Indigenous Knowledge and Biomarkers of Physiological Stress Inform Muskox Conservation in a Rapidly Changing Arctic

Home , Arctic reindeer, Inuvik (crater), Lipoptena cervi, Wildlife endocrinology

University of Calgary PRISM: University of Calgary's Digital Repository

Graduate Studies The Vault: Electronic Theses and Dissertations

2020-12-11 Indigenous Knowledge and Biomarkers of Physiological Stress Inform Muskox Conservation in a Rapidly Changing Arctic

Di Francesco, Juliette

Di Francesco, J. (2020). Indigenous Knowledge and Biomarkers of Physiological Stress Inform Muskox Conservation in a Rapidly Changing Arctic (Unpublished doctoral thesis). University of Calgary, Calgary, AB. http://hdl.handle.net/1880/112840 doctoral thesis

University of Calgary graduate students retain copyright ownership and moral rights for their thesis. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission. Downloaded from PRISM: https://prism.ucalgary.ca UNIVERSITY OF CALGARY

Indigenous Knowledge and Biomarkers of Physiological Stress Inform Muskox Conservation in a Rapidly Changing Arctic

Juliette Di Francesco

A THESIS

SUBMITTED TO THE FACULTY OF GRADUATE STUDIES

IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE

DEGREE OF DOCTOR OF PHILOSOPHY

GRADUATE PROGRAM IN VETERINARY MEDICAL SCIENCES

CALGARY, ALBERTA

DECEMBER, 2020

Glucocorticoids play a key role in energy regulation and are mediators of the physiological stress response in mammals. Their concentrations are commonly measured in wildlife to understand the effects of environmental changes and anthropogenic disturbances, but their use is associated with multiple challenges and there is a need for species-specific validation. Muskoxen (Ovibos moschatus) are an essential part of the Arctic ecosystem, where they have a strong economic, nutritional, and sociocultural value for Indigenous communities. Recent population declines and mortality events suggest that muskoxen may be threatened by the multiple environmental changes and associated stressors to which they are increasingly exposed. Overall, I sought to establish fecal glucocorticoid metabolites (FGM) and qiviut (woolly undercoat) cortisol as biomarkers of physiological stress in muskoxen, and to apply these tools together with Indigenous knowledge (IK) to explore potential causes and patterns of physiological stress in wild muskoxen. Through two repeated pharmacological challenges in captive muskoxen, I showed that qiviut cortisol and FGM levels accurately reflect long-term (over the period of the hair’s growth) and short-term changes in circulating cortisol, respectively. I also demonstrated that changes in circulating cortisol are not reflected in qiviut in the absence of growth and highlighted variations across body regions, significant differences in qiviut segments over time, and differences between shed and unshed qiviut. Additionally, I documented IK which provided novel insights on the potential stressors of muskoxen and their specific importance. Finally, I identified important factors influencing qiviut cortisol (sex, geographical location, season, and year), and found associations between qiviut cortisol and marrow fat and lungworm intensity. Findings were interpreted in part collaboratively with IK holders. This work has advanced our understanding of glucocorticoid deposition and stability in hair, and of the limitations and challenges associated with hair glucocorticoid interpretation. It has highlighted the multiple benefits of incorporating IK in wildlife endocrinology studies and provided a framework for doing so. Finally, identifying factors associated with qiviut cortisol is a key step to simultaneously investigating the causes and consequences, both at the individual and population levels, of physiological stress in muskoxen.

Preface

This thesis consists of manuscripts that have been published in peer-reviewed journals, submitted for publication, or are intended to be submitted for publication. Juliette Di Francesco designed this study, collected the data, analyzed the data, interpreted the results, and wrote the papers with guidance from her primary supervisor, her thesis committee, and other collaborators. All co-authors provided important technical and intellectual support while revising the papers. Written permission for reproduction in their entirety of the scientific articles included in this thesis has been obtained from the publishers and all co- authors.

Published manuscripts

Chapter 2 – Di Francesco J, Navarro-Gonzalez N, Wynne-Edwards K, Peacock S, Leclerc L-M, Tomaselli M, Davison T, Carlsson A, Kutz S (2017) Qiviut cortisol in muskoxen as a potential tool for informing conservation strategies. Conservation Physiology, 5(1):cox052.

Submitted manuscripts

Chapter 3 – Di Francesco J, Hanke A, Milton T, Leclerc L-M, Kugluktuk Angoniatit Association, Gerlach C, Kutz S. Documenting Indigenous knowledge to identify and understand the stressors that affect muskoxen (Ovibos moschatus). ARCTIC (submitted on September 25th, 2020; currently under review).

Chapter 4 – Di Francesco J, Mastromonaco GF, Rowell JE, Blake J, Checkley SL, Kutz S. Fecal glucocorticoid metabolites reflect hypothalamic–pituitary–adrenal axis activity in muskoxen (Ovibos moschatus). PLoS ONE (submitted on September 9th, 2020; currently under review).

Chapter 5 – Di Francesco J, Mastromonaco GF, Checkley SL, Blake J, Rowell JE, Kutz S. Qiviut cortisol reflects hypothalamic–pituitary–adrenal axis activity in muskoxen (Ovibos moschatus). General and Comparative Endocrinology (submitted on September 8th, 2020; currently under review).

Manuscripts intended to be submitted

Chapter 6 – Di Francesco J, Kwong GPS, Deardon R, Checkley SL, Mastromonaco GF, Mavrot F, Peacock S, Leclerc L-M, Kutz S. Intrinsic and extrinsic factors associated with increased qiviut cortisol in wild muskoxen (Ovibos moschatus). Intended to be submitted to Conservation Physiology or PeerJ.

Acknowledgements

I am extremely grateful to my supervisor, Dr. Susan Kutz, for giving me the opportunity to undertake this unique PhD research. Thank you for everything that you taught me, for letting me develop with a lot of liberty this amazing project with a lot of freedom while providing constant and invaluable guidance at every step of the way, and for always encouraging me. I also owe a great deal to my committee members, Drs. Gabriela Mastromonaco, Sylvia Checkley, and Craig Gerlach, for their availability and for providing me invaluable feedback and guidance both during the development of my PhD project and the analyses of the results. Thank you for your kindness, mentorship, and support throughout my PhD. This graduate experience has been extremely enriching both professionally and personally, and this has a lot to do with all of you. Drs. Steven Cooke and Mary Pavelka took the time to read my work: I am grateful for the stimulating discussions we had during my defense.

This work also benefitted from the input of several major collaborators, both at the University of Alaska Fairbanks and at the University of Calgary. Special thanks go to Drs. Jan Rowell, John Blake, Katherine Wynne-Edwards, Grace Kwong, Melanie Rock, Karin Orsel, and Rob Deardon for their invaluable contributions.

Numerous people participated in this project and helped me with sample collection in the field or processing in the lab. I am particularly grateful to all the staff at the Robert G. White Large Animal Research Station of the University of Alaska Fairbanks (Sarah Barcalow, Thalia Souza, Hanna Sfraga, Carla Wiletto, Christine Terzi, Megan Roberts, Jean Rein, Claire Kepner, and Charles Ashlock), Christine Gilman and Patricia Medd at the Endocrinology Laboratory of the Toronto Zoo, Kamala Sapkota, Felix Nwozu, Ruokun Zhou, my summer students (Akaysha Envik, Leslie Bottari, and Mélanie Meyer), and everyone who helped with the tedious qiviut sorting. A special thank you also to all the UCVM staff for their kindness and support (Joy Punsalan, Abir Bachir, Katrine Maurer, Robert Forsyth, and Kasia Wodjyla).

An important part of this project involved field work in the community of Kugluktuk, Nunavut. I am deeply grateful to all the hunters and community members who took part in muskox sample kit collection and participated in the interviews. This research would also not have been possible without support from the Kugluktuk Angoniatit Association, and particularly from Amanda Dumond, Larry Adjun, and Darlene Hokanak, or from the Government of Nunavut, especially Lisa-Marie Leclerc, Terry Milton, Kevin Methuen, Russell Akeeagok, and Allen Niptanatiak. It has been a real pleasure collaborating with all of you. I also thank Kugluktuk High School, who gave me the opportunity to do science outreach activities, and particularly Michael Valk, Haydn George, Brett MacCallum, and Rohan Hollingsworth. It was a very fun and enriching experience. Finally, I thank all the students I had the opportunity to teach and all the other wonderful people I got the chance to meet during my stays in the community, and especially Matt Stadnyk, Ronald Ladd, and Savannah Rose Hiko. The communities of Ulukhaktok, Ekaluktutiak, Paulatuk, and Sachs Harbour also participated in this project. My thanks go to the Government of the Northwest Territories, to the local Hunters and Trappers Committees/Organizations, and to all the harvesters who took part in sample kit collection.

This project would not have been possible without the multiple funding agencies who supported it: the Morris Animal Foundation, the NSERC-CREATE Host-Parasite Interactions Training Program, the NSERC-CREATE Integrated Training Program in Infectious Diseases, Food Safety and Public Policy, the University of Calgary Faculty of Graduate

iii

Studies Doctoral Scholarship, the University of Calgary Eyes High Seed grant, NSERC Discovery and NSERC Northern Supplement grants, Canada North Outfitting, the Government of Nunavut, Polar Knowledge Canada, the Government of the Northwest Territories, the Nunavut General Monitoring Plan, ArcticNet, and Irving Maritime Shipbuilding.

I was also lucky to be part of the NSERC-CREATE Host-Parasite Interactions Training Program and the NSERC-CREATE Integrated Training Program in Infectious Diseases, Food Safety and Public Policy. They provided a space for socializing and exchanging ideas, and I have fond memories of the workshops and externships that I was able to attend in connection with these programs.

In times of hardships, I was able to count on a very supportive work environment. A special thank you goes to all the Kutz lab members, Fabien Mavrot, Pratap Kafle, Stephanie Peacock, Anja Carlsson, Angeline McIntyre, Jesper Mosbacher, Tina Petersen, Andrea Hanke, Naima Jutha, Ale Aleuy, Matilde Tomaselli, Gilles Bourgoin, Consuelo Grassi, Filip Rakic, Regina Krohn, Kristin Bondo, Tessa Baker, Knut Madslien, Manigandan Lejeune, Collin Letain, and Xavier Fernandez Aguilar, as well as the Rock lab members, Dawn Rault, Ann Toohey, Amberlee Boulton, Taryn Graham, and Valli Fraser-Celin. It was awesome working with all of you! I also wish to thank James Wang, our amazing lab manager, for his kindness, cheerfulness, and for helping whenever I needed. Finally, thank you to Angie Schneider, our “lab mom”, not only for her invaluable support in the lab, but especially for always being there for me, I couldn’t have done it without you.

My stay in Calgary was filled with new and exciting experiences, and this had a lot to do with the amazing friends I made in Alberta: Sonja, Marija, Ana, Ravi, Camila, Alejandra, Anshula, Ivan, Kaitlyn, Barb, Edel, Rebecca, Roisin, Gina, Marta, Rogelio, Benjamin, Aïda, Méïssa, Pauline, and Adrian. You have given me wonderful memories that I will forever cherish, and I am already missing spending time with you!

I am highly grateful to my family, and particularly to my mom for always believing in me and for her eternal support. Thank you also to my brother and sister for always encouraging me.

Last, but not least, I am grateful to my husband Morgan. Thank you for undertaking this journey in Calgary with me, for always believing in me, and encouraging me at every step of the way. I love you with all my heart.

A special final thought goes to the non-human animals who contributed to this project: all the muskoxen who took part in my study and my dog, Maestro, for his unconditional love and for walking this path with me.

Table of Contents

Abstract ...... i Preface ...... ii Acknowledgements ...... iii Table of Contents ...... v List of Tables ...... ix List of Figures and Illustrations ...... xi List of Symbols, Abbreviations, and Nomenclature ...... xv

CHAPTER 1. INTRODUCTION ...... 1 1.1. Background information and rationale ...... 1 1.1.1. Why measure glucocorticoid levels in wildlife? ...... 1 1.1.2. Why measure glucocorticoid levels in muskoxen? ...... 4 1.1.3. What are the methods for measuring glucocorticoid levels? ...... 5 1.1.4. Importance of validating glucocorticoid measurement methods ...... 7 1.1.5. Measuring glucocorticoid levels to study the causes and consequences of physiological stress in wildlife and challenges encountered ...... 8 1.1.6. Value of incorporating Indigenous knowledge in endocrinology studies ...... 9 1.2. Thesis overview ...... 10 1.2.1. Study objectives ...... 10 1.2.2. Chapter outline ...... 10 1.2.3. Chapter contributions ...... 12

CHAPTER 2. QIVIUT CORTISOL IN MUSKOXEN AS A POTENTIAL TOOL FOR INFORMING CONSERVATION STRATEGIES ...... 13 2.1. Abstract ...... 14 2.2. Introduction ...... 15 2.3. Material and methods ...... 16 2.3.1. Study area ...... 16 2.3.2. Sample collection ...... 18 2.3.3. Sex determination ...... 19 2.3.4. Sampling procedures ...... 19 2.3.5. Sample preparation ...... 20 2.3.6. Cold wash procedure to remove surface contamination ...... 21 2.3.7. Extraction procedure ...... 21 2.3.8. Sample reconstitution ...... 21 2.3.9. Data processing ...... 22 2.3.10. Statistical analysis ...... 22 2.4. Results ...... 23 2.5. Discussion ...... 25 2.5.1. Findings ...... 25 2.5.2. Study limitations and future considerations ...... 29 2.6. Conclusion ...... 31 2.7. Acknowledgments ...... 31 2.8. Funding ...... 32

CHAPTER 3. DOCUMENTING INDIGENOUS KNOWLEDGE TO IDENTIFY AND UNDERSTAND THE STRESSORS THAT AFFECT MUSKOXEN (OVIBOS MOSCHATUS)33

3.1. Abstract ...... 34 3.2. Introduction ...... 35 3.3. Methods ...... 36 3.3.1. Study area ...... 36 3.3.2. Study overview ...... 37 3.3.3. Interviews ...... 38 3.3.3.1. Format and participant recruitment ...... 38 3.3.3.2. Interviewing process ...... 38 3.3.3.3. Analytical framework ...... 39 3.3.4. Validation process ...... 39 3.4. Results ...... 40 3.4.1. Characteristics of a healthy muskox ...... 40 3.4.2. Positive and negative factors affecting muskoxen ...... 41 3.4.2.1. Physical environment ...... 44 3.4.2.2. Biological environment ...... 46 3.4.2.3. Human/muskox interactions ...... 48 3.4.3. Collaborative re-interpretation of previously published stress results ...... 50 3.4.3.1. Sex differences ...... 50 3.4.3.2. Seasonal differences ...... 51 3.4.3.3. Yearly variations ...... 51 3.5. Discussion ...... 52 3.5.1. Characteristics of a healthy muskox ...... 52 3.5.2. Factors affecting muskoxen ...... 53 3.5.3. Collaborative re-interpretation of previously published stress results ...... 56 3.6. Final note ...... 59 3.7. Acknowledgements ...... 59 3.8. Funding ...... 59

CHAPTER 4. FECAL GLUCOCORTICOID METABOLITES REFLECT HYPOTHALAMIC– PITUITARY–ADRENAL AXIS ACTIVITY IN MUSKOXEN (OVIBOS MOSCHATUS) ...... 60 4.1. Abstract ...... 61 4.2. Introduction ...... 62 4.3. Material and methods ...... 63 4.3.1. Animals ...... 63 4.3.2. ACTH challenges and fecal sampling ...... 63 4.3.2.1. Winter challenge ...... 64 4.3.2.2. Summer challenge ...... 64 4.3.3. Hormone analyses ...... 65 4.4. Results ...... 66 4.4.1. EIA validation ...... 66 4.4.2. FGMs ...... 66 4.5. Discussion ...... 68 4.6. Conclusion ...... 71 4.7. Acknowledgements ...... 72 4.8. Funding ...... 72

CHAPTER 5. QIVIUT CORTISOL REFLECTS HYPOTHALAMO–PITUITARY–ADRENAL AXIS IN MUSKOXEN (OVIBOS MOSCHATUS) ...... 73 5.1. Abstract ...... 74 5.2. Introduction ...... 75 5.3. Material and methods ...... 77 5.3.1. Animals ...... 77 5.3.2. ACTH challenges ...... 77 5.3.3. Hair sampling ...... 78 5.3.3.1. Experiment 1 ...... 78 5.3.3.2. Experiment 2 ...... 78

5.3.4. Fecal sampling ...... 81 5.3.5. Immobilization procedures ...... 81 5.3.6. Hormone analyses and enzyme immunoassay validation ...... 81 5.3.6.1. Qiviut cortisol ...... 81 5.3.6.2. FGMs ...... 82 5.3.7. Statistical analyses ...... 83 5.3.7.1. Qiviut cortisol ...... 83 5.3.7.2. FGMs ...... 84 5.4. Results ...... 84 5.4.1. Experiment 1 – Are single and repeated ACTH injections reflected in qiviut in the absence of hair growth? ...... 84 5.4.2. Experiment 2 ...... 85 5.4.2.1. Are repeated ACTH injections reflected in qiviut when the hair is growing? ...... 85 5.4.2.2. Do qiviut cortisol levels differ between the neck, shoulder, and rump? ...... 86 5.4.2.3. Does the cortisol concentration in a specific segment remain the same over time? ...... 87 5.4.2.4. Is the cortisol concentration in shed qiviut the same as prior to shedding? ...... 88 5.4.2.5. Pre-injection FGM levels ...... 89 5.5. Discussion ...... 89 5.5.1. Response to ACTH in growing qiviut but not in the absence of growth ...... 90 5.5.2. Sources of qiviut cortisol variability and implications ...... 91 5.5.3. Sources of qiviut cortisol variability and implications ...... 92 5.6. Conclusion ...... 93 5.7. Acknowledgements ...... 94 5.8. Funding ...... 94

CHAPTER 6. INTRINSIC AND EXTRINSIC FACTORS ASSOCIATED WITH INCREASED QIVIUT CORTISOL IN WILD MUSKOXEN (OVIBOS MOSCHATUS) ...... 95 6.1. Abstract ...... 96 6.2. Introduction ...... 97 6.3. Material and methods ...... 98 6.3.1. Animals and sampling procedure ...... 98 6.3.2. Sample analyses ...... 103 6.3.2.1. Parasitology ...... 103 6.3.2.2. Hormone analyses ...... 104 6.3.2.3. Serology ...... 104 6.3.2.4. Marrow fat measurement and other indices of body condition ...... 105 6.3.2.5. Lower jaw analyses ...... 105 6.3.3. Statistical analyses ...... 105 6.4. Results ...... 107 6.5. Discussion ...... 107 6.6. Conclusion ...... 118 6.7. Acknowledgements ...... 119 6.8. Funding ...... 119

CHAPTER 7. CONCLUSIONS: SUMMARY AND FUTURE DIRECTIONS ...... 120 7.1. Main findings ...... 120 7.2. Study limitations, remaining knowledge gaps, and next steps ...... 123 7.3. Future research directions ...... 124 7.4. Final note ...... 126

References ...... 128

APPENDIX A. Relationship between qiviut cortisol and fecal glucocorticoid metabolites ...... 152 APPENDIX B. Longitudinal study of the fecal glucocorticoid metabolite response to repeated adrenocorticotropic hormone/saline injections ...... 155

vii

APPENDIX C. Comparison and main differences between the two methods used to measure qiviut cortisol levels ...... 161 APPENDIX D. Chapter 2 Supplementary material ...... 162 APPENDIX E. Summary of pharmacological challenges done in other wild and domestic even-toed ungulate species to validate the use of fecal glucocorticoid metabolite levels as a biomarker of hypothalamic–pituitary–adrenal axis activity ...... 164 APPENDIX F. Results from the analytical validations of the cortisol and corticosterone enzyme immunoassays ...... 169 APPENDIX G. Fecal cortisol results ...... 173 APPENDIX H. Summary of the studies measuring intestinal transit times in muskoxen ...... 175 APPENDIX I. Identification, sex, age, and experimental group of the animals included in Experiments 1 and 2 ...... 176 APPENDIX J. Summary of the pharmacological challenges carried out in wild and domestic mammalian species to evaluate whether hypothalamic–pituitary–adrenal axis activity is reflected in the hair (adapted from Koren et al., 2019) ...... 177 APPENDIX K. Results from the various cortisol enzyme immunoassay analytical validations ...... 182 APPENDIX L. Boxplots showing the coefficients of variation for the various qiviut sample groups ... 185 APPENDIX M. Details of the linear mixed models with qiviut cortisol as the dependent variable and coefficient estimates ...... 188 APPENDIX N. Details of the linear mixed models with quadruplicate coefficient of variation as the dependent variable and coefficient estimates ...... 193 APPENDIX O. Details of the linear mixed models with fecal glucocorticoid metabolites as the dependent variable and coefficient estimates ...... 194 APPENDIX P. Individual qiviut cortisol line-plots ...... 195 APPENDIX Q. Summary of the various studies evaluating differences in hair glucocorticoids among body regions in wild or domestic mammalian species ...... 197 APPENDIX R. Instructions provided to the hunters with the kits to collect the samples ...... 201 APPENDIX S. Information form provided to the hunters with the kits ...... 202 APPENDIX T. Investigation of biologically and ecologically plausible two-way interactions ...... 203 APPENDIX U. Model selection procedure by manual backward stepwise elimination ...... 204 APPENDIX V. Parameter trace plots ...... 205 APPENDIX W. Manuscript based thesis – copyright disclosure statement ...... 206

viii

List of Tables

Table 2.1. Median and range of qiviut cortisol levels (pg/mg) in hunter-harvested muskoxen from Nunavut and the Northwest Territories represented by location, season, and year of collection, and sex of the animal (n = sample size)...... 17 Table 2.2. Comparison of linear mixed-effect models including location as a random effect, with their corresponding AICc, ΔAICc in comparison to the best-fit model (bold), and degrees of freedom (DF)...... 24 Table 2.3. Ranges of hair cortisol concentrations previously determined in free-ranging or captive wild mammal species using ELISAs (n = sample size)...... 26 Table 4.1. Identification (ID), sex, age, and experimental group of the animals included in the winter and summer ACTH challenges...... 64 Table 4.2. Maximal percentage increase in fecal corticosterone (%) as compared to time 0 levels for the muskoxen given a single injection of ACTH or saline (control) during the winter (ACTH dose 1 IU/kg) and/or summer (ACTH dose 2 IU/kg) and the respective times post-injection (h) at which it was observed...... 67

Table 5.1. Median (range) rump qiviut cortisol levels (ng/g) in muskoxen pre- (t0) and post-administration (1 week [t0 + 1W] after a single injection or 1 week after termination of 5 weekly injections [t0 + 5W]) of saline (control) or 1 IU/kg ACTH during the winter (non-growing qiviut). There were no significant differences between sampling times within each experimental group...... 84

Table 5.2. Median (range) qiviut cortisol levels (ng/g) in muskoxen pre- (t0) and post-administration (2 weeks after termination of 5 weekly injections [t0 + 6W]) of saline (control, n = 6) or 2 IU/kg ACTH (n = 10) and percentage increase (%) in qiviut cortisol between pre-and post-administration during the summer (growing qiviut) by body region. Different letter subscripts indicate significant differences within each experimental group and body region...... 86 Table 5.3. Median (range) cortisol levels (ng/g) in the rump segments of qiviut grown during the summer ACTH challenge (between t0 and t0 + 6W) when collected 6 weeks [t0 + 6W], 3 months [t0 + 3M], and 6 months [t0 + 6M] after the start of the challenge. Different letter subscripts indicate significant differences within each experimental group...... 87 Table 5.4. Median (range) FGM levels (ng/g) measured the day before each of 5 weekly injections...... 89 Table 6.1. Variables, and their respective abbreviations, descriptions, and sample sizes, evaluated as potential predictors of qiviut cortisol levels in the n = 211 muskoxen. None of the continuous variables were normally distributed, so all are summarized as median (range). Missing values (i.e., insufficient sample for all laboratory analyses or information not recorded on data sheet), except those for back fat thickness and incisor breakage score, were estimated through the Bayesian analyses...... 101 Table 6.2. Final model parameter posterior estimates (median and 95% credible intervals (CrI)). The ‘-‘ indicates the reference group for each categorical variable...... 108

APPENDICES

Table B. 1. Identification (ID), sex, age, and experimental group of the animals included in the repeated pharmacological challenge...... 155 Table B. 2. Median (range) fecal corticosterone (ng/g wet feces) following each of five weekly injections...... 157 Table B. 3. Comparison of the linear mixed-effect models fit for fecal corticosterone with animal identity as a random effect, and all possible combinations of explanatory variables (experimental group (group), time of fecal sample collection (time_post_inj), and injection number (injection)) and two-by-

two and three-way interactions, with their corresponding AICc, ΔAICc in comparison to the best- fit model (bold), and degrees of freedom (DF)...... 157 Table B. 4. Comparison of the back-transformed coefficients (coef) and their associated 95% confidence intervals (CIs) for the top three models explaining fecal corticosterone and including as explanatory variables: (i) experimental group (group; the reference category is control), injection number (injection), and their interaction term, (ii) experimental group and injection number, or (iii) injection number only...... 158 Table B. 5. Comparison of the back-transformed coefficients and their associated 95% CIs for the best- fit model explaining fecal corticosterone fit with and without the injection n°4 data points of the 24 and 48 h sampling for the muskoxen that were immobilized at that time (n = 3/7 ACTH- injected and n = 1/3 control)...... 158 Table C. 1. Main differences between the EIA and LC-MS/MS procedures...... 161 Table D. 1. Parameter estimates of the final model explaining log-transformed cortisol levels and including sex (male versus female), year (2014, 2015, and 2016 versus 2013), and season (summer and winter versus fall) as fixed explanatory variables...... 162 Table D. 2. Estimated mean (95% CI) qiviut cortisol levels per sex, season, and year*...... 162 Table D. 3. Sample size description per season, year, and sex...... 162 Table F. 1. Polyclonal cortisol antibody R4866 cross-reactions (C. Munro, personal communication, 2010 (deceased in 2013)) ...... 169 Table F. 2. Polyclonal corticosterone antibody CJM006 cross-reactions (C. Munro, personal communication, 2010 (deceased in 2013)) ...... 170 Table G. 1. Maximal percentage increase in fecal cortisol (%) as compared to time 0 levels for the muskoxen given a single injection of ACTH or saline (control) during the winter (ACTH dose 1 IU/kg) and/or summer (ACTH dose 2 IU/kg) and the respective times post-injection (h) at which it was observed...... 173 Table K. 1. Polyclonal cortisol antibody R4866 cross-reactions (C. Munro, personal communication, 2010 (deceased in 2013))...... 183

List of Figures and Illustrations

Figure 1.1. Female muskox with a calf at the time of qiviut shedding (note the large tufts of qiviut hanging from the guard hairs; photo credit: Juliette Di Francesco) ...... 7 Figure 2.1. Map showing the location of the different communities from which muskox samples were obtained, and the geo-referenced hunting locations of the animals when available. Specific hunting location data were unavailable for muskoxen hunted in Ulukhaktok, Paulatuk, and for 17 of the 80 animals hunted in Cambridge Bay (map generated in QGIS version 2.8.9)...... 18 Figure 2.2. The annual cycle, showing the definition of winter, summer, and fall seasons used in this study; the period of qiviut growth from early April to the end of November, and; the timing of qiviut collection for samples used in this study. The sample size (number of individuals) is indicated by the size of the circle, while the color of the circle indicates the certainty in the date of sample collection (green: accurate to the day; yellow: accurate to the month; red: accurate to the season)...... 19 Figure 2.3. Boxplot showing combined qiviut cortisol values from all animals, seasons, years, sexes, and locations by sex (a), season (b), and year (c). The thick horizontal lines correspond to the medians, the triangles to the means, and the empty circles to the outliers...... 23 Figure 3.1. Map showing the communities of Kugluktuk, Ekaluktutiak, and Ulukhaktok (NWT) and the town of Yellowknife (NWT)...... 37 Figure 3.2. Factors affecting muskoxen negatively (red), positively (yellow) or both negatively and positively (orange) (muskox drawing by Jayninn Yue)...... 42 Figure 3.3. Yearly calendars summarizing interviews and validation sessions showing the timing: of key life events (a); and of factors that negatively affect muskoxen within their physical (b) and biological (c) environments. The four-color gradient represents the percentage of groups that had indicated the month during the interviews and was built as follows: 0% = white, 0% < light color ≤ 33.33%, 33.33% < mid-tone color ≤ 66.67%, and 66.67% < dark color ≤ 100%. If all groups indicated the same information, then only one color instead of three was used (i.e., corresponding to 100%). 43 Figure 4.1. Individual fecal corticosterone levels as a function of the time following a single injection of ACTH or saline (control) for the muskoxen sampled during the winter (ACTH dose 1 IU/kg – n = 6 ACTH-injected) and/or summer (ACTH dose 2 IU/kg – n = 5 ACTH-injected and n = 2 controls). Winter data are indicated as grey lines. Data for the ACTH-injected and control animals during the summer challenge correspond to the black and red lines, respectively...... 68 Figure 5.1. Summary of the possible sources of deposition and loss of GCs in hair (adapted from Henderson, 1993; Meyer and Novak, 2012; Sharpley et al., 2012). Glucocorticoids are thought to be deposited during the period of active hair growth at the level of the hair follicle as they diffuse directly from the blood vessel supplying the follicle (Meyer and Novak, 2012; Burnard et al., 2017; Kapoor et al., 2018). Hair GCs may also be derived from local synthesis in the skin and within the hair follicle by a functional equivalent of the HPA axis (Ito et al., 2005; Slominski et al., 2007; Keckeis et al., 2012). The connection between the local and central HPA axes, the extent to which local GC production is centrally or locally induced, and the relative contributions of locally produced and systemic GCs to “internal” hair concentrations remain unclear (Sharpley et al., 2012; Skobowiat and Slominski, 2015; Salaberger et al., 2016). Glucocorticoids originating from the blood, local production, or both may also be taken up by cells of the sebaceous and apocrine glands and be deposited via sebum and sweat onto the outer cuticle of the hair shaft (Meyer and Novak, 2012; Burnard et al., 2017). This contributes to the “external” GCs on the surface of the hair, which can be augmented by additional GCs deposited from extrinsic sources, such as saliva, urine, or feces. The washing step in most hair GC quantification processes aims to remove any potential “external” GCs (see Koren et al., 2019). Some studies suggest that “external” GCs may be incorporated into the hair shaft, and may contribute to the “internal” GC concentrations (Macbeth, 2013; Cattet et al., 2014; Russell et al., 2014; Otten et al., 2020). The magnitude of this phenomenon remains unclear, but it may be facilitated by moisture (Macbeth et al., 2010; Cattet et al., 2014) and

damage to the shaft, which may render the cuticle more permeable (Heimbürge et al., 2020a). “Washout” of GCs (i.e., leaching out of the hair) due to weather exposure (Heimbürge et al., 2019), UV radiation (Wester, 2016), or grooming (Acker et al., 2018) may also be possible and GC molecules may move along the hair shaft after being deposited even though studies have shown conflicting results (Carlitz et al., 2014; Kapoor et al., 2018; Heimbürge et al., 2020b)...... 76 Figure 5.2. Diagram showing the qiviut samples collected during Experiment 2 and the corresponding research questions. Not all photos were taken from the same animal. Photo credits: Juliette Di Francesco and Morgan Mouton and muskox drawing by Jayninn Yue...... 80

Figure 5.3. Rump qiviut cortisol levels in muskoxen pre- (t0) and post-administration (1 week [t0 + 1W] after a single injection or 1 week after termination of 5 weekly injections [t0 + 5W]) of saline (control, n = 5) or 1 IU/kg ACTH (n = 10) during the winter (non-growing qiviut). The triangles correspond to the means and the thick horizontal lines to the medians. Each thin dotted line indicates the values for an individual muskox and illustrates intra- and inter-individual variability...... 85

Figure 5.4. Qiviut cortisol levels in muskoxen pre- (t0) and post-administration (2 weeks after termination of 5 weekly injections [t0 + 6W]) of saline (control, n = 6) or 2 IU/kg ACTH (n = 10) during the summer (growing qiviut). The triangles correspond to the means and the thick horizontal lines to the medians. Each thin dotted line indicates the values for an individual muskox and illustrates inter-individual variability...... 86

Figure 5.5. Qiviut cortisol levels in all muskoxen (n = 16) pre-challenge (t0) by body region. The triangles correspond to the means and the thick horizontal lines to the medians. Each thin dotted line indicates the values for an individual muskox and illustrates intra- and inter-individual variability...... 87 Figure 5.6. Cortisol levels in the rump segments of qiviut grown during the summer ACTH challenge (between t0 and t0 + 6W) when collected 6 weeks [t0 + 6W], 3 months [t0 + 3M], and 6 months [t0 + 6M] after the start of the challenge in the control (n = 6) and ACTH-injected (n = 10) muskoxen. The triangles correspond to the means and the thick horizontal lines to the medians. Each thin dotted line indicates the values for an individual muskox and illustrates intra- and inter-individual variability...... 88 Figure 5.7. Cortisol levels in rump qiviut grown during the entire hair growth period when shaved in February 2019 and collected shed during qiviut combing in April-May 2019 in all muskoxen (n = 16). The triangles correspond to the means and the thick horizontal lines to the medians. Each thin dotted line indicates the values for an individual muskox and illustrates inter-individual variability...... 89 Figure 6.1. Map showing the five specific geographical locations from which muskox sampling kits were obtained (communities of Ulukhaktok, Kugluktuk, and Ekaluktutiak (black and white stars), and Lady Franklin Point and Kent Peninsula (black arrows)), with the geo-referenced harvesting locations of the muskoxen when available (blue points). Geographic coordinates were unavailable for 15 of the 211 muskoxen included in the statistical analyses; these animals were assigned to a specific geographical location based on the muskox management zone in which they were harvested and on the community from which the kit was submitted (Ekaluktutiak: n = 8; Kugluktuk: n = 2; Ulukhaktok: n = 3; Kent Peninsula: n = 2) (map generated in QGIS version 2.8.9)...... 100 Figure 6.2. Percentage of missing values for each variable (a) and missing data by animal (b)...... 103 Figure 6.3. Marginal posterior distribution of the parameters. Y axes correspond to the density and X axes to the parameters. The peak of each distribution corresponds to the most likely parameter estimate, its spread corresponds to uncertainty about the parameter estimate...... 109 Figure 6.4. Relative qiviut cortisol levels of male and female muskoxen in fall-early winter and mid-late winter. The dots show the posterior medians (the most likely estimates) and the lines the 95% credible intervals (the uncertainty about the estimates) for mean qiviut cortisol by sex and season. All other categorical variables were fixed at the reference group and continuous variables at the

xii

median. That is, the qiviut cortisol values predicted are for muskoxen from east mainland with qiviut grown in 2015, a metatarsus marrow fat of 0.91, and an Umingmakstrongylus pallikuukensis larval count of 43.08 lpg...... 110 Figure 6.5. Effect of Umingmakstrongylus pallikuukensis (Up) larval counts on qiviut cortisol levels in the three broad geographical locations. Solid lines show the posterior medians (the most likely estimates) and dashed lines the 95% credible intervals (the uncertainty about the estimates) for mean qiviut cortisol by location and Up larval counts. All other categorical variables were fixed at the reference group and continuous variables at the median. That is, the qiviut cortisol values predicted are for female muskoxen sampled in late fall-early winter with qiviut grown in 2015 and a metatarsus marrow fat of 0.91. The x-axis was truncated at the 3rd quartile of Up larval counts (maximum count = 1,669 lpg; Table 6.2)...... 111 Figure 6.6. Histogram of Umingmakstrongylus pallikuukensis (Up) larval counts by location. Red dotted lines indicate the medians and blue dotted lines indicate the means...... 112 Figure 7.1. Framework for the inclusion of Indigenous knowledge (IK) in wildlife endocrinology studies. Steps in dark red are those that can be informed by both IK and scientific knowledge (SK), whereas those in blue are mainly informed by SK. The interpretation of results and identification of stressors and influential factors may be an iterative process (thin arrow)...... 122 Figure 7.2. Intrinsic and extrinsic factors that may influence glucocorticoid (GC) secretion and/or may be affected by GCs...... 125

APPENDICES

Figure B. 1. Boxplots showing by experimental group (ACTH-injected – n = 7 and control – n = 3) the fecal corticosterone levels at each ACTH/saline injection and for each time of fecal sample collection...... 159 Figure D. 1. Plot showing the residuals against the fitted values...... 163 Figure D. 2. Plots showing (a) the conditional modes for the random effect level (i.e., estimated mean difference from the mean, conditional on the fixed-effects) for each location of sampling and (b) the conditional modes and the residuals on the same scale...... 163 Figure F. 1. Serial dilutions showing parallel displacement with the standard curve for the cortisol antibody...... 170 Figure F. 2. Serial dilutions showing parallel displacement with the standard curve for the corticosterone antibody...... 171 Figure F. 3. Recovery of exogenous cortisol added to a pooled muskox fecal extract. Samples were prepared as per protocol and spiked with cortisol standard at increasing concentrations...... 171 Figure F. 4. Recovery of exogenous corticosterone added to a pooled muskox fecal extract. Samples were prepared as per protocol and spiked with corticosterone standard at increasing concentrations...... 172 Figure G. 1. Individual fecal cortisol levels as a function of the time following a single injection of ACTH or saline (control) for the muskoxen sampled during the winter (ACTH dose 1 IU/kg – n = 6 ACTH-injected) and/or summer (ACTH dose 2 IU/kg – n = 5 ACTH-injected and n = 2 controls). Winter data are indicated as grey lines. Data for the ACTH-injected and control animals during the summer challenge correspond to the black and red lines, respectively...... 174 Figure K. 1. Extraction mass-dose response. Samples were processed as per protocol described in Chapter 5...... 182 Figure K. 2. Recovery of exogenous cortisol added to a pooled muskox qiviut extract. Samples were prepared as per protocol and spiked with cortisol standard at increasing concentrations...... 184 Figure K. 3. Serial dilutions showing parallel displacement with the standard curve...... 184

xiii

Figure L. 1. Boxplot of intra-sample variation (duplicate CV) pre- (t0) and post-administration (1 week [t0 + 1W] after a single injection or 1 week after termination of 5 weekly injections [t0 + 5W]) of saline (control, n = 5) or ACTH (treatment, n = 10) during the winter (non-growing qiviut, Experiment 1). The thick horizontal lines correspond to the medians, the triangles to the means and the empty circles to the outliers...... 185

Figure L. 2. Boxplot of intra-sample variation (quadruplicate CV) pre- (t0) and post-administration (2 weeks after termination of 5 weekly injections [t0 + 6W]) of saline (control, n = 6) or ACTH (n = 10) during the summer (growing qiviut, Experiment 2) by body region. The thick horizontal lines correspond to the medians, the triangles to the means and the empty circles to the outliers...... 186 Figure L. 3. Boxplot of intra-sample variation (quadruplicate CV) in the rump segments of qiviut grown during the summer ACTH challenge (between t0 and t0 + 6W) when collected 6 weeks [t0 + 6W], 3 months [t0 + 3M], and 6 months [t0 + 6M] after the start of the challenge in the control (n = 6) and ACTH-injected (n = 10) animals. The thick horizontal lines correspond to the medians, the triangles to the means and the empty circles to the outliers...... 187 Figure L. 4. Boxplot of intra-sample variation (duplicate CV) in rump qiviut grown during the entire hair growth period when shaved in February 2019 and collected shed during qiviut combing in April- May 2019 in all muskoxen (n = 16). The thick horizontal lines correspond to the medians, the triangles to the means and the empty circles to the outliers...... 187 Figure P. 1. Line plots showing, the neck (a), shoulder (b), and rump (c) qiviut cortisol levels for each individual at t0 and t0 + 6W in the control (n = 6) and ACTH-injected animals (n = 10). Detailed information regarding the animals can be found in Appendix I...... 195 Figure P. 2. Line plots showing the neck, shoulder, and rump qiviut cortisol levels for each individual at t0. Detailed information regarding the animals can be found in Appendix I...... 196

xiv

List of Symbols, Abbreviations, and Nomenclature

Abbreviation Definition ACTH Adrenocorticotropic hormone AICc Akaike Information Criterion for small sample sizes ATV All-terrain vehicle °C Degree Centigrade CA California CI Confidence interval cm Centimeter CrI Credible interval CV Coefficient of variation DF Degrees of Freedom DNA Deoxyribonucleic acid ECCC Environment and Climate Change Canada EIA Enzyme immunoassay ELISA Enzyme-linked immunosorbent assay eV Electron volt FGM Fecal glucocorticoid metabolite FTM Fecal testosterone metabolite g Gravitational force g Gram GC Glucocorticoid GCM Glucocorticoid metabolite h Hour ha Hectare HCC Hair cortisol concentration HGC Hair glucocorticoid HPA Hypothalamic–pituitary–adrenal i-ELISA Indirect enzyme-linked immunosorbent assay IK Indigenous knowledge IM Intramuscular IPA Isopropyl alcohol IS Internal standard IU International units km Kilometer l Liter LC Liquid chromatography

LC–MS/MS Liquid chromatography coupled to tandem mass spectrometry LIA Luminescence immunoassay mg Milligram min Minute ml Milliliter n Sample size NDVI Normalized Difference Vegetation Index ng Nanogram NJ New Jersey NU Nunavut NWT Northwest Territories OD Optical density p P-value pers. comm. Personal communication pers. obs. Personal observation pg Picogram psi Pound per square inch QC Quebec r Pearson’s correlation coefficient rs Spearman’s rank correlation coefficient ® Registered trademark R² Coefficient of determination RIA Radio immunoassay rpm Revolutions per minute σ² Variance s Second SD Standard deviation SK Scientific knowledge SPE Solid phase extraction TIK Traditional Inuit knowledge unpubl. obs. Unpublished observation unpubl. data Unpublished data Up Umingmakstrongylus pallikuukensis USA United States of America UV Ultraviolet µA Microampere µm Micrometer

xvi

V Volt Ve Varestrongylus eleguneniensis v/v Volume per volume % Percent

xvii

CHAPTER 1. INTRODUCTION

1.1. Background information and rationale

1.1.1. Why measure glucocorticoid levels in wildlife?

Climate change is occurring at an unprecedented pace globally and current trends will likely continue or accelerate in the future (IPCC, 2014). Effects of climate change are multiple and include warming land and ocean temperatures, rising sea levels, altered precipitation patterns, an increased frequency of extreme weather events, shrinking ice sheets, and ocean acidification (IPCC, 2013; Taherkhani et al., 2020). All of these modifications are highly impactful on both marine and terrestrial ecosystems worldwide (Ribera d’Alcalà, 2019), causing major shifts in the geographical distribution of plant and animal species and in the interactions among them (Chen et al., 2011), and altering pathogen transmission patterns (Burek et al., 2008; Kutz et al., 2013a, 2014). Other global changes due to anthropogenic activities (i.e., resource extraction, agriculture, tourism, urbanization, etc.) are also increasingly affecting natural ecosystems, through direct and incidental exploitation, growing interactions between wildlife and domestic animal species, as well as habitat fragmentation and alterations (Lebreton, 2011). Large mammalian species (i.e., > 3 kg) may be particularly vulnerable to these human-caused ecological changes, which are likely to continue intensifying as human population growth progresses (Cardillo et al., 2005). Thus, it is becoming critical to understand and monitor the effects that these multiple environmental changes are having on wildlife individuals and populations and to identify which associated stressors (i.e., unpredictable and predictable stimuli from the environment; Romero, 2004) are the most impactful. This is particularly important in the Arctic, which is especially influenced by climate and other anthropogenic changes as average temperatures are rising twice faster than in other areas of the world and human activities are rapidly developing (Post et al., 2013; AMAP, 2017). The theoretical framework developed by McEwen and Wingfield (2003) helps conceptualize the effect of stressors on wildlife (Wikelski and Cooke, 2006; Macbeth and Kutz, 2019). In order to survive, animals must maintain their internal physiological parameters (e.g., body temperature, pH, blood pressure, glucose levels) within a stable range of values. These homeostatic set points may fluctuate depending on a multitude of intrinsic (e.g., reproductive status) and extrinsic factors (e.g., season), a variability which is accounted for with the introduction of the “allostasis” concept or the process of “maintaining stability through change” (McEwen and Wingfield, 2003). Allostasis is achieved through behavioral and physiological changes which allow the acquisition or re-allocation of energy resources (McEwen and Wingfield, 2003). In their framework, McEwen and Wingfield (2003) also introduce two other notions: (i) “allostatic load,” which refers to the energetic requirements associated with the daily and seasonal activities necessary for food acquisition and survival, along with the additional energy needed to cope with unexpected challenges and periodic predictable life history events (e.g., migration, molting); and (ii) “allostatic overload,” a state in which energetic requirements exceed the organism’s available energy (i.e.,

1 that stored or acquired from food), the organism’s ability to deal with added challenges is reduced, and the potential for the occurrence of negative health effects is increased (McEwen and Wingfield, 2003; Busch and Hayward, 2009). Wikelski and Cooke (2006) suggest that allostatic overload may be useful for conservation physiologists to assess as it provides an integrated measure of the multiple natural and anthropogenic environmental stressors that are affecting wild animals. Glucocorticoids (GCs) (i.e., cortisol and/or corticosterone, depending on the species) are the primary effectors of allostasis through their key role in energy regulation (i.e., they are involved in all processes from the acquisition of energy to its storage and subsequent mobilization) (Landys et al., 2006; Busch and Hayward, 2009). In the absence of stressors, GCs circulate in the blood at low (i.e., baseline) levels to help maintain internal physiological parameters, such as glucose and salt levels, within their stable range of values (Sapolsky et al., 2000). They also present ultradian (i.e., hourly), circadian (i.e., daily), and seasonal low to moderate fluctuations within a “modulated range” to deal with the additional energetic requirements associated with various activities and life history events (e.g., feeding, daily activity patterns, migration, molt, reproduction) (Busch and Hayward, 2009). Glucocorticoids are also important mediators of the physiological stress response, which occurs when the organism is exposed to a stressor (e.g., disease, predation, storm) (Landys et al., 2006; Busch and Hayward, 2009). In mammals, the physiological stress response involves the stimulation of the sympathetic nervous system, which leads to the rapid release of catecholamines (epinephrine and norepinephrine) from the adrenal medulla, and the activation of the hypothalamic-pituitary-adrenal (HPA) axis, which results in the secretion of GCs by the adrenal cortex (Box 1.1) and in a “stress-related” increase of GCs in the blood (Landys et al., 2006; Romero and Butler, 2007). This “stress-related” increase is associated with the suppressive effects of GCs, as they mobilize and redirect resources towards activities more directly associated with immediate survival (Romero and Wingfield, 2016). These effects include the temporary inhibition of reproduction, growth, and the immune system; behavior modification, depending on the stressor and environmental context; and an increase in blood glucose concentrations, through an enhanced mobilization of protein and fat reserves and the inhibition of energy storage, such that more glucose becomes available to the tissues (i.e., muscles) involved in responding to the stressor (Sapolsky et al., 2000; Romero and Butler, 2007). Once the stressor has passed, GCs terminate the stress response through negative feedback signals and downregulation of the HPA axis (Box 1.1; Romero and Butler, 2007). An important distinction should be made between acute stressors that last a short period of time (e.g., attack by a predator, severe storm) and more persistent or chronic stressors that occur over long or repeated periods of time, typically lasting weeks to months (e.g., period of reduced food availability, habitat alterations), and encompass the majority of environmental changes and anthropogenic disturbances (Romero, 2004; Dantzer et al., 2014). In the case of acute stressors, the state of allostatic overload is generally either not reached or rapidly corrected (see Macbeth and Kutz, 2019). By contrast, if the stressor is chronic, organisms typically remain in a state of allostatic overload with high GC levels (Landys et al., 2006; Busch and Hayward, 2009). Severe and prolonged chronic stressors may eventually lead to a state of “homeostatic

Box 1.1. Hormonal cascade leading to the release of GCs in response to a stressor

Stressor

Negative feedback Hypothalamus

Arginine Corticotropin vasopressin releasing factor

Pituitary gland

Adrenocorticotropic hormone

Adrenal cortex

Glucocorticoids

failure,” in which GC concentrations collapse and death occurs (e.g., in the case of extreme starvation) (Romero et al., 2009). Chronic stress and its consequences are well-studied in humans and laboratory animals, and chronically stressed individuals typically exhibit a higher cumulative secretion of GCs over time due to increased baseline GC levels, a heightened or decreased response to stressors, and/or dysregulation of the negative feedback system leading to a slower return to baseline (reviewed in Dickens and Romero, 2013). The associated persistently elevated GC levels may have important detrimental effects on the individuals’ health and fitness (i.e., lifetime reproductive success) through physiological costs (e.g., mobilization of protein and fat stores) and negative impacts on the immune and reproductive systems (reviewed in Macbeth and Kutz, 2019). These long-term changes in GC levels would be measurable in individuals before impacts are visible at a population level, and may provide an opportunity to develop conservation initiatives pro-actively (Macbeth and Kutz, 2019). However, recent review studies have drawn attention to the fact that, in free-ranging wildlife, there is no consensus endocrine profile for chronically stressed individuals (Dickens and Romero, 2013) and evidence of the potential adverse effects of chronically elevated GC levels on health and fitness is inconsistent (Bonier et al., 2009a; Busch and Hayward, 2009; Crespi et al., 2013; Beehner and Bergman, 2017). The complex relationship between GC levels and measures of fitness needs to be further investigated, notably by exploring the context-dependency of this relationship and how it may be confounded by environmental co-variance, but also by assessing how GC levels might influence fitness (Dantzer et al., 2016). Glucocorticoids, as primary mediators of allostasis and biomarkers of the physiological stress response, are widely used by conservation biologists, endocrinologists, and ecologists to study the effects of various environmental changes and challenges on wildlife individuals and populations (Busch and Hayward, 2009; Dantzer et al., 2014). Identifying which factors are responsible for increased GC levels and

3 understanding the consequences of elevated GCs on fitness and health may greatly inform management and conservation actions (Beehner and Bergman, 2017; Kumar and Umapathy, 2019). As a side note, while GCs are commonly referred to as “stress hormones,” they are primarily metabolic hormones with a key role in the modulation of energy balance and they represent only one component of the stress response (Beehner and Bergman, 2017; MacDougall-Shackleton et al., 2019). Measuring GC levels, therefore, does not equate to measuring “stress” (MacDougall-Shackleton et al., 2019). However, in this thesis, for simplicity, I will sometimes use GC levels and “stress” interchangeably (e.g., Chapter 3).

1.1.2. Why measure glucocorticoid levels in muskoxen?

Muskoxen (Ovibos moschatus), known as umingmak (“the bearded one”) in Inuinnaqtun, are a singular and emblematic Artiodactyl (i.e., even-toed ungulate) Arctic species belonging to the Bovidae family and Caprinae subfamily. They occupy a wide geographical range throughout the circumpolar Arctic with endemic populations found in Canada and eastern Greenland, while other populations were introduced or re-introduced in parts of Quebec, Alaska, Russia, Western Greenland, Norway, and Sweden during the 20th century (Cuyler et al., 2019). Muskoxen are non-migratory and feed on grasses, sedges, and willows, depending on the season and forage available (Gunn and Adamczewski, 2003). They are an essential part of the northern ecosystems, where they have a strong economic, nutritional, and sociocultural value for Indigenous communities who have depended on them for generations (Lent, 1999; Tomaselli et al., 2018a). They contribute to regional biodiversity and represent a significant source of income for local communities through handicraft, the wool industry, and tourism because of their popularity in guided hunting (Tomaselli et al., 2018a; Cuyler et al., 2020). Muskoxen are also a nutritious and inexpensive source of meat for Indigenous people who harvest them for subsistence, in addition to playing a key role in community identity and cultural traditions (Lent, 1999; Tomaselli et al., 2018a). In northern Canada, the rate of food insecurity is currently alarmingly high, particularly in Nunavut (NU) where the prevalence of household insecurity was 57.0% in 2017-2018, of which 23.7% was categorized as severe, and 78.7% of the children lived in food insecure households (Tarasuk and Mitchell, 2020). It is, therefore, crucial to conserve muskoxen as a sustainable source of income, as well as a healthy and accessible food source for local communities. However, recent surveys on Banks Island and Victoria Island in the Northwest Territories (NWT) and NU, Canada, along with observations from community members in Ekaluktutiak (previously known as Cambridge Bay, NU), indicate that muskox populations have undergone substantial declines since the early 2000s and these are still ongoing in some areas (Tomaselli et al., 2018b; Cuyler et al., 2019). The cause of these population declines remains unclear, but is likely multifactorial and may be linked to icing events (Nagy and Gunn, 2009; Nagy et al., 2009a), ecological changes associated with climate warming, and an increased susceptibility to diseases and/or changes in exposure to pathogens (Kutz et al., 2017). An opportunistic zoonotic bacterium never reported before in muskoxen or in the Arctic, Erysipelothrix rhusiopathiae, has been identified as the cause of significant widespread mortality events of muskoxen on both islands (Kutz et al., 2015; Mavrot et al., 2020). Orf lesions

4 caused by a zoonotic parapoxvirus and Brucella-like syndromes have been increasingly described by Ekaluktutiak harvesters, and the apparent Brucella seroprevalence has increased in muskoxen on Victoria Island since the early 2000s (Tomaselli et al., 2016b, 2016a, 2019; Fernandez Aguilar et al., unpubl. obs.). Two protostrongylid lungworms, Umingmakstrongylus pallikuukensis and Varestrongylus eleguneniensis, have expanded their range on Victoria Island over the past decade (Kutz et al., 2013a; Kafle, 2018), severe dental anomalies, such as broken incisors, are more frequently observed (Mavrot et al., unpubl. obs.), and trace mineral deficiencies have been highlighted (Mosbacher et al., unpubl. obs.; Gamberg, 2017). Muskox mortality events due to multifactorial causes have also been reported in Norway, where they have been linked to pneumonia outbreaks, probably associated with periods of warm weather and high humidity (Ytrehus et al., 2008, 2015; Handeland et al., 2014), and in Alaska, where the occurrence of various disease syndromes, pathogens, and trace mineral deficiencies has been identified (Afema et al., 2017; Mavrot et al., 2020). These recent population declines and mortality events, along with the very low genetic diversity of muskoxen, suggest that they may be particularly threatened by the multiple environmental changes and associated stressors to which they are increasingly exposed in the Arctic (Kutz et al., 2017; Cuyler et al., 2019; Prewer et al., 2020). It is consequently becoming urgent to develop reliable tools to monitor GC levels in muskoxen, in order to study the effects that ecological changes are having on individuals and populations, and to identify which factors may be affecting muskoxen the most.

1.1.3. What are the methods for measuring glucocorticoid levels?

Glucocorticoids circulate in the blood and, because of their lipophilic nature, are primarily transported bound to corticosteroid binding globulins (CBGs) (Romero and Butler, 2007). The amount of GCs that diffuse out of the capillaries and reach the tissues is determined by the concentration of unbound (i.e., free) GCs in the blood, which increases dramatically during the stress response (Romero and Butler, 2007). Glucocorticoids and/or their metabolites (i.e., breakdown products) are incorporated and excreted in various biological matrices, including saliva, feces, urine, and hair, from which they can be readily quantified (Sheriff et al., 2011). Each of these matrices differ in the duration and type of GC measurement they provide and present various advantages and disadvantages for use in free-ranging wildlife (Box 1.2). In this thesis, based on the advantages they offer and on the samples already available through the hunter- based muskox health monitoring program, I decided to focus solely on feces and hair, as these are thought to provide an integrated measurement of GCs (i.e., both baseline activity and stress responses) over short- term and longer-term time periods, respectively (Box 1.2). Since hair GC (HGC) concentrations may vary with hair type (guard or undercoat; Macbeth et al., 2010; Dulude-de Broin et al., 2019), I concentrated on qiviut (Figure 1.1), the fine woolly undercoat of muskoxen, as it presents the advantages of growing during a defined time period (i.e., between early April and late November) and being shed annually, with shed qiviut allowing for non-invasive collection directly on the tundra (Flood et al., 1989).

Box 1.2. Advantages and drawbacks of the main matrices used to measure GC levels in free- ranging wildlife Plasma and serum were historically the main sample types from which GCs were measured in wild animals and are still commonly and widely used. They provide a measurement of circulating GCs at a single point in time. Additionally, numerous studies have shown that the plasma/serum levels of GCs measured may be biased by the sample collection procedure itself, as GC concentrations increase significantly two to five minutes following capture (e.g., Romero and Reed, 2005). This may particularly be the case in large mammals, such as muskoxen, for which blood sampling requires capture, handling, and physical restraint and/or chemical immobilization, most often occurring after a period of chase (Macbeth and Kutz, 2019). Plasma/serum GC levels will also be affected by natural fluctuations linked to ultradian and circadian rhythms and will generally represent total GC concentrations (i.e., bound to the corticosteroid binding globulin and unbound), unlike the levels measured in the other biological matrices, which reflect free GC concentrations (Russell et al., 2012). As soon as possible upon its collection, blood needs to be centrifuged and its derived plasma/serum frozen to minimize GC degradation (Sheriff et al., 2011). However, once frozen, plasma/serum GCs are stable over time, with no deterioration evidenced in samples stored at -20°C for decades (Stroud et al., 2007). Saliva started being used in the early 1990s to measure GC levels in wild animals (e.g., Dathe et al., 1992). Like blood, it provides a measurement of GCs at a single point in time and levels may be influenced by circadian rhythms (Russell et al., 2012). However, saliva collection is not as invasive and GC levels will generally be less affected by the stress of capture and handling associated with sample collection as they peak approximately 10-30 min following the onset of a stressor, depending on the species (Cook, 2002; Hernandez et al., 2014). While captive animals may successfully be trained to easily collect saliva samples (e.g., Gómez et al., 2004), this is much more difficult in free-ranging wildlife. Studies suggest that GC stability in saliva varies depending on the species. In humans, multiple freeze- thaw cycles and storage at 5°C for up to 3 months or at -20°C for at least one year did not affect GC levels (Garde and Hansen, 2005). By contrast, in dromedary camels, GC levels decreased by 40% within 24 h of storage at 4°C and were affected by multiple freeze-thaw cycles (Majchrzak et al., 2015). The use of excreta (feces and urine) to measure steroid hormone levels in wild animals started to develop in the early 1980s (e.g., Kassam and Lasley, 1981; Loskutoff et al., 1982). Free GCs circulating in the blood are primarily and extensively metabolized by the liver and subsequently excreted in urine and feces (Taylor, 1971). The proportion and structure of the GC metabolites (GCMs) excreted via the feces and urine differ significantly among mammalian species and sometimes between sexes (Palme et al., 2005). Because of the major logistical and technical difficulties associated with the collection of urine samples, this sample type is rarely employed in free-ranging wildlife, even though an important proportion of the GCMs may be excreted via urine (Sheriff et al., 2011). By contrast, feces have since been applied extensively and in a wide range of species as they have the major advantage of being easily and non-invasively collected without the need for any technical expertise or capture and handling of animals (see Palme, 2019). GCMs in urine and feces reflect an integrated average of the circulating free GCs secreted, metabolized, and subsequently eliminated over a species-specific period of time – generally several hours to days – depending on the frequency of urination and defecation, respectively (Palme et al., 1996, 2005; Touma and Palme, 2005). The short and regular pulsatile fluctuations in circulating GCs (i.e., ultradian rhythm) are smoothened when measuring GCMs, but diurnal variations (i.e., circadian rhythm) in GCM excretion generally occur, particularly in smaller species with high frequencies of urination and defecation (Touma and Palme, 2005; Palme, 2019). It is important to collect and freeze fecal samples as quickly as possible following defecation in order to minimize bacterial degradation (Möstl et al., 1999) and alteration of GCMs by environmental conditions (e.g., precipitation and temperature; Washburn and Millspaugh, 2002; Lafferty et al., 2019). Hair was first used to measure GC levels in wildlife in the early 2000s (Koren et al., 2002) and then in human subjects several years later (Sauvé et al., 2007). Unlike excreta, which give a short-term measurement of GC concentrations over hours to days and don’t allow to obtain long-term information without repeated sampling, hair is thought to provide an integrated and longer-term measure of the circulating free GCs secreted and deposited during the period of its growth (Sheriff et al., 2011; Russell et al., 2012). Hair is easily clipped and can be stored at room temperature in envelopes (see Koren et al., 2019). It offers the advantage of less invasive or opportunistic sampling (e.g., collection from harvested animals or mortality sites, hair traps, shed hair; Macbeth and Kutz, 2019) and it seems that GCs are stable in hair during months to years or more (Macbeth6 et al., 2010), as evidenced by their successful quantification from museum specimens and mummies, which provide additional sampling opportunities (Bechshøft et al., 2012; Webb et al., 2015).

2019). It offers the advantage of less invasive or opportunistic sampling (e.g., collection from harvested animals or mortality sites, hair traps, shed hair; Macbeth and Kutz, 2019) and studies indicate that GCs are stable in hair for months to years or more (Macbeth et al., 2010), as evidenced by their successful quantification from museum specimens and mummies, which provide additional sampling opportunities (Bechshøft et al., 2012; Webb et al., 2015).

