The Pennsylvania State University

The Graduate School

Department of Ecosystem Science and Management

A LANDSCAPE OF FEAR: MEASURING NUTRITIONAL AND PSYCHOLOGICAL

STRESS IN PREDATOR-PREY INTERACTIONS

A Dissertation in

Wildlife and Fisheries Science

by

Catharine Pritchard

© 2018 Catharine Pritchard

Submitted in Partial Fulfillment

of the Requirements

for the Degree of

Doctor of Philosophy

December 2018

The dissertation of Catharine Pritchard was reviewed and approved* by the following:

Tracy Langkilde Professor and Head of Biology, The Pennsylvania State University Dissertation Advisor

C. Paola Ferreri Associate Professor of Fisheries Management, The Pennsylvania State University Chair of Committee

Victoria Braithwaite Professor, The Pennsylvania State University

Matthew Marshall Adjunct Assistant Professor of Wildlife Conservation, The Pennsylvania State University

Michael Messina Professor and Head of Ecosystem Science and Management, The Pennsylvania State University

*Signatures are on file in the Graduate School

iii ABSTRACT

Animals commonly respond to stimuli, including risk of predation, nutritional deficits, and disturbance through the stress response. The most commonly measured stress hormones, glucocorticoids (GCs), then generally become elevated, energy is diverted away from non- essential processes, and behavior is modified to facilitate short-term survival. Because GCs can be collected noninvasively, they are candidates for evaluating health in wild animals. However, few studies have tested critical assumptions about GCs and their relationships with other health- relevant measures in wild animals, which are commonly observed in laboratory settings. I tested if these relationships held in wild animals, using pumas (Puma concolor) and vicuñas (Vicugna vicugna) in a dyadic predator-prey system. In Chapter 2, I test the predictions that: 1) GCs in four biological matrices (blood, saliva, feces, and hair) collected during captures will be positively related, allowing different matrices to be used interchangeably, and 2) that elevated GC levels indicate poor condition by correlating with downstream health-relevant traits. I found little support for these predictions, and suggest these be validated for each species to properly interpret

GC levels. In Chapter 3, I test the relationships between GC levels in different matrices and their downstream effects, collected during captures in two different habitats. I found that these relationships are not always consistent across habitat. In Chapter 4, I explore state-dependent foraging, using behavioral observations and fecal samples, from which I measured GCs and thyroid hormones to indicate levels of fear and nutritional deficits, respectively. There were no relationships between behavior and hormone concentrations, and I suggest this tradeoff may be difficult to detect in animals at baseline conditions. Lastly, in Chapter 5, I examine anti-predator behavior in response to estimates of direct and indirect risk. I report estimated home range sizes and successful chemical immobilization for vicuñas. By estimating dynamic interaction (across space and time), I found little response of vicuñas to direct risk, but found anticipatory anti-

iv predatory behaviors in response to indirect risk, through diel migrations to avoid risky habitats at the riskiest time of day. These results provide a cautionary tale of GCs as physiological indicators of population health, and highlight the insights provided by quantifying dynamic interactions between species.

v TABLE OF CONTENTS

LIST OF FIGURES ...... viii

LIST OF TABLES ...... x

ACKNOWLEDGEMENTS ...... xi

Chapter 1 Introduction ...... 1

The Physiological Stress Response ...... 2 The Hypothalamic-Pituitary-Thyroidal Axis ...... 3 Study System ...... 4 Research Objectives ...... 6 Literature Cited ...... 9

Chapter 2 Lack of Correspondence Among Different Measures of Stress in a Wild Animal ...... 18

Abstract ...... 18 Introduction ...... 19 Methods ...... 21 Study System and Capture Protocol ...... 21 Glucocorticoid Sample Collection and Analysis ...... 21 Downstream Effects Measurements and Analysis ...... 23 Statistical Analysis ...... 25 Animals Ethics Statement ...... 25 Results ...... 25 Glucocorticoid Measurements ...... 25 Downstream Measurements ...... 27 Discussion...... 27 Relationships Between Matrices ...... 28 Short-Term Integrators ...... 28 Short- and Long-Term Integrators ...... 29 Long-Term Integrators ...... 30 Ability to Predict Downstream Effects ...... 31 Conclusions ...... 35 Literature Cited ...... 38

Chapter 3 Few and Complex Relationships in Glucocorticoids and Fitness- Relevant Downstream Measures Between Habitats in a Wild Animal...... 62

Abstract ...... 62 Introduction ...... 62 Methods ...... 65 Study Organism and Study Site ...... 65 Biological Sample Collection ...... 66

vi Glucocorticoid Sample Collection and Analysis ...... 66 Downstream Effects Measurements and Analysis ...... 67 Statistical Analysis ...... 69 Animal Ethics Statement ...... 69 Results ...... 70 Discussion...... 72 Conclusions ...... 76 Literature Cited ...... 77

Chapter 4 Glucocorticoids and Triiodothyronine do not Correlate with Behavior in a Wild Mammal...... 88

Abstract ...... 88 Introduction ...... 88 Methods ...... 93 Study Animal and Study Sites ...... 93 Behavioral Analyses ...... 94 Fecal Sample Analyses...... 95 Statistical Analyses ...... 96 Animal Ethics Statement ...... 97 Results ...... 97 Discussion...... 98 Conclusion ...... 100 Literature Cited ...... 101

Chapter 5 Direct and Indirect Estimates of Risk and Associated Antipredator Responses ... 117

Abstract...... 117 Introduction ...... 117 Methods ...... 120 Study Animals ...... 120 Study Site ...... 121 Vicuña and Puma Capture ...... 122 Statistical Analyses ...... 122 Home Ranges...... 122 Utilization Distribution Overlap ...... 123 Dynamic Interactions...... 123 Results ...... 124 Home Ranges ...... 124 Utilization Distribution Overlap...... 125 Dynamic Interactions ...... 125 Discussion ...... 126 Conclusions ...... 130 Literature Cited ...... 131

vii Appendix Vicuña and Puma Capture and Chemical Immobilization ...... 140

Vicuña Chemical Immobilizations ...... 140 Vicuña Sampling and Collaring ...... 140 Puma Capture and Chemical Immobilization ...... 142 Literature Cited ...... 143

viii LIST OF FIGURES

Figure 2-1: Trendlines from linear models of relationships between matrices. Dependent variables on the y-axis are plasma cortisol (CORT), salivary CORT, and fecal glucocorticoid metabolites (FGCMs). Blank graphs indicate the dependent and independent physiological variables are the same, so no model was run...... 51

Figure 2-2: Trendlines from linear models of relationships between matrices. Dependent variables on the y-axis are leg hair cortisol (CORT), neck hair CORT, and rump hair CORT. Blank graphs indicate the dependent and independent physiological variables are the same, so no model was run...... 52

Figure 2-3: Trendlines from linear models of relationships between matrices and downstream effects. Dependent variables on the y-axis are glucose, hematocrit, and triiodothyronine (T3). X-axis independent variables include cortisol (CORT) in different matrices, and fecal glucocorticoid metabolites (FGCMs). Blank graphs indicate the dependent and independent physiological variables are the same, so no model was run...... 53

Figure 2-4: Trendlines from linear models of relationships between matrices and downstream effects. Dependent variables on the y-axis are Carbon:Nitrogen ratio, plasma progesterone, and body weight index residuals. X-axis independent variables include cortisol (CORT) in different matrices, and fecal glucocorticoid metabolites (FGCMs). Blank graphs indicate the dependent and independent physiological variables are the same, so no model was run...... 54

Figure 2-5: Relationships between glucocorticoid measurements in multiple matrices. Colors correspond to statistical significance—lightest boxes are significant at p < 0.05, medium-dark boxes indicate 0.05 < p < 0.10, and dark blue boxes indicate p > 0.10...... 55

Figure 2-6: Relationships between dependent downstream effects (rows), and independent glucocorticoid measurements in multiple matrices (columns). Colors correspond to significance—lightest boxes are significant at p < 0.05, medium-dark boxes indicate 0.05 < p < 0.10, and dark blue boxes indicate p > 0.10...... 56

Figure 3-1: Multiple linear regressions of glucocorticoid concentrations in different matrices plotted with 95% confidence intervals. Single solid black lines indicate models where there were no effects of habitat on the relationships between the two glucocorticoid measurements. When there were additive or interactive effects of habitat, each habitat has been graphed separately. Solid blue lines indicate canyons, and red dotted lines indicate meadows...... 84

Figure 3-2: Multiple linear regressions of glucocorticoid concentrations and downstream effects plotted with 95% confidence intervals. Single solid black lines indicate models where there were no effects of habitat on the relationships between the two glucocorticoid measurements. When there were additive or interactive effects of habitat,

ix each habitat has been graphed separately. Solid blue lines indicate canyons, and red dotted lines indicate meadows...... 85

Figure 3-3: Results from multiple linear regressions plotted with 95% confidence intervals. Single solid black lines indicate models where there were no effects of habitat on the relationships between the two glucocorticoid measurements. When there were additive or interactive effects of habitat, each habitat has been graphed separately. Solid blue lines indicate canyons, and red dotted lines indicate meadows...... 86

Figure 4-1. Pumas in tall grass of the meadows (left) and open habitat of the canyons (right). Both the high vegetation density in meadows, and the rocky outcroppings and steep slopes in canyons conceal pumas well and facilitate successful hunting...... 112

Figure 4-2. Relationships between physiological measurements and the percentage of time spent foraging. There was no relationship between an individual’s fecal glucocorticoid metabolite (GCs) or triiodothyronine (T3) concentrations and the percent time it spent foraging. Colors indicate individual animals...... 113

Figure 4-3. Relationships between physiological measurements and the percentage of time spent vigilant. There was no relationship between fecal glucocorticoid metabolites (GCs) or triiodothyronine (T3) and the percent time an individual exhibited vigilance behavior. Colors indicate individual animals...... 114

Figure 4-4. Proportion of time spent foraging and vigilant between the canyon and meadow habitats. Vicuñas spent approximately 11% more time foraging in meadows than canyons (p = 0.027). There was a trend for vicuñas to be more vigilant in canyons than in meadows by approximately 11% (p = 0.067)...... 115

Figure 4-5. Physiological measurements in canyon and meadow habitats. There were no differences between fecal glucocorticoid metabolite or triiodothyronine concentrations in canyons and meadows once individual ID was accounted for. Colors indicate individual animals...... 116

x LIST OF TABLES

Table 2-1: Downstream measurements influenced by glucocorticoid levels, predictions for individuals in good and poor condition, the adaptive role of the measure under acute stress, and references for previous studies where the effect has been discussed and measured...... 57

Table 2-2: Model parameters from linear models describing the relationships between glucocorticoid (CORT) in different matrices and downstream effects. Only variables with p < 0.10 were retained for this table, except for all variables of interest, the independent panel. Bold p-values indicate p < 0.10...... 58

Table 3-1. Expected and observed relationships between plasma cortisol concentrations (CORTc), fecal glucocorticoid metabolites (FGCMs), and neck hair cortisol concentrations and downstream effects. A “+” indicates a positive relationship, “-" indicates a negative relationship, and “=” indicates no relationship...... 87

Table 5-1. Description of female vicuña home ranges by year and habitat...... 136

Table 5-2. Description of puma home ranges by year. Puma home ranges covered both habitats...... 137

Table 5-3. Home range overlap between pumas and vicuñas. Overlap was greater in 2015 than in 2014 or 2016...... 138

Table 5-4. Spatial overlap (proportion) between vicuñas and pumas based on minimum convex polygons...... 139

xi ACKNOWLEDGEMENTS

There are many people without whom this work would not have been possible.

To my parents, William and Monica Pritchard, I am forever grateful for your unconditional support, love, encouragement, and wisdom. I would not be here without your strong, positive influence, and rock-solid support.

To my partner, I will be eternally grateful for your support, love, and understanding. You always seem to be at the top of whatever rabbit hole I have dug myself into, whether the garden or R, and that reassurance makes so many more things possible.

To my advisor, Dr. Tracy Langkilde, I would not be here without your constant support, optimism, and encouragement. Your continued commitment to encouraging the next generation of scientists through positive interactions, to facilitating opportunities for your students, and to upholding the highest of standards across a broad spectrum is something I will emulate. Your enthusiasm and joy of learning and sharing is contagious.

To my committee members, Drs. Victoria Braithwaite, C. Paola Ferrari, and Matt

Marshall, thank you for your support, provocative questions, hard work, and encouraging conversations. In your own way, each of you is a model for encouragement, productive conversations, and training the next generation of scientists. It was a pleasure to work with you.

Thank you for all of your guidance.

Lastly, but certainly not least, I am grateful to members of the broader community. Thank you to my Argentine technician and volunteers for all of their hard work under difficult conditions. This project could not have been completed without them. Thank you to members of the Langkilde lab, who provided critical feedback on rough drafts of this work, and thank you to my collaborators, Drs. Emiliano Donadio, Arthur Middleton, and Jon Pauli for their comments and suggestions.

1 Chapter 1

Introduction

Stress is a construct used to describe a set of physiological responses in vertebrates to unpredictable or uncontrollable stimuli that threaten homeostasis (Romero et al., 2009). Many stimuli, such as nutritional deficits (Bryan et al., 2014; Jeanniard du Dot et al., 2009; Kitaysky et al., 2007), predation (Davis and Gabor, 2015; Monclús et al., 2009; Narayan et al., 2013), and anthropogenic disturbances including tourism (Ellenberg et al., 2007; Rehnus et al., 2014;

Zwijacz-Kozica et al., 2013), traffic (Ayres et al., 2012; Creel et al., 2002; Hayward et al., 2011), urbanization (Davies et al., 2017; French et al., 2008; Scheun et al., 2015), noise (Barton et al.,

2018; Tennessen et al., 2014) and light pollution (Gaston et al., 2013) are recognized as stressors to wild animals. Critical for conservation and wildlife management is that tools exist to evaluate the impact of these disturbances in wild animals. Glucocorticoids (GCs), the mostly commonly measured stress hormones, have been put forward as potential candidates (Madliger et al., 2015;

Narayan, 2017; Wikelski and Cooke, 2006). As part of the endocrine system, however, glucocorticoids interact extensively with other body systems, playing important roles in other endocrine pathways, energy mobilization, and triggering behaviors (Landys et al., 2006; Sapolsky et al., 2000). Thus, critical to interpreting GC concentrations is understanding what other conditions they represent (Cooke and O’Connor, 2010).

The stress response is ubiquitous across vertebrates (Breuner and Orchinik, 2002; Landys et al., 2006), however evidence is increasingly emerging that the GC response is context- dependent and plastic (Archard et al., 2012; Bonier et al., 2009; Busch and Hayward, 2009; Davis and Gabor, 2015). To be a useful indicator of the impacts of exposure to stressors, it is crucial that GC measurements represent the same condition in multiple situations, for instance, across habitats where a species naturally occurs. By combining GCs with other measures of condition, such as thyroid hormones which respond to nutritional condition, I may be able to better identify

2 the source of stressors when multiple exist. These methods, if implemented and understood well, could serve to identify the effects, and potential sources, of disturbances and global change on wildlife.

The physiological stress response

When an animal encounters a stressor such as a predator, the hypothalamic-pituitary- adrenal axis (HPA axis) is activated. This ‘stress response’ consists of the hypothalamus producing corticotropin releasing hormone (CRH) and arginine vasopressin (AVP) into the hypophyseal portal system. These hormones cause the anterior pituitary to release adrenocorticotropic hormone (ACTH), which travels through the blood and causes the adrenal cortex to secrete GCs (cortisol in most mammals, including vicuñas and pumas, corticosterone, or a mixture in other animals; de Kloet et al., 1998; Vera et al., 2011; Wingfield and Sapolsky,

2003). At baseline levels, most circulating GCs are bound to corticosteroid binding globulin

(CBG) leaving only approximately 5% free and biologically active, and thus available to tissues

(Bright and Darmaun, 1995; Rosner, 1990). Once free circulating GCs are delivered to tissues, they bind to intracellular cytoplasmic receptors that begin working as transcription factors after arriving at the cell nucleus (Sapolsky et al., 2000). As transcription factors, GCs attach to GC response elements on DNA, subsequently promoting or inhibiting gene transcription, and changing protein synthesis and compositions— processes I will subsequently term ‘downstream effects’ (Sapolsky et al., 2000). These downstream effects can become evident within 30 minutes of a stressor and last days after the stressor has stopped (Sapolsky et al., 2000).

