 220°C over one hour, and then gradually cooled to room temperature over one hour. We transferred digested samples to Falcon tubes, diluted them to 4 mL with Type 1 water, and stored them at 5˚C until analysis. Hair mineral concentrations were measured at the Alberta Centre for Toxicology, University of Calgary. There, digested samples were further diluted 10X with Type 1 water and introduced to the inductively coupled plasma mass spectrometer (ICP-MS, 8800 Triple Quadrupole ICP-MS, Agilent) to analyze a panel of 15- minerals: calcium (Ca), chromium (Cr), cobalt (Co), copper (Cu), iron (Fe), magnesium (Mg), manganese (Mn), molybdenum (Mo), potassium (K), selenium (Se), sodium (Na), and zinc (Zn), along with three contaminant heavy metals: arsenic (As), cadmium (Cd), and lead (Pb). Instrument calibration verification was conducted before, during, and after sample analyses using certified reference materials (Trace 17 Elements in Natural Water [NIST1643f]; Multi-Element Standard [SCP Science]; and Environmental Calibration Standard [Agilent]). Within each analysis run, one digested hair sample was randomly selected to be run in duplicate, where a maximum deviation limit of 20% between duplicates was set for the results to be accepted. For samples run in duplicate, the average of the two mineral concentration values was used for analysis. The limit of quantification (LOQ; wet weight, digested sample) for Na was 3.0 mg/L, for Ca, Mg, and K was 1.0 mg/L, for Fe was 0.5 mg/L, for Cr, Cu, Zn, As, and Cd was 0.005 mg/L, for Mn and Se was 0.001 mg/L, and for Co, Mo, and Pb was 0.0001 mg/L. Mineral concentrations detected but falling below the LOQ were included in the analysis; values were omitted in cases where concentrations fell below the limit of detection (LOD). Quality assurance was verified in each batch using certified reference materials (NIST2976 freeze-dried mussel tissue, National Institute of Standards and Technology; and NRC DORM-4 "Fish Protein Certified Reference Material for Trace Metals", National Research Council Canada) as positive controls, along with blank samples as negative controls. We reported mineral concentrations on a µg/g dry hair weight basis. Environmental variables We surveyed the literature to identify explanatory variables to test a-priori hypotheses about how environmental conditions might predict variation in the hair mineral concentrations of moose. These variables corresponded with hypotheses within three explanatory categories: habitat composition, landscape disturbance, and climatic conditions (Table 2.1). Minerals are incorporated into hair from circulating blood during the period when hair is growing (Combs, 1987), which is assumed to begin in early summer and continue until late fall in moose, with moulting commencing the following early spring (Sokolov and Chernova, 1987). Accordingly, 18 hair samples collected in the winter (December–February) from female moose during captures reflect mineral deposition that occurred during the preceding early summer to late fall. To characterize the environmental conditions and habitat used by moose, we therefore used GPS collar locations collected between June 1 and October 31 in the year preceding hair collection. In instances where moose were captured for the first time and GPS collar data were not available prior to capture (n = 18 hair samples), we assumed site fidelity and used GPS collar locations from the June–October period succeeding the capture event (Van Dyke et al., 1995; Scheideman, 2018; McLaren and Patterson, 2021). Plant community composition and diversity have strong effects on the diversity of minerals available in forage plants (Ohlson and Staaland, 2001). In central BC, moose consume both coniferous and deciduous trees across seasons (Koetke et al., 2023), which vary in their nutritional properties, including the concentrations of several minerals (e.g., Ca, Cu, K, Mg; Richardson and Friedland, 2016). Therefore, we expected that the vegetation composition of a forest stand would affect the forage species moose consume, and consequently, an individual’s mineral status. To assess use of different stand types by individual moose, we calculated the proportion of GPS points for each individual moose in coniferous, mixed, and deciduous forest stands, using the Provincial Vegetation Resource Inventory (VRI; DataBC, 2024). We classified each stand as either coniferous or deciduous if the associated dominant leading species made up ≥70% of the trees in the patch. If the leading species made up <70%, we classified the stand based on the leading species if the second most common species belonged to the same category (i.e., both coniferous or both deciduous). A patch was classified as mixed if the leading tree species comprised <70% of trees in the patch and the second leading tree species was in the alternate category. 19 Wetlands and riparian habitat contain aquatic plant species that enable moose to meet essential mineral and nutrient requirements (Tischler et al., 2019). For instance, aquatic plants tend to have greater concentrations of Na, and to a lesser degree, greater concentrations of Co, Cu, Fe, and K, compared with terrestrial plants (Ohlson and Staaland, 2001; Staaland and White, 2001). We calculated the proportion of GPS points for each moose that fell within wetlands and riparian habitats using the Freshwater Atlas (FA) (DataBC, 2024). We defined riparian habitat as a 50 m-buffer around FA lakes, rivers, and streams (Koetke et al., 2023). Both natural and human disturbances shape forest structure and composition. Historically, wildfire has been a dominant disturbance type leading to the development of early seral conditions that increase the quality and quantity of preferred moose food types (e.g., young trees and shrubs; Maier et al., 2005; Lord and Kielland, 2015; Joly et al., 2017).Wildfires also create a pulse of plant-available nutrients in the soil which can be taken up by regenerating vegetation (Kelsall et al., 1977; Simard et al., 2001). Therefore, we used the BC Wildfire Service Historical Fire Perimeters data to calculate the proportion of moose GPS collar points in wildfire burns ≤10 years old (DataBC, 2024). We focused on burns ≤10 years old as they provide highquality forage for moose in the early seral stages of plant succession (Lord and Kielland, 2015). Additionally, most wildfires in our study areas occurred within the past decade (Mumma et al., 2024), and few moose were located near older burns, limiting our ability to assess moose use of older post-fire habitats. Due to a large number of zeros in the data, we treated fire as a factor, categorizing moose with a proportion of GPS-points > 0.1 as having access to burns and those with < 0.1 as not having access to burns. Forest harvesting is the primary resource-based land use in central BC. Forest harvesting, like wildfire, promotes the growth of early successional vegetation that serves as forage for 20 moose (Fisher and Wilkinson, 2005). However, the removal of forest canopy alters the conditions (e.g., sunlight, water, soil nutrients) that influence the growth and composition of regenerating vegetation (Hjeljord et al., 1990; Wurtz and Zasada, 2001). These shifts in growing conditions can also affect plant chemical composition, such as the production of plant secondary metabolites, which may reduce the digestibility and bioavailability of key nutrients for herbivores (Roberge, 2023). Therefore, we calculated the average age of forest stands across GPS points using Harvested Areas of BC (Consolidated Cutblocks) data (DataBC, 2024). In instances where stand age data were missing (e.g., non-harvested areas), we used the VRI data to estimate stand age. Climate change is altering weather patterns, including temperature variability and the amount of precipitation, which can affect plant growth, phenology, and forage quality (Rustad et al., 2001; Post et al., 2008; Park et al., 2020; Brown et al., 2022). These changes in climate also drive the decomposition of soil parent materials and the subsequent release of nutrients that are taken up by plants (Kabata-Pendias, 2010). Thus, we calculated seasonal averages of the mean monthly temperature and total monthly precipitation (June–October) across GPS collar locations for each moose using monthly climate data from ClimateBC version 7.42 (Wang et al., 2016). 21 Table 2.1. Variables considered to explain patterns of essential mineral concentrations in the hair (n = 60) of adult female moose (Alces alces) in central British Columbia, Canada, from 2020 to 2022. Explanatory variable category All Habitat Composition Landscape Disturbance Explanatory variable Description Data Source Study area Year Deciduous forest Categorical; PGS or BP Categorical; 2020, 2021, or 2022 Proportion of GPS points in deciduous leading forest stands Observation Observation VRI Mixed forest Proportion of GPS points in mixed forest stands VRI Wetland Riparian Proportion of GPS points within wetlands Proportion of GPS points within riparian habitat, defined as a 50 m buffer around lakes, rivers, and streams Factor; 1 = “burns” (proportion of GPS points in wildfire burns ≤10 years old is > 0.1), 0 = “no burns” (proportion of GPS points in wildfire burns ≤10 years old is < 0.1) Average forest age class across GPS points Freshwater Atlas Freshwater Atlas Wildfire Forest stand age Climatic Conditions Temperature Average mean monthly temperature across GPS points (°C, Jun–Oct) Precipitation Average total monthly precipitation across GPS points (mm, Jun–Oct) Abbreviations: VRI, Vegetation Resources Inventory 22 Historical Fire Perimeters Consolidated Cutblocks, VRI ClimateBC ClimateBC Statistical analysis We calculated descriptive statistics (i.e., mean, standard deviation, median, and range) for the 15 minerals across study areas, years, and individuals (i.e., including repeated measures). For our environmental analysis, we focused on minerals considered essential for metabolism (Underwood, 2012) and excluded As, Cd, and Pb from further consideration. We also excluded Cr, Co, and Na from further analysis as their concentrations fell below the LOQ in more than 85% of hair samples. We used pair-plots (Pearson r ≥ 0.70) to test for collinearity between minerals (Zuur et al., 2010). Most minerals were weakly correlated (Pearson r ≤ 0.70), except for Ca and Mg, which were highly correlated (r = 0.90); thus, we removed Ca to avoid measuring the same signal unnecessarily. We used Cleveland dot-plots to evaluate potential outliers for each mineral (Zuur et al., 2010) and removed one extreme outlier from Fe that was nearly 18 times higher than the median value. The cause of this extreme value is unknown; however, it could be due to external contamination from another substance that was not fully removed during the laboratory wash procedure. We developed a set of 17 candidate models to investigate the variance observed in hair minerals according to habitat composition, landscape disturbance, and climatic conditions (Table 2.2). These models were developed a-priori to reflect plausible relationships between minerals and environmental characteristics based on the literature. We applied the same set of models to each mineral individually. Given that mineral values were always positive and typically displayed a positively skewed distribution, we fit generalized linear mixed effects models using a γ distribution and log link (package glmmTMB; Brooks et al., 2017). We included a random effect of individuals to account for heterogeneity and potential non-independence of repeated measures. We included study area and year as fixed effects in all models to explain potential 23 spatial and temporal variation in mineral concentrations not explained by our environmental predictors. We centered temperature and precipitation within each study area and year, which allowed us to assess relative deviations from local climatic conditions. By contrast, including year and study area in our models accounted for broader temporal and geographical differences in climate. We tested for multicollinearity between continuous explanatory variables using variance inflation factors (VIF; Zuur et al., 2010). All variables were weakly correlated (VIF < 3), except for coniferous forest with both broadleaf deciduous forest and mixed forest (VIF > 3), so we removed coniferous forest from our models. We standardized continuous predictor variables by subtracting the mean from the observed values and dividing by the standard deviation. We compared our a-priori models and a null model using Akaike’s Information Criterion corrected for small sample sizes (AICc) and model weights (Burnham and Anderson, 2002). We considered models with ΔAICc < 2 to be equally likely hypotheses (Burnham and Anderson, 2002), provided they did not include uninformative parameters (i.e., those that include one extra parameter without meaningfully improving the model's log-likelihood but are ranked close to more parsimonious models with lower AIC values; Leroux, 2019). We interpreted parameters within models to be influential if their 85% confidence intervals (CIs) did not overlap zero. We used 85% CIs as this confidence level reflects the significance threshold consistent with the decision-making framework of AIC-based model selection (Sutherland et al., 2023). We evaluated model fit of our top models by visualizing scaled residuals simulated from the fitted model to assess uniformity, dispersion, and overall model assumptions using the DHARMa package (Hartig, 2024). All statistical analyses were performed using R (version 4.3.1; R Core Team, 2023). 24 Table 2.2. A-priori mixed effects models used to explain concentrations of minerals in hair (n = 60) collected from adult female moose (Alces alces) during the winters of 2020–2022 in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model. Model category Habitat composition Landscape disturbance Climatic conditions Habitat composition and landscape disturbance Habitat composition and climatic conditions Landscape disturbance and climatic conditions Global model Null model Independent variables Study area + Year + Deciduous forest + Mixed forest Study area + Year + Wetland + Riparian Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian Study area + Year + Wildfire Study area + Year + Forest stand age Study area + Year + Wildfire + Forest stand age Study area + Year + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Temperature + Precipitation Study area + Year + Wetland + Riparian + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Temperature + Precipitation Study area + Year + Wildfire + Temperature + Precipitation Study area + Year + Forest stand age + Temperature + Precipitation Study area + Year + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year 25 Results In total, we analyzed mineral concentrations in 60 hair samples (nBP = 31, nPGS = 29) collected between 2020 and 2022 in the two study areas. These samples corresponded to 31 individuals, of which eight had only one measurement, 17 had two measurements in two years, and six had three measurements in three years. We found detectable concentrations (i.e., above the LOD) of all 15 minerals in hair from female moose (Table 2.3). Concentrations in hair samples fell below the LOQ for the macro minerals K (1.7% samples) and Na (98.3% samples), and for the trace minerals Co (86.7% samples), Cr (91.7% samples), Mo (5% samples), and Se (1.7% samples). For non-essential minerals, the concentrations fell below the LOQ for each of the heavy metals tested: As (88.3% samples), Cd (96.7% samples), and Pb (5% samples). For As, 11.7% of samples (n = 7) fell below the LOD. Table 2.3. Mean, standard deviation (SD), median, and range mineral concentrations (µg/g, dry weight) in hair (n = 60) from female moose (Alces alces) collected in the winters of 2020–2022 from populations in Bonaparte (BP; n = 31) and Prince George South (PGS; n = 29). Mineral Mean ± SD Median Range # 0.10 threshold). One possible explanation is high site fidelity, as moose may remain in established home ranges that have not recently burned, even if nearby burned areas offer improved habitat conditions. Alternatively, moose might avoid recently burned areas due to increased risks or costs associated with open habitats, such as greater thermal stress and predation risk, especially during warm weather and parturition (Bowyer et al., 1999; Melin et al., 2014). Previous studies suggest that moose show the strongest selection for burned habitats at intermediate successional stages (~11– 30 years old; Maier et al., 2005; Joly et al., 2017), likely because these areas provide a balance of high forage availability and sufficient cover. Since most burns in our study areas were less than 10 years old, we could not assess whether intermediate or older burns were associated with variation in mineral concentrations. However, recent work by Mumma et al. (2024) in this system suggests that the use of burns by moose will increase as vegetation regrows and structural 37 cover improves. Hence, our findings provide early evidence that younger burns may offer a short-term improvement in forage quality, while highlighting the need for long-term monitoring to determine whether nutritional gains persist or shift as burns mature. Although mineral deficiencies are more commonly documented in wild ungulate populations, excessive intake of some minerals can also