Figure 1.1. Female muskox with a calf at the time of qiviut shedding (note the large tufts of qiviut hanging from the guard hairs; photo credit: Juliette Di Francesco)

Immunoassays (i.e., radioimmunoassays (RIAs) and enzyme immunoassays (EIAs)), based on an antibody directed against the steroid hormone of interest, are the most commonly used to measure FGM and HGC levels in wildlife (Sheriff et al., 2011; Koren et al., 2019). More complex methods based on liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) have also been increasingly applied over recent years to identify and/or measure HGCs in wild mammals (Kapoor et al., 2016; Weisser et al., 2016; Rakotoniaina et al., 2017). The major advantage they present is to avoid the antibody cross-reactivity issues that can potentially occur with immunoassays (Handelsman and Wartofsky, 2013). In this thesis, I used both LC-MS/MS (Chapter 2) and a cortisol (the main GC in muskoxen; Koren et al., 2012) EIA (Chapters 5 and 6) to measure qiviut cortisol levels.

1.1.4. Importance of validating glucocorticoid measurement methods

Validating GC measurement methods is a key prerequisite to proper results interpretation. Different types of validations should be distinguished. Assay validations ensure that GCs or their metabolites are properly quantified in the species and biological matrix of interest. For EIAs, these include measures of precision (i.e., intra- and inter-assay coefficients of variation (CVs) of low and high concentration standard or pooled samples), specificity (i.e., cross-reactivity of the EIA antibody with other

7 potentially present steroids and parallel displacement of standard and test sample curves), sensitivity (i.e., detection limit of the assay), and accuracy (i.e., amount of matrix interference based on the recovery of standard hormone added over a range of concentrations to a sample extract of known concentration) (Sheriff et al., 2011; Palme, 2019). Equally essential are physiological and/or biological validations, which provide evidence that the levels measured in the species and biological matrix of interest accurately reflect changes in circulating GC levels. While such validations have been done for feces in multiple wildlife species, few have been conducted to date for hair (Koren et al., 2019; Palme, 2019). Physiological validations involve the pharmacological stimulation or suppression of HPA axis activity, followed by the measurement of FGM levels or HGCs to determine whether the anticipated changes are reflected in the feces or hair, respectively. The common and “gold-standard” procedure is to stimulate the production of GCs by the adrenal cortex through the administration of synthetic adrenocorticotropic hormone (ACTH) (Koren et al., 2019; Palme, 2019). Biological validations include measuring FGM levels or HGCs before and after a stressful episode, such as a relocation event (e.g., Yamanashi et al., 2016a), transportation (e.g., Hein et al., 2020), or altered social conditions (e.g., Davenport et al., 2006), or removal from a stressful environment (e.g., Malcolm et al., 2013) (Koren et al., 2019; Palme, 2019).

1.1.5. Measuring glucocorticoid levels to study the causes and consequences of physiological stress in wildlife and challenges encountered

Fecal GC metabolites and HGCs have been widely measured in free-ranging wildlife to investigate potential causes of GC variation, such as anthropogenic activities and disturbances (e.g., tourism (Behie et al., 2010; Zwijacz-Kozica et al., 2013; Rehnus et al., 2014), hunting (Bryan et al., 2015; Santos et al., 2018), poaching (Gobush et al., 2008), urbanization (Scheun et al., 2015; Brunton et al., 2020), logging (Mastromonaco et al., 2014; Ewacha et al., 2017), wind-turbine noise (Agnew et al., 2016), pastoralist activity (Van Meter et al., 2009)), but also predation risk (Creel et al., 2009; Périquet et al., 2017; Lavergne et al., 2020), weather conditions and events (Huber et al., 2003; Cizauskas et al., 2015; Fardi et al., 2018), and pathogens (Goldstein et al., 2005; George et al., 2014; Madslien et al., 2020). More rarely, FGMs and HGCs have also been used to examine the possible consequences of elevated GCs on health and fitness (survival probability (Pride, 2005; Cabezas et al., 2007; Wey et al., 2015; Rakotoniaina et al., 2017), neonate birthweight and pregnancy success (Downs et al., 2018), fetal loss (Beehner et al., 2006), gastro-intestinal (GI) parasite load (Cizauskas et al., 2015), and GI parasite richness and infection presence (Gillespie et al., 2013)). Finally, a small number of studies have attempted to demonstrate the link between a stressor, elevated GCs, and the effect of these hormones on health and fitness (but see Chapman et al., 2015; Dulude-de Broin et al., 2020). Multiple extrinsic (e.g., season, habitat, food availability, social status) and intrinsic (e.g., sex, age, reproductive status, previous experiences) factors may affect baseline GC levels and influence the magnitude of the response to stressors (Dantzer et al., 2014; Heimbürge et al., 2019; Palme, 2019). Other factors may more specifically influence FGM (e.g., diet) or HGC (e.g., hair color, body location) levels

(Goymann, 2012; Heimbürge et al., 2019). These sources of variation need to be identified in the species of interest and taken into account when designing studies and interpreting FGM and HGC levels. Controlling for these variables in the analyses allows for an accurate characterization and/or monitoring of responses to various environmental changes and disturbances, as well as to possible management actions (Baker et al., 2013). Interpretation of hair GC levels is associated with its own additional challenges, as the use of this matrix is still recent (see Box 1.2), and mechanisms of GC deposition in hair are highly complex and remain poorly understood. Hair GC levels may reflect not only concentrations of GCs over the period of the hair’s growth as these hormones are incorporated inside the shaft from both local (i.e., skin and hair follicles) and systemic (i.e., blood) sources, but also more current and short-term levels from glandular or extrinsic sources (i.e., urine, saliva or fecal contamination) (see Chapter 5 for details; Kalliokoski et al., 2019; Koren et al., 2019). This emphasizes the need for thorough physiological and/or biological validations, which remain currently scarce, particularly in wild species (reviewed in Koren et al., 2019).

1.1.6. Value of incorporating Indigenous knowledge in endocrinology studies

The value of bringing together scientific knowledge (SK) and Indigenous knowledge (IK) is increasingly recognized in the fields of wildlife health and ecology (Huntington, 2011; Kutz and Tomaselli, 2019). Data collection for scientific knowledge is generally driven by specific research objectives and done in a systematic manner, but tends to be limited in time and space, especially in remote settings such as the Arctic. The use of SK alone is, therefore, often associated with fragmented data and field observations, which may be challenging to interpret (Lubin and Massom, 2006; Kutz and Tomaselli, 2019). Indigenous knowledge is defined as "a cumulative body of knowledge and beliefs handed down through generations by cultural transmission about the relationship of living beings, (including humans) with one another and with their environment" (Gadgil et al., 1993). Bridging SK and IK allows to circumvent some of the disadvantages encountered by the use of SK alone, and to gain a broader, more holistic understanding of ecological systems and how they are affected by environmental changes (Huntington, 2011; Kutz and Tomaselli, 2019). The knowledge of Sámi reindeer herders was, for example, documented in Sweden and Norway to characterize snow types and profiles, to gain a deeper comprehension of how reindeer (Rangifer tarandus tarandus) are affected by various snow conditions, and to evaluate the possible impacts that long- term changes in snow and ice conditions will have on reindeer herding (Riseth et al., 2011). Similarly, the knowledge from local Elders and hunters in the Northern Eeyou Marine Region (Quebec, Canada) provided important insights on the distribution of polar bears (Ursus maritimus), as well as on their denning activity, foraging habits and terrestrial habitat use (Laforest et al., 2018). The knowledge of Dene people was also collected concomitantly with genetic data to comprehensively characterize caribou (Rangifer tarandus granti) types in the Sahtu Region (NWT, Canada) and generate insights into the evolutionary histories that likely contributed to such population differentiation (Polfus et al., 2016).

Indigenous communities in the Arctic and elsewhere experience and witness first-hand the impacts of climate and other anthropogenic changes on the environment. Indigenous knowledge holders are, consequently, arguably the best placed to identify the factors that may be impacting local species and their effects, and to comprehend the patterns observed in GC levels. Incorporating IK in endocrinology studies investigating the causes of physiological stress is, therefore, extremely relevant, but has largely been overlooked.

1.2. Thesis overview

1.2.1. Study objectives

Fecal GC metabolite and HGC levels open exciting perspectives for monitoring the effects that environmental changes are having on muskox individuals and populations, identifying which stressors may be the most impactful, and ultimately informing conservation strategies for this species. However, before they are used as such tools, FGM and HGC levels need to be thoroughly validated as biomarkers of physiological stress and potential sources of variation in GC levels identified. The overarching goal of this PhD research was to establish FGM and qiviut cortisol levels as reliable biomarkers of physiological stress in muskoxen, and to subsequently apply these tools in combination with documenting IK to explore potential causes and patterns of physiological stress in wild muskoxen. More specifically, the main objectives of this study were to: 1. Pharmacologically validate the use of qiviut cortisol and FGM levels as biomarkers of HPA axis activity in muskoxen. 2. Document IK to determine which factors are influencing muskoxen in their rapidly changing environment and to improve our understanding of the sex, seasonal, and annual patterns observed in qiviut cortisol levels. 3. Establish baseline data, describe the natural variability of qiviut cortisol levels in wild muskoxen, and assess the effects of various extrinsic and intrinsic factors on qiviut cortisol levels.

1.2.2. Chapter outline

Chapter 2 presents the development of a method to measure qiviut cortisol levels using LC- MS/MS, along with its application to assess inter-individual variability and the effect of several extrinsic (i.e., year and season) and intrinsic (i.e., sex) factors on qiviut cortisol levels. This work also served to establish baseline longitudinal data on the qiviut cortisol levels of wild muskoxen in different geographical locations. Such baseline data are a key prerequisite to monitor the potential changes in qiviut cortisol levels that may occur within muskox populations, and to subsequently identify causative factors. Chapter 3 summarizes the findings from small group interviews and validation sessions that were done in Kugluktuk, NU to identify and gain a better understanding of the stressors that potentially affect muskoxen. More specifically, IK was documented on the characteristics used by harvesters to identify whether a muskox is healthy and on the factors that influence muskoxen in a positive and/or negative way,

10 when they occur throughout the year, and how they have changed over time. Results from Chapter 2 were also re-interpreted with IK holders to improve our understanding of the sex, seasonal, and annual patterns observed in qiviut cortisol levels through the identification of some of their potential underlying explanations. Danielsen et al., (2009) proposed a five-category classification of natural resource monitoring programs based on the degree of community involvement. This work represents a meaningful advancement in the process of transitioning the muskox health monitoring program, which depends in large part on hunter-based sampling, to the “collaborative monitoring with local data interpretation” category by actively involving communities not only in data collection, but also at other steps of the research such as data interpretation. It also shows how studies in the complex field of wildlife endocrinology, particularly those involving species found exclusively in highly remote settings, can benefit from involving IK holders in data interpretation. Chapters 4 and 5 present the pharmacological and EIA validations of the methods used to quantify GCs and their metabolites in the feces and qiviut of muskoxen, respectively. For this, two repeated ACTH challenges, involving weekly administrations of saline (control group) or ACTH over five consecutive weeks, were done on captive muskoxen at the University of Alaska Fairbanks, one in winter (no hair growth) and the other in summer (maximal hair growth). These two chapters provide evidence that the FGM and HGC levels measured in muskox feces and qiviut accurately reflect changes in HPA axis activity. They also contribute to furthering our understanding of GC deposition and stability in hair, and of the limitations and challenges associated with FGM and HGC interpretation. This work adds muskoxen to the small number of wildlife species in which HGCs have been properly validated as a biomarker of long- term HPA axis activity. Chapter 6 builds upon Chapter 2 by further identifying the intrinsic (i.e., sex, age, body condition, and incisor breakage) and extrinsic abiotic (i.e., geographical location, season, and year) and biotic (i.e., lungworm and gastro-intestinal parasite infection intensities, parasite richness, bacteria exposure) factors that influence qiviut cortisol levels using the data from the muskox sampling kits collected by hunters from the communities of Kugluktuk, Ekaluktutiak, and Ulukhaktok (NWT) between January 2016 and May 2019. In this work, a Bayesian modelling approach is employed to deal with the issue of multiple missing data. Chapter 7 discusses the limitations of this study, some of the challenges encountered and remaining knowledge gaps, as well as future research directions. Appendix A includes additional work assessing the relationship between qiviut cortisol and FGM levels in muskoxen. Appendix B presents the results from the longitudinal study of the FGM response to the repeated ACTH/saline injections administered during the pharmacological challenge undertaken in the summer of 2018. Appendix C compares and presents the main differences between the two methods used to measure qiviut cortisol levels in this thesis.

1.2.3. Chapter contributions

Chapter 2: SK conceptualized the study. KWE oversaw method development using LC-MS/MS, in which NNG and AC were also actively involved. NNG, L-ML, TD, and JD coordinated the hunter- based sample collection. NNG and JD coordinated the sample analyses and JD also analyzed part of the samples. JD and SP conducted the statistical analyses. JD wrote the manuscript, which was edited by all co- authors. Chapter 3: JD ideated the study and designed it with the guidance of SK and CG. JD conducted the small group interviews and validation sessions with the assistance of AH and TM. JD transcribed the interviews and TM verified the accuracy of all transcripts. JD analyzed the data and developed the coding framework with the help of AH. KAA provided advice and support at all steps of the study. JD summarized the IK documented and wrote the manuscript, which was edited by SK, AH, CG, and L-ML. Chapters 4 and 5: JD initiated the collaboration with JB and JR from the University of Alaska Fairbanks. JD conceptualized the study and designed it with the guidance and contribution of all co-authors. JD collected the hair and fecal samples and sorted the qiviut. GM oversaw method development using the EIAs and supervised sample analyses. JD conducted the statistical analyses and consulted regularly with SC. JD wrote the manuscripts, which were edited by all co-authors. Chapter 6: JD and SK conceptualized the study. JD and FM coordinated hunter-based sample collection in Kugluktuk and Ulukhaktok, respectively. JD coordinated and/or participated in sample analyses. AS performed Erysipelothrix rhusiopathiae serology, JD and AS performed the fecal parasite analyses, JD analyzed the metatarsals, FM and JD analyzed the lower jaws, and GM supervised the FGM and qiviut cortisol analyses. GK and RD conducted the statistical analyses with direction from JD. JD wrote the chapter, which was edited by all co-authors.

CHAPTER 2. QIVIUT CORTISOL IN MUSKOXEN AS A POTENTIAL TOOL FOR INFORMING CONSERVATION STRATEGIES

Juliette Di Francesco1, Nora Navarro-Gonzalez1, Katherine Wynne-Edwards2, Stephanie Peacock3, Lisa- Marie Leclerc4, Matilde Tomaselli1, Tracy Davison5, Anja Carlsson1, Susan Kutz1

1Department of Ecosystem and Public Health, Faculty of Veterinary Medicine, University of Calgary, 3280 Hospital Drive NW, Calgary, Alberta, Canada T2N 4Z6

2Department of Comparative Biology and Experimental Medicine, Faculty of Veterinary Medicine, University of Calgary, 3280 Hospital Drive NW, Calgary, Alberta, Canada T2N 4Z6

3Department of Biological Sciences, Faculty of Science, University of Calgary, 507 Campus Drive NW, Calgary, Alberta, Canada T2N 4V8

4Department of Environment, Government of Nunavut, P.O. Box 377, Kugluktuk, Nunavut, Canada X0B 0E0

5Department of Environment and Natural Resources, Government of the Northwest Territories, P.O. Box 2749, Inuvik, Northwest Territories, Canada X0E 0T0

Manuscript published in the journal Conservation Physiology

Di Francesco J, Navarro-Gonzalez N, Wynne-Edwards K, Peacock S, Leclerc L-M, Tomaselli M, Davison T, Carlsson A, Kutz S (2017) Qiviut cortisol in muskoxen as a potential tool for informing conservation strategies. Conservation Physiology, 5(1):cox052. https://doi.org/10.1093/conphys/cox052

Full length manuscript. This is an open access article distributed under the terms of the Creative Commons CC BY license (https://creativecommons.org/licenses/). This manuscript has been slightly modified from its original version for consistency with the United States English spelling, style, terms, abbreviations, as well as figure and table numbering used throughout the thesis.

2.1. Abstract

Muskoxen (Ovibos moschatus) are increasingly subject to multiple new stressors associated with unprecedented climate change and increased anthropogenic activities across much of their range. Hair may provide a measurement of stress hormones (glucocorticoids) over periods of weeks to months. We developed a reliable method to quantify cortisol in the qiviut (wooly undercoat) of muskoxen using liquid chromatography coupled to tandem mass spectrometry. We then applied this technique to determine the natural variability in qiviut cortisol levels among 150 wild muskoxen, and to assess differences between sexes, seasons and years of collection. Qiviut samples were collected from the rump of adult muskoxen by subsistence and sport hunters in seven different locations in Nunavut and the Northwest Territories between 2013 and 2016. Results showed a high inter-individual variability in qiviut cortisol concentrations, with levels ranging from 3.5 to 48.9 pg/mg (median 11.7 pg/mg). Qiviut cortisol levels were significantly higher in males than females, and varied seasonally (summer levels were significantly lower than in fall and winter), and by year (levels significantly increased from 2013 to 2015). These differences may reflect distinct environmental conditions and the diverse stressors experienced, as well as physiological and/or behavioral characteristics. Quantification of qiviut cortisol may serve as a valuable tool for monitoring health and informing conservation and management efforts.

Key words: Ovibos moschatus, stress, hair, Arctic, liquid chromatography coupled to tandem mass spectrometry

2.2. Introduction

Climate change, taking place at an unprecedented pace in the Arctic, is resulting in multiple new stressors for wildlife (e.g., Kutz et al., 2014), including increased heat stress (Ytrehus et al., 2008, 2015; van Beest and Milner, 2013), a higher frequency of extreme weather events (IPCC, 2013) and changes in exposure to pathogens (Burek et al., 2008; Hoberg et al., 2008; Ytrehus et al., 2008, 2015; Kutz et al., 2013a, 2014). These stressors are having, and will continue to have, important impacts on wildlife (Altizer et al., 2013; Post et al., 2013; Ytrehus et al., 2015). Muskoxen (Ovibos moschatus), large herbivores that reside at Arctic and subarctic latitudes, may be especially vulnerable to ongoing changes in the Arctic (MacCarthy et al., 2001; Weller, 2005; Kutz et al., 2017) as they are exceptionally well adapted to cold environments (Gunn and Adamczewski, 2003), but also have very low genetic diversity (Groves, 1997; MacPhee et al., 2005). Muskoxen are hunted for subsistence by aboriginal communities for whom they are a nutritious and affordable source of food, and serve as a key element in cultural traditions (Nuttall et al., 2005). However, recent surveys in the Canadian North indicate that the two largest muskox populations, those on Banks and Victoria islands, Northwest Territories and Nunavut, have declined substantially, and, in some areas, are still declining (Nagy et al., 2006, 2009a, 2009b; Davison et al., 2013, 2017; Tomaselli et al., 2016b). The cause of these declines remains uncertain, but is likely multifactorial, linked to icing events (Nagy and Gunn, 2009; Nagy et al., 2009a), ecological changes associated with climate warming, and disease emergence (Kutz et al., 2015, 2017; Tomaselli et al., 2016a). Ecological changes (e.g., climate change, habitat loss and fragmentation, fluctuations in food availability, human-caused disturbances, etc.) are increasingly recognized to be associated with chronic stress (chronic implying the stress occurs over long periods of time such as weeks to months), and may in turn lead to reduced health, fitness, and survival in free-ranging wildlife (Bonier et al., 2009b; Busch and Hayward, 2009; Ellis et al., 2012; Koren et al., 2012a). The stress response is mediated by the activation of the hypothalamic–pituitary–adrenal (HPA) axis, which leads to the secretion of glucocorticoids (GCs; mainly corticosterone or cortisol depending on the species) and subsequent mobilization of energy stores in mammals. While the short-term release of GCs plays an important role in allowing animals to cope with environmental change or challenges and to escape from life-threatening situations (Wingfield et al., 1997; McEwen and Wingfield, 2003; Romero, 2004; Busch and Hayward, 2009), chronically elevated levels of GCs have been associated with physiological costs and detrimental effects including: increased susceptibility and vulnerability to diseases, a decline in immune responses, and reduced reproductive success (Wingfield et al., 1997; Moore et al., 2005; Acevedo-Whitehouse and Duffus, 2009; Busch and Hayward, 2009). Hair, through its slow growth, is thought to give an integrated measure of GC concentrations over long periods of time, weeks to months, depending on the species-specific hair turnover rate (Sheriff et al., 2011). Over the past decade, the measurement of GCs in hair and feathers has shown promise as a biomarker of long-term stress in a variety of wild species, in which hair and feather GC concentrations have been associated with important fitness and health characteristics, as well as environmental factors. Hair GC levels, for example, have been negatively associated with proxies of fitness like body condition in polar

15 bears (Ursus maritimus) (Macbeth et al., 2012), and feather GC levels may be promising biomarkers of future survival in wild house sparrows (Passer domesticus) (Koren et al., 2012a). Hair GC levels were also positively associated with certain long-term stressors like high hunting pressure in wolves (Canis lupus) (Bryan et al., 2015), bile collection in Asiatic black bears (Ursus thibetanus) (Malcolm et al., 2013), and low food availability in grizzly bears (Ursus arctos) (Bryan et al., 2013). These studies suggest that hair GC levels may serve as a valuable tool to monitor wildlife health and inform conservation strategies. The fur of muskoxen is one of their distinct features. They possess both a thick undercoat wool called qiviut, that is grown between early April and late November every year and shed in its entirety the following spring (between May and July), and long guard hairs that are produced continuously over several years and form the characteristic “skirt” of muskoxen (Gray, 1987; Flood et al., 1989; Mosbacher et al., 2016). We propose that cortisol levels in qiviut may provide a measure of stress over the course of its growth, which in turn may give quantitative information about the health of the individuals and populations, and how they are affected by ecological changes in the Arctic. In this study, we aimed to: (i) develop a reliable technique for cortisol quantification in the qiviut of muskoxen using liquid chromatography coupled with tandem mass spectrometry (LC–MS/MS); (ii) determine the natural variability in qiviut cortisol among wild muskoxen; and (iii) assess the relationship between qiviut cortisol levels and sex, season and year of collection.

2.3. Material and methods

2.3.1. Study area

Adult muskoxen harvested by subsistence and sport hunters were sampled from January 2013 to August 2016 in the Canadian Arctic near the communities of Paulatuk, Sachs Harbour and Ulukhaktok in the Northwest Territories (NWT), Kugluktuk and Cambridge Bay in Nunavut (NU), and away from communities on the Kent Peninsula and Lady Franklin Point (NU) (Figure 2.1 and Table 2.1). These locations were selected based on traditional muskox harvesting grounds and access to samples through collaborations with local Hunters and Trappers Committees (NWT) or Hunter and Trapper Organizations (NU), and sport hunt outfitters. Muskoxen harvested in Cambridge Bay and Lady Franklin Point belong to the Nunavut management unit MX-07, whereas those harvested in Kugluktuk and on Kent Peninsula are in the management unit MX-11. Each location was, however, treated separately, as they are geographically distant and characterized by different environmental conditions. Muskoxen from Lady Franklin Point and Kugluktuk are hunted by the community of Kugluktuk. Our collaboration with their Hunter and Trapper Organization gave us access to samples from community and subsistence hunts where both males and females are taken. Conversely, muskoxen from the Kent Peninsula and Cambridge Bay areas are hunted by the community of Cambridge Bay, with whom our collaboration gave us access mainly to samples from outfitted sport hunts where males are exclusively harvested. Muskox populations on Banks and Victoria Islands have declined substantially over the last decade, whereas those from the mainland sites have

16 remained stable or increased (Davison and Williams, 2013; Davison et al., 2013, 2017; Davison and Branigan, 2014; Leclerc, 2014; Tomaselli et al., 2016b).

Location Season and year of Females median Males median collection pg/mg (range) pg/mg (range) Cambridge Bay Winter 2014 - 11.3 (6.8–14.2) n = 6 Summer 2014 - 8.2 (3.5–15.3) n = 15 Fall 2014 18.8 13.61 (7.8–48.9) n = 1 n = 21 Winter 2015 - 19.59 (13.9–30.3) n= 10 Fall 2015 12.7 (6.1–18.5) 23.3 (3.6–27.1) n = 5 n = 9 Winter 2016 - 21.4 (11.3–24.6) n = 8 Summer 2016 - 9.24 (7.9–9.7) n = 4 Kent Peninsula Winter 2014 - 10.5 n = 1 Winter 2016 - 15.3 (10.5–23.1) n = 8 Kugluktuk Winter 2014 7.6 (4.3–38.3) 17.2 (6.9–22.5) n = 6 n = 6 Winter 2015 12.2 (9.1–15.7) 21.8 n = 8* n = 1 Lady Franklin Point Winter 2015 17.3 17.3 (13.7–20.9) n = 1 n = 2 Paulatuk Winter 2013 7.73 (5.3–11.4) 5.51 (4.3–6.8) n = 7 n = 2 Sachs Harbour Winter 2013 7.7 (4.2–14.5) 6.8 (12.5–20.2) n = 7† n = 5 Ulukhaktok Fall 2014 9.8 (7.4–15.5) 9.37 (5.6–14.0) n = 12 n = 5 *Six were pregnant. †One was pregnant and three were lactating.

Figure 2.1. Map showing the location of the different communities from which muskox samples were obtained, and the geo-referenced hunting locations of the animals when available. Specific hunting location data were unavailable for muskoxen hunted in Ulukhaktok, Paulatuk, and for 17 of the 80 animals hunted in Cambridge Bay (map generated in QGIS version 2.8.9).

2.3.2. Sample collection

Samples of muskox qiviut were obtained through individual subsistence and community hunts (Kugluktuk, Paulatuk and Sachs Harbour), individual subsistence and sport hunts (Cambridge Bay, Lady Franklin Point and Kent Peninsula), and the qiviut marketing pathway (Ulukhaktok). These are regular activities in the communities and no animals were culled specifically for our study. Samples were obtained under Animal Care and Use Permit #AC13-0121, the Wildlife Research Permit #2013-035, 2014-053, 2015- 068 and 2016-058 for Nunavut, and the Wildlife Research Permit #WL500098, WL500158 and WL500257 for the Northwest Territories. The timing of sample collection was directly linked to the traditional muskox harvesting seasons and samples were classified to different seasons (winter, summer or fall) based on their collection date. For the purposes of this study, samples collected between January and early April were considered winter samples, those collected late July through August were classified as summer, and those collected October through to mid-December as fall (Figure 2.2). Qiviut growth extends from early April to late November (Flood et al., 1989). Thus samples tested in the fall would represent the end of the qiviut growth cycle, those from the summer the middle of the cycle, and the “winter” samples would be a period of no qiviut growth (Figure 2.2).

Figure 2.2. The annual cycle, showing the definition of winter, summer, and fall seasons used in this study; the period of qiviut growth from early April to the end of November, and; the timing of qiviut collection for samples used in this study. The sample size (number of individuals) is indicated by the size of the circle, while the color of the circle indicates the certainty in the date of sample collection (green: accurate to the day; yellow: accurate to the month; red: accurate to the season).

2.3.3. Sex determination

Hunters were asked to record the sex of the harvested muskoxen on a provided form. When this information was left blank or marked “unknown,” a skin sample was tested to determine the sex of the animal using laboratory genetic analyses that were performed within the Kyle Laboratory at Trent University, Peterborough, Ontario, Canada. DNA from each sample was extracted using the Qiagen DNAeasy Blood & Tissue Kit (Qiagen Inc., Mississauga, Ontario) following the recommended protocol supplied by the manufacturer. Sex was determined by amplification of the SRY and ZFX genes using the primers SRY-Y53-3C, SRY-Y53-3D (Fain and LeMay, 1995), ZFX-P2-3EZ and ZFX-P1-5EZ (Aasen and Medrano, 1990).

2.3.4. Sampling procedures

Hunters were directed to sample the qiviut when skinning the animals by cutting a piece of hide measuring approximately 10 × 10 cm from the rump, near, but lateral to the base of the tail. The location

19 of sample collection was standardized as hair cortisol levels may vary depending on the body region (Macbeth et al., 2010; Ashley et al., 2011; Terwissen et al., 2013). Specifically, the rump was chosen because a sample can be taken without causing significant damage to the hide as it is along an edge that is cut during butchering. The sampling location is particularly important to consider when the hides are used by hunters for taxidermy or commercially for the fiber industry as it allows hunters to preserve the economic value. Additionally, guard hairs in this area grow continuously throughout the year (Flood et al., 1989) and have been archived for future studies. Similar analytical methods can be used for guard hair cortisol quantification and guard hairs may also be used to assess the nutritional history of these muskoxen (Mosbacher et al., 2016). Hunters and guides were expressly asked not to contaminate the fur with blood, urine or feces, and samples that did not conform to the standardized sampling location or required quality were removed from this study (n = 4). For the samples from Ulukhaktok, hides were dried under ambient environmental winter conditions in the community, and then shipped to Calgary, Alberta, Canada from where they were exported for qiviut processing. In Calgary, hide samples were collected from the rump area, as per the hunter protocol. All samples were stored at -20°C until laboratory processing.

2.3.5. Sample preparation

Qiviut samples from each animal were obtained by cutting guard hairs and qiviut away from the skin using a scalpel blade. Care was taken not to collect qiviut that was grossly contaminated with dirt, blood, urine or feces and to not disturb the skin or collect hair roots. Guard hairs and qiviut were manually separated using clean forceps. Between 20 and 100 mg of qiviut were used for analysis, and each sample was run as a true experimental duplicate, collecting hair from the hide twice. Fifteen samples from Sachs Harbour were run in duplicate on two separate occasions to assess the repeatability of the method. Muskoxen are cortisol-dominant in serum (Koren et al., 2012c), and initial method development detected extremely low to non-quantifiable levels of corticosterone in qiviut, thus, only cortisol was quantified in this study. Qiviut cortisol was quantified using LC–MS/MS. The use of LC–MS/MS eliminates potential issues of antibody cross-reactivity that are associated with enzyme linked immunosorbent assays (ELISAs) and has become the preferred method of hair steroid analysis in human clinical research (Handelsman and Wartofsky, 2013). It was previously applied by Koren et al. (2012a) to quantify testosterone, corticosterone and cortisol in the feathers of wild house sparrows, by Koren et al. (2012b) to quantify multiple steroids in serum samples from a wide range of captive wild mammals and birds (including muskoxen), and by Gesquiere et al. (2014) to measure testosterone concentrations in fecal samples from wild baboons (Papio cynocephalus) (Koren et al., 2012a, 2012b; Gesquiere et al., 2014). The goal in sample preparation for LC–MS/MS is to provide as clean a steroid extraction as possible. For this reason, sample preparation differed from methods using ELISAs (Koren et al., 2002; Davenport et al., 2006; Macbeth et al., 2010) in two ways: (i) the hair was not ground in a ball mill or cut with scissors and (ii) the sample preparation process was kept cold at all times (no stages in the washing

20 and extraction were above 4°C). Both of these modifications were implemented to minimize mechanical disturbance and to maintain the surface oils, waxes and esters on the qiviut hair shaft in solid rather than liquid state, and with minimal surface area. As methanol is an excellent solvent for steroids, and the hair shaft is not thick, the methanol was expected to equilibrate and extract the steroid hormones present inside the hair shaft, while reducing potential matrix effects from other compounds on the hair. Thus, the concentrations of cortisol reported are the integrated sum of steroids external and internal to the hair shaft remaining after the cold washing procedure.

2.3.6. Cold wash procedure to remove surface contamination

Each qiviut sample was placed in a 50 ml Falcon tube. A soap solution was prepared with 2 ml of Neutrogena Body Clear® Salicylic acid body cleanser for acne-prone skin (chosen for the specific claim that no residue was left after rinsing) added to 2 l of cold tap water and gently stirred for 3 min. The soap solution was then chilled on ice and 20 ml was added to each of the Falcon tubes containing the qiviut. The qiviut was washed by vortexing for 30 s, then the soapy water was decanted and the qiviut rinsed thoroughly with cold tap water, and patted dry in a clean paper towel. As a final rinse, the qiviut was next transferred to a new Falcon tube, and 20 ml of isopropyl alcohol (IPA) pre-chilled to -20°C, was added and mixed by gentle inversion for 10 s, then decanted to waste. Qiviut was air-dried in paper towel for at least 24 h before extraction.

2.3.7. Extraction procedure

The qiviut samples were re-weighed before being placed in 13 × 100 mm borosilicate glass test tubes, and submerged under 6 ml of methanol pre-chilled to -20°C. A 100 μl spike of bio-identical, deuterated internal standard (IS—cortisol-d4 Catalogue D-5280, CDN isotopes, Pointe Claire, QC) in water/methanol (50/50, v/v) was added, with calibrators (Catalogue Q3880-000, Steraloids, Newport RI) and three quality control pools (low, medium and high calibrator pools in methanol) spiked at the same time. Samples were extracted for 20 h at 4°C in upright tubes (i.e., no vortexing or spinning). Cold supernatant was pipetted off the hair, transferred into new culture tubes, dried at 40°C under nitrogen (Techne® Sample Concentrator), and then stored, capped, at 4°C until reconstitution.

2.3.8. Sample reconstitution

Each dry sample was reconstituted with 200 μl of water/methanol (100/100, v/v), and vortexed for 30 s. The total volume was transferred to a 600 μl microcentrifuge tube and centrifuged at 14,000 rpm for 20 min at 4°C. One hundred and fifty microliters of supernatant were immediately transferred into liquid chromatography (LC) autosampler vials for subsequent analysis. Solid phase extraction (SPE) was not used as, during method development, we found a high correlation between the qiviut cortisol levels from twelve paired samples, whose analysis differed only by the addition or not of a SPE step (Pearson’s correlation coefficient (r) = 0.94, p < 0.001), and there was no

21 significant difference between the mean qiviut cortisol levels obtained with or without SPE (paired samples t-test: t = −0.65, p = 0.53). All samples were analyzed by using an Agilent 1200 binary LC system coupled to an AB SCIEX QTRAP® 5500 tandem mass spectrometer equipped with an atmospheric pressure chemical ionization (APCI) source in positive mode. Liquid chromatography separation was performed on an Agilent Poroshell 120 C18 column (50 × 3 mm, 2.7 μm particle size) at 45°C. The mobile phase A was water/methanol (75/25, v/v) and the mobile phase B was methanol/IPA (90/10, v/v). The 8.5 min gradient was 20–40% B (0–1.0 min), 40–60% B (1.0–5.0 min), 60–100% B (5.0–5.5 min), 100% B (5.5–6.5 min), 100–20% B (6.5–7.0 min) and held at 20% B (7.0–8.5 min). The flow rate was 0.6 ml/min and the injection volume was 20 μl. Nitrogen was used as the source, nebulizer and collision gas (curtain gas 28 psi; temperature 500°C; source gas 50 psi; collision gas medium; nebulizer current 8 μA). Mass resolution in Q1 and Q3 was set to unit resolution. Three transitions were monitored, the cortisol quantifier (-1), and qualifier (-2), and the deuterated cortisol IS quantifier transitions. Specific MRM (Multiple Reaction Monitoring) conditions were: m/z = mass-to-charge ratio (363.2/121.1; 363.2/115.1; 367.2/121.1), DP = declustering potential (all 85 V), EP = entrance potential (all 10 V), CE = collision energy (35; 114; 32 eV) and CXP = collision cell exit potential (all 20 V).

2.3.9. Data processing

Cortisol peak integration was performed using Analyst 1.5.1 software (AB SCIEX). Sample quantitation used the area ratio between the analyte peak and the matched internal standard peak. Calibration curves (1/x weighted linear regression) covered the range from 0.25 through 250 ng/ml with r ≥ 0.99 in each run. The intra-assay variability was on average 8.3% and the inter-assay variability was 13.5%. Each cortisol result was divided by qiviut mass to obtain pg/mg. Duplicate quantitation was repeated on a second run, with new hair samples, whenever the coefficient of variation (CV), calculated as (standard deviation/mean) × 100 was ≥ 15%. The average from the two new replicates was then used for statistical analyses.

2.3.10. Statistical analysis

Only adult animals, defined as two years or older based on hunter assessment, were included in the statistical analyses. To assess the effect of sex, season and year on log-transformed qiviut cortisol, we fit several different linear mixed-effects models with possible fixed effects including sex (male or female), year (2013, 2014, 2015 or 2016), season of collection (winter, summer or fall), and interactions between year and season, sex and season, and sex and year. In all models, we included a random effect for location of sampling to account for possible differences in qiviut cortisol among locations due to, for example, slight differences in hunter methodology or intrinsic differences in qiviut cortisol among locations linked to environmental (e.g., food quality, quantity and availability, weather etc.) or other location specific conditions. Qiviut cortisol levels were log-transformed to satisfy the linear model assumptions.

Models with different fixed effects were fit using restricted maximum likelihood and were compared using the adaptation of Akaike Information Criterion for small sample sizes (AICc) as the ratio of (sample size)/(number of parameters) was small (Burnham and Anderson, 2002). To avoid spurious relationships from fitting all possible combinations of fixed effects and interactions, models were developed in the following order: first, the effect of sex was tested, second the effects of year and season were assessed, and finally, the effect of sex was checked in the final model. This process resulted in testing a total of 14 different models (Table 2.2). The optimal subset of fixed effects that explained qiviut cortisol corresponded to the effects included in the model with the lowest AICc. This top model was re-fit using maximum likelihood to obtain the final parameter estimates. All models were fit using the R software (R Core Team, 2019) and the library lme4 for mixed-effects models (Bates, 2010). Marginal (fixed effects only) and conditional (fixed and random effects) coefficients of determination (R²) for mixed-effects models were calculated using the MuMIn library (Bartoń, 2015).

2.4. Results

Qiviut was sampled from 150 adult muskoxen between January 2013 and August 2016 (Table 2.1). Cortisol levels followed a right-skewed distribution and ranged from 3.5 to 48.9 pg/mg with a median of 11.7 pg/mg. This corresponds to an almost 14-fold variation. Qiviut cortisol levels by sex, season and year can be found in Figure 2.3.

Figure 2.3. Boxplot showing combined qiviut cortisol values from all animals, seasons, years, sexes, and locations by sex (a), season (b), and year (c). The thick horizontal lines correspond to the medians, the triangles to the means, and the empty circles to the outliers.

Model comparison using AICc showed similar support for three top models, all including the variables sex, year and season (Table 2.2). We based parameter estimates on the most parsimonious of these three, which had also the lowest AICc, and which included sex, year and season of collection as fixed effects, and no interaction terms (AICc = 158.86). The two other top models also included interactions between sex and season, and sex and year, respectively. Both the marginal and conditional R² were relatively low at a rounded value of 0.33, respectively. The best-fit model conformed to the assumptions of normality and constant variance (see Appendix D).

Table 2.2. Comparison of linear mixed-effect models including location as a random effect, with their corresponding AICc, ΔAICc in comparison to the best-fit model (bold), and degrees of freedom (DF).

Model fixed effects AICc ΔAICc DF sex, year, season 158.86 0 9 sex, year, season, sex:season, 159.63 0.77 10 sex, year, season, sex:year 160.41 1.55 11 sex, season, year:season 162.50 3.64 12 sex, year:season 162.50 3.64 12 sex, year, year:season 162.50 3.64 12 sex, year, season, year:season 162.50 3.64 12 sex, year, season, sex:season, year:season 162.52 3.66 13 sex, year, season, sex:year, year:season 163.60 4.74 13 year, season 167.13 8.27 8 year, season, year:season 170.45 11.59 11 sex, season 171.94 13.08 6 sex, year 176.30 17.44 7 sex 199.42 40.56 4

Based on the best-fit parameter estimates, qiviut cortisol levels were on average 1.30 (95% confidence interval (CI) = 1.11–1.53) times higher in males than in females. Season also had a significant effect, with qiviut cortisol levels in fall and winter respectively 1.57 (95% CI = 1.26–1.95) and 1.70 (95% CI = 1.37– 2.13) times higher on average than that in summer. Levels in fall and winter were not significantly different. Regarding differences between years, qiviut cortisol levels were on average 1.28 times (95% CI = 1.03– 1.58), 1.67 (95% CI = 1.33–2.10) and 1.60 (95% CI = 1.21–2.10) times higher in 2014, 2015 and 2016, respectively, compared to 2013. Estimated qiviut cortisol values according to sex, season and year can be found in Appendix D. The variance among locations was low (σ² location = 0.0015) compared to the residual variation (σ² resid = 0.16), but the conditional modes for the levels of the random effect at each location suggest slight differences in cortisol among locations that are consistent with the local muskox population trends, with qiviut cortisol concentrations that tend to be higher in declining populations on Banks and Victoria islands (Appendix D).

2.5. Discussion

2.5.1. Findings

This is the first study measuring hair cortisol levels in muskoxen and it has provided important baseline data on variability within and among populations, sexes, seasons and years that will inform future studies. Our first objective was to develop a reliable technique for cortisol quantification in the qiviut of muskoxen using LC–MS/MS. We chose to use LC–MS/MS as it has several advantages over the more commonly used ELISAs. It has a high specificity in molecular identification, as well as a calibration curve that is both linear and broad (in this case, 4 orders of magnitude). By contrast, in ELISAs, four-parameter logistic calibration curves will have quantitation errors outside of the linear portion of the curve. Therefore, when using LC–MS/MS, even when cortisol is present at a very low concentration, the detection of a signal corresponding to this hormone can be attributed to it with high confidence (Shackleton, 2010; Gale et al., 2015). Additionally, although not reported in this study, multiple steroids can be quantified simultaneously with LC–MS/MS. This is particularly useful for wildlife studies as samples are often difficult to collect and thus gaining as much information as possible from a single sample is desirable (Shackleton, 2010; Koren et al., 2012b). Furthermore, in LC–MS/MS, the addition of an internal standard to every sample (quality controls, unknowns and calibrants) early during their preparation process (just after the cold washing step in our case), intrinsically corrects for any subsequent analyte loss, and, therefore, improves the accuracy and precision of the method (Gale et al., 2015). Finally, LC–MS/MS offers a better reproducibility in comparison with ELISAs that are poorly reproducible between laboratories (Shackleton, 2010). However, LC–MS/MS is expensive, and requires substantial technical expertise and highly specialized equipment. In contrast, ELISAs come with advantages, such as technical simplicity, considerable lower costs associated with the equipment, and easy transportation of the equipment to remote field locations (Crowther, 2009). The various advantages offered by each method render them complementary, with a valuable use in tandem. For example, LC–MS/MS could first be applied to prove the presence of a certain compound in a new sample type or species and to validate the ELISAs, which could subsequently be used more widely to expand the technique to a wider group of research teams that may have limited access to LC–MS/MS. We found a high variability in qiviut cortisol levels among individual muskoxen, with levels ranging from 3.5 to 48.9 pg/mg, corresponding to an almost 14-fold variation. This wide range of values is consistent with other studies, where approximately 1.5 to 40-fold variations are reported depending on the study, number of animals included, and species of interest (Table 2.3). This highlights the high inter- individual variability of hair cortisol concentrations within a species, which may reflect the different health and social status of individuals (Creel, 2001; Busch and Hayward, 2009; Dantzer et al., 2014). Muskoxen are highly social and organized by a strong dominance hierarchy (Gray, 1987) and our results may, in part, reflect social interactions and status of individuals.

Table 2.3. Ranges of hair cortisol concentrations previously determined in free-ranging or captive wild mammal species using ELISAs (n = sample size).

Species Median (M) or mean (m) pg/mg (range) N Reference Rhesus macaque 110.3 (m) (32.1-254.3) 20 Davenport et al., 2006 (Macaca mulatta) Caribou 2.31 (M) (1.57-3.86) 12 Macbeth, 2013 (Rangifer tarandus tarandus) Reindeer 2.88 (M) (2.21-3.40) 12 Macbeth, 2013 (Rangifer tarandus granti) Wolves (Canis lupus) Tundra-taiga Females: 17.3 (M) (9.95-32.2) 48 Bryan et al., 2015 Males: 15.8 (M) (8.91-40.4) 55 Northern boreal forest Females: 14.6 (M) (7.6-34.0) 24 Bryan et al., 2015 Males: 12.3 (M) (4.8-26.8) 21 Polar bears 9.5 (m) (5.5-19.9) 17 Bechshøftet al., 2011 (Ursus maritimus) 12.75 (m) (3.98-24.42) 88 Bechshøftet al., 2012 0.48 (M) (0.16-2.) 185 Macbeth et al., 2012 Grizzly bears 2.84 (M) (0.62-43.33) 151 Macbeth et al., 2010 (Ursus arctos) 8.1 (M) (5.3-26.1) 113 Bryan et al., 2013

Particularly high qiviut cortisol concentrations, greater than 28 pg/mg, were measured in four individuals. Although all samples were washed the same way, the potential remains for cryptic blood to persist on the hair surface and be extracted. Since blood concentrations of cortisol are considerably higher than hair concentrations, blood contamination could substantially increase measured results. However, these exceptionally high “outlier” hides were re-tested and similar results attained, suggesting that these are real values, and not a result of undetected, local blood contamination. Moreover, these results are consistent with other studies where individuals with higher cortisol levels are commonly identified (e.g., Macbeth et al., 2012; Malcolm et al., 2013), and such outliers may represent highly stressed animals. Although we were challenged by small sample sizes and a poor representation across groups (Table 2.1 and Appendix D), we observed significant effects of year, season and sex on qiviut cortisol levels with a conditional R² of 0.33. Other potentially relevant variables including, but not limited to, body condition, disease and pregnancy, or weather characteristics like snow depth, temperature and humidity may have contributed to the remaining 66% of the variability in qiviut cortisol levels but were not assessed in this study. Qiviut cortisol levels varied by season, with levels in summer being significantly lower compared to fall or winter, and a non-significant increase from fall to winter. If we are measuring only a negligible part of the external cortisol, then based on the qiviut growth cycle (beginning in April and ending in November, with possibly slight latitudinal variation (Flood et al., 1989)), August collected samples would reflect stressors from April to the time of collection and fall and winter collected samples would reflect the physiological stress experienced during the entire qiviut growth season. However, if external deposition of cortisol contributes substantially to the measured levels (see discussion on qiviut cortisol deposition in

26 section 2.5.2), we would predict that differences between fall and winter samples should reflect the season- specific stressors. Seasonal variations in GC concentrations have been found in numerous ungulate species, often in association with environmental conditions (food abundance and quality, climatic characteristics, human activities, etc.). For example, fecal glucocorticoid metabolite (FGM) levels in bighorn sheep (Ovis canadensis) were significantly lower in the winter than in all the other seasons (Goldstein et al., 2005). Similarly, among free-ranging elk (Cervus canadensis) in South Dakota, USA, FGM concentrations were at their lowest in winter and peaked in the summer, conceivably due to increased human disturbance, high temperatures, or normal seasonal metabolic patterns (Millspaugh et al., 2001). Conversely, FGM levels peaked during the winter in a captive herd of red deer (Cervus elaphus) in Austria, with a significant effect of snow and minimum ambient temperature (Huber et al., 2003). These contrasting results may be due to different responses among species to stressors or to variations in local environmental conditions. Fecal glucocorticoid metabolite levels were also significantly higher during the dry season in African elephants (Loxodonta africana) presumably due to less nutritious and more sparsely distributed sources of food, and an increased competition between individuals (Foley et al., 2001). Similar seasonal patterns in FGM levels were described in zebras (Equus quagga) and springboks (Antidorcas marsupialis) during the dry season in relation to decreased food and water availability (Cizauskas et al., 2015). For muskoxen, stressors experienced between April and August may include heat extremes and insect harassment. Higher temperatures in the Arctic, particularly in the summer, have been described with increasing climate warming (MacCarthy et al., 2001), and an outbreak of pneumonia in the muskox population of Dovrefjell, Norway during the summer of 2006 was associated with unusually high temperatures and humidity (Ytrehus et al., 2008). Similarly, even though insect harassment effects on muskoxen have not been described, changes in insect phenology in the Arctic are already observed due to climate warming with an earlier occurrence, higher abundance, and an increased frequency of insect outbreaks, along with the observation of new species (Weller, 2005). Insect disturbance is a major stressor for caribou, and likely has negative effects on muskoxen as well. For example, oestrid fly and mosquito harassment affects the behavior of caribou (Rangifer tarandus tarandus) and reindeer (Rangifer tarandus granti) by significantly reducing their time spent foraging, and by augmenting their energy expenditure through increased movements and time spent standing (Toupin et al., 1996; Vors and Boyce, 2009). It also has negative effects on reindeer body condition and productivity (Weladji et al., 2003). Muskox females may experience additional stress associated with calving (mid-April to early June (Adamczewski and Flood, 1997; Gunn and Adamczewski, 2003)) and lactation. Female muskoxen lose much more fat during the first weeks of their lactation period following calving than during the previous 4 to 5 months (Adamczewski et al., 1992). While this period is thought to be stressful for most ungulates, lactation was not associated with increased FGM levels in red deer (Huber et al., 2003). Unfortunately, we were unable to sample female muskoxen during the summer, thus we cannot draw conclusions about this potential stressor from our data. Males may undergo additional stress due to the first agonistic interactions that are associated with the rut

27 and begin in early to mid-July (Gray, 1987). Despite the multiple stressors occurring in summer, this season also corresponds to the period of highest food availability and quality in the Arctic, with a peak in July followed by a rapid decline (Gray, 1987; Adamczewski et al., 1992), and would, therefore, be expected to be a season of low nutritional stress for muskoxen (Gunn and Adamczewski, 2003). This may contribute to explaining the lower qiviut cortisol levels observed in summer. Through late summer and early fall, muskoxen experience social stress associated with breeding. Glucocorticoids peak during the breeding season in reptiles, amphibians, birds, and certain mammals (e.g., Touma and Palme, 2005). Reyes et al. (1997) reported a peak in serum cortisol concentrations in male pudu (Pudu puda) during the rutting season (Reyes et al., 1997). Peak GC levels have also been reported during the breeding season for the northern muriqui (Brachyteles arachnoids hypoxanthus) (Strier et al., 2003) and the tufted capuchin monkey (Sapajus apella) (Lynch et al., 2002) and are attributed to the competition for breeding partners and an increased occurrence of aggressive encounters between males. Beginning in the fall and extending into all of the winter period, muskoxen may experience stress due to higher human disturbance associated with hunting, low food quality and availability, and extreme cold temperatures. Anthropogenic stress may be higher than in the summer as people have more access to the land through the use of snowmobiles, thus increasing the disturbance of the animals (noise pollution and human encounters). Muskoxen react to snowmobile activity (McLaren and Green, 1985) and in the area around Cambridge Bay, behavioral changes, with animals having a much longer flight distance, have been observed with increased snowmobile activity (pers. comm. from Matilde Tomaselli, University of Calgary). Fecal glucocorticoid metabolite levels in elk (Creel et al., 2002) and caribou (Freeman, 2008) are also positively associated with snowmobile activity. Lower food abundance and quality (Adamczewski et al., 1992), along with increased agonistic behaviors (displacements) related to food accessibility often restricted to feeding craters (Gray, 1987), may lead to higher nutritional and social stress in the fall and winter. Muskoxen are highly adapted to the extremely cold environmental conditions in the Arctic, with one of the lowest metabolic rates recorded in ruminants both in the summer and winter, as well a high capacity to digest low-quality forage (Gunn and Adamczewski, 2003). However, reduced food intake and consumption of low-quality forage in the winter may be associated with increased cortisol concentrations indicating a shift towards a catabolic metabolism and the mobilization of energy stores (Parker and Rainey, 2004). Qiviut cortisol levels were significantly higher in males than in females. However, there was also weak support for interactions between sex and season, and sex and year, so a more balanced sampling is required to confirm this finding. This difference contrasts with results from a previous study of hunter- harvested muskoxen which showed no significant difference in serum cortisol values between males and non-pregnant females aged more than one year (Wilcoxon rank sum test: W = 65, p = 0.98) (Katherine Wynne-Edwards, pers. obs. based on samples reported in Koren et al., 2012c). However, as these were hunted muskoxen, any sex-specific differences may have been overshadowed by the immediate stress of hunting. Other studies have shown varied results regarding sex-specific differences in FGM levels. A study in captive goral (Naemorhedus griseus), one of the muskox’s closest taxonomic relatives, described higher

FGM levels in males (Khonmee et al., 2014), but many other studies in ungulates report no difference between sexes (captive red deer (Huber et al., 2003), free-ranging elk (Millspaugh et al., 2001), free-ranging American bison (Bison bison) (Ranglack et al., 2016), and captive black (Diceros bicornis) and white (Ceratotherium simum) rhinoceros (Brown et al., 2001)). Differences in FGM or hair GC levels between sexes may reflect distinct physiological and/or behavioral characteristics that certainly vary among species and across seasons. We also observed a difference in qiviut cortisol levels among years. The trend for increasing cortisol from 2013 to 2015 suggests increasing population-level stressors. These may reflect environmental conditions (food availability, snow characteristics, temperature, humidity, occurrence of extreme weather events, predation risk, hunting pressure, etc.), exposure to pathogens, or a combination of these and other factors that could have affected both long and shorter-term stress levels. During this time period, the average annual surface air temperatures in Arctic Tundra region of Canada from 2013 to 2015 were 1°C, 1.1°C and 1.3°C above the 1961–1990 reference period, with average summer air temperatures 0.8°C, 1.1°C and 1.2°C, above the reference period (data from Environment Canada). These warm temperatures, and other associated ecological changes, may in part be contributing to increasing stress levels. All the samples collected from 2013 to winter 2015 were analyzed during the summer of 2015, whereas the samples collected onwards were analyzed during the subsequent summer. Possible hormone degradation during storage may, therefore, have had a confounding effect on the yearly differences we observed. However, multiple studies, including a small pilot study by our team, have shown that cortisol is highly stable over long periods of time in hair at room temperature and in other matrices (e.g., serum, saliva and feces) at subzero temperatures (Hunt and Wasser, 2003; Garde and Hansen, 2005; Stroud et al., 2007; Macbeth et al., 2012; Yamanashi et al., 2016b). These studies support that the differences in qiviut cortisol observed between years are real and not entirely driven by possible hormone degradation because of long- term storage conditions.

2.5.2. Study limitations and future considerations

It is important to note that the hair cortisol levels we measured in muskoxen are not directly comparable to the levels measured in other species. Indeed, all the studies in Table 2.3 used one of two different ELISAs: the Oxford EA-65 Cortisol EIA kit, Oxford Biomedical, Lansing, MI, USA (Macbeth et al., 2010, 2012; Macbeth, 2013) or the Salivary Cortisol ELISA Kit, Salimetrics, Philadelphia, PA, USA (Davenport et al., 2006; Bechshøft et al., 2011, 2012; Bryan et al., 2013, 2015). A study comparing cortisol concentrations of both human and vervet monkey (Chlorocebus pygerythrus) hair measured using four different immunoassays (Alpco ELISA (Alpco, Salem, NH), DRG International ELISA (DRG Instruments GmbH, Marburg, Germany), Salimetrics ELISA (Salimetrics, LLC, State College, PA) and IBL luminescence immunoassay (LIA) (IBL International, Hamburg, Germany)) with ones measured using LC–MS/MS (Russell et al., 2015), showed a high correlation (r were between 0.88 and 0.98, P < 0.0001), but hair cortisol concentrations were 2.5 to 20 times higher when measured using immunoassays than with LC–MS/MS. The Salimetrics ELISA and IBL LIA gave the closest results to LC–MS/MS with concentrations only

29 approximately 2.5 times higher (Russell et al., 2015). These results suggest that hair cortisol levels measured by LC–MS/MS and ELISAs are correlated, but not equivalent, and require validation against reference standards before comparison of concentrations across methods. The qiviut cortisol concentrations we measured represent the sum of external and internal cortisol remaining after the cold washing procedure. The internal cortisol is assumed to be incorporated in the hair during the course of hair growth. It may derive both from circulating cortisol concentrations (Russell et al., 2012) and from local synthesis in the skin (Ito et al., 2005; Keckeis et al., 2012). Because qiviut growth occurs from April to November, it is assumed that no additional cortisol enters the hair outside of this time period. However, cortisol originating from either blood, local production or both, and secreted by sebaceous and eccrine glands surrounding the hair follicle may be deposited on the outer cuticle of the hair shaft in association with sebum and sweat throughout the year (Raul et al., 2004; Russell et al., 2014). This external cortisol from sebum and sweat may, therefore, reflect more recent and ongoing events. In this regard, however, qiviut might be less affected by cutaneous secretions than many other wildlife hair samples. In muskoxen, sebaceous glands and apocrine sweat glands are associated with the primary follicles that produce guard hairs, whereas no sweat glands and only small sebaceous lobules are associated with a minority of the secondary follicles that produce qiviut (Flood et al., 1989). Qiviut is, consequently, a much dryer fiber than wool, with approximately 7% grease and a very low amount of suint (dried sweat), so the amount of cortisol from sebum and sweat on the hair shaft is probably limited (Rowell et al., 2001; Helfferich, 2008). Nevertheless, our data, with a slight trend for increasing cortisol from fall to winter, suggests that recent events may be contributing to the cortisol levels measured. Storage of qiviut prior to processing was not expected to affect the uptake of external cortisol. Qiviut samples were cut from pieces of hide that were stored at -20°C and taken out only for the duration of sample cutting. Condensation and humidity due to thawing may have increased the permeability of the hair, as water is known to extend the hair cuticle, potentially causing the cortisol from sebum, sweat and possible external contaminants (blood, urine, feces) to leach inside the hair shaft, thereby confounding our results (Kidwell and Blank, 1996; Macbeth, 2013; Cattet et al., 2014). Once this external cortisol has been incorporated into the hair matrix, it cannot be removed by decontamination protocols (Kidwell and Blank, 1996). Its contribution to the concentrations measured in our study remains unknown but is likely consistent across study areas and time periods as samples all experienced an initial freezing event. Additional research is needed to provide evidence that qiviut cortisol levels reflect adrenal activity, which will enable us to link high qiviut cortisol levels with stress. Adrenocorticotropic hormone (ACTH) challenges involving repeated weekly injections over several weeks have been used to validate that the measurement of cortisol in hair reflects adrenal changes in several species of domestic or captive wild animals, including eastern chipmunks (Tamias striatus) (Mastromonaco et al., 2014), Canada lynx (Lynx canadensis) (Terwissen et al., 2013) and domestic cattle (Bos taurus) (González-de-la-Vara et al., 2011). By contrast, a study in caribou and reindeer showed that hair cortisol levels were not affected by a single ACTH injection (Ashley et al., 2011). Consequently, in order for changes in adrenal activity to be detectable in the

30 hair, it seems necessary to administer a prolonged ACTH challenge (e.g., including repeated injections over time), mimicking chronic stress. Such a validation study is a focus of our future work. Finally, the determination of the qiviut growth cycle was based on the observation of captive animals in Saskatoon, Saskatchewan, Canada. The exact timing of the qiviut growth cycle in wild muskoxen in the Arctic remains unknown, but is unlikely to differ substantially as qiviut is shed at a similar time (Gray, 1987; Flood et al., 1989). Qiviut growth rate varied throughout the cycle in captive muskoxen in Alaska, USA, with a peak in August and a slow-down in October, although growth was observed until the end of November (Robertson, 2000). The differences in qiviut growth rate throughout the cycle or among individuals may affect cortisol deposition in the qiviut hair shaft. A better understanding of the qiviut growth patterns is important for interpretation of results and seasonal data.

2.6. Conclusion

The Arctic ecosystem is currently experiencing one of the greatest rates of climate and ecological change in the world (Overland et al., 2016). How Arctic-adapted species will persist in the face of this rapidly changing environment and increasing cumulative stressors is of concern, both from a conservation and human point of view, as these species have important economic, nutritious and socio-cultural values for Arctic communities (Nuttall et al., 2005; Kutz et al., 2014). The health of many Arctic mammal species is already changing, and is predicted to be increasingly affected by climate warming and its multiple impacts (loss of sea ice, increased occurrence of extreme weather events, elevation of sea level, higher air temperatures, etc.), which may alter food webs and pathogen transmission patterns (Weller, 2005; Burek et al., 2008; Altizer et al., 2013; Post et al., 2013; Kutz et al., 2014). There is a crucial need, therefore, to develop robust methods for monitoring individual and population health of wildlife species in order to inform management. Research and surveillance of wildlife species is challenging, with many logistic and financial restrictions, along with an access to often restricted sample sizes and diagnostic tools (Ryser-Degiorgis, 2013). These difficulties are exacerbated in remote locations like the Arctic, where the environment is extreme (Curry et al., 2011). Hair sampling could be incorporated with minimal additional effort into community-based wildlife health surveillance programs, notably through the collaboration with hunters for sample collection, and without major financial constraints (Brook et al., 2009; Carlsson et al., 2016b). We have shown that qiviut GC levels can be reliably quantified in muskoxen using LC–MS/MS, and found a high inter-individual variability in qiviut cortisol levels, along with sex, seasonal and year effects. These results suggest that qiviut cortisol may become a valuable tool for monitoring individual and population health in muskoxen and informing conservation and management efforts.

2.7. Acknowledgments

We wish to thank our two anonymous reviewers for their help in making this paper stronger. We are grateful to all the hunters and guides who participated in sample collection, and to the Ekaluktutiak Hunters and Trappers Organization, the Kugluktuk Angoniatit Association, the Sachs Harbour Hunters

31 and Trappers Committee, the Paulatuk Hunter and Trappers Committee, the Olokhaktomiut Hunter and Trappers Committee, and the Nunavut Harvesters Association. We are also grateful to Canada North Outfitting and Qiviuk Group/Jacques Cartier Clothier, and Erin Prewer and Christopher Kyle for their help in muskox sex determination. We wish to thank Felix Nwosu, Lea J. Bond, Ruokun Zhou and Kamala Sapkota for their hard work in method development and sample analysis. We finally thank James Wang, Angie Schneider and Angeline McIntyre.