The HPA axis can be chronically activated in response to a prolonged stimulus, causing responses greater in magnitude and duration. This is generally attributed to weak negative feedback by reducing glucocorticoid receptor densities in the hypothalamus (Romero, 2004;

Sapolsky et al., 1984). Animals exposed to frequent stressors that activate the HPA-axis, such as

3 risk from predation, may demonstrate a chronic stress response (Romero, 2004). Chronic activation of the stress response can be detrimental, with a variety of physiological consequences, including suppression of the immune and reproductive systems, mental impairment, and elevated baseline GC levels (Boonstra et al., 1998; Dickens and Romero, 2013; Sapolsky et al., 2000).

Persistent downstream effects can lead to detrimental effects on health, condition, and reproduction.

Glucocorticoids are essential for numerous physiological functions over a variety of timescales (Dallman et al., 1993; Landys-Ciannelli et al., 2002; Landys et al., 2006). On a daily basis, GCs are involved in promoting daily cycles such as alertness, hunger, and activity levels

(Leibowitz & Hoebel 1997). Glucocorticoid levels can vary seasonally (Breuner and Orchinik,

2001), and for some species, may mediate predictable annual activities such as migration, fasting, and reproduction (Boonstra, 2005; Kenagy and Place, 2000; Landys et al., 2006). At the longest timescale, GCs may play a role in selection and evolution through maternal stress effects, priming offspring for environmental pressures into which they are born (Hadany et al., 2006). Thus, GC levels vary predictably over an individual’s lifetime. Glucocorticoids are also crucial in responding to short-term acute stressors such as predation (Boonstra, 2013; Sapolsky et al.,

2000), by promoting behaviors and prioritizing energy expenditure to facilitate survival and a return to homeostasis (McEwen and Wingfield, 2003; Oitzl et al., 2010; Wingfield et al., 1998).

Because GCs respond to many stimuli, however, measuring them in concert with other, more specific, indicators of a stressor may facilitate the identification of the sources of stress.

The hypothalamic-pituitary-thyroidal axis

Measuring thyroid hormones (THs) in concert with GCs may provide the means to evaluate the relative influences of predation and food on individual condition or population health

(Wasser et al., 2011, 2010). In addition to altering GC levels, nutritional deficits can alter TH

4 levels, specifically triiodothyronine (T3) and thyroxine (T4; Douyon and Schteingart, 2002;

Kitaysky et al., 2005; Rosen and Kumagai, 2008; Samuels and McDaniel, 1997; Schew et al.,

1996). These hormones are important in regulating thermoregulation and metabolism, and are secreted via the hypothalamic-pituitary-thyroidal (HPT) axis. When activated, the hypothalamus secretes thyrotropin-releasing hormone (TRH), stimulating thyrotropes in the pituitary to release thyroid-stimulating hormone (TSH). The thyroid subsequently releases T3 and T4. The HPT axis is regulated through the negative feedback at the hypothalamus, the pituitary, and the thyroid glands. T4 is considered an inactive TH, and is produced at higher concentrations than T3

(approximately 10:1; Norman & Litwack 1997). However, T4 is converted via deionization in peripheral tissues to T3, which is considered the biologically active portion of the hormone

(Robbins, 1981; Tomasi, 1991). As nutritional deficits increase, T3 concentrations tend to decrease, lowering basal metabolic rate and resting energy expenditure (Douyon and Schteingart,

2002; Kitaysky et al., 2005; Rosen and Kumagai, 2008; Samuels and McDaniel, 1997; Schew et al., 1996) and thus may be useful to indicate nutritional deficits. For example, T3 levels decreased significantly throughout a 28-day restricted diet phase in Stellar sea lions, and only returned to baseline levels at the end of the controlled recovery phase (Jeanniard du Dot et al., 2009).

Study system

My dissertation research was conducted in a strongly dyadic predator-prey system composed of pumas (Puma concolor) and vicuñas (Vicugna vicugna) in San Guillermo National

Park (SGNP), Argentina. As ambush predators, pumas use rocky outcroppings and tall vegetation to stalk prey, and thus, prey can use habitat features as consistent cues of predation risk. Pumas are the only predators of adult vicuñas in this system, contributing to the majority of adult vicuña deaths (Donadio et al., 2012).

5 Vicuñas are moderately sized ungulates (40-50 kg) native to South America at elevations above ~3,300 m, and are the smallest members of the Camelidae family (Fowler 2011). Vicuñas are largely group-dwelling, with individuals belonging to either bachelor (males up to 4 years of age) or family groups (one male, at least one female, and offspring; Koford, 1957), but solitary individuals can occur. Males either form bachelor groups or roam as solitary individuals, while females join family groups. Sexually monomorphic, the only distinguishing trait between males and females at a distance is their behavior—males tend to appear more vigilant and territorial towards male conspecifics and other threats than females (Koford, 1957) and females are also often accompanied by their offspring (crias—young of the year). Breeding occurs in late January and early February (Donadio et al., 2012; Koford, 1957), gestation lasts approximately 50 weeks, and parturition is highly synchronous—approximately 83% of calves are born during the first 3 weeks in January (Donadio et al., 2012). Juveniles wean at approximately 6-8 months, and both males and females disperse by the first year when new young arrive (Koford 1957). Sexual maturation likely does not occur until 2 years (Koford 1957), and males begin their family groups at 3-4 years of age and can maintain them for up to 6 years (Koford 1957). Puma predation is the major cause of mortality my study population, accounting for 65% of mortality within the first year of life, and 91% of mortality in adults (Donadio et al. 2012).

The landscape at SGNP consists of three habitat types which vary in both risk and foraging potential. Meadows are the most productive of the habitats, and are characterized by dense vegetation, with both high forage quality and quantity. However, these areas are particularly risky—480% more vicuña deaths were recorded in the meadows than would be expected given their spatial extent (meadows comprise 4% of the landscape; Donadio and

Buskirk, 2016). Canyons are characterized by steep clines and rocky outcroppings, with intermediate forage quality and quantity, and are also relatively risky, with 90% more vicuña kills via predation than would be expected (canyons comprise 15% of the landscape; Donadio and

6 Buskirk, 2016). Plains dominate the landscape, and are nearly devoid of vegetation and rocky outcroppings, so successful predation occurs 30% less frequently than expected (plains comprise

81% of the landscape; Donadio and Buskirk, 2016). Due to the strongly dyadic nature of the puma-vicuña predator-prey relationship and the distinct habitat types within SGNP, this system presents an excellent opportunity to test questions of tradeoffs between predation risk and foraging potential.

Research objectives

In Chapter 2, I examine the relationships among glucocorticoid measurements in different biological matrices and downstream effects. GC levels are most often measured via blood, but can also be measured non-invasively through saliva, feces, and hair. GCs enter these alternate matrices through different pathways: passively from the blood into saliva, after being metabolized and secreted via the bile into feces, and for hair these mechanisms are not yet fully understood

(Millspaugh et al., 2003; Stalder and Kirschbaum, 2012; Vining et al., 1983b). Furthermore, each matrix may incorporate GCs over a period of minutes (saliva; Vining et al., 1983b, 1983a), hours

(feces; Arias et al., 2013), or weeks to months (hair; Ashley et al., 2011; Davenport et al., 2006;

Macbeth et al., 2010). As an increasing number of researchers and animal welfare managers begin to use these alternate matrices (Cooke et al., 2013; Wikelski and Cooke, 2006) it is critical to test whether GC levels in each matrix reflects the true condition of the animal, as determined by downstream effects. Although some studies have measured individual condition using multiple matrices and downstream effects (Boonstra et al., 1998; Eeva et al., 2005; Fanson et al., 2013), few have simultaneously quantified GC levels in all four matrices and compared these measurements with many downstream indices of health. By comparing these measurements simultaneously, I suggest that they should be used cautiously, and should be well-studied in the species of interest before they are interpreted.

7 In Chapter 3, I continue to examine these relationships between a subset of these measurements, but examine them in two habitats. Integral to the effective and appropriate use of

GCs as conservation-relevant hormones, is that they represent the condition of the animal consistently across multiple habitats. Habitats may differ in many factors, each of which may individually affect GC levels, such as predation, food availability, climate, and environmental characteristics. For instance, in one species, if elevated GCs correlate with decreased reproductive hormone levels in one habitat, I would expect them to correlate with the same reproductive condition in another habitat. If they do not, and managers falsely assume that low GCs reflect low reproductive condition, management and conservation practices may be placed inappropriately or unnecessarily.

In Chapter 4, I examine the relationships between anti-predator behaviors, foraging, and physiological condition. Prey can respond to predation threats by altering behaviors, which can incur trade-offs including between foraging and vigilance (Lima, 1988; Lima and Dill, 1990;

Peckarsky and McIntosh, 1998). State-dependent foraging theory suggests that decisions on how to balance such trade-offs should be condition-dependent (Liesenjohann et al., 2015; Naslund and

Johnsson, 2016; Pettersson and Bronmark, 1993). I use behavioral observations and non- invasively collected fecal samples to measure GCs and THs, which may represent fear and nutritional condition, respectively (Wasser et al., 2010). To date, few studies have evaluated how fearfulness and nutritional deficits (via GC and TH levels) drive behavior in wild animals, or investigated these interactions in wild animals under natural, rather than disturbed conditions.

In Chapter 5, I examine vicuña antipredator behavior in response to both direct and indirect risk. I quantify temporal and spatial overlap in pumas and vicuñas, recorded via GPS collars, in response to these potential cues of risk. Specifically, direct cues of risk may be predator presence within a distance that represents the opportunity for a predation attempt.

Indirect cues of risk may include habitat features that facilitate the successful hunting of

8 predators, and risky times, when predators are most active (Brown and Kotler, 2004; Laundré et al., 2010). I explore and describe these puma-vicuña dynamic interactions across the landscape. I also estimate vicuña home ranges based on GPS technology, and provide information on successful chemical immobilization in vicuñas in the wild.

Chapters 2 and 3 demonstrate that caution should be used in the interpretation of condition based on GC measurements in some species, and across habitats. Chapter 4 demonstrates foraging trade-offs may be challenging to observe in wild animals under natural, undisturbed conditions. Chapter 5 demonstrates limited antipredator responses to direct risk, but anticipatory responses to indirect cues of risk.

Throughout this dissertation, I approach each aspect in a conservation context. Both glucocorticoids and thyroid hormones are touted as strong potential candidates for identifying the magnitude, consequences, and in some cases, sources of a disturbance to wild animals (Ayres et al., 2012; Jesmer et al., 2017; Vynne et al., 2014), largely because they respond to many sources of disturbance, and because they can be collected noninvasively in many wild animals (Wasser et al., 2010, 2000). I found that GC levels in different matrices are not interchangeable, do not correlate consistently with downstream effects, and that these relationships are not consistent across habitats, suggesting cautious use of these biomarkers when they are not well-explored in the species of interest. I also found no relationships between foraging and antipredator behaviors and these biomarkers, suggesting that state-dependent relationships may be difficult to observe when animals are in baseline conditions. This work adds to a growing body of literature that GCs may be difficult to properly interpret in wild animals (Dantzer et al., 2014). Lastly, I found that dynamic interactions incorporating spatial and temporal overlap can inform the nature of interactions between species of concern and movement between habitats.

9 Literature Cited

Archard GA, Earley RL, Hanninen AF, Braithwaite VA (2012) Correlated behaviour and stress

physiology in fish exposed to different levels of predation pressure. Funct Ecol 26: 637–

645.

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.

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.

Ayres KL, Booth RK, Hempelmann JA, Koski KL, Emmons CK, Baird RW, Balcomb-Bartok K,

Hanson MB, Ford MJ, Wasser SK (2012) Distinguishing the impacts of inadequate prey

and vessel traffic on an endangered killer whale (Orcinus orca) population. PLoS One 7.

doi:10.1371/journal.pone.0036842

Barton BT, Hodge ME, Speights CJ, Autrey AM, Lashley MA, Klink VP (2018) Testing the

AC/DC hypothesis: Rock and roll is noise pollution and weakens a trophic cascade. Ecol

Evol 1–8.

Bonier F, Martin PR, Moore IT, Wingfield JC (2009) Do baseline glucocorticoids predict fitness?

Trends Ecol Evol 24: 634–642.

Boonstra R (2005) Equipped for life: the adaptive role of the stress axis in male mammals. J

Mammal 86: 236–247.

Boonstra R (2013) Reality as the leading cause of stress: rethinking the impact of chronic stress in

nature. Funct Ecol 27: 11–23.

Boonstra R, Hik D, Singleton GR, Tinnikov A (1998) The impact of predator-induced stress on

the snowshoe hare cycle. Ecol Monogr 79: 371–394.

10 Breuner CW, Orchinik M (2001) Seasonal regulation of membrane and intracellular

corticosteroid receptors in the house sparrow brain. J Neuroendocrinol 13: 412–420.

Breuner CW, Orchinik M (2002) Plasma binding proteins as mediators of corticosteroid action in

vertebrates. J Endocrinol 175: 99–112.

Bright GM, Darmaun D (1995) Corticosteroid-binding globulin modulates cortisol concentration

responses to a given production rate. J Clin Endocrinol Metab 80: 764–769.

Brown JS, Kotler BP (2004) Hazardous duty pay and the foraging cost of predation. Ecol Lett 7:

999–1014.

Bryan HM, Darimont CT, Paquet PC, Wynne-Edwards KE, Smits JEG (2014) Stress and

reproductive hormones reflect inter-specific social and nutritional conditions mediated by

resource availability in a bear – salmon system. Conserv Physiol 2: 1–18.

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.

Cooke SJ, O’Connor CM (2010) Making conservation physiology relevant to policy makers and

conservation practitioners. Conserv Lett 3: 159–166.

Cooke SJ, Sack L, Franklin CE, Farrell AP, Beardall J, Wikelski M, Chown SL (2013) What is

conservation physiology? Perspectives on an increasingly integrated and essential

science. Conserv Physiol 1: 1–23.

Creel SR, Fox JE, Hardy A, Sands J, Garrott RA, Peterson RO (2002) Snowmobile activity and

glucocorticoid stress responses in wolves and elk. Conserv Biol 16: 809–814.

Dallman MF, Strack AM, Akana SF, Bradbury MJ, Hanson ES, Scribner KA, Smith M (1993)

Feast and famine: critical role of glucocorticoids with insulin in daily energy flow. Front

Neuroendocrinol 14: 303–347.

11 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: 1–18.

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.

Davies S, Haddad N, Ouyang JQ (2017) Stressful city sounds: glucocorticoid responses to

experimental traffic noise are environmentally dependent. Biol Lett 13: 20170276.

Davis DR, Gabor CR (2015) Behavioral and physiological antipredator responses of the San

Marcos salamander, Eurycea nana. Physiol Behav 139: 145–149. de Kloet ER, Vreugdenhil E, Oitzl MS, Joels M (1998) Brain corticosteroid receptor balance in

health and disease. Endocr Rev 19: 269–301.

Dickens MJ, Romero LM (2013) A consensus endocrine profile for chronically stressed wild

animals does not exist. Gen Comp Endocrinol 191: 177–189.

Donadio E, Buskirk SW (2016) Linking predation risk, ungulate antipredator responses, and

patterns of vegetation in the high Andes. J Mammal 97: 966–977.

Donadio E, Buskirk SW, Novaro AJ (2012) Juvenile and adult mortality patterns in a vicuña

(Vicugna vicugna) population. J Mammal 93: 1536–1544.

Douyon L, Schteingart DE (2002) Effect of obesity and starvation on thyroid hormone, growth

hormone, and cortisol secretion. Endocrinol Metab Clin North Am 31: 173–189.