lead to toxicity or adverse health effects. For example, the macro minerals K and Mg are essential for muscle and nerve function, but excessive dietary levels can interfere with rumen health and lead to digestive or metabolic dysfunction (Puls, 1988; McDowell, 1992). Moreover, trace minerals such as Se, though critical for immune function and antioxidant defense, can lead to hair loss, hoof deformities, and reproductive impairment when chronically elevated (Puls, 1988; Flueck et al., 2012; Raisbeck, 2020). Moose must therefore navigate a narrow nutritional window—ensuring they meet mineral requirements without exceeding tolerable limits. This delicate nutritional balancing act is further complicated by interactions among minerals that can affect their absorption and utilization within the body (Underwood, 2012). Although our study identifies environmental correlates of minerals as reflected in hair, we acknowledge that higher hair concentrations do not necessarily equate to better health. Further understanding of the mechanisms moose use to regulate mineral intake to avoid deficiency and excess is essential for fully interpreting these patterns and informing effective management strategies. Another important consideration to our study is that mineral imbalances in wild ungulates are not always due to insufficient mineral concentrations in forage alone but may also arise from digestive challenges associated with seasonal dietary shifts. When moose switch from highly fibrous winter diets to lush, low-fiber spring vegetation, their ability to absorb or retain minerals may be compromised, even if mineral levels in forage are adequate (Ayotte et al., 2006). Moose 38 often seek mineral licks, especially in spring and summer (Tankersley and Gasaway, 1983; Rea et al., 2013; Huxter et al., 2024), where elevated levels of essential minerals such as Ca, Fe, Mg (Rea et al., 2013), as well as carbonates (Ayotte et al., 2006), are available. Nutritional supplementation from these licks can support rumen function, improve nutrient absorption, and enhance overall body condition (Kreulen, 1985). Moose may also use mineral licks to counteract other aspects of forage quality such as the increasing presence of plant defensive compounds during the summer (Ayotte et al., 2008). Due to a lack of spatial data on mineral lick locations within our study areas, we were unable to evaluate their role directly; however, this represents a valuable direction for future research. Nonetheless, because hair integrates mineral intake over an extended period, we expect that forage remains a major contributor to mineral acquisition in moose, and thus an important influence on hair mineral concentrations, although the relative importance of forage versus other sources such as mineral licks likely varies among individual minerals, individual animals, and forage nutrient concentrations. In summary, by integrating hair mineral concentrations with measures of habitat composition, landscape disturbance, and climatic conditions, we offer novel insights into patterns of mineral variation in moose in response to environmental change. The tight linkage between mineral status and population stability in herbivores is well documented (O’Hara et al., 2001; Flueck et al., 2012), as many minerals play a crucial role in maintaining health and supporting reproduction and survival (Kincaid, 2000; Underwood, 2012). Therefore, understanding the factors associated with mineral concentrations in moose is necessary for sound population management. Moreover, our findings provide a baseline to support the monitoring of mineral status in hair and underscore the importance of continued monitoring of minerals as a tool for assessing moose responses to environmental change. Minerals are only one of many nutritional 39 currencies essential for the health and fitness of moose, and these results underscore the importance of considering multiple environmental and physiological factors when studying wildlife nutrition. Ultimately, our preliminary study could pave the way for novel research on the causes and consequences of mineral variability in moose and other large herbivores in landscapes subject to environmental change. 40 CHAPTER 3: Immune biomarkers vary in relation to body fat, trace mineral status, and parasite exposure in moose Introduction Wildlife today face widespread changes in habitat and climate that can affect their ability to contend with parasitic infections. Parasites, including microparasites (e.g., viruses, bacteria) and macroparasites (e.g., helminths, ticks), can compromise the health of individuals and persistence of wildlife populations, shaping a range of life-history traits in their hosts (Gulland, 1995). Central to host–parasite dynamics is the immune system, which acts as the primary line of defense against the establishment, replication, and spread of infectious agents (Schmid-Hempel, 2021). Although parasites are ubiquitous in natural systems, hosts differ markedly in their ability to mount effective immune responses. This variation is often influenced by environmental conditions, which can suppress, modify, or enhance immune function through factors such as temperature fluctuations, pollution, and limited food availability (Acevedo-Whitehouse and Duffus, 2009; Jolles et al., 2014; Becker et al., 2019; Ohmer et al., 2021). Under such constraints, individuals may face trade-offs in the allocation of resources to immune defenses versus other physiological demands, such as reproduction (Sheldon and Verhulst, 1996; Lochmiller and Deerenberg, 2000). Indeed, a central concept in eco-immunology is that immunity is costly to maintain and deploy (Lochmiller and Deerenberg, 2000). Given a finite amount of nutritional and energetic resources, individuals must balance investments across competing physiological demands such as immunity, growth, maintenance, and reproduction (Stearns, 1989; Sheldon and Verhulst, 1996), and these trade-offs have been observed across diverse taxa (e.g., Ilmonen et al., 2000; Ruiz et al., 2011; Hayward et al., 2019; Encel et al., 2023). For example, female Soay sheep 41 (Ovis aries) with higher antibody responsiveness showed improved survival during harsh winters but also exhibited reduced reproductive output (Graham et al., 2010). However, such trade-offs are not universal; they may vary depending on the type of immune response (Albery et al., 2019; Ruoss et al., 2019) and can be influenced by social and ecological contexts (French et al., 2007; Wallace et al., 2023). For instance, when resources are abundant, hosts may afford simultaneous investment in immune defense and other energetically costly traits (van Noordwijk and de Jong, 1986; Becker et al., 2019). Understanding the context-dependent nature of these trade-offs is essential to understanding how fluctuating environmental conditions relate to wildlife health. Nutritional condition plays a central role in mediating physiological trade-offs and shapes an individual's ability to resist and tolerate infectious agents (Downs and Stewart, 2014; Ohmer et al., 2021). In general, individuals in good nutritional condition are expected to invest in immune responses that balance effective parasite control with minimal self-damage (Downs et al., 2014; Downs and Stewart, 2014), whereas those in poor condition often exhibit reduced immunocompetence and increased susceptibility to disease (Beldomenico and Begon, 2010). Experimental studies have demonstrated that food restriction can suppress multiple components of immunity, such as cellular responses to gastrointestinal parasites and antibody production (Valderrábano et al., 2006; French et al., 2007; Martin et al., 2007; Bourgeon et al., 2010). Yet, in natural populations, these relationships are often more complex. For example, in roe deer (Capreolus capreolus), immune responses shift with body condition, but the direction and magnitude of change vary depending on the specific immune biomarker examined (GilotFromont et al., 2012). Therefore, individuals may adjust their investment in immune strategies depending on their nutritional condition. A more nuanced understanding of these dynamics in 42 wild animals is essential for clarifying how nutritional condition relates to immunocompetence under natural ecological pressures. Limited macronutrient (e.g., energy, protein) and micronutrient (e.g., vitamins, minerals) intake can influence immune responses and underlie trade-offs across physiological systems (French et al., 2007; Brunner et al., 2014; Downs et al., 2018; Bariod et al., 2024). Among micronutrients, trace minerals such as copper (Cu), selenium (Se), and zinc (Zn) play particularly important roles in regulating immune responses and may influence immunity independently of energetic constraints (Jolles et al., 2014). These minerals act as essential cofactors in enzymatic pathways that support immune cell activation, proliferation, and differentiation, and contribute to reactive oxygen species generation, oxidative stress modulation, and regulation of inflammation and immune resolution (Kincaid, 2000; Underwood, 2012; Paul and Dey, 2015). As such, deficiencies or imbalances in trace minerals can impair immune system function, leaving individuals more vulnerable to infection and influencing the trajectory of disease outcomes (Chandra, 1996; Kincaid, 2000). Although these relationships are well-documented in domestic animals (McClure, 2003; McClure, 2008), they remain relatively underexplored in wild populations. Furthermore, the physiological mechanisms underlying these relationships in wildlife are often not fully investigated or understood. Measuring immunological biomarkers in wildlife enables an understanding of immune function and its connection to physiological condition, fitness, and parasite exposure (Ohmer et al., 2021). The immune system is a complex and multifaceted defense network, and interpreting variation in immune function requires assessing multiple immune indices that capture different components of this system. Cytokines are signaling proteins secreted by various immune cells that regulate immune responses by either enhancing or suppressing activity, and include, for 43 example, interferons (INF), interleukins (IL), interferon gamma-induced protein (IP), and tumor necrosis factors (TNF) (Graham et al., 2007; Zimmerman et al., 2014). Cytokines are multifunctional molecules that can promote pro- or anti-inflammatory responses and modulate specific subsets of immune cells in both innate and adaptive immunity. Acute phase proteins (APP), such as haptoglobin, are synthesized in the liver in response to the production inflammatory cytokines (Libera et al., 2022). Haptoglobin scavenges free hemoglobin, inhibits microbial iron uptake, and reduces oxidative damage (Peck et al., 2016). Production of haptoglobin is typically upregulated during the course of both acute and chronic bacterial or viral infections, reflecting increased erythrocyte turnover linked to inflammatory processes (Gruyse et al., 2005; Peck et al., 2016; Libera et al., 2022). Total globulins, encompassing alpha, beta, and gamma proteins, represent a broad group of serum proteins critical to immune defense, including antibodies, enzymes, and carrier proteins (Alberghina et al., 2011; Couch et al., 2017). Globulins contribute to innate and adaptive immune processes, including humoral responses, while also playing roles in inflammation and molecule transport, making their levels useful indicators of immune activation or ongoing infection. Therefore, measuring concentrations of a panel of cytokines alongside haptoglobin and total globulins provides broad coverage of the immune system, offering complementary insights into multiple aspects of immune function and enabling a more integrated assessment of immune status in wild animals. Moose (Alces alces) in central British Columbia (BC), Canada, are well suited for studying immune variation within a dynamic and changing ecological context. This region has faced extensive pine tree mortality caused by a severe mountain pine beetle (Dendroctonus ponderosae) outbreak, which led to large-scale salvage logging operations (Alfaro et al., 2015). Concurrent with these landscape changes, some moose populations experienced steep declines of 44 up to 70% (Kuzyk et al., 2018). Early findings from long-term research and monitoring suggest that bottom-up factors may be contributing to fluctuations in the viability of populations, as indicated by observed starvation and health-related mortalities, coupled with suboptimal pregnancy rates (Thacker et al., 2019). Furthermore, moose face exposure to multiple pathogens reported to cause mortality and reproductive failure in other wild ungulates (das Neves et al., 2010; Thacker et al., 2019). To our knowledge, no previous studies have comprehensively quantified a broad range of immune biomarkers in moose reflecting different components of immunity nor examined patterns of variation in these responses. Measuring immune biomarkers in moose offers the opportunity to assess associations with parasitic infections and physiological condition and ultimately informs conservation efforts aimed at improving wildlife health. The aim of the present study was to explore how individual physiology and immune challenges are associated with immune function in wild moose. To this end, we analyzed serum samples from adult female moose collected during the winters of 2020–2022 from two study areas in central British Columbia. We quantified the concentrations of haptoglobin, total globulins, and 13 cytokines, which represent the first such cytokine data for moose. We then examined correlations between immune responses and pregnancy status, body fat, key trace minerals, and exposure to micro- and macro-parasites. We expected that immune responses might differ between pregnant and non-pregnant individuals, as pregnancy is an energetically demanding stage. We also expected that individuals with greater body fat could show variation in immune responses, reflecting differences in energetic reserves. Because trace minerals serve as essential cofactors in many immune processes, we anticipated that greater mineral concentrations could be associated with differences in immune biomarker levels. Finally, we predicted that parasite exposure would be associated with increased immune biomarker levels due to immune 45 activation. We sought to provide a foundational understanding of the dynamic immune responses in free-ranging moose and lay groundwork for monitoring health and physiological adaptations in wildlife facing environmental change. Methods Ethics statement The capture and handling of adult female moose was undertaken by the Province of British Columbia as part of a large-scale, provincial-wide moose research project. All procedures adhered to the British Columbia Wildlife Act (permit CB17-277227). Data analysis for this project was approved by the Animal Care and Use Committee at the University of Northern British Columbia (ACUC Protocol Number 2021-01). Study system We carried out our study on female moose from two populations in central British Columbia, Canada, located in the Bonaparte Plateau (BP) and Prince George South (PGS) regions. These populations have been studied as part of a long-term monitoring effort. The BP (6,800 km2) is located north of Kamloops, BC (51°13′ N, 120°81′ W), on the traditional territory of the Secwépemc First Nation. The PGS area (11,000 km²) is situated southwest of Prince George, BC (53°56′ N, 123°63′ W), on the traditional territories of the Lheidli T’enneh and Saik’uz First Nations. Both regions experience a humid continental climate with short, warm summers and long, cold winters. Average annual temperatures are higher in BP (9.5°C) than in PGS (4.3°C), with BP receiving less precipitation overall (216 mm rain, 63 cm snow) compared to PGS (432 mm rain, 204 cm snow; Environment and Climate Change Canada, 2024). The BP landscape spans three Biogeoclimatic Ecosystem Classification (BEC) zones—Interior Douglas- 46 fir (IDF), Sub-Boreal Pine–Spruce (SBPS), and Montane Spruce (MS)—whereas PGS is dominated by the Sub-Boreal Spruce (SBS) zone (Meidinger and Pojar, 1991). Both areas include mixed coniferous and deciduous forests at varying successional stages, alongside lakes and wetlands. Historically shaped by wildfire, these ecosystems are now primarily influenced by industrial forestry and have undergone widespread salvage logging following mountain pine beetle outbreaks (Alfaro et al., 2015). Alongside moose, local wildlife includes mule deer (Odocoileus hemionus), white-tailed deer (O. virginianus), elk (Cervus canadensis), and predators such as wolves (Canis lupus), black bears (Ursus americanus), grizzly bears (U. arctos), and cougar (Puma concolor). Moose densities were estimated at 254 ± 41/1,000 km² in BP (2017–2018) and 400 ± 78/1,000 km² in PGS (2016–2017; Kuzyk et al., 2018). Sample collection To identify how physiological variables and immune challenges may relate to variation in immune biomarker levels, we analyzed serum samples from adult female moose (≥1.5 years old, n = 31) captured during early winter (December–February) from 2020 to 2022. For simplicity, individuals captured between December and February were grouped into a single study year (e.g., moose captured between December 2019 and February 2020 were classified as part of the 2020 study year). Comprehensive