2.8. Funding

This work was supported by University of Calgary Eyes High Seed Grant, Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery and Northern Supplement grants, Canada North Outfitting, Government of Nunavut, Polar Knowledge Canada, Government of the Northwest Territories and the Nunavut General Monitoring Plan, and ArcticNet Network Center of Excellence. Juliette Di Francesco and Nora Navarro-Gonzalez were supported by the NSERC-Collaborative Research and Training Experience (CREATE) Host-Parasite Interactions Training Program student scholarship and postdoctoral fellowship, respectively. Juliette Di Francesco and Matilde Tomaselli were supported by the NSERC-CREATE Integrated Training Program in Infectious Diseases, Food Safety and Public Policy Student Scholarship. Stephanie Peacock was supported by an NSERC Postdoctoral Fellowship.

CHAPTER 3. DOCUMENTING INDIGENOUS KNOWLEDGE TO IDENTIFY AND UNDERSTAND THE STRESSORS THAT AFFECT MUSKOXEN (OVIBOS MOSCHATUS)

Juliette Di Francesco1, Andrea Hanke1, Terry Milton2, Lisa-Marie Leclerc2, Kugluktuk Angoniatit Association, Craig Gerlach3, Susan Kutz1

1Department of Ecosystem and Public Health, Faculty of Veterinary Medicine, University of Calgary, 3280 Hospital Drive NW, Calgary, Alberta, Canada T2N 4Z6

2Department of Environment, Government of Nunavut, P.O. Box 377, Kugluktuk, Nunavut, Canada X0B 0E0

3School of Architecture, Planning and Landscape, University of Calgary, 2500 University Drive NW, Calgary, Alberta, Canada T2N 1N4

Manuscript submitted on September 25th, 2020 to the journal ARCTIC.

Full length manuscript.

3.1. Abstract

Indigenous knowledge is now recognized as an invaluable source of information on wildlife health and ecology, contributing to a broader understanding of the patterns and phenomena observed. Muskoxen (Ovibos moschatus), an important species for the subsistence and culture of Inuit communities in the Arctic, are increasingly exposed to a wide diversity of stressors linked to rapid climate change and other anthropogenic changes. Identifying and understanding these stressors and their impacts on muskoxen will inform management, health monitoring, and future research. To achieve this, we documented Indigenous knowledge through seven semi-structured small group interviews, each involving two to three purposely chosen harvesters in Kugluktuk, Nunavut, Canada to: (i) establish the characteristics of healthy muskoxen; (ii) determine the factors that impact muskoxen; and (iii) collaboratively re-interpret a study on the sex, seasonal, and annual patterns of glucocorticoids (described as “stress hormones” for the purposes of the interviews) in muskox hair. Key outcomes include: (i) a more holistic understanding of muskox health and what it encompasses; (ii) recognition and exploration of a rich One Health perspective expressed by participants around factors influencing muskoxen in a “changing world” and highlighting the multiple socio-ecological connections; and (iii) a broader comprehension of the glucocorticoid (stress) patterns measured in muskox hair, the various factors that influence them, and their interrelations. This study represents a meaningful advancement in the process of actively involving communities at all steps of the research and not only in data collection, and highlights the importance of bridging scientific and Indigenous knowledge in general, but also more specifically in the complex field of wildlife endocrinology research.

Key words: Muskox, Ovibos moschatus, wildlife, stress, Indigenous knowledge, Arctic

3.2. Introduction

Conservationists and ecologists are increasingly focused on understanding physiological stress responses in wildlife, including their causes and consequences, and how they may be affected by various ecological changes and environmental challenges (Bonier et al., 2009a; Busch and Hayward, 2009; Dantzer et al., 2014; Koren et al., 2019). The standard approach has been to measure glucocorticoids, which are important mediators of the physiological stress response (Romero and Butler, 2007), and examining associations with potential stressors (e.g., anthropogenic disturbance and habitat alterations (Mastromonaco et al., 2014; Fourie et al., 2015b; Ewacha et al., 2017)). Identifying stressors and understanding their impacts on different wildlife species are all the more important now in light of rapidly changing and sensitive environments, such as the Arctic where climate change is occurring at an unprecedented pace and anthropogenic activities are accelerating (Post et al., 2013; AMAP, 2017). Bridging scientific and Indigenous knowledge (IK) is an increasingly recognized strategy to address complex conservation problems (Huntington, 2011; Kutz and Tomaselli, 2019). Rarely, however, if ever, has IK been taken into account to identify and assess the impacts of stressors on wildlife. Indigenous knowledge is "a cumulative body of knowledge and beliefs handed down through generations by cultural transmission about the relationship of living beings, (including humans) with one another and with their environment" (Gadgil et al., 1993:151). It is constantly evolving from the experience and observations of its holders and provides information on historical conditions and local processes at a detailed scale (Gadgil et al., 1993; Usher, 2000). Scientific knowledge (SK) is often considered the “gold standard” for conservation action, driven as it is by specific research questions that rely on objective and quantitative measures. However, data collection, even though systematic, tends to be heavily circumscribed in time and space, particularly in remote areas such as the Arctic, where financial and logistical barriers are obstacles to extensive fieldwork. This dependence on SK can lead to fragmented and limited data and field observations that can be difficult to interpret (Lubin and Massom, 2006; Kutz and Tomaselli, 2019). Bringing IK and SK together allows for the possibility to circumvent some of the challenges that SK alone faces (Brook et al., 2009; Tomasini, 2018; Kutz and Tomaselli, 2019), provides the potential to arrive at a much richer and deeper understanding of ecological systems (Sefa Dei et al., 2000; Robinson and Wallington, 2012; Kutz and Tomaselli, 2019; Baker and Constant, 2020), and addresses mandates set out by land claims agreements (e.g., Indian and Northern Affairs, 1984, 1993) and calls for Indigenous sovereignty in research (Schnarch, 2004; Inuit Tapiriit Kanatami and Nunavut Research Institute, 2006; Inuit Tapiriit Kanatami, 2018). Indigenous knowledge is now recognized as an invaluable source of information on wildlife health and ecology, and it has increasingly been used to advance this field of research (Kutz and Tomaselli, 2019). With respect to wildlife ecology, for instance, the knowledge of Dene people was documented concomitantly with genetic data to comprehensively characterize caribou (Rangifer tarandus) types in the Sahtu Region (Northwest Territories (NWT), Canada) (Polfus et al., 2016) and the knowledge of Iñupiat from Point Lay (Alaska, USA) and satellite telemetry data were combined to gain a more complete understanding of the ecology of beluga whales (Delphinapterus leucas) (Huntington et al., 2004; for other

35 examples see Kumpula et al., 2011; Riseth et al., 2011; Laforest et al., 2020). With respect to wildlife heath, IK documented in the community of Ekaluktutiak (Nunavut, Canada) was mobilized with scientific data to describe the population trends and health status of muskoxen (Ovibos moschatus) and caribou in the area (Tomaselli et al., 2018b). Participants’ observations provided insights into previously undocumented mortality events and possible disease-associated causes of decline (Tomaselli et al., 2018b). These examples demonstrate how systematically documenting IK can lead to an expanded collective understanding of wildlife health and ecology. Muskoxen, a taxonomically unique and emblematic ungulate species in the Canadian Arctic, are an essential part of the local ecosystem and are important for the subsistence, economy, and cultural identity of Inuit communities who have depended on them for generations (Lent, 1999; Tomaselli et al., 2018a). Muskoxen are increasingly exposed to a wide diversity of stressors linked to rapid climate change and other anthropogenic changes (AMAP, 2017; Kutz et al., 2017; Cuyler et al., 2019). The effects of these stressors are likely cumulative, extremely complex, and remain poorly understood. Muskoxen are highly adapted to life in the Arctic, but recent, substantial, and ongoing population declines in the western Canadian Arctic Archipelago suggest that they are particularly vulnerable to a shifting environment (Kutz et al., 2017; Cuyler et al., 2019). Cortisol levels (the dominant glucocorticoid in muskoxen (Koren et al., 2012c)) have been reliably measured in the hair of muskoxen, with their interpretation solely based on the current state of SK on wildlife endocrinology and muskox ecology (Di Francesco et al., 2017). However, identifying the stressors that affect muskoxen more specifically, and understanding their impacts more holistically, will greatly inform endocrinological studies, as well as population management, current health monitoring programs, and future research directions. The goal of this research was to document IK to identify and gain a better understanding of the stressors that affect muskoxen. The specific objectives were to (i) document IK on the characteristics harvesters use to identify whether a muskox is healthy; (ii) document IK on the factors that affect muskoxen in a positive and/or negative way, their importance and impacts, when they occur throughout the year, and how they have changed over time; and (iii) improve our understanding of the stress patterns observed in muskoxen by working collaboratively with IK holders to re-interpret the results from a previously published study on hair cortisol levels measured in locally harvested wild muskoxen (Di Francesco et al., 2017).

3.3. Methods

3.3.1. Study area

The study took place in the hamlet of Kugluktuk, in the Kitikmeot region of Nunavut, Canada (Figure 3.1). The population in 2016 was 1,491, the majority of whom are Inuit (approximately 90%) (Statistics Canada, 2016). Muskoxen, caribou, and moose (Alces alces) are the three terrestrial ungulate species found in the area and are harvested by residents of the community. Three types of hunts can be distinguished based on their purpose: (i) subsistence, in which community residents harvest the animal for their own consumption or for sharing among a wider network of kin and community relations; (ii)

36 community, in which local harvesters are hired by the Kugluktuk Angoniatit Association (local Hunters' and Trappers' Organization) to harvest meat for residents in need; and (iii) guided, in which males are selectively harvested for their trophy characteristics by guided hunters (locally referred to as “sport hunts”). Muskoxen are typically hunted by Kugluktuk residents in three different management zones including MX- 07 (Nunavut part of Victoria Island), MX-09 (mainland west of the Coppermine River), and MX-11 (mainland east of the Coppermine River) (Department of Environment, Government of Nunavut, 2018). Since 2014, Kugluktuk has actively participated in a regional hunter-based muskox health monitoring program, which is a partnership among the communities of Kugluktuk, Ekaluktutiak, and Ulukhaktok, government biologists, guided hunting organizations, and academic researchers. Many of the samples analyzed as part of the previously published study on hair cortisol levels in wild muskoxen were collected through this hunter-based sampling program (Di Francesco et al., 2017).

Figure 3.1. Map showing the communities of Kugluktuk, Ekaluktutiak, and Ulukhaktok (NWT) and the town of Yellowknife (NWT).

3.3.2. Study overview

We consulted with the Kugluktuk Angoniatit Association in April 2018 to discuss community interests in this study in the context of the ongoing hunter-based sampling program. We then did small

37 group interviews in January-February 2019 and subsequently hosted validation sessions in January-February 2020. Preliminary results from the small group interviews were presented in January 2020 during the Annual General Meeting of the Kugluktuk Angoniatit Association and final results will be reported in 2021 during the same meeting. Participation in the study was voluntary and interviewees were free to withdraw at any time. Participants received honoraria for their time in agreement with the standards established by the Kitikmeot Inuit Association. The research was approved both by the Conjoint Faculties Research Ethics Board of the University of Calgary (REB16-1214) and the Nunavut Research Institute (Scientific Research License No. 04 002 19R-M). After consulting with the Kugluktuk Angoniatit Association, participants who were fluent in both English and Inuinnaqtun provided translation support during both the small group interviews and the validation sessions.

3.3.3. Interviews

Here, we use traditional Inuit knowledge (TIK) instead of IK as study participants unanimously agreed on this term to refer to their knowledge.

3.3.3.1. Format and participant recruitment

We initially documented TIK through seven semi-structured small group interviews, each involving two to three muskox harvesters who knew each other. We recruited participants with extensive hunting experience and knowledge on muskoxen and the land through purposive sampling, with their inclusion recommended by the Kugluktuk Angoniatit Association and/or the local government wildlife officers, and snowball sampling, with participants identifying additional informants (Green and Thorogood, 2014). Many of the recruited participants were actively taking part in the hunter-based sampling program. To facilitate equal participation, and because the experience, knowledge, and perspectives of participants were likely to differ based on their age, groups of Elders and non-Elders were formed (Green and Thorogood, 2014; Roller and Lavrakas, 2015). All participants included in the Elder groups confirmed that they self-identified as Elders. JD was the study interviewer and either AH or TM assisted her during the interviews.

3.3.3.2. Interviewing process

To document TIK on “the characteristics of a healthy muskox,” we asked the participants “how they recognize that a muskox is healthy” and “what are the characteristics that they look for.” To document TIK on “the factors that affect muskoxen in a positive/negative way,” we first focused the discussion on the negative, and then, on the positive factors. We formulated this question broadly instead of using the term “stress” so as to keep our approach holistic rather than reductionist, and with the goal of capturing a greater breadth of participant knowledge. Consequently, some of the factors presented in the results may not be direct “stressors,” but rather are things that impact muskoxen in a more general and/or indirect way. In this interview phase, we employed different strategies to engage participants and facilitate discussion. As a starting point, each group was asked to show on a yearly calendar when muskoxen calve, mate, and when

38 males fight. For the factors having a negative and/or positive impact on muskoxen, participants were also asked to indicate their monthly occurrence on the calendar and their temporal variability over the broader time-scales of participants’ experiences on a 70-year timeline. During the last section of the interview, participants were asked to interpret the results from a previously published study on hair cortisol levels measured in locally harvested wild muskoxen (Di Francesco et al., 2017). To do this, the interviewer presented, and invited discussion, around each major finding. Findings included: (i) higher hair cortisol levels in males than in females; (ii) higher hair cortisol concentrations in the fall and winter than in the summer, and (iii) yearly variations, with an increase in hair cortisol levels between 2013 and 2015 and no significant difference between 2015 and 2016 (Di Francesco et al., 2017). For the purposes of presenting the research to the participants, the term “stress” was substituted for “hair cortisol.”

3.3.3.3. Analytical framework

All interviews were audio-recorded and then fully transcribed by JD. TM, an Inuk resident of the community familiar with the cultural context and place names, verified the accuracy of all transcripts. These were subsequently analyzed with NVivo software (QSR International, Burlington, Massachusetts, USA) using thematic content analysis, with the purpose to identify themes in the participants’ accounts (Green and Thorogood, 2014). More specifically, holistic and in-vivo coding were first used to identify broad categories, preserving the participants’ words where appropriate, and were then followed by descriptive coding and subcoding, which allowed for the identification of more detailed sub-categories (Saldaña, 2013). To ensure the validity of the coding process, coding was done independently by JD and AH, who then developed a common coding scheme, which was subsequently applied to all transcripts and validated using one of the interviews to verify intercoder reliability (Roller and Lavrakas, 2015).

3.3.4. Validation process

The information documented during the small group interviews was summarized and presented back to the community in validation sessions to confirm, refine, modify, and/or clarify where needed. These sessions included many of the original interviewees, as well as new participants who had not been interviewed previously. Five sessions, each including two to six participants grouped by Elders and non- Elders, and one drop-in session where individuals could come in and provide feedback, were implemented. To rank and determine the relative importance of the characteristics used by harvesters to establish whether a muskox is healthy, we organized a “dot-voting” activity involving all the characteristics that emerged during the small group interviews, and any new ones added during particular validation sessions (e.g., Gittelsohn, 2010; Ferrell et al., 2014). Participants were given 15 dots and instructed to place between zero and three dots next to each characteristic depending on the importance they attributed to it (i.e., zero = low importance → three = high importance). The relative importance of each characteristic then corresponded to the total percentage of dots placed next to it by participants during the dot-voting activities. Some groups added new characteristics during the validation sessions and these were included in the exercise for those groups.

All factors brought up during the initial interviews as affecting muskoxen were presented to the participants for feedback and participants were asked if anything was missing. The same process was followed to validate the findings from the collaborative re-interpretation of the hair cortisol results. To build the final yearly calendars, we presented those summarizing the initial interviews and adjusted where necessary based on validation group feedback.

3.4. Results

Sixteen men and one woman participated in the small group interviews. Ten were Elders (median age = 62 years; range = 51–86 years) and seven were non-Elders (median age = 36 years; range = 22–43 years). Nineteen people attended the validation sessions (two women and 17 men – with one individual attending the drop-in session), of which 14 had participated in the original small group interviews. Ten were Elders (median age = 68.5 years; range = 54–87 years) and nine were non-Elders (median age = 34 years; range = 25–44 years). All participants self-identified as Inuit.

3.4.1. Characteristics of a healthy muskox

Participants indicated that, through their knowledge and experience, they can tell which muskoxen are healthy: “How we recognize are the animals, if they’re healthy or stressed, we learn to read them because that’s how we were brought up. From a young age, we’ve been going out hunting with our Elders […] and, you know, I learned a lot from all the Elders.” Participants identified “external” and “internal” characteristics that they assessed before shooting an animal and during butchering, respectively, to determine if a muskox is healthy. The two most important “external” characteristics identified were body shape (16.5% of dots), indicating that the animal is in good body condition, and good quality skin and fur (16.2% of dots). A healthy muskox has a nice, shiny and darker-colored coat, along with a rounded rump. Harvesters also looked for a large herd size (12.1% of dots) as loners are often (but not always) older and/or less healthy muskoxen that have been left behind from the herd. Other important characteristics were the absence of limping and lumps on the legs (11% of dots) and the animal’s behavior (10.7% of dots). Healthy muskoxen are more alert, react more quickly, and run away when approached. They are also capable of defending themselves and have good body movements. One participant stated that “when they’re really healthy, they prance, they almost prance.” Finally, healthy muskoxen have good speed and endurance (7.7% of dots), whereas “the poor muskox would get behind when running.” Body size and horns were brought up during the small group interviews as an indicator of the types of animals (young animals or medium size females) that have tender meat, but received low priority during the dot-voting for what identifies a “healthy” muskox (4.4% and 3.7% of dots, respectively). While butchering, participants would examine all the internal organs to determine if the animal is healthy; however, they mentioned focusing mostly on the lungs and joints (identified in the small group interviews – 9.2% and 8.5% of dots, respectively), as well as on the meat and liver (added during the validation sessions). They would look for clear fluid when cutting the joints, and would assess the size,

40 color, quantity of lungworms, presence of lumps, and adherence to the ribcage when examining the lungs. Two validation groups mentioned that they also evaluate the color and texture of the liver, as well as the occurrence of small, clear, fluid-filled cysts (3.3% of the dots of these two groups). The appearance of the meat was of very high importance as an indicator of health in one validation group (i.e., 20% of this group’s dots were placed on this characteristic), and participants described observing a gel-like substance in the layers of the meat in really stressed muskoxen. Because liver and meat were brought up later in the validation sessions by only two groups and one group, respectively, they could only be incorporated into the dot- voting for those groups.

3.4.2. Positive and negative factors affecting muskoxen

Participants repeatedly alluded to the “changing world.” Within this overarching theme, we classified the factors that participants indicated as affecting muskoxen into three broad interlinked categories: the “physical environment” that includes weather and climatic factors; the “biological environment” that comprises plants and animals; and the “human-muskox interactions” that encompass anthropogenic elements and disturbances (Figure 3.2). We summarized the timing of key life history events (i.e., calving, mating, and males fighting) as indicated by participants on a yearly calendar (Figure 3.3a). Interviewees reported observations of male muskoxen fighting all year round, with early August to mid-November being a period of increased fighting and competition between males due to mating. During the rest of the year, there were more typically “practice” battles: “There’s a difference between fighting and learning how to fight. These are mostly going to be young punks, […] whereas this [pointing to August to mid-November] is real fighting, mating.” Calving was described from March to mid-June, with the majority of the births observed in April, and then mating from July to mid-November.

Figure 3.2. Factors affecting muskoxen negatively (red), positively (yellow) or both negatively and positively (orange) (muskox drawing by Jayninn Yue).

Figure 3.3. Yearly calendars summarizing interviews and validation sessions showing the timing: of key life events (a); and of factors that negatively affect muskoxen within their physical (b) and biological (c) environments. The four-color gradient represents the percentage of groups that had indicated the month during the interviews and was built as follows: 0% = white, 0% < light color ≤ 33.33%, 33.33% < mid-tone color ≤ 66.67%, and 66.67% < dark color ≤ 100%. If all groups indicated the same information, then only one color instead of three was used (i.e., corresponding to 100%).

3.4.2.1. Physical environment

Interviewees identified that climate warming and precipitation have both positive and negative impacts on muskoxen, while other aspects of the physical environment, including smoke from forest fires and/or volcanic eruptions, high snow depth and/or hardness, slush, muddy terrain, and thawing permafrost, have only negative effects (Figure 3.2). The seasonal timing of some of these factors is illustrated (Figure 3.3b). All participants reported a rise in temperatures, particularly during the summer, since the 1980s- 1990s. Most felt that warmer temperatures increased calf survival; however, one interviewee added that the decreased amount of snow makes it easier for predators to go after muskoxen, especially the newborns: “It probably gives them a better chance because they don’t get the extreme cold temperatures right off the bat when they’re still small and still developing their hide. And they got easier access to food and more growth of better food. So yeah, it might, but at the same time, it’s the perfect chance for a predator too. There’s no deep snow to get left behind in.” Regarding the effect of high heat, some interviewees thought that muskoxen are able to cope fairly well with the heat and that their fur contributes to keeping them cool. They would remain active, continue feeding normally, and generally stay near water sources where the air is windier and cooler. By contrast, other participants believed that muskoxen are sensitive to the heat, their thick and dark fur rendering them particularly vulnerable: “It must suck being a muskox in that thick fur during the summer!” This would cause them to move around less, spend more time laying down, and become easier prey. One interviewee described: “They get left behind, they get all sluggish, they won’t run. Anything could just walk in and get them. They don’t have the energy, they’re sweating too much.” Another participant mentioned that it was the long periods of high heat (i.e., a couple of weeks) that stress muskoxen. A few participants thought that the longer summers and warmer temperatures may confuse the animals and delay the start of the rut, while others had not observed or disagreed with this. Some participants did not express an opinion as they rarely observed muskoxen during the summer. Smoke from forest fires occurring in southern Canada was thought to negatively affect muskoxen. Although none of the participants had directly observed the effects of smoke on muskoxen, many believed that they would be impacted and that their big lungs, particularly when infected by lungworms, would render them vulnerable. Two validation session groups brought up distant volcanic eruptions that would also cause the release of thick smoke and ashes in the air and may affect the vegetation and animals. Several participants thought that high snow hardness, due to temperature fluctuations causing thaw-freeze cycles or mainly to freezing rain (i.e., rain on snow events), negatively impacted muskoxen through difficulties in accessing their food. Such circumstances occur infrequently and their timing, location, and importance varies between years, but their effects are highest if they happen in the fall and/or spring. Several participants did not think that snow hardness impacted muskoxen substantially as they tend to stay away from the highly affected areas and “yeah, it would be harder for them to dig, but that’s a huge animal” and “they are big strong animals, and they know how to manage.” Similarly, several participants mentioned that deep snow caused difficulties for muskoxen to access their food. During validation sessions, one participant

44 added that deep snow may render them more vulnerable to predators by slowing them down when running away because of their short legs. Conversely, some participants felt that it would not affect them because muskoxen typically remain in windblown areas with little snow (i.e., on top of hills). While heavy snowfalls still occur, most participants observed a gradual general decrease in the amount of snow since the 1960s (Elders), and even more so since the early 2000s (Elders and non-Elders). The amount of snow is highly variable between years and also depends on the area, with some locations more affected than others. All participants concurred that reduced snow fall makes it easier for muskoxen to access food. One interviewee mentioned: “They have a better chance winter time around here nowadays. There’s not as much snow as there used to be, so they don’t have to dig as far down to get to their food.” All participants emphasized high variations in rainfall (i.e., “it’s up and down all the time”) between years, with the occurrence of very rainy years. Some mentioned observing a general decrease in rain over the past 10-15 years, whereas others described an increase or no change. According to most participants, only freezing rain has negative impacts on muskoxen. Otherwise, rain was considered positive, relieving them from the heat and “making the land grow,” consequently allowing the animals to get fat. One interviewee pointed out: “It [the rain] helps them lots. It keeps the bugs down, keeps them [muskoxen] cool. They can have water anywhere, they don’t always have to go find water at a river, there’s puddles all over the place, they can keep hydrated. It keeps better growing for the plants, so they got more food, it’s easier to get.” Many participants described the “land drying up” since roughly the 1990s, with lakes, creeks, and ponds drying out, and river levels going down. Victoria Island was more significantly affected by this phenomenon than the mainland. With respect to Victoria Island, one interviewee observed: “There are places where you go and it’s like walking on gravel, it’s so dry.” Factors contributing to the “land drying up” included a general decrease in snow, a marked increase in permafrost thaw, longer summers and higher temperatures, as well as a low amount of rain; once again, there are substantial seasonal and annual variations. The impacts of the “land drying up” on muskoxen are mostly through the quality and abundance of the vegetation (see section 3.4.2.2), but they are also sometimes forced to move away. For example, over the past decade, multiple interviewees observed migrations of large muskox herds in early spring before the sea-ice starts melting from Victoria Island to the mainland. They connected these migrations to the vegetation on the mainland, which is better both in quantity and quality compared to Victoria Island. A few participants thought that a large amount of slush during snow melt causes muskoxen to slow down or get stuck and to sometimes even die of exhaustion, but this happens only on rare occasions and mostly to weaker and older animals. Two validation groups also linked muddy terrain to areas with permafrost thawing. Such terrain is difficult for muskoxen to maneuver and may cause them to get stuck. One participant shared his observation of a likely sick muskox who got stuck in the mud and ended up dying. One validation group added that global warming has caused the permafrost to thaw and the release of methane in the atmosphere. Participants were concerned that this might negatively affect the lungs of muskoxen.

3.4.2.2. Biological environment

Aspects of the biological environment, such as predators and vegetation, were identified to have both positive and negative impacts on muskoxen, while others, including low caribou abundance, insect harassment, and lungworms were thought to have a negative effect (Figure 3.2). The timing of predator and insect activity is illustrated in Figure 3.3c, but the timing of the other negative factors, such as poor vegetation and lungworms, was not discussed in depth. Predators, including wolverines, wolves, and grizzly bears, had both positive and negative impacts on muskoxen. Participants unanimously observed no change in the abundance of wolverines over time, but noted an increase in the abundance of both wolves and grizzly bears since the 1980s-1990s, which was more evident on Victoria Island than on the mainland. Interviewees also observed an increase in the number of cubs per grizzly bear sow over the past decades. Predation on weak and diseased muskoxen was considered positive as it would limit the spread of diseases throughout the herd. However, mortality caused by predation was significant, as was the stress caused by predators through chasing and scattering the animals and herds. Participants indicated that wolverines, wolves, and grizzly bears are capable of killing any muskox, even very healthy ones, but they generally favor “easier” prey such as late term pregnant females, and young, old, injured, or diseased muskoxen that fall behind when the herd is chased, as well as lone animals. Wolverines and wolves were consistently observed year-round, whereas grizzly bear activity was restricted because of annual hibernation (Figure 3.3c). Grizzly bears are mainly active between April and October with most hibernating by November, excluding a few who are not fat enough and hibernate later. One participant told the story of a very slim bear that killed two dogs at the edge of Kugluktuk in November and another mentioned: “A big bear who doesn’t have enough to eat yet, he still needs to fatten up some, so he’d wander around a little bit longer than everybody else.” While some odd grizzly bear sightings were reported between December and March, these would be very rare occurrences, probably corresponding to very hungry bears. Finally, one participant added that: “Highest activity would be April-May and September-October. When they first come out, they come out almost all at the same time, few weeks apart, so you’re gonna have a big flood, and they’re all starving right? And then just before they’re gonna go back to sleep, they’re going to pack on as much pounds as they can, and that’s all of them.” One interview group suggested that changes in plant species may have contributed to the muskox declines on Victoria Island: “other plants are growing more and then they’re just not right for the muskox diet. […] The vegetation there [Victoria Island], it’s probably from that, they might be starving and getting sick from other plants that their bodies are not used to.” A few participants described changes in plant species on the mainland and expressed their worry that some of the newly established plants may be invasive and take over native species in the future. However, most participants in validation sessions did not express an opinion (i.e., this is not something they have been paying attention to) or thought muskoxen would only eat plants that they like: “If muskoxen don’t like the other plants, they won’t eat them, they’ll only eat what they like to eat.” People would, however, “just have to wait and see” what impacts changing vegetation will have on muskoxen.

The “land drying up” was raised multiple times. Participants voiced that this negatively impacts plant growth and, consequently, negatively affects muskoxen through decreased food availability, abundance, and quality. However, a few participants believed that the impact on muskoxen would be limited as they would be able to find food elsewhere, in less affected areas such as along river banks or around wetlands. Several participants mentioned that Victoria Island, described by them as having poorer vegetation than the mainland, was more significantly impacted by the “land drying up.” Conversely, longer summers could lead to longer periods of high vegetation growth, depending in part on the amount of rainfall, which increases both the abundance and quality of the vegetation. This positively affects muskoxen by giving them “a lot more to eat for a lot longer” and “they would get fatter because they’re grazing a lot more.” The relative importance of opposing impacts of climate warming, i.e., “land drying up” versus a longer growing season and better vegetation quality and quantity, varies depending on the year, area, and weather conditions (amount of rain, temperature, etc.). Declines in caribou abundance were thought to have negative impacts on muskoxen, which are harvested more intensively to compensate for caribou scarcity. However, one interviewee commented that muskoxen would not be significantly affected because their harvesting is regulated via a tag system administered through the Nunavut Government. Lungworms were thought to reduce lung capacity and endurance of muskoxen, rendering them more vulnerable to predators and making it more difficult for them to search for and find food. Insects in general, and mosquitos in particular, were discussed by all participants. Diversity of insect species has increased since the 1980s-1990s, possibly due to a combination of climate warming (i.e., increasing temperatures and longer summers) and airplanes and sea lifts bringing up new species. One participant had worked with a biologist during the previous summer and trapped hundreds of different fly and mosquito species, many of which were previously unknown in the area. While most interviewees described no change in the abundance of insects but strong yearly variations, a couple participants thought there had been an increase over the past decades. Several people also mentioned that the duration of insect activity was getting longer. A few participants had difficulties expressing an opinion regarding the effect of insects on muskoxen as they rarely observed muskoxen during the summer. However, some mentioned that muskoxen are less harassed by insects and less sensitive than caribou because of their thick skin and long fur that protect them: “they don’t seem too bothered by them [insects] because of their long hair” and “I don’t think they do anything to the muskox. Cause you got, you know, a good 6 to 8 inches of hair, and underneath that you got a good thick wool, that thing might overheat just trying to get in there! Probably knows it too!” Muskoxen are, however, affected on their face and legs, where the hair is shorter, and younger animals might be particularly sensitive because of their thinner skin and hair. Muskoxen were rarely seen running from insects, even when these are highly abundant, but sometimes muskoxen would stay in rivers to avoid insect harassment, and summer days with no wind exacerbate increased harassment. Insect harassment can cause moderate restlessness with animals

47 moving more and eating less. One participant concluded that the herd would generally not be affected and remain calm, but that the “odd” muskox may be seen running around “sick and tired of bugs buzzing in [its] ear.”

3.4.2.3. Human/muskox interactions

Breakdown of Inuit knowledge transmission, non-renewable resource exploration and development, hunting and other anthropogenic disturbance were identified to negatively impact muskoxen; conversely, hunting regulations were identified to have a positive effect (Figure 3.2). Participants felt that the lack of knowledge transmission with respect to hunting and butchering led to people sometimes “not knowing what they’re doing” and engaging in improper harvesting practices, and this had a negative impact on muskoxen. These practices include: (i) not approaching muskoxen properly and chasing them more, causing increased disturbance and stress; (ii) unfamiliarity in identifying healthy muskoxen, with abnormalities in the organs and meat, and in preparing meat and organs to ensure food safety – these knowledge gaps may result in meat wastage; and (iii) not having appropriate training on siting in firearms and targeting the right animal for harvest. These improper hunting practices, although infrequent, occurred not only with youth, but also adults. Interviewees thought that youth/Elder and inexperienced/experienced harvester programs that promote the sharing of knowledge and cultural traditions among generations could address this breakdown in knowledge transmission. Some participants were worried about air, ground, and water pollution, while others believed that pollution was limited in the Arctic, and still others had no particular opinion on the matter. Mining activities were of particular concern for several participants. The chemicals released in the environment, as well windborne dusts and fumes, were thought to degrade vegetation and to contaminate lakes and other water sources, with a consequent decrease in water quality. Participants believed that all wildlife species would be affected by pollution from mines although they were unsure of the specific impacts. One participant who used to work in the mines described that “sometimes, there would be a huge yellow cloud and the wind carries it, I don’t know how far, but I believe it affects most animals,” and another interviewee mentioned that “anything with mines will affect the wildlife one way or another.” A few participants suggested that because mining activities were done on a small scale, the adverse effects would be very limited. Those who had observed muskoxen around the mines reported animals as generally scarce, but also described a recent increase in their abundance in proximity of the mines. Additionally, participants believed that the pollutants released would be carried over long distances and affect muskoxen even if they were located further away. Several participants considered hunting as causing disturbance to muskoxen. Sport hunting in particular was mentioned to lead to the loss of “prime” bulls whose genes would potentially produce strong and healthy progeny: “They always take the champ. And you kind of want the champ to be the one to pass on the genes, strong, you know good line.” While most participants agreed with this effect, a few had a more reserved opinion on the matter, as (i) there would always be other males to take over, especially if the population is abundant; (ii) the males with the biggest bosses and horns targeted by trophy hunters may actually be quite old, not

48 reproducing anymore, and not necessarily the healthiest; and (iii) the number of males harvested for trophy would generally be low and their loss would consequently have a limited impact on the population. Several participants thought that subsistence and community hunting caused the loss of young animals and possibly pregnant females. Most interviewees agreed with this impact and one suggested the implementation of seasonal hunting regulations which would ban people from harvesting female muskoxen around calving time. However, a few had reservations and believed that when the population is abundant and healthy, this loss would not have a significant effect especially with the number of muskoxen hunted regulated through the tag system. One participant mentioned that it would even have a positive effect by keeping the population at a reasonable number and preventing overgrazing of the land. Additionally, some Elders consider muskox fetuses as a delicacy, and thus there would be very limited meat wastage if pregnant females were harvested. Participants of the first validation session added helicopter activity and the resulting noise pollution to the factors negatively affecting muskoxen and most interviewees subsequently agreed with this addition. Helicopters were described as very noisy, consequently disturbing and scaring muskoxen, and causing them to run away. Conversely, a few participants did not believe this impact to be significant; one person mentioned: “I’ve been in a helicopter and flown by them. They will just run a little bit, and then they’ll stop and resume what they were doing, feeding and what not.” More generally, interviewees thought that airplanes flying low over the land would affect muskoxen, but these were described as rare occurrences. Seasonal patterns of disturbance were further explored during validation sessions. Snowmobile activity generally begins between October and December and lasts until May or June, depending on the timing of freeze-up and break-up, respectively. Airplanes operate year-round, while helicopters, which are mainly linked to exploration and mining activities, are observed primarily between April and September. All-terrain vehicles (ATVs) are used only during the summer months, and most interviewees believed that they would not cause major disturbance to muskoxen as people can’t travel very far with them. Moreover, muskoxen are generally scarce in the areas accessible by ATVs. One participant indicated that when he comes across muskoxen on Victoria Island, where it is easier to travel by ATV because the land is flatter: “Unless we’re trying to get close to them, they don’t even really bother with us. They’ll stop and just watch us go by, and carry on.” Participants indicated that anthropogenic disturbance is lowest between June and September, as airplanes, helicopters, and ATVs generally cause lower stress than snowmobiles, whereas it is highest during periods of high hunting activity, in November-December and March-April, when a lot of people are out on the land with snowmobiles. All participants agreed that measures favoring predator management such as increased hunting incentives and reduced regulations (e.g., trapping allowed all year round and no tag system) were positive for muskoxen as they reduced predator abundance and consequently predation. Similarly, all participants concurred that muskox hunting regulations (i.e., tag system) were positive as they controlled the number of muskoxen hunted and kept it at a reasonable level. However, some interviewees suggested that when the population is abundant, these regulations could be lifted for subsistence and community hunting.

3.4.3. Collaborative re-interpretation of previously published stress results

We asked the participants to discuss the results from a previously published study on locally harvested wild muskoxen that showed differences in hair cortisol levels across sex class, seasons, and years (Di Francesco et al., 2017). Throughout the discussions, hair cortisol levels were colloquially referred to as “stress levels.”

3.4.3.1. Sex differences

Di Francesco et al. (2017) reported higher hair cortisol levels in males than in females. None of the seven groups were surprised by this finding and all groups attributed this in a large part to reproduction (i.e., rut and mating). In particular, rut was considered a time of high stress with some bulls getting killed during fights, and others dying from exhaustion and starvation afterwards, as they would have been eating less and expended all their energy and fat stores by the end of the rut. Additionally, bulls were thought to be weaker and more vulnerable to predators after the rut. While all participants agreed, one interviewee thought that reproduction was “normal” and that stress during that period would mainly be due to injuries. During a validation session, one person not previously interviewed mentioned being surprised by the sex differences measured and believed that female muskoxen may be stressed by their role of raising the young. Three groups mentioned that males had the important role of protecting and taking care of the herd. They represent the “first line of defense” against predators and other threats and have to find food for the rest of the herd, i.e., “Once the male has established the herd, then he’ll take care of them and they don’t have to worry about finding food, he’ll find them food.” The role of protection was all the more emphasized with the increase in the abundance of wolves and grizzly bears observed and would not be restricted to the dominant male but would also involve the younger males. However, one participant felt that all adult muskoxen, and not necessarily the males only, contribute to protecting and taking care of the herd. Interviewees also thought that if a herd is split up (i.e., by harvesters or predators), it would be particularly stressful for the male as he would lose some of the females he was responsible for protecting. Several participants identified sport hunting as a male stressor because sport hunters generally try to find bachelor herds with only bulls, although it was also highlighted that the male groups are not the only ones bothered and females would not be aware that they’re not the target. One interviewee illustrated this viewpoint: “It doesn’t matter how many males or females you got in a herd, as soon as they see danger or sense danger, everybody comes together, in a circle. It could be all males, it could be all females, it doesn’t matter […]. No matter what, no matter who’s hunting, sports hunters or me, we’re all going to have the same stress on that animal. Cause they don’t know what you’re coming for, you come to kill us all, or are you just here for me? They got no concept of that. […] It’s all a predator no matter what it is, everybody’s in danger, at all times, so it’s the same stress level.” Some interviewees also thought that male muskoxen were more sensitive to the heat than females because they move around more to protect and find food for the herd. However, while most participants agreed with this observation, others did not express an opinion or disagreed as “they move around more, but slowly.” One participant added that “the higher heat would affect the male more cause he’s all worked up too.”

Factors more specific to younger males were also brought up. Young males get chased away from the herd more frequently and consequently have a harder time finding a good place to feed. This generally occurs only during the rut: “When you get into the herd, there’s never only one male, there’s lots, and there’s young ones, and there’s big ones. They don’t really get kicked out per se, only rutting time.” Young males also have to search more for food to respond to their higher nutritional needs due to their rapid growth compared to females. Finally, they tend to fight more frequently than older bulls.

3.4.3.2. Seasonal differences

We then asked participants to discuss the previous findings of lower hair stress levels in muskoxen in the summer than in the fall and winter (Di Francesco et al., 2017). One of the seven groups was surprised by this finding, and summer stressors, including heat stress, insect harassment, and predators were identified by several interview groups. Helicopter activity was unanimously added to the list during the validation sessions and airplanes and ATVs, low air quality (from pollution, forest fire smoke, etc.), poor vegetation (which would depend highly on the year), and muddy terrain were added by some of the validation groups. Specific to the fall, all participants identified reproduction as the main stressor (i.e., rut and mating, including injuries from fighting which may take a while to heal). Participants suggested various factors to explain the higher stress levels observed during fall and winter. These included: (i) lower food availability, quality, and accessibility (4 groups), as muskoxen have to break through the snow and/or dig to access their food, and in some places the snow may get very hard because of freezing rain events and/or temperatures variations; (ii) higher harvesting pressure and human disturbance (3 groups) as these are the main hunting seasons and access to the land is easy using snowmobiles (i.e., “It’s mostly in the winter time cause we have access, we can travel by snowmobile to them. We can go anywhere.”); and (iii) higher pressure from predators (1 group) as they have access to a lower diversity of prey than in the summer.

3.4.3.3. Yearly variations

Discussions around yearly variations (i.e., increase in stress levels from 2013 to 2015 and no significant difference between 2015 and 2016) were more challenging as participants had difficulties recalling events that occurred during specific years. Interviewees did, however, discuss various factors that could contribute to interannual differences in stress levels. These included variations in competition between muskoxen, predator numbers, weather conditions, and human disturbance. Some interviewees indicated that competition between muskoxen mainly takes place during the rut and increases with the number of males in a herd as well as with the general abundance of muskoxen: “because then a big male wants to keep his harem, he’s gonna fight a lot more than he would if there was less [male muskoxen].” During the validation sessions, all participants unanimously agreed with these factors, and one group added the importance of annual variability in air quality, which would be impacted, for example, by mining activities, forest fire smoke, and/or volcanic eruptions.

3.5. Discussion

In this study, we documented TIK to identify and gain a better understanding of the stressors affecting muskoxen. We employed a three-pronged approach during the small group interviews to: (i) establish the characteristics of healthy muskoxen; (ii) determine the factors that impact muskoxen; and (iii) collaboratively re-interpret a study on the sex, seasonal, and yearly patterns in muskox hair cortisol levels. Key outcomes include (i) a more holistic understanding of muskox health and what it encompasses; (ii) recognition and exploration of a rich One Health perspective expressed by participants around factors influencing muskoxen in a “changing world” and highlighting the multiple socio-ecological connections; and (iii) a broader comprehension of the glucocorticoid (stress) patterns measured in muskox hair, the various factors that influence them, and their interrelations. Further emanating from this study is: (i) the importance of bridging SK and IK, and how the Inuit perspective reflected a One Health approach in all the sections of this study; (ii) a meaningful advancement in the process of actively involving communities at all steps of the research and not only in data collection; and (iii) advances in how SK and IK can contribute to informing various indicators of health and their measurable variables to be used in quantitative species status assessments.

3.5.1. Characteristics of a healthy muskox

A primary goal of this research was to gain a more holistic understanding of muskox health through documenting IK. Recent work has focused on updating and broadening the definition of wildlife health, and it now encompasses not just the absence of disease, but also the concepts of vulnerability, resilience, and population sustainability (Hanisch et al., 2012; Stephen, 2014). However, examples of the inclusion of IK holders in developing this definition are scarce. As an illustration, two recent studies used the Delphi method, which aims at soliciting and subsequently summarizing the knowledge of a group of experts, to establish a definition of wildlife and polar bear health, respectively (Hanisch et al., 2012; Patyk et al., 2015). However, there was no explicit connection to Indigenous worldviews and all experts who participated in these studies were identified by affiliation to government agencies, research or academic institutions, or non-profit organizations (Hanisch et al., 2012; Patyk et al., 2015). Indigenous peoples’ concept of health is relatively well established in human medicine. Rather than focusing on the individual, it generally includes the health of the entire community and its surrounding ecosystem, while incorporating a multitude of dimensions (i.e., social, physical, psychological, spiritual, and ecological) (Stephens et al., 2005; Janska, 2008). This holistic view of health extended to wildlife was highlighted in our study by the multiplicity and wide diversity of the characteristics brought up by the participants when we asked them to identify what enables them to establish if a muskox is healthy. Indeed, these encompassed “external” and “internal” elements, physical and behavioral traits, and individual and herd features. Additionally, some of these characteristics were intrinsically broad, such as the muskox’s behavior, which comprises a variety of aspects including their attitudes, reactions, movements, and capacity to defend themselves. Stephen (2014) emphasizes that the health of wildlife species results from the

52 cumulative effect of multiple biological, social, and environmental factors that act on individuals and populations, and affect their capacity to cope with change (Stephen, 2014). This integrated definition resonates strongly with, and could be further informed by Indigenous ways of knowing, as IK holders simultaneously examine animals and their complex and changing environment, including its human dimensions (Kutz and Tomaselli, 2019).

3.5.2. Factors affecting muskoxen

The holistic perspectives strongly expressed through identifying what characterizes a healthy muskox were continued throughout subsequent discussions on the factors that affect muskoxen. The TIK documented once again illustrated a One Health perspective relevant to this question, with the various factors influencing muskoxen discussed simultaneously and their multiple interconnections emphasized. An overarching theme that emerged from the interviews was that muskoxen are living in a “changing world” (Figure 3.2) and within it, they are highly influenced by a multitude of interrelated elements across a variety of spatial and temporal scales, including weather and climatic factors, animals and plants sharing their ecosystem, as well as the “human dimensions” of the world they live in. The Intergovernmental Panel on Climate Change (IPCC) clearly documents changes in the global climate (IPCC, 2014); the impacts of those changes on muskoxen, and more broadly, on the “physical,” “biological,” and cultural environment in the Arctic, were articulated in rich detail by the participants. Key concerns highlighted by participants included the direct and/or indirect effects on muskoxen of (i) climate warming, altered precipitation patterns and drying out of the tundra, smoke from forest fires and volcano eruptions, high snow depth and/or hardness, slush, muddy terrain, and thawing permafrost within their “physical environment”; (ii) increasing insects and predators, shifting plant communities, poor vegetation growth and quality, and low caribou abundance within their “biological environment”; and (iii) breakdown of Inuit knowledge transmission, non-renewable resource exploration and development, along with hunting and other anthropogenic disturbance within the “human-muskox interactions” (Figure 3.2). Factors thought to have a positive impact on muskoxen included (i) climate warming and altered precipitation patterns within their “physical environment”; (ii) predators and high vegetation growth and quality within their “biological environment”; and (iii) regulations within the “human-muskox interactions.” It is important to note that since we decided to formulate our questions broadly for the purposes already indicated in section 3.3.3.2, some of the factors discussed during the interviews and summarized in Figure 3.2 are not direct “stressors” of muskoxen, but rather are elements that would impact them in a more general and/or indirect way. We intentionally decided to retain inclusion of the “indirect stressors” in order to address the breadth of knowledge documented and not revert to a reductionist approach. Also, this decision is responsive to discrepancies between IK and SK ontologies, thereby creating space for diverse ways of knowing (Wilson, 2008; Kovach, 2009). These “indirect stressors” included (i) the breakdown of Inuit knowledge transmission on good harvesting and butchering practices which generate meat wastage; (ii) regulations regarding predator management through their effects on predator abundance;

53 as well as (iii) muskox hunting regulations (i.e., number of tags, tag allocation to sport versus subsistence/community hunting, seasonal regulations, etc.) and caribou abundance which control and influence the number of muskoxen harvested, respectively. Participants’ observations and perceptions regarding the factors that affect muskoxen often concurred with, and expanded on, the current scientific understanding of Arctic ecology, but also generated novel insights and new questions. While excessive and prolonged heat was considered an important factor having a direct influence on muskox health and well-being, muskoxen were also able to behaviorally mitigate this stressor by reducing their activity level and remaining in windier areas. The scientific literature on the effects of heat on muskoxen is sparse, with only a single study linking heat to an outbreak of fatal pneumonia in the muskox population of Dovrefjell in Norway (Ytrehus et al., 2008, 2015) and the effects of high heat on muskox behavior and physiology otherwise remaining poorly understood (Kutz et al., 2017). Further research bringing together both IK and SK is needed to investigate the interactions of warming temperatures with other possible stressors and their resulting cumulative effects on muskoxen. Some of the hypotheses generated in this study could be explored through targeted SK studies, as has previously been done for brucellosis in muskoxen (Tomaselli et al., 2018b, 2019). Increased insect diversity, changes in insect abundance, and longer periods of activity were reported by the participants. Similar findings have been described through other IK studies in the Canadian Artic (Huntington and Fox, 2005; Prno et al., 2011). Of particular note was that participants had observed the effects of insects on muskoxen. While the impacts of insect harassment are well documented in caribou (e.g., Witter et al., 2012; Raponi et al., 2018), only one study has investigated these in muskoxen. Jingfors (1982) found evidence that muskoxen increased the proportion of time spent walking and standing, while they decreased the proportion of time spent feeding when insect harassment level was high (Jingfors, 1982). Participants, therefore, provided valuable descriptions and unique insights regarding the negative effects of insect harassment on muskoxen, which based on their accounts, are probably limited but non-negligible. Participants described an increased wolf and grizzly bear abundance, which was more evident on Victoria Island than on the mainland. Similar observations, and particularly the northward range expansion of grizzly bears into the Canadian Arctic Archipelago, have been reported in other SK and IK studies (SARC, 2013, 2017; ECCC, 2018; Tomaselli et al., 2018b). While participants emphasized the negative effect of predators through the disturbance of muskoxen and direct mortality, particularly of the more vulnerable individuals, they also highlighted that predation limits the spread of diseases throughout the herd. This role of predators in the control of infectious diseases has been discussed in various scientific studies (Packer et al., 2003; Hall et al., 2005; Wild et al., 2011) and was previously suggested by Kugluktuk IK holders: “Wolves keep the caribou in good health. If there would be no wolves, there would be lots of sick caribou” (Dumond, 2007:17). Participants indicated a general decrease in the amount of snow. While this had a positive effect on muskoxen through easier food accessibility and availability, and decreased vulnerability to predators, it also contributed to the land getting dryer, particularly on Victoria Island. Similar observations of enhanced drying of local lakes and rivers with decreased water levels have been documented through other IK studies

54 in Nunavut and the Yukon (Huntington and Fox, 2005; Wrona et al., 2005). This phenomenon of the “land drying up” indirectly negatively affects muskoxen mainly by decreasing the vegetation quality and abundance, but also by forcing them to move away. Multiple participants recounted muskox migrations from Victoria Island to the mainland in early spring. These accounts are further supported by the findings of a recent study analyzing muskox gut microbiome and microsatellite data, which demonstrated that a few of the muskoxen sampled on the mainland clustered with those from Victoria Island (Bird et al., 2019). Participants also reported the negative effects of high snow hardness and snow depth on muskoxen through reduced food availability and accessibility. The frequency of rain on snow events is predicted to increase in the Arctic (Hansen et al., 2011; Langlois et al., 2017), so their impacts on muskoxen will likely become more severe in the future. The complexities of the interactions between altered precipitation (i.e., both rain and snow), warming temperatures, thawing permafrost, and other consequences of climate change, as well as the diversity of their effects on the land, vegetation growth, and other animals (i.e., insects and predators) were highlighted throughout the participants’ narratives as they considered all of these factors simultaneously and comprehensively. As current climate change trends are predicted to continue (IPCC, 2014), it is likely that the direct and/or indirect impacts of these various factors on muskoxen will intensify in the future. Poor air quality, whether it be due to pollution from non-renewable resource extraction, or to smoke from distant volcano eruptions or forest fires further south, was a major concern for the participants as it negatively affects muskoxen by directly and indirectly impacting their lungs and environment, respectively. Among other anthropogenic factors, participants mentioned a negative but probably limited impact of sport hunting on muskoxen through the loss of “prime” bulls with valuable trophy phenotypic characteristics (i.e., big boss and horns). This recalls the findings of scientific studies in bighorn sheep (Ovis canadensis) and mouflon (Ovis gmelini musimon), where unlimited sport hunting resulted in the decline over time in those populations of some of the heritable traits trophy-harvested rams were selected for (e.g., body and horn size) (Coltman et al., 2003; Garel et al., 2007). Participants also emphasized the negative impacts of human disturbance in general (i.e., due to the use of various vehicles and to hunting activities), which was further enhanced by the breakdown of Inuit knowledge transmission on proper hunting practices that may lead to an incorrect approach of the animals. These observations, that not approaching muskoxen properly generates higher stress, echo the findings from a review study of ungulate flight responses to human disturbance, which found that ungulates flee at greater distances when they are approached in a more threatening manner by humans (i.e., directly or rapidly) (Stankowich, 2008). All of the negative impacts of human disturbance on muskoxen mentioned by the participants are also likely to intensify in the future with increasing anthropogenic activities in the Arctic (Post et al., 2013; AMAP, 2017). The content of IK is inextricably linked to its spatial context (Berkes, 2008; Wilson, 2008; Kovach, 2009). This was exemplified in this study by the way participants detailed the spatial context of their knowledge in addition to muskox-specific content. For instance, they described different phenomena

55 depending on their areas of observation, which generally corresponded to where they had lived and travelled. This connection between context and content was further highlighted by some of the differences specified between Victoria Island and the mainland around Kugluktuk regarding environmental changes, their severity, and their impacts on muskoxen. Variations in the observations and analyses of participants from the same community reflect the context and relationality that led to the creation of the knowledge (Wilson, 2008; Kovach, 2009; Maxwell and Mittapalli, 2010). These variations also prompt us to pay close attention to IK areas of observation and to consider individual-specific nuances in IK when trying to abstract to the community-level (Armitage and Kilburn, 2015; Martinez-Levasseur et al., 2017). Because of logistical and financial constraints, we were only able to include Kugluktuk IK holders in this study. However, this research would have greatly benefitted from involving all the communities participating in the muskox health monitoring program to capture a broader range of reflections and observations, and maybe even identify other factors, which may have been missed because they were not occurring around Kugluktuk. Since muskoxen are typically non-migratory animals (Gunn and Adamczewski, 2003), the content-context link suggests that Kugluktuk-specific observations would well represent the muskoxen around Kugluktuk for the seasons that harvesters interact with the animals. However, participants in this study indicated that muskoxen were emigrating from Victoria Island to the mainland, adding another component that requires consideration. Hanke et al. 2020 emphasize the importance of including different community perspectives for understanding the Dolphin and Union caribou herd. In their study, IK from Kugluktuk and Ekaluktutiak reflected two different states of the caribou herd, where harvesters from Ekaluktutiak described a stable caribou abundance with healthy animals that were close to the community in 2003 while harvesters from Kugluktuk described a declining caribou abundance with sick animals located far from the community in 2003. They conclude that, since one community’s IK only represents a portion of the herd for a specific seasonal period, information from communities and IK holders throughout the range needs to be considered in order to develop a herd-level understanding (Hanke et al., 2020). Similar for muskoxen, a multi-community study would be useful to further understand their health, ecology, and behavior across different landscapes.

3.5.3. Collaborative re-interpretation of previously published stress results

In the last part of this study, we collaboratively re-interpreted with TIK holders data on the sex, seasonal, and yearly variations in hair cortisol levels of wild muskoxen. This work of collaboratively interpreting physiological stress results with IK holders is, to our knowledge, unique and novel in the field of wildlife endocrinology research. Kutz and Tomaselli (2019) highlight that combining and contrasting findings from different knowledge systems provides greater insights as they can, among other things, compensate for each other’s uncertainties (Kutz and Tomaselli, 2019). In the published study, results were discussed using the available SK on wildlife endocrinology and muskoxen in general (Di Francesco et al., 2017). Participants identified additional explanations and hypotheses for the patterns observed and supported and complemented

56 previously recognized stressors. They also specified, through their holistic view and close intimacy with their environment, the complex interconnections between the various factors mentioned. For example, regarding seasonal variations (i.e., lower stress levels in the summer than in the fall and winter), heat extremes and insect harassment were identified both through the scientific literature review and by the participants as possible summer stressors. Interviewees, with the depth and diversity of their accounts and observations, provided invaluable information regarding the actual level of the effects that these two factors have on muskoxen, the strategies of muskoxen to cope with them, and their implications for the future as they intensify with increasing climate warming. Additionally, poor air quality from pollution and/or smoke, muddy terrain, and anthropogenic disturbance from helicopter, airplanes and/or ATV activity were factors identified by participants but not considered in the peer reviewed publication. As for the fall and winter, reproduction (i.e., rut and mating) in the fall, the higher harvesting pressure and human disturbance, as well as the lower food availability, quality, and accessibility were discussed both in the published paper and during the interviews as possible explanations for the higher stress levels measured during these two seasons. Once again, participants provided additional information on the level of the impact that these stressors have on muskoxen and its timing, as well as on their interconnections with other factors (see Figures 3.2 and 3.3). The higher pressure from predators was a new factor added by the participants. Participants identified several stressors in summer and were surprised that these were not reflected in the summer hair stress levels. However, it is important to note that the time spent by the proximal part of the hair in the follicle before it surfaces from within the skin is unknown in muskoxen. Consequently, late summer stressors (e.g., increased fighting between males and mating) may mainly be measured in the hair collected during the fall. Findings from the interviews allowed us to gain a broader understanding of sex specific stressors and the functions, particularities, and behavioral characteristics of male muskoxen that may lead them to having higher stress levels. Additionally, the fact that none of the participants were surprised by these sex differences increased our confidence in the published results, which were based on limited and unbalanced sample sizes. Finally, for yearly variations, findings from the interviews supported and expanded on the discussion from the published study. Our study is an example of how IK can inform quantitative species status assessments and conservation measures. As suggested by Peacock et al. (2020), the IK we documented enabled us to identify relevant indicators of muskox health and measurable variables that could be used to inform these indicators (Peacock et al., 2020). Potential indicators, which are analogous to “determinants of health” and emanated from the participants’ accounts, include multiple extrinsic factors such as indexes of helicopter traffic and snowmobile activity, measures of plant abundance and quality, indexes of insect harassment taking into account both the abundance of insects and the duration of their activity, measures of air quality, etc. Future work will focus on developing such indicators and their measurable variables, using both SK and IK, so they can be used to further understand individual and/or population-level hair cortisol concentrations. Our aim through this research is to strengthen the integration of IK and SK to gain a better understanding of the stressors that affect muskoxen. Danielsen et al. (2009) propose a five-category

57 classification of natural resource monitoring programs based on their degree of community involvement, and ranging from “externally driven and professionally executed,” without any participation of local stakeholders, to “autonomous local monitoring,” without any direct involvement of external agencies (Danielsen et al., 2009). The hunter-based muskox health monitoring program, from which the muskox hair samples were obtained, for a long time fit in Danielsen et al.’s third category of “collaborative monitoring with external data interpretation, which involves local people in data collection and management-oriented decision making, but the design of the scheme and the data analysis are undertaken by external scientists” (Danielsen et al., 2009:33). Indeed, the concerns around muskox health were raised by the communities, and the sampling kits were designed in collaboration with local harvesters to allow for the collection of a standardized set of samples and were adapted to the extreme field conditions encountered in the Arctic (Tomaselli, 2019). However, all sample and data analyses were conducted by researchers with results regularly brought back to the communities for further discussions. Our current study, in addition to the training of an Inuk resident of the community (TM) in sample analyses, moves this program closer to Danielsen et al.’s fourth category, corresponding to “collaborative monitoring with local data interpretation, which involves local stakeholders in data collection, interpretation or analysis, and management decision making, although external scientists may provide advice and training” (Danielsen et al., 2009:33–34). Ethical and successful bridging of SK and IK, allowing for this combination to reach its full potential, should engage IK holders across all stages of the research, from study design to data interpretation and validation (Ban et al., 2018; Bélisle et al., 2018). Furthermore, the Nunavut Land Claims Agreement Act led to the establishment of the Nunavut Wildlife Management Board, a co-management board mandated to equitably consider all available SK and IK in their wildlife management recommendations (Indian and Northern Affairs, 1993). Transitioning northern research to categories four and five on Danielsen et al.’s hierarchy is congruent with calls from the Nunavut Research Institute and Inuit Tapiriit Kanatami (Inuit Tapiriit Kanatami and Nunavut Research Institute, 2006; Inuit Tapiriit Kanatami, 2018). It also creates the conditions for fruitful and durable collaborations, and in the case of this program, positions it to more effectively achieve the goal of promoting the health and sustainability of muskoxen for the communities that depend on them, and strengthens the co-management process. Our study has addressed three points integral to transitioning into a more collaborative program: (i) documenting TIK to determine the characteristics of a healthy muskox and identify the stressors affecting them, (ii) collaboratively re-interpreting the results from a related SK study, and (iii) partnering with an Inuk resident of the community (TM) in the interviews and data analyses. We have done this by giving a platform to the participants to speak about muskoxen, including many harvesters who are actively involved in the hunter-based sampling program. Also, our use of small group interviews and validation sessions acted as another opportunity for exchange and co-production of knowledge among the various parties participating in the project. For example, through the process, JD responded to interviewees’ questions on sample analysis methods and other study findings, while TIK holders shared their detailed observations, including possible explanations and hypotheses, to inform the interpretation of results.

Finally, TM, who co-coordinates the hunter-based sampling program in Kugluktuk and does the initial sample analyses, also participated in the interviews and data analyses.

3.6. Final note

This study highlights the importance of bridging SK and IK. The holistic One Health perspective expressed by the TIK holders allowed for a broader and deeper comprehension of the stressors affecting muskoxen, their complex interconnections, and how these contribute to the sex, seasonal, and annual patterns observed. It also represented a meaningful advancement in the process of transitioning the muskox health monitoring program to the “collaborative monitoring with local data interpretation” category by actively involving communities at all steps of the research and not only in data collection. More broadly, this work shows that many studies, particularly in the complex field of wildlife endocrinology research, could benefit from involving IK holders in data interpretation to gain a more holistic understanding of the patterns measured.