Eeva T, Hasselquist D, Langefors Å, Tummeleht L, Nikinmaa M, Ilmonen P (2005) Pollution

related effects on immune function and stress in a free-living population of pied

flycatcher Ficedula hypoleuca. J Avian Biol 36: 405–412.

12 Ellenberg U, Setiawan AN, Cree A, Houston DM, Seddon PJ (2007) Elevated hormonal stress

response and reduced reproductive output in yellow-eyed penguins exposed to

unregulated tourism. Gen Comp Endocrinol 152: 54–63.

Fanson K V., Lynch M, Vogelnest L, Miller G, Keeley T (2013) Response to long-distance

relocation in Asian elephants (Elephas maximus): monitoring adrenocortical activity via

serum, urine, and feces. Eur J Wildl Res 59: 655–664.

French SS, Fokidis HB, Moore MC (2008) Variation in stress and innate immunity in the tree

lizard (Urosaurus ornatus) across an urban-rural gradient. J Comp Physiol B Biochem

Syst Environ Physiol 178: 997–1005.

Gaston KJ, Bennie J, Davies TW, Hopkins J (2013) The ecological impacts of nighttime light

pollution: a mechanistic appraisal. Biol Rev 88: 912–927.

Hadany L, Beker T, Eshel I, Feldman MW (2006) Why is stress so deadly? An evolutionary

perspective. Proc Biol Sci 273: 881–885.

Hayward LS, Bowles AE, Ha JC, Wasser SK (2011) Impacts of acute and long-term vehicle

exposure on physiology and reproductive success of the Northern spotted owl. Ecosphere

2: Art65.

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.

Jesmer BR, Goheen JR, Monteith KL, Kauffman MJ (2017) State-dependent behavior alters

endocrine-energy relationship: implications for conservation and management. Ecol Appl

27: 2303–2312.

13 Kenagy GJ, Place NJ (2000) Seasonal changes in plasma glucocorticosteroids of free-living

yellow pine chipmunks: effects of reproduction and capture and handling. Gen Comp

Endocrinol 117: 189–199.

Kitaysky AS, Piatt JF, Wingfield JC (2007) Stress hormones link food availability and population

processes in seabirds. Mar Ecol Prog Ser 352: 245–258.

Kitaysky AS, Romano MD, Piatt JF, Wingfield JC, Kikuchi M (2005) The adrenocortical

response of tufted puffin chicks to nutritional deficits. Horm Behav 47: 606–619.

Koford CB (1957) The vicuña and the puna. Ecol Monogr 27: 153–219.

Landys-Ciannelli MM, Ramenofsky M, Piersma T, Jukema J, Wingfield JC (2002) Baseline and

stress-induced plasma corticosterone during long-distance migration in the bar-tailed

godwit, Limosa lapponica. Physiol Biochem Zool 75: 101–10.

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.

Laundré JW, Hernández L, Ripple WJ (2010) The landscape of fear: ecological implications of

being afraid. Open Ecol J 3: 1–7.

Liesenjohann T, Liesenjohann M, Trebaticka L, Sundell J, Haapakoski M, Ylönen H, Eccard JA

(2015) State-dependent foraging: lactating voles adjust their foraging behavior according

to the presence of a potential nest predator and season. Behav Ecol Sociobiol 69: 747–

754.

Lima SL (1988) Initiation and termination of daily feeding in dark-eyed juncos: influences of

predation risk and energy reserves. Oikos 53: 3–11.

Lima SL, Dill LM (1990) Behavioral decisions made under the risk of predation: a review and

prospectus. Can J Zool 68: 619–640.

14 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.

Madliger CL, Semeniuk CAD, Harris CM, Love OP (2015) Assessing baseline stress physiology

as an integrator of environmental quality in a wild avian population: implications for use

as a conservation biomarker. Biol Conserv 192: 409–417.

McEwen BS, Wingfield JC (2003) Response to commentaries on the concept of allostasis. Horm

Behav 43: 28–30.

Millspaugh JJ, Washburn BE, Milanick M, Slotow R, Van Dyk G (2003) Effects of heat and

chemical treatments on fecal glucocorticoid measurements: implications for sample

transport. Wildl Soc Bull 31: 399–406.

Monclús R, Palomares F, Tablado Z, Martínez-Fontúrbel A, Palme R (2009) Testing the threat-

sensitive predator avoidance hypothesis: physiological responses and predator pressure in

wild rabbits. Oecologia 158: 615–623.

Narayan EJ (2017) Evaluation of physiological stress in Australian wildlife: embracing

pioneering and current knowledge as a guide to future research directions. Gen Comp

Endocrinol 244: 30–39.

Narayan EJ, Cockrem JF, Hero J-M (2013) Sight of a predator induces a corticosterone stress

response and generates fear in an amphibian. PLoS One 8: e73564.

Naslund J, Johnsson JI (2016) State-dependent behavior and alternative behavioral strategies in

brown trout (Salmo trutta L.) fry. Behav Ecol Sociobiol 70: 2111–2125.

Oitzl MS, Champagne DL, van der Veen R, de Kloet ER (2010) Brain development under stress:

hypotheses of glucocorticoid actions revisited. Neurosci Biobehav Rev 34: 853–866.

15 Peckarsky BL, McIntosh AR (1998) Fitness and community consequences of avoiding multiple

predators. Oecologia 113: 565–576.

Pettersson LB, Bronmark C (1993) Trading off safety against food: state-dependent habitat

choice and foraging in Crucian carp. Oecologia 95: 353–357.

Rehnus M, Wehrle M, Palme R (2014) Mountain hares Lepus timidus and tourism: stress events

and reactions. J Appl Ecol 51: 6–12.

Robbins J (1981) Factors altering thyroid hormone metabolism. Environ Health Perspect Vol. 38:

65–70.

Romero LM (2004) Physiological stress in ecology: lessons from biomedical research. Trends

Ecol Evol 19: 249–255.

Romero LM, Dickens MJ, Cyr NE (2009) The reactive scope model - a new model integrating

homeostasis, allostasis, and stress. Horm Behav 55: 375–389.

Rosen DAS, Kumagai S (2008) Hormone changes indicate that winter is a critical period for food

shortages in Steller sea lions. J Comp Physiol B Biochem Syst Environ Physiol 178: 573–

583.

Rosner W (1990) The functions of corticosteroid-binding globulin and sex hormone-binding

globulin: recent advances. Endocr Rev 11: 80–91.

Samuels M, McDaniel P (1997) Thyrotropin levels during hydrocorticosterone infusions that

mimic fasting-induced cortisol elevations: a clinical research study. J Clin Endocrinol

Metab 82: 3700–3704.

Sapolsky RM, Krey LC, McEwen BS (1984) Stress down-regulates corticosteroid receptors in a

site-specific manner in the brain. Endocrinology 114: 287–292.

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.

16 Scheun J, Bennett NC, Ganswindt A, Nowack J (2015) The hustle and bustle of city life:

monitoring the effects of urbanization in the African lesser bushbaby. Sci Nat 102: 57.

Schew WA, McNabb FM, Scanes CG (1996) Comparison of the ontogenesis of thyroid

hormones, growth hormone, and insulin-like growth factor-I in ad libitum and food-

restricted (altricial) European starlings and (precocial) Japanese quail. Gen Comp

Endocrinol 101: 304–316.

Stalder T, Kirschbaum C (2012) Analysis of cortisol in hair - state of the art and future directions.

Brain Behav Immun 26: 1019–1029.

Tennessen JB, Parks SE, Langkilde T (2014) Traffic noise causes physiological stress and

impairs breeding migration behaviour in frogs. Conserv Physiol 2: 1–8.

Tomasi TE (1991) Utilization rates of thyroid hormones in mammals. Comp Biochem Physiol -

Part A Physiol 100: 503–516.

Vera F, Antenucci CD, Zenuto RR (2011) Cortisol and corticosterone exhibit different seasonal

variation and responses to acute stress and captivity in tuco-tucos (Ctenomys talarum).

Gen Comp Endocrinol 170: 550–557.

Vining RF, McGinley RA, Maksvytis J, Ho K (1983a) Salivary cortisol: a better measure of

adrenal cortical function than serum cortisol. Ann Clin Biochem 20: 329–335.

Vining RF, McGinley RA, Symons RG (1983b) Hormones in saliva: mode of entry and

consequent implications for clinical interpretation. Clin Chem 29: 1752–1756.

Vynne C, Booth RK, Wasser SK (2014) Physiological implications of landscape use by free-

ranging maned wolves (Chrysocyon brachyurus) in Brazil. J Mammal 95: 696–706.

Wasser SK, Azkarate JC, Booth RK, Hayward LS, Hunt K, Ayres KL, Vynne C, Gobush K,

Canales-Espinosa D, Rodríguez-Luna E (2010) Non-invasive measurement of thyroid

hormone in feces of a diverse array of avian and mammalian species. Gen Comp

Endocrinol 168: 1–7.

17 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.

Wasser SK, Keim JL, Taper ML, Lele SR (2011) The influences of wolf predation, habitat loss,

and human activity on caribou and moose in the Alberta oil sands. Front Ecol Environ 1:

376–382.

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

Wingfield JC, Maney DL, Breuner CW, Jacobs JD, Lynn SE, Ramenofsky M, Richardson RD

(1998) Ecological bases of hormone—behavior interactions: the “emergency life history

stage. Integr Comp Biol 38: 191–206.

Wingfield JC, Sapolsky RM (2003) Reproduction and resistance to stress: when and how. J

Neuroendocrinol 15: 711–724.

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.

18 Chapter 2

Lack of Correspondence Between Glucocorticoids and Downstream Effects in a Wild Animal

Abstract

Wildlife managers and conservationists are searching for accurate non-invasive metrics to measure population or individual condition in the face of environmental and anthropogenic change. The most commonly measured stress hormones, glucocorticoids (GCs), are attractive candidates for various reasons. First, GCs respond to various factors including anthropogenic disturbance, climate change, food restriction, and predation, making them widely applicable.

Secondly, they are highly conserved throughout vertebrates, allowing for inferences across species. Lastly, they can be collected noninvasively via saliva, feces, hair and feathers. Critical, however, is that what we have learned about downstream fitness-relevant consequences of GCs in the lab can be translated accurately to wild animals in the field. Glucocorticoid measurements must also reflect the condition of individuals or populations as indicated by correlations with fitness-relevant downstream effects. I tested these assumptions in a wild ungulate, the vicuña

(Vicugna vicugna) in the Andes Mountains of Argentina. I captured 33 adult females and measured GC levels in four matrices: blood, saliva, feces, and hair (from three body regions). I also measured fitness-relevant proposed downstream effects of GCs including glucose, hematocrit, C:N ratios, triiodothyronine, progesterone, and body mass index. My results suggest that: 1) GC measurements from different matrices can rarely be used interchangeably; 2) GC measurements from a single matrix were never related to all measured downstream effects; 3) some downstream effects were not related to any GC measurements; and 4) the relationships between a single downstream effect and GC measurements from different matrices could be opposite. My data demonstrate that GC measurements should be used cautiously when assessing condition in wild animals.

19

Introduction

Understanding how environmental and anthropogenic disturbances impact animal survival and reproduction, and how these scale to population dynamics, is of urgent current importance (Cooke et al., 2013; Robert et al., 2015; Trombulak and Frissell, 2007). Such stressors activate the hypothalamic-pituitary-adrenal (HPA) axis of vertebrates, releasing stress-relevant glucocorticoid hormones (GCs) within the body. For example, climate change, including elevated temperatures (Bechshoft et al., 2013; Jessop et al., 2016; Wilkening et al., 2015), extreme climatic events (Meylan et al., 2012; Wingfield et al., 2017), predation risk (Archard et al., 2012;

Davis and Gabor, 2015; Monclús et al., 2009), competition (Creel et al., 2013; Hackländer et al.,

2003; Sapolsky, 2005), and food availability (Fokidis et al., 2012; Kitaysky et al., 2007; Pedersen and Greives, 2008) can alter individual GC levels. Anthropogenic stressors, such as increased traffic intensity (Strasser and Heath, 2013; Tennessen et al., 2014) and tourism (Creel et al., 2002;

Ellenberg et al., 2007; Rehnus et al., 2014), can also alter GC levels. Thus, measures of GCs could provide predictive information on the condition of an individual experiencing a particular stressor (Tarlow and Blumstein, 2007; Wingfield, 2008). As such, GC levels are increasingly being used to indicate condition in wild and captive animal populations (Champagne et al., 2016;

Hayward et al., 2011; Hellgren et al., 1993). Although GCs are essential for allowing animals to respond to and cope with both predictable and unpredictable events (Boonstra, 2013; Wingfield and Kitaysky, 2002), GC levels are intimately tied to individual performance and long-term elevations may negatively impact fitness (Bonier et al., 2009; Boonstra, 2013; Breuner et al.,

2008).

Glucocorticoid levels can be measured in a variety of biological matrices, most often within the blood, but recent non-invasive techniques using saliva, feces, hair, and feathers are becoming more common (Cook, 2012; Otovic and Hutchinson, 2015). I will refer to these

20 different sample types as matrices. Glucocorticoids enter these matrices from the blood through different pathways and reflect changes in blood GCs over different time periods: saliva integrates

GCs over minutes; feces over hours to days; and hair and feathers over weeks to months, as plasma GCs are largely incorporated during hair and feather growth (Warne et al., 2015). Lastly, biological matrices can represent different portions of GCs. For instance, unlike plasma GC concentrations, which can represent both total and free steroids, GCs in saliva, feces, and hair represent the free portion of GCs (Möstl, 2014).

Glucocorticoids also have important downstream effects. Once they enter the cell, they bind to intracellular cytoplasmic receptors and begin working as transcription factors within the cell nucleus (Sapolsky et al., 2000) promoting or inhibiting gene transcription, and changing protein synthesis and composition (Sapolsky, Romero & Munck 2000; Rosner 1990; Bright and

Darmaun 1995). These downstream effects can become evident within minutes of an individual experiencing a stressor and may last for days after the stressor has stopped (Sapolsky et al.,

2000). It is these GC-induced effects that lead to changes in individual growth, immunity, reproduction, and survival (Sapolsky et al., 2000), and make GCs measurements a valuable indicator of individual condition. However, we have little understanding of how GCs measured in the different matrices correlate to these downstream effects, knowledge that is critical if we are to use GCs as a predictive tool for animal condition. Although a combination of GC and downstream effect indices have been used to estimate the impact of a stressor on individual condition (e.g., Boonstra et al., 1998; Eeva et al., 2005; Fanson et al., 2013), few have investigated the relationship between GC measurements in multiple matrices and their downstream effects in wild, free-ranging animals.

My goals in this study were to evaluate: 1) the relationships between GC levels in blood, saliva, feces, and hair (matrices), and 2) whether GC levels in each matrix were correlated to body condition. I collected blood, salivary, fecal, and hair samples from adult female vicuñas

21 (Vicugna vicugna), and subsequently measured glucocorticoids and various downstream measurements of condition immediately after capture (see Table 2-1). I expected that blood and saliva would be strongly and positively correlated, due to similar short GC integration times, but to have a weaker relationship with feces and hair because of their long integration time. Second, I expected GC concentrations in matrices with short integration times to relate with downstream effects that change quickly. Similarly, I expected GC concentrations in matrices with long integration times to relate strongly with downstream effects that change slowly. My results indicate that GC levels in different matrices cannot be used interchangeably and may not be closely related to fitness-relevant downstream effects in some wild animals.

Methods

Study system and capture protocol

This study was conducted in San Guillermo National Park (SGNP), Argentina. Vicuñas are medium-sized ungulates (40-50 kg) and the smallest extant members of the Camelidae family.

Native to South America, vicuñas inhabit at elevations above ~3,300 m (Fowler 2011, Franklin

2011). Vicuñas (n = 33 females) were captured and sampled in fall of 2014 and 2015 with a dart rifle and chemical immobilization using Thiafentanil or Carfentanil; Naltrexone served as the reversal agent (Appendix). Blood, saliva, feces and hair were collected from each individual (see details below).