descriptions of animal capture, sampling, and monitoring have been previously described (Kuzyk et al., 2018, Procter et al., 2020). In brief, adult females were captured using either aerial net-gunning with physical restraint or chemical immobilization via aerial darting from a helicopter. Each moose underwent a comprehensive physical examination and health assessment performed by the attending wildlife veterinarian or an experienced wildlife biologist. Blood samples (20–35 mL) were drawn from the jugular vein using an 18gauge, 1.5-inch needle, and placed into 6.0 mL royal blue top collection tubes (for trace mineral 47 testing) and 5.0 mL serum separator tubes (gold-top SST; for all other serological testing). Within 12 hours, samples were centrifuged for 15 minutes, and the resulting serum was decanted into cryovials and frozen at −20°C until laboratory analysis. Recaptures were conducted in successive years when possible to facilitate longitudinal monitoring of individual health status (Thacker et al., 2019). Immune biomarkers We submitted frozen serum samples to Eve Technologies Corporation, Calgary, Alberta, for the multiplexed quantification of 13 cytokines using Luminex xMAP technology. Due to the absence of cross-reactivity data for moose, we initially submitted three pilot samples for testing using commercial assay kits developed for bovine, ovine, and porcine species. Based on these pilot samples, we selected the porcine panel for analysis of the remaining samples (n = 57), as it produced the highest detection rates across individual cytokines and yielded the greatest number of values within the assay’s dynamic range. Cytokine concentrations were quantified using Eve Technologies' Porcine Cytokine 13-Plex Discovery Assay® (MilliporeSigma, Burlington, Massachusetts, USA) on the Luminex™ 200 system (Luminex, Austin, Texas, USA), following the manufacturer's protocol. The assay measures fluorescence intensity emitted by fluorescently labeled beads bound to cytokines, and cytokine concentrations are derived by comparing fluorescence signals to a standard curve created from known concentrations. The 13-plex assay measured granulocyte-macrophage colony-stimulating factor (GM-CSF), interferon gamma (IFNγ), interleukin-1 alpha (IL-1α), IL-1β, IL-1 receptor antagonist (IL-1ra), IL-2, IL-4, IL-6, IL-8, IL-10, IL-12, IL-18, and tumor necrosis factor alpha (TNF-α). Each sample was analyzed in duplicate, and the average of the two replicate values was used in the final analysis, reported in pg/mL. Assay sensitivities of the biomarkers range from 5–42 pg/mL. 48 Serum was also submitted to the Animal Health Laboratory, University of Guelph, for the quantification of globulin (Glb) and haptoglobin (Hp). To calculate globulin concentrations (g/L), total protein and albumin were first measured directly using colorimetric assays on the Roche Cobas 6000 c501 biochemistry analyzer (Roche Diagnostics, Indianapolis, Indiana, USA); globulin was then derived by subtracting albumin from total protein. Haptoglobin (g/L) concentrations were determined using a photometric method on the same analyzer, following protocols based on the methods of Makimura and Suzuki (1982) and Skinner et al. (1991). These methods rely on the peroxidase activity of the hemoglobin–haptoglobin complex at low pH, which is measured spectrophotometrically. Physiological and parasitic variables Reproduction requires substantial energy in mammals and, as a result, pregnant individuals may have reduced capacity to invest in other costly physiological processes such as immune function. To explore how immunity may relate to pregnancy, we submitted serum samples to the Herd Health Diagnostics Center (Pullman, Washington, USA) for pregnancy testing using a BioPRYN enzyme-linked immunosorbent assay (ELISA). Adult females were classified as pregnant if serum concentrations of pregnancy-specific protein B (PSPB) exceeded 0.21 mg/mL. Body fat serves as the primary energetic reserve in ungulates, fluctuating seasonally and in response to environmental conditions, making it a key indicator of nutritional status. Moreover, winter body fat in our study system has been shown to reflect whether or not a female successfully raised a calf in the previous year, and therefore provides insight into both recent reproductive investment and current energetic condition (Jefferies, 2024). Therefore, to assess how immunity may vary in relation to energy, we estimated body fat using maximum rump fat 49 thickness (MAXFAT) measured at capture via ultrasonography (FUJIFILM Sonosite M-Turbo®, Toronto, ON, or Ibex® Pro, Loveland, CO), and converted values to ingesta-free body fat percentage (IFBFAT; hereafter, body fat) using the equation: IFBFAT = 5.61 + 2.05 × MAXFAT (Stephenson et al., 1998). Trace minerals are essential micronutrients that play critical roles in maintaining immune function, supporting antioxidant defenses, and regulating inflammatory responses (Kincaid, 2000; Underwood, 2012; Paul and Dey, 2015). To assess how variation in mineral status may be associated with variation immune responses, we submitted serum samples collected in royal blue top tubes to the Animal Health Laboratory at the University of Guelph. Concentrations of seven minerals were quantified using inductively coupled plasma mass spectrometry (ICP-MS). For this study, we focused on copper (Cu), selenium (Se), and zinc (Zn), as these elements have wellestablished roles in immune regulation, oxidative stress response, and inflammation in model organisms (Underwood, 2012). To evaluate immune responses in relation to parasitic challenge, we selected one macroparasite and one microparasite based on their prevalence in our study population. As part of the broader provincial moose health monitoring program, serum samples were screened for exposure to a wide array of infectious agents; however, most pathogens exhibited very low seroprevalence (≤10%) and were excluded from analysis. Full diagnostic protocols and prevalence estimates for all screened pathogens are reported in Thacker et al. (2019). For this study, we retained two parasites with sufficient prevalence to support statistical modeling: winter tick (Dermacentor albipictus) infestation and an alphaherpesvirus. Winter tick is a single-host ectoparasite that can cause substantial energetic costs through blood loss, skin damage, and thermoregulatory stress, especially in late winter (Samuel, 2004). Winter tick 50 infestation was assessed at capture at two body sites: the upper shoulder and the rump. At each site, ticks were counted along four parallel 10 cm transect lines spaced 2 cm apart. The total number of ticks across all transects at both sites was summed to provide an overall infestation score for each individual. Due to several zero counts in the data, we treated winter tick infestation as a binary variable, classifying individuals as tick-positive if ticks were observed at either site and tick-negative if no ticks were present. Alphaherpesviruses are a group of viruses that can establish latent infections and reactivate under stress or immunosuppression. In domestic livestock, bovine herpesvirus-1 (BoHV-1), which causes infectious bovine rhinotracheitis, is a well-known alphaherpesvirus linked to respiratory, reproductive, and systemic illnesses (Muylkens et al., 2007). We assessed seropositivity to an alphaherpesvirus, which has not yet been identified in moose, using an ELISA at the BC Animal Health Centre in Abbotsford, British Columbia. This assay detects antibodies specific to bovine herpesvirus-1. Seropositivity was treated as a binary variable indicating prior exposure or infection. Statistical analysis We computed descriptive statistics (i.e., mean, standard deviation, median, and range) for the observed concentrations of 13 cytokines, total globulin, and haptoglobin, across study areas, years, and individuals (i.e., including repeated measures). All cytokines exhibited some out-ofrange (OOR) concentrations below the assay’s detection limit, although the extent of OOR values varied among biomarkers (see results). Low or undetectable cytokine levels still represent biologically meaningful variation in immune expression; therefore, we assigned concentrations equal to half the lowest observed concentration for each biomarker to OOR samples and retained them in analyses (Balle et al., 2020). Concentrations of cytokines that were considered OOR 51 were, however, excluded from descriptive statistics. In addition to concentrations of biomarkers, we reported descriptive statistics for raw fluorescence intensity values for each cytokine analyte on the full dataset. Fluorescence intensity is a more reliable indicator of sample-to-sample variation, particularly in samples with low concentrations of an analyte, as values do not require defining a limit of detection (Breen et al., 2016). By contrast, concentrations are determined from raw fluorescence values through comparisons with a standard curve and can therefore only be quantified for values that fall within the linear range of the standard curve. We selected immune biomarkers for further analysis by first examining patterns of covariation using pairwise Spearman’s rank correlations. Cytokines GM-CSF, IFN‐γ, IL-8, and TNF-α were excluded from pairwise correlations and subsequent analyses due to a high proportion of samples falling below the minimum detectable concentration. The remaining nine cytokines were all highly correlated (all ρ > 0.93; see results). Based on these results, we selected four cytokines for further analyses that are known to have distinct, non-redundant biological roles in an attempt to capture meaningful variation in immune function: IL-1β, a proinflammatory cytokine important in innate immune activation; IL-10, an anti-inflammatory cytokine that regulates immune suppression and resolution; IL-12, which promotes Th1-type cellular immunity; and IL-4, which supports Th2-type humoral responses (Graham et al., 2007; Zimmerman et al., 2014). Total globulin and haptoglobin were also retained, as they were weakly correlated with cytokines (ρ ≤ 0.20) and represent different axes of immune function. We used generalized linear mixed effects models (GLMMs) implemented in the glmmTMB package (Brooks et al., 2017) to identify variables associated with the six selected immune biomarkers. Given that cytokines (i.e., IL-1β, IL-12, IL-10, IL-4) and haptoglobin were restricted to positive values and their distribution is right skewed, we assumed a Gamma 52 distribution of the residuals and used a logarithmic link function. In contrast, globulin concentrations were approximately normally distributed and therefore modeled using a Gaussian distribution with an identity link. Prior to fitting models, we used Cleveland-plots to evaluate outliers for each biomarker (Zuur et al., 2010) and removed two severe outliers from haptoglobin that were approximately 9 and 12 times higher than the median value, respectively. These values (1.59 and 2.08 g/L) were considered outliers as they were substantially higher than the next highest reported haptoglobin level of 0.68 g/L in our dataset and exceeded levels previously documented in another moose population and other cervid species (McDonough et al., 2022; Lamb et al., 2024). Model selection was conducted using a model dredging approach via the dredge function in the MuMIn package (Bartoń, 2023), which fits all possible subsets of a global model and ranks them based on Akaike’s Information Criterion corrected for small sample size (AICc; Burnham and Anderson, 2002). The global models included the following fixed effects to reflect biologically plausible relationships: pregnancy status, body fat, serum concentrations of Cu, Se, and Zn, winter tick infestation, and alphaherpesvirus serostatus. Study area and year were retained as conditional fixed effects in all models to control for spatial and temporal variation, allowing us to isolate the effects of our variables of interest. Individual ID of female moose was included as a random intercept to account for repeated sampling. We assessed multicollinearity among explanatory variables within our global models using variance inflation factors (VIFs; Zuur et al., 2010) and found all variables to be weakly correlated with VIFs less than two. Continuous predictors were standardized by centering on the mean and scaling by the standard deviation prior to analysis. 53 We considered models with ΔAICc < 2 to be equally supported (Burnham and Anderson, 2002), provided they did not include uninformative parameters (i.e., those that include one extra parameter without meaningfully improving the model's log-likelihood but are ranked close to more parsimonious models with lower AIC values; Leroux, 2019). If the set of similarly supported models contained the null model (study area and year only), we considered that the most parsimonious model. We interpreted parameters within models to be influential if their 85% confidence intervals (CIs) did not overlap zero. We used 85% CIs as this confidence level reflects the significance threshold consistent with the decision-making framework of AIC-based model selection (Sutherland et al., 2023). We evaluated model fit of our top models by visualizing scaled residuals simulated from the fitted model to assess uniformity, dispersion, and overall model assumptions using the DHARMa package (Hartig, 2024). All statistical analyses were conducted using R (version 4.3.1; R Core Team, 2023). Given the exploratory nature of our analysis and the number of model comparisons conducted, the reported covariate effects should be interpreted as preliminary hypotheses requiring validation with independent data. Results We sampled 31 unique adult female moose during winter between 2020 and 2022 in two study areas in central British Columbia: the Bonaparte Plateau (BP; n = 17) and Prince George South (PGS; n = 14). Of these individuals, eight were sampled once, 17 were sampled twice in two different years, and six were sampled three times in three years, resulting in a total of 60 serum samples in which immune biomarkers were quantified. We detected measurable concentrations of all tested immune biomarkers in moose serum (Table 3.1). Of the 13 cytokines measured using the porcine cytokine multiplex test, the proportion of samples within the assay’s sensitivity range varied from 23.3% for GM-CSF to 96.7% for both IL-1β and IL-4. Cytokines 54 GM-CSF, IFN-γ, IL-8, and TNF- α fell below the minimum detectable concentration limit in 77%, 30%, 57%, and 53% of samples, respectively, and were removed from analyses. All other immune biomarkers were detectable in at least 80% of samples. No samples exceeded the assay’s upper sensitivity limit. All immune biomarkers, except total globulins, showed moderate to strong right-skewed distributions, indicating that most individuals had low concentrations and a few had markedly higher levels. Summary statistics for key predictor variables used in our models are presented in Table 3.2. The nine cytokines retained were strongly and positively correlated, with Spearman’s rank coefficients ranging from 0.94 to 0.98 (all p-values <0.001; Appendix B Fig. B.1). In contrast, correlations between cytokines and total globulins were weak but generally positive, ranging from ρ = 0.10 to 0.20 (all p-values > 0.13). Correlations between cytokines and haptoglobin were weak and mostly negative, ranging from ρ = –0.14 to –0.04 (all p-values >0.28). Total globulins and haptoglobin were effectively uncorrelated (ρ = 0.03, p-value = 0.65). 