3.7. Acknowledgements

We thank all of the Kugluktuk Angoniatit Association board members from 2018 to 2020, including Larry Adjun, Kevin Klengenberg, Bobby Anavilok, Colin Adjun, Jayko Palongayak, Margo Nivingalok, Peter Taktogon, Stanley Carpenter, Sammy Angnaluak, Bessie Sitatak for their continued support. We also thank Amanda Dumond, Russell Akeeagok, and Allen Niptanatiak for their invaluable guidance and help throughout this study, as well as the Hamlet of Kugluktuk and Kugluktuk High School for their assistance.

3.8. Funding

Juliette Di Francesco was funded by the Morris Animal Foundation Fellowship Training Grant D18ZO-407. Andrea Hanke was funded by the Natural Sciences and Engineering Research Council Canada Graduate Scholarship – Master’s, Northern Scientific Training Program, Mitacs Accelerate IT14527, Association of Canadian Universities for Northern Studies, Environment and Climate Change Canada Project GCXE20C347. This work was supported by Polar Knowledge Canada Grant NST-1718-0015, the Natural Sciences and Engineering Research Council Discovery Grant RGPIN/04171-2014, the Natural Sciences and Engineering Research Council Northern Supplement - RGPNS/316244-2014, ArcticNet, and Irving Maritime Shipbuilding.

CHAPTER 4. FECAL GLUCOCORTICOID METABOLITES REFLECT HYPOTHALAMIC–PITUITARY–ADRENAL AXIS ACTIVITY IN MUSKOXEN (OVIBOS MOSCHATUS)

Juliette Di Francesco1, Gabriela F. Mastromonaco2, Janice E. Rowell3, John Blake4, Sylvia L. Checkley1, Susan Kutz1

1Department of Ecosystem and Public Health, Faculty of Veterinary Medicine, University of Calgary, 3280 Hospital Drive NW, Calgary, Alberta, Canada T2N 4Z6

2Reproductive Sciences Unit, Toronto Zoo, 361A Old Finch Avenue, Scarborough, Ontario, Canada M1B 5K7

3Agricultural and Forestry Experiment Station, University of Alaska Fairbanks, Fairbanks, Alaska, USA 99775-7500

4Animal Resources Center, University of Alaska Fairbanks, 1033 Sheenjek Drive, Fairbanks, Alaska, USA 99775-6980

Manuscript submitted on September 9th, 2020 to the journal PLOS ONE.

Full length manuscript.

4.1. Abstract

Muskoxen (Ovibos moschatus), a taxonomically unique Arctic species, are increasingly exposed to climate and other anthropogenic changes. It is critical to develop and validate reliable tools to monitor their physiological stress response in order to assess the impacts of these changes. Here, we measured fecal glucocorticoid metabolite (FGM) levels in response to the administration of adrenocorticotropic hormone (ACTH) in the winter (1 IU/kg) and summer (2 IU/kg) using two enzyme immunoassays (EIAs), one targeting primarily cortisol and the other targeting primarily corticosterone. Fecal cortisol varied substantially within and among individuals, and none of the animals in either challenge showed an increase in fecal cortisol following the injection of ACTH. By contrast, two of six (winter) and two of five (summer) muskoxen showed a clear response in fecal corticosterone levels (i.e., maximal percentage increase as compared to time 0 levels > 100%). Increases in fecal corticosterone post-ACTH injection occurred earlier and were of shorter duration in the summer than in the winter and fecal corticosterone levels were, in general, lower during the summer. These seasonal differences in FGM responses may be related to seasonal variations in the metabolism and excretion of glucocorticoids, intestinal transit time, voluntary food intake, and fecal output and moisture content. Results from this study support using FGMs as a biomarker of hypothalamic–pituitary–adrenal axis activity in muskoxen, advance our understanding of the physiological adaptations of mammals living in highly seasonal and extreme environments such as the Arctic, and emphasize the importance of considering seasonality in other species when interpreting FGM levels.

Key words: muskox, glucocorticoid metabolites, feces, wildlife, ACTH, Arctic

4.2. Introduction

Muskoxen (Ovibos moschatus) are an emblematic and taxonomically unique Arctic species (Kutz et al., 2017). Climate and other anthropogenic changes are taking place at an unprecedented pace in the Arctic and leading to the occurrence of multiple new stressors, including a higher frequency of extreme weather events, changes in vegetation abundance and diversity, modifications in species distribution and associations, and altered exposure to pathogens (Kutz et al., 2014, 2017; AMAP, 2017; Cuyler et al., 2019). Although muskoxen are extremely well adapted to the Arctic environment, their very low genetic diversity renders them particularly vulnerable to these new and accelerating environmental changes, and it is becoming crucial to develop and validate reliable tools to monitor their physiological stress response (Cuyler et al., 2019; Prewer et al., 2020). The hypothalamic–pituitary–adrenal (HPA) axis is an important mediator of the stress response (Romero and Butler, 2007). A stressor activates the hypothalamus, which then secretes corticotropin- releasing factor and arginine vasopressin to stimulate the anterior pituitary. The pituitary in turn secretes adrenocorticotropic hormone (ACTH) that stimulates the adrenal glands to produce glucocorticoids (GCs), mainly cortisol in muskoxen (Koren et al., 2012c). Free GCs circulating in the plasma are primarily metabolized by the liver, and the resulting metabolites are excreted via the bile into the intestine, where further metabolism may occur, and some of the metabolites may also be reabsorbed (enterohepatic circulation) (Möstl and Palme, 2002). Glucocorticoid metabolites consequently appear in the feces after a species-specific time delay, approximately corresponding to the intestinal transit time, from duodenum to rectum. Fecal GC metabolite (FGM) levels thus are thought to reflect the cumulative secretion and elimination of GCs over several hours to days (Palme et al., 1996; Wasser et al., 2000; Morrow et al., 2002; Touma and Palme, 2005). Fecal GC metabolites have been increasingly and widely used over the past 25 years as biomarkers of the physiological stress response in free-ranging wildlife. Fecal sampling, and therefore FGM analysis, offer the advantage of easy and non-invasive sample collection, lack of capture and handling feedback, and smoothing of the short and regular pulsatile fluctuations in circulating GCs (Palme et al., 2005; Touma and Palme, 2005; Sheriff et al., 2011). Use of FGMs as a biomarker of HPA axis activity in a novel species, such as muskoxen, requires confirmation that increases in FGM levels reflect changes in adrenal function. Pharmacological challenges involving administration of synthetic ACTH and measurement of the resulting adrenal response have been done across many taxa and are the “gold standard” for validation (Palme, 2019). Glucocorticoids are generally heavily metabolized, which results in multiple metabolites, and usually little to no native hormones, being excreted in the feces (Möstl et al., 1999, 2002; Palme et al., 2005). The cortisol and corticosterone enzyme immunoassays (EIAs) commonly used to measure FGMs, therefore, rely mostly on the cross-reactivities of their antibodies to detect this diversity of metabolites (Möstl et al., 2005). Glucocorticoid metabolism and excretion, the types and proportions of metabolites formed, and consequently, which EIA will be most effective for detecting them, can vary among species (Appendix E) (Palme et al., 2005). For example, in Roosevelt elk (Cervus canadensis roosevelti), Wasser et al.

(2000) demonstrated that the corticosterone antibody used was superior to the cortisol antibody to detect changes in FGMs following a pharmacological challenge (Wasser et al., 2000). The goal of this study was to validate the use of FGMs as a biomarker of HPA axis activity in muskoxen. More specifically, the objective was to determine whether a single pharmacological stimulation of the adrenal glands (i.e., through the administration of ACTH) was reflected in the FGM levels of muskoxen, measured using two EIAs, one targeting primarily cortisol and the other targeting primarily corticosterone.

4.3. Material and methods

4.3.1. Animals

This study was approved by the Institutional Animal Care and Use Committee, University of Alaska Fairbanks (protocol #1138945), the Veterinary Sciences Animal Care Committee, University of Calgary (protocol #AC16-0259), and the Morris Animal Foundation Animal Welfare Advisory Board. It was done at the Robert G. White Large Animal Research Station at the University of Alaska Fairbanks (USA), where a population of captive muskoxen is maintained for research and teaching purposes. The muskoxen were housed in groups of two to six individuals based on the established dominance hierarchies and their affinities. They were kept in outdoor pens of mixed pasture dominated by smooth brome grass and boreal forest, varying in size from 0.4 to 11.3 ha. All animals had access to seasonally available forage. They were also provided ad libitum fresh grass hay (brome and bluegrass), received a daily pelleted supplement (custom milled by Alaska Pet and Garden, Anchorage, USA), and had access to plain salt blocks. The muskoxen had ad libitum access to water as snow in winter and in troughs throughout the rest of the year. All animals were accustomed to routinely move from their pens to a smaller handling area, which led to a modified bison standing squeeze chute with a load scale for measures of body mass (± 1 kg).

4.3.2. ACTH challenges and fecal sampling

We did two ACTH challenges, one in winter, and the other in summer, 2018. All males were intact, none of the females were pregnant, and some of the same individuals were used in both studies (Table 4.1).

Table 4.1. Identification (ID), sex, age, and experimental group of the animals included in the winter and summer ACTH challenges.

Winter challenge Summer challenge Animal ID Sex Age Experimental Age Experimental (years) group (years) group MX-738 M 0.75 NA* 1.25 ACTH MX-740 F 0.75 ACTH 1.25 Control MX-741 M 0.75 ACTH 1.25 ACTH MX-620 M 1.75 ACTH 2.25 NA MX-621 F 1.75 ACTH 2.25 ACTH MX-597 F 2.75 NA 3.25 ACTH MX-283 F 5.75 ACTH 6.25 Control MX-1169 F 6.75 ACTH 7.25 ACTH *Not included in the challenge.

4.3.2.1. Winter challenge

This served as a pilot study to test the ACTH dose. On February 5th, 2018, six muskoxen (Table 4.1) received a 1 IU/kg intramuscular (IM) injection of Corticotrophin (Wedgewood Pharmacy, Swedesboro, NJ, USA, a long-term release gel formulation of ACTH; concentration of 80 IU/ml) in the shoulder. This dosage was chosen based on similar studies done in other ungulate species (Wasser et al., 2000; Dehnhard et al., 2001; Ashley et al., 2011; Ganswindt et al., 2012). All injections were given between 8:20 and 11:20 AM. The animals were part of a broader study that involved hair sampling at the time of the injection (Chapter 5). Fecal samples were collected from the six animals at 0–3 (referred to as time 0), 3–8, 22–25, 29– 33, 45–52, 53–56, 68–75, and 92–99 h after the ACTH injection. Some of the muskoxen were not sampled at each time-period because of technical constraints due to the small number of workers and to limited daylight, while others were sampled several times. To collect fecal samples, animals (all well habituated to humans) were observed from a distance until they defecated. They were then approached slowly and the entire fecal pile was collected with gloves from the ground. Feces were immediately placed in a Whirlpack® and then in a cooler with icepacks for a maximum of 4 h before being stored frozen at -20°C. Samples were shipped frozen to the Endocrinology Laboratory of the Toronto Zoo for FGM analysis, no later than 3 months’ post-collection.

4.3.2.2. Summer challenge

On July 23rd, 2018, five muskoxen received a 2 IU/kg IM injection of Corticotrophin in the shoulder. Two control animals were administered an equivalent volume of physiological saline (0.9% of sodium chloride), IM (Table 4.1). The dose of 2 IU/kg, twice that of the previous challenge, was given because there were some non-responders, based on FGM analyses, during the winter challenge (see section 4.4.2). All injections were given between 9:00 AM and 12:00 PM. The animals were once again part of a broader study that involved hair sampling at the time of the injection (Chapter 5).

Feces, sampled as indicated for the winter challenge, were collected the day before (referred to as time 0), and then approximately 8, 24, 32, 48, 56, 72, and 96 h after the first ACTH/saline injection from the five ACTH-injected and two control animals.

4.3.3. Hormone analyses

All hormone analyses were done at the Endocrinology Laboratory of the Toronto Zoo. Both cortisol and corticosterone EIAs were used to quantify FGMs in this study. To validate the EIAs, immunological similarities between the standard and sample hormones were evaluated by assessing parallel displacement between the standard curve and a serial dilution of a pooled muskox fecal extract. Sample dilution was selected based on 50% binding of the pooled sample curve. The recovery of exogenous hormone added to pooled muskox fecal extracts was also tested. The percentage recovery was calculated as (amount observed/amount expected) × 100, with the amount observed corresponding to the value obtained for the spiked sample minus the amount of endogenous cortisol in the unspiked fecal extract and the amount expected corresponding to the amount of standard cortisol added. As detailed in Carlsson et al. (2016a), GCs were extracted from 0.5 g of each fecal sample by rotating overnight (16–18 h) at room temperature in 5 mL of 80 % methanol-distilled water. Samples were then centrifuged for 10 min at 2,400g and the supernatant (fecal extract) was decanted and stored in glass vials at -20°C until further analysis (Carlsson et al., 2016a). Samples were removed from the freezer and thawed at room temperature prior to analysis. For cortisol analysis, 40 μl of fecal extract was evaporated in a fume hood at room temperature and the dried extracts were then reconstituted in 160 μl of assay buffer for a 1:4 dilution. For corticosterone analysis, 80 μl of fecal extract was evaporated in a fume hood at room temperature and the dried extracts were then reconstituted in 160 μl of assay buffer for a 1:2 dilution. Fecal cortisol and corticosterone metabolites were measured using the appropriate EIA protocols described by Majchrzak et al. (2015) and Baxter-Gilbert et al. (2014), respectively (Baxter-Gilbert et al., 2014; Majchrzak et al., 2015). Cortisol antibody and cortisol horseradish peroxidase conjugate dilutions were 1:10,250 and 1:33,400, respectively. Corticosterone antibody and corticosterone-HRP conjugate dilutions were 1:298,000 and 1:100,000, respectively. The detection limits of these assays were 34.5 pg/ml (cortisol) and 82.1 pg/ml (corticosterone). The cross- reactivities of the cortisol and corticosterone EIAs used were 100% to the parent hormone and < 10 or < 15% with other GCs, respectively (C. Munro, University of California, Davis, CA, USA; Appendix F). All concentrations were assayed in duplicate, with the mean of the two results presented as data. Only the duplicates with coefficients of variation (CVs, calculated as (standard deviation/mean) × 100) < 10% were accepted, and if CV was ≥ 10 %, duplicate quantitation was repeated on a second run. Data are presented as nanograms of hormone per gram of wet feces (ng/g). Due to the use of both a cortisol- and a corticosterone-specific EIA for GC metabolite detection in the fecal extracts, the terms “fecal cortisol” and “fecal corticosterone” will be used for simplicity in the results to refer to the FGMs detected

65 by the cortisol and corticosterone EIAs, respectively. Results are presented as descriptive data and plots were done using the R software version 3.4.4 (R Core Team, 2019).

4.4. Results

4.4.1. EIA validation

Serial dilutions of a pooled muskox fecal extract showed parallel displacement with the standard curves for both the cortisol (Pearson’s correlation coefficient (r) = 0.987, p < 0.01) and corticosterone (r = 0.971, p < 0.01) EIAs (Appendix F). Recovery of exogenous hormone added to a pooled muskox fecal extract was 117.6 ± 2.4% (r = 0.999, p < 0.001) for cortisol and 75.5 ± 2.8% (r = 0.999, p < 0.001) for corticosterone (Appendix F). Intra-assay CVs were 6.0 and 9.5%, and inter-assay CVs were 9.2 and 3.7%, for the cortisol and corticosterone assays, respectively.

4.4.2. FGMs

Fecal cortisol levels varied substantially within and among individuals, and none of the animals in either challenge showed an increase in fecal cortisol following the injection of ACTH. Results from the cortisol EIA are consequently presented in Appendix G. Fecal corticosterone levels varied among individuals during both the winter and summer challenges. Two of six (winter) and two of five (summer) muskoxen showed a clear response following the injection of ACTH (Figure 4.1 and Table 4.2). Increases in fecal corticosterone post-ACTH injection occurred earlier and were of shorter duration in the summer than in the winter (Fig 4.1 and Table 4.2). Fecal corticosterone levels were also, in general, lower during the summer than during the winter challenge (Figure 4.1). During the winter challenge, MX-283 exhibited a maximal percentage increase in fecal corticosterone of 229% at 47 h post-ACTH injection and levels subsequently decreased by 55 h. For MX- 1169, fecal corticosterone concentrations started increasing after 25 h before reaching a peak at 55 h and decreasing back to levels similar to those measured at time 0 by 95 h. The other four muskoxen (MX-740, 741, 620, and 621) had maximal percentage increases less than 70% (Figure 4.1 and Table 4.2). During the summer challenge, for MX-741 and MX-1169, fecal corticosterone concentrations increased by 228 and 2,814%, respectively, at approximately 7 h post-ACTH injection before decreasing back to levels similar to those measured at time 0 by 22 h. MX-597 exhibited a 776% increase in fecal corticosterone at 55 h. The other two ACTH-injected animals (MX-738 and 621) had maximal percentage increases lower than 70% (Figure 4.1 and Table 4.2). The two control animals, MX-740 and MX-283, exhibited a maximal percentage increase of 237% at 94 h and 25% at 45 h, respectively (Figure 4.1 and Table 4.2).

Table 4.2. Maximal percentage increase in fecal corticosterone (%) as compared to time 0 levels for the muskoxen given a single injection of ACTH or saline (control) during the winter (ACTH dose 1 IU/kg) and/or summer (ACTH dose 2 IU/kg) and the respective times post-injection (h) at which it was observed.

Winter challenge Summer challenge Maximal percentage Time Maximal percentage Time Animal Experimental Experimental increase in fecal post- increase in fecal post- ID group group corticosterone injection corticosterone injection MX-738 / / / ACTH 64 7 MX-740 ACTH 13 53 Control 237 94 MX-741 ACTH 17 25 ACTH 228 7 MX-620 ACTH 0* 0* / / / MX-621 ACTH 69 45 ACTH 0* 0* MX-597 / / / ACTH 776 55 MX-283 ACTH 229 47 Control 25 45 MX-1169 ACTH 394 55 ACTH 2,814 7 *The maximal fecal corticosterone concentration was measured at time 0.

Figure 4.1. Individual fecal corticosterone levels as a function of the time following a single injection of ACTH or saline (control) for the muskoxen sampled during the winter (ACTH dose 1 IU/kg – n = 6 ACTH-injected) and/or summer (ACTH dose 2 IU/kg – n = 5 ACTH-injected and n = 2 controls). Winter data are indicated as grey lines. Data for the ACTH-injected and control animals during the summer challenge correspond to the black and red lines, respectively. 4.5. Discussion

In this study, we demonstrated that a corticosterone EIA could detect changes in FGM levels following pharmacological stimulation of the HPA axis with the administration of ACTH. By contrast none of the animals in either challenge showed an increase in fecal cortisol in response to the pharmacological challenge. The cortisol EIA will not be discussed further and the FGMs of muskoxen consequently refer to those measured with the corticosterone EIA in the rest of this manuscript. We also observed seasonal differences in the timing and duration of the increase in FGMs post-ACTH injection, as well as in FGM

68 levels in general. This work advances our understanding of, and ability to interpret patterns of FGMs in muskoxen, and provides insights into their interpretation for other species. During both the winter and summer challenges, there was a high variability among individuals in the presence and magnitude of their response to a single dose of ACTH, as well as in the timing of the peak and duration of the increase in FGMs post-injection. Similar inter-individual variations have been reported in other ungulates (e.g., Rocky mountain goats (Oreamnos americanus) (Dulude-de Broin et al., 2019), reindeer (Rangifer tarandus tarandus) (Özkan Gülzari et al., 2019); Appendix E). These could be due to a variety of factors, including sex, age, reproductive status, health condition, and/or “individual” characteristics, such as genetic background or past and recent experiences, which may influence the responsiveness of the HPA axis and the metabolism and excretion of FGMs (Dantzer et al., 2014; Palme, 2019). In this study, the same individual (i.e., MX-1169) exhibited the highest increase in both challenges. Additionally, differences in the age and sex of the muskoxen may explain some of the inter-individual variability in FGM responses. During the summer challenge, MX-1169 (2,814% increase) was an adult female, whereas MX-738 and MX-741, which exhibited peaks of lower magnitude (64 and 228% increase, respectively), were male yearlings. A lower responsiveness of males was found during a pharmacological challenge in reindeer (Ashley et al., 2011) and increased HPA axis responsiveness with age has been described in other vertebrate species (Dantzer et al., 2014). The fecal sampling regime may also have contributed to the variations observed. If collections are not sufficiently frequent, the peak response for some animals may be missed. For example, by not collecting the fecal samples voided between 8 and 24 h post-injection during the summer challenge, we likely detected only the beginning of the FGM increase in MX-738 and MX-741 and probably missed the response entirely for MX-597 and MX-621. The possibility of having missed peak samples in some animals emphasizes the importance of collecting and testing, when feasible, feces from all fecal voidance events during the expected time of response. We observed relatively low maximal percentage increases of 13% at 53 h, 17% at 25 h, and 69% at 45h in three animals (MX-740, MX-741, and MX-621, respectively) during the winter challenge (Figure 4.1), and two other animals (MX-620 – winter and MX-621 – summer) had their maximal FGM levels measured at time 0. These zero and low maximal increases may reflect the unusually high FGM concentrations measured at time 0. We are unaware of the occurrence of a previous stressful event that may have caused these high time 0 levels. Ideally, because of intra-individual variations, multiple samples (rather than just a single sample) should be collected prior to adrenal stimulation to establish an FGM baseline (Palme, 2019). Finally, it is possible that some animals exhibited a low or an absence of response to the ACTH administered as has been highlighted in other studies (Ashley et al., 2011; Coradello et al., 2012; Cantarelli et al., 2017). The timing and magnitude of the peaks in FGMs that we detected, as well as the timing of return to baseline levels, are comparable to those measured following pharmacological challenges in other even- toed ungulate species (Appendix E). The 2,814% increase observed in MX-1169 during the summer challenge sits at the high end of the ranges of FGM peak magnitudes. The summer FGM peak timing of 7 h with a return to time 0 levels by 22 h post-injection would be closest to the responses measured in

69 reindeer (Özkan Gülzari et al., 2019), caribou (Rangifer tarandus granti) (Ashley et al., 2011), cattle (Bos taurus) (Palme et al., 1999; Morrow et al., 2002), and sheep (Ovis aries) (Palme et al., 1999). The later peaks and long responses observed during the winter challenge (i.e., peak in FGMs at 47 h with a decrease by 55 h in MX- 283 and increase after 25 h followed by a return to time 0 levels by 95 h in MX-1189) are at the high end of the ranges measured in other even-toed ungulate species, but such responses have been observed in reindeer (Ashley et al., 2011) and dromedary camels (Camelus dromedaries) (Sid-Ahmed et al., 2013). Our data suggest that metabolism of GCs and intestinal transit time were faster in the summer than in the winter (Figure 4.1). However, comparisons between the two challenges must consider that different doses of ACTH were administered (1 IU/kg in the winter and 2 IU/kg in the summer) and sampling times varied. The few studies that have compared seasonal responses of FGMs to ACTH administration have had differing results depending on the species. A study of reindeer in Norway in which FGMs were measured following a pharmacological challenge in the winter and a stressful event (i.e., calf marking) in the summer found no major differences in the timing of the FGM elevation (Özkan Gülzari et al., 2019). A study in white-tailed deer (Odocoileus virginianus) detected earlier peaks in FGMs following a pharmacological challenge in the winter (10–13 h post-ACTH injection) compared to the fall (20–24 h) (Millspaugh et al., 2002). Finally, a study in cattle found longer lag times between elevated plasma GC levels and peak FGM concentrations following ACTH challenges in the autumn (14.8 ± 0.47 h (mean ± SD)) compared to the spring (8.61 ± 0.26 h) (Morrow et al., 2002). Seasonal differences in FGM responses may be related to seasonal variations in type and quantity of food intake, as well as differences in liver metabolism and conjugation rates, bacteria action as they further metabolize GCs in the intestines, and fecal output and moisture content (Millspaugh et al., 2002; Morrow et al., 2002). Muskoxen live in a highly seasonal environment characterized by a short summer with access to abundant forage of high quality, during which they accumulate important fat reserves, followed by a long winter of restricted access to generally limited and low-quality forage (Gunn and Adamczewski, 2003). To conserve energy in the extreme winter conditions, muskoxen down-regulate their metabolism and lower their energy expenditure and body temperature (Adamczewski and Flood, 1997; Lawler and White, 1997; Schmidt et al., 2020), which would explain a possible slower metabolism of GCs during this period. Studies on intestinal transit times in muskoxen have consistently found that these were slowest during the winter (Appendix H) (Holleman et al., 1984; Adamczewski et al., 1994; Barboza et al., 2006). A study measuring fecal testosterone metabolites (FTMs) following the administration of testosterone IM to four adult castrated male muskoxen in September demonstrated that FTM levels peaked between 14 and 24 h post-injection and had generally almost returned to baseline by 36 h (Flood et al., personal communication). Based on the high seasonal, inter and intra-individual variations highlighted in these studies (Appendix H), we cannot exclude the range of 7-hour to 47-hour intestinal transit time that we detected depending on the season and based on the timing of the FGM peak. Finally, voluntary food intake (White et al., 1984) and frequency of defecation (Di Francesco, pers. obs.) were higher during the summer, which may have also contributed to the seasonal differences we observed. Radiometabolism studies, which involve the injection of a radiolabeled steroid

70 hormone and subsequent collection of all the excreta voided, typically represent the “gold standard” to determine the metabolism and excretion pathways of steroid hormones (Palme, 2019). Undertaking such a study in muskoxen both in winter and summer would help to refine the seasonal variations in this species. FGM levels were also, in general, higher during the winter than during the summer challenge (Figure 4.1). This finding is in line with the lower hair cortisol concentrations measured in the summer than during the fall and winter in wild muskoxen (Di Francesco et al., 2017). Similar seasonal variations have been observed in captive goral (Naemorhedus griseus) (Khonmee et al., 2014) and red deer (Cervus elaphus) (Huber et al., 2003). These may reflect the reduced metabolic rate and voluntary food intake of muskoxen in winter with a shift towards catabolic metabolism (Huber et al., 2003; Goymann, 2012). Two muskoxen in the summer challenge had significant increases in FGMs at 55 h (776%) and 94 h (237%) post-ACTH and saline injection, respectively. It is likely that these correspond to an independent stimulus, rather than a delayed experimental response. While we were not able to identify a possible stressful event that could explain these increases, this may support the corticosterone EIA’s potential to detect biological HPA axis responses. The performance of the cortisol and corticosterone EIAs used in this study differed. While the corticosterone EIA detected responses to the pharmacological challenge in several muskoxen, the cortisol EIA did not detect any changes post-ACTH injection (Appendix G) and was thus not used in subsequent studies. Multiple studies have highlighted differences in assay performance for FGM quantification depending on the species. Due to species-specific differences in hormone metabolism during transit through the gut, the excreted metabolites vary in structure and proportion, and consequently, will only be effectively detected using an antibody that cross-reacts with the specific structures present (Appendix E). For example, a study in giraffes (Giraffa camelopardalis) found variations both between the animals and among the six EIAs used to quantify FGMs following an ACTH injection (Bashaw et al., 2016), and the contrasting results of two pharmacological challenges done in reindeer may have been due to the use of different EIAs (see Appendix E for study details) (Ashley et al., 2011; Özkan Gülzari et al., 2019). While the corticosterone EIA detected responses in several muskoxen, an EIA more specifically targeting the metabolites excreted by this species may have allowed us to more reliably detect changes in FGMs with greater consistency across animals. Access to high-performance liquid chromatography data would have enabled us to separate and characterize the metabolites excreted by the muskoxen, and consequently to determine whether another assay would have been better suited to measure FGMs in this species.

4.6. Conclusion

Our results demonstrate that the FGM response following a pharmacological challenge can be measured in muskoxen using a corticosterone EIA. This is encouraging with respect to using FGMs as a biomarker of HPA axis activity in muskoxen. Fecal GC metabolites have been widely used to inform wildlife conservation in a variety of species and settings (e.g., Zwijacz-Kozica et al., 2013; Zbyryt et al., 2018; Atwood et al., 2020). This tool can now be applied to assess the impact of various stressors (e.g., weather conditions,

71 anthropogenic activities) in muskoxen and to investigate the relationship between FGM levels and health indicators (e.g., parasite diversity and infestation intensity or exposure to various bacterial and viral pathogens). This is particularly important to evaluate in the Arctic, where rapid climate warming is leading to increased environmental changes and altered host-pathogen interactions, as evidenced, for example, by the recent range expansion of two major muskox lungworms in the Canadian Arctic Archipelago (Kutz et al., 2013a; Kafle, 2018). This study also highlighted seasonal variations in the metabolism and excretion of GCs, as well as in intestinal transit time in muskoxen. These findings advance our understanding of the physiological adaptations of mammals living in highly seasonal and extreme environments such as the Arctic, and emphasize the importance of considering seasonality in other species when interpreting FGM levels.

4.7. Acknowledgements

We are grateful to all the staff and summer students working at the Robert G. White Large Animal Research Station of the University of Alaska Fairbanks who helped in sample collection and animal handling, particularly to Sarah Barcalow, Thalia Souza, Hanna Sfraga, Carla Wiletto, Christine Terzi, Megan Roberts, Jean Rein, Claire Kepner, and Charles Ashlock. We wish to thank Christine Gilman and Patricia Medd from the Endocrinology Laboratory of the Toronto Zoo for their hard work in method development and fecal sample analyses. We finally thank James Wang for his ongoing support in the lab and Katherine Wynne-Edwards for her input during the conceptualization of this study. A special thought also goes to all the muskoxen who took part in this study.

4.8. Funding

Juliette Di Francesco was funded by the Morris Animal Foundation Fellowship Training Grant D18ZO-407. This work was supported by Polar Knowledge Canada Grant NST-1718-0015, the Natural Sciences and Engineering Research Council Discovery Grant RGPIN/04171-2014, the Natural Sciences and Engineering Research Council Northern Supplement RGPNS/316244-2014, and ArcticNet.

CHAPTER 5. QIVIUT CORTISOL REFLECTS HYPOTHALAMO–PITUITARY–ADRENAL AXIS IN MUSKOXEN (OVIBOS MOSCHATUS)

Juliette Di Francesco1, Gabriela F. Mastromonaco2, Sylvia L. Checkley1, John Blake3, Janice E. Rowell4, Susan Kutz1

1Department of Ecosystem and Public Health, Faculty of Veterinary Medicine, University of Calgary, 3280 Hospital Drive NW, Calgary, Alberta, Canada T2N 4Z6

2Reproductive Sciences Unit, Toronto Zoo, 361A Old Finch Avenue, Scarborough, Ontario, Canada M1B 5K7

3Animal Resources Center, University of Alaska Fairbanks, 1033 Sheenjek Drive, Fairbanks, Alaska, USA 99775-6980

4Agricultural and Forestry Experiment Station, University of Alaska Fairbanks, Fairbanks, Alaska, USA 99775-7500

Manuscript submitted on September 8th, 2020 to the journal General and Comparative Endocrinology.

Full length manuscript.

5.1. Abstract

Muskoxen (Ovibos moschatus) are increasingly exposed to a broad diversity of stressors in their rapidly changing Arctic environment. There is an urgent need to develop validated tools to monitor the impact of these stressors on the hypothalamic–pituitary–adrenal (HPA) axis activity of muskoxen to help inform conservation actions. Here, we evaluated whether muskox qiviut (dense wooly undercoat) cortisol accurately reflects changes in HPA axis activity. Two repeated pharmacological challenges, involving weekly administrations of saline (control group) or adrenocorticotropic hormone (ACTH) during five consecutive weeks, were done on captive muskoxen, in winter (no hair growth) and summer (maximum hair growth). Pre-challenge qiviut cortisol levels were significantly higher in the shoulder than in the neck, but neither differed from rump concentrations. Qiviut cortisol levels significantly increased (p < 0.001) in response to the administration of exogenous ACTH during the hair growth phase, but not in the absence of growth (p = 0.84). Cortisol levels in the qiviut segment grown during the summer challenge increased significantly over a six-month period in the ACTH-injected muskoxen with a similar trend occurring in the control animals. Finally, cortisol levels in shed qiviut were significantly higher and not correlated to those of fully grown qiviut shaved three months earlier. Our results show that cortisol is deposited in qiviut during its growth and that qiviut cortisol can thus be used as an integrated measure of HPA axis activity over the period of the hair’s growth. Differences in qiviut cortisol across body regions, significant differences in qiviut segments over time, and differences between shed qiviut versus unshed qiviut, highlight the limitations of qiviut cortisol as a biomarker of HPA axis activity in muskoxen and the importance of strict design and methodology for sample collection and analyses in order to account for sources of variation.

Key words: muskox, hair cortisol, wildlife, ACTH, qiviut, Arctic

5.2. Introduction

Over the past two decades, hair glucocorticoids (GCs) have become a popular tool to study long- term hypothalamic–pituitary–adrenal (HPA) axis activity in wildlife (Dantzer et al., 2014). Hair GCs have been used in a variety of wild species as a biomarker of HPA axis response to a wide range of biological, social, and environmental conditions, including anthropogenic disturbances (Bourbonnais et al., 2013; Bryan et al., 2013, 2015; Mastromonaco et al., 2014; Agnew et al., 2016; Salas et al., 2016; Ewacha et al., 2017; Fardi et al., 2018; Santangeli et al., 2019). Hair GCs have also been used to assess relationships between HPA axis activity and health indicators, such as parasitism and body condition (Cattet et al., 2014; Carlsson et al., 2016a; Mislan et al., 2016; Potratz et al., 2019; Madslien et al., 2020), and proxies of fitness, such as survival and reproductive success (Rakotoniaina et al., 2017; Downs et al., 2018). Despite the growing use of hair GCs in wildlife studies, validation of these techniques is often lacking (Koren et al., 2019). Hair, unlike other matrices in which GCs are incorporated, is thought to give an integrated measure of circulating GC concentrations over periods of weeks to months, depending on the species-specific patterns of hair growth (Sheriff et al., 2011; Russell et al., 2012). However, mechanisms of GC deposition and stability in hair are highly complex, continuously debated, and poorly understood. “Internal” hair GC concentrations may reflect not only those GCs accumulated from both local and systemic sources inside the shaft during hair growth, but also more recent and short-term levels from glandular secretions and extrinsic sources such as urine, saliva or fecal contamination (Figure 5.1; Kalliokoski et al., 2019; Koren et al., 2019). Before hair GCs are used as a biomarker of HPA axis activity, evidence that concentrations accurately reflect biologically meaningful changes in HPA axis activity and the temporal context that they represent should be determined for the species of interest. This is generally done through biological (i.e., tracking the effect of a stressful episode, such as a relocation event) or physiological validations (i.e., pharmacological stimulation or suppression of the HPA axis), but few such studies exist in wildlife (Koren et al., 2019; Palme, 2019). Muskoxen (Ovibos moschatus), a taxonomically unique Arctic ungulate species, are increasingly exposed to a broad variety of stressors in their rapidly changing environment (AMAP, 2017; Kutz et al., 2017; Cuyler et al., 2019). Hair GCs, mainly cortisol in muskoxen (Koren et al., 2012c), once validated, could be a useful tool to evaluate the impacts that these stressors have on individuals and populations. The overarching goal of this study was to understand the patterns of cortisol deposition, distribution, and stability in muskox hair, and to validate the use of hair cortisol as a biomarker of HPA axis activity in this species. Outcomes from this study provide novel insights and further our understanding of GC deposition and stability patterns in hair, and contribute to advancing the challenging research field of wildlife endocrinology.

Figure 5.1. Summary of the possible sources of deposition and loss of GCs in hair (adapted from Henderson, 1993; Meyer and Novak, 2012; Sharpley et al., 2012). Glucocorticoids are thought to be deposited during the period of active hair growth at the level of the hair follicle as they diffuse directly from the blood vessel supplying the follicle (Meyer and Novak, 2012; Burnard et al., 2017; Kapoor et al., 2018). Hair GCs may also be derived from local synthesis in the skin and within the hair follicle by a functional equivalent of the HPA axis (Ito et al., 2005; Slominski et al., 2007; Keckeis et al., 2012). The connection between the local and central HPA axes, the extent to which local GC production is centrally or locally induced, and the relative contributions of locally produced and systemic GCs to “internal” hair concentrations remain unclear (Sharpley et al., 2012; Skobowiat and Slominski, 2015; Salaberger et al., 2016). Glucocorticoids originating from the blood, local production, or both may also be taken up by cells of the sebaceous and apocrine glands and be deposited via sebum and sweat onto the outer cuticle of the hair shaft (Meyer and Novak, 2012; Burnard et al., 2017). This contributes to the “external” GCs on the surface of the hair, which can be augmented by additional GCs deposited from extrinsic sources, such as saliva, urine, or feces. The washing step in most hair GC quantification processes aims to remove any potential “external” GCs (see Koren et al., 2019). Some studies suggest that “external” GCs may be incorporated into the hair shaft, and may contribute to the “internal” GC concentrations (Macbeth, 2013; Cattet et al., 2014; Russell et al., 2014; Otten et al., 2020). The magnitude of this phenomenon remains unclear, but it may be facilitated by moisture (Macbeth et al., 2010; Cattet et al., 2014) and damage to the shaft, which may render the cuticle more permeable (Heimbürge et al., 2020a). “Washout” of GCs (i.e., leaching out of the hair) due to weather exposure (Heimbürge et al., 2019), UV radiation (Wester, 2016), or grooming (Acker et al., 2018) may also be possible and GC molecules may move along the hair shaft after being deposited even though studies have shown conflicting results (Carlitz et al., 2014; Kapoor et al., 2018; Heimbürge et al., 2020b).

5.3. Material and methods

5.3.1. Animals

Experiments were approved by the Institutional Animal Care and Use Committee, University of Alaska Fairbanks (protocol #1138945), the Veterinary Sciences Animal Care Committee, University of Calgary (protocol #AC16-0259), and the Morris Animal Foundation Animal Welfare Advisory Board. They were done at the Robert G. White Large Animal Research Station at the University of Alaska Fairbanks, USA, using their herd of captive muskoxen. Muskoxen were kept in groups of 2–6 individuals based on their dominance hierarchies and affinities. They were housed in outdoor pens, varying in size from 0.4–11.3 ha, and constituted of mixed pasture dominated by smooth brome grass and boreal forest. All animals had access to seasonally available forage, complemented with a daily pelleted supplement (custom milled by Alaska Pet and Garden, Anchorage, Alaska), fresh grass hay provided ad libitum (brome and bluegrass), and access to plain salt blocks. Water was available ad libitum as snow in winter and in water troughs throughout the rest of the year. Muskoxen were previously habituated to the handling facility, which includes a chute system and a load scale for measuring body mass (± 1 kg). All except five animals had been trained for basic procedures, such as hair clipping, to be done while standing in the chute. Muskoxen have three types of hair: (i) thick guard hairs that grow continually (Flood et al., 1989); (ii) a dense and fine woolly undercoat called qiviut, which is grown annually between early April and late November and shed in its entirety the following May (Flood et al., 1989); and (iii) intermediate hairs that are fine like qiviut near their root but coarsen substantially toward their distal end, and are shed every year with the qiviut (Rowell et al., 2001). Hair cortisol concentrations (HCCs) have been shown to vary with hair type (Macbeth et al., 2010; Dulude-de Broin et al., 2019). We consequently focused solely on qiviut as it grows during a defined time period, is shed annually, and shed qiviut can easily be collected non-invasively from the tundra, thus potentially offering another way of monitoring muskoxen.

5.3.2. ACTH challenges

Two experiments were done. Experiment 1, in winter 2018, tested whether single and repeated ACTH injections resulted in elevated qiviut cortisol when the hair was not growing. Experiment 2, from summer 2018 to spring 2019: (i) assessed if repeated ACTH injections resulted in elevated qiviut cortisol in growing hair; (ii) compared qiviut cortisol levels between the neck, shoulder and rump of muskoxen; (iii) determined consistency of cortisol concentrations once deposited into the qiviut (i.e., does the cortisol concentration in a specific segment remain the same over time?) and (iv) when the qiviut is shed (i.e., is the cortisol concentration in shed qiviut the same as prior to shedding?). Two repeated five-week ACTH challenges were performed in each experiment: one in winter (February 5th to March 5th, 2018 – Experiment 1), in the absence of qiviut growth, and one in summer (July

23rd to August 21st, 2018 – Experiment 2), during the period of maximal qiviut growth (Robertson, 2000). Fifteen muskoxen were included in Experiment 1, and 16 were included in Experiment 2 (Appendix I). Ten animals were used in both experiments. Animals were allocated to the control (saline) and ACTH experimental groups using a randomized block design (Appendix I). They were divided into groups of three similar individuals for sex, age, reproduction status, and/or need for immobilization, from which one was randomly assigned to the control group and the other two to the ACTH group. In Experiment 2, the two young bulls and the two steers were randomized so that one individual was randomly assigned to the control group and the other to the ACTH group. Animals of the ACTH groups received weekly intramuscular (IM) injections in the shoulder or hip of a long-term release gel formulation of ACTH (Corticotrophin, stock concentration of 80 IU/ml, Wedgewood Pharmacy, Swedesboro, NJ, USA) over a period of five weeks, while animals of the control groups were administered an equivalent volume of physiological saline (0.9% of sodium chloride). The number of doses and their spacing were chosen based on similar studies in other mammalian species (Terwissen et al., 2013; Mastromonaco et al., 2014; Dulude-de Broin et al., 2019; Appendix J). A dosage of 1 IU/kg (Experiment 1) or 2 IU/kg (Experiment 2) was administered weekly in a single injection using an 18-gauge 1.5-inch needle. The initial 1 IU/kg dosage was chosen based on studies in other ungulate species (Wasser et al., 2000; Dehnhard et al., 2001; Ashley et al., 2011; Ganswindt et al., 2012). This dose was raised to 2 IU/kg for Experiment 2 because fecal glucocorticoid metabolite (FGM) analyses in Experiment 1 revealed some non/poor-responders following the first ACTH injection (Chapter 4).

5.3.3. Hair sampling

Hair samples were collected using electric clippers while the muskoxen were restrained in the chute or immobilized. The clipper blade was cleaned between each animal using a brush. Hair samples were placed in paper envelopes, which were left open during several hours to allow the samples to air-dry if they were moist, and subsequently closed and stored at room temperature for 5-18 months until the samples were analyzed.

5.3.3.1. Experiment 1

Five control and ten ACTH-injected animals were used. Three adjacent vertical bands of approximately 5 × 15 cm were shaved from the rump at time 0 (t0 – first injection), t0 + 1 week (t0 + 1W – second injection), and at t0 + 5W (1 week after fifth and last injection). A small 1 cm-wide band of hair was left between each shaving location to reduce likelihood of a previous shaving event affecting adjacent samples.

5.3.3.2. Experiment 2

Six control and ten ACTH-injected animals were used. A square patch of hair was shaved at t0 (first injection) from the neck and shoulder (approximately 10 × 10 cm area for each) and from the rump on the

78 same side (approximately 20 × 20 cm) (Figure 5.2). At t0 + 6W (2 weeks after fifth and last injection), the same neck and shoulder patches were re-shaved entirely, while a vertical band 20 × 5 cm was re-shaved from one side of the rump patch leaving a 20 × 15 cm area containing the hair grown since t0 intact on the rump (Figure 5.2). The t0 + 6W time frame (versus + 5W in Experiment 1) was chosen so that the qiviut grown during the challenge would have time (2 weeks) to grow out from the skin after the last injection. To assess if the cortisol concentration within a specific segment of the qiviut shaft remains the same over time and, consequently, whether segmental analysis of qiviut can be used to refine timescales of stressful events, approximately half (10 × 15 cm horizontal band) of the remaining rump patch, was shaved at t0 + 3 months (t0 + 3M – November 2018) and the other half (10 × 15 cm horizontal band) at t0 + 6M

(February 2019) (Figure 5.2). The adjacent 10 × 5 cm patch of hair grown between t0 + 6W and t0 + 3 or 6 months was shaved at the same time so that its length could be used as reference to delimit and cut using scissors the hair grown during the period of the ACTH challenge (from t0 to t0 + 6W) (Figure 5.2). Body regions were chosen to represent the most likely sites to be sampled during hunter-based sampling programs, and particular focus was placed on the rump as it is the region from which samples have been collected by hunters since 2013 as part of an ongoing muskox health monitoring program (see Di Francesco et al., 2017). To assess if the cortisol concentration from shed qiviut is the same as prior to shedding, an additional sample (5 × 10 cm) grown during the entire qiviut growth period (i.e., April–November 2018) was shaved in February 2019 from the rump on the side of the animal not previously shaved. Shed qiviut from an adjacent area was collected the following April/May 2019, during qiviut combing (Figure 5.2).

Figure 5.2. Diagram showing the qiviut samples collected during Experiment 2 and the corresponding research questions. Not all photos were taken from the same animal. Photo credits: Juliette Di Francesco and Morgan Mouton and muskox drawing by Jayninn Yue.

5.3.4. Fecal sampling

To better understand the duration of the release of the ACTH formulation injected, during Experiment 2, we sampled feces the day before each injection from three control and seven ACTH-injected muskoxen, which were all well habituated to humans. To collect fecal samples, animals were observed from a distance until they defecated. They were then approached slowly and the entire fecal pile was collected with gloves from the ground. Feces were immediately placed in a Whirlpack® and then in a cooler with icepacks for a maximum of 4 h before being stored frozen at -20°C. Samples were shipped frozen to the Endocrinology Laboratory of the Toronto Zoo for FGM analysis, no later than 3 months’ post-collection.

5.3.5. Immobilization procedures

Five animals (n = 1 in Experiment 1 and n = 4 in Experiment 2) required immobilization for each hair clipping procedure (4–5 mg/kg ketamine, 0.1 mg/kg xylazine, and 0.1 mg/kg azaperone administered IM; reversed with 1 mg/kg tolazoline IM 30 min or more after the last dose of ketamine). Immobilized animals had supplemental oxygen, and a pulse oximeter was used to monitor heart rate and blood oxygenation level. During Experiment 2, but unrelated to the experimental design, two control and five ACTH- injected mature females were immobilized at the time of the 4th injection in order to insert a controlled internal drug release intravaginal device for estrus synchronization; one additional control female was immobilized two days after the last injection for the same reason. During Experiment 2, blood was sampled from the same four control and five ACTH-injected animals at t0, t0 + 3W, and t0 + 6W to monitor complete blood counts and serum biochemistry.

5.3.6. Hormone analyses and enzyme immunoassay validation

5.3.6.1. Qiviut cortisol

Qiviut was manually separated from guard and intermediate hairs using forceps and sent in paper envelopes to the Endocrinology Laboratory of the Toronto Zoo for analysis. There, as per Mastromonaco et al. (2014), the qiviut was cut into 5 mm pieces and placed into a 7 ml glass scintillation vial so that 0.05 g could be weighed (see Appendix K for the extraction mass dose response). All qiviut samples were washed with 4 ml of 100% methanol by vortexing for 10 s and immediately removing all of the methanol with a pipettor. This wash step was done to remove from the surface of the qiviut shaft glandular secretions and possible extrinsic contamination by biological fluids such as urine, feces, or saliva. Immediately afterwards, 100% methanol was added to the samples, at a ratio of 0.01 g/ml, and samples were placed on a rotator for 24 h (Fisher Scientific Platelet Mixer 348, USA). Samples were then centrifuged for 5 min at 2,400g and the supernatant (qiviut extract) was pipetted off into clean glass vials and 450 μl was evaporated in a fume hood at room temperature (Mastromonaco et al., 2014). Dried extracts were stored at -20 °C until further analysis.

Samples were brought to room temperature and reconstituted in 150 μl of assay buffer for a three- fold concentration. Cortisol concentrations in the reconstituted extracts were measured using the enzyme immunoassay (EIA) protocol previously described by Majchrzak et al. (2015). Cortisol antibody and cortisol horseradish peroxidase conjugate dilutions were 1:10,250 and 1:33,400, respectively. The detection limit of this assay was 34.5 pg/ml. The cross-reactivities of the cortisol assay were 100% to the parent hormone and < 10% with other GCs (C. Munro, University of California, Davis, CA, USA; Appendix K). To validate the cortisol EIA for muskox qiviut, we tested the recovery of standard cortisol added to a pooled muskox qiviut extract. The percentage recovery was calculated as amount observed/amount expected × 100, with the amount observed corresponding to the value obtained for the spiked sample minus the amount of endogenous cortisol in the unspiked qiviut extract and the amount expected corresponding to the amount of standard cortisol added. Recovery of standard hormone added to a pooled muskox qiviut extract was 101.1 ± 3.2% and the amount of cortisol recovered was correlated with the amount added (Pearson’s correlation coefficient (r) = 0.999, p < 0.001; Appendix K). We also assessed the parallel displacement between the standard curve and a serial dilution of a pooled muskox qiviut extract. Sample concentration was chosen based on 50% binding of the pooled sample curve and serial dilutions showed parallel displacement with the standard curve (r = 0.992, p < 0.001; Appendix K). Intra- and inter-assay coefficients of variation (CVs, calculated as (standard deviation/mean) × 100) based on pooled hair extracts run on each plate as controls were 8.5 and 6.6%, respectively. All cortisol concentrations were assayed in duplicate on the plate with the mean of the two results used as data. Only the duplicates with CVs < 10% were accepted, while duplicate quantitation was repeated for those with CV ≥ 10%. Intra-sample quantitation variation for qiviut cortisol levels (obtained by running 10 separate extractions from one hair pool) was 9.97 ± 0.78% for non-growing winter qiviut from a captive muskox (collected in February) and 41.78 ± 14.49% for growing summer qiviut from a captive muskox (collected in July; discussed in section 5.5.2). To account for this variability, for all qiviut samples (i.e., same collection site and sampling time), two (non-growing winter and spring-collected samples) or four (summer- collected/grown samples) qiviut subsamples were tested independently as true experimental replicates, with the mean of the two or four results presented as data (see Appendix L for the CVs of the various sample groups). Data are presented as nanograms of cortisol per gram of qiviut (ng/g).

5.3.6.2. FGMs

Fecal GC metabolite levels were measured at the Endocrinology Laboratory of the Toronto Zoo using a corticosterone EIA as described in Chapter 4. Briefly, 0.5 g of fecal sample was extracted by rotating overnight at room temperature in 5 mL of 80 % methanol-distilled water (Carlsson et al., 2016a). Fecal extracts were stored in glass vials at -20°C until further analysis. Prior to analysis, 80 μl of fecal extract was evaporated in a fume hood at room temperature and then reconstituted in 160 μl of assay buffer for a 1:2 dilution. Fecal GC metabolite levels were measured in duplicate using the corticosterone EIA protocol described by Baxter-Gilbert et al. (2014). Corticosterone antibody and corticosterone-HRP conjugate

82 dilutions were 1:298,000 and 1:100,000, respectively. The detection limit of the assay was 82.1 pg/ml. The cross-reactivities of the corticosterone EIA was 100% to the parent hormone and < 15% with other GCs (C. Munro, University of California, Davis, CA, USA). Data are presented as nanograms of hormone per gram of wet feces (ng/g). Intra-assay and inter-assay CVs were 9.5% and 3.7%, respectively.

5.3.7. Statistical analyses

Data are presented as median and range. All analyses were performed using the R software version 3.4.4 (R Core Team, 2019). The significance level was set at p < 0.05 and normality was assessed using the Shapiro-Wilk test.

5.3.7.1. Qiviut cortisol

We used linear mixed-effect models to determine if there were differences in qiviut cortisol among sampling times (t0, t0 + 1W, and t0 + 5W in Experiment 1; t0 and t0 + 6W and t0 + 6W, t0 + 3M, and t0 + 6M in Experiment 2), among body regions (neck, shoulder, and rump), and between fully grown shaved versus shed qiviut. Animal identity was fitted as a random effect in all models to account for the repeated sampling of individual animals. The models were fit separately in the control and ACTH-injected animals, except when specified otherwise, using the lme function from the nlme package and restricted maximum likelihood was used to obtain coefficient estimates. The potential effects of sex (male and female in Experiment 1; male, female, and steer in Experiment 2) and age (juvenile (< 2 years old) and adult) were originally assessed in all models, but they were subsequently removed since their inclusion did not modify the effects of the variables tested. When necessary, qiviut cortisol levels were log-transformed to satisfy the linear mixed model assumptions of normal distribution and homoscedasticity. These assumptions were assessed by review of the residual plots for each model both at the observation and animal (random effect) levels. All model details and coefficients are included in Appendix M. When a log-transformation was performed, the coefficients are also presented as back-transformed.

When assessing the difference in qiviut cortisol levels between t0 and t0 + 6W in Experiment 2, models were also fit separately in each body region. When comparing qiviut cortisol levels among body regions, all animals were grouped together in the model and only pre-challenge (t0) data was used. When comparing cortisol levels between fully grown shaved and shed qiviut, even though the duration of the ACTH challenge was very short when compared to the entire qiviut growth period, we originally corrected for experimental group (control or ACTH) in the model. However, it did not have a significant effect (p = 0.64) and its inclusion did not modify the coefficient for shed versus shaved qiviut, so we grouped all animals together in the final model (Appendix M). The two-by-two correlations between the neck, shoulder, and rump qiviut cortisol levels were measured using Pearson’s correlation coefficient (r) as these were normally distributed, whereas the correlation between fully grown shaved and shed qiviut was measured using Spearman’s rank correlation coefficient (rs) (McCrum-Gardner, 2008).

To compare intra-sample variability among body regions, we used a linear mixed-effect model with animal identity as a random effect to determine if there were differences in pre-challenge (t0) quadruplicate CVs (i.e., determined for each qiviut sample as the CV of the cortisol concentrations measured in four subsamples tested independently as true experimental replicates) among body regions. Quadruplicate CVs were log-transformed to satisfy the linear mixed model assumptions and all model details and coefficients are presented in Appendix N.

5.3.7.2. FGMs

To compare pre-injection FGM levels between the five injections, a linear mixed-effect model with animal identity as a random effect was fit to test the effect of injection number (1–5) on FGMs separately in the control and ACTH-injected animals. We chose to treat injection number as a continuous rather than as an ordinal variable, as the number of observations in our dataset was small and we were more interested in the effect of an additional injection than in comparing specific injections. Fecal GC metabolite levels were log-transformed, where necessary (i.e., ACTH-injected animals), to satisfy the linear mixed model assumptions and all model details and coefficients are included in Appendix O.

5.4. Results

5.4.1. Experiment 1 – Are single and repeated ACTH injections reflected in qiviut in the absence of hair growth?

In the absence of hair growth, there was no significant difference in qiviut cortisol levels pre- injection, one week after a single injection, or one week after termination of five weekly injections in the control (p = 0.61) and ACTH-injected (p = 0.84) animals (Table 5.1, Figure 5.3, and Appendix M). In both control and ACTH-injected animals, qiviut cortisol variability was high among animals, but low within the same individual (Figure 5.3 and Appendix M).

Control animals ACTH-injected animals

(n = 5) (n = 10) t0 11.24 (8.81–26.19) 13.25 (6.61–44.79) t0 + 1W 11.78 (8.53–25.79) 15.25 (7.45–34.13) t0 + 5W 11.21 (8.10–29.56) 14.73 (7.13–28.61)

5.4.2.1. Are repeated ACTH injections reflected in qiviut when the hair is growing?

In control animals, qiviut cortisol concentrations did not significantly differ in the rump or shoulder pre- and post-challenge, however, they were significantly higher post-challenge in the neck (p = 0.01). In ACTH-injected animals, qiviut cortisol levels were significantly higher post-challenge compared to pre- challenge in all three body regions (p < 0.001) (Table 5.2, Figure 5.4, and Appendix M). There was variability among muskoxen and among body regions within the same individual in the magnitude of the response to ACTH (Figure 5.4 and Appendix P).

Neck Shoulder Rump ACTH- ACTH- ACTH- Control Control Control injected injected injected animals animals Animals animals animals animals 7.01a 7.40a 8.09a 8.71a 8.84a 7.38a t 0 (2.82–9.04) (4.71–13.82) (5.16–11.47) (6.88–15.42) (3.66–11.86) (4.09–12.51) 10.67b 17.57b 10.15a 17.99b 11.35a 14.64b t + 6W 0 (7.68–15.19) (13.65–26.82) (8.81–17.36) (12.83–34.15) (8.05–14.45) (11.53–18.00) Percentage +52.48 +151.40 +39.83 +119.62 +39.19 +105.32 increase (20.50–238.46) (15.43–305.83) (5.75–201.86) (12.34–227.36) (-23.64–150.68) (33.95–316.85)

Qiviut cortisol concentrations pre-challenge differed significantly among body regions (n = 16; p = 0.02). They increased from neck (median (range) = 7.40 ng/g (2.82–13.82)) to rump (median (range) = 8.06 ng/g (3.66–12.51)) to shoulder (median (range) = 8.49 ng/g (5.16–15.42)), but only the neck and shoulder differed significantly (p = 0.01; Figure 4.5 and Appendix M). Neck qiviut cortisol concentrations correlated with shoulder (r = 0.82, p < 0.001) and rump concentrations (r = 0.57, p = 0.02),

86 but rump and shoulder qiviut cortisol levels were not correlated (r = 0.30, p = 0.26). Intra-sample variability did not differ significantly among body regions (p = 0.27) and was 15.70% (8.67–39.95) (median quadruplicate CV (range)) in the rump, 17.05% (6.31–37.59) in the neck, and 21.70% (8.72–38.52) in the shoulder (Appendices L and N).

In the control animals, the cortisol concentrations measured in the rump segments of qiviut grown during the summer ACTH challenge did not differ significantly over time (p = 0.36; Table 5.3, Figure 5.6, and Appendix M). In the ACTH-injected animals, cortisol concentrations measured in the rump segments of qiviut grown during the summer ACTH challenge increased significantly from t0 + 6W to t0 + 3M and from t0 + 6W to t0 + 6M (p < 0.001; Table 5.3, Figure 5.6, and Appendix M).

Table 5.3. Median (range) cortisol levels (ng/g) in the rump segments of qiviut grown during the summer ACTH challenge (between t0 and t0 + 6W) when collected 6 weeks [t0 + 6W], 3 months [t0 + 3M], and 6 months [t0 + 6M] after the start of the challenge. Different letter subscripts indicate significant differences within each experimental group.

Control animals ACTH-injected animals

(n = 6) (n = 10) t0 + 6W 11.35 (8.05–14.45)a 14.64 (11.53–18.01)a t0 + 3M 12.38 (5.45–31.79)a 21.52 (16.24–41.08)bc t0 + 6M 15.07 (12.64–65.68)a 35.46 (12.64–65.68)c

Figure 5.6. Cortisol levels in the rump segments of qiviut grown during the summer ACTH challenge (between t0 and t0 + 6W) when collected 6 weeks [t0 + 6W], 3 months [t0 + 3M], and 6 months [t0 + 6M] after the start of the challenge in the control (n = 6) and ACTH-injected (n = 10) muskoxen. The triangles correspond to the means and the thick horizontal lines to the medians. Each thin dotted line indicates the values for an individual muskox and illustrates intra- and inter-individual variability.

5.4.2.4. Is the cortisol concentration in shed qiviut the same as prior to shedding?

Cortisol concentrations from shaved rump qiviut collected in February 2019 (median (range) = 15.30 ng/g (9.79–47.51)) were significantly lower than those measured from shed rump qiviut in April-May 2019 (median (range) = 21.52 ng/g (15.54–36.81)) (p = 0.03; Figure 5.7 and Appendix M). There was no correlation between shaved and shed qiviut cortisol levels (rs = 0.03, p = 0.92).

Figure 5.7. Cortisol levels in rump qiviut grown during the entire hair growth period when shaved in February 2019 and collected shed during qiviut combing in April-May 2019 in all muskoxen (n = 16). The triangles correspond to the means and the thick horizontal lines to the medians. Each thin dotted line indicates the values for an individual muskox and illustrates inter-individual variability.

5.4.2.5. Pre-injection FGM levels

Pre-injection FGM levels did not differ significantly between injections in the control (p = 0.84) and ACTH-injected animals (p = 0.20) (Table 5.4 and Appendix O). The same results were found when the injection number was included as an ordinal variable in the model (results not shown).

Table 5.4. Median (range) FGM levels (ng/g) measured the day before each of 5 weekly injections.

Control animals ACTH-injected animals Injection (n = 3) (n = 7) 1 9.36 (7.22–11.30) 10.15 (4.58–15.22) 2 9.76 (7.51–13.78) 10.48 (2.91–37.11) 3 4.23 (2.62–9.85) 6.72 (3.30–17.11) 4 6.94 (6.42–9.74) 10.43 (2.41–35.36) 5 7.43 (5.87–16.47) 12.48 (8.30–64.88)

5.5. Discussion

Interpretation of hair cortisol is a controversial and challenging field. In this study, our goal was to pharmacologically stimulate the adrenal glands of captive muskoxen through the administration of ACTH to address some of the key questions on hair cortisol deposition, stability, and variability across body regions. Our results illustrate that repeated ACTH injections resulted in elevated cortisol levels in qiviut only during its growth phase, and not in the absence of growth. We also highlight multiple sources of qiviut cortisol variability (i.e., among individuals, among body regions, and within a body region). Finally, we

89 demonstrate that qiviut does not seem to be a reliable tool for segmental analyses and that cortisol levels in shed qiviut are higher and not correlated with cortisol levels in qiviut shaved during the telogen (i.e., non- growing) phase.