Glucocorticoid sample collection and analyses

All blood, saliva, and fecal samples collected in the field were immediately placed on ice until processing at the field station (up to 8 hours later), and subsequently stored at -20˚C until they were transported frozen (on dry ice) to The Pennsylvania State University, where they were stored at -20°C until processing.

22 Blood (~5 mL) was collected from the jugular vein of each vicuña with an 18-guage needle and transferred into an EDTA blood tube. Upon returning to the field station, EDTA tubes were centrifuged for 9 min at 4000 gs. Plasma and red blood cells were separated and transferred into 2 mL microcentrifuge tubes and frozen until assayed. Total cortisol was measured using a cortisol 125I radioimmunoassay kit (ImmunoChemTM, MP Biomedicals, Orangeburg, NY) following manufacturer’s instructions.

Saliva was aspirated directly from a vicuña’s mouth using a transfer pipette, transferred into a 2 mL tube, and frozen until assayed. Salivary GCs were measured with an enzyme immunoassay (EIA; No. 1-3002, Salimetrics, State College, PA) following manufacturer’s instructions.

All available fecal pellets were taken directly from the rectum of each individual, placed in 4 oz. clinical jars, and frozen until assayed. Fecal glucocorticoid metabolite (FGCM) levels were measured with an enzyme immunoassay using 11-oxoaetiocholanolone (Goymann et al.,

1999), validated for this species (Arias et al. 2013). Briefly, fecal pellets were lyophilized for 48 hours and ground to a fine powder with mortar and pestle. Then 0.500 ± 0.05 g were extracted with 5 mL 80% methanol for 5 minutes on a vortexer, and centrifuged at 2,500 gs for 30 minutes.

The supernatant was diluted (1:10) with assay buffer before assay.

Glucocorticoid levels in hair can vary with the region of the body from which the hair was collected (Macbeth et al., 2010; Terwissen et al., 2014). Therefore, to answer my questions most robustly, I obtained hair samples from three regions of the body of each individual: the inside thigh of a hind leg, the base of the neck, and the hindquarter (hereafter referred to as leg, neck, and rump samples, respectively). Hair was collected with scissors and clippers. Samples of approximately 3 cm2 were cut to within approximately 0.5 cm of the skin. Samples were stored in paper coin envelopes and placed on ice. The hair was thoroughly ground to a fine powder using a dry ball mixer. I then combined 100 ± 5 mg powder with 100 μL HPLC-grade methanol per mg

23 hair in a glass scintillation vial. Samples were sonicated for 30 min, and then rotated at 60 rpm for

18 h at 50°C. Samples were then centrifuged at 3000 rpm for 45 minutes and 0.7 mL of the resulting supernatant was transferred to a 2.0 mL microcentrifuge tube. The methanol was then evaporated under a stream of nitrogen in a water bath at 50°C. I then used a salivary cortisol enzyme immunoassay (EIA) kit to measure hair cortisol (Bryan et al. 2014; No. 1-3002,

Salimetrics, State College, PA, USA). Hair was not washed prior to assay. Vicuña hair is only

~12 um in diameter (Fowler 2011) and chemical treatment can easily fragment the hair, removing

GCs and other hormones from the interior of the hair during the washing process (Vineis et al.,

2010). Washing hair with even water can cause changes in cortisol levels in some species (Novak et al., 2013). Therefore, it is possible that minimal external contaminants (such as dust) may have affected my results.

Downstream effects measurements and analysis

I quantified several commonly measured and informative indicators of condition (Table

2-1): glucose, hematocrit, C:N ratios, thyroid hormone (triiodothyronine; T3), body mass index

(BMI) residuals, and plasma progesterone levels. Glucose levels should rise in response to an acute stressor (such as capture) within individuals in good condition, indicating the animal’s ability to mobilize energy quickly in response to a stressor (Boonstra et al., 1998; Sapolsky et al.,

2000; Weissman, 1990). Furthermore, elevated glucose levels indicate that energy stores are not depleted. Hematocrit levels (or the percentage of blood which is red blood cells) should be high within individuals in good condition. Red blood cells are costly to produce, and production generally has a positive relationship with condition (Boonstra et al., 1998; Hellgren et al., 1993).

Carbon:Nitrogen ratios are a relatively new metric, especially in mammals, but ratios should be low within individuals in good condition. Carbohydrates are easily digestible and converted into short-term energy, whereas nitrogen can be stored long-term as proteins (Hawlena and Schmitz,

24 2010). Triiodothyronine (T3) is a thyroid hormone responsible for controlling much of the metabolism and regulating body temperature. Triiodothyronine is high in individuals in good condition, allowing for a normal metabolism. As the body goes into energetic distress, T3 levels decrease, dropping the metabolic rate, in an effort to preserve energy stores until conditions improve (Douyon and Schteingart, 2002; Wasser et al., 2010). Body mass index (BMI) is a commonly used metric for body condition and provides important insight into energy reserves.

Thus, body mass index residuals should be high for individuals in good condition (Cabezas et al.,

2007; Graves et al., 2012; Schulte-Hostedde et al., 2005). Glucocorticoids can inhibit reproduction, leading to decreased reproductive hormone levels (Romero and Butler, 2007).

Thus, GCs may provide a proxy for reproductive hormone levels, such as progesterone, which can be used as an index of reproductive potential and investment (Deschner et al., 2003;

Schwarzenberger et al., 1996; Stead-Richardson et al., 2010).

Immediately upon taking a blood sample, blood glucose was measured using a OneTouch

Ultra®2 glucometer (Johnson & Johnson, Chestbrook, PA). Hematocrit was measured in duplicate from whole blood in microcapillary tubes, centrifuged for 9 min. at 4000 gs and read as percentage of the packed red cells in relation to the volume of whole blood. Carbon:Nitrogen ratios were measured from the lyophilized fecal samples (~0.25 g) and analyzed by The

Laboratory for Isotopes and Metals in the Environment at PSU using a Coztech elemental analyzer connected to a Thermo Conflo IV and a Thermo Delva V Advantage analyzer. Fecal T3 was measured with L-triiodothyronine 125I- RIAs (06B254216; MP Biomedicals, Orangeburg,

NY), following Wasser et al. (2010) and manufacturer instructions. Plasma progesterone was measured with an 125I RIA kit (ImmuChemTM, MP Biomedicals Orangeburg, NY), following manufacturer’s instructions. Body weight was measured with a Big Game digital game scale and body length measured from nose to tail tip. From this, I estimated a body mass index (BMI) as the residuals of a linear model of log-transformed body weight by log-transformed body length.

25

Statistical analyses

To test for relationships between physiological measurements, I used multiple linear regressions (function ‘lm’ in R), adding year as a fixed factor. In the field, I also measured the time between when vicuñas first began responding to us behaviorally, an estimate of when the

HPA axis may have been activated to an acute stressor, and the time when I took samples from each biological matrix (hereafter referred to as time lapse). I included this time lapse for each matrix (blood, saliva, feces, and hair) in the models. I used backwards stepwise selection to remove all covariates where p > 0.10, except our independent physiological measurement. All analyses were performed in R (v3.3.1, R Core Team, 2013). Summary data are presented in

Figures 2-5 and 2-6.

Animal ethics statement

All protocols were approved by The Pennsylvania State University Institutional Animal

Care and Use Committee under protocol #45139. Samples were imported under U.S. Fish and

Wildlife Service permitting for threatened animals, with Federal Fish and Wildlife Permit

#MA70993B-2.

Results

Glucocorticoid measurements

Plasma cortisol (Figure 2-1; Table 2-2): Plasma (CORT) increased with salivary CORT and fecal glucocorticoid metabolites (FGCMs), but had no relationship with leg, neck, or rump hair CORT. In all models, plasma CORT tended to increase with blood time lapse. When salivary

26 CORT was the predictor, year was also a significant factor, where plasma CORT was higher in

2014 than 2015 (1.29  0.15 vs. 1.06  0.13 g/dL) .

Salivary CORT (Figure 2-1; Table 2-2): Salivary CORT increased with plasma CORT, but this was driven by one data point which may be an outlier. However, as this point has both high plasma and salivary CORT, there is no indication this result in an error. There was no relationship with FGCMs, leg, neck, or rump hair CORT. When rump hair CORT was the predictor, salivary CORT was greater in 2014 than 2015 (0.212  0.7 vs. 0.15  0.2 pg/dL). When plasma and rump hair CORT were predictors, salivary CORT tended to increase with salivary time lapse.

FGCMs (Figure 2-1; Table 2-2): There were no relationships between FGCMs and

CORT levels in any matrix. When leg hair CORT was the predictor, there was a trend for FGCMs to increase with fecal time lapse. There were no differences in FGCMS between years in any model.

Leg hair CORT (Figure 2-2, Table 2-2): Leg hair CORT was not predicted by CORT in any matrix. There was no effect of leg hair time lapse in any model. In all models, leg hair CORT tended to be greater in 2015 than 2014 (0.53  0.11 vs 0.25  0.03 pg/mg).

Neck hair CORT (Figure 2-2, Table 2-2): Neck hair CORT was not related to CORT in any other matrix. When CORT in every matrix except saliva was the predictor, neck hair CORT tended to decrease with neck hair time lapse. When CORT in every matrix except saliva was the predictor, neck hair CORT tended to be greater in 2015 than 2014 (0.67  0.12 vs. 0.42  0.05 pg/mg).

Rump hair CORT (Figure 2-2, Table 2-2): Rump hair CORT increased with neck hair

CORT, but there was no relationship with plasma or salivary CORT, FGCMs, or leg hair CORT.

There was no effect of rump hair time lapse or year in any model.

27

Downstream measurements

Glucose (Figure 2-3): Glucose levels increased with leg hair CORT, but had no relationship with GC levels in any other matrix. When leg, neck, and rump hair CORT were predictors, glucose tended to increase with blood time lapse. When leg hair CORT was the predictor, glucose was higher in 2015 than 2014 (211  15 vs. 195  11 mmol/L) .

Average hematocrit: Average hematocrit increased with salivary and leg hair CORT, but there was no relationship with plasma CORT, FGCMs, neck hair, or rump hair CORT. There was no effect of blood time lapse in any model. There was no effect of year in any model.

Triiodothyronine (T3): No CORT measurement predicted T3 levels. When plasma, saliva, and rump hair CORT were the predictors, T3 tended to increase with fecal time lapse.

There was no effect of year in any model.

C:N ratios (Figure 2-4): No CORT measurement in any matrix predicted C:N ratios.

Fecal time lapse and year were not significant in any model.

Progesterone (PROG): PROG was not predicted by CORT measurements in any matrix.

There was never a significant effect of blood time lapse. When plasma, neck hair and rump hair

CORT were predictors, PROG was higher in 2015 than 2014 (5.00  0.85 vs. 2.11  0.75 pg/mL).

Body Mass Index (BMI): BMI was not predicted by CORT measurements in any matrix.

When plasma CORT, FGCMs, and leg hair and rump hair CORT were the predictors, BMI tended to be higher in 2014 than 2015 (0.02  0.01 vs. -0.03  0.02). When BMI was the dependent variable, no time lapse covariate was included.

Discussion

I evaluated the relationship between glucocorticoid levels measured in various matrices – blood, saliva, feces, and hair from three different body regions – with each other and with several

28 condition traits that represent potential downstream effects of GCs in a wild mammal. My findings suggest that GC measurements in different matrices are not interchangeable.

Furthermore, although I found that GCs measured in some matrices were related to downstream condition measures, these relationships were specific to certain matrices, indicating that matrices cannot be interchangeably used to infer condition. Below, I suggest mechanisms for the observed relationships, and posit future directions for consideration.

Relationships between matrices

Short-term integrators

I predicted that blood and salivary GCs would be positively correlated due to their similarly rapid integration timelines (as per Kirschbaum & Hellhammer 1994; Negrão et al. 2004;

Peeters et al. 2011). I found support for this prediction, once I accounted for time lapse. In this study, I measured total plasma CORT, but there are also often positive relationships between total plasma CORT, free plasma CORT, and salivary CORT (Fell et al., 1985; Greenwood and Shutt,

1992; Negrão et al., 2004). However, a relationship between total or free plasma CORT and salivary CORT is not universal (Blackshaw and Blackshaw, 1989; Levine et al., 2007).

Short- and long-term integrators

I also predicted that there would be no relationship between plasma and salivary CORT, and fecal glucocorticoid metabolite levels (FGCMs), due to different integration and response times. However, I found a weak relationship between plasma CORT and FGCMs. Fecal GCMs are an integrated measure of exposure to GCs and reflect the free portion of GCs which have been well-metabolized and excreted, while plasma and salivary CORT can respond quickly (within minutes) to a capture stress and were measured as total GCs. Interestingly, in vicuñas, peaks in

29 FGCMs can occur up to 24-48 hours after a stressor (Arias et al., 2013), which is longer than is found in other species. FGCMs often correlate with total plasma CORT (Cavigelli, 1999; Harper and Austad, 2012; Mateo and Cavigelli, 2005). The trend towards a relationship in this study is unexpected, given the acute stress event elevating plasma CORT levels, while fecal time lapse appears to have little effect on FGCM levels.

I also predicted there would be no relationship between plasma and saliva, and hair

CORT, as again, the integration time due to acute stress is quite disparate. My data supported this prediction, and there was no ability of plasma or salivary CORT to predict hair CORT. Although some have found trends towards significant relationships between plasma and hair CORT

(Rothwell et al., 2011; Waterhouse et al., 2017), it is most likely to be seen in a stable environment, or with baseline sample collection, rather than my capture protocol which may have been perceived as an intense, acute stressor. Likewise, although some have found relationships between salivary and hair CORT, subjects were often trained to provide salivary samples (and were thus without additional stress), resulting in positive relationships between salivary and hair

CORT (Bennett and Hayssen, 2010; Davenport et al., 2006; van Holland et al., 2012). However, others have also found no relationships between salivary and hair CORT (Sauvé et al., 2007), which may be caused by higher variability in salivary CORT between subjects than other metrics such as hair CORT (Bennett and Hayssen, 2010). Interestingly, the body region of hair collection may also influence its relationship with saliva. For instance, in beef cattle, salivary CORT and hair CORT collected from the tail and hip were positively related, and there was a trend towards a positive relationship with head hair, but no relationship with neck and shoulder hair CORT

(Moya et al., 2013).

30 Long-term integrators

I found no relationships between FGCMs and hair CORT from any region of the body. I had not expected to find any relationships, given the integration times are disparate, although some have observed relationships between FGCMs and hair CORT (Accorsi et al., 2008;

Mastromonaco et al., 2014; Moya et al., 2013), while others have not (Waterhouse et al., 2017).

However, this again may depend on body region. For instance, in beef cattle, FGCMs and neck and tail hair, but not head, shoulder and hip or hip hair were correlated (Moya et al. 2013). These relationships may also depend on the type of stressor. For example, in chipmunks, FGCMs and hair CORT were related in response to an ACTH challenge, but not in response to a natural, chronic stressor (Mastromonaco et al. 2014). This needs further exploration.

Lastly, I predicted that hair CORT from different body regions would be positively related to one another but found limited supporting evidence. Neck and rump hair CORT were related, while leg hair was not related to either neck nor rump hair. Further, hair CORT from the three different body regions were never related to the same downstream variable. Importantly, hair CORT levels may vary with characteristics such as hair color (Bennett and Hayssen, 2010).

In vicuñas, hair and epidermal characteristics on the dorsal side (soft tan fleece with a higher ratio of primary to secondary hairs) are different than the belly and legs (white hair of thicker diameter;

Chamut, Cancino & Black-Decima 2016).

Conceivably, the lack of relationships between hair CORT levels and other measurements could be due to large variation in hair length between individuals (Wosu et al., 2013). In vicuñas, variation is between 20-55 mm (Quispe et al., 2010). Vicuñas shed annually (Fowler 2011), and

CORT is incorporated throughout the growing period, causing long-haired individuals to have more dispersed CORT, leaving short-haired individuals to have higher CORT levels in my analyses. Therefore, although I standardized the weight of hair I analyzed, the density of cortisol within that hair may vary between individuals. Lastly, another source of variation could be my

31 lab protocol; I chose not to wash the hair. The thinness of vicuña hair (Fowler 2011) also means that the CORT assay processing could have fragmented the hair, releasing hormones from the interior of the hair (Novak et al., 2013; Vineis et al., 2010).