55 Table 3.1. Serum immune biomarker concentrations (n = 60) from adult female moose (Alces alces) sampled in winter (2020–2022) from two populations in central British Columbia, Canada: the Bonaparte Plateau (BP; n = 31) and Prince George South (PGS; n = 29). Cytokine concentrations are reported in pg/mL, and globulin and haptoglobin in g/L. Cytokine concentration values that fell below the assay detection range (out of range; OOR) and could not be extrapolated were excluded from descriptive statistics. Fluorescence intensity values, which do not have a defined limit of detection, are reported for the full sample size. Immune Mean conc. ± SD biomarker Median Conc. range OOR conc. Mean FI (n/60) Median FI range FI GM-CSF 26.07 ± 25.54 17.92 1.85–81.85 46 12.78 ± 4.16 11.50 8.50 – 29.50 IFN-γ 3034.56 ± 5014.00 1340.81 24.84–28771.74 18 84.53 ± 157.92 36.40 13.80 – 1066.00 IL-1α 75.41 ± 141.69 25.42 0.12–705.58 6 264.59 ± 503.95 81.90 9.50 – 2650.00 IL-1β 601.35 ± 1193.03 123.45 1.01–5703.34 2 256.13 ± 469.29 67.15 9.00 – 2419.50 IL-1ra 704.01 ± 1416.55 244.57 4.71–6261.22 10 282.17 ± 507.38 91.15 14.50 – 2619.80 IL-2 557.37 ± 1236.10 142.45 0.57–5744.94 6 259.50 ± 526.49 61.50 11.30 – 2636.30 IL-4 3849.75 ± 10376.39 522.76 2.84–51308.42 2 285.49 ± 554.89 87.75 13.30 – 2860.80 IL-6 252.20 ± 546.26 75.54 0.47–2760.12 12 249.95 ± 481.83 75.40 11.50 – 2690.80 IL-8 67.03 ± 73.05 48.46 4.55–329.72 34 56.37 ± 80.72 27.75 11.50 – 444.30 IL-10 2745.18 ± 6281.27 842.35 4.76–36118.42 10 254.30 ± 519.88 63.75 9.50 – 2896.30 IL-12 363.76 ± 671.91 115.88 0.35–3031.10 11 247.73 ± 470.66 65.90 10.80 – 2620.00 IL-18 3594.05 ± 8397.79 1028.07 4.52–38674.41 9 228.85 ± 462.09 55.75 12.00 – 2332.50 TNFα 34.79 ± 25.51 29.64 6.15–100.81 32 21.84 ± 6.37 20.75 12.00 – 46.80 Globulin 29.63 ± 7.26 28.00 17.00–58.00 0 - - - Haptoglobin 0.25 ± 0.31 0.17 0.12–2.08 0 - - - Abbreviations: conc., concentration, FI, fluorescence intensity, GM-CSF, granulocyte-macrophage colony-stimulating factor; IFN-γ, interferon gamma; IL, interleukin; TNFα, tumor necrosis factor alpha 56 Table 3.2. Summary of physiological and parasitic variables (n = 60) used to predict immune biomarker concentrations in female moose (Alces alces) serum sampled during the winters of 2020–2022 in two populations: the Bonaparte Plateau (BP; n = 31) and Prince George South (PGS; n = 29), central British Columbia, Canada. Continuous variables (body fat percentage and serum trace minerals Cu, Se, and Zn) are presented as mean ± standard deviation. Categorical variables (pregnancy status, winter tick presence, and alphaherpesvirus serostatus) are presented as counts with corresponding percentages in parentheses. These values represent averages and proportions calculated across both study areas, all three sampling years, and include repeated measurements from individuals. Variable Body fat (%) Serum Cu (µg/mL) Serum Se (µg/mL) Serum Zn (µg/mL) Pregnancy status — Pregnant Winter tick — Presence Alphaherpesvirus — Seropositive Type Continuous Continuous Continuous Continuous Categorical Categorical Categorical Summary 9.00 ± 2.00 0.40 ± 0.07 0.05 ± 0.02 0.56 ± 0.08 42 (70%) 43 (72%) 45 (75%) Our model selection approach indicated that body fat, or a combination of body fat and serum Zn concentrations, best predicted IL-12 concentrations in the serum of adult female moose (Fig. 3.1). The two top-ranked models with ΔAICc scores less than two accounted for a combined 24% of the cumulative model weight (Table 3.3). Moose with greater amounts of body fat had greater concentrations of IL-12 (1st ranked model: β = 0.69, SE = 0.24, 85% CI = [0.35, 1.03]; 2nd ranked model: β = 0.59, SE = 0.25, 85% CI = [0.23, 0.95]). Additionally, moose with higher serum Zn concentrations also had elevated IL-12 concentrations (1st ranked model: β = 0.76, SE = 0.33, 85% CI = [0.11, 1.40]). Concentrations of IL-12 were similar between the two study areas (1st ranked model: β = 0.73, SE = 0.98, 85% CI = [-0.69, 2.15]; 2nd ranked model: β = 0.74, SE = 0.96, 85% CI = [-0.65, 2.12]). Moose sampled in 2022 had higher IL-12 concentrations compared to those sampled in 2020 in the first ranked model (β = 1.07, SE = 0.55, 85% CI = [0.29, 1.86]); however, the 85% confidence interval overlapped zero in the second ranked model (β = 0.27, SE = 0.48, 85% CI = [−0.42, 0.96]). Moose sampled in 2021 had similar 57 IL-12 concentrations to those sampled in 2020 in both models (1st ranked model: β = −0.05, SE = 0.54, 85% CI = [−0.83, 0.73]; 2nd ranked model: β = −0.69, SE = 0.54, 85% CI = [−1.46, 0.09]). Figure 3.1. Coefficient estimates and 85% confidence intervals (CI) for independent variables in the top-ranked generalized linear mixed effects models (< 2 ΔAICc) explaining interleukin-12 concentrations in female moose (Alces alces) serum. Serum samples (n = 60) were collected during capture events in the winters of 2020–2022 from two study areas (Prince George South and the Bonaparte Plateau) in central British Columbia, Canada. An independent variable has an influential relationship with the mineral if the CI does not overlap 0. Continuous predictors serum Zn and body fat were standardized to a mean of zero and a standard deviation of one prior to analysis. Coefficients were ordered from most positive to most negative based on standardized estimates. 58 Table 3.3. Model selection statistics used to predict serum immune biomarker concentrations (n = 60 for most biomarkers; haptoglobin: n = 58 due to removal of two outliers) in adult female moose (Alces alces) in two study areas in central British Columbia, Canada, from 2020–2022. Models that were within the top model set (< 2 ΔAICc) and ranked higher than the null model, as well as null models, are included. Variables in bold font indicate an influential relationship with the dependent variable (85% CI does not overlap 0). Predictor variables did not explain substantially more variation in the data than the null model for immune biomarkers IL-1β, IL-10, and IL-4, and therefore, these results are not presented. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as conditional fixed effects in every model. Dependent variable IL-12 Rank Independent variables 1 Study area + Year + Body fat + Serum Zn 2 Study area + Year + Body fat 13 Study area + Year (null) Total 1 Study area + Year + Winter globulins tick + Alphaherpesvirus + Serum Zn 2 Study area + Year + Winter tick + Alphaherpesvirus 3 Study area + Year + Winter tick 18 Study area + Year (null) Haptoglobin 1 Study area + Year + Serum Cu 2 Study area + Year + Serum Cu + Alphaherpesvirus 3 Study area + Year + Serum Cu + Serum Cu 4 Study area + Year + Serum Cu + Winter tick 7 Study area + Year (null) Abbreviations: IL, interleukin 59 AICc ΔAIC AICc wi 674.23 0.00 0.17 676.15 678.83 381.12 1.92 4.59 0.00 0.07 0.02 0.11 382.11 0.99 0.07 382.73 1.61 0.05 385.09 -172.21 -171.05 3.97 0.00 1.17 0.02 0.10 0.07 -170.87 1.34 0.05 -170.45 1.76 0.04 -169.73 2.45 0.03 Among the models evaluated to predict total globulin concentrations, three had ΔAICc scores below two and collectively accounted for 23% of the cumulative model weight (Fig. 3.2, Table 3.3). Winter tick presence consistently emerged as an important predictor of total globulins; female moose infested with winter ticks had elevated concentrations of total globulins (1st ranked model: β = 3.31, SE = 1.41, 85% CI = [1.29, 5.34]; 2nd ranked model: β = 3.20, SE = 1.44, 85% CI = [1.13, 5.26]; 3rd ranked model: β = 3.27, SE = 1.43, 85% CI = [1.21, 5.33]). Additionally, total globulin concentrations were greater in moose seropositive for alphaherpesvirus (1st ranked model: β = 3.89, SE = 1.78, 85% CI = [1.33, 6.46]; 2nd ranked model: β = 3.23, SE = 1.80, 85% CI = [0.63, 5.82]). Total globulin concentrations decreased with higher serum Zn levels (1st ranked model: β = -1.53, SE = 0.78, 85% CI = [-2.65, -0.41]). Total globulin concentrations were greater in moose sampled in 2022 (e.g., 1st ranked model: β = 4.88, SE = 1.52, 85% CI = [2.69, 7.08]) and marginally greater in moose sampled in 2021 (e.g., 2nd ranked model: β = 3.00, SE = 1.52, 85% CI = [0.82, 5.19]) relative to 2020. Concentrations of total globulins in female moose were similar between study areas (e.g., 1st ranked model: β = 0.33, SE = 1.95, 85% CI = [-2.47, 3.14]). Four models predicting haptoglobin concentrations showed ΔAICc scores less than two, collectively comprising 25% of the cumulative model weight (Fig. 3.3, Table 3.3). Across all top models, serum Cu was the only consistent and influential predictor of haptoglobin (e.g., 1st ranked model: β = 0.10, SE = 0.04, 85% CI = [0.04, 0.17]). Although the second through fourth ranked models included additional predictors such as alphaherpesvirus serostatus, serum Se, or winter tick presence, they differed from the top-ranked model by only one parameter each and had higher AICc scores. This pattern suggests they likely contain uninformative parameters and that the most parsimonious model is the first ranked model with serum Cu alone. Haptoglobin 60 concentrations were similar between study areas (1st ranked model: β = 0.01, SE = 0.11, 85% CI = [-0.16, 0.16]) and across years (1st ranked model — 2021: β = -0.06, SE = 0.07, 85% CI = [0.16, 0.05]; 2022: β = 0.02, SE = 0.06, 85% CI = [-0.06, 0.12]). Figure 3.2. Coefficient estimates and 85% confidence intervals (CI) for independent variables in the top-ranked generalized linear mixed effects models (< 2 ΔAICc) explaining total globulin concentrations in female moose (Alces alces) serum. Serum samples (n = 60) were collected during capture events in the winters of 2020–2022 from two study areas (Prince George South and the Bonaparte Plateau) in central British Columbia, Canada. An independent variable has an influential relationship with the mineral if the CI does not overlap 0. The continuous predictor serum Zn was standardized to a mean of zero and a standard deviation of one prior to analysis. Coefficients were ordered from most positive to most negative based on standardized estimates. AHV = Alphaherpesvirus. 61 Figure 3.3. Coefficient estimates and 85% confidence intervals (CI) for independent variables in the top-ranked generalized linear mixed effects models (< 2 ΔAICc) explaining haptoglobin concentrations in female moose (Alces alces) serum. Serum samples (n = 58; two outliers removed) were collected during capture events in the winters of 2020–2022 from two study areas (Prince George South and the Bonaparte Plateau) in central British Columbia, Canada. An independent variable has an influential relationship with the mineral if the CI does not overlap 0. Continuous predictors serum Cu and serum Se were standardized to a mean of zero and a standard deviation of one prior to analysis. Coefficients were ordered from most positive to most negative based on standardized estimates. AHV = Alphaherpesvirus. For the remaining immune biomarkers (i.e., IL-1β, IL-10, and IL-4), our predictor variables did not explain more variation in the data than the null model, which included study area and year. For IL-4, the first ranked model included serum Zn as a predictor, which was 62 positively associated with IL-4 concentrations (β = 0.45, SE = 0.28, 85% CI = [0.06, 0.85]), and the confidence interval did not encompass zero. The null model, however, ranked second and was within two AICc units, suggesting the addition of Zn did not meaningfully improve the model. For cytokines IL-1β and IL-10, the null model was the first ranked model. Female moose in the PGS study area had higher concentrations of IL-4 (β = 1.15, SE = 0.75, 85% CI = [0.07, 2.23]) and IL-10 (β = 1.21, SE = 0.82, 85% CI = [0.04, 2.39]) compared to females in the BP study area. However, IL-1β concentrations were similar between study areas (β = 0.68, SE = 0.69, 85% CI = [−0.32, 1.67]). No year-to-year variation was detected for IL-1β (2021: β = −0.46, SE = 0.41, 85% CI = [−1.05, 0.13]; 2022: β = −0.16, SE = 0.38, 85% CI = [−0.71, 0.40]), IL-10 (2021: β = −0.34, SE = 0.49, 85% CI = [−1.05, 0.36]; 2022: β = 0.23, SE = 0.47, 85% CI = [−0.45, 0.90]), or IL-4 (2021: β = −0.51, SE = 0.44, 85% CI = [−1.13, 0.12]; 2022: β = −0.03, SE = 0.41, 85% CI = [−0.62, 0.57]). Discussion Immunity plays a central role in helping animals cope with physiological and parasiterelated challenges. Measures of immune function can therefore offer valuable insights into individual health and population resilience (Ohmer et al., 2021). In this study, we demonstrated that serum concentrations of immune biomarkers in adult female moose reflected variation in both physiological condition and parasite exposure. We found that moose with greater fat reserves had greater concentrations of IL-12, suggesting that sufficient energetic reserves could enable greater investment in immune function. In addition, we found that total globulin levels were elevated in individuals exposed to micro- and macro-parasites, consistent with immune activation. We also found that Zn concentration was associated with both IL-12 and total globulin levels, and that Cu was associated with haptoglobin, underscoring the potential role of 63 trace minerals in modulating immune responses. Importantly, we believe this study represents the first to characterize cytokine expression in moose, establishing baseline values for a suite of immune biomarkers within this demographic—including cytokines, acute phase proteins, and total globulins—and revealing variation across years, populations, and immune biomarkers. Collectively, these findings underscore the complex and context-dependent interactions among nutrition, parasitism, and immune function in wild ungulates, and provide a foundation for future efforts to monitor moose health in the face of environmental change. Our findings suggest that energetic reserves could relate to immune investment in wild adult female moose, consistent with theoretical expectations that individuals in better nutritional condition can afford the costs of mounting stronger immune responses (Downs and Stewart, 2014; Ohmer et al., 2021). Specifically, we found that female moose with more body fat had greater concentrations of interleukin-12 (IL-12), a pro-inflammatory cytokine central to cellmediated immunity. Interleukin-12 promotes Th1 differentiation and enhances interferon-gamma (IFN-γ) production, both of which are critical for controlling intracellular pathogens (Graham et al., 2007; Zimmerman et al., 2014). The production of IL-12 is metabolically costly, as it contributes to inflammation, which can incur physiological damage if not tightly regulated. Thus, our results support the idea that individuals with greater energy stores may be better able to invest in demanding immune functions. Similar patterns have been reported in other mammals, such as northern elephant seals (Mirounga angustirostris), where body reserves positively influenced levels of pro-inflammatory cytokines across life history stages (Peck et al., 2016). This relationship could also potentially relate to reproductive history, as another study conducted on this same moose population found that females in better condition were less likely to have raised a calf in the previous year (Jefferies, 2024). This may suggest that females who avoided 64 the energetic costs of lactation may have had additional energy available to invest in their immune system. Another alternative explanation is that elevated IL-12 could partly reflect increased adipose tissue, as fat cells are known to secrete cytokines including IL-12 (ClementeSuárez et al., 2023). Overall, this finding provides an important initial insight into the potential ways by which energy balance may shape immune function in wild moose, highlighting the need for further research to better understand the mechanism underlying this relationship in natural populations. We found that total globulin concentrations were elevated in moose with winter tick infestations and in individuals seropositive for an alphaherpesvirus, consistent with immune activation following pathogen infection or exposure. Globulins represent a broad class of serum proteins involved in immune defense—such as antibodies and complement proteins—and often increase during prolonged or repeated immune challenges as the body responds to foreign antigens (Alberghina et al., 2011; Couch et al., 2017). Macro-parasitic infections, including winter tick infestations, induce type 2 immune responses that promote antibody production (Jolles et al., 2015), suggesting that infested adult female moose could have exhibited an adaptive immune reaction reflected in elevated globulin levels. This finding aligns with previous work on moose calves, where gamma-globulin concentrations increased shortly after the onset of winter tick engorgement (Glines and Samuel, 1989), highlighting the immune challenge posed by tick infestation. However, it remains unclear whether this immune response can effectively reduce or clear tick infestations, or whether it primarily helps moose cope with the physiological impacts of parasitism. Similarly, alphaherpesvirus seropositivity indicates prior viral exposure and the long-term persistence of specific antibodies (das Neves et al., 2010), which would also be reflected in total globulin levels. This relationship may also help to explain the temporal 65 variation in total globulins observed across years, as alphaherpesvirus seroprevalence increased from 52% in 2020 to 83% in 2022, paralleling increases in globulin concentrations. These findings suggest that total globulin concentrations could serve as reliable biomarkers of general immune activation and parasite exposure in moose, a consideration that becomes increasingly important as the prevalence of parasitic infections rises under changing environmental conditions. We found associations between trace minerals and several immune biomarkers, which may support the known roles of trace mineral status in immune regulation. Trace minerals are critical cofactors for numerous enzymes that support immune function, including antioxidant defense, cell proliferation, and the maintenance of physical barriers, including skin and mucous membranes (Kincaid, 2000; Underwood, 2012; Paul and Dey, 2015). Both Cu and Zn play pivotal roles in regulating innate and adaptive immunity by modulating cytokine production, lymphocyte activity, and controlling inflammation. In line with these functions, we found a strong positive association between serum Zn and the cytokine IL-12, and weaker evidence for a positive relationship between Zn and IL-4, although the latter model did not clearly outperform the null model. We also found a positive association between Cu and haptoglobin. Conversely, the negative association between Zn and globulin concentrations could potentially indicate that individuals with lower Zn levels are more vulnerable to chronic or persistent infections, as elevated globulins reflect sustained immune activation. Zinc deficiency could impair the ability to control infections effectively (Scott and Koski, 2000), leading to prolonged immune responses and increased globulin production. Because serum minerals were measured at a single time point, it is possible that animals experiencing immune challenges may have depleted mineral levels as a consequence of increased immune activity, rather than mineral status