5.5.1. Response to ACTH in growing qiviut but not in the absence of growth

We demonstrated that repeated stimulation of the adrenal glands through weekly administrations of ACTH during the hair growth phase was reflected in qiviut as an increase in cortisol concentration in all three body regions tested. The qiviut cortisol levels measured in response to the five weekly injections of 2 IU/kg ACTH were low when compared to the range of concentrations measured in wild muskoxen using the same method (3.83–84.54 ng/g; n = 211; Di Francesco et al., unpubl. data). These findings demonstrate that even small changes in HPA axis activity can be detected in qiviut and that qiviut cortisol can consequently be used as a biomarker of long-term HPA axis activity in muskoxen. The percentage increases in qiviut cortisol levels for ACTH-injected muskoxen (Table 5.2) were variable, but within the ranges reported in other wild mammalian species (Terwissen et al., 2013; Mastromonaco et al., 2014; Dulude-de Broin et al., 2019; see Appendix J for details). Inter-individual variability in the magnitude of the response to ACTH may be influenced by a diversity of intrinsic factors and their interactions. These include the health condition of individuals (Mastromonaco et al., 2014), their sex, age, reproductive status, as well as “individual” differences caused by developmental, early life and recent experiences that may sensitize or desensitize the HPA axis response (see Dantzer et al., 2014). A larger sample size would be required to adequately examine the effects of these other variables. A post-challenge increase in qiviut cortisol concentrations was observed in the controls, but was much less pronounced than that of the ACTH-injected muskoxen (Figure 5.4 and Appendix P). This moderate increase could be due to a combination of experimental and natural factors. Stress due to handling and to the injections may have influenced both experimental groups. Shaving may have facilitated local skin irritation caused by biting insects which may have activated the local HPA axis in the skin and led to increased qiviut cortisol levels. In a study on domestic sheep (Ovis aries), extensive brushing led to a higher local production of cortisol and increased HCCs (Salaberger et al., 2016). Natural stressors due to environmental conditions (e.g., heat) and/or life history events (e.g., increased social interactions with the approach of the mating season) may also have contributed to this general increase. Finally, the shaved areas may have been more exposed to UV radiation from sunlight, which may have stimulated the synthesis of both local and central cortisol, as was demonstrated in mice (Skobowiat and Slominski, 2015). This study was conducted in Fairbanks over the summer when day lengths are long (i.e., approximate day lengths of 19 h at the start and 15 h at the end of the ACTH challenge). Cortisol can be incorporated into the hair via local (i.e., in the skin and follicle itself) and central production (Figure 5.1). We were unable to assess with our study design to what extent the local production of cortisol was activated by the ACTH injections and if it contributed to the “internal” cortisol

90 concentrations measured. The administration of radiolabeled cortisol would have allowed us to explore this question, but was not feasible at the time of this study. We also demonstrated that neither a single nor repeated ACTH injections altered qiviut cortisol levels in the absence of hair growth. The few available studies on HCCs during the telogen phase have conflicting results regarding the relative stability of this hormone (Ashley et al., 2011; Cattet et al., 2014). Studies on brown bears (Ursus arctos) showed that HCCs could change rapidly in response to capture stress (an acute, singular event) and that cortisol, progesterone, and testosterone levels in hair could increase markedly in the absence of hair growth (Cattet et al., 2014, 2017). However, a study on reindeer (Rangifer tarandus tarandus) and caribou (Rangifer tarandus granti) showed no effect of a single dose of ACTH, representing an acute event, on the cortisol concentration of hair collected one week later (Ashley et al., 2011). Although we used a lower dose of ACTH (1 IU/kg instead of 2 IU/kg in caribou and reindeer and during the summer ACTH challenge), this did stimulate HPA axis activity as evidenced by the response detected in the feces of at least two muskoxen (Chapter 4). Qiviut is very low in secretions from both sebaceous and apocrine (sweat) glands, which is evidenced by its high scoured yield (Rowell et al., 2001). Apocrine glands are associated with guard hair follicles and only small sebaceous lobules are associated with some of the secondary “qiviut” follicles (Flood et al., 1989). Thus, the amount of “external” cortisol from glandular sources on the qiviut shaft is likely limited and of negligible impact on the “internal” cortisol concentration. Divergent results in the effect of short-term stressors on HCCs during the telogen phase may be related to sample analysis procedures, the hair type analyzed (e.g., guard hairs in grizzly bears versus woolly undercoat in muskoxen), and/or to species-specific differences in the abundance of glandular secretions. This highlights the importance of validating the methods for each species and hair type of interest.

5.5.2. Sources of qiviut cortisol variability and implications

Understanding the variability of cortisol levels in hair from different body regions is important to ensure comparability among and within studies and individuals. We detected some variability among individuals and differences among body regions with not all muskoxen exhibiting the same patterns (Figure 5.5 and Appendix P). Studies in a variety of wild and domestic mammalian species have conflicting results regarding the variability in HCCs among body regions (Appendix Q). Hypotheses put forth when such variations are observed include differences in hair/wool fiber structure and characteristics (Ashley et al., 2011; Yamanashi et al., 2013; Fürtbauer et al., 2019), hair color (Ashley et al., 2011; Acker et al., 2018; Heimbürge et al., 2020a), hair growth and molting patterns (Macbeth et al., 2010; Burnard et al., 2017; Heimbürge et al., 2019), skin blood-flow (Carlitz et al., 2015), abundance of glandular secretions (Macbeth et al., 2010; Ashley et al., 2011), external contamination (Ghassemi Nejad et al., 2019; Heimbürge et al., 2020a), and rates of cortisol “washout” due to dissimilar weather exposure (Heimbürge et al., 2019) or grooming (Acker et al., 2018). For muskoxen, the most likely explanations are differences in (i) qiviut fiber diameter (Robertson, 2000); (ii) qiviut growth onset, timing, and rate, which may also vary among

91 individuals and depending on their age, sex, and lactating status (Robertson, 2000); (iii) the amount of dandruff in the sample (i.e., pieces of dead skin possibly containing cortisol), which were extremely difficult to remove; and (iv) cortisol “washout” due to grooming behaviors such as rubbing against rocks and trees or on the ground while sprawling (Gray, 1987). We also highlighted an important intra-sample variability in qiviut cortisol concentrations, which was lower for fully grown qiviut. These differences may be related to intrinsic variations in qiviut fiber structure and characteristics (i.e., length, diameter, dryness, etc.). Potential variations in qiviut cortisol concentrations among body regions emphasize the importance of standardized sampling. Selection of sampling site will be driven by logistical, cultural, and economic constraints, but should also be guided by sample quality and consistency of cortisol quantitation. If the trend towards lower variability in rump hair is confirmed, this may be a preferential site for sampling. Regardless of sampling site, duplicate extractions should be standard due to the within site variability.

5.5.3. Sources of qiviut cortisol variability and implications

Cortisol levels in the qiviut segment grown during the summer challenge increased significantly over time in the ACTH-injected muskoxen; a similar trend also occurred in the control animals (Table 5.3 and Figure 5.6). This may have been in part due to the imprecision in isolating the exact hair segment in that hair could not be cut precisely because of the very fine and static nature of qiviut fibers (see section 5.3.3.2 and Figure 5.2). However, it is unlikely that this imprecision would be biased towards increased qiviut cortisol as there was no evidence of major stressors outside of the ACTH injections. Another possibility is the movement of cortisol molecules within the qiviut shaft after being deposited which has been demonstrated in hair of rhesus monkeys using radiolabeled [3H]-cortisol (Kapoor et al., 2018). The ACTH injected was a slow-release gel formulation with little known on the pharmacokinetics in muskoxen, particularly regarding the duration of its release and possible residual effects or sequestration in the fat. Ongoing release with subsequent deposition of cortisol in qiviut post-challenge may have influenced cortisol concentrations beyond t0 + 6W, particularly if cortisol is mobile in the qiviut shaft. However, the similar FGM levels measured before each ACTH injection suggest an absence of such longer-term effects. Another hypothesis, which would be valid both for the shed qiviut and the segmental analyses, involves the possible damage to the surface structure of the shaft as it ages and is exposed to external elements (i.e., rain, snow, wind, UV radiation from sunlight, etc.) and friction from the surrounding hairs (Richena and Rezende, 2016; Heimbürge et al., 2020b, 2020a). The “external” cortisol from glandular secretions and extrinsic sources may be more easily incorporated into a damaged shaft (Heimbürge et al., 2020b, 2020a; Figure 5.1). Wool fibers differ from hair fibers by their smaller diameter and thinner cuticle (Feughelman, 1997) and qiviut fibers are extremely fine (Rowell et al., 2001), which may render them particularly fragile. However, this phenomenon of “external” cortisol incorporation would have affected control and ACTH- injected animals equally and seems unlikely in muskoxen as glandular secretions are limited (Flood et al., 1989; Rowell et al., 2001), qiviut is protected by the surrounding guard and intermediate hairs, and qiviut is

92 extremely dense with muskoxen having a higher secondary to primary follicle ratio than any other wild or domestic ruminant species, thereby providing even more protection (Flood et al., 1989). Other phenomena causing the qiviut fibers to become lighter, such as dehydration, may have contributed to the cortisol increases observed both in shed qiviut and segmental analyses. We measure cortisol per mass of qiviut, not by number of fibers, thus if the qiviut dehydrates over time, we would have included more qiviut fibers at later sampling periods, including when it was shed, which would have led to higher concentrations per gram, but not necessarily per fiber. Some studies in cattle (Bos taurus), orangutans (Pongo spp.), and humans suggest that segmental analysis of hair can be used as a retrospective calendar of HPA axis activity, providing a more refined temporal calendar of past stressful events (Kirschbaum et al., 2009; Carlitz et al., 2014; Heimbürge et al., 2020b). Our results for muskox qiviut indicate that cortisol may not be stationary in a fiber segment and/or that changes in shaft structure over time may influence segment concentrations. This discourages the use of qiviut fibers as a retrospective calendar of HPA axis activity in muskoxen. The higher cortisol concentrations in shed qivut may also have been influenced by contamination with the greater amount of dandruff present compared to the fully grown shaved qiviut (Di Francesco, pers. obs.). Shed fibers also included the follicle base, which may have contained higher concentrations of cortisol reflecting local synthesis and/or long-term storage within the follicle (see Cattet et al., 2017); shaved qiviut does not include that portion of the fiber shaft and base embedded in the skin. While many of the follicles were removed manually after visual identification before hormone analyses, it is likely that a significant proportion remained, which may have increased the cortisol levels measured (Cattet et al., 2017; Sergiel et al., 2020). The inclusion of skin fragments and follicles reflecting more recent time-frames of HPA axis activity may explain the absence of correlation between fully grown and shed qiviut cortisol levels. While it would be logistically and financially attractive to measure cortisol concentrations in shed qiviut samples collected directly on the tundra, our results demonstrate that these samples are not comparable and not correlated to samples of fully grown qiviut. Further studies are required to determine what the cortisol levels measured in shed qiviut reflect and whether these could serve as a potential monitoring tool. In any case, the potential issues due to follicle contamination highlighted by our findings in shed qiviut support standardizing sample collection to shaving/cutting versus plucking of hair.

5.6. Conclusion

Our results demonstrate that repeated stimulation of the adrenal glands through weekly administrations of ACTH was reflected in qiviut as an increase in cortisol concentration only during its growth phase, and not in the absence of growth. Qiviut cortisol concentrations are consequently not easily affected by short or long-term stressors during the telogen phase. The segmental analyses suggest that there is movement of cortisol within the qiviut shaft after deposition and/or significant changes in fiber structure over time, however, there is no evidence that cortisol disappears. Qiviut cortisol can, consequently, be used

93 as an integrated measure of HPA axis activity over the period of the hair’s growth, provided that the qiviut shaft is cut/shaved and analyzed in its entirety. The cortisol levels measured in response to the five weekly injections of 2 IU/kg ACTH were low when compared to the range of qiviut cortisol concentrations measured in wild muskoxen using the same method. Even small changes in HPA axis activity can, therefore, reliably be detected in qiviut. Now that qiviut cortisol has been validated as a biomarker of HPA axis activity, it may be added to the health assessment tool box and used as a monitoring tool to evaluate the impacts that multiple and increasing stressors linked to accelerated climate change and other anthropogenic disturbances have on wild muskox individuals and populations. This will be critical to inform management and conservation efforts. However, the variability among body locations and between shaved and shed qiviut highlight the importance of monitoring programs following strict design and methodology for sample collection and analyses in order to account for the multiple sources of variation. This work adds muskoxen to the very small number of wildlife species in which HCCs have been validated as a biomarker of long-term HPA axis activity. It also furthers our understanding of cortisol deposition and stability patterns in hair while simultaneously providing critical information on the applications and limitations of qiviut cortisol as a biomarker of HPA axis activity in muskoxen.

5.7. Acknowledgements

We are grateful to all the staff and summer students working at the Robert G. White Large Animal Research Station of the University of Alaska Fairbanks who helped in sample collection and animal handling, particularly to Sarah Barcalow, Thalia Souza, Hanna Sfraga, Carla Wiletto, Christine Terzi, Megan Roberts, Jean Rein, Claire Kepner, and Charles Ashlock. We wish to thank Christine Gilman and Patricia Medd from the Endocrinology Laboratory of the Toronto Zoo for their hard work in method development and sample analyses, as well as Angie Schneider, Pauline Humez, and Morgan Mouton for their help in sorting the qiviut. We finally thank James Wang for his support in the lab, Katherine Wynne-Edwards for her input during the conceptualization of this study, and Karin Orsel for her advice regarding the statistical analyses. A special thought also goes to all the muskoxen who took part in this study.

5.8. Funding

CHAPTER 6. INTRINSIC AND EXTRINSIC FACTORS ASSOCIATED WITH INCREASED QIVIUT CORTISOL IN WILD MUSKOXEN (OVIBOS MOSCHATUS)

Juliette Di Francesco1, Grace P.S. Kwong2, Rob Deardon3,4, Sylvia L. Checkley1, Gabriela F. Mastromonaco5, Fabien Mavrot1, Stephanie Peacock1, Lisa-Marie Leclerc6, Susan Kutz1

1Department of Ecosystem and Public Health, Faculty of Veterinary Medicine, University of Calgary, 3280 Hospital Drive NW, Calgary, Alberta, Canada T2N 4Z6

2Faculty of Veterinary Medicine, University of Calgary, 3280 Hospital Drive NW, Calgary, Alberta, Canada T2N 4Z6

3Department of Production Animal Health, Faculty of Veterinary Medicine, University of Calgary, 3280 Hospital Drive NW, Calgary, Alberta, Canada T2N 4Z6

4Department of Mathematics and Statistics, Faculty of Science, University of Calgary, 2500 University Drive NW, Calgary, AB, Canada T2N 1N4

5Reproductive Physiology Unit, Toronto Zoo, 361A Old Finch Avenue, Scarborough, Ontario, Canada M1B 5K7

6Department of Environment, Government of Nunavut, P.O. Box 377, Kugluktuk, Nunavut, Canada X0B 0E0

Chapter intended to be submitted to the journal Conservation Physiology or PeerJ

6.1. Abstract

Glucocorticoid levels (cortisol in muskoxen (Ovibos moschatus)) are increasingly and widely used as biomarkers of hypothalamic–pituitary–adrenal (HPA) axis activity to study the effects of various environmental changes and challenges on wildlife individuals and populations. In muskoxen, qiviut (fine woolly undercoat hair) cortisol concentrations can be used as an integrated biomarker of HPA axis activity over the course of the hair’s growth. We gathered cross-sectional data from harvested wild muskoxen in the Canadian Arctic between October 2015 and May 2019. Using this dataset, we examined the relationship between qiviut cortisol and various intrinsic (sex, age, body condition, and incisor breakage) and extrinsic biotic factors (lungworm and gastro-intestinal parasite infection intensities, parasite richness, exposure to bacteria), as well as broader non-specific landscape and temporal features (geographic location, season, and year). A Bayesian approach, which allows for the joint estimation of missing values in the data and model parameters estimates, was applied for the statistical analyses. Main findings included: (i) higher qiviut cortisol levels in males than in females; (ii) strong inter-annual variations; (iii) differences among geographical locations that were consistent with local muskox population trends (qiviut cortisol levels higher in the declining population than in the stable/increasing population); (iv) a negative association between qiviut cortisol and marrow fat percentage, but not with other metrics of body condition; and (v) a relationship between qiviut cortisol and the infection intensity of the lungworm Umingmakstrongylus pallikuukensis which varied depending on the geographical location, and otherwise an absence of association between qiviut cortisol and pathogen exposure/infection intensity metrics. Identifying factors influencing GC levels is a key step in endocrinology studies to ensure accurate monitoring of GC responses to environmental changes and disturbances. Additionally, this study provides important insights regarding the relationship between GC levels and pathogen exposure/infection intensity.

6.2. Introduction

Climate warming and other anthropogenic changes are currently occurring at an unprecedented pace worldwide and it is becoming crucial to monitor the effects that these changes and their associated stressors have on wildlife species (IPCC, 2014; Ribera d’Alcalà, 2019). This is particularly important in high latitude environments such as the Arctic, where these changes have been especially rapid and substantial (AMAP, 2017). Muskoxen (Ovibos moschatus), an iconic Arctic ungulate species, are essential for the nutrition, economy, and culture of local Indigenous communities (Tomaselli et al., 2018a; Cuyler et al., 2020). Over the past two decades, muskoxen have experienced large population declines on Banks and Victoria Islands in the western Canadian Arctic Archipelago, which are still ongoing today (Tomaselli et al., 2018b; Cuyler et al., 2019). These declines are likely linked to multiple factors, including ecological modifications associated with climate change, icing events, and an increased susceptibility to diseases and/or changes in exposure to pathogens (Nagy and Gunn, 2009; Nagy et al., 2009b; Kutz et al., 2017). Erysipelothrix rhusiopathiae, an opportunistic zoonotic bacterium, has been identified as the cause of significant widespread muskox mortality events on both islands (Kutz et al., 2015; Mavrot et al., 2020). Brucella-like syndromes have been increasingly described by Ekaluktutiak harvesters (Victoria Island, Nunavut (NU)) and the apparent Brucella seroprevalence has increased in muskoxen on Victoria Island since the early 2000s (Tomaselli et al., 2018b, 2019). Two protostrongylid lungworms, Umingmakstrongylus pallikuukensis (Up) and Varestrongylus eleguneniensis (Ve), have expanded their range on Victoria Island over the past decade (Kutz et al., 2013a; Kafle, 2018), and severe dental anomalies, such as broken incisors, are more frequently observed (Mavrot et al., unpubl. obs.). The recent population declines, along with the very low genetic diversity of muskoxen (Prewer et al., 2020), suggest that they may be particularly threatened by the multiple environmental changes and associated stressors to which they are increasingly exposed (Kutz et al., 2017; Cuyler et al., 2019). The activation of the hypothalamic–pituitary–adrenal (HPA) axis, which leads to the release of glucocorticoids (GCs) (mainly cortisol in muskoxen (Koren et al., 2012c)), is an important part of the physiological stress response in mammals and plays a key role in energy regulation (Romero and Butler, 2007; Busch and Hayward, 2009). Glucocorticoid levels are increasingly and widely used as biomarkers of HPA axis activity to study the effects of various environmental changes and challenges on wildlife individuals and populations (Busch and Hayward, 2009; Dantzer et al., 2014). These hormones are incorporated and can be readily quantified in various biological matrices including blood, saliva, feces and urine (in the form of metabolites), and hair (reviewed in Dantzer et al., 2014). Among these matrices, blood, saliva, urine, and feces provide short-term information on an animal’s GC levels at a single point in time or over hours to days, whereas hair is thought to provide a cumulative measurement of GC levels during the period of the hair’s growth – this may be weeks to months depending on turnover patterns and sampling regime (Sheriff et al., 2011; Russell et al., 2012). Hair GC concentrations may, therefore, represent a non- specific integrative measure of all the stressors experienced by the animal over the course of the hair’s growth (Cattet et al., 2014), but are also influenced by a wide variety of extrinsic (e.g., season, social status)

97 and intrinsic (e.g., sex, reproductive status) factors that may affect both baseline GC levels and the magnitude of the response to stressors (Dantzer et al., 2014; Heimbürge et al., 2019). It is important that these influencing factors be identified and controlled for in the species of interest to ensure that GC responses to environmental changes and disturbances are accurately monitored. While many studies have assessed the relationship between GC levels and a single or several extrinsic and/or intrinsic factors (e.g., Fardi et al., 2018; Salas et al., 2016), the inclusion of multiple factors in the analyses is rare (but see Santos et al., 2018). In particular, the studies that have investigated associations between GC levels and pathogens (i.e., bacteria, fungi, viruses, and parasites (protozoa, helminths, and arthropods)) have generally focused on a single type of pathogen with inclusion of few other factors (e.g., Chapman et al., 2006; Madslien et al., 2020, but see Cizauskas et al., 2015; Hoby et al., 2006). Cortisol levels in the qiviut (fine woolly undercoat) of muskoxen can be used as biomarkers of HPA axis activity during the period of the hair’s growth (Chapter 5). Here, we use a large-scale cross- sectional study of wild muskoxen to examine the relationship between qiviut cortisol levels and various intrinsic (sex, age, body condition, and incisor breakage) and extrinsic biotic factors (lungworm and gastrointestinal (GI) parasite infection intensities, parasite richness, exposure to bacteria), as well as broader non- specific landscape and temporal features (geographic location, season, and year). Building on our previous work (Di Francesco et al., 2017), this study enables us to confirm and further identify important sources of variability in qiviut cortisol levels.

6.3. Material and methods

6.3.1. Animals and sampling procedure

This study was approved by the Veterinary Sciences Animal Care Committee, University of Calgary (protocol #AC13-0121). Samples were obtained under the Wildlife Research Permit #2016-058 for NU and the Wildlife Research Permits #WL500257, WL5004469, and WL500664 for the Northwest Territories (NWT). Samples from 219 muskoxen, harvested as part of a community-based muskox health surveillance program in Ekaluktutiak and Kugluktuk, NU and in Ulukhaktok, NWT, were collected by hunters between October 2015 and May 2019 (Figure 6.1). The surveillance program was developed through a partnership among these communities, a guided hunting organization, government biologists, and academic researchers. Samples were obtained through individual subsistence, community, and guided hunts. Hunters and guides collected samples using standardized kits available to them at the local wildlife office and/or Hunters’ and Trappers’ Organization. Participation in this program was voluntary and they received financial compensation for each completed kit. Kits were modified from those described by Kutz et al. (2013b) for caribou. A fully completed kit would include: blood collected on Nobuto filter paper strips (Toyo Roshi Kaisha, Ltd., Tokyo, Japan), feces, the lower left hind leg, the lower jaw, and a piece of skin with fur from the rump measuring approximately 10 cm × 10 cm (Appendix R). Information regarding the GPS coordinates of the kill location, the age class and sex of the muskox, a measure of its back fat thickness,

98 a subjective assessment of its body condition (i.e., choice between the categories “skinny,” “not bad,” “fat,” or “really fat”), and the description of any relevant abnormality observed during butchering was recorded by the hunter using the datasheet provided with the kit (Appendix S). All samples were stored at -20°C until analysis. Not all returned kits and information forms were complete, and the amount of a sample collected was sometimes not sufficient for all the tests to be done (e.g., fecal samples and parasitology testing), which led to missing values in the dataset (Table 6.1 and Figure 6.2). Samples were classified to two different seasons based on their collection date: late fall-early winter (last week of September to end of December) and mid-late winter (January to mid-May). This was decided for consistency with a previous seasonal classification (Di Francesco et al., 2017), and based on the timing of qiviut growth, which occurs approximately from early April to late November with no growth observed between mid-December and March (Flood et al., 1989; Robertson, 2000). The year was defined as the year of qiviut growth instead of the year of sample collection (i.e. muskoxen harvested in January through mid- May were assigned to the year prior to harvest as the qiviut sampled had grown during the previous year). When left blank or indicated as “unknown” on the information sheet, the sex of the muskox was determined using genetic analyses as previously detailed by Di Francesco et al. (2017). The age was categorized as “juvenile” (calves and yearlings) or “adult” (> 2 years) for the analyses. This was estimated either based first on the teeth eruption patterns when the lower jaw was collected (Henrichsen and Grue, 1980), or by the hunter if the jaw was not available. If neither of these sources of information were available, the age was determined based on the metatarsus length. We plotted the distribution of metatarsus length for animals of known age and developed a conservative cut-off of 16 cm above which muskoxen were classified as adults and below which the age remained “unknown.” The eight muskoxen with an “unknown” age were removed from the statistical analyses, as we chose not to impute the missing data for this variable, leaving a total of 211 animals to be included.

Figure 6.1. Map showing the five specific geographical locations from which muskox sampling kits were obtained (communities of Ulukhaktok, Kugluktuk, and Ekaluktutiak (black and white stars), and Lady Franklin Point and Kent Peninsula (black arrows)), with the geo-referenced harvesting locations of the muskoxen when available (blue points). Geographic coordinates were unavailable for 15 of the 211 muskoxen included in the statistical analyses; these animals were assigned to a specific geographical location based on the muskox management zone in which they were harvested and on the community from which the kit was submitted (Ekaluktutiak: n = 8; Kugluktuk: n = 2; Ulukhaktok: n = 3; Kent Peninsula: n = 2) (map generated in QGIS version 2.8.9).

100

Table 6.1. Variables, and their respective abbreviations, descriptions, and sample sizes, evaluated as potential predictors of qiviut cortisol levels in the n = 211 muskoxen. None of the continuous variables were normally distributed, so all are summarized as median (range). Missing values (i.e., insufficient sample for all laboratory analyses or information not recorded on data sheet), except those for back fat thickness and incisor breakage score, were estimated through the Bayesian analyses.

Number of Information Variable Variable description muskoxen (%) Categorical Season General fall-early winter 75 (35.5%) (season) mid-late winter 136 (64.5%) Categorical 2015 28 (13.3%) Year (i.e., hair growth year) 2016 41 (19.4%) (year) 2017 80 (37.9%) 2018 62 (29.4%) Categorical Sex female 91 (43.1%) (sex) male 120 (56.9%) Categorical Age adult 191 (90.5%) (age) juvenile 20 (9.5%) Categorical Ekaluktutiak 39 (18.5%) Specific geographical Kent Peninsula 30 (14.2%) location Kugluktuk 70 (33.2%) Lady Franklin Point 13 (6.2%) Ulukhaktok 59 (27.9%) Categorical

east mainland (Kent (14.2%) 30 Peninsula)

Broad geographical location Victoria Island (52.6%) 111 (location) (Ekaluktutiak, Lady

Franklin Point, and

Ulukhaktok) (33.3%) 70 west mainland (Kugluktuk) Umingmakstrongylus Continuous Lungworms pallikuukensis larval counts 43.08 lpg (0–1,669.81) 200 (94.8%) (Up_lpg) missing 11 (5.2%) Varestrongylus eleguneniensis Continuous larval counts 1.33 lpg (0–442.74) 200 (94.8%) (Ve_lpg) missing 11 (5.2%) Categorical 0 28 (13.3%) Lungworm richness 1 54 (25.6%) (lung_richness) 2 118 (55.9%) missing 11 (5.2%) Categorical Moniezia spp. infection no 152 (72.0%) GI parasites (moniezia_YN) yes 23 (10.9%) missing 36 (17.1%) Continuous Eimeria spp. oocyst counts 2.48 epg (0–822.60) 175 (82.9%) (eimeria_epg) missing 36 (17.1%) Nematodirines egg counts Continuous

(nematodirines_epg) 0 epg (0–16.10) 175 (82.9%)

101

missing 36 (17.1%) Marshallagia marshalli egg Continuous counts 0 epg (0–10.44) 175 (82.9%) (marshallagia_epg) missing 36 (17.1%) Categorical 0 5 (2.4%) 1 61 (28.9%) GI parasite richness 2 58 (27.5%) (GI_richness) 3 41 (19.4%) 4 10 (4.7%) missing 36 (17.1%) Metatarsus percent marrow Continuous Body fat+ 0.91 (0.10–0.95) 197 (93.4%) condition (marrow_fat) missing 14 (6.6%) Categorical skinny 21 (9.9%) Hunter condition not bad 53 (25.1%) assessment fat 93 (44.1%) (condition_hunter) really fat 20 (9.5%) missing 24 (11.4%) Continuous Back fat thickness 2 cm (0–6.35) 145 (68.7%) (back_fat) missing* 66 (31.3%) Erysipelothrix percent Continuous Bacteria positivity+ 0.22 (0.01–4.96) 203 (96.2%) exposure (erysipelothrix_PP) missing 8 (3.8%) Categorical Brucella serology negative 192 (91.0%)

(brucella_serology) positive 11 (5.2%) missing 8 (3.8%) Continuous Incisor breakage score Jaw health 0 (0–1) 152 (72.0%) (incisor_breakage) missing* 59 (28.0%) *Not estimated through Bayesian analyses. +Values presented here are percentages divided by 100 for the purposes of the statistical analyses.

102

Figure 6.2. Percentage of missing values for each variable (a) and missing data by animal (b). 6.3.2. Sample analyses

6.3.2.1. Parasitology

Approximately 5 g from each fecal sample was weighed (as little as 2 g were used if the amount was limited) and analyzed using the modified beaker Baermann technique for the presence of protostrongylid lungworm first stage larvae (L1) (Forrester and Lankester, 1997). Modifications were the following: (i) tissue paper was replaced by a single layer of cheesecloth; (ii) samples sat for 5 min before the supernatant was aspirated (using gently vacuum suction) down to 15 ml and transferred in a test tube; (iii) the beaker was rinsed with 2 ml of tap water, which were added to the test tube; and (iv) samples were then centrifuged, supernatant aspirated down to 2 ml and depending on the amount of L1 expected, either the entire sample or three 50 or 100 µl aliquots were analyzed. The L1 recovered were identified microscopically by an experienced observer under 40X magnification as Umingmakstrongylus pallikuukensis or Varestrongylus eleguneniensis based on their caudal morphology following the keys developed by Kafle et al. (2015). Each species was quantified separately and L1 counts are expressed as larvae per gram of feces (lpg). Approximately 2–4 g (depending on the amount available) from each fecal sample was weighed and evaluated for the presence of GI parasite eggs and oocysts using the modified Wisconsin double- centrifugation sugar flotation technique (Egwang and Slocombe, 1982). Modifications included straining the water-fecal mixture through a single layer of cheesecloth instead of a tea strainer and centrifuging the samples for 10 min at 1,500 rpm both times. The eggs and oocysts recovered were counted and identified

103 microscopically by an experienced observer under 100X magnification to the group or species level (nematodirines (Nematodirus spp. and Nematodirella spp.), Marshallagia marshallagi, Eimeria spp., or Moniezia spp.) based on their morphological characteristics. The eggs of Teladorsagia boreoarcticus and Ostertagia gruehneri, two major GI parasites of muskoxen, were not counted as egg production occurs mainly throughout the summer (reviewed in Kutz et al., 2012) and freezing of fecal samples results in a rapid decrease in the number of detectable eggs of these species (De Bruyn, 2010). All egg and oocyst counts are expressed as eggs per gram of feces (epg), except for Moniezia spp. for which only the absence or presence of eggs was recorded as entire proglottids are typically shed in the feces and quantification of eggs in feces is not considered representative of actual parasite burden (Taylor et al., 2016). For the purpose of this study, we assumed that larval and egg counts served as proxies for their corresponding parasite infection intensities (with the exception of Moniezia spp. as explained above). The correlations between the number of adult worms in the lungs or GI tract and the amount of L1, eggs or oocysts shed in the feces have not been investigated in muskoxen. However, a study in Svalbard reindeer (Rangifer tarandus platyrhynchus) showed that fecal egg counts were positively associated with adult worm burden for Marshallagia marshalli (Irvine et al., 2001). An experimental study on a limited number of muskoxen suggested a similar relationship between fecal larval counts and adult worm burden for Up (Kutz et al., 1999). Parasite richness was subsequently recorded for lungworms and GI parasites by counting the number of different species (lungworms) and species or groups (GI parasites) infecting each muskox.

6.3.2.2. Hormone analyses

For each animal, qiviut, intermediate hairs, and guard hairs from an area free from obvious blood or other contamination were cut away from the skin using a scalpel blade. Qiviut was manually separated from guard and intermediate hairs using forceps and was divided into two subsamples of 0.05 g, which were independently analyzed. Qiviut cortisol analyses were done at the Endocrinology Laboratory of the Toronto Zoo following the procedures detailed in Chapter 5 section 5.3.5. The mean concentration between the two subsamples, measured in nanograms of cortisol per gram of qiviut (ng/g), was used as the final response variable. Duplicate subsamples had a median coefficient of variation (CV, calculated as (standard deviation/mean) × 100) of 12.70% (range = 0.14–83.26%).

6.3.2.3. Serology

Shortly after reception of the samples, filter paper sets were removed from their envelopes and placed in racks at room temperature in a fume hood to air-dry for at least 24 h. They were then eluted following the method described by Curry et al. (2011) and eluates were stored at -20°C until testing. Eluates were tested for antibodies against the bacterium Erysipelothrix rhusiopathiae with a modified enzyme-linked immunosorbent assay (ELISA) (Mavrot et al., 2020). Results were standardized across all ELISA plates and expressed as percent positivity (PP) of a reference positive control corresponding to a pool of five muskox samples with values of optic density (OD) close to 1. The PP was calculated as

104

(ODsample - ODblank)/(ODcont - ODblank) × 100, with ODsample corresponding to the OD value measured for the sample, ODcont to the OD value of the positive control, and ODblank to the OD value of the ELISA plate’s blank well (Mavrot et al., 2020). Samples were tested for antibodies against the bacterium Brucella suis biovar 4 using two different indirect enzyme-linked immunoassays (i-ELISA). Samples from all hair growth year 2015 (n=28) were analyzed at the Arctic University of Norway, Research Group for Arctic Infection Biology (Tromsø, Norway; Nymo et al., 2013). The remaining samples (n = 183) were analyzed at the Canadian Food Inspection Agency National Brucellosis Reference Laboratory (Ottawa, Canada; Nielsen et al., 1994, 2004). In both i-ELISAs, the antigen was smooth lipopolysaccharide from Brucella abortus, and the enzyme conjugate was a peroxidase conjugated chimeric protein A/G (Nielsen et al., 2004; Nymo et al., 2013).

6.3.2.4. Marrow fat measurement and other indices of body condition

The percentage of metatarsus marrow fat was measured following the protocol described by the CircumArctic Rangifer Monitoring and Assessment (CARMA) Network (2008), except marrow samples were dried at room temperature instead of in an oven (CARMA, 2008). The percentage of fat was calculated as (dry weight/fresh weight) × 100. We used the percentage of metatarsus marrow fat as a metric of body condition, along with the back fat thickness and the subjective body condition assessment by the hunter.

6.3.2.5. Lower jaw analyses

Lower jaws were assessed for incisor breakage (both permanent and deciduous). The incisor breakage score was calculated as number of incisors broken/number of incisors examined.

6.3.3. Statistical analyses

The effects of the different intrinsic (sex, age, body condition, jaw health), extrinsic abiotic (season, hair growth year, and broad geographical location), and extrinsic biotic (lungworms, GI parasites, bacteria exposure) factors (Table 6.1) on qiviut cortisol were assessed using a Bayesian approach. This method allows for the joint estimation of missing values in the data and model parameters, and consequently takes into account the uncertainty due to missing data when estimating parameters (see Erler et al., 2016). Back_fat and incisor_breakage were not included in the original modeling procedure as more than 25% of records for these variables were missing. The effects of these two variables and their corresponding interactions on qiviut cortisol were later assessed using the final model chosen from subsets of the remaining variables, and the subset of animals for which they had been recorded. Correlation analyses were performed to check for multicollinearity between the explanatory variables using Spearman’s rank correlation coefficient (rs) as these were not normally distributed (McCrum- Gardner, 2008). Biologically and ecologically plausible two-way interactions were originally identified through discussions among co-authors and based on the current knowledge in the available scientific literature (e.g., Baker et al., 2013). Only those for which the data were fairly balanced across the various levels of the variables were considered. Tested interactions included those between sex and season, sex and

105 year, sex and age, year and season, age and season, sex and body condition (i.e., marrow_fat, condition_hunter, and back_fat), season and body condition, age and body condition, location and lungworm larval counts (i.e., Up_lpg and Ve_lpg), age and lungworm larval counts, age and GI parasite egg/oocyst counts (i.e., moniezia_YN, eimeria_epg, nematodirines_epg, and marshallagia_epg), season and incisor_breakage, year and disease exposure (i.e., brucella_serology and erysipelothrix_PP), sex and disease exposure, age and disease exposure, and location and erysipelothrix_PP. To determine which interactions would be important to keep for model building, we fit a model with all the main effects to which we added independently each identified interaction to check the impact on the deviance information criterion (DIC) (Spiegelhalter et al., 2002; Gelman et al., 2004) and whether it has an “important” effect on qiviut cortisol (based on the 95% credible interval (CrI)). For the purpose of parsimony, we only retained the three interactions in the model that produced the best improvement (i.e., largest reduction) of the DIC when compared with the model including only all the main effects (i.e., age and Ve_lpg, sex and year, year and erysipelothrix_PP), as well as some others which did not substantially improve the DIC, but which were deemed “important” (i.e., location and Up_lpg, sex and season) (Appendix T). The resulting model, which we term the “stage 2” model, included the five interaction terms and all the main effects, except for brucella_serology because of its absence of effect, low prevalence, and unbalanced data (e.g., no positive muskoxen in west mainland). From that “stage 2” model, we proceeded through manual backward elimination for simplification and sequentially removed the variables or group of variables (i.e., GI parasite egg/oocyst counts) that did not appear to have an “important” effect on qiviut cortisol when considering their 95% CrI, each time also checking the impact of their removal on the DIC and on the parameter posterior estimates of the other variables (Appendix U). As indicated in Spiegelhalter et al., (2002), a difference in DIC < 2 was considered “not important,” with variables being removed when their elimination resulted in such an increase. Checking the impact of each variable’s removal on the parameter estimates of the other variables ensured that possible multicollinearity issues were unlikely to affect the results. An interaction was never included without its main effects also in the model. For the Bayesian approach, we considered our data as missing at random (i.e., the propensity of a value to be missing is related to some of the observed data and does not depend on the missing data), and consequently the missing data mechanism as “ignorable” (Ma and Chen, 2018). For all model parameters, we assumed a relatively flat normal prior distribution with mean zero and standard deviation of 1,000, so that the parameter estimates were unlikely to be influenced by our choice of prior distribution. Additional prior information was included for the random imputation of certain variables (i.e., the values of some explanatory variables were taken into account when imputing the missing data): (i) specific geographical location for disease exposure; (ii) specific geographical location and age for lungworm larval counts; (iii) specific geographical location, age, and season for GI parasite egg/oocyst counts; (iv) sex, season, specific geographical location, Brucella serology, and lungworm and GI parasite egg/oocyst counts for metatarsus marrow fat, and subsequently the same variables and the marrow fat for the hunter condition assessment. For GI parasite egg/oocyst and lungworm larval counts, all animals with an imputed value ≤ 0.1 epg or lpg (the lowest amount of eggs/oocysts or L1 measured above zero) were considered not infected and assigned

106 a count of zero. We assumed that the response variable, qiviut cortisol, followed a gamma distribution as its values were positive and right-skewed, and Up_lpg was rescaled using a square-root (sqrt) transformation for better model fit. Two parallel Monte Carlo Markov chains (MCMC) were run with 900,000 iterations per chain and a burn-in period of 10,000 iterations (i.e., we discarded the first 10,000 samples prior to convergence to the stationary distribution). We kept every 1,000th sampled value, after a stationary distribution was reached, when estimating parameters to reduce the correlation between consecutive values of the chain. Marginal posterior distributions and the parameter trace plots were examined to check the convergence of the chains. Mixing of the chains was assessed visually using the MCMC trace plots for each parameter (Appendix V). Posterior median estimates and 95% CrIs are shown. All statistical analyses were performed using the R version 3.6.1 and rjags version 4-10 package was used for Bayesian data analysis (Plummer, 2019; R Core Team, 2019).

6.4. Results

We found a fairly high correlation only between Up_lpg and Ve_lpg (i.e., rs = 0.69, p < 0.001; all other rs were < 0.40). The final model obtained after manual backward elimination included sex, season, year, location, marrow_fat, sqrt(Up_lpg), as well as the interactions between location and sqrt(Up_lpg), and between sex and season. Overall, males had higher qiviut cortisol levels than females, and qiviut cortisol levels did not differ between late fall-early winter and mid-late winter (Table 6.2 and Figure 6.3). There appeared to be an interaction between sex and season, with males having lower qiviut cortisol levels in mid-late winter than in late fall-early winter, whereas females had similar qiviut cortisol levels during the two seasons (Figure 6.4). Metatarsus marrow fat was negatively associated with qiviut cortisol. Qiviut cortisol levels were lower in 2016 and 2017 than in 2015, but similar in 2015 and 2018, and decreased from east mainland (Kent Peninsula) to Victoria Island (Lady Franklin Point, Ekaluktutiak, and Ulukhaktok) to west mainland (Kugluktuk) (Table 6.2 and Figure 6.3). There appeared to be an interaction between Up_lpg and location. Qiviut cortisol levels tended to increase with Up larval counts on west mainland, whereas they tended to decrease on east mainland, and Up larval counts did not seem to have an effect on qiviut cortisol on Victoria Island (Table 6.2 and Figures 6.3 and 6.5). Up infection prevalence was 100% on the mainland (both east and west), whereas it was 72.4% on Victoria Island. However, Up infection intensity was considerably lower on east mainland than on west mainland (median (range) = 16.58 lpg (0.20–153.19); n = 28 versus median (range) = 236.46 lpg (0.43–1669.81); n = 67) (Figure 6.6). Up infection intensity was also generally very low on Victoria Island (median (range) = 6.67 (0–809); n = 105), although several animals had high infection intensities (Figure 6.6). Further analyses of subsets of data with back fat thickness and incisor breakage score, respectively, did not detect an effect of these factors, nor their corresponding interactions, on qiviut cortisol when added into the final model (results not shown).

107

Table 6.2. Final model parameter posterior estimates (median and 95% credible intervals (CrI)). The ‘-‘ indicates the reference group for each categorical variable.

Variable Levels Median 95%CrI intercept 4.34 (3.53, 5.21) season late fall-early winter - - mid-late winter -0.03 (-0.26, 0.21) year 2015 - - 2016 -0.30 (-0.56, -0.04) 2017 -0.36 (-0.58, -0.12) 2018 0.06 (-0.19, 0.31) location east mainland - - Victoria Island -0.40 (-0.81, 0.00*) west mainland -0.89 (-1.31, -0.45) marrow_fat -1.10 (-1.90, -0.40) sqrt(Up_lpg) -0.04 (-0.11, 0.03) sex female - - male 0.46 (0.23, 0.72) sqrt(Up_lpg) and location Victoria Island 0.03 (-0.03, 0.10) interaction west mainland 0.06 (-0.00+, 0.14) sex and season interaction male, mid-late winter -0.30 (-0.63, 0.01) *This value is 0.003. +This value is -0.002.

108

Figure 6.3. Marginal posterior distribution of the parameters. Y axes correspond to the density and X axes to the parameters. The peak of each distribution corresponds to the most likely parameter estimate, its spread corresponds to uncertainty about the parameter estimate.

109

Figure 6.4. Relative qiviut cortisol levels of male and female muskoxen in fall-early winter and mid-late winter. The dots show the posterior medians (the most likely estimates) and the lines the 95% credible intervals (the uncertainty about the estimates) for mean qiviut cortisol by sex and season. All other categorical variables were fixed at the reference group and continuous variables at the median. That is, the qiviut cortisol values predicted are for muskoxen from east mainland with qiviut grown in 2015, a metatarsus marrow fat of 0.91, and an Umingmakstrongylus pallikuukensis larval count of 43.08 lpg.

110

Figure 6.5. Effect of Umingmakstrongylus pallikuukensis (Up) larval counts on qiviut cortisol levels in the three broad geographical locations. Solid lines show the posterior medians (the most likely estimates) and dashed lines the 95% credible intervals (the uncertainty about the estimates) for mean qiviut cortisol by location and Up larval counts. All other categorical variables were fixed at the reference group and continuous variables at the median. That is, the qiviut cortisol values predicted are for female muskoxen sampled in late fall-early winter with qiviut grown in 2015 and a metatarsus marrow fat of 0.91. The x-axis was truncated at the 3rd quartile of Up larval counts (maximum count = 1,669 lpg; Table 6.2).

111

Figure 6.6. Histogram of Umingmakstrongylus pallikuukensis (Up) larval counts by location. Red dotted lines indicate the medians and blue dotted lines indicate the means. 6.5. Discussion

Hair GCs have been increasingly used in free-ranging wildlife to investigate potential causes of GC variation, such as anthropogenic disturbances, extreme weather events, and pathogens (e.g., Bryan et al., 2015; Ewacha et al., 2017; Fardi et al., 2018; Madslien et al., 2020). Less frequently, hair GCs have also been used to examine the possible consequences of elevated GCs on fitness indicators (e.g., survival probability, neonate birthweight, and pregnancy success (Rakotoniaina et al., 2017; Downs et al., 2018)), and even more rarely, studies have attempted to demonstrate the link between a stressor, elevated GCs, and the effect of these hormones on fitness (but see Dulude-de Broin et al., 2020). Here, we took a comprehensive approach to evaluate the associations between hair GCs and multiple intrinsic and extrinsic factors. We highlighted some important influential factors (i.e., sex, season, qiviut growth year, and geographical location) and relationships between hair GCs and metrics of health (i.e., body condition and pathogen exposure/infection intensity metrics). Qiviut cortisol levels were higher in males than in females, which is consistent with previous findings in wild muskoxen (Di Francesco et al., 2017). Studies in a variety of wild mammalian species have differing results regarding variations in hair cortisol concentrations (HCCs) between sexes. While multiple studies did not find differences in HCCs between sexes (Macbeth et al., 2010; Malcolm et al., 2013; Terwissen et al., 2013; Yamanashi et al., 2013; Carlitz et al., 2014; Caslini et al., 2016; Potratz et al., 2019), others reported higher HCCs in males (Lafferty et al., 2015; Schell et al., 2017; Azevedo et al., 2019; Santangeli et al., 2019; Madslien et al., 2020) or higher HCCs in females (Bechshøft et al., 2011; Laudenslager et al., 2012; Cattet et

112 al., 2014; Dettmer et al., 2014; Dulude-de Broin et al., 2019). These sex variations most likely reflect a diverging cortisol secretion due to possible differences in HPA axis function (i.e., basal activity and/or response to stressors; Levine, 2002; Rhodes and Rubin, 1999; Turner et al., 2002), and to the experience of distinct stressors (Di Francesco et al., 2017; Chapter 3). As was previously highlighted in several ungulate species in which increased FGM levels were measured during the rut (e.g., bison (Bison bison), Alpine chamois (Rupicapra rupicapra rupicapra), and white-tailed deer (Odocoileus virginianus)) and as indicated by Indigenous knowledge (IK) holders for muskoxen, male muskoxen probably experience high physiological stress during the rut (Mooring et al., 2006; McCoy and Ditchkoff, 2012; Corlatti et al., 2014; Chapter 3). This period is characterized by reduced nutritional intake and repeated and prolonged aggressive interactions with male counterparts, along with courtship behaviors, which are all highly energetically demanding (Chapter 3; Gray, 1987). The lower qiviut cortisol levels of males measured in mid-late winter may be due to the lower stress experienced after the rut (occurring from August to October; see Chapter 3 Figure 3.3a), which may have diluted slightly the qiviut cortisol levels, or to the highly stressed males being less likely to survive after the rut. Sex differences in HCCs could also be an artefact of hair growth patterns and fiber morphological characteristics, which may influence cortisol incorporation into hair (Dulude-de Broin et al., 2019). Male muskoxen have coarser qiviut fibers and a higher variability in fiber diameter than females (Rowell et al., 2001). Additionally, timing of qiviut shedding and onset of qiviut growth differ between sexes (Wilkinson, 1974; Gray, 1987; Robertson, 2000). Female qiviut cortisol levels were lower than those of males and did not differ between the two seasons. Based on the qiviut growth cycle, qiviut cortisol levels could be affected by early and late pregnancy, parturition, and lactation, characteristics that we were unable to assess in this study. While heightened GC levels have been measured in multiple mammalian species during pregnancy and around parturition, these patterns are not as straightforward in Artiodactyls, and studies have shown conflicting results (Burnett et al., 2015; Braun et al., 2017b; Edwards and Boonstra, 2018). Increased cortisol levels linked to parturition, if they occur in muskoxen, may be of too short duration for the signal to be measured in qiviut. Although increased GC levels possibly associated with the longer-term nutritional stress of lactation may be more likely to be detected in the hair, this was not observed for Rocky Mountain goats (Oreamnos americanus) nor red deer (Cervus elaphus) (Huber et al., 2003; Dulude-de Broin et al., 2019). We did not detect any age differences, but were limited by an unbalanced dataset with a low number of juveniles and by the coarse scale age classification (i.e., only two categories, Table 6.1). Similarly, HCCs did not vary significantly among fawns (< 1 year), yearlings (1-2 years), and adults (> 2 years) in white-tailed or red deer (Caslini et al., 2016; Potratz et al., 2019). By contrast, yearlings had lower HCCs than calves and adults in moose (Cervus elaphus) (Madslien et al., 2020), and calves and yearlings had higher HCCs than 2- years-olds and adults in Rocky mountain goats (Dulude-de Broin et al., 2019). Such differences in HCCs among age classes have been attributed to variations in HPA axis activity and/or in hair growth and molting patterns (Azevedo et al., 2019; Dulude-de Broin et al., 2019). A more precise classification, with the separation of calves and yearlings, may have allowed us to detect differences between these age classes.

113

Additionally, identification of older individuals in future studies may allow the detection of increasing HCCs with senescence as has been highlighted in baboons (Papio spp.) (Fourie et al., 2015a). Qiviut cortisol levels increased from west mainland to Victoria Island to east mainland. These differences between geographic locations may be due to variations in biotic and abiotic environmental characteristics. Kugluktuk area (west mainland) is located in the Circumpolar Arctic Bioclimate Subzone E, which is characterized by higher summer warmth index and mean July temperatures, as well as higher maximal NDVI (i.e., Normalized Difference Vegetation Index, an index of vegetation greenness) and phytomass compared to Subzone D, where Kent Peninsula (east mainland) and Victoria Island are situated (Walker et al., 2005). Additionally, Kugluktuk IK holders described a dryer land, poorer vegetation (both in quantity and quality), and a greater relative increase in predators (wolves and grizzly bears) on Victoria Island than around Kugluktuk (Chapter 3), which likely contribute to the observed variations. Our results also indicate differences in qiviut cortisol among geographical locations that are consistent with the local muskox population trends. The population of muskoxen on Victoria island is declining (Leclerc, 2014; Tomaselli et al., 2018b; Mavrot et al., unpubl. data), whereas it is stable to increasing around Kugluktuk (Leclerc, 2014), and its status remains unknown on the Kent Peninsula (Cuyler et al., 2019). Qiviut cortisol levels were lower in growth years 2016 and 2017 than they were in 2015, but increased in 2018 to levels similar to those of 2015. This contrasts with Di Francesco et al. (2017) where no difference was detected between 2015 and 2016 qiviut cortisol levels, however, that study compared the year of sample collection instead of the year of qiviut growth. Multiple biotic and abiotic factors likely contribute to inter-annual differences in qiviut cortisol levels, such as variations in intraspecific competition, predator abundance, weather conditions, exposure to pathogens, food availability and abundance, as well as human disturbance (Di Francesco et al., 2017; Cuyler et al., 2019; Kugluktuk IK holders (see Chapter 3)). Qiviut cortisol was inversely associated with marrow fat percentage, with the muskoxen in poorer body condition having higher qiviut cortisol levels. By contrast, qiviut cortisol was not associated with the other two body condition metrics (hunter assessment and back fat thickness). In ungulates, as an animal's nutritional status declines, fat stores are generally used sequentially, starting with subcutaneous deposits (back fat), followed by visceral deposits, and ultimately by bone marrow fat (Harris, 1945; Bear, 1971). The presence of an association between qiviut cortisol and marrow_fat, but not with back_fat may, therefore, attest to some resilience that muskoxen have up to a critical threshold at which the marrow fat starts declining and we begin observing an association with qiviut cortisol levels. Alternatively, the sample size (n = 145 instead of 211) may have been too small to detect an effect of back_fat in the final model. While the cause and effect relationship remains unclear, multiple studies in wild mammalian species have reported negative associations between body condition metrics and HCCs (e.g., bison, grizzly bears (Ursus arctos), polar bears (Ursus maritimus), and white-tailed deer (Macbeth et al., 2012; Cattet et al., 2014; Mislan et al., 2016; Potratz et al., 2019; Shave et al., 2019)) or fecal GC metabolite (FGM) levels (e.g., Asian elephants (Elephas maximus), giraffes (Giraffa camelopardalis), and European wild rabbits (Oryctolagus cuniculus) (Cabezas et al., 2007; Pokharel et al., 2017; Wolf et al., 2018)). Elevated levels of circulating GCs may cause a reduction

114 in body condition through the mobilization of energy and inhibition of protein synthesis (Sapolsky et al., 2000; Landys et al., 2006; Cabezas et al., 2007). Alternatively, nutritional stress may lead to reduced body condition and increased levels of circulating GCs as animals must mobilize energy stores (e.g., Jeanniard du Dot et al., 2009; Bryan et al., 2013). The combination of high qiviut cortisol and poor body condition in a muskox may, therefore, reflect nutritional stress, but also other long-term stressors (Cattet et al., 2014; Mislan et al., 2016). The relationship between GC secretion and pathogen infections is highly complex and likely bidirectional, influenced by a multitude of factors related to the host and pathogen (e.g., pathogen type, host sex and age) (reviewed in Defolie et al., 2020). The cross-sectional nature of our data did not allow us to establish causal relationships when associations between qiviut cortisol levels and pathogens were identified. We did not detect an association between GI parasite infection intensities or richness and qiviut cortisol levels. This is consistent with findings from the few studies in Artiodactyls which have assessed the relationship between GI parasite burdens in naturally infected animals and HCCs or FGM levels, and found a lack of association (e.g., Cizauskas et al., 2015; Trevisan et al., 2017). Our results are also consistent with those from a longitudinal experiment in captive reindeer (Rangifer tarandus tarandus) in which HCCs did not differ between animals infected with Ostertagia gruehneri and those that had been treated to remove infection (Carlsson et al., 2016a). The infection intensities and prevalence measured in this study for Eimeria spp. (prevalence = 89.14%), nematodirines (prevalence = 48.57%), and Marshallagia marshalli (prevalence = 43.43%) are within the range of those reported in wild muskoxen in the Canadian Arctic, and lower than those recorded in some other ungulate species, for example, in Dall sheep (Ovis dalli) (Kutz et al., 2012). In Dall sheep, the infection intensity of Marshallagia marshalli (measured by the number of adult worms isolated from the GI tract) was inversely associated with body condition and pregnancy status, highlighting its potential negative effects on fitness (Aleuy et al., 2018). However, the impacts of these helminths remain unknown in muskoxen (Kutz et al., 2012). Similarly, clinical disease associated with Eimeria spp. is uncommon in wild ruminants (Kutz et al., 2012) and would generally be associated with high numbers of shed oocysts (> 800,000 oocysts per gram of feces) (Oksanen et al., 1990), and the impact of this protozoan in muskoxen remains unclear (Kutz et al., 2012). Hosts use three different strategies to defend themselves against pathogens: avoidance (i.e., reduction of exposure risk), resistance (i.e., prevention of infection or reduction of pathogen burden once infected), and tolerance (i.e., reduction of the negative impacts of the infection on health and fitness without preventing infection or directly influencing pathogen burden) (Raberg et al., 2009; Medzhitov et al., 2012). The lack of association between qiviut cortisol and GI parasites indicates that muskoxen may use a tolerance strategy to deal with GI parasites, as has been previously suggested by Carlsson et al. (2016a) for reindeer and Cizauskas et al. (2015) for plains zebras (Equus quagga) and springboks (Antidorcas marsupialis). Tolerance mechanisms generally evolve when virulence is low and rate of transmission is high (Restif and Koella, 2004), which seems to be the case for the GI parasites of muskoxen investigated in this study.

115

This absence of association between qiviut cortisol and the GI parasites investigated may also be due to the fairly low infection intensities. Additionally, we were not able to include Teladorsagia boreoarcticus, which is the dominant GI parasite of wild muskoxen, in our study. The reasons were that (i) its egg production occurs mainly throughout the summer (Kutz et al., 2012); (ii) eggs do not survive freezing (De Bruyn, 2010); and (iii) fourth stage larvae may undergo inhibited development in the abomasal mucosa and the burden of hypobiotic larvae is not measurable through fecal analyses (Kutz et al., 2012). The emergence of larvae from the mucosa in May causes significant pathology in muskoxen (Kutz et al., 2004). This may have a measurable impact in qiviut, but even if we had been able to analyze the feces fresh, there would be a temporal mismatch between the abundance of parasites during the qiviut’s growth and that measured through fecal egg counts at the time of death in the fall or winter. We did not observe a relationship between qiviut cortisol levels and Ve larval counts or lungworm richness. By contrast, the association of Up larval counts with qiviut cortisol levels varied across locations (positive on west mainland (Kugluktuk), negative on east mainland (Kent Peninsula), and negligible on Victoria Island (Lady Franklin Point, Ekaluktutiak, and Ulukhaktok)). The difference between the two lungworm species may be due to the lower prevalence and infection intensity of Ve in all three locations, even though it exhibits the same latitudinal gradient in abundance, and/or to differences in the lungworm effects on muskoxen and in their relationship with GCs. The individual level impacts of Up and Ve infection in muskoxen remain poorly understood. Anecdotal evidence suggests that heavy Up infections may lead to impaired breathing, exercise intolerance, and reduced endurance (Gunn and Wobeser, 1993; Hoberg et al., 1995). Umingmakstrongylus pallikuukensis is indeed a large nematode (> 65 cm long), which forms big space occupying cysts in the lung parenchyma (Hoberg et al., 1995; Kutz et al., 1999). In contrast, Ve is shorter (< 25 mm long) and is generally found in low numbers in the terminal bronchioles and alveoli (Verocai et al., 2014). No gross pulmonary lesions, but histological evidence of light parasitic pneumonia, have been observed in muskoxen infected with Ve (see Verocai et al., 2014). Possible additive effects of these two lungworms in the case of co-infections are also unknown, but we would have predicted a positive association between qiviut cortisol and lungworm richness if these existed. To our knowledge, only two studies have investigated the relationship between GC levels and lungworm infections in ungulates and found conflicting results. A lungworm removal experiment in bighorn sheep (Ovis canadensis canadensis) showed no difference in FGM levels between the treated and control herds (Goldstein et al., 2005), whereas another study found a positive correlation between lungworm larval output and FGM levels in male Alpine chamois (Hoby et al., 2006). The geographic variation in the relationship between qiviut cortisol levels and Up larval counts may be related to differences in the abundance of this lungworm and in the timing of its establishment. Umingmakstrongylus pallikuukensis was originally discovered in a muskox near Kugluktuk in 1988 (Gunn et al., 1991), and since 2000 has rapidly expanded its range northward and eastward (Kafle et al., 2020). It invaded and established onto Victoria Island in the early 2000s, where it was detected for the first time in harvested muskoxen on Lady Franklin Point in 2008 and near Ekaluktutiak in 2012 (Kutz et al., 2013a). The positive relationship with qiviut cortisol on west mainland likely reflects a long-term

116 endemic parasite-host relationship with high prevalence and infection intensity of Up. In contrast, absence of a relationship on Victoria Island may reflect that the parasite cost is minimal at low intensities. The apparent inverse association on Kent Peninsula remains enigmatic and, given the very low intensities of infection, is perhaps a spurious relationship. Overall, our results suggest that qiviut cortisol may not be a good indicator of parasitism. Although exposure to most parasites would be predominantly in the summer, during the period of qiviut growth, and thus one might expect a positive correlation between worm burden and qiviut cortisol, we were only able to quantify larval/oocyst/egg counts. Many parasites, including protostrongylids and multiple strongyle species, have a highly seasonal pattern of shedding, thus late fall and winter egg counts may not reflect adult parasite burden at that time, or during the previous summer of hair growth (e.g., Halvorsen et al., 1985; Irvine et al., 2001). One exception to this is Marshallagia marshalli for which egg counts correlated with adult worm burdens in Svalbard reindeer (Irvine et al., 2001). However, as this parasite is transmitted throughout the year, mid-late winter burdens (as determined by fecal egg counts) may reflect recent acquisition during a time when the qiviut is no longer growing, and thus, is unable to reflect recent stressors. As for Eimeria spp., seasonal variations in oocyst counts have also been recorded, with high oocyst shedding generally occurring in response to recent stress (Pyziel et al., 2011; Chartier and Paraud, 2012). Again, this would not be detectable in the qiviut in the absence of growth. The larval/egg/oocyst output measured in feces, particularly in mid-late winter may, therefore, be better associated with a shorter-term measure of GC levels such as FGMs. We did not detect an association between qiviut cortisol levels and Erysipelothrix rhusiopathiae PP or Brucella serology status. Seropositivity indicates exposure to these two bacteria, but does not provide information regarding the timing of exposure, occurrence of clinical disease, or the subsequent possible recovery of the animal. Erysipelothrix rhusiopathiae causes a variety of clinical manifestations (reviewed in Opriessnig et al., 2020). It is associated with widespread mortality in muskoxen (Kutz et al., 2015; Forde et al., 2016), and while there is evidence of very few carrier animals, no chronic forms have been described in this species (see Forde et al., 2016). These findings suggest that Erysipelothrix rhusiopathiae is generally associated with death or recovery in muskoxen. Assuming that positive cases of Erysipelothrix rhusiopathiae among our harvested animals are recovered, we may not, therefore, have expected an association between exposure to this bacterium and qiviut cortisol levels. In contrast, Brucella suis biovar 4, reported increasingly from muskoxen (Tomaselli et al., 2019), is associated with chronic manifestations and lesions such as abscesses, granulomatous mastitis, endometritis, lymphadenitis, and nephritis (Tomaselli et al., 2018b, 2019). We would have predicted higher qiviut cortisol levels in Brucella-positive muskoxen as chronic illnesses have previously been associated with increased HCCs in domestic cattle (Bos Taurus) (Braun et al., 2017a). The absence of association, however, may be due to the animals being exposed to the bacterium after the qiviut had stopped growing and/or to the unbalanced dataset and low prevalence. We did not detect an association between qiviut cortisol and incisor breakage score. Severe wearing or breakage of incisors may affect browsing and food processing efficiency, and decrease the ability to

117 acquire sufficient forage (Kojola et al., 1998; Ericsson and Wallin, 2001), but impacts on muskox health remain unclear. We, consequently, might have expected a positive correlation between qiviut cortisol and incisor breakage score, but most of the effect of incisor breakage on qiviut cortisol would have manifested through its impacts on body condition. Additionally, we were restricted in our assessment by: (i) the smaller sample size used to assess the effect of incisor_breakage in the final model (n = 152 instead of 211) and (ii) the fact that the incisor breakage score takes into account only the number of incisors broken and not the severity of the breakage.