These results suggest that for this species, CORT measured in the different matrices do not provide the same information about an individual’s relative physiological stress status and exposure. Investigators should carefully consider their research goals, including the timeline of experience that they want to capture in the CORT measures, concordant rates of CORT accumulation, and information provided by free versus bound CORT, and select their matrix to sample accordingly. Of course, this decision can be constrained by logistics, however, and investigators need to be cognizant of these factors in interpreting their results.

Ability to predict downstream effects

Crucial to consider in determining in which matrix to measure CORT is what can be inferred from the results. Increasingly, wildlife biologists are using CORT as indicators of organismal and population condition (Busch and Hayward, 2009; Wikelski and Cooke, 2006).

This assumes that the CORT in the sampled matrix correlates with ecologically-relevant downstream effects. I tested this assumption with several physiological traits from indicators of energy reserves (blood glucose) to those of reproductive investment (progesterone levels). I found evidence that CORT in different matrices has few relationships with downstream effects in this wild animal.

I found only that leg hair CORT was related to glucose, although there were trends for both plasma CORT and FGCMs to also be related. However, the direction of the relationship varied with sampling matrix: there was a negative relationship with leg hair CORT but a positive trend with both FGCMs and plasma CORT. I had expected a strong relationship between CORT and glucose levels in matrices that quickly reflect acute stress, including plasma and saliva, as

32 glucose is supposed to be elevated during acute stress (Rand et al., 2002; Sapolsky et al., 2000) and with increasingly adverse conditions (Delehanty and Boonstra, 2011; Kempster et al., 2007).

Despite the fact that GCs are named for their ability to mobilize glucose, some studies have found no relationship between acute stress and glucose or have found that glucose decreased in response to acute stress (Corbel et al., 2010; Deviche et al., 2014; Fokidis et al., 2011).

Glucose mobilization can also be strongly affected by internal condition, including nutritional state, and this relationship is complex. For instance, captive European starlings subjected to a restraint and handling protocol during both day and night had elevated glucose levels during night treatments only (Remage-Healey and Romero, 2001). Glucose levels were not further amplified by the addition of an ACTH injection, suggesting glucose levels might become maximized quickly. In my study, all captures were made during the active period, when the great majority of activity is devoted to foraging. It might be difficult to mobilize glucose from stores, with glucose from foraging and subsequent release of insulin already in place (Remage-Healey et al., 2006). For example, a low dose of injected insulin deactivates the hyperglycemic response to a stress-inducing handling and restraint protocol (Remage-Healey et al., 2006). Interestingly, a subsequent CORT injection of supraphysiological concentrations could not overcome the effects of the low dose of insulin to produce the expected hyperglycemia (Remage-Healey et al., 2006).

Given that the nutritional condition of wild animals is difficult to quantify, and that I would expect high variation in nutritional condition even within a population, it is not surprising to see few relationships with short-term integrators of GCs. The relationships with long-term integrators of GCs (leg hair) need further exploration.

Interestingly, I found that a single short-term and a single long-term integrator were both related to average hematocrit. I found that salivary CORT and leg hair CORT levels were positively correlated with hematocrit. I expected to find a negative relationship between CORT and hematocrit, especially in long-term integrators of GCs (Hellgren et al., 1993; Lochmiller et

33 al., 1986; Moreno et al., 1998), as hematocrit often decreases with increasingly adverse conditions (Kempster et al. 2007a; Delehanty & Boonstra 2011, but see Brachetta et al. 2015).

However, if high hematocrit indicates good condition, as does a high magnitude response to an acute stressor, elevated salivary CORT and elevated hematocrit are in agreement in indicating condition. Average hematocrit levels increase in as little as 30 minutes in vicuñas injected with

ACTH (Bonacic et al., 2003), similar to the time I expect to observe elevations in salivary GCs.

The positive relationship between salivary GCs and average hematocrit could also be related to the ecology of this species. High-altitude animals, such as vicuñas, may need to maintain or increase their oxygen-carrying capacity (provided by hemoglobin in red blood cells) to facilitate rapid or prolonged movement, which can be achieved by increasing the number of red blood cells

(Monge and León-Velarde, 1991; Ramirez et al., 2007; Storz et al., 2010). Thus, decreasing the volume of red blood cells under stressful conditions could be detrimental. Interestingly, I found no correlation between plasma GCs and hematocrit, which needs further exploration.

I had expected to find a positive relationship between GC levels and C:N ratios, but this was not supported by our data. However, the positive relationship between C:N ratios and stress has primarily been examined in invertebrates (Hawlena & Schmitz 2010; Hawlena et al. 2012;

Leroux, Hawlena & Schmitz 2012; van Dievel, Janssens & Stoks 2015; but see Kempster et al.

2007a with birds), and to my knowledge, never has this prediction been tested in mammals.

Elevated CORT can increase the conversion of protein to glucose within the body (Sapolsky et al., 2000). As proteins are the largest nitrogen (N) pool in the body, the N concentrations should decrease as proteins are converted to glucose, and nitrogenous wastes are excreted at a high rate, resulting in increased C:N ratios under chronic stressors (Hawlena and Schmitz, 2010). Further, individuals under chronic stress may also select forage high in carbohydrates which are rich in carbon and more easily digestible, allowing energy obtained to be more readily available than proteins, which are more slowly digested. Recent studies, however, indicate that the mechanisms

34 between stress and stoichiometry are not fully understood (Guariento et al., 2015; van Dievel et al., 2015). For instance, high N incorporation efficiency and lower N excretion were observed in predator-stressed guppies compared to controls (Dalton and Flecker, 2014). Chronic predation risk decreases protein concentrations in damselflies exposed to dragonfly predators (Stoks et al.,

2005). However, my data suggest that in wild mammals, CORT may not be a good predictor of

C:N ratios.

I predicted a negative relationship between progesterone and CORT but found that only

FGCMs tended to have a negative relationship with progesterone. Plasma progesterone can be used as a proxy for potential reproductive success, which decreases under chronic stress (Cyr and

Michael Romero, 2007; Parker and Douglas, 2010). Glucocorticoids can inhibit the hypothalamic-pituitary-gonadal (HPG) axis (Kalantaridou et al., 2004), decreasing reproductive hormone output. In other studies, reproductive hormones remain unchanged in response to acute capture stress (Wingfield et al. 1984, Deviche et al. 2014, 2016), and the lack of a relationship with plasma and saliva CORT is easily rationalized, as the interaction between the HPA and HPG axes are likely not activated quickly enough to detect changes within my blood collection time.

However, I am then left with GC levels in the remaining matrices – feces and hair, which should represent baseline levels. The lack of correlation between these measurements may be again due to my study organism. In vicuñas, plasma progesterone levels do not correlate with the maintenance of the corpus luteum (Fowler 2011), and thus may not be representative of whether a female is pregnant. The use of an ultrasound in the field would be an ideal way to validate if a female was pregnant at the time of capture. However, given the resources and data I have, CORT does not appear to be a good indicator of pregnancy and reproductive potential in this species.

I predicted that there would be a negative relationship between CORT and triiodothyronine (T3) levels. However, I found no relationships between CORT in any matrix and

T3 levels. Many stressors can contribute to elevations in CORT, including nutritional deficits

35 (Jeanniard de Dot et al. 2009). As nutritional deficits increase, T3 concentrations tend to decrease, lowering basal metabolic rate and resting energy expenditure (Schew et al. 1996, Samuels &

McDaniel 1997, Douyon & Schteingart 2002, Kitaysky et al. 2005, Rosen & Kumagai 2008), and thus may be useful to indicate nutritional deficits contributing to elevated CORT. For example,

T3 levels decreased significantly throughout a 28-day restricted diet phase in Stellar sea lions, and only returned to baseline levels at the end of the controlled recovery phase (Jeanniard du Dot et al. 2009). The lack of relationship in our study between CORT and T3 may indicate this population is not experiencing nutritional deficits.

Lastly, I had predicted a negative relationship between CORT and body weight index

(BMI), but found no relationships. A recent metanalysis that found the most common response to chronic stress was a decrease in body weight (Dickens and Romero, 2013), caused by processes such as reduced conversion efficiency of food (Hawlena and Schmitz, 2010). However, in my study, there were a number of factors which could influence the BMI of individuals that I could not account for, including reproductive status. Additionally, the relationship between BMI and

GCs may be nonlinear; however, my limited sample size may be below the detection limit of such a relationship (Schorr et al., 2015). My data suggest that CORT may not be a good predictor of

BMI in wild animals, especially where other important covariates, such as reproductive status, cannot be accounted for.

Conclusions

Together, the results of my field study suggest that: 1) GC measures in one matrix may not correlate to measures in different matrices, 2) GC measures are not always good predictors of downstream effects, and 3) that different matrices may predict different downstream effects. This cautions that researchers need to carefully consider which matrix GCs to measure for their organisms of interest. In making this determination, researchers should consider the timeline of

36 the stress that they want to capture, concordant rates of accumulation of GCs, and the utility of free versus bound CORT. Furthermore, these data caution against using GCs to make inferences about condition without having strong evidence of this relationship within the matrix of interest within their study species. Many of these factors are poorly understood in most species.

As with monitoring population health of wild animals, results from studies in the field of animal welfare suggests there are limitations to the use of GCs to infer animal well-being (see

Otovic and Hutchinson, 2015; Ralph and Tilbrook, 2016; Rushin 1991, for comprehensive reviews). For example, stressful stimuli can evoke behavioral or physiological changes in the absence of detectable changes in GC levels (Owen et al. 2004; Rampacek et al. 1984). For instance, cats with a chronic pain syndrome (feline interstitial cystitis) have an increased neutrophil to leukocyte ratio, a decrease in lymphocytes, and higher expression of behaviors associated with sickness, including decreased appetite and elimination during stressful periods, compared to healthy cats (Stella et al. 2013). However, there was no difference in GC levels between sick and healthy cats (Stella et al. 2013). As these fields progress, it is critical to remember that the release of GCs is just one facet of the stress response. For instance, prior to the release of GCs, the sympathetic adrenal medullary system is activated, where catecholamines epinephrine and norepinephrine are released, causing fight, flight, or freeze behavior (Saplosky et al. 2000). It is thus possible that this system plays a role which is not detected through GC measures. Connections between the complex components of the stress response thus make GCs more difficult to interpret.

This study highlights the complexity of measuring GC and their applicability to understanding downstream effects within free-living vertebrates. It would be useful to examine these relationships across many species in different environments. Comparative evidence of this kind, combined with experimental manipulations to modify GCs and measure outcomes under

37 different scenarios, has great potential to clarify the utility of GCs to predict condition of organisms in the wild.

38 Literature Cited

Accorsi, P.A., Carloni, E., Valsecchi, P., Viggiani, R., Gamberoni, M., Tamanini, C., Seren, E.,

2008. Cortisol determination in hair and faeces from domestic cats and dogs. Gen. Comp.

Endocrinol. 155, 398–402.

Archard, G.A., Earley, R.L., Hanninen, A.F., Braithwaite, V.A., 2012. Correlated behaviour and

stress physiology in fish exposed to different levels of predation pressure. Funct. Ecol.

26, 637–645.

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.

Bechshoft, T.Ø., Sonne, C., Rigét, F.F., Letcher, R.J., Novak, M.A., Henchey, E., Meyer, J.S.,

Eulaers, I., Jaspers, V.L.B., Covaci, A., Dietz, R., 2013. Polar bear stress hormone

cortisol fluctuates with the North Atlantic Oscillation climate index. Polar Biol. 36,

1525–1529.

Bennett, A., Hayssen, V., 2010. Measuring cortisol in hair and saliva from dogs: coat color and

pigment differences. Domest. Anim. Endocrinol. 39, 171–180.

Blackshaw, J.K., Blackshaw, A.W., 1989. Limitation of salivary and blood cortisol

determinations in pigs. Vet. Res. Commun. 13, 265–271.

Bonacic, C., Macdonald, D.W., Villouta, G., 2003. Adrenocorticotrophin-induced stress response

in captive vicuñas (Vicugna vicugna) in the Andes of Chile. Anim. Welf. 12, 369–385.

Bonier, F., Martin, P.R., Moore, I.T., Wingfield, J.C., 2009. Do baseline glucocorticoids predict

fitness? Trends Ecol. Evol. 24, 634–642.

Boonstra, R., 2013. Reality as the leading cause of stress: rethinking the impact of chronic stress

in nature. Funct. Ecol. 27, 11–23.

39 Boonstra, R., Hik, D., Singleton, G.R., Tinnikov, A., 1998. The impact of predator-induced stress

on the snowshoe hare cycle. Ecol. Monogr. 79, 371–394.

Breuner, C.W., Patterson, S.H., Hahn, T.P., 2008. In search of relationships between the acute

adrenocortical response and fitness. Gen. Comp. Endocrinol. 157, 288–295.

Busch, D.S., Hayward, L.S., 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, T.A., Moreno, S., 2007. Physiological stress levels predict

survival probabilities in wild rabbits. Horm. Behav. 51, 313–320.

Cavigelli, S.A., 1999. Behavioural patterns associated with faecal cortisol levels in free-ranging

female ring-tailed lemurs, (Lemur catta). Anim. Behav. 57, 935–944.

Champagne, C.D., Kellar, N.M., Crocker, D.E., Wasser, S.K., Booth, R.K., Trego, M.L., Houser,

D.S., 2016. Blubber cortisol qualitatively reflects circulating cortisol concentrations in

bottlenose dolphins. Mar. Mammal Sci. 33, 134–153.

Chamut, S., Cancino, A.K., Black-Decima, P., 2016. The morphological basis of vicuña wool:

skin and gland structure in Vicugna vicugna (Molina 1782). Small Rumin. Res. 137, 124–

129.

Cook, N.J., 2012. Review: Minimally invasive sampling media and the measurement of

corticosteroids as biomarkers of stress in animals. Can. J. Anim. Sci. 92, 227–259.

Cooke, S.J., Sack, L., Franklin, C.E., Farrell, A.P., Beardall, J., Wikelski, M., Chown, S.L., 2013.

What is conservation physiology? Perspectives on an increasingly integrated and

essential science. Conserv. Physiol. 1, 1–23.

Corbel, H., Geiger, S., Groscolas, R., 2010. Preparing to fledge: the adrenocortical and metabolic

responses to stress in king penguin chicks. Funct. Ecol. 24, 82–92.

40 Creel, S.R., Christianson, D., Liley, S., Winnie, J.A., 2007. Predation risk affects reproductive

physiology and demography of elk. Science 315, 960.

Creel, S.R., Christianson, D., Winnie, J.A., 2011. A survey of the effects of wolf predation risk on

pregnancy rates and calf recruitment in elk. Ecol. Appl. 21, 2847–2853.

Creel, S.R., Dantzer, B., Goymann, W., Rubenstein, D.R., 2013. The ecology of stress: effects of

the social environment. Funct. Ecol. 27, 66–80.

Creel, S.R., Fox, J.E., Hardy, A., Sands, J., Garrott, R.A., Peterson, R.O., 2002. Snowmobile

activity and glucocorticoid stress responses in wolves and elk. Conserv. Biol. 16, 809–

814.

Cyr, N.E., Michael Romero, L., 2007. Chronic stress in free-living European starlings reduces

corticosterone concentrations and reproductive success. Gen. Comp. Endocrinol. 151,

82–89.

Dalton, C.M., Flecker, A.S., 2014. Metabolic stoichiometry and the ecology of fear in Trinidadian

guppies: consequences for life histories and stream ecosystems. Oecologia 176, 691–701.

Davenport, M.D., Tiefenbacher, S., Lutz, C.K., Novak, M.A., Meyer, J.S., 2006. Analysis of

endogenous cortisol concentrations in the hair of rhesus macaques. Gen. Comp.