directly 66 influencing immune responses. Nonetheless, our results highlight possible links between trace mineral status and immune function in moose, underscoring the need for further investigation into the interactions among mineral deficiencies and immune challenges in wild populations. Given the energetic demands of reproduction, we expected pregnant females to show different immune responses due to trade-offs around energy availability (Stearns, 1989; Sheldon and Verhulst, 1996); however, we found that immune biomarker levels were similar between pregnant and non-pregnant individuals. This may, in part, reflect the timing of sample collection, as females were in early pregnancy and may not yet have incurred substantial energetic costs associated with pregnancy. Late gestation and active lactation are more energetically demanding phases when trade-offs with costly functions might be more pronounced (Clutton-Brock et al., 1989; Christe et al., 2000; Beasley et al., 2010). In addition, some moose testing positive for pregnancy may not have carried a fetus to term or produced a viable calf, limiting the reliability of a single serum test as an indicator of reproductive investment. Furthermore, the immune biomarkers we measured may not have captured the specific facets of immune function modulated during pregnancy. For example, a study on Dall sheep (Ovis dalli dalli) found that pregnant individuals had reduced bacterial killing ability, a functional measure of innate immunity, but showed no relationship between pregnancy and haptoglobin (Downs et al., 2018), highlighting that different immune measures vary in sensitivity to reproductive status. Future research should include a broader range of immune biomarkers and target sampling during critical reproductive stages, such as late gestation and post-parturition, to better capture shifts in immune investment across the reproductive timeline. Given the constraints of wildlife research, including limited access to individuals, logistical challenges of long-term monitoring, and the stress associated with capture, sampling moose during the spring and 67 summer to better capture reproductive stages was not feasible for this project. However, noninvasive approaches may help address these gaps. For example, Albery et al. (2019) used fecal IgA concentrations to assess immune function across seasons and reproductive states in wild red deer (Cervus elaphus). Similar methods could be applied to moose to better understand how immune investment shifts across reproductive stages and energetic demands throughout the year. Although the cytokines measured in this study have distinct roles in immune regulation in model species, we found that all were strongly positively correlated. This pattern suggests that multiple cytokine pathways may be activated simultaneously, reflecting a broad, coordinated immune response rather than discrete or polarized signaling (Graham et al., 2007; Zimmerman et al., 2014). Such upregulation could result from environmental or physiological stressors such as infection, injury, or chronic inflammation that activate several immune pathways concurrently. Despite strong correlations among cytokines, we found no associations between any cytokine and other immune biomarkers such as haptoglobin or globulins, nor were haptoglobin and globulins correlated with each other. This lack of association likely reflects fundamental differences in the timing and function of these immune responses. Cytokines are typically produced in rapid, short-lived bursts at the site of immune activation and can quickly return to baseline levels (Graham et al., 2007; Zimmerman et al., 2014). In contrast, haptoglobin is synthesized in the liver in response to cytokine signaling (Gruyse et al., 2005; Peck et al., 2016; Libera et al., 2022), introducing a delay in its expression that depends on the timing and intensity of upstream signals. Globulins, which include circulating antibodies and other serum proteins, reflect longer-term or repeated immune stimulation and tend to change more gradually over time (Alberghina et al., 2011; Couch et al., 2017). Given that our measurements were taken from single-point serum samples, we may have missed temporal fluctuations and dynamic changes in 68 these biomarkers, underscoring the importance of repeated longitudinal sampling to capture the complexity of immune responses fully. The absence of co-variation among immune biomarkers may also reflect the nature of immune challenges currently experienced by this population, which may not elicit strong or temporally aligned responses across immune axes. These findings highlight the complexity and individuality of immune responses in free-ranging moose and emphasize the importance of assessing multiple immune components to characterize systemic immune activity in ecological studies. To our best knowledge, this study provides the first record of serum cytokines measured in free-ranging moose. Incorporating cytokines alongside traditional immune biomarkers expands the range of immunological responses that can be measured in wild animals and enhances our ability to assess individual immune status. We found that a commercial porcine multiplex assay exhibited strong cross-reactivity with moose serum, with nine of thirteen cytokines detectable in at least 80% of samples. The assay yielded consistent, replicable signals across individuals, and even those cytokines with lower detection frequencies displayed clear variation in fluorescence intensity—suggesting biologically meaningful differences. Individual cytokine concentrations varied with physiological metrics and across study areas, further supporting the sensitivity of these biomarkers to ecological and individual factors. However, this assay has not yet been formally validated for moose using functional tests such as immune stimulation tests. Future research should prioritize validation efforts to confirm assay specificity and sensitivity in this species, which will be essential for advancing cytokine-based immune monitoring in wildlife (Levin et al., 2014; Borque et al., 2020). In summary, our study revealed potential relationships between immunity, parasite exposure, and host physiology in free-ranging moose. The immune system plays a critical role in 69 protecting animals from infection, maintaining physiological homeostasis, and promoting population resilience (Schmid-Hempel, 2021). Therefore, understanding factors linked with immunity is essential for predicting infection outcomes and overall health of wild populations, which can inform targeted monitoring and management strategies (Ohmer et al., 2021). Although our analysis was exploratory in nature, we identified preliminary relationships with immune biomarkers that warrant further investigation; for example, cytokines may be particularly responsive to nutritional stressors in female moose, while globulins may better reflect chronic immune stimulation, such as persistent parasite exposure, suggesting these biomarkers could serve as useful indicators in ongoing health assessments. Our findings also provide valuable baseline data for immune biomarkers in adult female moose and underscore the importance of continued monitoring to evaluate how environmental change may affect wildlife health. Future research employing longitudinal sampling across multiple seasons and infection stages would be valuable to better capture the dynamic relationships between immune profiles, infection progression, parasite clearance, and fitness outcomes. Such approaches would provide critical insights into the functional role of immunity in mediating host-pathogen interactions under natural environmental conditions and support adaptive management efforts. Ultimately, this study provides a foundation for future efforts to understand the causes and consequences of immune variation in moose and other large herbivores inhabiting rapidly changing landscapes. 70 CHAPTER 4: Conclusions Research summary and implications The rapid pace of environmental change is profoundly altering the landscapes and ecological conditions on which wildlife depend, with important consequences for their nutrition, health, and population dynamics (Acevedo-Whitehouse and Duffus, 2009). Changes in climate and habitat quality, in particular, can affect the availability and nutritional value of forage, thereby influencing the physiological condition and fitness of large herbivores (Parker et al., 2009; Stephenson et al., 2020). Monitoring health biomarkers (i.e., measurable indicators that provide insight into wildlife health) offers a valuable approach to understanding how large herbivores physiologically respond to environmental stressors (Stephen, 2014; Thacker et al., 2019; Wittrock et al., 2019; Aleuy et al., 2023). For example, measuring essential mineral levels and immune biomarkers can reveal early signs of nutritional deficiencies or subclinical disease, thereby guiding adaptive conservation strategies that promote population resilience. Health biomarkers may also help elucidate the mechanisms behind recent population changes observed in some wildlife species. For example, in the early 2000s, a widespread mountain pine beetle (MPB) outbreak caused significant pine tree mortality across much of British Columbia, leading to intensified salvage logging that dramatically transformed the landscape. Concurrently, some moose populations experienced steep declines of up to 70% (Kuzyk et al., 2018), hypothesized to be caused by rapid landscape changes. Initial findings from long-term research and monitoring suggest that bottom-up factors (i.e., food quality and/or availability) may have contributed to these declines based on evidence of starvation and healthrelated mortalities combined with suboptimal pregnancy rates (Thacker et al., 2019). The mechanisms linking environmental variation, including food quality and quantity, with moose 71 health and population declines, however, are poorly understood. Considering the complex interplay of natural disturbances, human activities, and climate effects shaping moose habitat and population trends, a more thorough understanding of the factors that are associated moose health, and ultimately population numbers, is critical. Accordingly, my thesis aimed to identify the environmental and physiological conditions associated with essential mineral concentrations and immune responses in adult female moose from two study areas affected by landscape disturbances and shifts in climatic regimes. In Chapter 2, I examined associations between hair mineral concentrations and environmental variables using GPS collar data, spatial datasets, and hair samples collected during winter captures as part of the BC Provincial Moose Research Project. I found that essential mineral concentrations in moose hair varied in relation to climatic conditions, habitat composition, and landscape disturbances. Specifically, female moose with greater concentrations of Se and Zn experienced increased precipitation, suggesting a potential link between climate-driven vegetation changes and mineral uptake. Mineral concentrations of K and Mg were greater in female moose spending more time in deciduous forests, consistent with the nutritional characteristics of these habitats. In addition, moose with access to recently burned areas had greater Zn concentrations in their hair, indicating a potential connection between post-fire environments and forage quality and availability. I also documented the concentrations of a suite of macro minerals, trace minerals, and heavy metals in the hair of female moose from central British Columbia, revealing notable variation between study populations and across years. This chapter demonstrates that hair mineral analysis can serve as a potential indicator of environmental variation and provides baseline data important for ongoing health monitoring in moose. 72 In Chapter 3, I measured concentrations of several immune biomarkers in the serum of female moose and explored how these markers related to physiological condition and parasite exposure. Moose with more fat reserves had greater concentrations of IL-12, suggesting that individuals in better body condition may be able to allocate more resources toward immune function. Total globulin levels were elevated in moose exposed to both micro- and macroparasites, indicating immune activation in response to parasitic challenges. Additionally, I observed associations between zinc concentrations and both IL-12 and total globulin levels, whereas copper levels were linked to haptoglobin, highlighting the possible role of trace minerals in modulating immune responses. Altogether, these results underscore the intricate links between nutrition, parasitism, and immune function in wild ungulates, laying crucial groundwork for future monitoring of moose health under changing environmental conditions. Collectively, my findings offer important insights into patterns of moose responses to environmental change. Mineral status and immune biomarkers were dynamic and reflected variation in environmental and physiological conditions, underscoring their value for monitoring wildlife health. This work provides a foundation for future research aimed at refining how health biomarkers can be integrated into long-term monitoring frameworks, particularly in systems experiencing rapid landscape change. For example, patterns linking mineral concentrations to habitat type, disturbance history, and climatic conditions highlight opportunities to investigate how habitat composition, such as the extent of deciduous stands or post-fire regeneration, affects forage quality and nutritional status. Similarly, the observed associations between trace minerals, body condition, and immune biomarkers point to promising avenues for exploring the nutritional underpinnings of disease susceptibility and resilience in wild ungulates. Building on these preliminary relationships, future studies could clarify the pathways connecting environmental 73 pressures, nutrition, and immunity, and ultimately inform adaptive management and conservation strategies that support resilient moose populations. Limitations and future directions One limitation of my research is that I examined broad-scale relationships between habitat composition and trace mineral concentrations in moose hair using the proportion of GPS collar points within different habitat types. This approach provides valuable insights into habitat use patterns but does not directly reflect the specific plant species consumed by moose within those habitats. In addition, diet is not the only factor affecting mineral status; other environmental sources such as water intake and the use of natural mineral licks may also contribute to variation in mineral levels (Spears, 1994; Ayotte et al., 2006). Consequently, my findings are limited to generalizations about habitat-level influences on mineral exposure rather than precise connections that could link diet composition and other environmental inputs with essential mineral uptake. Future research that integrates habitat use data with dietary analyses, mineral profiling of forage, and assessments of alternative mineral sources would help clarify the mechanisms driving the spatial patterns observed in this study and enhance our understanding of how fine-scale foraging choices and environmental factors shape mineral status in moose populations. Further, my analysis is limited by a lack of understanding of the degree to which trace minerals in hair reflect levels found in the whole body. Although hair sampling is increasingly used in wildlife research due to its minimally invasive nature, the relationship between mineral concentrations in hair and those in storage organs remains poorly understood, and reference intervals have not been established, leading to ambiguity in how this metric should be interpreted. Studies validating hair mineral concentrations have yielded conflicting evidence 74 regarding their correspondence with organ levels; for instance, Jutha et al. (2022) found that hair concentrations reflected Co, Mo, and Se in liver and/or kidney, but not Cu, Fe, Mn, or Zn. To date, no studies have assessed the correlation between hair and organ mineral levels specifically in moose. In this study, paired samples were not available because only minimally invasive, liveanimal sampling was conducted during captures. However, this represents a critical knowledge gap and a valuable area for future research, as establishing these relationships in moose would improve the interpretation and application of hair mineral analysis for wildlife health monitoring. Such validations could be accomplished through targeted harvest-based sampling and coordinated surveillance efforts (Jutha et al., 2022). An additional limitation of my research lies in the interpretation that higher mineral concentrations in hair may be assumed to reflect better health or improved nutritional status in moose. However, elevated mineral levels do not necessarily indicate optimal physiological functioning, as both deficiencies and excesses of certain minerals can have adverse physiological consequences. Although many minerals are essential for key biological functions, excessive accumulation can lead to toxicity, metabolic disturbances, or impaired organ function. For example, while trace minerals such as Se are essential for immune function and antioxidant