6.6. Conclusion

In this study, we confirmed and further identified important factors influencing qiviut cortisol levels in muskoxen (i.e., sex, season, qiviut growth year, and geographical location) and highlighted relationships between qiviut cortisol and metrics of health (i.e., body condition and pathogen exposure/infection intensity metrics) in this species. Unfortunately, the cross-sectional nature of the data did not allow us to demonstrate temporal relationships from the associations that we detected (e.g., between metatarsus marrow fat or Up infection intensity and qiviut cortisol levels). Future research in muskoxen should focus on more specifically elucidating the relationship between a particular stressor, the increased cortisol levels it causes, and the subsequent effect this has on health and fitness. This could be done, for example, by monitoring marked individuals over several years, as has been done through a long-term study of mountain goats, which recently demonstrated a link between increased predation risk and breeding suppression through chronically elevated cortisol levels (Dulude-de Broin et al., 2020). Among the parasites and bacteria investigated, we only detected an association between qiviut cortisol levels and Up infection intensity, and their relationship varied depending on the geographical location. However, it is likely that the various pathogens have a cumulative effect on muskoxen. Future work should aim at developing a comprehensive summary health score based on the multiple body condition and pathogen exposure/infection intensity metrics, which may allow to identify a threshold at which qiviut cortisol levels start to be impacted. Finally, while we mostly focused on evaluating associations between qiviut cortisol levels and individual measures of health, our results also suggest differences in qiviut cortisol among geographical locations that are consistent with the local muskox population trends. Gathering long-term data from additional muskox populations for which population trends, and preferably annual productivity, are known would provide further insights as to whether these associations between population trajectory and qiviut cortisol levels hold true. If increased GC levels are shown to precede population declines, poor productivity, or mortality events, these could be used as predictive biomarkers of population health and fitness, and inform management and conservation actions pro-actively.

118

6.7. Acknowledgements

We are highly grateful to all the hunters and biologists who participated in sample collection, as well as the Ekaluktutiak, Kugluktuk, and Ulukhaktok Hunters and Trappers Organizations or Committees for their support and guidance. We also wish to thank all of the students and staff who helped coordinate the sampling program in the communities or contributed to sample processing and analyses, in particular Russell Akeeagok, Allen Niptanatiak, Terry Milton, Bessie Inuktalik, Denise Okheena, Angie Schneider, Christine Gilman, Patricia Medd, Mélanie Meyer, Sanchit Chopra, Tessa Bailey, Akaysha Envik, Marian Trudeau, Leslie Bottari, Kamala Sapkota, and James Wang. Finally, we thank Canada North Outfitting and the Governments of NWT and Nunavut for their support.

6.8. Funding

Juliette Di Francesco was supported by the NSERC-Collaborative Research and Training Experience (CREATE) Host-Parasite Interactions Training Program student scholarship, the NSERC- CREATE Integrated Training Program in Infectious Diseases, Food Safety and Public Policy Student Scholarship, the Faculty of Graduate Studies Doctoral Scholarship (University of Calgary), and by the Morris Animal Foundation Fellowship Training Grant D18ZO-407. This work was supported by Canada North Outfitting, Polar Knowledge Canada Grant NST-1718-0015, the Natural Sciences and Engineering Research Council Discovery Grant RGPIN/04171-2014, the Natural Sciences and Engineering Research Council Northern Supplement RGPNS/316244-2014, and ArcticNet.

119

CHAPTER 7. CONCLUSIONS: SUMMARY AND FUTURE DIRECTIONS

7.1. Main findings

Through my thesis research, I established fecal glucocorticoid metabolite (FGM) and qiviut cortisol levels as reliable biomarkers of physiological stress in muskoxen and then applied these tools in combination with documenting Indigenous knowledge (IK) to explore potential causes and patterns of physiological stress in wild muskoxen. This work addressed some of the major questions and challenges in the field of wildlife endocrinology, particularly regarding hair glucocorticoid (HGC) analyses. First, I validated two methods to measure qiviut cortisol levels using liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS; Chapter 2) and a cortisol enzyme immunoassay (EIA; Chapter 5), respectively, as well as a corticosterone EIA to measure FGM levels in muskoxen (Chapter 4). EIAs remain the most commonly used to measure HGC levels in wildlife (reviewed in Koren et al., 2019), but LC-MS/MS methods are increasingly applied (e.g., Carlitz et al., 2016; Weisser et al., 2016; Rakotoniaina et al., 2017; Jewgenow et al., 2020). Through two repeated pharmacological challenges in captive muskoxen, I also successfully demonstrated that qiviut cortisol and FGM levels accurately reflect long-term (i.e., over the period of the hair’s growth) and short-term changes in hypothalamic-pituitary-adrenal (HPA) axis activity, respectively (Chapters 4 and 5). Pharmacological challenges have been carried out in a wide variety of wild mammalian species to validate the use of FGMs as biomarkers of HPA axis activity (reviewed in Palme, 2019). However, these remain scarce for HGCs and all except one were done during hair growth (Ashley et al., 2011; Terwissen et al., 2013; Mastromonaco et al., 2014; Crill et al., 2019; Dulude-de Broin et al., 2019). Additionally, the study which was done in the absence of hair growth on captive caribou (Rangifer tarandus granti) and reindeer (Rangifer tarandus tarandus) measured the response in hair to a single injection of adrenocorticotropic hormone (ACTH) (Ashley et al., 2011). To my knowledge, I was, consequently, the first to measure the response to the repeated administration of ACTH in hair of a wildlife species, and I demonstrated that repeated ACTH injections are not reflected in qiviut in the absence of growth. With my experimental work on muskoxen, I also highlighted some important limitations of qiviut cortisol as a biomarker of HPA axis activity (i.e., variations across body regions, significant differences in qiviut segments over time, and differences between shed and unshed qiviut). These results underscore the importance of following strict experimental design and methodology for sample collection to reduce variation and avoid confounding variables in data analyses. Finally, I detected important seasonal variations in the timing and duration of the FGM response to ACTH administration. The few studies that have compared such responses among seasons have had differing results depending on the species (see Millspaugh et al., 2002; Morrow et al., 2002; Özkan Gülzari et al., 2019), and my findings (i.e., earlier and shorter peak in summer compared to winter) are coherent with the physiological adaptations of muskoxen to their highly seasonal and extreme Arctic environment.

120

Second, I documented IK about the factors that may be influencing muskoxen in their rapidly changing environment (Chapter 3). The benefits of applying IK in the monitoring and management of natural resources are increasingly recognized (e.g., Bell and Harwood, 2012; Carlsson et al., 2016b) and IK has been used as an invaluable source of information on wildlife health and ecology in multiple studies (e.g., Tomaselli et al., 2018b; Laforest et al., 2020). Rarely, however, if ever, has IK been taken into account in endocrinology studies to identify and assess the impacts of stressors on wildlife. The IK that I documented through small-group interviews with muskox harvesters in Kugluktuk provided invaluable insights regarding the potential stressors of muskoxen and their specific importance. For example, participants shared knowledge and observations regarding the impacts of high heat on muskoxen, which are poorly understood by the scientific community, revealing that muskoxen are mainly sensitive to prolonged periods of high heat, but are otherwise able to behaviorally mitigate the effects of this stressor. Third, I assessed associations between qiviut cortisol and the multiple health indicators/metrics recorded or measured through the sampling kits collected as part of a co-designed hunter-based muskox health monitoring program (Tomaselli, 2019). Like Peacock et al. (2020), I define indicators as broad “determinants of health” (e.g., body condition) and metrics as “specific measures of an indicator that can be obtained from qualitative or quantitative data” (e.g., back-fat thickness and relative body condition) (Peacock et al., 2020:3). While several studies have determined associations between HGC levels and metrics of body condition (e.g., Mislan et al., 2016; Potratz et al., 2019; Shave et al., 2019), the relationship between HGC levels and metrics of parasitism has seldom been assessed (but see Carlsson et al., 2016a; Trevisan et al., 2017), and only one study has investigated associations of HGCs with both body condition and parasitism metrics (Madslien et al., 2020). My work is, therefore, novel in the field of HGC research, as it simultaneously tested the effect on HGC levels of multiple intrinsic (i.e., sex, age, body condition, and incisor breakage) and extrinsic biotic factors (i.e., lungworm and gastro-intestinal parasite infection intensities, parasite richness, exposure to bacteria), as well as broad non-specific landscape and temporal features (i.e., geographical location, season, and year). Major findings from the statistical analyses included: (i) higher qiviut cortisol levels in males than in females; (ii) strong inter-annual variations; (iii) lower qiviut cortisol levels in the summer than in the fall and winter; (iv) increasing qiviut cortisol levels from west mainland (Kugluktuk area) to Victoria Island to east mainland (Kent Peninsula); (v) a negative association between qiviut cortisol and marrow fat percentage, but not with other metrics of body condition; and (iv) a relationship between qiviut cortisol and the infection intensity of the lungworm Umingmakstrongylus pallikuukensis which varied depending on the geographical location, and otherwise an absence of association between qiviut cortisol and pathogen exposure/infection intensity metrics. Fourth, I interpreted findings from the statistical analyses using the available scientific literature on wildlife endocrinology and muskox health and ecology (Chapters 2 and 6), but also collaboratively interpreted part of the results (i.e., sex, seasonal, and inter-annual variations) with IK holders, which allowed for a broader, more holistic comprehension of the observed patterns (Chapter 3). This collaborative work, along with the training of a local Inuk Kugluktuk resident in initial sample analyses during my fieldwork,

121 represented a meaningful advancement in the process of transitioning the hunter-based muskox health monitoring program to Danielsen et al. (2009)’s “collaborative monitoring with local data interpretation” category by actively involving communities at all steps of the research and not only in data collection. Fifth, results from this study served to establish baseline longitudinal data on the qiviut cortisol levels of wild muskoxen in different geographical locations (Chapters 2 and 6). Such knowledge is crucial to monitor the changes in qiviut cortisol levels that may occur, and to subsequently identify the factors they are associated with. Finally, although the various pieces occurred in a different chronological order in this thesis, this has enabled me to highlight some of the key steps in endocrinology studies and where IK can be incorporated. These steps are summarized in the framework proposed in Figure 7.1. Peacock et al. (2020) proposed a conceptual framework for bringing IK and SK together to inform wildlife status assessments, but no such framework had previously been developed for wildlife endocrinology studies. At an international level, the importance of IK in biodiversity conservation and sustainable use of natural resources is increasingly recognized and is highlighted in several international conventions and declarations (e.g., United Nations Declaration on the Rights of Indigenous Peoples (2007) and the Convention on Biological Diversity (1992)). In Canada, including IK in wildlife management is a federal legal requirement (e.g., Species at Risk Act (2002)) and is mandated in land claims agreements (e.g., Indian and Northern Affairs, 1984, 1993). Findings from my work were regularly shared with the various participating communities and partners from government through different meetings, such as the Annual General Meeting of the Kugluktuk Angoniatit Association. They are, consequently, readily available if needed for management and conservation actions.

Figure 7.1. Framework for the inclusion of Indigenous knowledge (IK) in wildlife endocrinology studies. Steps in dark red are those that can be informed by both IK and scientific knowledge (SK), whereas those in blue are mainly informed by SK. The interpretation of results and identification of stressors and influential factors may be an iterative process (thin arrow).

122

7.2. Study limitations, remaining knowledge gaps, and next steps

As with all studies, this research has limitations. First, even though the pharmacological challenge undertaken in captive muskoxen enabled me to validate the use of qiviut cortisol and FGM levels as biomarkers of HPA axis activity, some important knowledge gaps remain. For instance, I was unable to assess the extent to which the local production of cortisol in the skin and hair follicles was activated by the ACTH injections and if it contributed to the response measured in qiviut. Only the administration of radiolabeled cortisol would have provided the necessary information, and this approach is not feasible in non-sequestered (i.e. paddock ranging) large animals. Shed qiviut has been put forth as a tool for non- invasive monitoring of the physiological status of muskoxen. Unfortunately, my data showed that the cortisol levels in shed qiviut differed and were not correlated with those of fully grown qiviut shaved three months earlier. Before shed qiviut cortisol can be used as a separate monitoring tool, further studies are required to determine what the cortisol levels measured in shed qiviut actually reflect, as well as the impact of natural environmental conditions on shed qiviut cortisol concentrations (although this impact is likely minor; Macbeth et al., 2010). Developing methods to entirely remove the dandruff (i.e., pieces of dead skin possibly containing cortisol) and follicles, which may both be affected by more recent HPA axis activity, from the shed qiviut samples would be a good first step towards clarifying the time-frame of HPA axis activity reflected in shed qiviut. Finally, while qiviut cortisol represents an integrated measure of HPA axis activity over the period of its growth, the growth of this wooly undercoat is seasonal (i.e., between early April and late November; Flood et al., 1989). Qiviut cortisol, consequently, does not provide information regarding the HPA axis activity between December and March. This period, which corresponds to the main winter months of limited resource availability and to mid and late pregnancy in females, is of biological relevance. By contrast, guard hairs, which are much thicker than qiviut fibers and grow continuously (Rowell et al., 2001), would capture HPA axis activity during the entire year. Results from the experimental study showed that the cortisol concentration in a specific segment of qiviut changes over time and discourage the use of qiviut fibers as a retrospective calendar of HPA axis activity in muskoxen. However, it is possible that segmental analyses of guard hairs, which have successfully been used to measure nitrogen stable isotope ratios and reconstruct the dietary records of muskoxen (Mosbacher et al., 2016), could serve as a retrospective calendar of HPA axis activity as was shown in cattle (Bos taurus; Heimbürge et al., 2020b), orangutans (Pongo spp.; Carlitz et al., 2014), and humans (Kirschbaum et al., 2009). Samples of guard hair collected during the two pharmacological challenges have been archived and are readily available to carry out the same physiological and EIA validations as with qiviut cortisol. Second, in the course of this thesis I was only able to investigate associations between qiviut cortisol and the various intrinsic and extrinsic factors because the Bayesian approach was complex and time consuming computationally. Data are available to do these same analyses with FGM levels as the outcome. Unlike qiviut cortisol, FGM levels are biomarkers of shorter-term stress and, consequently, likely provide complementary and additional information. For instance, while I did not detect any association between

123 qiviut cortisol levels and gastrointestinal parasite infection intensities, the egg/oocyst output measured in feces may be better associated with a shorter-term measure of GC levels such as FGMs (see Chapter 6). Third, the idea of incorporating IK in the design and interpretation of this work was conceived of and matured through the course of my thesis. Ideally, I would have regularly incorporated IK throughout the thesis to inform the identification of potential stressors and influential factors, the development of specific indicators and associated metrics to measure them, and the interpretation of results (steps 2 and 5 in Figure 7.1), especially since these steps feedback to inform each other. Further work bringing together IK and SK is also needed to develop indicators/metrics for the various potential stressors of muskoxen identified through the interviews, so that these can be added in the statistical analyses to determine their associations with qiviut cortisol and FGM levels. Such indicators/metrics would include quantitative and/or qualitative measures of plant abundance and quality, indexes of insect harassment (some have already been developed for reindeer (Weladji et al., 2003) and caribou (Witter et al., 2012)), measures of air quality, etc. Additionally, pregnancy status was not incorporated in this study. Fecal reproductive hormone analyses (Desaulniers et al., 1989) and validating pregnancy tests based on the measurement of pregnancy- associated glycoproteins using filter paper blood, which is more easily collected through hunter-based sampling, would enable the inclusion of reproductive status among the intrinsic factors in future statistical analyses. These pregnancy tests have been validated for muskox serum (Greunz et al., 2018) and may allow to diagnose pregnancy at an early stage (i.e., as early as 28 days after conception in cattle (Green et al., 2005)). Finally, further work bringing together IK and SK to identify some of the potential cumulative effects among stressors would greatly inform our understanding of muskox vulnerability and of the conditions that may contribute to mortality events. The IK I documented during the interviews already highlights some of these cumulative effects, such as that of lungworm infections and air pollution. While periods of high heat during the summer have previously been associated with fatal outbreaks of pneumonia in muskoxen in Dovrefjell (Norway), and summer outbreaks of Erysipelothrix rhusopathiae in muskoxen in the western Canadian Arctic Archipelago have occurred, these mortality events were likely caused by multiple factors, some of which remain to be elucidated (Ytrehus et al., 2008, 2015; Handeland et al., 2014; Kutz et al., 2015). The negative behavioral and physiological impacts of long periods of high heat on muskoxen suggested by participants during the interviews could also be explored through targeted SK studies, as has previously been done for brucellosis in muskoxen (Tomaselli et al., 2018b, 2019).

7.3. Future research directions

In this thesis, I documented IK to identify some of the potential causes of physiological stress in muskoxen. I also investigated associations between GC levels and various intrinsic and extrinsic factors in muskoxen and highlighted some important sources of variability (i.e., sex, season, geographical location, etc.). However, the cross-sectional nature of the available data did not allow me to establish causal relationships (e.g., between GC secretion and Up infection). Future research in muskoxen, and more generally in the field of wildlife endocrinology, should focus on simultaneously determining the causes and

124 consequences of physiological stress. The relationships between a stressor, the increased GC levels it causes, and the subsequent effect this has on fitness (i.e., how GCs mediate the link between stressors/environmental changes and fitness effects) are difficult to disentangle for two main reasons: (i) many connections between stressors and GC levels are likely bidirectional (e.g., parasitic infections), meaning that some factors may be both a cause and a consequence of elevated GC levels; and (ii) several factors may directly and/or indirectly, through increased GC levels, impact fitness (Dantzer et al., 2016; Beehner and Bergman, 2017). In Figure 7.2, I have attempted to illustrate the complexity of these relationships, by summarizing the potential intrinsic and extrinsic (both abiotic and biotic) factors that influence GC secretion and/or may be affected by GCs.

Figure 7.2. Intrinsic and extrinsic factors that may influence glucocorticoid (GC) secretion and/or may be affected by GCs.

Future studies investigating these complex relationships will require long-term longitudinal data and large-scale experimental manipulations of individuals (Dantzer et al., 2016). Such studies are scarce in wild mammals but have, for example, previously been done in North American red squirrels (Sciurus vulgaris) in the Yukon, Canada, where the species experiences major inter-annual fluctuations in resource availability (reviewed in Dantzer et al., 2016). Briefly, two experiments increasing food abundance and perceived population density, respectively, showed that GC levels in female red squirrels were more influenced by population density than food abundance. These also suggested that increased maternal GCs during high population density years may actually enhance female reproductive success by increasing the postnatal growth rates of their offspring, which then are more likely to survive in a context of high competition (Dantzer et al., 2012, 2013, 2016). Undertaking such studies would be much more challenging in a large herbivore with a long life-span like the muskox. However, it is likely that following marked individuals over several years, and monitoring regularly GC levels, indicators of fitness (e.g., reproductive success, individual survival), and the various causes of GC secretion previously identified, would provide invaluable

125 information to begin elucidating the key question of how GCs mediate the relationship between environmental changes and fitness in this species. A long-term study of individually marked mountain goats (Oreamnos americanus), for example, recently demonstrated a link between increased predation risk and breeding suppression through chronically elevated GC levels (Dulude-de Broin et al., 2020). In this thesis, I investigated the factors associated with the GC levels of individual muskoxen, but evaluating the usefulness of GC levels as prospective indicators of individual fitness is also critical. Studies in a variety of vertebrate species have assessed GC levels as predictors of individual survival and have found inconsistent results. While some have shown that high GC levels were associated with a reduced survival probability (e.g., ring-tailed lemurs (Lemur catta; Pride, 2005), grey mouse lemurs (Microcebus murinus; Rakotoniaina et al., 2017), yellow-bellied marmots (Marmota flaviventris; Wey et al., 2015)), others have highlighted the opposite association (e.g., lizards (Lacerta vivipara; Cote et al., 2006), European wild rabbits (Oryctolagus cuniculus; Cabezas et al., 2007), zebra finches (Taeniopygia guttata; Marasco et al., 2015)), or even a curvilinear relationship (cliff swallows (Petrochelidon pyrrhonota; Brown et al., 2005)). Such variations may be due to differences in the nature of the studies (i.e., observational versus experimental), the time-frame of survival assessment (i.e., months or years), and/or in the stressor causing elevated GC levels. In any case, the variety of results highlights once again, the complexity and probable context-dependency of the relationship between GCs and fitness (Dantzer et al., 2016; Beehner and Bergman, 2017). Data from harvested muskoxen have been collected since 2013 through the hunter-based muskox health surveillance program from various geographical locations across the Northwest Territories and Nunavut, particularly the areas hunted by the communities of Ulukhaktok, Ekaluktutiak, and Kugluktuk. The qiviut cortisol data gathered between 2013 and 2019 allowed me to establish baseline knowledge and to identify years of “low” versus “high” GC levels. These data also suggest a trend for higher GC levels in the declining muskox populations of Victoria and Banks Islands than in stable or increasing populations, such as the population around Kugluktuk. Data from additional muskox populations for which population trends are known would allow for a comparative approach to understand the relationship between GC levels and population trends in the different areas, as has been done for Erysipelothrix rhusiopathiae seroprevalence (Mavrot et al., 2020). If increased GC levels are shown to precede population declines or mortality events, these could be used as predictive biomarkers of population health and fitness, and inform management and conservation actions pro-actively.

7.4. Final note

In this thesis I documented IK and used biomarkers of physiological stress to inform muskox conservation through identifying: (i) potential stressors of muskoxen; (ii) sources of variability in the GC levels of muskoxen and their potential underlying causes; and (iii) extrinsic and intrinsic factors associated with qiviut cortisol levels. These are key steps to simultaneously investigating the causes and consequences, both at the individual and population levels, of physiological stress in this species. More broadly, even though many questions remained unanswered in the novel field of wildlife HGC research, this work has

126 advanced our understanding of GC deposition and stability in hair, and of the limitations and challenges associated with HGC interpretation. Finally, this thesis has highlighted the multiple benefits of incorporating IK in wildlife endocrinology studies, and has provided a framework for doing so.

127

References

Aasen E, Medrano JF (1990) Amplification of the ZFY and ZFX genes for sex identification in humans, cattle, sheep and goats. Biotechnology 8: 1279–1281.

Acevedo-Whitehouse K, Duffus ALJ (2009) Effects of environmental change on wildlife health. Philos Trans R Soc B Biol Sci 364: 3429–3438.

Acker M, Mastromonaco G, Schulte-Hostedde AI (2018) The effects of body region, season and external arsenic application on hair cortisol concentration. Conserv Physiol 6: coy037.

Adamczewski J, Flood PF (1997) Seasonal patterns in body composition and reproduction of female muskoxen (Ovibos moschatus). J Zool 241: 245–269.

Adamczewski J, Gunn A, Laarveld B, Flood PF (1992) Seasonal changes in weight, condition and nutrition of free-ranging and captive muskox females. Rangifer 12: 179–183.

Adamczewski JZ, Flood PF, Chaplin RK, Schaefer JA (1994) Seasonal variation in intake and digestion of a high-roughage diet by muskoxen. Can J Anim Sci 74: 305–313.

Afema JA, Beckmen KB, Arthur SM, Huntington KB, Mazet JAK (2017) Disease complexity in a declining Alaskan muskox (Ovibos moschatus) population. J Wildl Dis 53: 311–329.

Agnew RCN, Smith VJ, Fowkes RC (2016) Wind turbines cause chronic stress in badgers (Meles meles) in Great Britain. J Wildl Dis 52: 459–467.

Aleuy OA, Ruckstuhl K, Hoberg EP, Veitch A, Simmons N, Kutz SJ (2018) Diversity of gastrointestinal helminths in Dall’s sheep and the negative association of the abomasal nematode, Marshallagia marshalli, with fitness indicators. PLOS ONE 13: e0192825.

Altizer S, Ostfeld RS, Johnson PTJ, Kutz S, Harvell CD (2013) Climate change and infectious diseases: From evidence to a predictive framework. Science 341: 514–519.

AMAP (2017) Snow, water, ice and permafrost in the Arctic (SWIPA) 2017.

Armitage P, Kilburn S (2015) Conduct of Traditional Knowledge Research: A Reference Guide. Wildlife Management Advisory Council North Slope, Whitehorse, Yukon, Canada, p 97.

Ashley NT, Barboza PS, Macbeth BJ, Janz DM, Cattet MRL, Booth RK, Wasser SK (2011) Glucocorticosteroid concentrations in feces and hair of captive caribou and reindeer following adrenocorticotropic hormone challenge. Gen Comp Endocrinol 172: 382–391.

Atwood MP, Kie JG, Millspaugh JJ, Matocq MD, Bowyer RT (2020) Condition of mule deer during winter: stress and spatial overlap with North American elk. Mammal Res 65: 349–358.

Azevedo A, Bailey L, Bandeira V, Dehnhard M, Fonseca C, de Sousa L, Jewgenow K (2019) Age, sex and storage time influence hair cortisol levels in a wild mammal population. PLOS ONE 14: e0221124.

Baker MR, Gobush KS, Vynne CH (2013) Review of factors influencing stress hormones in fish and wildlife. J Nat Conserv 21: 309–318.

Baker S, Constant NL (2020) Epistemic justice and the integration of local ecological knowledge for marine conservation: Lessons from the Seychelles. Mar Policy 117: 103921.

Ban NC, Frid A, Reid M, Edgar B, Shaw D, Siwallace P (2018) Incorporate Indigenous perspectives for impactful research and effective management. Nat Ecol Evol 2: 1680–1683.

128

Barboza PS, Peltier TC, Forster RJ (2006) Ruminal fermentation and fill change with season in an Arctic grazer: Responses to hyperphagia and hypophagia in muskoxen (Ovibos moschatus). Physiol Biochem Zool 79: 497–513.

Bartoń K (2015) MuMIn Package.

Bashaw MJ, Sicks F, Palme R, Schwarzenberger F, Tordiffe ASW, Ganswindt A (2016) Non-invasive assessment of adrenocortical activity as a measure of stress in giraffe (Giraffa camelopardalis). BMC Vet Res 12. doi:10.1186/s12917-016-0864-8

Bates DM (2010) Lme4: Mixed-Effects Modeling with R. Springer, New York, USA.

Baxter-Gilbert JH, Riley JL, Mastromonaco GF, Litzgus JD, Lesbarreres D (2014) A novel technique to measure chronic levels of corticosterone in turtles living around a major roadway. Conserv Physiol 2: cou036.

Bear GD (1971) Seasonal trends in fat levels of pronghorns, Antilocapra americana, in Colorado. J Mammal 52: 583–589.

Bechshøft Tø., Rigét FF, Sonne C, Letcher RJ, Muir DCG, Novak MA, Henchey E, Meyer JS, Eulaers I, Jaspers VLB, et al. (2012) Measuring environmental stress in East Greenland polar bears, 1892– 1927 and 1988–2009: What does hair cortisol tell us? Environ Int 45: 15–21.

Bechshøft Tø., Sonne C, Dietz R, Born EW, Novak MA, Henchey E, Meyer JS (2011) Cortisol levels in hair of East Greenland polar bears. Sci Total Environ 409: 831–834.

Beehner JC, Bergman TJ (2017) The next step for stress research in primates: To identify relationships between glucocorticoid secretion and fitness. Horm Behav 91: 68–83.

Beehner JC, Nguyen N, Wango EO, Alberts SC, Altmann J (2006) The endocrinology of pregnancy and fetal loss in wild baboons. Horm Behav 49: 688–699.

Behie AM, Pavelka MSM, Chapman CA (2010) Sources of variation in fecal cortisol levels in howler monkeys in belize. Am J Primatol 72: 600–606.

Bélisle AC, Asselin H, LeBlanc P, Gauthier S (2018) Local knowledge in ecological modeling. Ecol Soc 23: 14.

Bell RK, Harwood LA (2012) Harvest-based monitoring in the Inuvialuit Settlement Region: Steps for success. ARCTIC 65: 421–432.

Berkes F (2008) Sacred Ecology, 4th Edition. Routledge, Taylor & Francis Group, New York, USA.

Bird S, Prewer E, Kutz S, Leclerc L-M, Vilaça ST, Kyle CJ (2019) Geography, seasonality, and host‐ associated population structure influence the fecal microbiome of a genetically depauparate Arctic mammal. Ecol Evol 9: 13202–13217.

Bonier F, Martin PR, Moore IT, Wingfield JC (2009a) Do baseline glucocorticoids predict fitness? Trends Ecol Evol 24: 634–642.

Bonier F, Moore IT, Martin PR, Robertson RJ (2009b) The relationship between fitness and baseline glucocorticoids in a passerine bird. Gen Comp Endocrinol 163: 208–213.

Bourbonnais ML, Nelson TA, Cattet MRL, Darimont CT, Stenhouse GB (2013) Spatial analysis of factors influencing long-term stress in the grizzly bear (Ursus arctos) population of Alberta, Canada. PLOS ONE 8: e83768.

129

Braun U, Clavadetscher G, Baumgartner M, Riond B, Binz T (2017a) Hair cortisol concentration and adrenal gland weight in healthy and ill cows. Schweiz Arch Tierheilkd 159: 493–495.

Braun U, Michel N, Baumgartner MR, Hässig M, Binz TM (2017b) Cortisol concentration of regrown hair and hair from a previously unshorn area in dairy cows. Res Vet Sci 114: 412–415.

Brook RK, Kutz SJ, Veitch AM, Popko RA, Elkin BT, Guthrie G (2009) Fostering community-based wildlife health monitoring and research in the Canadian North. EcoHealth 6: 266–278.

Brown CR, Brown MB, Raouf SA, Smith LC, Wingfield JC (2005) Effects of endogenous steroid hormone levels on annual survival in cliff swallows. Ecology 86: 1034–1046.

Brown JL, Bellem AC, Fouraker M, Wildt DE, Roth TL (2001) Comparative analysis of gonadal and adrenal activity in the black and white rhinoceros in North America by noninvasive endocrine monitoring. Zoo Biol 20: 463–486.

Brunton EA, Clemente CJ, Burnett SE (2020) Not all urban landscapes are the same: Interactions between urban land use and stress in a large herbivorous mammal. Ecol Appl 30: e02055.

Bryan HM, Darimont CT, Paquet PC, Wynne-Edwards KE, Smits JEG (2013) Stress and reproductive hormones in grizzly bears reflect nutritional benefits and social consequences of a salmon foraging niche. PLOS ONE 8: e80537.

Bryan HM, Smits JEG, Koren L, Paquet PC, Wynne-Edwards KE, Musiani M (2015) Heavily hunted wolves have higher stress and reproductive steroids than wolves with lower hunting pressure. Funct Ecol 29: 347–356.

Burek KA, Gulland F, O’Hara TM (2008) Effects of climate change on Arctic marine mammal health. Ecol Appl 18: S126–S134.

Burnard C, Ralph C, Hynd P, Hocking Edwards J, Tilbrook A (2017) Hair cortisol and its potential value as a physiological measure of stress response in human and non-human animals. Anim Prod Sci 57: 401–414.

Burnett TA, Madureira AML, Silper BF, Tahmasbi A, Nadalin A, Veira DM, Cerri RLA (2015) Relationship of concentrations of cortisol in hair with health, biomarkers in blood, and reproductive status in dairy cows. J Dairy Sci 98: 4414–4426.

Burnham KP, Anderson DR (2002) Model Selection and Multimodel Inference: A Practical Information- Theoretic Approach, 2nd Edition. Springer, New York, USA.

Busch DS, Hayward LS (2009) Stress in a conservation context: A discussion of glucocorticoid actions and how levels change with conservation-relevant variables. Biol Conserv 142: 2844–2853.

Cabezas S, Blas J, Marchant TA, Moreno S (2007) Physiological stress levels predict survival probabilities in wild rabbits. Horm Behav 51: 313–320.

Cantarelli VI, Perez-Rueda MA, Kowalewski MM, Mastromonaco GF, Ponzio MF (2017) Validation of an enzyme immunoassay and comparison of fecal cortisol metabolite levels in black and gold howler monkeys (Alouatta caraya) inhabiting fragmented and continuous areas of the humid Chaco region, Argentina. Am J Primatol 79: e22625.

Cardillo M, Mace GM, Jones KE, Bielby J, Bininda-Emonds ORP, Sechrest W, Orme CDL, Purvis A (2005) Multiple causes of high extinction risk in large mammal species. Science 309: 1239–1241.

130

Carlitz EHD, Kirschbaum C, Miller R, Rukundo J, van Schaik CP (2015) Effects of body region and time on hair cortisol concentrations in chimpanzees (Pan troglodytes). Gen Comp Endocrinol 223: 9–15.

Carlitz EHD, Kirschbaum C, Stalder T, van Schaik CP (2014) Hair as a long-term retrospective cortisol calendar in orang-utans (Pongo spp.): New perspectives for stress monitoring in captive management and conservation. Gen Comp Endocrinol 195: 151–156.

Carlitz EHD, Miller R, Kirschbaum C, Gao W, Hänni DC, van Schaik CP (2016) Measuring hair cortisol concentrations to assess the effect of anthropogenic impacts on wild chimpanzees (Pan troglodytes). PLOS ONE 11: e0151870.

Carlsson AM, Mastromonaco G, Vandervalk E, Kutz S (2016a) Parasites, stress and reindeer: Infection with abomasal nematodes is not associated with elevated glucocorticoid levels in hair or faeces. Conserv Physiol 4: cow058.

Carlsson AM, Veitch AM, Popko RA, Behrens S, Kutz SJ (2016b) Monitoring wildlife health for conservation and food security in the Canadian North: A case study from the Sahtu settlement area in the Northwest Territories. In: Cork S, Hall D, Karen L, eds. One Health Case Studies: Addressing Complex Problems in a Changing World, 1st Edition. 5M Publishing, Essex, UK, pp 132–151.

CARMA (2008) CircumArctic Rangifer Monitoring and Assessment (CARMA) Network: Rangifer Health & Body Condition Monitoring, Monitoring Protocols Level II. p 54.

Caslini C, Comin A, Peric T, Prandi A, Pedrotti L, Mattiello S (2016) Use of hair cortisol analysis for comparing population status in wild red deer (Cervus elaphus) living in areas with different characteristics. Eur J Wildl Res 62: 713–723.

Cattet M, Macbeth BJ, Janz DM, Zedrosser A, Swenson JE, Dumond M, Stenhouse GB (2014) Quantifying long-term stress in brown bears with the hair cortisol concentration: A biomarker that may be confounded by rapid changes in response to capture and handling. Conserv Physiol 2: cou026.

Cattet M, Stenhouse GB, Janz DM, Kapronczai L, Anne Erlenbach J, Jansen HT, Nelson OL, Robbins CT, Boulanger J (2017) The quantification of reproductive hormones in the hair of captive adult brown bears and their application as indicators of sex and reproductive state. Conserv Physiol 5. doi:10.1093/conphys/cox032

Chapman CA, Schoof VAM, Bonnell TR, Gogarten JF, Calmé S (2015) Competing pressures on populations: Long-term dynamics of food availability, food quality, disease, stress and animal abundance. Philos Trans R Soc B Biol Sci 370: 20140112.

Chapman CA, Wasserman MD, Gillespie TR, Speirs ML, Lawes MJ, Saj TL, Ziegler TE (2006) Do food availability, parasitism, and stress have synergistic effects on red colobus populations living in forest fragments? Am J Phys Anthropol 131: 525–534.

Chartier C, Paraud C (2012) Coccidiosis due to Eimeria in sheep and goats, a review. Small Rumin Res 103: 84–92.

Chen I-C, Hill JK, Ohlemuller R, Roy DB, Thomas CD (2011) Rapid range shifts of species associated with high levels of climate warming. Science 333: 1024–1026.

Cizauskas CA, Turner WC, Pitts N, Getz WM (2015) Seasonal patterns of hormones, macroparasites, and microparasites in wild African ungulates: The interplay among stress, reproduction, and disease. PLOS ONE 10: e0120800.

131

Coltman DW, O’Donoghue P, Jorgenson JT, Hogg JT, Strobeck C, Festa-Bianchet M (2003) Undesirable evolutionary consequences of trophy hunting. Nature 426: 655–658.

Coradello MA, Morais RN, Roper J, Spercoski KM, Massuda T, Nogueira SSC, Nogueira-Filho SLG (2012) Validation of a fecal glucocorticoid metabolite assay for collared peccaries (Pecari Tajacu). J Zoo Wildl Med 43: 275–282.

Corlatti L, Palme R, Lovari S (2014) Physiological response to etho-ecological stressors in male Alpine chamois: timescale matters! Naturwissenschaften 101: 577–586.

Cote J, Clobert J, Meylan S, Fitze PS (2006) Experimental enhancement of corticosterone levels positively affects subsequent male survival. Horm Behav 49: 320–327.

Creel S (2001) Social dominance and stress hormones. Trends Ecol Evol 16: 491–497.

Creel S, Fox JE, Hardy A, Sands J, Garrott B, Peterson RO (2002) Snowmobile activity and glucocorticoid stress responses in wolves and elk. Conserv Biol 16: 809–814.

Creel S, Winnie JA, Christianson D (2009) Glucocorticoid stress hormones and the effect of predation risk on elk reproduction. Proc Natl Acad Sci 106: 12388–12393.

Crespi EJ, Williams TD, Jessop TS, Delehanty B (2013) Life history and the ecology of stress: How do glucocorticoid hormones influence life-history variation in animals? Funct Ecol 27: 93–106.

Crill C, Janz DM, Kusch JM, Santymire RM, Heyer GP, Shury TK, Lane JE (2019) Investigation of the utility of feces and hair as non-invasive measures of glucocorticoids in wild black-tailed prairie dogs (Cynomys ludovicianus). Gen Comp Endocrinol 275: 15–24.

Crowther JR (2009) The ELISA Guidebook. Humana Press, Totowa, USA.

Curry PS, Elkin BT, Campbell M, Nielsen K, Hutchins W, Ribble C, Kutz SJ (2011) Filter-paper blood samples for ELISA detection of Brucella antibodies in caribou. J Wildl Dis 47: 12–20.

Cuyler C, Daniel CJ, Enghoff M, Levermann N, Møller‐Lund N, Hansen PN, Damhus D, Danielsen F (2020) Using local ecological knowledge as evidence to guide management: A community‐led harvest calculator for muskoxen in Greenland. Conserv Sci Pract 2: e159.

Cuyler C, Rowell J, Adamczewski J, Anderson M, Blake J, Bretten T, Brodeur V, Campbell M, Checkley SL, Cluff HD, et al. (2019) Muskox status, recent variation, and uncertain future. Ambio 49: 805– 819.

Danielsen F, Burgess ND, Balmford A, Donald PF, Funder M, Jones JPG, Alviola P, Balete DS, Blomley T, Brashares J, et al. (2009) Local participation in natural resource monitoring: A characterization of approaches. Conserv Biol 23: 31–42.

Dantzer B, Boutin S, Humphries MM, McAdam AG (2012) Behavioral responses of territorial red squirrels to natural and experimental variation in population density. Behav Ecol Sociobiol 66: 865–878.

Dantzer B, Fletcher QE, Boonstra R, Sheriff MJ (2014) Measures of physiological stress: A transparent or opaque window into the status, management and conservation of species? Conserv Physiol 2: cou023.

Dantzer B, Newman AEM, Boonstra R, Palme R, Boutin S, Humphries MM, McAdam AG (2013) Density triggers maternal hormones that increase adaptive offspring growth in a wild mammal. Science 340: 1215–1217.

132

Dantzer B, Westrick SE, van Kesteren F (2016) Relationships between endocrine traits and life histories in wild animals: Insights, problems, and potential pitfalls. Integr Comp Biol 56: 185–197.

Davenport MD, Tiefenbacher S, Lutz CK, Novak MA, Meyer JS (2006) Analysis of endogenous cortisol concentrations in the hair of rhesus macaques. Gen Comp Endocrinol 147: 255–261.

Davison T, Branigan M (2014) Aerial Survey of Muskoxen (Ovibos Moschatus) in the Mainland of the Inuvialuit Settlement Region, March 2009 (Manuscript Report. No. 243). Department of Environment and Natural Resources, Government of the Northwest Territories, Canada, p 11.

Davison T, Pongracz J, Williams J (2013) Population survey of Peary caribou (Rangifer tarandus pearyi) and muskoxen (Ovibos moschatus) on Banks Island, Northwest Territories, July 2010. Rangifer 33: 135.

Davison T, Williams J (2013) Peary caribou (Rangifer tarandus pearyi) and muskoxen (Ovibos moschatus) on northwest Victoria Island, Northwest Territories. Rangifer 33: 129.

Davison T, Williams J, Adamczewski J (2017) Aerial Survey of Peary Caribou (Rangifer Tarandus Pearyi) and Muskoxen (Ovibos Moschatus) on Banks Island, July 2014 (Manuscript Report. No. 270). Department of Environment and Natural Resources, Government of the Northwest Territories, Canada.

De Bruyn NP (2010) Gastrointestinal Nematodes of Western Canadian Cervids: Molecular Diagnostics, Faunal Baselines and Management Considerations (Master’s thesis). Department of Biological Sciences, University of Calgary, Calgary, Alberta, Canada.

Defolie C, Merkling T, Fichtel C (2020) Patterns and variation in the mammal parasite–glucocorticoid relationship. Biol Rev 95: 74–93.

Dehnhard M, Clauss M, Lechner-Doll M, Meyer HHD, Palme R (2001) Noninvasive monitoring of adrenocortical activity in roe deer (Capreolus capreolus) by measurement of fecal cortisol metabolites. Gen Comp Endocrinol 123: 111–120.

Department of Environment, Government of Nunavut (2018) 2018/2019 Summary of the Nunavut hunting regulations.

Desaulniers DM, Goff AK, Betteridge KJ, Rowell JE, Flood PF (1989) Reproductive hormone concentrations in faeces during the oestrous cycle and pregnancy in cattle (Bos taurus) and muskoxen (Ovibos moschatus). Can J Zool 67: 1148–1154.

Dettmer AM, Novak MA, Meyer JS, Suomi SJ (2014) Population density-dependent hair cortisol concentrations in rhesus monkeys (Macaca mulatta). Psychoneuroendocrinology 42: 59–67.

Dickens MJ, Romero LM (2013) A consensus endocrine profile for chronically stressed wild animals does not exist. Gen Comp Endocrinol 191: 177–189.

Downs CJ, Boan BV, Lohuis TD, Stewart KM (2018) Investigating relationships between reproduction, immune defenses, and cortisol in Dall sheep. Front Immunol 9: 105.

Dulude-de Broin F, Côté SD, Whiteside DP, Mastromonaco GF (2019) Faecal metabolites and hair cortisol as biological markers of HPA-axis activity in the Rocky mountain goat. Gen Comp Endocrinol 280: 147–157.

133

Dulude-de Broin F, Hamel S, Mastromonaco GF, Côté SD (2020) Predation risk and mountain goat reproduction: Evidence for stress‐induced breeding suppression in a wild ungulate. Funct Ecol 34: 1003–1014.

Dumond M (2007) Western Kitikmeot Caribou Workshop (Final Wildlife Report No. 19). Department of Environment, Government of Nunavut, Canada, p 49.

ECCC (2018) Management Plan for the Barren-Ground Caribou (Rangifer Tarandus Groenlandicus), Dolphin and Union Population, in Canada: Adoption of the Management Plan for the Dolphin and Union Caribou (Rangifer Tarandus Groenlandicus x Pearyi) in the Northwest Territories and Nunavut. (Species at Risk Act Management Plan Series). Environment and Climate Change Canada (ECCC), Ottawa, Canada, p 107.

Edwards PD, Boonstra R (2018) Glucocorticoids and CBG during pregnancy in mammals: diversity, pattern, and function. Gen Comp Endocrinol 259: 122–130.

Egwang TG, Slocombe JO (1982) Evaluation of the Cornell-Wisconsin centrifugal flotation technique for recovering trichostrongylid eggs from bovine feces. Can J Comp Med 46: 133–137.

Ellis RD, McWhorter TJ, Maron M (2012) Integrating landscape ecology and conservation physiology. Landsc Ecol 27: 1–12.

Ericsson G, Wallin K (2001) Age-specific moose (Alces alces) mortality in a predator-free environment: Evidence for senescence in females. Écoscience 8: 157–163.

Erler NS, Rizopoulos D, Rosmalen J van, Jaddoe VWV, Franco OH, Lesaffre EMEH (2016) Dealing with missing covariates in epidemiologic studies: a comparison between multiple imputation and a full Bayesian approach. Stat Med 35: 2955–2974.

Ewacha MVA, Roth JD, Anderson WG, Brannen DC, Dupont DLJ (2017) Disturbance and chronic levels of cortisol in boreal woodland caribou. J Wildl Manag 81: 1266–1275.

Fain SR, LeMay JP (1995) Gender identification of humans and mammalian wildlife species from PRC amplified sex linked genes. Proc Am Acad Forensic Sci 1: 34.

Fardi S, Sauther MichelleL, Cuozzo FP, Jacky IAY, Bernstein RM (2018) The effect of extreme weather events on hair cortisol and body weight in a wild ring-tailed lemur population (Lemur catta) in southwestern Madagascar. Am J Primatol 80: e22731.

Ferrell EW, Nance CN, Torres AL, Torres SM (2014) Using participatory action research to address absenteeism. Action Learn Res Pract 11: 201–214.

Feughelman M (1997) Introduction to the physical properties of wool, hair & other alpha-keratin fibres. In: Feughelman M, ed. Mechanical Properties and Structure of Alpha-Keratin Fibres: Wool, Human Hair, and Related Fibres, 1st Edition. University of South Wales Press, Sydney, Australia, pp 1–15.

Flood PF, Stalker MJ, Rowell JE (1989) The hair follicle density and seasonal shedding cycle of the muskox (Ovibos moschatus). Can J Zool 67: 1143–1147.

Foley CAH, Papageorge S, Wasser SK (2001) Noninvasive stress and reproductive measures of social and ecological pressures in free-ranging African elephants. Conserv Biol 15: 1134–1142.

Forde TL, Orsel K, Zadoks RN, Biek R, Adams LG, Checkley SL, Davison T, De Buck J, Dumond M, Elkin BT, et al. (2016) Bacterial genomics reveal the complex epidemiology of an emerging pathogen in Arctic and boreal ungulates. Front Microbiol 7. doi:10.3389/fmicb.2016.01759

134

Forrester SG, Lankester MW (1997) Extracting protostrongylid nematode larvae from ungulate feces. J Wildl Dis 33: 511–516.

Fourie NH, Jolly CJ, Phillips-Conroy JE, Brown JL, Bernstein RM (2015a) Variation of hair cortisol concentrations among wild populations of two baboon species (Papio anubis, P. hamadryas) and a population of their natural hybrids. Primates 56: 259–272.

Fourie NH, Turner TR, Brown JL, Pampush JD, Lorenz JG, Bernstein RM (2015b) Variation in vervet (Chlorocebus aethiops) hair cortisol concentrations reflects ecological disturbance by humans. Primates 56: 365–373.

Freeman NL (2008) Motorized Backcountry Recreation and Stress Response in Mountain Caribou (Rangifer Tarandus Caribou) (Master’s thesis). University of British Columbia, Vancouver, British Columbia, Canada.

Fürtbauer I, Solman C, Fry A (2019) Sheep wool cortisol as a retrospective measure of long-term HPA axis activity and its links to body mass. Domest Anim Endocrinol 68: 39–46.

Gadgil M, Berkes F, Folke C (1993) Indigenous knowledge for biodiversity conservation. Ambio 22: 151– 156.

Gale JP, Yergey AL, Duncan MW (2015) Quantifying small molecules by mass spectrometry. LCGC 33: 34–41.

Gamberg M (2017) Muskox Health Program: Contaminants in Country Foods in Kitikmeot, NU Technical Report. Nunavut, Canada.

Ganswindt A, Tordiffe ASW, Stam E, Howitt MJ, Jori F (2012) Determining adrenocortical activity as a measure of stress in African buffalo (Syncerus caffer) based on faecal analysis. Afr Zool 47: 262–269.

Garde AH, Hansen ÅM (2005) Long‐term stability of salivary cortisol. Scand J Clin Lab Invest 65: 433–436.

Garel M, Cugnasse J-M, Maillard D, Gaillard J-M, Hewison AJM, Dubray D (2007) Selective harvesting and habitat loss produce long-term life history changes in a mouflon population. Ecol Appl 17: 1607–1618.

Gelman A, Carlin JB, Stern HS, Rubin DB (2004) Chapter 6: Model checking and improvement. In: Chatfield C, Tanner M, Zidek J, eds. Bayesian Data Analysis, 2nd Edition. Chapman & Hall/CRC, Boca Raton, London, New York, Washington D.C., pp 157–196.

George SC, Smith TE, Mac Cana PSS, Coleman R, Montgomery WI (2014) Physiological stress in the Eurasian badger (Meles meles): Effects of host, disease and environment. Gen Comp Endocrinol 200: 54–60.

Gesquiere LR, Ziegler TE, Chen PA, Epstein KA, Alberts SC, Altmann J (2014) Measuring fecal testosterone in females and fecal estrogens in males: Comparison of RIA and LC/MS/MS methods for wild baboons (Papio cynocephalus). Gen Comp Endocrinol 204: 141–149.

Ghassemi Nejad J, Lee B-H, Kim J-Y, Kim B-W, Chemere B, Park K-H, Sung K-I (2019) Comparing hair cortisol concentrations from various body sites and serum cortisol in Holstein lactating cows and heifers during thermal comfort zone. J Vet Behav 30: 92–95.

Gillespie TR, Barelli C, Heistermann M (2013) Effects of social status and stress on patterns of gastrointestinal parasitism in wild white‐handed gibbons (Hylobates lar). Phys Anthropol 150: 602– 608.

135

Gittelsohn J (2010) Participatory research for chronic disease prevention in Inuit communities. Am J Health Behav 34: 453–464.

Gobush KS, Mutayoba BM, Wasser SK (2008) Long-term impacts of poaching on relatedness, stress hhysiology, and reproductive output of adult female African elephants. Conserv Biol 22: 1590–1599.

Goldstein EJ, Millspaugh JJ, Washburn BE, Brundige GC, Raedeke KJ (2005) Relationships among fecal lungworm loads, fecal glucocorticoid metabolites, and lamb recruitment in free-ranging Rocky Mountain bighorn sheep. J Wildl Dis 41: 416–425.

González-de-la-Vara M del R, Valdez RA, Lemus-Ramirez V, Vázquez-Chagoyán JC, Villa-Godoy A, Romano MC (2011) Effects of adrenocorticotropic hormone challenge and age on hair cortisol concentrations in dairy cattle. Can J Vet Res 75: 216–221.

Goymann W (2012) On the use of non‐invasive hormone research in uncontrolled, natural environments the problem with sex, diet, metabolic rate and the individual. Methods Ecol Evloution 3: 757–765.

Gray DR (1987) The Muskoxen of Polar Bear Pass. Fitzhenry & Whiteside in trust for the National Museum of Natural Sciences, National Museums of Canada, Markham, Canada.

Green J, Thorogood N (2014) Qualitative Methods for Health Research, 3rd Edition. SAGE Publications, London, UK.

Green JA, Parks TE, Avalle MP, Telugu BP, McLain AL, Peterson AJ, McMillan W, Mathialagan N, Hook RR, Xie S, et al. (2005) The establishment of an ELISA for the detection of pregnancy-associated glycoproteins (PAGs) in the serum of pregnant cows and heifers. Theriogenology 63: 1481–1503.

Greunz EM, Stelvig M, Grøndahl C, Schmidt NM, Mosbacher JB, Hansen LH, Bertelsen MF (2018) Evaluation of two enzyme-linked immunosorbent assays measuring pregnancy-associated glycoproteins in the blood of muskoxen (Ovibos moschatus). J Zoo Wildl Med 49: 798–801.

Groves P (1997) Intraspecific variation in mitochondrial DNA of muskoxen, based on control-region sequences. Can J Zool 75: 568–575.

Gunn A, Adamczewski J (2003) Muskox (Ovibos moschatus). In: Wild Mammals of North America: Biology, Management, and Conservation, 2nd Edition. The John Hopkins University Press, Baltimore, USA and London, UK, pp 1076–1094.

Gunn A, Leighton T, Wobeser G (1991) Wildlife Diseases and Parasites in the Kitikmeot Region, 1984-90. (File Report No. 104). Department of Environment and Natural Resources, Government of the Northwest Territories, Canada, p 51.

Gunn A, Wobeser G (1993) Protostrongylid lungworm infection in muskoxen, Coppermine, N.W.T., Canada. Rangifer 13: 45.

Hall SR, Duffy MA, Caceres CE (2005) Selective predation and productivity jointly drive complex behavior in host‐parasite systems. Am Nat 165: 70–81.

Halvorsen O, Skorping A, Hansen K (1985) Seasonal cycles in the output of first stage larvae of the nematode Elaphostrongylus rangiferi from reindeer, Rangifer tarandus tarandus. Polar Biol 5: 49–54.

Handeland K, Tengs T, Kokotovic B, Vikøren T, Ayling RD, Bergsjø B, Sigurðardóttir ÓG, Bretten T (2014) Mycoplasma ovipneumoniae - A primary cause of Severe pneumonia epizootics in the Norwegian muskox (Ovibos moschatus) population. PLOS ONE 9: e106116.

136

Handelsman DJ, Wartofsky L (2013) Requirement for mass spectrometry sex steroid assays in the Journal of Clinical Endocrinology and Metabolism. J Clin Endocrinol Metab 98: 3971–3973.

Hanisch SL, Riley SJ, Nelson MP (2012) Promoting wildlife health or fighting wildlife disease: Insights from history, philosophy, and science. Wildl Soc Bull 36: 477–482.

Hanke AN, Angohiatok M, Leclerc L-M, Adams C, Kutz S (2020) The dynamic life of caribou: Inuit insights into abundance, distribution, ecology, and helath of the Dolphin and Union herd. ARCTIC Submitted September 30, 2020.

Hansen BB, Aanes R, Herfindal I, Kohler J, Sæther B-E (2011) Climate, icing, and wild arctic reindeer: Past relationships and future prospects. Ecology 92: 1917–1923.

Harris D (1945) Symptoms of malnutrition in deer. J Wildl Manag 9: 319.

Heimbürge S, Kanitz E, Otten W (2019) The use of hair cortisol for the assessment of stress in animals. Gen Comp Endocrinol 270: 10–17.

Heimbürge S, Kanitz E, Tuchscherer A, Otten W (2020a) Within a hair’s breadth – Factors influencing hair cortisol levels in pigs and cattle. Gen Comp Endocrinol 288: 113359.

Heimbürge S, Kanitz E, Tuchscherer A, Otten W (2020b) Is it getting in the hair? – Cortisol concentrations in native, regrown and segmented hairs of cattle and pigs after repeated ACTH administrations. Gen Comp Endocrinol 295: 113534.

Hein A, Palme R, Baumgartner K, von Fersen L, Woelfing B, Greenwood AD, Bechshoft T, Siebert U (2020) Faecal glucocorticoid metabolites as a measure of adrenocortical activity in polar bears (Ursus maritimus). Conserv Physiol 8: coaa012.

Helfferich D (2008) The Muskox: A New Northern Farm Animal. Agricultural and Forestry Experiment Station, School of Agriculture and Land Resources Management, University of Alaska Fairbanks, Fairbanks, USA, p 32.

Henderson GL (1993) Mechanisms of drug incorporation into hair. Forensic Sci Int 63: 19–29.

Henrichsen P, Grue H (1980) Age criteria in the muskox (Ovibos moschatus) from Greenland. Dan Rev Game Biol 11.

Hoberg EP, Polley L, Gunn A, Nishi JS (1995) Umingmakstrongylus pallikuukensis gen.nov. et sp.nov. (Nematoda: Protostrongylidae) from muskoxen, Ovibos moschatus, in the central Canadian Arctic, with comments on biology and biogeography. Can J Zool 73: 2266–2282.

Hoberg EP, Polley L, Jenkins EJ, Kutz SJ, Veitch AM, Elkin BT (2008) Integrated approaches and empirical models for investigation of parasitic diseases in northern wildlife. Emerg Infect Dis 14: 10–17.

Hoby S, Schwarzenberger F, Doherr MG, Robert N, Walzer C (2006) Steroid hormone related male biased parasitism in chamois, Rupicapra rupicapra rupicapra. Vet Parasitol 138: 337–348.

Holleman DF, White RG, Frisby K, Jourdan M, Henrichsen P, Tallas PG (1984) Food passage rates in captive muskoxen as measured with non-absorbed radiolabeled markers. Biol Pap Univ Alsk Spec Rep 188–192.

Huber S, Palme R, Arnold W (2003) Effects of season, sex, and sample collection on concentrations of fecal cortisol metabolites in red deer (Cervus elaphus). Gen Comp Endocrinol 130: 48–54.

137

Hunt KE, Wasser SK (2003) Effect of long‐term preservation methods on fecal glucocorticoid concentrations of grizzly bear and African elephant. Physiol Biochem Zool 76: 918–928.

Huntington H, Fox S (2005) The changing Arctic: Indigenous perspectives. In: Symon C, Arris L, Heal B, eds. Arctic Climate Impact Assessment—Scientific Report. Cambridge University Press, Cambridge, UK, pp 61–98.

Huntington HP (2011) Arctic science: The local perspective. Nature 478: 182–183.

Huntington HP, Suydam RS, Rosenberg DH (2004) Traditional knowledge and satellite tracking as complementary approaches to ecological understanding. Environ Conserv 31: 177–180.

Indian and Northern Affairs (1984) The Western Arctic claim: The Inuvialuit Final Agreement.

Indian and Northern Affairs (1993) Agreement between the Inuit of the Nunavut Settlement Area and Her Majesty in Right of Canada.

Inuit Tapiriit Kanatami (2018) National Inuit strategy on research.

Inuit Tapiriit Kanatami, Nunavut Research Institute (2006) Negotiating research partnership with Inuit communities: A guide for researchers.

IPCC (2013) Summary for Policymakers. In: Stocker T, Qin D, Plattner G, Tignor M, Allen S, Boschung J, Nauels A, Xia Y, Bex V, Midgley P, eds. Cambridge University Press, Cambridge, UK and New York, USA.

IPCC (2014) Climate Change 2014: Impacts, Adaptation, and Vulnerability. Contribution of Working Group II to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). Cambridge University Press, Cambridge, UK.

Irvine RJ, Stien A, Dallas JF, Halvorsen O, Langvatn R, Albon SD (2001) Contrasting regulation of fecundity in two abomasal nematodes of Svalbard reindeer (Rangifer tarandus platyrhynchus). Parasitology 122: 673–681.

Ito N, Ito T, Kromminga A, Bettermann A, Takigawa M, Kies F, Straub RH, Paus R (2005) Human hair follicles display a functional equivalent of the hypothalamic-pituitary-adrenal (HPA) axis and synthesize cortisol. FASEB J 19: 1332–1334.

Janska E (2008) Health Knowledge, Traditional. In: Kirch W, ed. Encyclopedia of Public Health. Springer, Dordrecht, Netherlands.

Jeanniard du Dot T, Rosen DAS, Richmond JP, Kitaysky AS, Zinn SA, Trites AW (2009) Changes in glucocorticoids, IGF-I and thyroid hormones as indicators of nutritional stress and subsequent refeeding in Steller sea lions (Eumetopias jubatus). Comp Biochem Physiol A Mol Integr Physiol 152: 524– 534.

Jewgenow K, Azevedo A, Albrecht M, Kirschbaum C, Dehnhard M (2020) Hair cortisol analyses in different mammal species: Choosing the wrong assay may lead to erroneous results. Conserv Physiol 8: coaa009.

Jingfors KT (1982) Seasonal activity budgets and movements of a reintroduced Alaskan muskox herd. J Wildl Manag 46: 344.

Kafle P (2018) Climate Change and Range Expansion of Protostrongylid Nematodes in the Canadian Arctic (PhD dissertation). Faculty of Veterinary Medicine, University of Calgary, Calgary, Alberta, Canada.