Endocrinol. 147, 255–261.

Davis, D.R., Gabor, C.R., 2015. Behavioral and physiological antipredator responses of the San

Marcos salamander, Eurycea nana. Physiol. Behav. 139, 145–149.

Delehanty, B., Boonstra, R., 2011. Coping with intense reproductive aggression in male arctic

ground squirrels: the stress axis and its signature tell divergent stories. Physiol. Biochem.

Zool. 84, 417–428.

Deschner, T., Heistermann, M., Hodges, K., Boesch, C., 2003. Timing and probability of

ovulation in relation to sex skin swelling in wild West African chimpanzees, Pan

troglodytes verus. Anim. Behav. 66, 551–560.

41 Deviche, P., Beouche-Helias, B., Davies, S., Gao, S., Lane, S., Valle, S., 2014. Regulation of

plasma testosterone, corticosterone, and metabolites in response to stress, reproductive

stage, and social challenges in a desert male songbird. Gen. Comp. Endocrinol. 203, 120–

131.

Deviche, P., Valle, S., Gao, S., Davies, S., Bittner, S., Carpentier, E., 2016. The seasonal

glucocorticoid response of male Rufous-winged sparrows to acute stress correlates with

changes in plasma uric acid, but neither glucose nor testosterone. Gen. Comp.

Endocrinol. 235, 78–88.

Dickens, M.J., Romero, L.M., 2013. A consensus endocrine profile for chronically stressed wild

animals does not exist. Gen. Comp. Endocrinol. 191, 177–189.

Douyon, L., Schteingart, D.E., 2002. Effect of obesity and starvation on thyroid hormone, growth

hormone, and cortisol secretion. Endocrinol. Metab. Clin. North Am. 31, 173–189.

Eeva, T., Hasselquist, D., Langefors, Å., Tummeleht, L., Nikinmaa, M., Ilmonen, P., 2005.

Pollution related effects on immune function and stress in a free-living population of pied

flycatcher Ficedula hypoleuca. J. Avian Biol. 36, 405–412.

Ellenberg, U., Setiawan, A.N., Cree, A., Houston, D.M., Seddon, P.J., 2007. Elevated hormonal

stress response and reduced reproductive output in yellow-eyed penguins exposed to

unregulated tourism. Gen. Comp. Endocrinol. 152, 54–63.

Escribano-Avila, G., Pettorelli, N., Virgós, E., Lara-Romero, C., Lozano, J., Barja, I., Cuadra,

F.S., Puerta, M., 2013. Testing Cort-Fitness and Cort-Adaptation hypotheses in a habitat

suitability gradient for roe deer. Acta Oecologica 53, 38–48.

Fanson, K. V., Lynch, M., Vogelnest, L., Miller, G., Keeley, T., 2013. Response to long-distance

relocation in Asian elephants (Elephas maximus): monitoring adrenocortical activity via

serum, urine, and feces. Eur. J. Wildl. Res. 59, 655–664.

42 Fell, L., Shutt, D., Bentley, C., 1985. Development of a salivary cortisol method for detecting

changes in plasma “free” cortisol arising from acute stress in sheep. Aust. Vet. J. 62,

403–406.

Fokidis, H.B., des Roziers, M.B., Sparr, R., Rogowski, C., Sweazea, K., Deviche, P., 2012.

Unpredictable food availability induces metabolic and hormonal changes independent of

food intake in a sedentary songbird. J. Exp. Biol. 215, 2920–2930.

Fokidis, H.B., Hurley, L., Rogowski, C., Sweazea, K., Deviche, P., 2011. Effects of captivity and

body condition on plasma corticosterone, locomotor behavior, and plasma metabolites in

curve-billed thrashers. Physiol. Biochem. Zool. 84, 595–606.

Gobush, K., Booth, R.K., Wasser, S.K., 2014. Validation and application of noninvasive

glucocorticoid and thyroid hormone measures in free-ranging Hawaiian monk seals. Gen.

Comp. Endocrinol. 195, 174–182.

Goymann, W., Möstl, E., Van’t Hof, T., East, M.L., Hofer, H., 1999. Noninvasive fecal

monitoring of glucocorticoids in spotted hyenas, Crocuta crocuta. Gen. Comp.

Endocrinol. 114, 340–348.

Graves, G.R., Newsome, S.D., Willard, D.E., Grosshuesch, D.A., Wurzel, W.W., Fogel, M.L.,

2012. Nutritional stress and body condition in the Great Gray Owl (Strix nebulosa)

during winter irruptive migrations. Can. J. Zool. 90, 787–797.

Greenwood, P., Shutt, D., 1992. Salivary and plasma cortisol as an index of stress in goats. Aust.

Vet. J. 69, 161–163.

Guariento, R.D., Carneiro, L.S., Jorge, J.S., Borges, A.N., Esteves, F.A., Caliman, A., 2015.

Interactive effects of predation risk and conspecific density on the nutrient stoichiometry

of prey. Ecol. Evol. 5, 4747–4756.

Hackländer, K., Möstl, E., Arnold, W., 2003. Reproductive suppression in female Alpine

marmots, Marmota marmota. Anim. Behav. 65, 1–8.

43 Harper, J.M., Austad, S.N., 2012. Fecal glucocorticoids: a noninvasive method of measuring

adrenal activity in wild and captive rodents. Physiol. Biochem. Zool. 73, 12–22.

Hawlena, D., Schmitz, O.J., 2010. Physiological stress as a fundamental mechanism linking

predation to ecosystem functioning. Am. Nat. 176, 537–556.

Hawlena, D., Strickland, M.S., Bradford, M. a, Schmitz, O.J., 2012. Fear of predation slows

plant-litter decomposition. Science 336, 1434–1438.

Hayward, L.S., Bowles, A.E., Ha, J.C., Wasser, S.K., 2011. Impacts of acute and long-term

vehicle exposure on physiology and reproductive success of the Northern spotted owl.

Ecosphere 2, Art65.

Hellgren, E.C., Rogers, L.L., Seal, U.S., 1993. Serum chemistry and hematology of black bears:

physiological indexes of habitat quality or seasonal patterns? J. Mammal. 74, 304–315.

Hobson, K.A., Alisauskas, R.T., Clark, R.G., 1993. Stable-nitrogen isotope enrichment in avian

tissues due to fasting and nutritional stress: implications for isotopic analyses of diet.

Condor 95, 388–394.

Hood, L.C., Boersma, P.D., Wingfield, J.C., 1998. The adrenocortical response to stress in

incubating Magellanic penguins (Spheniscus magellanicus). Auk 115, 76–84.

Jessop, T.S., Lane, M.L., Teasdale, L., Stuart-Fox, D., Wilson, R.S., Careau, V., Moore, I.T.,

2016. Multiscale evaluation of thermal dependence in the glucocorticoid response of

vertebrates. Am. Nat. 188, 342–356.

Kalantaridou, S.N., Makrigiannakis, A., Zoumakis, E., Chrousos, G.P., 2004. Stress and the

female reproductive system. J Reprod. Immunol.. 62, 61–68.

Kempster, B., Zanette, L.Y., Longstaffe, F.J., MacDougall-Shackleton, S.A., Wingfield, J.C.,

Clinchy, M., 2007. Do stable isotopes reflect nutritional stress? Results from a laboratory

experiment on song sparrows. Oecologia 151, 365–371.

44 Kirschbaum, C., Hellhammer, D.H., 1994. Salivary cortisol in psychoneuroendocrine research:

recent developments and applications. Psychoneuroendocrinology 19, 313–333.

Kitaysky, A.S., Piatt, J.F., Wingfield, J.C., 2007. Stress hormones link food availability and

population processes in seabirds. Mar. Ecol. Prog. Ser. 352, 245–258.

Leroux, S.J., Hawlena, D., Schmitz, O.J., 2012. Predation risk, stoichiometric plasticity and

ecosystem elemental cycling. Proc. Biol. Sci. 279, 4183–4191.

Levine, A., Zagoory-Sharon, O., Feldman, R., Lewis, J.G., Weller, A., 2007. Measuring cortisol

in human psychobiological studies. Physiol. Behav. 90, 43–53.

Lochmiller, R., Hellgren, E., Varner, L., Grant, W., 1986. Serum and urine biochemical indicators

of nutritional status in adult collared peccaries Tayassu tajacu (Tayassuidae). Comp.

Biochem. Physiol. A: Physiol. 83, 477–488.

Long, J.A., Holberton, R.L., 2004. Corticosterone secretion, energetic condition, and a test of the

migration modulation hypothesis in the hermit thrush (Catharus guttatus), a short-

distance migrant. Auk 121, 1094–1102.

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.

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.

Mateo, J.M., Cavigelli, S.A., 2005. A validation of extraction methods for noninvasive sampling

of glucocorticoids in free-living ground squirrels. Physiol Biochem Zool. 78, 1069–1084.

Meylan, S., Miles, D.B., Clobert, J., 2012. Hormonally mediated maternal effects, individual

strategy and global change. Philos. Trans. R. Soc. B Biol. Sci. 367, 1647–1664.

45 Monclús, R., Palomares, F., Tablado, Z., Martínez-Fontúrbel, A., Palme, R., 2009. Testing the

threat-sensitive predator avoidance hypothesis: physiological responses and predator

pressure in wild rabbits. Oecologia 158, 615–623.

Monge, C., León-Velarde, F., 1991. Physiological adaptation to high altitude: oxygen transport in

mammals and birds. Physiol. Rev. 71, 1135–1172.

Moreno, J., de León, A., Fargallo, J.A., Moreno, E., 1998. Breeding time, health and immune

response in the Chinstrap penguin Pygoscelis antarctica. Oecologia 115, 312–319.

Moya, D., Schwartzkopf-Genswein, K.S., Veira, D.M., 2013. Standardization of a non-invasive

methodology to measure cortisol in hair of beef cattle. Livest. Sci. 158, 138–144.

Negrão, J.A., Porcionato, M.A., de Passillé, A.M., Rushen, J., 2004. Cortisol in saliva and plasma

of cattle after ACTH administration and milking. J. Dairy Sci. 87, 1713–1718.

Novak, M.A., Hamel, A.F., Kelly, B.J., Dettmer, A.M., Meyer, J.S., 2013. Stress, the HPA axis,

and nonhuman primate well-being: a review. Appl. Anim. Behav. Sci. 143, 135–149.

Otovic, P., Hutchinson, E., 2015. Limits to using HPA axis activity as an indication of animal

welfare. ALTEX 32, 41–50.

Owen, M.A., Swaisgood, R.R., Czekala, N.M., Steinman, K., Lindburg, D.G. 2004. Monitoring

stress in captive giant pandas (Ailuropoda melanoleuca): behavioral and hormonal

responses to ambient noise. Zoo Biol. 23, 147-164.

Parker, V.J., Douglas, A.J., 2010. Stress in early pregnancy: maternal neuro-endocrine-immune

responses and effects. J. Reprod. Immunol. 85, 86–92.

Pedersen, A.B., Greives, T.J., 2008. The interaction of parasites and resources cause crashes in a

wild mouse population. J. Anim. Ecol. 77, 370–377.

Peeters, M., Sulon, J., Beckers, J.F., Ledoux, D., Vandenheede, M., 2011. Comparison between

blood serum and salivary cortisol concentrations in horses using an adrenocorticotropic

hormone challenge. Equine Vet. J. 43, 487–493.

46 Quispe, E.C., Ramos, H., Mayhua, P., Alfonso, L., 2010. Fibre characteristics of vicuña (Vicugna

vicugna mensalis). Small Rumin. Res. 93, 64–66.

Ramirez, J.-M., Folkow, L.P., Blix, A.S., 2007. Hypoxia tolerance in mammals and birds: from

the wilderness to the clinic. Annu. Rev. Physiol. 69, 113–143.

Ralph, C., Tilbrook, A. 2016. Invited review: the usefulness of measuring glucocorticoids for

assessing animal welfare.

Rampacek, G.B., Kraeling, R.R., Fonda, E.J. and Barb, C.R., 1984. Comparison of physiological

indicators of chronic stress in confined and non-confined gilts. J. Anim. Sci., 58:401-

408.

Rand, J.S., Kinnaird, E., Baglioni, A., Blackshaw, J., Priest, J., 2002. Acute stress hyperglycemia

in cats is associated with struggling and increased concentrations of lactate and

norepinephrine. J. Vet. Intern. Med. 16, 123–132.

Rehnus, M., Wehrle, M., Palme, R., 2014. Mountain hares Lepus timidus and tourism: Stress

events and reactions. J. Appl. Ecol. 51, 6–12.

Remage-Healey, L., Nowacek, D.P., Bass, A.H., 2006. Dolphin foraging sounds suppress calling

and elevate stress hormone levels in a prey species, the Gulf toadfish. J. Exp. Biol. 209,

4444–4451.

Remage-Healey, L., Romero, L.M., 2001. Corticosterone and insulin interact to regulate glucose

and triglyceride levels during stress in a bird. Am. J. Physiol. Regul. Integr. Comp.

Physiol. 281, R994–R1003.

Robert, K.A., Chambers, B.K., Lesku, J.A., Partecke, J., Chambers, B., 2015. Artificial light at

night desynchronizes strictly seasonal reproduction in a wild mammal. Proc. R. Soc. B

282, 1–7.

Romero, L.M., Butler, L.K., 2007. Endocrinology of stress. Int. J. Comp. Psychol. 20, 89–95.

47 Rothwell, E.S., Bercovitch, F.B., Andrews, J.R.M., Anderson, M.J., 2011. Estimating daily

walking distance of captive African elephants using an accelerometer. Zoo Biol. 30, 579–

591.

Rushin, J. 1991. Problems associated with the interpretation of physiological data in the

assessment of animal welfare. Appl. Anim. Behav. Sci. 28, 381-386.

Sapolsky, R.M., 2005. Review: the influence of social hierarchy on primate health. Science 308,

648–652.

Sapolsky, R.M., Romero, L.M., Munck, A.U., 2000. How do glucocorticoids influence stress

responses? Integrating permissive, suppressive, stimulatory and preparative actions.

Endocr. Rev. 21, 55–89.

Sauvé, B., Koren, G., Walsh, G., Tokmakejian, S., Van Uum, S.H.M., 2007. Measurement of

cortisol in human hair as a biomarker of systemic exposure. Clin. Investig. Med. 30, 183–

192.

Schorr, M., Lawson, E.A., Dichtel, L.E., Klibanski, A., Miller, K.K., 2015. Cortisol measures

across the weight spectrum. J. Clin. Endocrinol. Metab. 100, 3313–3321.

Schulte-Hostedde, A.I., Zinner, B., Millar, J.S., Hickling, G.J., 2005. Restitution of mass-size

residuals: validating body condition indices. Ecology 86, 155–163.

Schwarzenberger, F., Möstl, E., Palme, R., Bamberg, E., 1996. Faecal steroid analysis for non-

invasive monitoring of reproductive status in farm, wild and zoo animals. Anim. Reprod,

Sci. 14, 515–526.

Stead-Richardson, E., Bradshaw, D., Friend, T., Fletcher, T., 2010. Monitoring reproduction in

the critically endangered marsupial, Gilbert’s potoroo (Potorous gilbertii): Preliminary

analysis of faecal oestradiol-17β, cortisol and progestagens. Gen. Comp. Endocrinol. 165,

155–162.

48 Stella, J., Croney, C., Buffington, T. 2013. Effects of stressors on the behavior and physiology of

domestic cats. Appl. Anim. Behav. Sci. 143, 157-163.

Stoks, R., De Block, M., McPeek, M.A., 2005. Alternative growth and energy storage responses

to mortality threats in damselflies. Ecol. Lett. 8, 1307–1316.

Storz, J.F., Scott, G.R., Cheviron, Z.A., 2010. Phenotypic plasticity and genetic adaptation to

high-altitude hypoxia in vertebrates. J. Exp. Biol. 213, 4125–4136.

Strasser, E.H., Heath, J.A., 2013. Reproductive failure of a human-tolerant species, the American

kestrel, is associated with stress and human disturbance. J. Appl. Ecol. 50, 912–919.