defense, chronically elevated levels can result in adverse effects including hair loss, hoof deformities, and reproductive impairments (Puls, 1988; Flueck et al., 2012; Raisbeck, 2020). Moose, like other wildlife, must maintain mineral intake within a narrow optimal range, and the balance among different minerals can further complicate how they are absorbed and utilized in the body (Underwood, 2012). In this study, I was not able to determine whether the observed hair mineral concentrations fell within normal, deficient, or potentially harmful thresholds, 75 underscoring the need for future research to establish reference ranges and to investigate how mineral levels relate to health indicators, fitness outcomes, and survival in moose populations. Similarly, interpreting immune biomarkers also presents challenges because higher levels of immune activity do not necessarily indicate better health or effective parasite resistance. Immune responses are energetically costly and can reflect ongoing infection, inflammation, or physiological stress rather than successful parasite clearance or recovery. Elevated immune biomarkers may signal trade-offs where animals allocate resources toward immune function at the expense of other vital processes such as growth or reproduction (Stearns, 1989; Sheldon and Verhulst, 1996). Although some immune biomarkers in this study were associated with parasite exposure, whether these responses correspond to effective parasite management or simply immune activation remains unclear. Furthermore, my sample included only moose with repeated measures across multiple years, which limited my ability to directly link immune variation to survival outcomes. Without longitudinal data tracking immune biomarkers alongside fitness measures such as survival or reproductive success, I was unable to assess whether elevated immunity reflects beneficial or detrimental health states. This limitation highlights the need for future research that integrates immune profiling with long-term monitoring across multiple seasons and parasite infection stages to better understand the effects of immune function and parasite exposure on wild moose populations. Finally, a major limitation of my research, and of the broader provincial moose research project, is that only adult female moose were captured and collared for long-term monitoring. As a result, males and calves were not represented in the data. Male moose and calves may exhibit different behaviours, such as variations in diet, habitat use, and movement patterns (Bowyer, 2004; Oehlers et al., 2011), which could influence their exposure to environmental conditions 76 and lead to distinct mineral or immune profiles that were not captured by this study. Additionally, within the group of adult females, detailed age classifications were not available, which limited the ability to examine age-related variation in mineral status and immune profiles (Nussey et al., 2012; Draghi et al., 2023). Expanding future research to include males, calves, and more precise age distinctions among females would provide a more comprehensive understanding of how different individuals within the moose population interact with their environment and how these differences may influence health and fitness outcomes, ultimately affecting overall moose population performance. 77 References Abdulla M, Chmielnicka J (1979) New aspects on the distribution and metabolism of essential trace elements after dietary exposure to toxic metals. Biol Trace Elem Res 23: 25–53. https://doi.org/10.1007/BF02917176. Acevedo-Whitehouse K, Duffus ALJ (2009) Effects of environmental change on wildlife health. Proc Biol Sci 364: 3429–3438. https://doi.org/10.1098/rstb.2009.0128. Aguilar XF, Leclerc LM, Kugluktuk Angoniatit Association, Ekaluktutiak Hunters & Trappers Organization, Olokhaktomiut Hunters & Trappers Committee, Mavrot F, RobertoCharron A, Tomaselli M, Mastromonaco G, Gunn A, et al. (2023) An integrative and multi-indicator approach for wildlife health applied to an endangered caribou herd. Sci Rep 13: 16524. https://doi.org/10.1038/s41598-023-41689-y. Alberghina D, Giannetto C, Vazzana I, Ferrantelli V, Piccione G (2011) Reference intervals for total protein concentration, serum protein fractions, and albumin/ globulin ratios in clinically healthy dairy cows. J Vet Diagn Invest 23: 111–114. https://doi.org/10.1177/104063871102300119. Albery GF, Watt KA, Keith R, Morris S, Morris A, Kenyon F, Nussey DH, Pemberton JM (2019) Reproduction has different costs for immunity and parasitism in a wild mammal. Funct Ecol 34: 229–239. https://doi.org/10.1111/1365-2435.13475. Aleuy OA, Kutz S, Mallory ML, Provencher JF (2023) Wildlife health in environmental impact assessments: are we missing a key metric? Environ Rev 31: 348–359. https://doi.org/10.1139/er-2022-0023. Alfaro RI, van Akker L, Hawkes B (2015) Characteristics of forest legacies following two mountain pine beetle outbreaks in British Columbia, Canada. Can J For Res 45: 1387– 1396. https://doi.org/10.1139/cjfr-2015-0042. Ali H, Khan E (2019) Trophic transfer, bioaccumulation, and biomagnification of non-essential hazardous heavy metals and metalloids in food chains/webs—Concepts and implications for wildlife and human health. Hum Ecol Risk Assess 25: 1353–1376. https://doi.org/10.1080/10807039.2018.1469398. Ayotte JB, Parker KL, Arocena JM, Gillingham MP (2006) Chemical composition of lick soils: functions of soil ingestion by four ungulate species. J Mammal 87: 878–888. https://doi.org/10.1644/06-MAMM-A-055R1.1. Ayotte JB, Parker KL, Gillingham MP (2008) Use of natural licks by four species of ungulates in northern British Columbia. J Mammal 89: 1041–1050. https://doi.org/10.1644/07MAMM-A-345.1. 78 Baj J, Flieger W, Barbachowska A, Kowalska B, Flieger M, Forma A, Teresiński G, Portincasa P, Buszewicz G, Radzikowska-Büchner E, et al. (2023) Consequences of disturbing manganese homeostasis. Int J Mol Sci 24: 14959. https://doi.org/10.3390/ijms241914959. Balle C, Konstantinus IN, Jaumdally SZ, Havyarimana E, Lennard K, Esra R, Barnabas SL, Happel A-U, Moodie Z, Gill K, et al. (2020) Hormonal contraception alters vaginal microbiota and cytokines in South African adolescents in a randomized trial. Nat Commun 11: 5578. https://doi.org/10.1038/s41467-020-19382-9. Barboza PS, Parker KL, Hume ID (2009) Metabolic Constituents: Water, Minerals and Vitamins. In: Barboza PS, Parker KL, Hume ID, eds. Integrative Wildlife Nutrition. Springer, Berlin, Heidelberg, pp 157–206. Bariod L, Saïd S, Calenge C, Scheifler R, Fritsch C, Peroz C, Benabed S, Bidault H, Chabot S, Débias F, et al. (2024) Essential mineral elements in roe deer: Associations with parasites and immune phenotypes in two contrasting populations. Ecol Evol 14: e11613. https://doi.org/10.1002/ece3.11613. Bartoń K (2023) MuMIn: Multi-Model Inference. R package version 1.47.5. https://CRAN.Rproject.org/package=MuMIn. Beasley AM, Kahn LP, Windon RG (2010) The periparturient relaxation of immunity in Merino ewes infected with Trichostrongylus colubriformis: Parasitological and immunological responses. Vet Parasitol 168: 60–70. https://doi.org/10.1016/j.vetpar.2009.08.028. Becker DJ, Albery GF, Kessler MK, Lunn TJ, Falvo CA, Czirják GÁ, Martin LB, Plowright RK (2019) Macroimmunology: The drivers and consequences of spatial patterns in wildlife immune defence. J Anim Ecol 89: 972–995. https://doi.org/10.1111/1365-2656.13166. Beldomenico PM, Begon M (2010) Disease spread, susceptibility and infection intensity: vicious circles? Trends Ecol Evol 25: 21–27. https://doi.org/10.1016/j.tree.2009.06.015. Blakley BR, Kutz SJ, Tedesco SC, Flood PF (2000) Trace mineral and vitamin concentrations in the liver and serum of wild muskoxen from Victoria Island. J Wildl Dis 36: 301–307. https://doi.org/10.7589/0090-3558-36.2.301. Bondo KJ, Macbeth B, Schwantje H, Orsel K, Culling D, Culling B, Tryland M, Nymo IH, Kutz S (2019) Health survey of boreal caribou (Rangifer tarandus caribou) in northeastern British Columbia, Canada. J Wildl Dis 55: 544–562. https://doi.org/10.7589/2018-01-018. Bourgeon S, Kauffmann M, Geiger S, Raclot T, Robin J-P (2010) Relationships between metabolic status, corticosterone secretion and maintenance of innate and adaptive humoral immunities in fasted re-fed mallards. J Exp Biol 213: 3810–3818. https://doi.org/10.1242/jeb.045484. 79 Bourque J, Desforges JP, Levin M, Atwood TC, Sonne C, Dietz R, Jensen TH, Curry E, McKinney MA (2020) Climate-associated drivers of plasma cytokines and contaminant concentrations in Beaufort Sea polar bears (Ursus maritimus). Sci Total Environ 745: 140978. https://doi.org/10.1016/j.scitotenv.2020.140978. Bowyer RT (2004) Sexual segregation in ruminants: definitions, hypotheses, and implications for conservation and management. J Mammal 85: 1039–1052. https://doi.org/10.1644/BBL-002.1. Bowyer RT, Van Ballenberghe V, Kie JG, Maier JAK (1999) Birth-site selection by Alaskan moose: Maternal strategies for coping with a risky environment. J Mammal 80: 1070– 1083. https://doi.org/10.2307/1383161. Breen EJ, Tan W, Khan A (2016) The statistical value of raw fluorescence signal in Luminex xMAP based multiplex immunoassays. Sci Rep 6: 26996. https://doi.org/10.1038/srep26996. Breithaupt K, Rea RV, Gillingham MP, Aitken DA, Hodder DP (2024) Using winter diet composition and forage plant availability to determine browse selection and importance for moose (Alces alces) in a landscape modified by industrial forestry. Forestry 98: 167– 180. https://doi.org/10.1093/forestry/cpae019. Brooks ME, Kristensen K, van Benthem KJ, Magnusson A, Berg CW, Nielsen A, Skaug HJ, Maechler M, Bolker BM (2017) glmmTMB Balances Speed and Flexibility Among Packages for Zero-inflated Generalized Linear Mixed Modeling. R J 9: 378–400. https://doi.org/10.32614/RJ-2017-066. Brown CL, Coe PK, Clark DA, Wisdom MJ, Rowland MM, Averett JP, Johnson BK (2022) Climate change effects on understory plant phenology: implications for large herbivore forage availability. Environ Res Ecol 1: 011002. https://doi.org/10.1088/2752664X/ac7fb0. Brown GE, Foster AL, Ostergren JD (1999) Mineral surfaces and bioavailability of heavy metals: a molecular-scale perspective. Proc Natl Acad Sci USA 96: 3388–3395. https://doi.org/10.1073/pnas.96.7.3388. Brunner FS, Schmid‐Hempel P, Barribeau SM (2014) Protein‐poor diet reduces host‐specific immune gene expression in Bombus terrestris. Proc Biol Sci 281: 20140128. https://doi.org/10.1098/rspb.2014.0128. Burnham KP, Anderson DR (2002) Model Selection and Multimodel Inference: A Practical information-Theoretic Approach. 2nd ed. Springer. 80 Ceballos G, Ehrlich PR, Barnosky AD, García A, Pringle RM, Palmer TM (2015) Accelerated modern human–induced species losses: Entering the sixth mass extinction. Sci Adv 1: e1400253. https://doi.org/10.1126/sciadv.1400253. Chandra RK (1996) Nutrition, immunity and infection: from basic knowledge of dietary manipulation of immune responses to practical application of ameliorating suffering and improving survival. Proc Natl Acad Sci USA 93: 14304–14307. https://doi.org/10.1073/pnas.93.25.14304. Christe P, Arlettaz R, Vogel P (2000) Variation in intensity of a parasitic mite (Spinturnix myoti) in relation to the reproductive cycle and immunocompetence of its bat host (Myotis myotis). Ecol Lett 3: 207–212. https://doi.org/10.1046/j.1461-0248.2000.00142.x. Clemente-Suárez VJ, Redondo-Flórez L, Beltrán-Velasco AI, Martín-Rodríguez A, MartínezGuardado I, Navarro-Jiménez E, Laborde-Cárdenas CC, Tornero-Aguilera JF (2023) The role of adipokines in health and disease. Biomedicines 11: 1290. https://doi.org/10.3390/biomedicines11051290. Clutton-Brock TH, Albon SD, Guinness FE (1989) Fitness costs of gestation and lactation in wild mammals. Nature 337: 260–262. https://doi.org/10.1038/337260a0. Combs DK (1987) Hair analysis as an indicator of mineral status of livestock. J Anim Sci 65: 1753–1758. https://doi.org/10.2527/jas1987.6561753x. Cooke SJ, Sack L, Franklin CE, Farrell AP, Beardall J, Wikelski M, Chown SL (2013) What is conservation physiology? Perspectives on an increasingly integrated and essential science. Conserv Physiol 1: cot001. https://doi.org/10.1093/conphys/cot001. Couch CE, Movius MA, Jolles AE, Gorman ME, Rigas JD, Beechler BR (2017) Serum biochemistry panels in African buffalo: Defining reference intervals and assessing variability across season, age and sex. PLoS One 12: e0176830. https://doi.org/10.1371/journal.pone.0176830. das Neves CG, Roth S, Rimstad E, Thiry E, Tryland M (2010) Cervid herpesvirus 2 infection in reindeer: a review. Vet Microbiol 143: 70–80. https://doi.org/10.1016/j.vetmic.2010.02.015. DataBC (2024) BC Data Catalogue. Province of British Columbia. https://www2.gov.bc.ca/gov/content/data/bc-data-catalogue. Dickinson ER, Mosbacher JB, Arnison C, Beckmen K, Côté SD, Francesco JD, Hansson SV, Jahromi EZ, Kinniburgh DW, Le Roux G, et al. (2025) Qiviut trace and macro element profile reflects muskox population trends. Ecol Evol 15: e71020. https://doi.org/10.1002/ece3.71020. 81 Downs C, Adelman J, Demas G (2014) Mechanisms and methods in ecoimmunology: integrating within-organism and between organism processes. Integr Comp Biol 54: 340–352. https://doi.org/10.1093/icb/icu082. Downs CJ, Boan BV, Lohuis TD, Stewart KM (2018) Investigating relationships between reproduction, immune defenses, and cortisol in Dall sheep. Front Immunol 9: 105. https://doi.org/10.3389/fimmu.2018.00105. Downs CJ, Stewart KM (2014) A primer in ecoimmunology and immunology for wildlife research and management. Calif Fish Game 100: 371–395. Retrieved from: https://nrm.dfg.ca.gov/FileHandler.ashx?DocumentID=93569. Draghi S, Agradi S, Riva F, Tarhan D, Bilgiç B, Dokuzeylül B, Ercan AM, Or ME, Brecchia G, Vigo D, et al. (2023) Roe deer (Capreolus capreolus) hair as a bioindicator for the environmental presence of toxic and trace elements. Toxics 11: 49. https://doi.org/10.3390/toxics11010049. Encel SA, Simpson EK, Schaerf TM, Ward AJW (2023) Immune challenge affects reproductive behaviour in the guppy (Poecilia reticulata). R Soc Open Sci 10: 230579. http://doi.org/10.1098/rsos.230579. Environment and Climate Change Canada (2024) Canadian Climate Normals 1991–2020. Retrieved from: https://climate.weather.gc.ca/climate_normals/index_e.html. Fisher JT, Wilkinson L (2005) The response of mammals to forest fire and timber harvest in the North American boreal forest. Mamm Rev 35: 51–81. https://doi.org/10.1111/j.13652907.2005.00053.x. Flueck WT (1994) Effect of trace elements on population dynamics: Selenium deficiency in freeranging black-tailed deer. Ecol 75: 807–812. https://doi.org/10.2307/1941736. Flueck WT, Smith-Flueck JM, Mionczynski J, Mincher BJ (2012) The implications of selenium deficiency for wild herbivore conservation: a review. Eur J Wildl Res 58: 761–780. https://doi.org/10.1007/s10344-012-0645-z. Flynn A, Franzmann AW, Arneson PD, Oldemeyer JL (1977) Indications of copper deficiency in a subpopulation of Alaskan moose. J Nutr 107: 1182–1189. https://doi.org/10.1093/jn/107.7.1182. Frank A, McPartlin J, Danielsson R (2004) Nova Scotia moose mystery--a moose sickness related to cobalt- and vitamin B12 deficiency. Sci Total Environ 318: 89–100. https://doi.org/10.1016/S0048-9697(03)00374-7. Franzmann AW (1981). Alces alces. Mamm Species 154: 1–7. https://doi.org/10.2307/3503876. 82 French AS, Shaw D, Gibb SW, Taggart MA (2017) Geochemical landscapes as drivers of trace and toxic element profiles in wild red deer (Cervus elaphus). Sci Total Environ 601: 1606–1618. https://doi.org/10.1016/j.scitotenv.2017.05.210. French SS, Johnston GIH, Moore MC (2007) Immune activity suppresses reproduction in foodlimited female tree lizards Urosaurus ornatus. Funct Ecol 21: 1115–1122. https://doi.org/10.1111/j.1365-2435.2007.01311.x. Gilot-Fromont E, Jégo M, Bonenfant C, Gibert P, Rannou B, Klein F, Gaillard J-M (2012) Immune phenotype and body condition in roe deer: individuals with high body condition have different, not stronger immunity. PLoS One 7: e45576. https://doi.org/10.1371/journal.pone.0045576. Glines MV, Samuel WM (1989) Effect of Dermacentor albipictus (Acari: Ixodidae) on blood composition, weight gain and hair coat of moose, Alces alces. Exp Appl Acarol 6: 197– 213. https://doi.org/10.1007/BF01193980. Graham AL, Cattadori IM, Lloud-Smith JO, Ferrari MJ, Bjørnstad ON (2007) Transmission consequences of coinfection: cytokines writ large? Trends Parasitol 23: 284–291. https://doi.org/10.1016/j.pt.2007.04.005. Graham AL, Hayward AD, Watt KA, Pilkington JG, Pemberton JM, Nussey DH (2010) Fitness correlates of heritable variation in antibody responsiveness in a wild mammal. Sci 330: 662–665. https://doi.org/10.1126/science.1194878. Gruyse E, Toussaint MJM, Niewold TA, Koopmans SJ (2005) Acute phase reaction and acute phase proteins: a review. J Zhejiang Univ Sci B 6B: 1045–1056. https://doi.org/10.1631/jzus.2005.B1045. Gulland FMD (1995) Impact of Infectious Diseases on Wild Animal Populations: a Review. In: Grenfell BT, Dobson AP, eds. Ecology of Infectious Diseases in Natural Populations. Cambridge University Press, Cambridge, pp 20–51. Hartig F (2024) DHARMa: Residual Diagnostics for Hierarchical (Multi-Level / Mixed) Regression Models. R package version 0.4.7. https://CRAN.Rproject.org/package=DHARMa. Hayward AD, Pilkington JG, Wilson K, McNeilly TN, Watt KA (2019) Reproductive effort influences intra-seasonal variation in parasite-specific antibody responses in wild Soay sheep. Funct Ecol 33: 1307–1320. https://doi.org/10.1111/1365-2435.13330. Herdt TH, Hoff B (2011) The use of blood analysis to evaluate trace mineral status in ruminant livestock. Vet Clin North Am Food Anim Pract 27: 255–283. https://doi.org/10.1016/j.cvfa.2011.02.004. 