138

Kafle P, Lejeune M, Verocai GG, Hoberg EP, Kutz SJ (2015) Morphological and morphometric differentiation of dorsal-spined first stage larvae of lungworms (Nematoda: Protostrongylidae) infecting muskoxen (Ovibos moschatus) in the central Canadian Arctic. Int J Parasitol Parasites Wildl 4: 283–290.

Kafle P, Peller P, Massolo A, Hoberg E, Leclerc L-M, Tomaselli M, Kutz S (2020) Range expansion of muskox lungworms track rapid arctic warming: Implications for geographic colonization under climate forcing. Sci Rep 10: 17323.

Kalliokoski O, Jellestad FK, Murison R (2019) A systematic review of studies utilizing hair glucocorticoids as a measure of stress suggests the marker is more appropriate for quantifying short-term stressors. Sci Rep 9: 11997.

Kapoor A, Lubach GR, Ziegler TE, Coe CL (2016) Hormone levels in neonatal hair reflect prior maternal stress exposure during pregnancy. Psychoneuroendocrinology 66: 111–117.

Kapoor A, Schultz-Darken N, Ziegler TE (2018) Radiolabel validation of cortisol in the hair of rhesus monkeys. Psychoneuroendocrinology 97: 190–195.

Keckeis K, Lepschy M, Schöpper H, Moser L, Troxler J, Palme R (2012) Hair cortisol: A parameter of chronic stress? Insights from a radiometabolism study in guinea pigs. J Comp Physiol B 182: 985– 996.

Khonmee J, Brown JL, Rojanasthien S, Aunsusin A, Thumasanukul D, Kongphoemphun A, Siriaroonrat B, Tipkantha W, Punyapornwithaya V, Thitaram C (2014) Gender, season and management affect fecal glucocorticoid metabolite concentrations in captive goral (Naemorhedus griseus) in Thailand. PLOS ONE 9: e91633.

Kidwell DA, Blank DL (1996) Environmental exposure – the stumbling block of hair testing. In: Kintz P, ed. Drug Testing in Hair, 1st Edition. CRC Press, Boca Raton, Florida, USA, pp 17–64.

Kirschbaum C, Tietze A, Skoluda N, Dettenborn L (2009) Hair as a retrospective calendar of cortisol production–—Increased cortisol incorporation into hair in the third trimester of pregnancy. Psychoneuroendocrinology 34: 32–37.

Kojola I, Helle T, Huhta E, Niva A (1998) Foraging conditions, tooth wear and herbivore body reserves: A study of female reindeer. Oecologia 117: 26–30.

Koren L, Bryan H, Matas D, Tinman S, Fahlman Å, Whiteside D, Smits J, Wynne‐Edwards K (2019) Towards the validation of endogenous steroid testing in wildlife hair. J Appl Ecol 56: 547–561.

Koren L, Mokady O, Karaskov T, Klein J, Koren G, Geffen E (2002) A novel method using hair for determining hormonal levels in wildlife. Anim Behav 63: 403–406.

Koren L, Nakagawa S, Burke T, Soma KK, Wynne-Edwards KE, Geffen E (2012a) Non-breeding feather concentrations of testosterone, corticosterone and cortisol are associated with subsequent survival in wild house sparrows. Proc R Soc B Biol Sci 279: 1560–1566.

Koren L, Ng ESM, Soma KK, Wynne-Edwards KE (2012b) Sample preparation and liquid chromatography-tandem mass spectrometry for multiple steroids in mammalian and avian circulation. PLOS ONE 7: e32496.

Koren L, Whiteside D, Fahlman Å, Ruckstuhl K, Kutz S, Checkley S, Dumond M, Wynne-Edwards K (2012c) Cortisol and corticosterone independence in cortisol-dominant wildlife. Gen Comp Endocrinol 177: 113–119.

139

Kovach M (2009) Indigenous Methodologies: Characteristics, Conversations, and Contexts. University of Toronto Press, Toronto, Canada.

Kumar V, Umapathy G (2019) Non-invasive monitoring of steroid hormones in wildlife for conservation and management of endangered species — A review. Indian J Exp Biol 57: 307–314.

Kumpula T, Pajunen A, Kaarlejärvi E, Forbes BC, Stammler F (2011) Land use and land cover change in Arctic Russia: Ecological and social implications of industrial development. Glob Environ Change 21: 550–562.

Kutz S, Bollinger T, Branigan M, Checkley S, Davison T, Dumond M, Elkin B, Forde T, Hutchins W, Niptanatiak A, et al. (2015) Erysipelothrix rhusiopathiae associated with recent widespread muskox mortalities in the Canadian Arctic. Can Vet J 56: 560–563.

Kutz S, Checkley S, Verocai GG, Dumond M, Hoberg EP, Peacock R, Wu JP, Orsel K, Seegers K, Warren AL, et al. (2013a) Invasion, establishment, and range expansion of two parasitic nematodes in the Canadian Arctic. Glob Change Biol 19: 3254–3262.

Kutz S, Ducrocq J, Cuyler C, Elkin B, Gunn A, Kolpashikov L, Russell D, White RG (2013b) Standardized monitoring of Rangifer health during International Polar Year. Rangifer 91–114.

Kutz S, Rowell J, Adamczewski J, Gunn A, Cuyler C, Aleuy OA, Austin M, Berger J, Blake J, Bondo K, et al. (2017) Muskox Health Ecology Symposium 2016: Gathering to share knowledge on Umingmak in a time of rapid change. ARCTIC 70: 225.

Kutz S, Tomaselli M (2019) “Two-eyed seeing” supports wildlife health. Science 364: 1135–1137.

Kutz SJ, Ducrocq J, Verocai GG, Hoar BM, Colwell DD, Beckmen KB, Polley L, Elkin BT, Hoberg EP (2012) Chapter 2 - Parasites in ungulates of Arctic North America and Greenland: A view of contemporary diversity, ecology, and impact in a world under change. Adv Parasitol 79: 99–252.

Kutz SJ, Hoberg EP, Molnár PK, Dobson A, Verocai GG (2014) A walk on the tundra: Host–parasite interactions in an extreme environment. Int J Parasitol Parasites Wildl 3: 198–208.

Kutz SJ, Hoberg EP, Nagy J, Polley L, Elkin BT (2004) “Emerging” Parasitic Infections in Arctic Ungulates. Integr Comp Biol 44: 109–118.

Kutz SJ, Hoberg EP, Polley L (1999) Experimental infections of muskoxen (Ovibos moschatus) and domestic sheep with Umingmakstrongylus pallikuukensis (Nematoda: Protostrongylidae): Parasite development, population structure, and pathology. Can J Zool 77: 1562–1572.

Lafferty DJR, Laudenslager ML, Mowat G, Heard D, Belant JL (2015) Sex, diet, and the social environment: Factors influencing hair cortisol concentration in free-ranging black bears (Ursus americanus). PLOS ONE 10: e0141489.

Laforest BJ, Hébert JS, Obbard ME, Thiemann GW (2018) Traditional ecological knowledge of polar bears in the northern Eeyou Marine Region, Québec, Canada. ARCTIC 71: 40–58.

Landys MM, Ramenofsky M, Wingfield JC (2006) Actions of glucocorticoids at a seasonal baseline as compared to stress-related levels in the regulation of periodic life processes. Gen Comp Endocrinol 148: 132–149.

Langlois A, Johnson C-A, Montpetit B, Royer A, Blukacz-Richards EA, Neave E, Dolant C, Roy A, Arhonditsis G, Kim D-K, et al. (2017) Detection of rain-on-snow (ROS) events and ice layer formation using passive microwave radiometry: A context for Peary caribou habitat in the Canadian Arctic. Remote Sens Environ 189: 84–95.

140

Laudenslager ML, Jorgensen MJ, Fairbanks LA (2012) Developmental patterns of hair cortisol in male and female nonhuman primates: Lower hair cortisol levels in vervet males emerge at puberty. Psychoneuroendocrinology 37: 1736–1739.

Lavergne SG, Peers MJL, Mastromonaco G, Majchrzak YN, Nair A, Boutin S, Boonstra R (2020) Hair cortisol as a reliable indicator of stress physiology in the snowshoe hare: Influence of body region, sex, season, and predator–prey population dynamics. Gen Comp Endocrinol 294: 113471.

Lawler JP, White RG (1997) Seasonal changes in metabolic rates in muskoxen following twenty- four hours of starvation. Rangifer 17: 135.

Lebreton J-D (2011) The impact of global change on terrestrial vertebrates. C R Biol 334: 360–369.

Leclerc L-M (2014) Muskox Aerial Survey (Ovibos Moschatus) of the Kitikmeot Region, Nunavut (Final Report). Department of Environment, Government of Nunavut, Canada.

Lent PC (1999) Muskoxen and Their Hunters: A History. University of Oklahoma Press, Norman, USA.

Levine JE (2002) Editorial: Stressing the importance of sex. Endocrinology 143: 4502–4504.

Lubin D, Massom R (2006) Introduction. In: Polar Remote Sensing. Volume I: Atmosphere and Oceans. Springer Berlin Heidelberg, Germany in association with Praxis Publishing Ltd, Chichester, UK, pp 1–60.

Lynch JW, Ziegler TE, Strier KB (2002) Individual and seasonal variation in fecal testosterone and cortisol levels of wild male tufted capuchin monkeys, Cebus apella nigritus. Horm Behav 41: 275–287.

Ma Z, Chen G (2018) Bayesian methods for dealing with missing data problems. J Korean Stat Soc 47: 297– 313.

Macbeth BJ (2013) An Evaluation of Hair Cortisol Concentration as a Potenital Biomarker of Long-Term Stress in Free-Ranging Grizzly Bears (Ursus arctos), Polar Bears (Ursus maritimus) and Caribou (Rangifer tarandus sp.) (PhD dissertation). University of Saskatchewan, Saskatoon, Saskatchewan, Canada.

Macbeth BJ, Cattet MRL, Obbard ME, Middel K, Janz DM (2012) Evaluation of hair cortisol concentration as a biomarker of long-term stress in free-ranging polar bears. Wildl Soc Bull 36: 747–758.

Macbeth BJ, Cattet MRL, Stenhouse GB, Gibeau ML, Janz DM (2010) Hair cortisol concentration as a noninvasive measure of long-term stress in free-ranging grizzly bears (Ursus arctos): Considerations with implications for other wildlife. Can J Zool 88: 935–949.

Macbeth BJ, Kutz SJ (2019) Rangifer health: A holistic perspective. In: Tryland M, Kutz SJ, eds. Reindeer and Caribou: Health and Disease, 1st Edition. CRC Press, Taylor & Francis Group, Boca Raton, USA, pp 91–106.

MacCarthy JJ, Canziani OF, Leary NA, Dokken DJ, White KS (2001) Impacts, Adaptation, and Vulnerability; Contribution of Working Group II to the Third Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge, UK.

MacDougall-Shackleton SA, Bonier F, Romero LM, Moore IT (2019) Glucocorticoids and “stress” are not synonymous. Integr Org Biol 1: obz017.

MacPhee RD, Tikhonov AN, Mol D, Greenwood AD (2005) Late Quaternary loss of genetic diversity in muskox (Ovibos). BMC Evol Biol 5: 1.

141

Madslien K, Stubsjøen SM, Viljugrein H, Ytrehus B, Solberg EJ, Kapronczai L, Mysterud A, Godfroid J, Janz DM, Cattet M (2020) Hair cortisol concentration and body mass in moose (Alces alces) infested with deer keds (Lipoptena cervi). J Wildl Dis 2019-07–185.

Majchrzak YN, Mastromonaco GF, Korver W, Burness G (2015) Use of salivary cortisol to evaluate the influence of rides in dromedary camels. Gen Comp Endocrinol 211: 123–130.

Malcolm KD, McShea WJ, Van Deelen TR, Bacon HJ, Liu F, Putman S, Zhu X, Brown JL (2013) Analyses of fecal and hair glucocorticoids to evaluate short- and long-term stress and recovery of Asiatic black bears (Ursus thibetanus) removed from bile farms in China. Gen Comp Endocrinol 185: 97–106.

Marasco V, Boner W, Heidinger B, Griffiths K, Monaghan P (2015) Repeated exposure to stressful conditions can have beneficial effects on survival. Exp Gerontol 69: 170–175.

Martinez-Levasseur LM, Furgal CM, Hammill MO, Burness G (2017) Challenges and strategies when mapping local ecological knowledge in the Canadian Arctic: The importance of defining the geographic limits of participants’ common areas of observations. Polar Biol 40: 1501–1513.

Mastromonaco GF, Gunn K, McCurdy-Adams H, Edwards DB, Schulte-Hostedde AI (2014) Validation and use of hair cortisol as a measure of chronic stress in eastern chipmunks (Tamias striatus). Conserv Physiol 2: cou055.

Mavrot F, Orsel K, Hutchins W, Adams LG, Beckmen K, Blake JE, Checkley SL, Davison T, Di Francesco J, Elkin B, et al. (2020) Novel insights into serodiagnosis and epidemiology of Erysipelothrix rhusiopathiae, a newly recognized pathogen in muskoxen (Ovibos moschatus). PLOS ONE 15: e0231724.

Maxwell JA, Mittapalli K (2010) Realism as a stance for mixed methods research. In: Tashakkori A, Teddlie C, eds. SAGE Handbook of Mixed Methods in Social & Behavioral Research, 2nd Edition. SAGE Publications, Thousand Oaks, USA, pp 145–168.

McCoy JC, Ditchkoff SS (2012) Patterns of fecal hormones in a fenced population of white-tailed deer. Wildl Soc Bull 36: 641–646.

McCrum-Gardner E (2008) Which is the correct statistical test to use? Br J Oral Maxillofac Surg 46: 38–41.

McEwen BS, Wingfield JC (2003) The concept of allostasis in biology and biomedicine. Horm Behav 43: 2– 15.

McLaren MA, Green JE (1985) The reactions of muskoxen to snowmobile harassment. ARCTIC 188–193.

Medzhitov R, Schneider DS, Miguel P. Soares (2012) Disease tolerance as a defense strategy. Science 335.

Meyer JS, Novak MA (2012) Minireview: Hair cortisol: A novel biomarker of hypothalamic-pituitary- adrenocortical activity. Endocrinology 153: 4120–4127.

Millspaugh JJ, Washburn BE, Milanick MA (2002) Non-invasive techniques for stress assessment in white- tailed deer. Wildl Soc Bull 30: 899–907.

Millspaugh JJ, Woods RJ, Hunt KE, Raedeke KJ, Brundige GC, Washburn BE, Wasser SK (2001) Fecal glucocorticoid assays and the physiological stress response in elk. Wildl Soc Bull 899–907.

Mislan P, Derocher AE, St. Louis VL, Richardson E, Lunn NJ, Janz DM (2016) Assessing stress in Western Hudson Bay polar bears using hair cortisol concentration as a biomarker. Ecol Indic 71: 47–54.

142

Moore IT, Greene MJ, Lerner DT, Asher CE, Krohmer RW, Hess DL, Whittier J, Mason RT (2005) Physiological evidence for reproductive suppression in the introduced population of brown tree snakes (Boiga irregularis) on Guam. Biol Conserv 121: 91–98.

Mooring MS, Patton ML, Lance VA, Hall BM, Schaad EW, Fetter GA, Fortin SS, McPeak KM (2006) Glucocorticoids of bison bulls in relation to social status. Horm Behav 49: 369–375.

Morrow CJ, Kolver ES, Verkerk GA, Matthews LR (2002) Fecal glucocorticoid metabolites as a measure of adrenal activity in dairy cattle. Gen Comp Endocrinol 126: 229–241.

Mosbacher JB, Michelsen A, Stelvig M, Hendrichsen DK, Schmidt NM (2016) Show me your rump hair and I will tell you what you ate – The dietary history of muskoxen (Ovibos moschatus) revealed by sequential stable isotope analysis of guard hairs. PLOS ONE 11: e0152874.

Möstl E, Maggs JL, Schrötter G, Besenfelder U, Palme R (2002) Measurement of cortisol metabolites in faeces of ruminants. Vet Res Commun 26: 127–139.

Möstl E, Messmann S, Bagu E, Robia C, Palme R (1999) Measurement of glucocorticoid metabolite concentrations in faeces of domestic livestock. Transbound Emerg Dis 46: 621–631.

Möstl E, Palme R (2002) Hormones as indicators of stress. Domest Anim Endocrinol 23: 67–74.

Möstl E, Rettenbacher S, Palme R (2005) Measurement of corticosterone metabolites in birds’ droppings: An analytical approach. Ann N Y Acad Sci 1046: 17–34.

Nagy JA, Gunn A (2009) Productivity of Peary Caribou and Muskoxen on Banks and Melville Islands, NT, July 2004 (Manuscript Report No. 204). Department of Environment and Natural Resources, Government of the Northwest Territories, Canada, p 25.

Nagy JA, Gunn A, Wright WH (2009a) Population Estimates for Peary Caribou (Minto Inlet Herd), Dolphin and Union Caribou, and Muskox on Northwest Victoria Island, NT, July 2005 (Manuscript Report No. 203). Department of Environment and Natural Resources, Government of the Northwest Territories, Canada, p 39.

Nagy JA, Larter N, Wright WH (2006) Population Estimates for Peary Caribou and Muskox on Banks Island, Northwest Territories, July 2001. (Manuscript Report No. 199). Government of the Northwest Territories, Department of Environment and Natural Resources, Canada, p 47.

Nagy JA, Wright WH, Larter NC (2009b) Population Estimates for Peary Caribou (Minto Inlet Herd), Dolphin and Union Caribou, and Muskox on Northwest Victoria Island, NT, July 2001 (Manuscript Report No. 202). Department of Environment and Natural Resources, Government of the Northwest Territories, Canada, p 36.

Nielsen K, Kelly L, Gall D, Smith P, Bosse J, Nicoletti P, Kelly W (1994) The use of divalent cation chelating agents (EDTA/EGTA) to reduce non-specific serum protein interaction in enzyme immunoassay. Vet Res Commun 18: 433–437.

Nielsen K, Smith P, Yu W, Nicoletti P, Elzer P, Vigliocco A, Silva P, Bermudez R, Renteria T, Moreno F, et al. (2004) Enzyme immunoassay for the diagnosis of brucellosis: chimeric Protein A–Protein G as a common enzyme labeled detection reagent for sera for different animal species. Vet Microbiol 101: 123–129.

Nuttall M, Berkes F, Forbes B, Kofinas G, Vlassova T, Wenzel G (2005) Hunting, herding, fishing, and gathering: Indigenous peoples and renewable resource use in the Arctic. In: Symon C, Arris L, Heal B, eds. Arctic Climate Impact Assessment—Scientific Report. Cambridge University Press, Cambridge, UK, pp 649–690.

143

Nymo IH, Godfroid J, Åsbakk K, Larsen AK, das Neves CG, Rødven R, Tryland M (2013) A protein A/G indirect enzyme-linked immunosorbent assay for the detection of anti-Brucella antibodies in Arctic wildlife. J Vet Diagn Invest 25: 369–375.

Oksanen A, Nieminen M, Soveri T, Kumpula K, Heiskari U, Kuloharju V (1990) The establishment of parasites in reindeer calves. Rangifer 10: 20.

Opriessnig T, Forde T, Shimoji Y (2020) Erysipelothrix spp.: Past, present, and future directions in vaccine research. Front Vet Sci 7: 174.

Otten W, Heimbürge S, Kanitz E, Tuchscherer A (2020) It’s getting hairy – External contamination may affect the validity of hair cortisol as an indicator of stress in pigs and cattle. Gen Comp Endocrinol 295: 113531.

Overland J, Hanna E, Hanssen-Bauer I, Kim S, Walsh J, Wang M, Bhatt US, Thoman R (2016) Surface air temperature. In: Richter-Menge J, Overland JE, Mathis J, eds. Arctic Report Card 2016. pp 18–24.

Özkan Gülzari Ş, Jørgensen GHM, Eilertsen SM, Hansen I, Hagen SB, Fløystad I, Palme R (2019) Measuring faecal glucocorticoid metabolites to assess adrenocortical activity in reindeer. Animals 9: 987.

Packer C, Holt RD, Hudson PJ, Lafferty KD, Dobson AP (2003) Keeping the herds healthy and alert: implications of predator control for infectious disease. Ecol Lett 6: 797–802.

Palme R (2019) Non-invasive measurement of glucocorticoids: Advances and problems. Physiol Behav 199: 229–243.

Palme R, Fischer P, Schildorfer H, Ismail MN (1996) Excretion of infused 14C-steroid hormones via faeces and urine in domestic livestock. Anim Reprod Sci 43: 43–63.

Palme R, Rettenbacher S, Touma C, El-Bahr SM, Möstl E (2005) Stress hormones in mammals and birds: Comparative aspects regarding metabolism, excretion, and noninvasive measurement in fecal samples. Ann N Y Acad Sci 1040: 162–171.

Palme R, Robia C, Messmann S, Hofer J, Möstl E (1999) Measure of faecal cortisol metabolites in ruminants: A non-invasive parameter for adrenal function. Wien Tierärztl Monatsschrift 86: 237–241.

Parker K, Rainey W (2004) The adrenal glands. In: Griffin J, Ojeda S, eds. Textbook of Endocrine Physiology, 5th Edition. Oxford University Press, Oxford, UK, pp 319–348.

Patyk KA, Duncan C, Nol P, Sonne C, Laidre K, Obbard M, Wiig Ø, Aars J, Regehr E, Gustafson LL, et al. (2015) Establishing a definition of polar bear (Ursus maritimus) health: A guide to research and management activities. Sci Total Environ 514: 371–378.

Peacock SJ, Mavrot F, Tomaselli M, Hanke A, Fenton H, Nathoo R, Aleuy OA, Di Francesco J, Aguilar XF, Jutha N, et al. (2020) Linking co-monitoring to co-management: Bringing together local, traditional, and scientific knowledge in a wildlife status assessment framework. Arct Sci 1–20.

Périquet S, Richardson P, Cameron EZ, Ganswindt A, Belton L, Loubser E, Dalerum F (2017) Effects of lions on behaviour and endocrine stress in plains zebras. Ethology 123: 667–674.

Plummer M (2019) Rjags: Bayesian Graphical Models Using MCMC.

Pokharel SS, Seshagiri PB, Sukumar R (2017) Assessment of season-dependent body condition scores in relation to faecal glucocorticoid metabolites in free-ranging Asian elephants. Conserv Physiol 5: cox039.

144

Polfus JL, Manseau M, Simmons D, Neyelle M, Bayha W, Andrew F, Andrew L, Klütsch CFC, Rice K, Wilson P (2016) Łeghágots’enetę (learning together): The importance of indigenous perspectives in the identification of biological variation. Ecol Soc 21: 18.

Post E, Bhatt US, Bitz CM, Brodie JF, Fulton TL, Hebblewhite M, Kerby J, Kutz SJ, Stirling I, Walker DA (2013) Ecological consequences of sea-ice decline. Science 341: 519–524.

Potratz EJ, Brown JS, Gallo T, Anchor C, Santymire RM (2019) Effects of demography and urbanization on stress and body condition in urban white-tailed deer. Urban Ecosyst 22: 807–816.

Prewer E, Kutz S, Leclerc LM, Kyle CJ (2020) Already at the bottom? Demographic declines are unlikely further to undermine genetic diversity of a large Arctic ungulate: muskox, Ovibos moschatus (Artiodactyla: Bovidae). Biol J Linn Soc 129: 459–469.

Pride RE (2005) High faecal glucocorticoid levels predict mortality in ring-tailed lemurs (Lemur catta). Biol Lett 1: 60–63.

Prno J, Bradshaw B, Wandel J, Pearce T, Smit B, Tozer L (2011) Community vulnerability to climate change in the context of other exposure-sensitivities in Kugluktuk, Nunavut. Polar Res 30: 7363.

Pyziel AM, Kowalczyk R, Demiaszkiewicz AW (2011) The annual cycle of shedding Eimeria oocysts by European bison (Bison bonasus) in the Bialowieza Primeval Forest, Poland. J Parasitol 97: 737–739.

QSR International (2018) NVivo. QSR International, Burlington, USA.

R Core Team (2019) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria.

Raberg L, Graham AL, Read AF (2009) Decomposing health: Tolerance and resistance to parasites in animals. Philos Trans R Soc B Biol Sci 364: 37–49.

Rakotoniaina JH, Kappeler PM, Kaesler E, Hämäläinen AM, Kirschbaum C, Kraus C (2017) Hair cortisol concentrations correlate negatively with survival in a wild primate population. BMC Ecol 17: 30.

Ranglack DH, Neuman-Lee LA, French SS, du Toit JT (2016) Considerations of context and scale when using fecal glucocorticoids to indicate stress in large mammals: A study of wild American plains bison. PeerJ Prepr 4: e2092v1.

Raponi M, Beresford DV, Schaefer JA, Thompson ID, Wiebe PA, Rodgers AR, Fryxell JM (2018) Biting flies and activity of caribou in the boreal forest. J Wildl Manag 82: 833–839.

Raul J-S, Cirimele V, Ludes B, Kintz P (2004) Detection of physiological concentrations of cortisol and cortisone in human hair. Clin Biochem 37: 1105–1111.

Rehnus M, Wehrle M, Palme R (2014) Mountain hares Lepus timidus and tourism: Stress events and reactions. J Appl Ecol 51: 6–12.

Restif O, Koella JC (2004) Concurrent evolution of resistance and tolerance to pathogens. Am Nat 164: E90–E102.

Reyes E, Bubenik GA, Lobos A, Schams D, Bartos L (1997) Seasonal levels of cortisol, IGF-1 and triiodothyronine in adult male pudu (Pudu puda). Folia Zool 46: 109–116.

Rhodes ME, Rubin RT (1999) Functional sex differences (‘sexual diergism’) of central nervous system cholinergic systems, vasopressin, and hypothalamic–pituitary–adrenal axis activity in mammals: A selective review. Brain Res Rev 30: 135–152.

145

Ribera d’Alcalà M (2019) Similarities, differences and mechanisms of climate impact on terrestrial vs. marine ecosystems. Nat Conserv 34: 505–523.

Richena M, Rezende CA (2016) Morphological degradation of human hair cuticle due to simulated sunlight irradiation and washing. J Photochem Photobiol B 161: 430–440.

Riseth JÅ, Tømmervik H, Helander-Renvall E, Labba N, Johansson C, Malnes E, Bjerke JW, Jonsson C, Pohjola V, Sarri L-E, et al. (2011) Sámi traditional ecological knowledge as a guide to science: Snow, ice and reindeer pasture facing climate change. Polar Rec 47: 202–217.

Robertson MA (2000) Maximizing Qiviut Growth in Muskoxen (Master’s thesis). University of Alaska Fairbanks, Fairbanks, Alaska, USA.

Robinson CJ, Wallington TJ (2012) Boundary work: Engaging knowledge systems in co-management of feral animals on Indigenous lands. Ecol Soc 17: art16.

Roller MR, Lavrakas PJ (2015) Applied Qualitative Research Design: A Total Quality Framework Approach. The Guilford Press, New York, USA.

Romero LM (2004) Physiological stress in ecology: Lessons from biomedical research. Trends Ecol Evol 19: 249–255.

Romero LM, Butler LK (2007) Endocrinology of stress. Int J Comp Psychol 20: 89–95.

Romero LM, Dickens MJ, Cyr NE (2009) The reactive scope model — A new model integrating homeostasis, allostasis, and stress. Horm Behav 55: 375–389.

Romero LM, Wingfield JC (2016) Tempests, Poxes, Predators, and People: Stress in Wild Animals and How They Cope. Oxford University Press, Oxford, UK.

Rowell JE, Lupton CJ, Robertson MA, Pfeiffer FA, Nagy JA, White RG (2001) Fiber characteristics of qiviut and guard hair from wild muskoxen (Ovibos moschatus). J Anim Sci 79: 1670.

Russell E, Kirschbaum C, Laudenslager ML, Stalder T, De Rijke Y, van Rossum EF, Van Uum S, Koren G (2015) Toward standardization of hair cortisol measurement: Results of the first international interlaboratory round robin. Ther Drug Monit 37: 71–75.

Russell E, Koren G, Rieder M, Van Uum S (2012) Hair cortisol as a biological marker of chronic stress: Current status, future directions and unanswered questions. Psychoneuroendocrinology 37: 589–601.

Russell E, Koren G, Rieder M, Van Uum SH (2014) The detection of cortisol in human sweat: Implications for measurement of cortisol in hair. Ther Drug Monit 36: 30–34.

Ryser-Degiorgis M-P (2013) Wildlife health investigations: Needs, challenges and recommendations. BMC Vet Res 9: 223.

Salaberger T, Millard M, Makarem SE, Möstl E, Grünberger V, Krametter-Frötscher R, Wittek T, Palme R (2016) Influence of external factors on hair cortisol concentrations. Gen Comp Endocrinol 233: 73– 78.

Salas M, Temple D, Abáigar T, Cuadrado M, Delclaux M, Enseñat C, Almagro V, Martínez-Nevado E, Quevedo MÁ, Carbajal A, et al. (2016) Aggressive behavior and hair cortisol levels in captive Dorcas gazelles (Gazella dorcas) as animal-based welfare indicators. Zoo Biol 35: 467–473.

Saldaña J (2013) The Coding Manual for Qualitative Researchers, 2nd Edition. SAGE Publications, Los Angeles, USA.

146

Santangeli A, Wistbacka R, Morosinotto C, Raulo A (2019) Hair cortisol concentration in Siberian flying squirrels is unrelated to landscape and social factors. Sci Nat 106: 29.

Santos JPV, Acevedo P, Carvalho J, Queirós J, Villamuelas M, Fonseca C, Gortázar C, López-Olvera JR, Vicente J (2018) The importance of intrinsic traits, environment and human activities in modulating stress levels in a wild ungulate. Ecol Indic 89: 706–715.

Sapolsky RM, Romero LM, Munck AU (2000) How do glucocorticoids influence stress responses? Integrating permissive, suppressive, stimulatory, and preparative actions. Endocr Rev 21: 55–89.

SARC (2013) Species Status Report for Dolphin and Union Caribou (Rangifer Tarandus Groenlandicus x Pearyi) in the Northwest Territories. Species at Risk Committee (SARC), Yellowknife, Canada, p 118.

SARC (2017) Species Status Report for Grizzly Bears (Ursus Arctos) in the Northwest Territories. Species at Risk Committee (SARC), Yellowknife, Canada, p 153.

Schell CJ, Young JK, Lonsdorf EV, Mateo JM, Santymire RM (2017) Investigation of techniques to measure cortisol and testosterone concentrations in coyote hair. Zoo Biol 36: 220–225.

Scheun J, Bennett NC, Ganswindt A, Nowack J (2015) The hustle and bustle of city life: Monitoring the effects of urbanisation in the African lesser bushbaby. Sci Nat 102: 57.

Schmidt NM, Grøndahl C, Evans AL, Desforges J-P, Blake J, Hansen LH, Beumer LT, Mosbacher JB, Stelvig M, Greunz EM, et al. (2020) On the interplay between hypothermia and reproduction in a high arctic ungulate. Sci Rep 10: 1514.

Schnarch B (2004) Ownership, control, access, and possession (OCAP) or self-determination applied to research: A critical analysis of contemporary First Nations research and some options for First Nations communities. J Aborig Health 1: 80–95.

Sefa Dei GJ, Hall BL, Rosenberg DG, eds. (2000) Indigenous Knowledges in Global Contexts: Multiple Readings of Our World. University of Toronto Press, Scholarly Publishing Division, Toronto, Canada.

Sergiel A, Cattet M, Kapronczai L, Janz DM, Selva N, Bartoń KA, Swenson JE, Zedrosser A (2020) Do follicles matter? Testing the effect of follicles on hair cortisol levels. Conserv Physiol 8: coaa003.

Shackleton C (2010) Clinical steroid mass spectrometry: A 45-year history culminating in HPLC–MS/MS becoming an essential tool for patient diagnosis. J Steroid Biochem Mol Biol 121: 481–490.

Sharpley CF, McFarlane JR, Slominski A (2012) Stress-linked cortisol concentrations in hair: What we know and what we need to know. Rev Neurosci 23: 111–121.

Shave JR, Derocher AE, Cherry SG, Thiemann GW (2019) Chronic stress and body condition of wolf- killed prey in Prince Albert National Park, Saskatchewan. Conserv Physiol 7: coz037.

Sheriff MJ, Dantzer B, Delehanty B, Palme R, Boonstra R (2011) Measuring stress in wildlife: Techniques for quantifying glucocorticoids. Oecologia 166: 869–887.

Sid-Ahmed O-E, Sanhouri A, Elwaseela B-E, Fadllalah I, Mohammed G-EE, Möstl E (2013) Assessment of adrenocortical activity by non-invasive measurement of faecal cortisol metabolites in dromedary camels (Camelus dromedarius). Trop Anim Health Prod 45: 1453–1458.

Skobowiat C, Slominski AT (2015) UVB activates hypothalamic–pituitary–adrenal axis in C57BL/6 mice. J Invest Dermatol 135: 1638–1648.

147

Slominski A, Wortsman J, Tuckey RC, Paus R (2007) Differential expression of HPA axis homolog in the skin. Mol Cell Endocrinol 265: 143–149.

Spiegelhalter DJ, Best NG, Carlin BP, van der Linde A (2002) Bayesian measures of model complexity and fit. J R Stat Soc Ser B Stat Methodol 64: 583–639.

Stankowich T (2008) Ungulate flight responses to human disturbance: A review and meta-analysis. Biol Conserv 141: 2159–2173.

Statistics Canada (2016) Census profile, 2016 Census, Kugluktuk, Hamlet, Nunavut.

Stephen C (2014) Toward a modernized definition of wildlife health. J Wildl Dis 50: 427–430.

Stephens C, Nettleton C, Porter J, Willis R, Clark S (2005) Indigenous peoples’ health—why are they behind everyone, everywhere? The Lancet 366: 10–13.

Strier KB, Lynch JW, Ziegler TE (2003) Hormonal changes during the mating and conception seasons of wild northern muriquis (Brachyteles arachnoides hypoxanthus). Am J Primatol 61: 85–99.

Stroud LR, Solomon C, Shenassa E, Papandonatos G, Niaura R, Lipsitt LP, LeWinn K, Buka SL (2007) Long-term stability of maternal prenatal steroid hormones from the National Collaborative Perinatal Project: Still valid after all these years. Psychoneuroendocrinology 32: 140–150.

Taherkhani M, Vitousek S, Barnard PL, Frazer N, Anderson TR, Fletcher CH (2020) Sea-level rise exponentially increases coastal flood frequency. Sci Rep 10: 6466.

Tarasuk V, Mitchell A (2020) Household Food Insecurity in Canada, 2017-18. Research to identify policy options to reduce food insecurity (PROOF), Toronto, Canada.

Taylor M, Coop R, Wall R, eds. (2016) Veterinary Parasitology, 4th Edition. Wiley Blackwell, Oxford, UK.

Terwissen CV, Mastromonaco GF, Murray DL (2013) Influence of adrenocorticotrophin hormone challenge and external factors (age, sex, and body region) on hair cortisol concentration in Canada lynx (Lynx canadensis). Gen Comp Endocrinol 194: 162–167.

Tomaselli M (2019) Improved Wildlife Health and Disease Surveillance through the Combined Use of Local Knowledge and Scientific Knowledge (PhD dissertation). Faculty of Veterinary Medicine, University of Calgary, Calgary, Alberta, Canada.

Tomaselli M, Dalton C, Duignan PJ, Kutz S, van der Meer F, Kafle P, Surujballi O, Turcotte C, Checkley S (2016a) Contagious ecthyma, rangiferine brucellosis, and lungworm infection in a muskox (Ovibos moschatus) from the Canadian Arctic, 2014. J Wildl Dis 52: 719–724.

Tomaselli M, Elkin B, Kutz S, Harms NJ, Nymo HI, Davison T, Leclerc L-M, Branigan M, Dumond M, Tryland M, et al. (2019) A transdisciplinary approach to Brucella in muskoxen of the western Canadian Arctic 1989–2016. EcoHealth 16: 488–501.

Tomaselli M, Gerlach SC, Kutz SJ, Checkley SL, The Community of Iqaluktutiaq (2018a) Iqaluktutiaq voices: Local perspectives about the importance of muskoxen, contemporary and traditional use and practices. ARCTIC 71: 1–14.

Tomaselli M, Kutz S, Checkley S (2016b) Local Knowledge from Ikaluktutiak, Nunavut, on Muskoxen and Dolphin and Union Caribou (Interim Report Prepared for the Kitikmeot Inuit Association, Ekaluktutiak Hunters and Trappers Organization, and Nunavut Research Institute).

148

Tomaselli M, Kutz S, Gerlach C, Checkley S (2018b) Local knowledge to enhance wildlife population health surveillance: Conserving muskoxen and caribou in the Canadian Arctic. Biol Conserv 217: 337–348.

Tomasini S (2018) Unpacking the Red List: Use (and Misuse?) of expertise, knowledge, and power. Conserv Soc 16: 505.

Touma C, Palme R (2005) Measuring fecal glucocorticoid metabolites in mammals and birds: The importance of validation. Ann N Y Acad Sci 1046: 54–74.

Toupin B, Huot J, Manseau M (1996) Effect of insect harassment on the behaviour of the Riviere George caribou. ARCTIC 375–382.

Trevisan C, Montillo M, Prandi A, Mkupasi EM, Ngowi HA, Johansen MV (2017) Hair cortisol and dehydroepiandrosterone concentrations in naturally Taenia solium infected pigs in Tanzania. Gen Comp Endocrinol 246: 23–28.

Turner A, Canny B, Hobbs R, Bond J, Clarke I, Tilbrook A (2002) Influence of sex and gonadal status of sheep on cortisol secretion in response to ACTH and on cortisol and LH secretion in response to stress: importance of different stressors. J Endocrinol 173: 113–122.

Usher PJ (2000) Traditional ecological knowledge in environmental assessment and management. ARCTIC 52: 183–193. van Beest FM, Milner JM (2013) Behavioural responses to thermal conditions affect seasonal mass change in a heat-sensitive northern ungulate. PLOS ONE 8: e65972.

Van Meter PE, French JA, Dloniak SM, Watts HE, Kolowski JM, Holekamp KE (2009) Fecal glucocorticoids reflect socio-ecological and anthropogenic stressors in the lives of wild spotted hyenas. Horm Behav 55: 329–337.

Verocai GG, Kutz SJ, Simard M, Hoberg EP (2014) Varestrongylus eleguneniensis sp. n. (Nematoda: Protostrongylidae): A widespread, multi-host lungworm of wild North American ungulates, with an emended diagnosis for the genus and explorations of biogeography. Parasit Vectors 7: 556.

Vors LS, Boyce MS (2009) Global declines of caribou and reindeer. Glob Change Biol 15: 2626–2633.

Walker DA, Raynolds MK, Daniëls FJA, Einarsson E, Elvebakk A, Gould WA, Katenin AE, Kholod SS, Markon CJ, Melnikov ES, et al. (2005) The Circumpolar Arctic vegetation map. J Veg Sci 16: 267– 282.

Wasser SK, Hunt KE, Brown JL, Cooper K, Crockett CM, Bechert U, Millspaugh JJ, Larson S, Monfort SL (2000) A generalized fecal glucocorticoid assay for use in a diverse array of nondomestic mammalian and avian species. Gen Comp Endocrinol 120: 260–275.

Weisser JJ, Hansen M, Björklund E, Sonne C, Dietz R, Styrishave B (2016) A novel method for analysing key corticosteroids in polar bear (Ursus maritimus) hair using liquid chromatography tandem mass spectrometry. J Chromatogr B 1017–1018: 45–51.

Weladji RB, Holand Ø, Almøy T (2003) Use of climatic data to assess the effect of insect harassment on the autumn weight of reindeer (Rangifer tarandus) calves. J Zool 260: 79–85.

Weller G (2005) Summary and synthesis of the Arctic Climate Impact Assessment (ACIA). In: Symon C, Arris L, Heal B, eds. Arctic Climate Impact Assessment—Scientific Report. Cambridge University Press, Cambridge, UK, pp 990–1020.

149

Wester VL (2016) Hair cortisol and cortisone are decreased by natural sunlight. Psychoneuroendocrinology 72: 94–96.

Wey TW, Lin L, Patton ML, Blumstein DT (2015) Stress hormone metabolites predict overwinter survival in yellow-bellied marmots. Acta Ethologica 18: 181–185.

White RG, Holleman DF, Wheat P, Tallas PG, Jourdan M (1984) Seasonal changes in voluntary intake and digestibility of diets by captive muskoxen. Biol Pap Univ Alsk Spec Rep 193–194.

Wikelski M, Cooke SJ (2006) Conservation physiology. Trends Ecol Evol 21: 38–46.

Wild MA, Hobbs NT, Graham MS, Miller MW (2011) The role of predation in disease control: A comparison of selective and nonselective removal on prion disease dynamics in deer. J Wildl Dis 47: 78–93.

Wilkinson PF (1974) Wool shedding in musk oxen. Biol J Linn Soc 6: 127–141.

Wilson S (2008) Research Is Ceremony: Indigenous Research Methods. Fernwood Publishing, Black Point, Canada.

Wingfield J, Hunt K, Breuner C, Dunlap K, Fowler G, Freed L, Lepson J (1997) Environmental stress, field endocrinology, and conservation biology. In: Clemmons JR, Buchholz R, eds. Behavioral Approaches to Conservation in the Wild, 1st Edition. Cambridge University Press, Cambridge, UK, pp 95–131.

Witter LA, Johnson CJ, Croft B, Gunn A, Gillingham MP (2012) Behavioural trade-offs in response to external stimuli: Time allocation of an Arctic ungulate during varying intensities of harassment by parasitic flies. J Anim Ecol 81: 284–295.

Wolf TE, Valades GB, Simelane P, Bennett NC, Ganswindt A (2018) The relationship between physical injury, body condition and stress-related hormone concentrations in free-ranging giraffes. Wildl Biol 2018.

Wrona FJ, Prowse TD, Reist JD (2005) Freshwater ecosystems and fisheries. In: Symon C, Arris L, Heal B, eds. Arctic Climate Impact Assessment—Scientific Report. Cambridge University Press, Cambridge, UK, pp 353–452.

Yamanashi Y, Morimura N, Mori Y, Hayashi M, Suzuki J (2013) Cortisol analysis of hair of captive chimpanzees (Pan troglodytes). Gen Comp Endocrinol 194: 55–63.

Yamanashi Y, Teramoto M, Morimura N, Hirata S, Inoue-Murayama M, Idani G (2016a) Effects of relocation and individual and environmental factors on the long-term stress levels in captive chimpanzees (Pan troglodytes): Monitoring hair cortisol and behaviors. PLOS ONE 11: e0160029.

Yamanashi Y, Teramoto M, Morimura N, Hirata S, Suzuki J, Hayashi M, Kinoshita K, Murayama M, Idani G (2016b) Analysis of hair cortisol levels in captive chimpanzees: Effect of various methods on cortisol stability and variability. MethodsX 3: 110–117.

Ytrehus B, Bretten T, Bergsjø B, Isaksen K (2008) Fatal pneumonia epizootic in musk ox (Ovibos moschatus) in a period of extraordinary weather conditions. EcoHealth 5: 213–223.

Ytrehus B, Davidson RK, Isaksen K (2015) Single causative factor for severe pneumonia epizootics in muskoxen? EcoHealth 12: 395–397.

Zbyryt A, Bubnicki JW, Kuijper DPJ, Dehnhard M, Churski M, Schmidt K (2018) Do wild ungulates experience higher stress with humans than with large carnivores? Behav Ecol 29: 19–30.

150

Zwijacz-Kozica T, Selva N, Barja I, Silván G, Martínez-Fernández L, Illera JC, Jodłowski M (2013) Concentration of fecal cortisol metabolites in chamois in relation to tourist pressure in Tatra National Park (South Poland). Acta Theriol (Warsz) 58: 215–222.

151

APPENDIX A Relationship between qiviut cortisol and fecal glucocorticoid metabolites

1. Introduction

Multiple studies have assessed the relationship between hair glucocorticoids (HGCs) and the GC levels measured in other matrices, such as blood, saliva, urine, and feces (reviewed in Burnard et al., 2017; Kalliokoski et al., 2019). Results from a recent meta-analysis including both human and animal studies showed that HGCs correlate most strongly with fecal glucocorticoid metabolite (FGM) concentrations (summary correlation coefficient of 0.56 (95% confidence interval = 0.21–0.78)), which is likely due to the greater timeframe of several hours to days reflected in the feces (Kalliokoski et al., 2019). Additionally, the correlation between HGCs and FGMs generally increases with the number of fecal samples collected during the corresponding period of hair growth (Burnard et al., 2017). Using both multiple fecal samples collected during the corresponding period of hair growth in captive muskoxen and single point samples from wild muskoxen, we assessed the strength of the correlation between FGMs and qiviut cortisol.

2. Material and methods

2.1. Captive muskoxen

A repeated ACTH challenge was done during the summer of 2018. Seven animals received weekly intramuscular injections in the shoulder or hip of a long-term release gel formulation of ACTH (Corticotrophin, Wedgewood Pharmacy, Swedesboro, NJ, USA) at a dosage of 2 IU/kg over a period of five weeks, while three control animals were administered an equivalent volume of physiological saline (0.9% of sodium chloride). Hair samples were collected while the muskoxen were restrained in the chute. A square patch of hair was shaved at time 0 (t0 – first injection) from the neck, shoulder, and rump on the same side using electric clippers and these patches were re-shaved entirely at t0 + 6 weeks (t0 + 6W – 2 weeks after fifth and last injection). The clipper blade was cleaned between each animal using a brush. Hair samples were placed in paper envelopes, which were left open during several hours to allow the samples to air-dry if they were moist, and subsequently closed and stored at room temperature for 12–14 months until the samples were analyzed. Fecal samples were collected in their entirety, non-invasively, and with minimal interactions with the muskoxen. Animals were observed from a distance until they defecated voluntarily. They were then approached slowly in order to collect the fecal sample with gloves from the ground. These were immediately placed into a Whirlpack® and then a cooler with icepacks for a maximum of 4 h, before being stored frozen at -20°C. Fecal samples were collected three times a week during the ACTH challenge: the day before each

152 injection, and again at 24 and 48 h post-injection. Samples were shipped frozen to the Toronto Zoo for FGM analysis, no later than 3 months’ after their collection. All hormone analyses and enzyme immunoassay validations were done at the Endocrinology Laboratory of the Toronto Zoo. These are detailed in Chapter 4 section 4.3.3 for FGMs and Chapter 5 section 5.3.6.1 for qiviut cortisol. We averaged for each week the FGM levels over the three sampling days, and further summarized the FGM levels over the period of the ACTH challenge using the area under the curve with respect to ground (AUCg) (Pruessner et al., 2003). We assessed the correlation between the cortisol levels in qiviut grown during the summer ACTH challenge (between t0 and t0 + 6W) on the neck, shoulder, or rump and both the AUCg and the mean FGM concentration using Pearson’s correlation coefficient (r) as these were all normally distributed.

2.2. Wild muskoxen

Fecal and hair samples were collected between January 2016 and May 2019 from 198 harvested muskoxen as part of a regional hunter-based muskox health monitoring program (see Chapter 6 section 6.3.1 for details). We assessed the correlation between qiviut cortisol levels and FGM concentrations using

Spearman’s rank correlation coefficient (rs) as neither variable was normally distributed.

3. Results

Qiviut cortisol levels in the neck, shoulder, and rump of captive muskoxen were not significantly correlated with mean FGM concentrations (neck: r = 0.24, p = 0.50; shoulder: r = -0.15, p = 0.67; rump: r = 0.28, p = 0.44) or with the AUCg (neck: r = 0.31, p = 0.39; shoulder: r = -0.14, p = 0.70; rump: r = 0.29, p = 0.41). Qiviut cortisol levels in wild muskoxen were not significantly correlated with FGM concentrations

(rs = -0.08, p = 0.24).

4. Discussion

Despite finding a non-significant correlation between qiviut cortisol and FGMs, as expected, there was a trend for an increased correlation when multiple fecal samples were collected over the period of corresponding hair growth than with single point fecal samples. Interestingly, FGMs were negatively correlated with shoulder qiviut cortisol, whereas the correlation was positive with the rump and neck qiviut. These differences may be due to the higher intra-sample variability in the shoulder (see Chapter 5). Similar non-significant correlations between FGMs and HGCs were found in other studies in which multiple fecal samples were collected periodically to three times per week during the period of corresponding hair growth (Bryan et al., 2013; Moya et al., 2013; Yamanashi et al., 2013; Mastromonaco et al., 2014; Tallo-Parra et al., 2015; Carlsson et al., 2016). In the study on captive muskoxen, we collected three samples per week, but this number is still limited, and the FGM levels probably do not represent accurately

153 the total hormone output over the period of hair growth, which explains the positive but non-significant correlation measured. This would have been even more the case for the single point samples in wild muskoxen, which is consistent with the almost null correlation measured.

References

Bryan HM, Adams AG, Invik RM, Wynne-Edwards KE, Smits JE (2013) Hair as a meaningful measure of baseline cortisol levels over time in dogs. J Am Assoc Lab Anim Sci 52: 189–196.

Carlsson AM, Mastromonaco G, Vandervalk E, Kutz S (2016) Parasites, stress and reindeer: Infection with abomasal nematodes is not associated with elevated glucocorticoid levels in hair or faeces. Conserv Physiol 4: cow058.

Moya D, Schwartzkopf-Genswein KS, Veira DM (2013) Standardization of a non-invasive methodology to measure cortisol in hair of beef cattle. Livest Sci 158: 138–144.

Pruessner JC, Kirschbaum C, Meinlschmid G, Hellhammer DH (2003) Two formulas for computation of the area under the curve represent measures of total hormone concentration versus time- dependent change. Psychoneuroendocrinology 28: 916–931.

Tallo-Parra O, Manteca X, Sabes-Alsina M, Carbajal A, Lopez-Bejar M (2015) Hair cortisol detection in dairy cattle by using EIA: Protocol validation and correlation with faecal cortisol metabolites. Animal 9: 1059–1064.

Yamanashi Y, Morimura N, Mori Y, Hayashi M, Suzuki J (2013) Cortisol analysis of hair of captive chimpanzees (Pan troglodytes). Gen Comp Endocrinol 194: 55–63.

154

APPENDIX B Longitudinal study of the fecal glucocorticoid metabolite response to repeated adrenocorticotropic hormone/saline injections

The objective of this study was to determine whether muskoxen exhibited a similar response in fecal glucocorticoid metabolites (FGMs) to each of five weekly adrenocorticotropic hormone (ACTH) injections.

1. Material and methods

Beginning on July 23rd, 2018, seven muskoxen (Table B.1) received five weekly intramuscular (IM) injections in the shoulder or hip of 2 IU/kg of Corticotrophin (Wedgewood Pharmacy, Swedesboro, NJ, USA, a long-term release gel formulation of ACTH; concentration of 80 IU/ml). Three control muskoxen (Table B.1) were administered an equivalent volume of physiological saline (0.9% of sodium chloride), IM, over the same five-week period. All injections were given between 9:00 AM and 12:00 PM every week.

Table B. 1. Identification (ID), sex, age, and experimental group of the animals included in the repeated pharmacological challenge.

Animal ID Sex Age (years) Experimental group MX-740 F 1.25 Control MX-596 F 3.25+ Control MX-283 F 6.25 Control MX-738 M 1.25 ACTH MX-741 M 1.25 ACTH MX-619 M 2.25 ACTH MX-621 F 2.25 ACTH MX-597 F 3.25+ ACTH MX-1169 F 7.25+ ACTH MX-247 F 11.25+ ACTH +Immobilized at the time of the 4th ACTH/saline injection (see text for explanation).

Feces were collected as described in Chapter 4 section 4.3.2 from the seven ACTH-injected and three control animals three times a week during the entire challenge: the day before each ACTH/saline injection (referred to as time 0), and again at approximately 24 and 48 h post-injection. These sampling times were chosen based on the approximate peak in FGM levels determined during the winter challenge (see Chapter 4) and considering the faster gut transit time of muskoxen in the summer (Adamczewski et al., 1994). The animals were part of a broader study that involved hair sampling at the time of the first injection and two weeks following the last injection (see Chapter 5). At the time of the 4th injection and unrelated to the ACTH challenge, four females (Table B.1) were immobilized (4–5 mg/kg ketamine, 0.1 mg/kg xylazine, and 0.1 mg/kg azaperone administered IM, and a reversal dose of 1 mg/kg tolazoline IM) to insert an intravaginal device used for estrus synchronization.

155

All hormone analyses were done at the Endocrinology Laboratory of the Toronto Zoo. A corticosterone enzyme immunoassay (EIA) was used to quantify FGMs as described in Chapter 4 section 4.3.3. “Fecal corticosterone” is used to refer to the FGMs detected by the corticosterone EIA.

All statistical analyses were performed using the R software version 3.4.4 (R Core Team, 2019). The significance level was set at p < 0.05. Linear mixed-effect models were fit using the lme function from the nlme package to test the effects of experimental group (ACTH or control), time of fecal sample collection (time 0, 24 h, and 48 h post-injection) and injection number (1–5), as well as their two-by-two and three- way interactions, on fecal corticosterone. We chose to treat the injection number as a continuous rather than as an ordinal variable because the number of observations in our dataset was small and we were more interested in the effect of an additional injection than in comparing specific injections. To account for the clustering effect of repeated sampling of individual animals, animal identity was fitted as a random effect. Fecal corticosterone was logarithmically transformed to satisfy the linear mixed model assumptions of normal distribution and homoscedasticity. Model simplification was performed by manual backward elimination from a maximal model including all variables and interactions that were significant at the p-value cut-off level of 0.25 in univariate analyses. Model comparison was done using likelihood ratio tests, where lme models were fitted by maximum likelihood. To help choose the best model, all possible combinations of fixed effects and two- by-two and three-way interactions were fit and models were compared using the adaptation of Akaike Information Criterion for small sample sizes (AICc) since the ratio of (sample size)/(number of parameters) was small (Burnham and Anderson, 2002). The top model was finally re-fit using restricted maximum likelihood to obtain the final parameter estimates. Linear mixed model assumptions were assessed by review of the residual plots for the top model both at the observation and animal (random effect) levels. Marginal (fixed effects only) and conditional (fixed and random effects) coefficients of determination (R2) were calculated using the MuMIn package (Bartoń, 2015).

2. Results

Feces were collected the day before each injection (time 0) and 24 (range = 19.72–24.87 h) and 48 h (range = 44.30–49.50 h) post-injection. The fecal corticosterone levels following each of five weekly injections are summarized in Table B.2 by experimental group and time of fecal sample collection.

156

Table B. 2. Median (range) fecal corticosterone (ng/g wet feces) following each of five weekly injections.

Injection Time 0 24 h post-injection 48 h post-injection Control animals (n = 3) 1 9.36 (7.22–11.30) 5.26 (5.15–7.17) 9.04 (4.74–14.44) 2 9.76 (7.51–13.78) 9.97 (6.93–13.21) 6.99 (4.50–9.16) 3 4.23 (2.62–9.85) 8.35 (6.60–14.48) 5.46 (4.30–11.72) 4 6.94 (6.42–9.74) 11.11 (5.53–26.31) 26.51 (4.65–65.34) 5 7.43 (5.87–16.47) 6.15 (5.50–14.05) 15.74 (4.92–17.01) ACTH-injected animals (n = 7) 1 10.15 (4.58–15.22) 6.02 (3.70–8.68) 6.49 (4.60–31.99) 2 10.48 (2.91–37.11) 11.61 (10.07–17.37) 6.35(3.01–9.41) 3 6.72 (3.30–17.11) 8.28 (5.56–38.51) 12.88 (7.23–16.57) 4 10.43 (2.41–35.36) 21.58 (3.46–73.15) 24.28 (5.17–112.89) 5 12.48 (8.30–64.88) 15.11 (10.42–62.86) 15.42 (2.32–25.93)

We tested the effects of experimental group (group: ACTH or control), time of fecal sample collection (time_post_inj: time 0, 24 h, and 48 h post-injection) and injection number (injection), as well as their two-by-two and three-way interactions on fecal corticosterone. The model with best fit included injection number. However, this model was not significantly different from the models with injection number and experimental group or with injection number, experimental group and their interaction term (i.e., ΔAICc < 2; Table B.3). Parameter estimates for these three models can be found in Table B.4. The best-fit model conformed to the assumptions of normality and constant variance both at the observation and animal levels. The marginal and conditional R² were 0.09 and 0.25, respectively.

Table B. 3. Comparison of the linear mixed-effect models fit for fecal corticosterone with animal identity as a random effect, and all possible combinations of explanatory variables (experimental group (group), time of fecal sample collection (time_post_inj), and injection number (injection)) and two-by-two and three-way interactions, with their corresponding AICc, ΔAICc in comparison to the best-fit model (bold), and degrees of freedom (DF).

DF AICc ΔAICc injection 4 312.37 0.00 injection, group, injection:group 6 313.12 0.75 injection, group 5 313.13 0.76 injection, time_post_inj 6 315.22 2.85 injection, group, time_post_inj, injection:group 8 316.07 3.70 injection, time_post_inj, injection:time_post_inj 8 316.72 4.35 injection, group, time_post_inj, injection:time_post_inj 9 317.61 5.24 injection, group, time_post_inj, injection:group, injection:time_post_inj 10 317.66 5.29 injection, group, time_post_inj, group:time_post_inj 9 320.06 7.69 injection, group, time_post_inj, injection:group, group:time_post_inj 10 320.15 7.78 injection, group, time_post_inj, injection:time_post_inj, group:time_post_inj 11 321.74 9.37 injection, group, time_post_inj, injection:group, injection:time_post_inj, 12 321.85 9.48 group:time_post_inj injection, group, time_post_inj, injection:group, injection:time_post_inj, 14 326.16 13.79 group:time_post_inj, injection:group:time_post_inj null model (i.e., with intercept only) 3 326.32 13.95 group 4 327.05 14.68 time_post_inj 5 329.27 16.90 group, time_post_inj 6 330.06 17.69 group, time_post_inj, group:time_post_inj 8 334.07 21.70

157

Table B. 4. Comparison of the back-transformed coefficients (coef) and their associated 95% confidence intervals (CIs) for the top three models explaining fecal corticosterone and including as explanatory variables: (i) experimental group (group; the reference category is control), injection number (injection), and their interaction term, (ii) experimental group and injection number, or (iii) injection number only.

group, injection, group, injection injection group:injection Coef 95% CI Coef 95% CI Coef 95% CI Intercept 7.03 4.17–11.87 5.49 3.65–8.24 6.55 4.83–8.88 Injection 1.07 0.94–1.22 1.16 1.08–1.25 1.16 1.08–1.25 Group = ACTH 0.90 0.44–1.87 1.29 0.80–2.08 - - Group ACTH:injection 1.13 0.96–1.32 - - - -

Fecal corticosterone levels were more variable at 24 and 48 h post-injection n°4, when some of the females were immobilized (n = 3/7 ACTH-injected and n = 1/3 control), than after the other injections (Table B.2 and Figure B.1). Immobilization has been shown to affect the FGM response in ACTH- and saline-injected red deer (Cervus elaphus) (Huber et al., 2003) and to elicit a FGM response in adult African buffalo (Syncerus caffer) (Spaan et al., 2017). We, therefore, re-fit the best-fit model without these data points to check their influence. However, none of the coefficients were significantly modified and our conclusions remained unchanged (Table B.5), so we kept them in the model.

Table B. 5. Comparison of the back-transformed coefficients and their associated 95% CIs for the best- fit model explaining fecal corticosterone fit with and without the injection n°4 data points of the 24 and 48 h sampling for the muskoxen that were immobilized at that time (n = 3/7 ACTH-injected and n = 1/3 control).

Without data during All data immobilization Coef 95% CI Coef 95% CI Intercept 6.55 4.83–8.88 6.76 5.10–8.95 Injection 1.16 1.08–1.25 1.13 1.05–1.21

158

3. Discussion

The response in FGMs to a repeated ACTH challenge has not previously been reported in the literature. We detected a significant increase in fecal corticosterone levels with each additional injection. We are, unfortunately, limited in the interpretation of these findings as we did not have a large enough sample

159 size and sufficient power to properly assess the effects of experimental group, time of fecal sample collection, injection number, and their various interactions on fecal corticosterone levels. Additionally, because we based the timing of fecal collections on the results from the winter ACTH challenge, we very likely missed the peak fecal corticosterone levels with the 24 and 48 h sampling. Based on the intensive sampling done after the first injection in the summer (Chapter 4), this peak likely occurred around 7 h with a return to time 0 levels by 22 h and we were, therefore, not able to compare the magnitude and duration of the response between the successive ACTH injections. Furthers studies would be necessary to confirm our findings and more comprehensively describe the adrenal response to consecutive ACTH injections.

References

Adamczewski JZ, Flood PF, Chaplin RK, Schaefer JA (1994) Seasonal variation in intake and digestion of a high-roughage diet by muskoxen. Can J Anim Sci 74: 305–313.

Bartoń K (2015) MuMIn Package.

Burnham KP, Anderson DR (2002) Model Selection and Multimodel Inference: A Practical Information- Theoretic Approach, 2nd edition. Springer, New York, USA.

Huber S, Palme R, Zenker W, Mostl E (2003) Non-invasive monitoring of the adrenocortical response in red deer. J Wildl Manag 67: 258.

R Core Team (2019) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria.