Tarlow, E.M., Blumstein, D.T., 2007. Evaluating methods to quantify anthropogenic stressors on

wild animals. Appl. Anim. Behav. Sci. 102, 429–451.

Tennessen, J.B., Parks, S.E., Langkilde, T., 2014. Traffic noise causes physiological stress and

impairs breeding migration behaviour in frogs. Conserv. Physiol. 2, 1–8.

Terwissen, C. V., Mastromonaco, G.F., Murray, D.L., 2014. Enzyme immunoassays as a method

for quantifying hair reproductive hormones in two felid species. Conserv. Physiol. 2, 1–9.

Trombulak, S.C., Frissell, C.A., 2007. Review of ecological effects of roads on terrestrial and

aquatic communities. Conserv. Biol. 14, 18–30. van Dievel, M., Janssens, L., Stoks, R., 2015. Short- and long-term behavioural, physiological

and stoichiometric responses to predation risk indicate chronic stress and compensatory

mechanisms. Oecologia. 181, 347–357. van Holland, B.J., Frings-Dresen, M.H.W., Sluiter, J.K., 2012. Measuring short-term and long-

term physiological stress effects by cortisol reactivity in saliva and hair. Int. Arch. Occup.

Environ. Health 85, 849–852.

Vineis, C., Aluigi, A., Tonin, C., 2010. Outstanding traits and thermal behaviour for the

identification of specialty animal fibres. Text. Res. J. 81, 264–272.

49 Warne, R.W., Proudfoot, G.A., Crespi, E.J., 2015. Biomarkers of animal health: integrating

nutritional ecology, endocrine ecophysiology, ecoimmunology, and geospatial ecology.

Ecol. Evol. 5, 557–566.

Wasser, S.K., Azkarate, J.C., Booth, R.K., Hayward, L.S., Hunt, K., Ayres, K.L., Vynne, C.,

Gobush, K., Canales-Espinosa, D., Rodríguez-Luna, E., 2010. Non-invasive

measurement of thyroid hormone in feces of a diverse array of avian and mammalian

species. Gen. Comp. Endocrinol. 168, 1–7.

Waterhouse, M.D., Sjodin, B., Ray, C., Erb, L., Wilkening, J.L., Russello, M.A., 2017.

Individual-based analysis of hair corticosterone reveals factors influencing chronic stress

in the American pika. Ecol. Evol. 7, 4099–4108.

Weissman, C., 1990. The metabolic response to stress: an overview and update. Anesthesiology.

Wikelski, M., Cooke, S.J., 2006. Conservation physiology. Trends Ecol. Evol. 21, 38–46.

Wilkening, J.L., Ray, C., Varner, J., 2015. Relating sub-surface ice features to physiological

stress in a climate sensitive mammal, the American pika Ochotona princeps. PLoS One

10, 1–17.

Wingfield, J.C., 2008. Organization of vertebrate annual cycles: implications for control

mechanisms. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 363, 425–41.

Wingfield, J.C., O'Reilly, M., Astheimer, L.B., 1994. Modulation of the adrenocortical response

to stress in birds. Amer. Zool. 35, 285-294

Wingfield, J.C., Kitaysky, A.S., 2002. Endocrine responses to unpredictable environmental

events: stress or anti-stress hormons? Integr. Comp. Biol. 42, 600–609.

Wingfield, J.C., Pérez, J.H., Krause, J.S., Word, K.R., González-Gómez, P.L., Lisovski, S.,

Chmura, H.E., 2017. How birds cope physiologically and behaviourally with extreme

climatic events. Philos. Trans. R. Soc. B Biol. Sci. 372, 20160140.

50 Wingfield, J.C., Silverin, B., 1986. Effects of corticosterone on territorial behavior of free- living

male song sparrows Melospiza melodia. Horm. Behav. 20, 405–417.

Wosu, A.C., Valdimarsdóttir, U., Shields, A.E., Williams, D.R., Williams, M.A., 2013. Correlates

of cortisol in human hair: Implications for epidemiologic studies on health effects of

chronic stress. Ann. Epidemiol. 23, 797–811.

1.0

T

R

O

)

C

g

r

n

i

/

a

g

h

p

( p

m 0.5

u R

0.0

0 1 2 3 4 0.0 0.2 0.4 0.6 0 10 20 30 40 0.0 0.2 0.4 0.6 0.0 0.5 1.0 0.0 0.4 0.8 1.2 Plasma CORT Salivary CORT FGCMs Leg hair CORT Neck hair CORT Rump hair CORT (ug/dL) (pg/dL) (ng/g) (pg/ng) (pg/ng) (ng/g) ####################################################################################################

grid::grid.newpage() grid::grid.draw(rbind(plasma,saliva, fgcms, size="last"))

51 )

L 5.0

d

/

g u

( 4.0

T

R 3.0

O

C

a 2.0

m

s a

l 1.0

P

T 0.6

R

)

O

L

C

d 0.4

/

y

r

g

a

p

(

v i

l 0.2

a S 0.0

40

30

s

)

g

M

/ g

C 20

n

G ( F 10

0 1 2 3 4 0.0 0.2 0.4 0.6 0 10 20 30 40 0.0 0.5 1.0 0.0 0.5 1.0 1.5 0.0 0.5 1.0 Plasma CORT Salivary CORT FGCMs Leg hair CORT Neck hair CORT Rump hair CORT (ug/dL) (pg/dL) (ng/g) (pg/ng) (pg/ng) (pg/ng)

Figure 2-1: Trendlines from linear models of relationships between matrices. Dependent variables on the y axis are plasma cortisol (CORT), salivary CORT, and fecal glucocorticoid metabolites (FGCMs). Blank graphs indicate the dependent and independent physiological variables are the same, so no model was run.

134

grid::grid.newpage()

grid::grid.draw(rbind(l, n ,r, size="last")) 52

T 1.0

R

)

O

g

C

n

r

/

i

g a

p 0.5

h

(

g

e L 0.0

1.5

T

R O

) 1.0

C

g

r

n

i

/

a

g

h

p

( 0.5

k

c

e N

0.0

T

R 1.0

O

)

C

g

r

n

i

/

a

g

h

p 0.5

(

p

m u R 0.0 0 1 2 3 4 0.0 0.2 0.4 0.6 0 10 20 30 40 0.0 0.2 0.4 0.6 0.0 0.5 1.0 0.0 0.4 0.8 1.2 Plasma CORT Salivary CORT FGCMs Leg hair CORT Neck hair CORT Rump hair CORT (ug/dL) (pg/dL) (ng/g) (pg/ng) (pg/ng) (ng/g)

Figure 2-2: Trendlines from linear models of relationships between matrices. Dependent variables on the y axis are leg hair cortisol (CORT), neck hair CORT, and rump hair CORT. Blank graphs indicate the dependent and independent physiological variables are the same, so no model was run.

135 grid::grid.newpage() grid::grid.draw(rbind(glucose, hem, t3, size="last" )) 53

) 350

L

/ l

o 300 m

m 250

(

e

s 200

o c

u 150 l G 100

45

)

% (

40

t

i

r c

o 35

t a

m 30

e H 25

5500

)

g

/

g

n (

5000

3 T

4500 0 1 2 3 0.0 0.2 0.4 0.6 0 10 20 30 40 0.0 0.5 1.0 0.0 0.5 1.0 1.5 0.0 0.5 1.0 Plasma CORT Salivary CORT FGCMs Leg hair CORT Neck hair CORT Rump hair CORT (ug/dL) (pg/dL) (ng/g) (pg/ng) (pg/ng) (pg/ng) cn<-(cbind(ggplotGrob(plot.cn.plasma.cort), ggplotGrob(plot.cn.salivary.cort), ggplotGrob(plot.cn.fgcms), ggplotGrob(plot.cn.l), ggplotGrob(plot.cn.n), ggplotGrob(plot.cn.r), size="last"))

## Warning: Removed 12 rows containing non-finite values (stat_smooth). Figure 2-3: Trendlines## Warning: fromRemoved linear models12 rows of relationshipscontaining missingbetween matricesvalues (geom_point). and downstream effects. Dependent variables on the y axis are glucose, hematocrit,## Warning: and triiodothyronineRemoved 1 rows (T3)containing. X-axis independentnon-finite variablesvalues include(stat_smooth). cortisol (CORT) in different matrices, and fecal ## Warning: Removed 1 rows containing missing values (geom_point). glucocorticoid## metabolitesWarning: (FGCMs).Removed 3 Blankrows graphscontaining indicatenon-finite the dependentvalues and(stat_smooth). independent physiological variables are the same, so no model was run. ## Warning: Removed 3 rows containing missing values (geom_point). ## Warning: Removed 2 rows containing non-finite values (stat_smooth). ## Warning: Removed 2 rows containing missing values (geom_point). ## Warning: Removed 1 rows containing non-finite values (stat_smooth). ## Warning: Removed 1 rows containing missing values (geom_point). ## Warning: Removed 1 rows containing non-finite values (stat_smooth).

## Warning: Removed 1 rows containing missing values (geom_point). grid::grid.newpage() grid::grid.draw(cn)

138 grid::grid.newpage() grid::grid.draw(rbind(cn, prog, bwi, size="last")) 54

38

n

e

g o

r 34

t

o

i

i

t

N

:

a

r n

o 30

b

r a

C 26

e n

o 15

r

e

t

)

s L

e 10

g

m

/

o

r

g

p

n

(

a 5

m

s a

l 0 P

0.2

x

e

d

n i

s 0.1

l

t

a

h

u

g d

i 0.0

i

e

s

w

e

r

y −0.1

d o

B −0.2 0 1 2 3 4 0.0 0.2 0.4 0.6 0 10 20 30 40 0.0 0.5 1.0 0.0 0.5 1.0 1.5 0.0 0.5 1.0 Plasma cortisol Salivary cortisol Fecal GCMs Leg hair CORT Neck hair CORT Rump hair CORT (ug/dL) (pg/dL) (ng/g) (pg/ng) (pg/ng) (pg/ng)

Figure 2-4: Trendlines from linear models of relationships between matrices and downstream effects. Dependent variables on the y-axis are

Carbon:Nitrogen ratio, plasma progesterone, and body weight index residuals. X-axis independent variables include cortisol (CORT) in different matrices, and fecal glucocorticoid metabolites (FGCMs).Blank graphs indicate the dependent and independent physiological variables are the same, so no model was run. 50

141 55

Figure 2-5. Relationships between glucocorticoid measurements in multiple matrices. Colors correspond to statistical significance—lightest boxes are significant at p < 0.05, medium-dark boxes indicate 0.05 < p < 0.10, and dark blue boxes indicate p > 0.10.

56

Figure 2-6. Relationships between dependent downstream effects (rows), and independent glucocorticoid measurements in multiple matrices (columns). Colors correspond to significance— lightest boxes are significant at p < 0.05, medium-dark boxes indicate 0.05 < p < 0.10, and dark blue boxes indicate p > 0.10.

57 Table 2-1. Downstream measurements influenced by glucocorticoid levels, predictions for individuals in good and poor condition, the adaptive role of the measure under acute stress, and references for previous studies where the effect has been discussed and measured.

Downstream Measure Predictions under acute Adaptive role under acute References stress stress Glucose Good condition Mobilize energy available as easily Weissman 1990; Boonstra et al. 1998; Sapolsky et al. High concentrations accessible fuel 2000; Landys et al. 2006 Poor condition Low concentrations

Hematocrit Good condition Increase oxygen carrying capacity Boonstra et al., 1998; Delehanty and Boonstra, 2011; High percentage Hellgren et al., 1993; Kempster et al., 2007b Poor condition Low percentage C:N Good condition Increases protein catabolism, increasing Hawlena & Schmitz 2010; Hobson, Alisauskas & Clark Low ratio available energy reserves 1993; Warne et al. 2015 Poor condition High ratio Triiodothyronine (T3) Good condition Decreases to reduce metabolic rate and Douyon & Schteingart 2002; Wasser et al. 2010; Gobush, High concentrations stored energy consumption Booth & Wasser 2014 Poor condition Low concentrations Progesterone (Plasma) Good condition Reduce immediate reproductive efforts to Creel et al., 2011, 2007; Escribano-Avila et al., 2013; High concentrations prioritize energy expenditure to promote Kalantaridou et al., 2004 Poor condition survival over long-term fitness Low concentrations Body Mass Index Good condition Continually mobilizing energy depletes fat Hood, Boersma & Wingfield 1998; Long & Holberton High index reserves in favor of immediate survival 2004; Landys et al. 2006; Dickens & Romero 2013; Poor condition Cabezas et al. 2007 Low index

58 Table 2-2. Model parameters from linear models describing the relationships between glucocorticoid (CORT) in different matrices and downstream effects. Only variables with p < 0.10 were retained for this table, except for all variables of interest, the independent panel. Bold p-values indicate p < 0.10.

INDEPENDENT PANEL DEPENDENT MODEL VARIABLE ESTIMATE STANDARD T P PANEL ERROR SALIVARY CORT Salivary CORT 2.57 0.819 3.137 0.012 Year -0.77 0.218 -3.427 0.007 Blood Time Lapse 0.03 0.015 1.904 0.089 FGCMS FGCMs 0.03 0.011 2.397 0.026 Blood Time Lapse 0.03 0.015 2.132 0.046 LEG HAIR CORT Plasma CORT Leg hair CORT -0.04 0.306 -1.227 0.232 Blood Time Lapse 0.01 0.005 2.520 0.019 NECK HAIR CORT Neck hair CORT 0.003 0.268 0.012 0.990 Blood Time Lapse 0.02 0.005 2.596 0.015 RUMP HAIR CORT Rump hair CORT -0.39 0.414 -0.947 0.352 Blood Time Lapse 0.02 0.007 2.243 0.033 PLASMA CORT Plasma CORT 0.01 0.039 3.529 0.006 Saliva Time Lapse 0.02 0.004 2.011 0.075 FGCMS Salivary CORT FGCMs 0.01 0.003 1.781 0.105 LEG HAIR CORT Leg hair CORT 0.05 0.050 1.091 0.301 NECK HAIR CORT Neck hair CORT 0.01 0.044 0.184 0.858 RUMP HAIR CORT Rump hair CORT -0.01 0.147 -0.065 0.950 PLASMA CORT FGCMs Plasma CORT 6.77 3.351 2.020 0.056 SALIVARY CORT Salivary CORT 45.45 25.513 1.781 0.105

59 LEG HAIR CORT Leg hair CORT -9.95 8.563 -1.162 0.263 Fecal time lapse -0.57 0.317 -1.782 0.095 NECK HAIR CORT Neck hair CORT 3.81 6.200 0.614 0.546 RUMP HAIR CORT Rump hair CORT 13.83 8.857 1.561 0.134 PLASMA CORT Leg hair CORT Plasma CORT -0.06 0.110 -0.573 0.572 Year 0.03 0.108 2.474 0.021 SALIVARY CORT Salivary CORT 1.87 1.545 1.210 0.257 Year 0.37 0.173 2.138 0.061 FGCMS FGCMs -0.00 0.000 -0.658 0.520 Year 0.37 0.130 2.734 0.015 NECK HAIR CORT Neck hair CORT 0.22 0.160 1.309 0.204 Year 0.23 0.115 1.958 0.063 RUMP HAIR CORT Rump hair CORT -0.02 0.203 -0.104 0.918 Year 0.28 0.111 2.468 0.022 PLASMA CORT Neck hair CORT Plasma CORT 0.08 0.130 0.568 0.575 Year 0.34 0.120 2.777 0.010 Neck hair time lapse -0.01 0.004 -2.328 0.028 SALIVARY CORT Salivary CORT 0.41 2.247 0.184 0.858 FGCMS FGCMs 0.00 0.007 0.234 0.818 Year 0.35 0.154 2.285 0.035 Neck hair time lapse -0.02 0.013 -1.744 0.099 LEG HAIR CORT Leg hair CORT 0.35 0.023 1.501 0.148 Year 0.25 0.014 1.750 0.948 Neck hair time lapse -0.01 0.004 -2.300 0.032 RUMP HAIR CORT Rump hair CORT 0.41 0.217 1.881 0.072 Year 0.25 0.119 2.058 0.051 Neck hair time lapse -0.01 0.004 -1.895 0.070 PLASMA CORT Rump hair CORT Plasma CORT -0.10 0.078 -1.333 0.193