83 Herrada A, Bariod L, Saïd S, Rey B, Bidault H, Bollet Y, Chabot S, Débias F, Duhayer J, Pardonnet S, et al. (2024) Minor and trace element concentrations in roe deer hair: A non-invasive method to define reference values in wildlife. Ecol Indic 159: 111720. https://doi.org/10.1016/j.ecolind.2024.111720. Hjeljord OG, Hövik N, Pedersen HB (1990) Choice of feeding sites by moose during summer, the influence of forest structure and plant phenology. Ecography 13: 281–292. https://doi.org/10.1111/j.1600-0587.1990.tb00620.x. Hollingsworth KA, Shively RD, Glasscock SN, Light JE, Tolleson DR, Barboza PS (2021) Trace mineral supplies for populations of little and large herbivores. PLoS One 16: e0248204. https://doi.org/10.1371/journal.pone.0248204. Huxter CE, Rea RV, Otter KA, Hesse G (2024) Behaviours of moose at roadside mineral licks in British Columbia: Implications for moose-vehicle collisions. Appl Anim Behav Sci 275: 106292. https://doi.org/10.1016/j.applanim.2024.106292. Ilmonen P, Taarna T, Hasselquist D (2000) Experimentally activated immune defence in female pied flycatchers results in reduced breeding success. Proc R Soc Lond B 267: 665–670. http://doi.org/10.1098/rspb.2000.1053. Jefferies C-E (2024) Effects of landscape change on cow moose body fat and physiology in central British Columbia. MSc Thesis. University of Northern British Columbia. Jolles AE, Beechler BR, Dolan BP (2014) Beyond mice and men: environmental change, immunity and infections in wild ungulates. Parasite Immunol 37: 255–266. https://doi.org/10.1111/pim.12153. Joly K, Sorum MS, Craig T, Julianus EL (2017) The effects of sex, terrain, wildfire, winter severity, and maternal status on habitat selection by moose in north-central Alaska. Alces 52: 101–115. Retrieved from: https://alcesjournal.org/index.php/alces/article/view/165. Jutha N, Jardine C, Schwantje H, Mosbacher J, Kinniburgh D, Kutz S (2022) Evaluating the use of hair as a non-invasive indicator of trace mineral status in woodland caribou (Rangifer tarandus caribou). PLoS One 17: e0269441. https://doi.org/10.1371/journal.pone.0269441. Kabata-Pendias A (2010) Trace elements in soils and plants. 4th ed. CRC Press, Inc. Kasischke ES, Turetsky MR (2006) Recent changes in the fire regime across the North American boreal region—spatial and temporal patterns of burning across Canada and Alaska. Geophys Res Lett 33: L09703. https://doi.org/10.1029/2006GL025677. Kelsall JP, Telfer ES, Wright TD (1977) The effects of fire on the ecology of the Boreal Forest with particular reference to the Canadian North and selected bibliography. Occasional 84 Paper Number 32. Fisheries and Environment Canada, Canadian Wildlife Service, Ottawa, ON. Kincaid RL (2000) Assessment of trace mineral status of ruminants: A review. J Anim Sci 77: 1– 10. https://doi.org/10.2527/jas2000.77E-Suppl1x. Koetke LJ, Hodder DP, Rea RV, Johnson CJ, Marshall S (2023) Landscape disturbance alters the composition and diversity of the diet of moose, a generalist herbivore. For Ecol Manag 530: 120760. https://doi.org/10.1016/j.foreco.2022.120760. Kreulen DA (1985) Lick use by large herbivores: a review of benefits and banes of soil consumption. Mamm Rev 15: 107–123. https://doi.org/10.1111/j.13652907.1985.tb00391.x. Kuzyk G, Heard D (2014) Research design to determine factors affecting moose population change in British Columbia: testing the landscape change hypothesis. Ministry of Forests, Lands, and Natural Resource Operations, Victoria, BC. Kuzyk G, Marshall S, Procter C, Schindler H, Schwantje H, Gillingham M, Hodder D, White S, Mumma M (2018) Determining factors affecting moose population change in British Columbia: Testing the landscape change hypothesis, 2018 Progress Report: February 2012–April 2018. Ministry of Forests, Lands, Natural Resource Operations and Rural Development, Victoria, BC. Lamb CT, Dubman E, McNay RS, Giguere L, Majchrzak Y, Thacker C, Slater O, Macbeth B, Owens-Beek N, Muir B, et al. (2024) Assessing the health-fitness dynamics of endangered mountain caribou and the influence of maternal penning. Can J Zool 102: 673–690. https://doi.org/10.1139/cjz-2023-0032. Landhausser SM, Wein RW (1993) Postfire vegetation recovery and tree establishment at the arctic treeline: climate-change-vegetation-response hypotheses. J Ecol 81: 665–672. https://doi.org/10.2307/2261664. Leroux SJ (2019) On the prevalence of uninformative parameters in statistical models applying model selection in applied ecology. PLoS One 14: e0206711. https://doi.org/10.1371/journal.pone.0206711. Levin M, Romano T, Matassa K, De Guise S (2014) Validation of a commercial canine assay kit to measure pinniped cytokines. Vet Immunol Immunopathol 160: 90–96. https://doi.org/10.1016/j.vetimm.2014.04.001. Libera K, Szopka W, Ratajczak AM, Pomorska-Mól M (2022) Acute phase proteins in wildlife and their domesticated relatives. Med Weter 78: 272–278. https://doi.org/10.21521/mw.6662. 85 Lochmiller RL, Deerenberg C (2000) Trade-offs in evolutionary immunology: just what is the cost of immunity? Oikos 88: 87–98. https://doi.org/10.1034/j.1600-0706.2000.880110.x. Lord R, Kielland K (2015) Effects of variable fire severity on forage production and foraging behavior of moose in winter. Alces 51: 23–34. Retrieved from: https://alcesjournal.org/index.php/alces/article/view/147. Maier JAK, Ver Hoef JM, McGuire AD, Bowyer RT, Saperstein L, Maier HA (2005) Distribution and density of moose in relation to landscape characteristics: effects of scale. Can J For Res 35: 2233–2243. https://doi.org/10.1139/x05-123. Makimura S, Suzuki N (1982) Quantitative determination of bovine serum Haptoglobin and its elevation in some inflammatory diseases. Jpn J Vet Sci 44: 15–21. https://doi.org/10.1292/jvms1939.44.15. Martin LB 2nd, Navara KJ, Weil ZM, Nelson RJ (2007) Immunological memory is compromised by food restriction in deer mice Peromyscus maniculatus. Am J Physiol Regul Integr Comp Physiol 292: R316 –R320. https://doi.org/10.1152/ajpregu.00386.2006. McClure SJ (2003). Mineral nutrition and its effects on gastrointestinal immune function of sheep. Aust J Exp Agric 43: 1455–1461. https://doi.org/10.1071/EA03002. McClure SJ (2008). How minerals may influence the development and expression of immunity to endoparasites in livestock. Parasite Immunol 30: 89–100. https://doi.org/10.1111/j.1365-3024.2007.00996.x. McDonough TJ, Thompson DP, Crouse JA, Dale BW, Badajos OH (2022) Evaluation of impacts of vaginal implant transmitter use in moose. Wildl Soc Bull 46: e1378. https://doi.org/10.1002/wsb.1378. McDowell LR (1992) Minerals in Animal and Human Nutrition. Academic Press, New York, USA. McLaren AAD, Patterson BR (2021) There’s no place like home — site fidelity by female moose (Alces alces) in central Ontario, Canada. Can J Zool 99: 557–563. https://doi.org/10.1139/cjz-2021-0010. Meidinger D, Pojar J (1991) Ecosystems of British Columbia. BC Ministry of Forests and Range Research Branch, Victoria, BC. Melin MJ, Matala J, Mehtätalo L, Tiilikainen R, Tikkanen OP, Maltamo M, Pusenius J, Packalen P (2014) Moose (Alces alces) reacts to high summer temperatures by utilizing thermal shelters in boreal forests - an analysis based on airborne laser scanning of the canopy structure at moose locations. Glob Change Biol 20: 1115–1125. https://doi.org/10.1111/gcb.12405. 86 Mosbacher JP, Desforges JP, Michelsen A, Hansson SV, Stelvig M, Eulaers I, Sonne C, Dietz R, Jenssen BM Ciesielski TM, et al. (2022) Hair mineral levels as indicator of wildlife demographics?—a pilot study of muskoxen. Polar Res 41: 8543. https://doi.org/10.33265/polar.v41.8543. Mumma MA, Bevington AR, Marshall S, Gillingham MP (2024) Delineating wildfire burns and regrowth using satellite imagery to assess moose (Alces alces) spatial responses to burns. Ecosphere 15: e4793. https://doi.org/10.1002/ecs2.4793. Murray DL, Cox EW, Ballard WB, Whitlaw HA, Lenarz MS, Custer TW, Barnett T, Fuller TK (2006) Pathogens, nutritional deficiency, and climatic influences on a declining moose population. Wildl Monogr 166: 1–30. https://doi.org/10.2193/00840173(2006)166[1:PNDACI]2.0.CO;2. Muylkens B, Thiry J, Kirten P, Schynts F, Thiry E (2007) Bovine herpesvirus 1 infection and infectious bovine rhinotracheitis. Vet Res 38: 181–209. https://doi.org/10.1051/vetres:2006059. Newbold T, Hudson LN, Hill SLL, Contu S, Lysenko I, Senior RA, Börger L, Bennett DJ, Choimes A, Collen B, et al. (2015) Global effects of land use on local terrestrial biodiversity. Nat 520: 45–50. https://doi.org/10.1038/nature14324. Newby JR, DeCesare NJ (2020) Multiple nutritional currencies shape pregnancy in a large herbivore. Can J Zool 98: 307–315. https://doi.org/10.1139/cjz-2019-0241. Nussey DH, Watt K, Pilkington JG, Zamoyska R, McNeilly TN (2012) Age-related variation in immunity in a wild mammal population. Aging Cell 11: 178–180. https://doi.org/10.1111/j.1474-9726.2011.00771.x. Oehlers SA, Bowyer RT, Huettmann F, Person DK, Kessler WB (2011) Sex and scale: implications for habitat selection by Alaskan moose Alces alces gigas. Wildl Biol 17: 67– 84. https://doi.org/10.2981/10-039. O’Hara TM, Carroll G, Barboza P, Mueller K, Blake J, Woshner V, Willetto C (2001) Mineral and heavy metal status as related to a mortality event and poor recruitment in a moose population in Alaska. J Wildl Dis 37: 509–522. https://doi.org/10.7589/0090-355837.3.509. Ohlson M, Staaland H (2001) Mineral diversity in wild plants: benefits and bane for moose. Oikos 94: 442–454. https://doi.org/10.1034/j.1600-0706.2001.940307.x. Ohmer MEB, Costantini D, Czirják GÁ, Downs CJ, Ferguson LV, Flies A, Franklin CE, Kayigwe AN, Knutie S, Richards-Zawacki CL, et al. (2021) Applied ecoimmunology: using immunological tools to improve conservation efforts in a changing world. Conserv Physiol 9: coab074. https://doi.org/10.1093/conphys/coab074. 87 Oster KW, Barboza PS, Gustine DD, Joly K, Shively RD (2018) Mineral constraints on arctic caribou (Rangifer tarandus): a spatial and phenological perspective. Ecosphere 9: e02160. https://doi.org/10.1002/ecs2.2160. Oster KW, Gustine DD, Smeins FE, Barboza PS (2024) Estimating mineral requirements of wild herbivores: modelling arctic caribou (Rangifer tarandus granti) in summer. Animals 14: 868. https://doi.org/10.3390/ani14060868. Parfitt B (2007) Over-cutting and waste in B.C.'s interior: a call to rethink B.C.'s pine beetle logging strategy. Canadian Centre for Policy Alternatives BC Office, Vancouver, BC. Park H, Jeong S, Peñuelas J (2020) Accelerated rate of vegetation green-up related to warming at northern high latitudes. Glob Chang Biol 26: 6190–6202. https://doi.org/10.1111/gcb.15322. Parker KL, Barboza PS, Gillingham MP (2009) Nutrition integrates environmental responses of ungulates. Funct Ecol 23: 57–69. https://doi.org/10.1111/j.1365-2435.2009.01528.x. Paul SS, Dey A (2015) Nutrition in health and immune function of ruminants. Indian J Anim Sci 85: 103–112. https://doi.org/10.56093/ijans.v85i2.46557. Peck HE, Costa DP, Crocker DE (2016) Body reserves influence allocation to immune responses in capital breeding female northern elephant seals. Funct Ecol 30: 389–397. https://doi.org/10.1111/1365-2435.12504. Pollock B (2005) Trace elements status of white-tailed deer (Odocoileus virginianus) and moose (Alces alces) in Nova Scotia. Canadian Cooperative Wildlife Health Centre: Newsletters & Publications 45. Post E, Pederson C, Wilmers CC, Forchhammer MC (2008) Warming, plant phenology and the spatial dimension of trophic mismatch for large herbivores. Proc Biol Sci 275: 2005– 2013. https://doi.org/10.1098/rspb.2008.0463. Procter C, Anderson M, Scheideman M, Marshall S, Schindler H, Schwantje H, Hodder D, Blythe E (2020) Factors affecting moose population declines in British Columbia, 2020 Progress report: Feb 2012–May 2020. Ministry of Forests, Lands, Natural Resource Operations and Rural Development, Victoria BC. Puls R (1988) Mineral levels in animal health. Diagnostic data. Sherpa International, Clearbrook. R Core Team, 2023. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/. Raisbeck MF (2020) Selenosis in ruminants. Vet Clin North Am Food Anim Pract 36: 775–789. https://doi.org/10.1016/j.cvfa.2020.08.013. 88 Rakic F (2022) Hair Biomarkers to Support Barren-ground Caribou Health Monitoring and Management. MSc Thesis. University of Calgary. Rea RV, Hodder DP, Child KN (2013) Year-round activity patterns of moose (Alces alces) at a natural mineral lick in north central British Columbia, Canada. Can Wildl Biol Manag 2: 37–41. Retrieved from: https://cwbm.ca/year-round-activity-patterns-of-moose-alcesalces-at-a-natural-mineral-lick-in-north-central-british-columbia-canada/. Richardson JB, Friedland AJ (2016) Influence of coniferous and deciduous vegetation on major and trace metals in forests of northern New England, USA. Plant Soil 402: 363–378. https://doi.org/10.1007/s11104-016-2805-5. Rioux E, Pelletier F, Mosbacher JP, Lesmerises F, St-Louis R, Kutz S, St-Laurent MH (2022) Links between individual performance, trace elements and stable isotopes in an endangered caribou population. Glob Ecol Conserv 38: e02234. https://doi.org/10.1016/j.gecco.2022.e02234. Robbins CT (1983) Wildlife Feeding and Nutrition. Academic Press, New York, USA. Roberge C (2023) Forage quality and moose (Alces alces) nutrition in a logged landscape. MSc Thesis. Thompson Rivers University. Rocha-Santos L, Pessoa MS, Cassano CR, Talora DC, Orihuela RLL, Mariano-Neto E, MoranteFilho JC, Faria D, Cazetta E (2016) The shrinkage of a forest: Landscape-scale deforestation leading to overall changes in local forest structure. Biol Conserv 196: 1–9. https://doi.org/10.1016/j.biocon.2016.01.028. Ruiz M, Wang D, Reinke BA, Demas GE, Martins EP (2011) Trade-offs between reproductive coloration and innate immunity in a natural population of female sagebrush lizards, Sceloporus graciosus. Herp J 21: 131–134. Retrieved from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4231297/. Ruoss S, Becker NI, Otto MS, Czirják GÁ, Encarnação JA (2019) Effect of sex and reproductive status on the immunity of the temperate bat Myotis daubentonii. Mamm Biol 94: 120– 126. https://doi.org/10.1016/j.mambio.2018.05.010. Rustad L, Campbell J, Marion G, Norby R, Mitchell M, Hartley A, Cornelissen J, Gurevitch J, GCTE-NEWS (2001) A meta-analysis of the response of soil respiration, net nitrogen mineralization, and aboveground plant growth to experimental ecosystem warming. Oecologia 126: 543–562. https://doi.org/10.1007/s004420000544. Samuel WM (2004) White as a ghost: Winter ticks and moose. Federation of Alberta Naturalists, Edmonton, AB. 89 Scheideman M (2018) Use and selection at two spatial scales by female moose (Alces alces) across central British Columbia following a mountain pine beetle outbreak. MSc Thesis. University of Northern British Columbia. Schmid-Hempel P (2021) Evolutionary parasitology: the integrated study of infections, immunology, ecology, and genetics. 2nd ed. Oxford University Press, Oxford. Schmidt IK, Jonasson S, Shaver GR, Michelsen A, Nordin A (2002) Mineralization and distribution of nutrients in plants and microbes in four arctic ecosystems: responses to warming. Plant Soil 242: 93–106. https://doi.org/10.1023/A:1019642007929. Scott ME, Koski KG (2000) Zinc deficiency impairs immune responses against parasitic nematode infections at intestinal and systemic sites. J Nutr 130:1412S–1420S. https://doi.org/10.1093/jn/130.5.1412S. Sheldon BC, Verhulst S (1996) Ecological immunology: costly parasite defences and trade-offs in evolutionary ecology. Trends Ecol Evol 11: 317–321. https://doi.org/10.1016/01695347(96)10039-2. Simard DG, Fyles JW, Paré D, Nguyen T (2001) Impacts of clearcut harvesting and wildfire on soil nutrient status in the Quebec boreal forest. Can J Soil Sci 81: 229–237. https://doi.org/10.4141/S00-028. Skinner JG, Brown RA, Roberts L (1991) Bovine haptoglobin response in clinically defined field conditions. Vet Rec 128:147–149. https://doi.org/10.1136/vr.128.7.147. Smith PN, Cobb GP, Godard-Codding C, Hoff D, McMurry ST, Rainwater TR, Reynolds KD (2007) Contaminant exposure in terrestrial vertebrates. Environ Pollut 150: 41–64. https://doi.org/10.1016/j.envpol.2007.06.009. Sokolov V, Chernova O (1987) Morphology of the skin of moose. Swedish Wildl Res Suppl 1: 367–375. Retrieved from: https://www.researchgate.net/profile/Olga-Chernova4/publication/260120646_Morphology_of_the_skin_of_Moose_Alces_alces_L/links/57f 39e9608ae886b897d9d1d/Morphology-of-the-skin-of-Moose-Alces-alces-L.pdf. Spears JW (1994) Minerals in forages. In: Fahey GCJ, ed. Forage Quality, Evaluation, and Utilization. American Society of Agronomy, Madison, WI, USA, pp 281–317. Spitzer R, Ericson M, Felton AM, Heim M, Raubenheimer D, Solberg EJ, Wam HK, Rolandsen CM (2024) Camera collars reveal macronutrient balancing in free-ranging male moose during summer. Ecol Evol 14: e70192. https://doi.org/10.1002/ece3.70192. Staaland H, White RG (2001) Regional variation in mineral contents of plants and its significance for migration by Arctic reindeer and caribou. Alces 37: 497–509. Retrieved from: https://www.alcesjournal.org/index.php/alces/article/view/611. 