Spaan JM, Pitts N, Buss P, Beechler B, Ezenwa VO, Jolles AE (2017) Noninvasive measures of stress response in African buffalo (Syncerus caffer) reveal an age-dependent stress response to immobilization. J Mammal 98: 1288–1300.

160

APPENDIX C Comparison and main differences between the two methods used to measure qiviut cortisol levels

In this thesis, we first used liquid chromatography coupled with tandem mass spectrometry (LC- MS/MS) to measure qiviut cortisol levels (Chapter 2). However, the LC-MS/MS laboratory closed when we were halfway through the analyses, and we subsequently collaborated with the Endocrinology Laboratory of the Toronto Zoo, which was already carrying out the hormone analyses for the fecal samples, to develop an alternative method using a cortisol enzyme immunoassay (EIA) to measure qiviut cortisol levels (Chapters 5 and 6). The main differences between the two methods are summarized in Table C.1. As an initial check, we compared the qiviut cortisol levels measured with the cortisol EIA and LC- MS/MS following the initial cold wash and extraction procedure, which gave us a strong correlation (Pearson’s correlation coefficient (r) = 0.97, p < 0.001, n = 10). Using the standard sample preparation protocol from the Toronto Zoo, qiviut cortisol levels measured with the cortisol EIA were positively, but not significantly correlated to those measured using the LC-MS/MS method (Spearman’s correlation coefficient (rs) = 0.45, p = 0.19) or to those measured using the cold wash and extraction procedure from the LC-MS/MS method coupled with the cortisol quantitation with the EIA (rs = 0.49, p = 0.15). For consistency, we consequently reported in Chapters 5 and 6 only the qiviut cortisol results obtained following the Toronto Zoo sample preparation protocol and cortisol quantitation with the EIA.

Table C. 1. Main differences between the EIA and LC-MS/MS procedures.

Amount of Qiviut cortisol Wash procedure Extraction procedure qiviut analyzed quantification Soapy water (30 s), 20 h in 6 mL methanol at LC- 20–100 mg water rinse, IPA 4°C; evaporation at 40°C LC-MS/MS MC/MS (10 s) – all cold under nitrogen 24 h in 5 mL methanol at room temperature on EIA 50 mg Methanol (10 s) Cortisol EIA rotator; evaporation in fume hood at room temperature

161

APPENDIX D Chapter 2 Supplementary material

Table D. 1. Parameter estimates of the final model explaining log-transformed cortisol levels and including sex (male versus female), year (2014, 2015, and 2016 versus 2013), and season (summer and winter versus fall) as fixed explanatory variables.

Parameter estimate Standard Error t value (Intercept) 1.5453 0.1436 10.7585 sex-male 0.2656 0.0800 3.3194 year-2014 0.2435 0.1106 2.2011 year-2015 0.5142 0.1163 4.4218 year-2016 0.4682 0.1399 3.3467 season-fall 0.4513 0.1104 4.0886 season-winter 0.5334 0.1129 4.7257

Table D. 2. Estimated mean (95% CI) qiviut cortisol levels per sex, season, and year*.

Year Season Sex Mean (95% CI) 2013 winter male 10.43 (8.36-13.01) female 7.99 (6.41-9.97) 2014 winter male 11.65 (9.33-14.53) female 8.93 (7.16-11.14) summer male 6.83 (5.50-8.49) fall male 10.73 (8.64-13.32) female 8.23 (6.63-10.21) 2015 winter male 11.71 (9.39-14.61) female 8.98 (7.20-11.20) fall male 10.79 (8.69-13.39) female 8.27 (6.66-10.27) 2016 winter male 11.99 (9.61-14.96) summer male 7.03 (5.35-9.25) *Only those combinations of factors for which we had data (Table 2.1) are shown. Table D. 3. Sample size description per season, year, and sex.

Female Male Winter 2013* 7 5 2014 13 6 2015 13 9 2016 0 16 Summer 2013 0 0 † 2014 0 15 2015 0 0 2016† 0 4

Fall 2013* 7 2 2014 13 26 2015† 5 9 *Samples collected in a single location (other than Cambridge Bay); †Samples collected only in Cambridge Bay

162

Figure D. 1. Plot showing the residuals against the fitted values.

Figure D. 2. Plots showing (a) the conditional modes for the random effect level (i.e., estimated mean difference from the mean, conditional on the fixed-effects) for each location of sampling and (b) the conditional modes and the residuals on the same scale.

163

APPENDIX E Summary of pharmacological challenges done in other wild and domestic even-toed ungulate species to validate the use of fecal glucocorticoid metabolite levels as a biomarker of hypothalamic–pituitary–adrenal axis activity

NS indicates non-specified information.

ACTH formulation (F), Timing of FGM peak (h Magnitude of FGM peak (% Timing of return to Sample Species dosage (D), and route post-injection – range unless increase from baseline – range baseline (h post- Assay** Reference size of administration (R)+ specified otherwise) unless specified otherwise) injection) Wild species F: Synacthen Depot, 1 ♂ ♂ = 8-22 h (11,17-DOA) and ♂ = 1,500% (11,17-DOA) and Hoffman-La 11,17-DOA EIA African buffalo yearling 22 h (3α,11oxo-CM) 1,200% (3α,11oxo-CM); ♂ = 48 h (11,17-DOA) Ganswindt et Roche AG and 3α,11oxo-CM (Syncerus caffer) and 1 ♀ = 7-21 h (both EIAs) ♀ = 1,000% (11,17-DOA) and ♀ = 29 h (11,17-DOA) al., 2012 D: 150 IU total EIA adult ♀ 900% (3α,11oxo-CM) R: IM RIA: average = 20 h RIA: average = 371% Corticosterone RIA F: Synacthen Depot, 3 adult ♂ (95% CI = 13-27 h) EIA: average = 1,247% (MP Biomedicals) African buffalo Novartis NS (> 50 h for RIA and Spaan et al., and 5 EIA: average = 23 h Both are expressed after and cortisol EIA (Syncerus caffer) D: 1 IU/kg EIA*) 2017 adult ♀ (95% CI = 15-30 h) accounting for week, age, sex, and (Enzo Life Sciences R: IM buffalo ID Inc.) 3 intact ♂, 3 F: Synacthen, Defiante Alpaca (Vicugna castrated Farmaceutica Median = 33 h > 48 h (no samples Arias et al., 963% (value NS) 11,17-DOA EIA pacos) ♂, and 3 D: 25 IU total collected past this time) 2013 ♀ all R: IV adults F: Synacthen Depot, Brown brocket 1 adult ♂ Novartis ♂ = 29 h* ♂ = 313% ♂ = 40 h* Cortisol EIA Christofoletti et deer (Mazama and 1 D: 25 IU total ♀ = 24 h* ♀ = 334% ♀ = 33 h* (Munro) al., 2010 gouazoubira) adult ♀ R: IM ♂ = 8 h ♂ = 190% F: ACTH gel form, Meds 5 adult ♂ ♀ = 8 h ♀ = 154% Caribou (Rangifer for Vets Inc. ♂ = 24 h Corticosterone RIA Ashley et al., and 5 Determined after averaging Determined after averaging values tarandus granti) D: 2 IU/kg ♀ = 48 h (MP Biomedicals) 2011 adult ♀ values of all animals of all animals R: IM

164

F: Synacthen Depot, Collared peccary Novartis Cortisol EIA Coradello et al., 6 adult ♂ 24 h 30-390% 48 h (Pecari tajacu) D: 0.25 IU/kg (Munro) 2012 R: IM F: Synacthen Depot, Dromedary 2 adult ♂ ♂ = 24 h Novartis ♂ = 210-734% ♂ = 36 h Sid-Ahmed et camel (Camelus and 3 ♀ = 36-48 h 3α,11oxo-CM EIA D: 50 IU total ♀ = 330-702% ♀ = 72 h al., 2013 dromedarius) adult ♀ R: IV F: ACTH Sigma- 1-24 Deer 1 = 18 h Deer 1 = 290% Fallow deer 3 ♂ Aldrich NS (no samples collected Konjević et al., Deer 2 = 22 h Deer 2 = 430% 11,17-DOA EIA (Dama dama) yearlings D: 0.6 IU/kg past 22 h) 2011 Deer 3 = 22 h Deer 3 = 290% R: IM Giraffe 1: Giraffe 1: Giraffe 1: 21 h (cortisol) 13.5-19.5 h (cortisol) 160% (cortisol) 53.5 h (corticosterone) 50.5-52.5 h (corticosterone) 60% (corticosterone) 51.5 h (11,17-DOA) 13.5-50.5 h (11,17-DOA) 3,240% (11,17-DOA) Cortisol EIA 51.5 h (3α,11oxo-CM) F: Synacthen Depot, 11.5-50.5 h (3α,11oxo-CM) 1720% (3α,11oxo-CM) (Munro), 37 h (3β,11β-diol CM) Novartis 13.5-35.5 h (3β,11β-diol CM) 310% (3β,11β-diol CM) corticosterone EIA Giraffe 35.5 h (3α,11β-diol CM) D: 1 IU/kg (giraffe 1) and 30-35 h (3α,11β-diol CM) 160% (3α,11β-diol CM) (Munro), 11,17- Bashaw et al., (Giraffa 2 adult ♂ Giraffe 2: 0.7 IU/kg (giraffe 2) Giraffe 2: Giraffe 2: DOA EIA, 2016 camelopardalis) No peak (cortisol) R: IM No peak (cortisol) No peak (cortisol) 3α,11oxo-CM EIA, No peak (corticosterone) No peak (corticosterone) No peak (corticosterone) 3β,11β-diol CM, and 44.7 h (11,17-DOA) 20.5-27.5 h (11,17-DOA) 860% (11,17-DOA) 3α,11β-diol CM 44.7 h (3α,11oxo-CM) 24-29 h (3α,11oxo-CM) 190% (3α,11oxo-CM) 44.7 h (3β,11β-diol CM) 23.2-28.2 h (3β,11β-diol CM) 190% (3β,11β-diol CM) No peak (3α,11β-diol No peak (3α,11β-diol CM) No peak (3α,11β-diol CM) CM) F: Synacthen, Defiante 3 ♀ and 3 Llama (Lama Farmaceutica Median = 28 h > 48 h (no samples Arias et al., ♂ all 805% (value NS) 11,17-DOA EIA glama) D: 25 IU total collected past this time) 2013 adults R: IV F: Synacthen, Ciba-Geigy Average = 19 h Red deer (Cervus 6 ♀ (age Huber et al., D: 0.5 IU/kg (range* = 16-29 h) 668-2,021% 20-45 h* 3α,11oxo-CM EIA elaphus) NS) 2003 R: IM F: Synacthen, CD Reindeer 7 h in 5 individuals and 14, 16, Pharmaceuticals AB Highly variable among Özkan Gülzari (Rangifer tarandus 8 adult ♂ 24 h in 3 other individuals 100-1,580% 3α,11oxo-CM EIA D: 25 IU total individuals (NS) et al., 2019 tarandus) Magnitude: R: IM

165

F: ACTH gel form, Meds ♂ = no response (NS) ♂ = no response (NS) Reindeer 5 ♂ and 5 for Vets Inc. ♀ = 24 h ♀ = 85% ♂ = no response (NS) Corticosterone RIA Ashley et al., (Rangifer tarandus ♀ all ≈ 1 D: 8 IU/kg Determined after averaging Determined after averaging values ♂ > 72 h (MP Biomedicals) 2011 tarandus) year R: IM values of all animals of all animals F: Synacthen Depot, Rocky mountain 1 adult ♂ ♂ = 32 h ♂ = 192% NS in 2 ♀ as not sampled Dulude-de Novartis Cortisol EIA goat (Oreamnos and 2 ♀ n°1 = 21 h ♀ n°1 = 223% long enough and 40 h in Broin et al., D: 50 IU total (Munro) americanus) adult ♀ ♀ n°2 = 20 h ♀ n°2 = 131% ♂* 2019 R: IM 5 Roe deer F: Synacthen, Ciba-Geigy castrated Dehnhard et (Capreolus D: 25 IU total 6-23 h 504-1,350% 28-31 h 11,17-DOA EIA ♂ all 9-11 al., 2001 capreolus) R: IM months 4 adult ♀ F: Cortrosyn, Organon White-tailed deer (2 in Corticosterone RIA Inc. October: 20-24 h October: 150% 30-35 h in October and Millspaugh et (Odocoileus October (ICN D: 50 IU total March: 10-13 h March: 240-253% March* al., 2002 virginianus) and 2 in Pharmaceuticals) R: IM March) Domestic species 3 ♀ and 3 F: Synacthen, Ciba-Geigy 6.0-19 h (after peak in blood Palme et al., Cattle (Bos taurus) ♂ (age D: 100 IU total 230-2,440% 18-44 h 11,17-DOA EIA cortisol) 1999 NS) R: IV 10 adult F: Synacthen Depot, Autumn: 528% (11,17-DOA) and Autumn: 21 h (both 11,17-DOA EIA ♀ (5 in Novartis 196% (corticosterone) Autumn: 14-18 h (both EIAs) EIAs*) and corticosterone Morrow et al., Cattle (Bos taurus) autumn D: 5 IU total twice (at a 2- Spring: 227% (corticosterone) Spring: 8-10 h (corticosterone) Spring: 21 h RIA (ICN 2002 and 5 in h interval) All values were calculated between (corticosterone*) Pharmaceuticals) spring) R: IV the baseline and peak averages F: Synacthen Depot, 11,17-DOA EIA Goats (Capra 20 adult Novartis Mean ± SD = 13 ± 1 h (both 11,17-DOA: 510-3,160% NS (≈ 24 h, but > 24 h in Kleinsasser et and 3α,11oxo-CM hircus) ♀ D: 50 IU total EIAs) 3α,11oxo-CM: 540-3,050% some individuals) al., 2010 EIA R: IM 3 ♀ and 3 F: Synacthen, Ciba-Geigy 6.0-19 h (after peak in blood Palme et al., Sheep (Ovis aries) ♂ (age D: 50 IU total 230-1,510% 18-44 h 11,17-DOA EIA cortisol) 1999 NS) R: IV +IM for intramuscular and IV for intravenous. *Values read from graphs presented in publications – these are approximate values. **The 11,17-DOA EIA is described in Palme and Möstl (1997), the 3α,11oxo-CM EIA in Möstl et al. (2002), the 3β,11β-diol CM EIA in Touma et al. (2003), and the 3α,11β-diol CM EIA in Ganswindt et al. (2003).

166

References

Arias N, Requena M, Palme R (2013) Measuring faecal glucocorticoid metabolites as a non-invasive tool for monitoring adrenocortical activity in South American camelids. Anim Welf 22: 25–31.

Christofoletti MD, Pereira RJG, Duarte JMB (2010) Influence of husbandry systems on physiological stress reactions of captive brown brocket (Mazama gouazoubira) and marsh deer (Blastocerus dichotomus)—noninvasive analysis of fecal cortisol metabolites. Eur J Wildl Res 56: 561–568.

Ganswindt A, Palme R, Heistermann M, Borragan S, Hodges JK (2003) Non-invasive assessment of adrenocortical function in the male African elephant (Loxodonta africana) and its relation to musth. Gen Comp Endocrinol 134: 156–166.

Huber S, Palme R, Zenker W, Mostl E (2003) Non-invasive monitoring of the adrenocortical response in red deer. J Wildl Manag 67: 258.

Kleinsasser C, Graml C, Klobetz-Rassam E, Barth K, Waiblinger S, Palme R (2010) Physiological validation of a non-invasive method for measuring adrenocortical activity in goats. Wien Tierärztl Monatsschrift 97: 259–262.

Konjević D, Janicki Z, Slavica A, Severin K, Krapinec K, Božić F, Palme R (2011) Non-invasive monitoring of adrenocortical activity in free-ranging fallow deer (Dama dama L.). Eur J Wildl Res 57: 77–81.

167

Millspaugh JJ, Washburn BE, Milanick MA (2002) Non-invasive techniques for stress assessment in white-tailed deer. Wildl Soc Bull 30: 899–907.

Morrow CJ, Kolver ES, Verkerk GA, Matthews LR (2002) Fecal glucocorticoid metabolites as a measure of adrenal activity in dairy cattle. Gen Comp Endocrinol 126: 229–241.

Möstl E, Maggs JL, Schrötter G, Besenfelder U, Palme R (2002) Measurement of cortisol metabolites in faeces of ruminants. Vet Res Commun 26: 127–139.

Palme R, Möstl E (1997) Measurement of cortisol metabolites in faeces of sheep as a parameter of cortisol. Int J Mamm Biol 62 (Supplement 2): 192–197.

Palme R, Robia C, Messmann S, Hofer J, Möstl E (1999) Measure of faecal cortisol metabolites in ruminants: A non-invasive parameter for adrena function. Wien Tierärztl Monatsschrift 86: 237–241.

Spaan JM, Pitts N, Buss P, Beechler B, Ezenwa VO, Jolles AE (2017) Noninvasive measures of stress response in African buffalo (Syncerus caffer) reveal an age- dependent stress response to immobilization. J Mammal 98: 1288–1300.

Touma C, Sachser N, Möstl E, Palme R (2003) Effects of sex and time of day on metabolism and excretion of corticosterone in urine and feces of mice. Gen Comp Endocrinol 130: 267–278.

168

APPENDIX F Results from the analytical validations of the cortisol and corticosterone enzyme immunoassays

Table F. 1. Polyclonal cortisol antibody R4866 cross-reactions (C. Munro, personal communication, 2010 (deceased in 2013))

Steroid % Cross Reaction Cortisol 100.0 Prednisolone 9.9 Prednisone 6.3 Compound S 6.2 Cortisone 5.0 Corticosterone 0.7 Desoxycorticosterone 0.3 21-desoxycortisone 0.5 11-desoxycortisol 0.2 Progesterone 0.2 17α-hydroxyprogesterone 0.2 Pregnenolone 0.1 17α-hydroxypregnenolone 0.1 Androstenedione 0.1 Testosterone 0.1 Androsterone 0.1 Dehyroepiandrosterone 0.1 Dehydroisoandrosterone-3-sulfate 0.1 Aldosterone 0.1 Estradiol-17β 0.1 Estrone 0.1 Estriol 0.1 Spironolactone 0.1 Cholesterol 0.1

169

Table F. 2. Polyclonal corticosterone antibody CJM006 cross-reactions (C. Munro, personal communication, 2010 (deceased in 2013))

Steroid % Cross Reaction Corticosterone 100.00 Desoxycorticosterone 14.25 Tetrahydrocorticosterone 0.90 11-Deoxycortisol 0.03 Prednisone < 0.01 Prednisolone 0.07 Cortisol 0.23 Cortisone < 0.01 Progesterone 2.65 Testosterone 0.64 Estradiol 17β < 0.01

Figure F. 1. Serial dilutions showing parallel displacement with the standard curve for the cortisol antibody.

170

Figure F. 2. Serial dilutions showing parallel displacement with the standard curve for the corticosterone antibody.

Figure F. 3. Recovery of exogenous cortisol added to a pooled muskox fecal extract. Samples were prepared as per protocol and spiked with cortisol standard at increasing concentrations.

171

Figure F. 4. Recovery of exogenous corticosterone added to a pooled muskox fecal extract. Samples were prepared as per protocol and spiked with corticosterone standard at increasing concentrations.

172

APPENDIX G Fecal cortisol results

Fecal cortisol levels were generally higher in the summer than in the winter, but exhibited important intra- and inter-individual variability and none of the animals showed a clear response following the ACTH injection during either of the challenges.

Table G. 1. Maximal percentage increase in fecal cortisol (%) as compared to time 0 levels for the muskoxen given a single injection of ACTH or saline (control) during the winter (ACTH dose 1 IU/kg) and/or summer (ACTH dose 2 IU/kg) and the respective times post-injection (h) at which it was observed.

Winter challenge Summer challenge Maximal Time Maximal Time Animal Experimental Experimental percentage increase post- percentage increase post- ID group group in fecal cortisol injection in fecal cortisol injection MX-738 / / / ACTH 43 23 MX-740 ACTH 19 3 Control 118 8 MX-741 ACTH 28 25 ACTH 70 31 MX-620 ACTH 9 54 / / / MX-621 ACTH 57 45 ACTH 0* 0* MX-597 / / / ACTH 41 25 MX-283 ACTH 30 95 Control 105 22 MX-1169 ACTH 69 5 ACTH 264 91 *The maximal fecal cortisol concentration was measured at time 0.

173

Figure G. 1. Individual fecal cortisol levels as a function of the time following a single injection of ACTH or saline (control) for the muskoxen sampled during the winter (ACTH dose 1 IU/kg – n = 6 ACTH- injected) and/or summer (ACTH dose 2 IU/kg – n = 5 ACTH-injected and n = 2 controls). Winter data are indicated as grey lines. Data for the ACTH-injected and control animals during the summer challenge correspond to the black and red lines, respectively.

174

APPENDIX H Summary of the studies measuring intestinal transit times in muskoxen

Study Intestinal retention time (IRT) Description and study design Fluid marker Particulate (h)* marker (h) Markers (chromium-51 complexed with EDTA Summer (fluid) and ruthenium-103 chloride (particulate)) fed by way of a capsule or by dosing a portion Holleman et al., Pen-fed 6.5 7.1 of the food to 3–9 young muskoxen (age and sex 1984 Pasture 3.2 4.4 not specified) Winter Calculation of IRT using a compartmental Pen-fed 9.1 9.5 model Pasture 4.8 0.7

Chromium-mordanted hay (particulate) fed to 6 Adamczewski et March: 21.0 ± 1.1 h (mean ± SE) non-breeding adult ♀ al., 1994 July: 14.7 ± 0.5 h Calculation of IRT using a compartmental model Fluid Particulate Cobalt-EDTA (fluid) and ytterbium-fiber marker (h)+ marker (h) (particulate) given in solution and in gel Barboza et al., May-June 32.1 ± 20.3 29.4 ± 15.0 capsules, respectively, at a duodenal cannula to 2006 Aug-Sept 28.1 ± 9.1 26.9 ± 10.3 4–6 castrated adult ♂ Feb-March 33.5 ± 13.9 36.8 ± 12.1 Direct measurement of IRT

*Mean IRTs were calculated using the rumen turnover time (RTT), transit time (TT), and total mean retention time (TMRT) provided in the publication as IRT = TMRT – (RTT + TT). No SD or SE could, however, be calculated without access to the raw data. +All results are presented as mean ± SD

References

Adamczewski JZ, Flood PF, Chaplin RK, Schaefer JA (1994) Seasonal variation in intake and digestion of a high-roughage diet by muskoxen. Can J Anim Sci 74: 305–313.

175

APPENDIX I Identification, sex, age, and experimental group of the animals included in Experiments 1 and 2

Abbreviations: Identification (ID), male (M), female (F), and steer (S).

Each line corresponds to the same individual and colors used for shading indicate the different randomization groups.

Experiment 1 Experiment 2 Animal ID Study Age Experimental Study Age Experimental Feces ID* (years) group ID* (years) group sampled MX-738 M1-1 0.75 Control M1-2 1.25 ACTH Yes MX-740 F1-1 0.75 ACTH F1-2 1.25 Control Yes MX-741 M2-1 0.75 ACTH M2-2 1.25 ACTH Yes MX-619 M3-1 1.75 Control M3-2 2.25 ACTH Yes MX-620 M4-1 1.75 ACTH M4-2 2.25 Control No MX-621 F2-1 1.75 ACTH F2-2 2.25 ACTH Yes MX-596 / / F3-2 3.25 Control Yes MX-597 F3-1 2.75 Control F4-2 3.25 ACTH Yes MX-283 F4-1 5.75 ACTH F5-2 6.25 Control Yes MX-1169 F5-1 6.75 ACTH F6-2 7.25 ACTH Yes MX-247 / / / F7-2 11.25 ACTH Yes MX-253 F6-1 9.75 Control F8-2 10.25 ACTH No MX-541 M5-1 9.75 ACTH / / / MX-279 M6-1 5.75 Control / / / MX-280 M7-1 5.75 ACTH / / / MX-463 M8-1 3.75 ACTH / / / MX-598 M9-1 2.75 ACTH / / / MX-276 / / / S1-2 6.25 ACTH No MX-278 / / / S2-2 6.25 Control No MX-347 / / / F9-2 5.25 Control No MX-167 / / / F10-2 7.25 ACTH No *None of the females were pregnant during either Experiment 1 or Experiment 2 ACTH challenges.

176

APPENDIX J Summary of the pharmacological challenges carried out in wild and domestic mammalian species to evaluate whether hypothalamic–pituitary–adrenal axis activity is reflected in the hair (adapted from Koren et al., 2019)

Challenge description Response Sample Timing, protocol & (ACTH formulation, Number of in serum, Response in hair (timing) Species description (size ACTH dose body region of hair Reference administration route*, controls, doses (timing) saliva or & magnitude (n), sex, age) sampling timing) faeces Captive and free-ranging wild mammals ACTH: n = 9 ♀ Timing: week -4 (n = 9 (1 not sampled ACTH & n = 3 controls Yes (serum, Formulation: Cortrosyn, Amstar post-challenge) (group 2), otherwise not separate Black-tailed Pharmaceuticals, USA Control group 1: specified), 12-49 days 3-5 injections ACTH prairie dogs Route: IM n = 36 ♀ after last injection (mean over 6-28 days, challenge) 12 IU/kg No Crill et al., 2019 (Cynomys Controls: un-manipulated (group (8 not sampled of 33 ± 3 days) (mean of 16 ± 2 No (feces, ludovicianus) 1) or saline (group 2) post-challenge) Protocol: same patch re- days) separate Timing: hair growth (March-Aug) Control group 2: shaved ACTH n = 9 ♀ Body region: right hind challenge) Age = adults leg Formulation: Corticotrophin, Timing: day 0, week 6 (1 Wedgewood Pharmacy, USA (n = ACTH: n = 3 week after last ACTH Yes (week 6)† Canada lynx 2) or Synacthen Depot, Novartis (1 intact , 1 ♂ injection) Lynx 1: 164% increase Terwissen et (Lynx Pharmaceuticals, Canada (n = 1) 20 IU/kg 5 (1/week) NA intact ♀, and 1 Protocol: same patch re- Lynx 2: 50% increase al., 2013 canadensis) Route: IM spayed ♀) shaved Lynx 3: 172% increase Controls: none Age = adults Body region: hind leg Timing: hair growth (Oct-Nov) Formulation: gel form, Meds for ACTH: n = 10 Timing: days -7, 7 Vets, USA Caribou (5 ♀ & 5 ♂) Protocol: new patch Route: IM Ashley et al., (Rangifer Controls: n = 2 2 IU/kg shaved for each sample 1 (day 0) Yes (feces) No Controls: saline 2011 tarandus granti) Body region: neck, Timing: absence of hair growth (1♀ & 1 ♂) shoulder, rump (March) Age > 1.8 years Formulation: Synacthen Depot, ACTH: n = 5 Timing: day 0, 6-26 days ACTH: 5.4 ± 1.8 Yes (6-26 days after last Eastern Novartis Pharmaceuticals, Canada (3 ♂, 2 ♀) after last injection except (mean ± SD) injection)† Mastromonaco chipmunks 10 IU/kg NA Route: IM Controls: n = 7 one control shaved 64 injections over ACTH: median increase = et al., 2014 (Tamias striatus) Controls: none (3 ♂, 4 ♀) days after 64.2 ± 13.8 days 637% (range = -31–1677%)

177

Timing: hair growth (May-Aug) Age = adults Protocol: same patch re- Controls: 5.7 ± 3 n = 2 non-responders shaved injections over Controls: median increase = Body region: right hind 54.9 ± 20.8 days 102% (range = -51–277%) limb Timing: day 0, week 6 (1 Formulation: Synacthen Depot, ACTH: n = 4 week after last ACTH Yes (feces, Adult ♀ 1: 264% increase Mountain Novartis Pharmaceuticals, Canada Dulude-de (1 adult , 2 injection) separate Adult ♀ 2: 147% increase goats (Oreamnos Route: IM ♂ 25 IU total 5 (1/week) Broin et al., Protocol: same patch re- ACTH Yearling ♀: 240% increase americanus) Controls: none adult ♀, 1 yearling 2019 shaved challenge) Timing: hair growth (Aug-Sept) ♀) Adult ♂: 233%. increase Body region: rump Formulation: gel form, Meds for ACTH: n = 10 (5 Timing: days -7, 7 Vets, USA ♀ & 5 ♂) Protocol: new patch Reindeer (R. t. Route: IM Ashley et al., Controls: n = 2 2 IU/kg shaved for each sample 1 (day 0) No (feces) No tarandus) Controls: saline 2011 Body region: neck, Timing: absence of hair growth (1♀ & 1 ♂) shoulder, rump (March) Age = 0.8 years Formulation: gel form, Meds for ACTH: n = 10 (5 Timing: days -7, 14, 134 Yes (feces, Vets, USA ♀ & 5 ♂) Protocol: new patch but low Reindeer (R. t. Ashley et al., Route: IM Controls: n = 2 8 IU/kg shaved for each sample 1 (day 0) response No tarandus) 2011 Controls: saline (1♀ & 1 ♂) Body region: neck, and in ♀ Timing: hair growth (June) Age ≈ 1 year shoulder, rump only) Domestic mammals Yes (days 14 and 28)+ ACTH: day 0 = 19 ± 2.5 ACTH: n = 5 ♀ Formulation: porcine ACTH, Timing: days 0, 14, 28, 44 ng/g; day 14 = 88 ± 27.5 Control group 1: n Sigma-Aldrich, USA Protocol: shave at day 0, ng/g; day 28 = 76 ± 39 ng/g; = 5 ♀ González-de- Cattle (Bos Route: IV same patch divided into 3 Yes day 44 = 16.5 ± 1 ng/g Control group 2: n 0.15 IU/kg 3 (days 0,7,14) la-Vara et al., taurus) Controls: nothing (group 1) or and re-shaved at days 14, (serum) Control group 2 (control = 5 ♀ 2011 saline (group 2) 28, 44 group 1, results similar): day 0 Age = 17-22 Timing: hair growth Body region: not specified = 26 ± 1 ng/g; day 14 = 21.5 months ± 0.8 ng/g; day 28 = 16 ± 1 ng/g; day 44 = 19 ± 1 ng/g Formulation: porcine ACTH, ACTH: n = 12 Timing: days -14, 0, 14 Sigma-Aldrich, USA ♂ Protocol: same patch re- Cattle (Bos Yes Tallo-Parra et Route: IV Controls: n = 12 1 IU/kg shaved 2 (days 0, 7) No taurus) (serum) al., 2017 Controls: nothing ♂ Body region: forehead, Timing: hair growth Age ≈ 4.5 months hip

178

Regrown – yes (weeks 4, 8)+ ACTH: week 4 = 66 ± 6 (least square mean ± SE); week 8 = 70 ± 14; week 12 = 5 ± 2.5 pg/mg Controls: week 4 = 11.5 ± 6; Timing: day 0, weeks 4 week 8 = 5.5 ± 13.5; week 12 Formulation: Synacthen Depot, (end of treatment), 8, 12 = 4.5 ± 2 pg/mg ACTH: n = 17 ♀ Alfasigma S.p.A., Italy Protocol: same patch re- 14 (1 every Natural – yes (weeks 4, 8, Cattle (Bos Controls: n = 17 Heimbürge et Route: IM 100 IU total shaved (regrown) and new second day for 4 Yes (saliva) 12)+ taurus) ♀ al., 2020 Controls: saline patch shaved (natural) weeks) ACTH: day 0 = 8.5 ± 0.5 Age ≈ 7.9 months Timing: hair growth (Sept-Jan) Body region: caudo-dorsal (least square mean ± SE); region of the back week 4 = 29 ± 2; week 8 = 42 ± 3.5; week 12 = 34 ± 2.5 pg/mg Controls: day 0 = 9 ± 0.5; week 4 = 9.5 ± 2; week 8 = 11.5 ± 3; week 12 = 8 ± 3 pg/mg Formulation: Cortrosyn, Daiichi Pharmaceutical Co., Japan ACTH: n = 6 ♀ Timing: months 0, 1, 2 Goats (Capra 7 (1/day, days 0- Yes Endo et al., Route: IM Controls: n = 6 ♀ 0.0625 IU/kg Protocol: unclear No hircus) 6) (serum) 2018 Controls: saline Age = adults Body region: rump Timing: hair growth (Aug) Yes (month 1)+ ACTH: month 0 = 1.64 ± 0.76 pg/mg (mean ± SD); Formulation: Cortrosyn, Daiichi month 1 = 3.36 ± 1.36 Pharmaceutical Co., Japan ACTH: n = 6 ♀ Timing: months 0, 1, 2 Goats (Capra 28 (2/day, days Yes pg/mg: month 2 = 2.84 ± Endo et al., Route: IM Controls: n = 6 ♀ 0.0625 IU/kg Protocol: unclear hircus) 0-13) (serum) 0.28 pg/mg 2018 Controls: saline Age = adults Body region: rump Controls: month 0 = 2.6 ± Timing: hair growth (Oct) 0.52 pg/mg; month 1 = 2.6 ± 0.92; month 2 = 3.04 ± 1.32 pg/mg Formulation: Synacthen Depot, Timing: day 0, weeks 4 Regrown – yes (week 4)+ ACTH: n = 19 ♀ Alfasigma S.p.A., Italy (end of treatment), 8, 12 14 (1 every ACTH: week 4 = 67 ± 8 Pig (Sus Controls: n = 19 Heimbürge et Route: IM 100 IU total Protocol: same patch re- second day for 4 Yes (saliva) (least square mean ± SE); domesticus) ♀ al., 2020 Controls: saline shaved (regrown) and new weeks) week 8 = 39 ± 4; week 12 = Age ≈ 4.5 months Timing: hair growth (March-Aug) patch shaved (natural) 48 ± 6.5 pg/mg

179

Body region: caudo-dorsal Controls: week 4 = 49.5 ± 8; region of the back week 8 = 38.5 ± 4; week 12 = 53.5 ± 6.5 pg/mg Natural – yes (week 4, both ACTH and controls)+ ACTH: day 0 = 56 ± 9 (least square mean ± SE); week 4 = 105 ± 10; week 8 = 83 ± 11.5; week 12 = 93 ± 7.5 pg/mg Controls: day 0 = 75 ± 9; week 4 = 118 ± 10; week 8 = 104 ± 11.5; week 12 = 101.5 ± 7.5 pg/mg *IM for intramuscular and IV for intravenous. +No raw data was available, so these numbers were determined graphically from the figure published (i.e., they are highly approximate). †Raw data obtained from the authors.

References

Crill, C., Janz, D.M., Kusch, J.M., Santymire, R.M., Heyer, G.P., Shury, T.K., Lane, J.E., 2019. Investigation of the utility of feces and hair as non-invasive measures of glucocorticoids in wild black-tailed prairie dogs (Cynomys ludovicianus). Gen Comp Endocrinol 275: 15–24. https://doi.org/10.1016/j.ygcen.2019.02.003

Endo N, Yamane H, Rahayu LP, Tanaka T (2018) Effect of repeated adrenocorticotropic hormone administration on reproductive function and hair cortisol concentration during the estrous cycle in goats. Gen Comp Endocrinol 259: 207–212.

Heimbürge S, Kanitz E, Tuchscherer A, Otten W (2020) Is it getting in the hair? – Cortisol concentrations in native, regrown and segmented hairs of cattle and pigs after repeated ACTH administrations. Gen Comp Endocrinol 295: 113534.

180

Koren L, Bryan H, Matas D, Tinman S, Fahlman Å, Whiteside D, Smits J, Wynne‐Edwards K (2019) Towards the validation of endogenous steroid testing in wildlife hair. J Appl Ecol 56: 547–561.

Mastromonaco, G.F., Gunn, K., McCurdy-Adams, H., Edwards, D.B., Schulte-Hostedde, A.I., 2014. Validation and use of hair cortisol as a measure of chronic stress in eastern chipmunks (Tamias striatus). Conserv Physiol 2: cou055–cou055. https://doi.org/10.1093/conphys/cou055

Tallo-Parra O, Lopez-Bejar M, Carbajal A, Monclús L, Manteca X, Devant M (2017) Acute ACTH-induced elevations of circulating cortisol do not affect hair cortisol concentrations in calves. Gen Comp Endocrinol 240: 138–142.

Terwissen, C.V., Mastromonaco, G.F., Murray, D.L., 2013. Influence of adrenocorticotrophin hormone challenge and external factors (age, sex, and body region) on hair cortisol concentration in Canada lynx (Lynx canadensis). Gen Comp Endocrinol 194: 162–167. https://doi.org/10.1016/j.ygcen.2013.09.010

181

APPENDIX K Results from the various cortisol enzyme immunoassay analytical validations

Figure K. 1. Extraction mass-dose response. Samples were processed as per protocol described in Chapter 5.

182

Table K. 1. Polyclonal cortisol antibody R4866 cross-reactions (C. Munro, personal communication, 2010 (deceased in 2013)).

Steroid %Cross Reaction Cortisol 100,0 Prednisolone 9,9 Prednisone 6,3 Compound S 6,2 Cortisone 5,0 Corticosterone 0,7 Desoxycorticosterone 0,3 21-desoxycortisone 0,5 11-desoxycortisol 0,2 Progesterone 0,2 17α-hydroxyprogesterone 0,2 Pregnenolone 0,1 17α-hydroxypregnenolone 0,1 Androstenedione 0,1 Testosterone 0,1 Androsterone 0,1 Dehyroepiandrosterone 0,1 Dehydroisoandrosterone-3-sulfate 0,1 Aldosterone 0,1 Estradiol-17β 0,1 Estrone 0,1 Estriol 0,1 Spironolactone 0,1 Cholesterol 0,1

183

Figure K. 2. Recovery of exogenous cortisol added to a pooled muskox qiviut extract. Samples were prepared as per protocol and spiked with cortisol standard at increasing concentrations.

Figure K. 3. Serial dilutions showing parallel displacement with the standard curve.

184

APPENDIX L Boxplots showing the coefficients of variation for the various qiviut sample groups

To account for intra-sample variability, for all qiviut samples, two (non-growing winter and spring- collected samples) or four (summer-collected/grown samples) subsamples were tested independently as true experimental replicates (i.e., duplicates or quadruplicates, respectively).

185

186

Figure L. 3. Boxplot of intra-sample variation (quadruplicate CV) in the rump segments of qiviut grown during the summer ACTH challenge (between t0 and t0 + 6W) when collected 6 weeks [t0 + 6W], 3 months [t0 + 3M], and 6 months [t0 + 6M] after the start of the challenge in the control (n = 6) and ACTH-injected (n = 10) animals. The thick horizontal lines correspond to the medians, the triangles to the means and the empty circles to the outliers.

Figure L. 4. Boxplot of intra-sample variation (duplicate CV) in rump qiviut grown during the entire hair growth period when shaved in February 2019 and collected shed during qiviut combing in April-May 2019 in all muskoxen (n = 16). The thick horizontal lines correspond to the medians, the triangles to the means and the empty circles to the outliers.

187

APPENDIX M Details of the linear mixed models with qiviut cortisol as the dependent variable and coefficient estimates

Abbreviations: Coef for coefficient, SE for standard error, 95%CI for 95% confidence interval, DF for degrees of freedom, R² for coefficient of determination, and ICC for intraclass correlation coefficient.

1) Experiment 1 – Are single and repeated ACTH injections reflected in qiviut in the absence of hair growth?

Models assessing the effect of sampling time (t0, t0 + 1W and t0 + 5W) on log-transformed rump qiviut cortisol separately in each experimental group.

a) Control animals (n = 5)

Model estimates (the reference category for sampling time is t0).

Coef SE DF t-value p-value Intercept 2.59 0.22 8 11.70 < 0.001 Sampling time = t0 + 1W 0.06 0.06 8 0.95 0.37 Sampling time = t0 + 5W 0.01 0.06 8 0.15 0.88

Back-transformed coefficients and their associated 95%CIs (the reference category for sampling time is t0)

Intercept 13.32 (7.99–22.19) Sampling time = t0 + 1W 1.06 (0.92–1.22) Sampling time = t0 + 5W 1.01 (0.88–1.16)

Model details

Number of observations 15 Number of groups 5 Between animal variance 0.24 Error variance (within animal) 0.01 Marginal R² < 0.01 Conditional R² 0.96 ICC 0.96

b) ACTH-injected animals (n = 10)

Model estimates (the reference category for sampling time is t0)

Coef SE DF t-value p-value Intercept 2.78 0.17 18 16.42 < 0.001 Sampling time = t0 + 1W 0.03 0.12 18 0.21 0.84 Sampling time = t0 + 5W -0.05 0.12 18 -0.38 0.71

188

Back-transformed coefficients and their associated 95%CIs (the reference category for sampling time is t0)

Intercept 16.18 (11.33–23.10) Sampling time = t0 + 1W 1.02 (0.80–1.31) Sampling time = t0 + 5W 0.96 (0.75–1.22)

Model details

Number of observations 30 Number of groups 10 Between animal variance 0.22 Error variance (within animal) 0.07 Marginal R² < 0.01 Conditional R² 0.76 ICC 0.76

2) Experiment 2 – Are repeated ACTH injections reflected in qiviut when the hair is growing?

Models assessing the effect of sampling time (t0 and t0 + 6W) on qiviut cortisol separately in each experimental group and body location.

a) Control animals (n = 6)

Model estimates for each body region (the reference category for sampling time is t0)

Coef SE DF t-value p-value 95% CI Neck Intercept 6.36 1.02 5 6.21 < 0.01 3.73–8.99 Sampling time = t0 + 6W 4.50 1.04 5 4.32 0.01 1.82–7.18 Shoulder Intercept 8.11 1.26 5 6.42 < 0.01 4.86–11.36 Sampling time = t0 + 6W 3.79 1.54 5 2.45 0.06 -0.18–7.76 Rump Intercept 8.53 1.19 5 7.19 < 0.001 5.48–11.58 Sampling time = t0 + 6W 2.83 1.33 5 2.13 0.09 -0.59–6.26

Model details

Neck Shoulder Rump Number of observations 12 12 12 Number of groups 6 6 6 Between animal variance 3.02 2.42 3.12 Error variance (within animal) 3.26 7.16 5.33 Marginal R² 0.47 0.29 0.21 Conditional R² 0.72 0.47 0.5 ICC 0.48 0.25 0.37

189

b) ACTH-injected animals (n = 10)

Model estimates for each body region (the reference category for sampling time is t0)

Coef SE DF t-value p-value 95%CI Neck Intercept 8.09 1.20 9 6.72 < 0.001 5.37–10.81 Sampling time = t0 + 6W 10.36 1.50 9 6.93 < 0.001 6.98–13.74 Shoulder Intercept 10.05 1.71 9 5.86 < 0.001 6.17–13.93 Sampling time = t0 + 6W 10.53 1.99 9 5.29 < 0.001 6.03–15.02 Rump Intercept 7.66 0.78 9 9.83 < 0.001 5.9–9.42 Sampling time = t0 + 6W 7.25 0.84 9 8.66 < 0.001 5.36–9.14

Model details

Neck Shoulder Rump Number of observations 20 20 20 Number of groups 10 10 10 Between animal variance 3.3 9.63 2.57 Error variance (within animal) 11.18 19.77 3.5 Marginal R² 0.66 0.5 0.69 Conditional R² 0.74 0.66 0.82 ICC 0.23 0.33 0.42

3) Experiment 2 – Do qiviut cortisol levels differ between the neck, shoulder, and rump?

Models assessing the effect of body region (neck, shoulder, and rump) on pre-challenge (t0) qiviut cortisol in all the animals (n = 16).

Model estimates (the reference category for body location is neck)

Coef SE DF t-value p-value 95%CI Intercept 7.44 0.72 30 12.94 < 0.001 5.97–8.91 Body location = shoulder 1.88 0.67 30 2.82 0.01 0.52–3.24 Body location = rump 0.55 0.67 30 0.82 0.42 -0.81–1.91

Model details

Number of observations 48 Number of groups 16 Between animal variance 4.76 Error variance (within animal) 3.55 Marginal R² 0.07 Conditional R² 0.60 ICC 0.57

190

4) Experiment 2 – Does the cortisol concentration in a specific segment of qiviut remain the same over time?

Model assessing the effect of sampling time (t0 + 6W, t0 + 3M and t0 + 6M) on the log-transformed cortisol concentrations in the segments of qiviut grown during the summer ACTH challenge on the rump in each experimental group.

a) Control animals (n = 6)

Model estimates (the reference category for sampling time is t0 + 6W)

Coef SE DF t-value p-value Intercept 2.40 0.21 10 11.29 < 0.001 Sampling time = t0 + 3M 0.11 0.24 10 0.46 0.65 Sampling time = t0 + 6M 0.35 0.24 10 1.47 0.17

Back-transformed coefficients and their associated 95%CIs (the reference category for sampling time is t0 + 6W)

Intercept 11.05 (6.88–17.76) Sampling time = t0 + 3M 1.12 (0.66–1.89)

Sampling time = t0 + 6M 1.42 (0.83–2.41)

Model details

Number of observations 18 Number of groups 6 Between animal variance 0.10 Error variance (within animal) 0.17 Marginal R² 0.08 Conditional R² 0.42 ICC 0.38

b) ACTH-injected animals

Model estimates (the reference category for sampling time is t0 + 6W)

Coef SE DF t-value p-value Intercept 2.69 0.12 18 22.85 < 0.001 Sampling time = t0 + 3M 0.50 0.14 18 3.50 < 0.01 Sampling time = t0 + 6M 0.78 0.14 18 5.42 < 0.001

Back-transformed coefficients and their associated 95%CIs (the reference category for sampling time is t0 + 6W)

Intercept 14.76 (6.88–17.76) Sampling time = t0 + 3M 1.66 (1.22–2.24)

Sampling time = t0 + 6M 2.18 (1.61–2.95)

191

Model details

Number of observations 30 Number of groups 10 Between animal variance 0.04 Error variance (within animal) 0.10 Marginal R² 0.44 Conditional R² 0.58 ICC 0.25

5) Experiment 2 – Is the cortisol concentration in shed qiviut the same as prior to shedding?

Model assessing the effect of qiviut collection method (shed and shaved) on the cortisol levels of qiviut grown during the entire hair growth period on the rump of all the animals (n = 16).

Model estimates (the reference category for qiviut collection method is shaved)

Coef SE DF t-value p-value Intercept 2.82 0.09 15 32.02 < 0.001 Qiviut collection method = shed 0.26 0.11 15 2.47 0.03

Back-transformed coefficients and their associated 95%CIs (the reference category for qiviut collection method is shaved)

Intercept 16.84 (13.95–20.32) Qiviut collection method = shed 1.30 (1.04–1.63)

Model details

Number of observations 32 Number of groups 16 Between animal variance 0.04 Error variance (within animal) 0.09 Marginal R² 0.12 Conditional R² 0.37 ICC 0.28

Comparison of coefficients between the model with only qiviut collection method and the model with both qiviut collection method and experimental group (the reference category for qiviut collection method and experimental group are shaved and ACTH-injected, respectively)

Coef p-value Coef p-value Intercept 2.82 < 0.001 2.87 < 0.001 Qiviut collection method = shed 0.26 0.03 0.26 0.03 Experimental group = ACTH-injected . . -0.07 0.64

192

APPENDIX N Details of the linear mixed models with quadruplicate coefficient of variation as the dependent variable and coefficient estimates

Models assessing the effect of body region (neck, shoulder, and rump) on pre-challenge (t0) quadruplicate CV in all the animals (n = 16).

Model estimates (the reference category for body location is neck)

Coef SE DF t-value p-value Intercept 2.87 0.11 30 24.96 < 0.001 Body location = shoulder 0.17 0.16 30 1.04 0.31 Body location = rump -0.10 0.16 30 -0.61 0.55

Back-transformed coefficients and their associated 95%CIs (the reference category for body location is neck)

Intercept 17.59 (13.91–22.24) Body location = shoulder 1.18 (0.85–1.65) Body location = rump 0.91 (0.65–1.26)

Model details

Number of observations 48 Number of groups 16 Between animal variance < 0.001 Error variance (within animal) 0.21 Marginal R² 0.06 Conditional R² 0.06 ICC < 0.001

193

APPENDIX O Details of the linear mixed models with fecal glucocorticoid metabolites as the dependent variable and coefficient estimates

Model assessing the effect of injection number on pre-injection FGM levels in each experimental group.

1) Control animals (n = 3)

Model estimates

Coef SE DF t-value p-value 95% CI Intercept 8.98 2.22 11 4.05 < 0.01 4.10–13.87 Injection -0.14 0.67 11 -0.21 0.84 -1.61–1.33

Model details

Number of observations 15 Number of groups 3 Between animal variance < 0.001 Error variance (within animal) 13.4 Marginal R² < 0.01 Conditional R² < 0.01 ICC < 0.001

2) ACTH-injected animals (n = 7)

Model estimates

Coef SE DF t-value p-value Intercept 2.01 0.28 27 7.28 < 0.001 Injection 0.11 0.08 27 1.30 0.20

Back-transformed coefficients and their associated 95%CIs

Intercept 7.48 (4.24–13.19) Injection 1.11 (0.94–1.31)

Model details

Number of observations 35 Number of groups 7 Between animal variance 0.03 Error variance (within animal) 0.46 Marginal R² 0.04 Conditional R² 0.11 ICC 0.07

194

APPENDIX P Individual qiviut cortisol line-plots

Figure P. 1. Line plots showing, the neck (a), shoulder (b), and rump (c) qiviut cortisol levels for each individual at t0 and t0 + 6W in the control (n = 6) and ACTH-injected animals (n = 10). Detailed information regarding the animals can be found in Appendix I.

195

Figure P. 2. Line plots showing the neck, shoulder, and rump qiviut cortisol levels for each individual at t0. Detailed information regarding the animals can be found in Appendix I.

196

APPENDIX Q Summary of the various studies evaluating differences in hair glucocorticoids among body regions in wild or domestic mammalian species

Species Sample size (n) Body regions compared Results Reference Captive and free-ranging wild mammals One location on foot with lower Canada lynx (Lynx canadensis) 20 6 locations (3 on foot & 3 on leg) Terwissen et al., 2013 levels Caribou (Rangifer tarandus 12 Neck, shoulder, rump Neck > shoulder & rump Ashley et al., 2011 granti)α Back (n = 46), forearm Chimpanzees (Pan troglodytes) (n = 48), shoulder Back, forearm, shoulder, chest Chest > shoulder > forearm > back Carlitz et al., 2015 (n = 47), chest (n = 47) Arm (n = 25), side (n = Side > arm Chimpanzees (Pan troglodytes) Arm, back, side Yamanashi et al., 2013 22), back (n = 22) Back > armζ Neck, shoulders, mid-back, Coyotes (Canis latrans) 12 No difference Schell et al., 2017 abdomen, hips, above tail Eastern gray kangaroos Ventral (hypogastric), dorsal 25 Dorsal > ventral Sotohira et al., 2017 (Macropus giganteus) (lumbar) Grizzly bear (Ursus arctos) 15 Neck, shoulder, abdomen, rump Neck > shoulder, abdomen & rump Macbeth et al., 2010 Right & left wrists (n = 12), stomach (n = 14), Right & left wrists, stomach, Orang-utans (Pongo spp.) No difference Carlitz et al., 2014 back (n = 11), right & left back, right & left shoulders shoulders (n = 15) Polar bears (Ursus maritimus) 20 Neck, shoulder, abdomen, rump No difference Macbeth et al., 2012 Reindeer (Rangifer tarandus 12 Neck, shoulder, rump Neck < shoulder & rump Ashley et al., 2011 tarandus)α Reindeer (Rangifer tarandus 10 Neck, rump No difference Carlsson et al., 2016 tarandus)β Vancouver Island marmots Forelimb, chest, back, hindlimb, 8 Hindlimb > forelimb, rump & back Acker et al., 2018 (Marmota vancouverensis) rump Yellow baboons (Papio 5 Shoulder, thigh, base of tail Thigh > base of tail Fourie et al., 2016 cynocephalus)

197

Domestic mammals Cattle (Bos taurus) 18 Shoulder, top line, hip, tail switch Tail switch > shoulder Burnett et al., 2014 Cattle (Bos taurus)γ 12 Head, neck, shoulder, hip, tail No difference Moya et al., 2013 Cattle (Bos taurus) 40 Shoulder, back, tail tip Tail tip > back & shoulder Heimbürge et al., 2020 Ghassemi Nejad et al., Cattle (Bos taurus) 70 Forehead, withers, rump No difference 2019 Horse (Equus ferus caballus)ε 10 Mane, tail Mane > tail Duran et al., 2017 Pigs (Sus scrofa domesticus) 46 Neck, back, tail tip Tail tip > back > neck Heimbürge et al., 2020 Dorsal area of neck Dorsal area of neck, dorsolumbar Dorsolumbar region > dorsal area of Pigs (Sus scrofa domesticus)δ (n = 31), dorsolumbar Casal et al., 2017 region neck region (n = 42) Sheep (Ovis aries) 30 Shoulder, back Shoulder > back Fürtbauer et al., 2019 New Zealand white rabbits 8 26 locations No difference Comin et al., 2012 (Oryctolagus cuniculus) αResults presented correspond to those of the hair samples collected before the ACTH challenge. βResults presented correspond to those of the hair samples collected before removal of parasites. γResults presented correspond to those of the hair samples collected on day 1 of the study. δResults presented correspond to those of the hair samples collected during experimental week 8 of the study (i.e., sample 1). εDifference significant for the first three 6-cm proximal segments of hair tested corresponding to 3 months of hair growth each, but not significant for the fourth and most distal segment. ζDifference at the limit of significance (p-value = 0.051).

References

Acker, M., Mastromonaco, G., Schulte-Hostedde, A.I., 2018. The effects of body region, season and external arsenic application on hair cortisol concentration. Conserv Physiol 6 (1): cox075. https://doi.org/10.1093/conphys/coy037

Burnett TA, Madureira AML, Silper BF, Nadalin A, Tahmasbi A, Veira DM, Cerri RLA (2014) Short communication: Factors affecting hair cortisol concentrations in lactating dairy cows. J Dairy Sci 97: 7685–7690.

Carlitz, E.H.D., Kirschbaum, C., Miller, R., Rukundo, J., van Schaik, C.P., 2015. Effects of body region and time on hair cortisol concentrations in chimpanzees (Pan troglodytes). Gen Comp Endocrinol 223: 9–15. https://doi.org/10.1016/j.ygcen.2015.09.022

Carlitz, E.H.D., Kirschbaum, C., Stalder, T., van Schaik, C.P., 2014. Hair as a long-term retrospective cortisol calendar in orang-utans (Pongo spp.): New perspectives for stress monitoring in captive management and conservation. Gen Comp Endocrinol 195: 151–156. https://doi.org/10.1016/j.ygcen.2013.11.002

198

Carlsson, A.M., Mastromonaco, G., Vandervalk, E., Kutz, S., 2016. Parasites, stress and reindeer: infection with abomasal nematodes is not associated with elevated glucocorticoid levels in hair or faeces. Conserv Physiol 4 (1): cow058. https://doi.org/10.1093/conphys/cow058

Casal N, Manteca X, Peña L R, Bassols A, Fàbrega E (2017) Analysis of cortisol in hair samples as an indicator of stress in pigs. J Vet Behav 19: 1–6.

Comin A, Zufferli V, Peric T, Canavese F, Barbetta D, Prandi A (2012) Hair cortisol levels determined at different body sites in the New Zealand white rabbit. World Rabbit Sci 20: 149–154.

Duran MC, Janz DM, Waldner CL, Campbell JR, Marques FJ (2017) Hair cortisol concentration as a stress biomarker in horses: Associations with body location and surgical castration. J Equine Vet Sci 55: 27–33.

Fourie NH, Brown JL, Jolly CJ, Phillips-Conroy JE, Rogers J, Bernstein RM (2016) Sources of variation in hair cortisol in wild and captive non-human primates. Zoology 119: 119–125.

Fürtbauer I, Solman C, Fry A (2019) Sheep wool cortisol as a retrospective measure of long-term HPA axis activity and its links to body mass. Domest Anim Endocrinol 68: 39–46.

Heimbürge S, Kanitz E, Tuchscherer A, Otten W (2020) Within a hair’s breadth – Factors influencing hair cortisol levels in pigs and cattle. Gen Comp Endocrinol 288: 113359.

Macbeth BJ, Cattet MRL, Obbard ME, Middel K, Janz DM (2012) Evaluation of hair cortisol concentration as a biomarker of long-term stress in free-ranging polar bears. Wildl Soc Bull 36: 747–758.

Macbeth, B.J., Cattet, M.R.L., Stenhouse, G.B., Gibeau, M.L., Janz, D.M., 2010. Hair cortisol concentration as a noninvasive measure of long-term stress in free- ranging grizzly bears (Ursus arctos): Considerations with implications for other wildlife. Can J Zool 88: 935–949. https://doi.org/10.1139/Z10-057

Moya D, Schwartzkopf-Genswein KS, Veira DM (2013) Standardization of a non-invasive methodology to measure cortisol in hair of beef cattle. Livest Sci 158: 138– 144.

Schell CJ, Young JK, Lonsdorf EV, Mateo JM, Santymire RM (2017) Investigation of techniques to measure cortisol and testosterone concentrations in coyote hair. Zoo Biol 36: 220–225.

199

Sotohira Y, Suzuki K, Sano T, Arai C, Asakawa M, Hayashi H (2017) Stress assessment using hair cortisol of kangaroos affected by the lumpy jaw disease. J Vet Med Sci 79: 852–854.

Yamanashi, Y., Morimura, N., Mori, Y., Hayashi, M., Suzuki, J., 2013. Cortisol analysis of hair of captive chimpanzees (Pan troglodytes). Gen Comp Endocrinol 194: 55– 63. https://doi.org/10.1016/j.ygcen.2013.08.013

200

APPENDIX R Instructions provided to the hunters with the kits to collect the samples

201

APPENDIX S Information form provided to the hunters with the kits

202

APPENDIX T Investigation of biologically and ecologically plausible two-way interactions

Interactions were added individually to the model including all main effects (sex, age, year, location, Up_lpg, Ve_lpg, lung_richness, nematodirus_epg, eimeria_epg, marshallagia_epg, moniezia_YN, GI_richness, condition_hunter, marrow_fat, erysipelothrix_PP, brucella_serology) and their impact on the deviance information criterion (DIC) and potential effect were assessed. Interactions that gave the best improvement (i.e., highest decrease) of the DIC and those which had a potential effect on qiviut cortisol were retained for model building and are indicated in bold.

Interaction DIC Potential effect None (i.e., model with all fixed effects) 6630.17 - season:condition_hunter 6638.72 no sex:erysipelothrix_PP 6638.47 no year:brucella_serology 6636.61 no age:monizeia_YN 6635.83 no sex:age 6635.61 no age:marrow_fat 6634.91 no sex:condition_hunter 6634.57 no age:season 6633.99 no age:brucella_serology 6633.99 no age:erysipelothrix_PP 6633.54 no age:nematodirines_epg 6633.36 no age:marshallagia_epg 6633.08 no age:condition_hunter 6632.07 no location:erysipelothrix_PP 6631.71 no age:eimeria_epg 6630.98 no age:Up_lpg 6628.90 no location:Up_lpg 6628.66 yes year:season 6626.94 no season:marrow_fat - - (coefficients not converged) sex:season 6624.81 yes sex:brucella_serology 6624.53 no sex:marrow_fat 6622.71 no location:Ve_lpg 6622.10 no age:Ve_lpg 6620.41 no sex:year 6618.06 yes year:erysipelothrix_PP 6616.51 no

203

APPENDIX U Model selection procedure by manual backward stepwise elimination

DIC “Stage 2” starting model (sex, age, year, location, Up_lpg, Ve_lpg, lung_richness, nematodirus_epg, eimeria_epg, marshallagia_epg, moniezia_YN, GI_richness, condition_hunter, 6634.36 marrow_fat, erysipelothrix_PP, age:Ve_lpg, sex:year, year:erysipelothrix_PP, location:Up, sex:season) Removal of GI_richness 6630.78 Removal of GI parasites egg/oocyst counts (nematodirines_epg, eimeria_epg, marshallagia_epg, 6616.10 moniezia_YN) Removal of condition_hunter 6204.61 Removal of lung_richness 6198.32 Removal of year:erysipelothrix_PP 6188.88 Removal of erysipelothrix_PP 6048.36 Removal of sex:year 6038.72 Removal of age:Ve_lpg 6039.56 Removal of Ve_lpg 6036.66 Removal of age (final model with sex, year, location, Up_lpg, marrow_fat, 6033.43 location:Up, sex:season)

204

APPENDIX V Parameter trace plots

Red and black represent the two Monte Carlo Markov chains. Plots show adequate mixing and convergence of the two chains.

205

APPENDIX W Manuscript based thesis – copyright disclosure statement

November 12, 2020

The co-authors and journal editors gave permission to Juliette Di Francesco to include the manuscripts specified below in her doctoral thesis to be submitted to the Faculty of Graduate Studies, University of Calgary. The co-authors and journal editors were made aware that University of Calgary theses are licensed under the University of Calgary Non-Exclusive Distribution License (license available at: http://grad.ucalgary.ca/files/grad/university-of-calgary-non-exclusive-distribution-licence.pdf). They were also made aware that University of Calgary theses are harvested by Library and Archives Canada and will be publicly available through the University of Calgary Theses Repository, The Vault.

Supporting documents that prove that all the applicable permissions were obtained have been submitted to the Faculty of Graduate Studies, University of Calgary.

Juliette Di Francesco University of Calgary Faculty of Veterinary Medicine Department of Ecosystem and Public Health

206