90 Stearns SC (1989) Trade-offs in life-history evolution. Funct Ecol 3: 259–268. https://doi.org/10.2307/2389364. Stephen C (2014) Toward a modernized definition of wildlife health. J Wildl Dis 50: 427–430. https://doi.org/10.7589/2013-11-305. Stephenson TR, Crouse JA, Hundertmark KJ, Keech MA (2001) Vitamin E, selenium, and reproductive losses in Alaskan moose. Alces 37: 210–206. Retrieved from: https://alcesjournal.org/index.php/alces/article/view/567. Stephenson TR, German DW, Cassirer EF, Walsh DP, Blum ME, Cox M, Stewart KM, Monteith KL (2020) Linking population performance to nutritional condition in an alpine ungulate. J Mammal 101: 1244–1256. https://doi.org/10.1093/jmammal/gyaa091. Stephenson TR, Hundertmark KJ, Schwartz CC, Van Ballenberghe V (1998) Predicting body fat and body mass in moose with ultrasonography. Can J Zool 76: 717–722. https://doi.org/10.1139/z97-248. Sutherland C, Hare D, Johnson PJ, Linden DW, Montgomery RA, Droge E (2023) Practical advice on variable selection and reporting using Akaike information criterion. Proc R Soc B 290: 20231261. https://doi.org/10.1098/rspb.2023.1261. Suttle NF (2010) Mineral Nutrition of Livestock. 4th ed. CABI, Wallingford, UK. Tankersley NG, Gasaway WC (1983) Mineral lick use by moose in Alaska. Can J Zool 61: 2242–2249. https://doi.org/10.1139/z83-296. Thacker C, Macbeth BJ, Kuzyk G, Marshall S, Procter C, Schwantje H (2019) British Columbia Provincial Moose research project health assessment summary 2013–2018. Prepared for Ministry of Forests, Lands, Natural Resource Operations and Rural Development, Victoria, BC. Timmermann HR, Rodgers AR (2017) The status and management of moose in North America – circa 2015. Alces 53: 1–22. Retrieved from: https://www.alcesjournal.org/index.php/alces/article/view/177. Tischler KB, Severud WJ, Peterson RO, Bump JK (2019) Aquatic macrophytes are seasonally important dietary resources for moose. Divers 11: 209. https://doi.org/10.3390/d11110209. Underwood E (2012) Trace Elements in Human and Animal Nutrition. 5th ed. Academic Press. Valderrábano J, Gomez-Rincón C, Uriarte J (2006) Effect of nutritional status and fat reserves on the periparturient immune response to Haemonchus contortus infection in sheep. Vet Parasitol 141: 122–131. https://doi.org/10.1016/j.vetpar.2006.04.029. 91 van Beest FM, Schmidt NM, Stewart L, Hansen LH, Michelsen A, Mosbacher JB, Gilbert H, Le Roux G, Hansson SV (2023) Geochemical landscapes as drivers of wildlife reproductive success: Insights from a high-Arctic ecosystem. Sci Total Environ 903: 166567. https://doi.org/10.1016/j.scitotenv.2023.166567. van Beest FM, van Moorter B, Milner JM (2012) Temperature-mediated habitat use and selection by a heat-sensitive northern ungulate. Anim Behav 84: 723–735. https://doi.org/10.1016/j.anbehav.2012.06.032. Van Dyke F, Probert BL, Van Beek GM (1995) Moose home range fidelity and core area characteristics in south-central Montana. Alces 31: 93–104. Retrieved from: https://www.alcesjournal.org/index.php/alces/article/view/883. van Noordwijk AJ, de Jong G (1986) Acquisition and allocation of resources: Their influence on variation in life history tactics. Am Nat 128: 137–142. https://doi.org/10.1086/284547. Vikøren T, Kristoffersen AB, Lierhagen S, Handeland K (2011) A comparative study of hepatic element levels in wild moose, roe deer, and reindeer from Norway. J Wildl Dis 47: 661– 672. https://doi.org/10.7589/0090-3558-47.3.661. Wallace KME, Hart DW, Venter F, Janse van Vuuren AK, Bennett NC (2023) The best of both worlds: no apparent trade-off between immunity and reproduction in two group-living African mole-rat species. Phil Trans R Soc B 378: 20220310. http://doi.org/10.1098/rstb.2022.0310. Wang L, Wang D, He Z, Liu G, Hodgkinson KC (2010) Mechanisms linking plant species richness to foraging of a large herbivore. J Appl Ecol 47: 868–875. https://doi.org/10.1111/j.1365-2664.2010.01837.x. Wang T, Hamann A, Spittlehouse D, Carroll C (2016) Locally downscaled and spatially customizable climate data for historical and future periods for north America. PLoS One 11: e0156720. https://doi.org/10.1371/journal.pone.0156720. Whitman E, Parisien MA, Thompson DK, Flannigan MD (2019) Short-interval wildfire and drought overwhelm boreal forest resilience. Sci Rep 9: 18796. https://doi.org/10.1038/s41598-019-55036-7. Wikelski M, Cooke SJ (2006) Conservation physiology. Trends Ecol Evol 21: 38–46. https://doi.org/10.1016/j.tree.2005.10.018. Wittrock J, Duncan C, Stephen C (2019) A determinants of health conceptual model for fish and wildlife health. J Wildl Dis 55: 285–297. https://doi.org/10.7589/2018-05-118. 92 Wurtz TL, Zasada JC (2001) An alternative to clear-cutting in the boreal forest of Alaska: a 27year study of regeneration after shelterwood harvesting. Can J For Res 31: 999–1011. https://doi.org/10.1139/x01-014. Zimmerman LM, Bowden RM, Vogel LA (2014) A vertebrate cytokine primer for ecoimmunologists. Funct Ecol 28: 1061–1073. https://doi.org/10.1111/1365-2435.12273. Zimmerman TJ, Jenks JA, Leslie DM, Neiger RD (2008) Hepatic minerals of white-tailed and mule deer in the southern Black Hills, South Dakota. J Wildl Dis 44: 341–350. https://doi.org/10.7589/0090-3558-44.2.341. Zuur AF, Ieno EN, Elphick CS (2010) A protocol for data exploration to avoid common statistical problems. Methods Ecol Evol 1: 3–14. https://doi.org/10.1111/j.2041210X.2009.00001.x. 93 APPENDIX A: Supplemental Information for Chapter 2 Table A.1. Full candidate models and model selection statistics used to explain potassium (K) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model. Model Study area + Year + Deciduous forest + Mixed forest Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian Study area + Year + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Temperature + Precipitation Study area + Year + Wetland + Riparian + Temperature + Precipitation Study area + Year + Forest stand age + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Temperature + Precipitation Study area + Year + Wildfire + Temperature + Precipitation Study area + Year + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year (null) Study area + Year + Forest stand age Study area + Year + Wildfire Study area + Year + Wetland + Riparian Study area + Year + Wildfire + Forest stand age 94 df logLik AICc 7 -368.80 756.42 0.00 0.59 9 -367.76 760.01 3.59 0.10 7 -370.97 760.76 4.34 0.07 9 -368.23 760.95 4.53 0.06 9 -368.30 761.09 4.67 0.06 8 -370.29 762.18 5.76 0.03 10 -367.63 762.76 6.34 0.02 10 -367.76 763.02 6.60 0.02 11 -366.24 763.13 6.71 0.02 8 -370.95 763.50 7.08 0.02 9 -370.28 765.06 8.64 0.01 11 -367.62 765.89 9.47 0.01 13 -365.64 768.61 12.19 0.00 5 6 6 7 7 -379.97 -379.89 -379.94 -379.61 -379.87 773.51 775.94 776.03 778.05 778.56 0.00 0.00 0.00 0.00 0.00 ΔAIC AICc wi 17.09 19.52 19.61 21.63 22.14 Table A.2. Full candidate models and model selection statistics used to explain magnesium (Mg) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model. Model Study area + Year + Deciduous forest + Mixed forest Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian Study area + Year + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Temperature + Precipitation Study area + Year (null) Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Forest stand age Study area + Year + Wetland + Riparian Study area + Year + Forest stand age + Temperature + Precipitation Study area + Year + Wildfire + Temperature + Precipitation Study area + Year + Wetland + Riparian + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Temperature + Precipitation Study area + Year + Wildfire Study area + Year + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age Study area + Year + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year + Wildfire + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age + Temperature + Precipitation 95 df logLik AICc 8 -284.27 587.36 0.00 0.44 10 -282.52 589.53 2.17 0.15 8 -285.99 590.81 3.46 0.08 10 -283.30 591.08 3.73 0.07 6 -289.26 592.11 4.76 0.04 11 -282.44 592.37 5.02 0.04 11 -282.51 592.52 5.17 0.03 8 -286.86 592.54 5.19 0.03 9 -285.82 593.24 5.89 0.02 9 -285.99 593.59 6.23 0.02 10 -284.66 593.81 6.45 0.02 12 -281.63 593.91 6.55 0.02 7 7 -289.24 -289.25 594.63 7.27 594.66 7.31 0.01 0.01 12 -282.43 595.50 8.15 0.01 10 -285.82 596.13 8.77 0.01 8 -289.22 597.27 9.92 0.00 14 -281.63 600.59 13.24 0.00 ΔAIC AICc wi Table A.3. Full candidate models and model selection statistics used to explain copper (Cu) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model. Model df logLik Study area + Year (null) Study area + Year + Temperature + Precipitation Study area + Year + Wildfire Study area + Year + Deciduous forest + Mixed forest Study area + Year + Forest stand age Study area + Year + Wetland + Riparian Study area + Year + Forest stand age + Temperature + Precipitation Study area + Year + Wildfire + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Temperature + Precipitation Study area + Year + Wildfire + Forest stand age Study area + Year + Wetland + Riparian + Temperature + Precipitation Study area + Year + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age + Temperature + Precipitation 6 8 7 -75.44 -73.39 -74.94 ΔAIC AICc wi 164.46 0.00 0.29 165.60 1.14 0.16 166.03 1.58 0.13 8 -73.86 166.55 2.09 0.10 7 8 -75.41 -74.73 166.98 2.52 168.28 3.82 0.08 0.04 9 -73.38 168.36 3.90 0.04 9 -73.39 168.37 3.91 0.04 10 -72.04 168.57 4.11 0.04 8 -74.93 168.69 4.23 0.03 10 -72.92 170.32 5.87 0.02 10 -73.38 171.25 6.79 0.01 10 -73.61 171.71 7.26 0.01 11 -73.25 174.00 9.54 0.00 11 -73.26 174.02 9.56 0.00 12 -71.77 174.19 9.73 0.00 12 -72.98 176.59 12.14 0.00 14 -71.66 180.66 16.20 0.00 96 AICc Table A.4. Full candidate models and model selection statistics used to explain iron (Fe) concentrations in the hair (n = 59) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model. Model df logLik AICc Study area + Year (null) Study area + Year + Wildfire Study area + Year + Forest stand age Study area + Year + Deciduous forest + Mixed forest Study area + Year + Temperature + Precipitation Study area + Year + Wildfire + Forest stand age Study area + Year + Wetland + Riparian Study area + Year + Wildfire + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Temperature + Precipitation Study area + Year + Forest stand age + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian Study area + Year + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year + Wetland + Riparian + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age + Temperature + Precipitation 6 7 7 8 -230.49 -230.23 -230.31 -229.32 474.60 476.65 476.82 477.53 ΔAIC AICc wi 0.00 0.39 2.04 0.14 2.21 0.13 2.93 0.09 8 8 8 9 -229.73 -230.03 -230.11 -228.98 478.34 478.94 479.09 479.64 3.74 4.34 4.49 5.04 0.06 0.04 0.04 0.03 10 -228.22 481.02 6.42 0.02 9 -229.68 481.04 6.44 0.02 10 -228.58 481.74 7.14 0.01 10 -228.95 482.48 7.87 0.01 10 -229.01 482.61 8.01 0.01 11 -228.30 484.22 9.61 0.00 11 -228.51 484.63 10.03 0.00 12 -227.65 486.08 11.47 0.00 12 -228.22 487.23 12.62 0.00 14 -226.86 491.27 16.67 0.00 97 Table A.5. Full candidate models and model selection statistics used to explain manganese (Mn) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model. Model df logLik AICc Study area + Year (null) Study area + Year + Temperature + Precipitation Study area + Year + Wetland + Riparian Study area + Year + Wildfire Study area + Year + Forest stand age Study area + Year + Deciduous forest + Mixed forest Study area + Year + Wildfire + Temperature + Precipitation Study area + Year + Forest stand age + Temperature + Precipitation Study area + Year + Wetland + Riparian + Temperature + Precipitation Study area + Year + Wildfire + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian Study area + Year + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age + Temperature + Precipitation 6 8 8 7 7 -150.82 -148.95 -149.10 -150.72 -150.82 315.23 316.72 317.03 317.60 317.78 8 -149.67 318.16 2.93 0.07 9 -148.80 319.20 3.97 0.04 9 -148.95 319.49 4.26 0.04 10 -147.61 319.71 4.48 0.03 8 -150.72 320.26 5.03 0.02 10 -148.41 321.31 6.08 0.01 10 -148.62 321.73 6.50 0.01 10 -148.80 322.08 6.85 0.01 11 -148.46 324.43 9.20 0.00 11 -148.59 324.69 9.46 0.00 12 -147.55 325.74 10.51 0.00 12 -148.45 327.53 12.30 0.00 14 -147.50 332.34 17.11 0.00 98 ΔAIC AICc wi 0.00 0.30 1.49 0.14 1.80 0.12 2.37 0.09 2.56 0.08 Table A.6. Full candidate models and model selection statistics used to explain molybdenum (Mo) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model. Model df logLik AICc Study area + Year (null) Study area + Year + Forest stand age Study area + Year + Wildfire Study area + Year + Temperature + Precipitation Study area + Year + Wetland + Riparian Study area + Year + Deciduous forest + Mixed forest Study area + Year + Wildfire + Forest stand age Study area + Year + Forest stand age + Temperature + Precipitation Study area + Year + Wildfire + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian Study area + Year + Wetland + Riparian + Temperature + Precipitation Study area + Year + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age + Temperature + Precipitation 6 7 7 8 8 122.60 123.49 122.62 123.80 123.58 -231.61 -230.83 -229.08 -228.78 -228.34 ΔAIC AICc wi 0.00 0.32 0.78 0.22 2.53 0.09 2.82 0.08 3.27 0.06 8 123.55 -228.28 3.33 0.06 8 123.54 -228.25 3.36 0.06 9 124.22 -226.83 4.77 0.03 9 123.98 -226.36 5.24 0.02 10 124.77 -225.05 6.56 0.01 10 124.51 -224.52 7.08 0.01 10 124.42 -224.36 7.25 0.01 10 124.08 -223.67 7.94 0.01 11 125.37 -223.25 8.36 0.00 11 124.82 -222.14 9.47 0.00 12 125.46 -220.28 11.32 0.00 12 125.16 -219.68 11.92 0.00 14 126.01 -214.68 16.93 0.00 99 Table A.7. Full candidate models and model selection statistics used to explain selenium (Se) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model. Model df logLik AICc Study area + Year + Temperature + Precipitation Study area + Year + Wildfire + Temperature + Precipitation Study area + Year + Forest stand age + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Temperature + Precipitation Study area + Year + Wetland + Riparian + Temperature + Precipitation Study area + Year + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age Study area + Year (null) Study area + Year + Forest stand age Study area + Year + Wetland + Riparian Study area + Year + Wildfire Study area + Year + Wildfire + Forest stand age 8 78.44 -138.05 ΔAIC AICc wi 0.00 0.51 9 78.47 -135.34 2.71 0.13 9 78.46 -135.31 2.74 0.13 10 79.65 -134.81 3.24 0.10 10 79.45 -134.41 3.64 0.08 10 78.48 -132.48 5.57 0.03 12 80.86 -131.09 6.96 0.02 14 80.91 -124.48 13.57 0.00 8 71.47 -124.11 13.94 0.00 10 73.34 -122.20 15.85 0.00 11 74.05 -120.61 17.45 0.00 11 73.40 -119.30 18.75 0.00 12 74.06 -117.49 20.56 0.00 6 7 8 7 8 61.88 62.64 63.61 61.96 62.74 -110.18 -109.12 -108.39 -107.77 -106.66 27.87 28.93 29.66 30.28 31.39 0.00 0.00 0.00 0.00 0.00 100 Table A.8. Full candidate models and model selection statistics used to explain zinc (Zn) concentrations in the hair (n = 60) of female moose (Alces alces) in two study areas in central British Columbia, Canada. All models included a random intercept for Individual ID to account for repeated measures. Study area and year were included as fixed effects in every model. Model Study area + Year + Wildfire + Temperature + Precipitation Study area + Year + Temperature + Precipitation Study area + Year + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year + Forest stand age + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Temperature + Precipitation Study area + Year (null) Study area + Year + Wetland + Riparian + Temperature + Precipitation Study area + Year + Wildfire Study area + Year + Forest stand age Study area + Year + Wetland + Riparian Study area + Year + Deciduous forest + Mixed forest Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Temperature + Precipitation Study area + Year + Wildfire + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age + Temperature + Precipitation Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Forest stand age Study area + Year + Deciduous forest + Mixed forest + Wetland + Riparian + Wildfire + Forest stand age 101 df logLik AICc ΔAIC AICc wi 9 -162.98 347.56 0.00 0.37 8 -164.53 347.89 0.33 0.31 10 -162.98 350.44 2.89 0.09 9 -164.53 350.66 3.10 0.08 10 -163.72 351.92 4.37 0.04 6 -169.23 352.04 4.48 0.04 10 -164.19 352.87 5.31 0.03 7 7 8 -169.09 -169.21 -168.36 354.34 354.57 355.55 6.78 7.02 7.99 0.01 0.01 0.01 8 -168.72 356.27 8.71 0.00 12 -163.14 356.91 9.35 0.00 8 -169.08 356.98 9.43 0.00 14 -161.56 360.45 12.89 0.00 10 -168.00 360.50 12.94 0.00 11 -167.86 363.23 15.67 0.00 11 -168.00 363.50 15.94 0.00 12 -167.86 366.36 18.80 0.00 APPENDIX B: Supplemental Information for Chapter 3 Figure B.1. Spearman correlation matrix showing paired correlation coefficients between immune biomarkers concentrations measured in serum samples (n = 60) from adult female moose sampled in winter (2020–2022) from two populations in central British Columbia, Canada: the Bonaparte Plateau (BP; n = 31) and Prince George South (PGS; n = 29). Each coloured box represents the strength and direction of the paired correlation, with colours ranging from red (negative correlation) to blue (positive correlation). The correlation coefficient ranges from -1 (perfect negative correlation) to 1 (perfect positive correlation), with 0 indicating no correlation. Immune biomarkers GM-CSF, IFN-γ, IL-8, and TNF-α were excluded from this analysis due to having less than 80% detectable samples. 102