60 SALIVARY CORT Salivary CORT -0.044 0.679 -0.065 0.950 FGCMS FGCMs 0.007 0.005 1.561 0.134 LEG HAIR CORT Leg hair CORT 0.070 0.197 0.355 0.726 NECK HAIR CORT Neck hair CORT 0.378 0.137 2.762 0.010 PLASMA CORT Glucose Plasma CORT 28.32 14.480 1.955 0.060 SALIVARY CORT Salivary CORT 49.53 109.080 0.454 0.659 FGCMS FGCMs 1.552 0.800 1.941 0.065 LEG HAIR CORT Leg hair CORT -88.96 34.30 -2.593 0.017 Year 42.82 20.060 2.135 0.044 Blood time lapse 0.93 51.760 1.800 0.086 NECK HAIR CORT Neck hair CORT -3.48 25.143 -0.133 0.895 Blood time lapse 1.39 0.552 2.524 0.018 RUMP HAIR CORT Rump hair CORT -9.13 34.053 -0.268 0.791 Blood time lapse 1.12 0.555 2.009 0.054 PLASMA CORT Average Plasma CORT 0.56 1.445 0.39 0.700 SALIVARY CORT hematocrit Salivary CORT 41.73 14.728 2.833 0.020 FGCMS FGCMs -0.08 0.065 -1.206 0.242 LEG HAIR CORT Leg hair CORT 5.98 2.515 2.378 0.026 NECK HAIR CORT Neck hair CORT 3.32 2.311 1.438 0.162 RUMP HAIR CORT Rump hair CORT 2.49 2.665 0.935 0.358 PLASMA CORT Triiodothyronine Plasma CORT -68.27 98.374 -0.694 0.494 Fecal time lapse 7.08 4.016 1.763 0.091 SALIVARY CORT Salivary CORT 449.06 468.33 0.959 0.363 Fecal time lapse -27.41 10.12 -2.710 0.024 FGCMS FGCMs 1.72 5.400 0.319 0.753 LEG HAIR CORT Leg hair CORT 100.4 257.4 0.39 0.701 NECK HAIR CORT Neck Hair CORT -22.63 187.58 -0.121 0.905 RUMP HAIR CORT Rump hair CORT 7.26 3.695 1.964 0.061

61 PLASMA CORT Carbon:Nitrogen Plasma CORT -0.64 1.105 -0.58 0.567 SALIVARY CORT Salivary CORT -0.83 9.175 -0.091 0.929 FGCMS FGCMs -0.02 0.067 -0.283 0.780 LEG HAIR CORT Leg hair CORT -2.37 2.606 -0.911 0.372 NECK HAIR CORT Neck hair CORT 0.64 2.165 0.297 0.769 RUMP HAIR CORT Rump hair CORT 2.40 2.732 0.879 0.387 PLASMA CORT Plasma Plasma CORT 0.13 0.988 0.128 0.899 Progesterone Year 2.92 1.221 2.391 0.024 SALIVARY CORT Salivary CORT -7.82 8.441 -0.926 0.374 FGCMS FGCMs -0.12 0.062 -1.893 0.072 LEG HAIR CORT Leg hair CORT 1.36 2.484 0.547 0.590 NECK HAIR CORT Neck hair CORT 0.48 2.058 0.234 0.817 Year 2.58 1.394 1.847 0.076 RUMP HAIR CORT Rump hair CORT -0.38 1.527 -0.233 0.818 Year 3.58 0.811 4.410 0.0001 PLASMA CORT BWI Plasma CORT 0.03 0.020 1.513 0.141 Year -0.05 0.025 -1.797 0.083 SALIVARY CORT Salivary CORT 0.18 0.171 1.037 0.322 FGCMS FGCMs -0.00 0.00 -0.592 0.560 Year -0.06 0.030 -2.155 0.043 LEG HAIR CORT Leg hair CORT 0.11 0.050 2.190 0.039 Year -0.03 0.029 -2.873 0.009 NECK HAIR CORT Neck hair CORT -0.05 0.039 -1.272 0.214 RUMP HAIR CORT Rump hair CORT -0.07 0.049 -1.391 0.175 Year -0.05 0.026 -1.799 0.083

62

Chapter 3

Few and Complex Relationships in Glucocorticoids and Fitness- Relevant Downstream Measures Between Habitats in a Wild Animal

Abstract Wildlife managers, conservationists, and researchers are increasingly in need of easily accessible tools to quantify the condition and health of wild animals in the face of global change, including anthropogenic disturbance, invasive species, and habitat degradation. Stress hormones

(glucocorticoids, GCs) have been proposed to fill this need. Critical, however, is that results from laboratory studies indicating the relationships between glucocorticoids in multiple biological matrices and downstream effects (including glucose, hematocrit, triiodothyronine levels, carbon:nitrogen ratios, progesterone levels, and body mass index) accurately predict these relationships in wild animals. Further, they need to carry over under a variety of conditions, including across habitats. Few studies have investigated these relationships in wild animals. Here,

I measured GC levels in blood, feces, and hair and examined their relationships with downstream effects. I found few relationships overall and found that the relationships among the measurements can vary with habitat type. I suggest that to get a holistic view of animal health or condition, GCs in multiple matrices and multiple downstream effects should be collected simultaneously.

Introduction

Physiological condition can impact an animal’s ability to survive and reproduce, thus affecting fitness (Love et al., 2005; Schreck et al., 2001; Silverin, 1986). The hypothalamic- pituitary adrenal (or HP interrenal) axis mediates many of these fitness-relevant conditions, and is highly conserved across vertebrates (Sapolsky et al., 2000). These axes are responsible for releasing stress hormones, or glucocorticoids (GCs), such as cortisol and corticosterone. Under

63 acute stress, this response is adaptive, helping an animal cope and respond appropriately, while chronic stress can result in detrimental effects on fitness-relevant traits, such as decreased immunity (Martin, 2009; McEwen and Wingfield, 2003), reproduction (Boonstra and Singleton,

1993; Tsigos and Chrousos, 2002; Wingfield and Sapolsky, 2003), and condition (Boonstra and

Singleton, 1993; Dickens and Romero, 2013).

When an animal is exposed to a stressor and GCs are released, non-essential processes begin shutting down, and energetic resources are diverted to those systems that are needed most to survive. For instance, under acute stress, glucose is mobilized from stores and its uptake in peripheral tissues is inhibited, allowing overall glucose levels to increase, fueling muscles and the brain to respond to the stressor (Boonstra and Singleton, 1993; Wingfield, 1994). As these glucose stores become depleted under chronic stress conditions, the ability to mobilize glucose in response to an acute stressor decreases. Reproductive hormone levels, such as progesterone in females, may decrease, signaling a decrease in reproductive potential as energy is diverted from future fitness to current survival (Hackländer et al., 2003; Sapolsky et al., 2000; Wingfield and

Sapolsky, 2003). These downstream effects are thought to result from reallocation of energy away from non-critical functions (Romero, 2004; Sapolsky, 2000)

Glucocorticoid levels are widely accepted to serve as a proxy for these downstream effects, therefore reflecting differences in animal condition between habitats. For this reason, GCs are widely used to determine population health (Ahlering et al., 2013; Bauer et al., 2013;

Ellenberg et al., 2007; Hik et al., 2001). However, given that these downstream effects are energetic measures, they may be predicted to vary with environment. For instance, responses to elevated GCs may be stronger if energy is limited (French et al., 2010, 2007). Similarly, the nature of the effects of GCs on downstream trait may vary between habitats. For example, when infection risk is high, immune function should be prioritized and thus not suppressed by GCs as it

64 may be when infection risk is low (Avitsur et al., 2001; McCormick and Langkilde 2018, submitted).

Careful experimental manipulations and familiar study systems have demonstrated a direct causal relationship between GCs and these downstream effects. In contrast to tightly controlled laboratory systems where a single (or at most, a few) parameters are manipulated and firmly constrained to ensure little variation, habitats outside of the laboratory differ in many parameters simultaneously, such as forage quantity and quality, predation risk (and thus both acute and chronic stress), weather and climate, and conspecific density. Furthermore, these characteristics within a single habitat may change over time. Thus, when multiple parameters and stimuli change simultaneously, can we still expect there to be predictable relationships between

GCs and downstream effects? It is in the wild that these selective pressures should act upon individuals and promote phenotypic and genotypic adaptive responses (Boonstra, 2013; Lavergne et al., 2014). Thus, testing effects of elevated GCs on downstream effects across multiple habitats in the wild is integral for better understanding these consequences.

Here, I went to two different habitats and collected both invasive and noninvasive physiological samples from female vicuñas (Vicugna vicugna). I then tested if the relationships between GC measurements from three biological matrices (plasma, feces, and hair) and downstream effects were consistent across the two habitats (Table 3-1). In summary, I found that

GC measurements may not be a good predictor of fitness-relevant downstream effects and, crucially, that any predictions GC measurements may make about fitness-relevant downstream may differ (sometimes in opposite directions) depending on the habitat in which animals reside.

65 Methods

Study organism and study site

Vicuñas (Vicugna vicugna) are medium-sized ungulates (~45-50 kg), endemic to high altitude regions (above 3000 m) of South America. They belong to the Camelid family, and are relatives of alpaca, llamas, and guanacos. Their wool is highly prized because of its fine diameter, which lead to overharvesting, and resulting endangerment until the late 1990’s. My study site,

San Guillermo National Park (SGNP), was established specifically to protect this population of vicuñas, and is free from infrastructure, including roads, electricity, and permanent inhabitants.

As there are few anthropogenic disturbances or habitat alterations to account for, it is therefore an ideal location to study stress.

Pumas (Puma concolor) are the only predators of adult vicuñas in this system (Donadio

2012). In SGNP, 91% of adult vicuña deaths can be attributed to puma predation (Donadio et al.,

2012). As ambush predators, pumas use rocky outcroppings and tall vegetation to stalk prey, and thus, prey are provided with consistent cues of predation risk via habitat features. I captured vicuñas in two habitat types in SGNP, which vary in both risk and foraging potential.

Importantly, vicuñas have high site fidelity (Pritchard, unpublished), and so are consistently exposed to one type of habitat. Meadows (n = 32 vicuña captures; see Chapter 2 for details) are the most productive, and are characterized by dense vegetation, with both high forage quality and quantity. However, pumas heavily utilize these areas—approximately 480% more kills were recorded in the meadows than would be expected given their spatial extent (meadows comprise

4% of the landscape; Donadio 2012). Canyons (n=14 vicuña captures) are characterized by steep clines and rocky outcroppings, with intermediate forage quality and quantity, and are less risky— approximately 90% more kills via predation than would be expected (canyons comprise 15% of the landscape; Donadio 2012).

66 Biological sample collection

Glucocorticoid Sample Collection and Analyses:

In the field, all samples collected were immediately placed on ice until processing at the field station. They were then stored at -20˚C until they were transported on dry ice to

Pennsylvania State University, where they were stored at -80°C until processing.

Blood (~5 mL) was collected from each vicuña via the jugular vein and transferred into an EDTA blood tube. For processing at the field station, EDTA tubes were centrifuged for 9 min at 4000 gs. Plasma and red blood cells were separated into separate microcentrifuge tubes and frozen until assay. Total cortisol was measured using a cortisol 125I radioimmunoassay (RIA) kit

(ImmunoChemTM, MP Biomedicals, Orangeburg, NY) following manufacturer’s instructions.

Saliva was collected from a vicuña’s mouth using a transfer pipette, transferred into a microcentrifuge tube, and frozen until assay. Salivary GCs were measured with an enzyme immunoassay (EIA; No. 1-3002, Salimetrics, State College, PA) following manufacturer’s instructions.

Fecal pellets were taken directly from each individual, placed in 4 oz. clinical jars, and frozen until assay. Fecal GCM levels were evaluated with an EIA validated for this species (Arias et al. 2013), using 11-oxoaetiocholanolone (Goymann et al., 1999), Briefly, pellets were lyophilized for 48 hours and ground to a fine powder with mortar and pestle. 0.500 ± 0.05 g were extracted for 5 minutes on a vortexer with 5 mL of 80% methanol. Pellets were then centrifuged for 2,500 gs for 30 minutes. The supernatant was then diluted (1:10) with assay buffer.

Hair samples were collected with scissors from the base of the neck (~ 3 cm2), clipping as close to the skin as possible. Samples were stored in paper coin envelopes and placed on ice. The hair was thoroughly ground to a fine powder using a dry ball mixer. I then combined 100 ± 5 mg powder with 100 μL HPLC-grade methanol per mg hair in a glass scintillation vial. Samples were

67 sonicated for 30 min, and then rotated at 60 rpm for 18 h at 50°C. Samples were then centrifuged at 3000 rpm for 45 minutes, and 0.7 mL of the resulting supernatant was transferred to a 2.0 mL microcentrifuge tube and the methanol was evaporated under a stream of nitrogen in a water bath at 50°C. I then used a salivary cortisol enzyme immunoassay kit to measure hair cortisol (Bryan et al. 2014; EIA; No. 1-3002, Salimetrics, State College, PA, USA). Hair was not washed prior to assay. Vicuña hair is only ~12 μm in diameter and chemical treatment can easily fragment the hair, removing GCs and other hormones from the interior of the hair during the washing process

(Vineis et al., 2010). Washing hair with even water can cause changes in cortisol levels in some species (Novak et al., 2013). Therefore, it is possible that minimal external contaminants (such as dust) may have affected my results.

Downstream Effects Measurements and Analysis

I quantified some commonly measured and informative indicators of condition: glucose, hematocrit, C:N ratios, thyroid hormone (triiodothyronine; T3), body mass index (BMI), and plasma progesterone (PROG) levels. Glucose levels in response to an acute stressor (such as capture) should be high within individuals in good condition, indicating their ability to mobilize energy quickly in response to a stressor (Sapolsky et al., 2000). Furthermore, it is an indication that their energy stores are not depleted. Hematocrit levels (or the percentage of blood which is red blood cells) should be high within individuals in good condition. Red blood cells are costly to produce, and production generally has a positive relationship with condition (Boonstra et al.,

1998). C:N ratios are a relatively new metric, especially in mammals, but ratios should be low within individuals in good condition. Carbohydrates are easily digestible and converted into short-term energy, whereas nitrogen can be stored long-term as proteins (Hawlena and Schmitz,

2010). Triiodothyronine (T3) is a thyroid hormone responsible for controlling much of the metabolism, and regulating body temperature. Triiodothyronine is high in individuals in good

68 condition, allowing for a normal metabolism. As the body goes into energetic distress, T3 levels decrease, dropping the metabolic rate, in an effort to preserve energy stores until conditions improve (Wasser et al., 2010). Body mass index (BMI) is one of the most commonly used metrics for body condition, and provides important insight into energy reserves. Body mass index should be high for healthy individuals. Progesterone levels are used as an index of reproductive potential and investment, and thus provide a proxy for fitness (Kalantaridou et al., 2004). This measure of potential reproductive success using GCs has previously been used in wildlife physiology (Creel et al., 2011, 2007; Escribano-Avila et al., 2013).

Immediately upon taking a blood sample, blood glucose was measured using a OneTouch

Ultra®2 glucometer (Johnson & Johnson, Chestbrook, PA). Hematocrit was measured in duplicate from whole blood in microcapillary tubes, centrifuged for 9 min. at 4000 gs and read as percentage of the packed red cells in relation to the volume of whole blood. Carbon:Nitrogen ratios were measured from the lyophilized fecal samples (~0.25 g) and analyzed by The

Laboratory for Isotopes and Metals in the Environment at PSU using a Coztech elemental analyzer connected to a Thermo Conflo IV and a Thermo Delva V Advantage analyzer. Fecal T3 was measured with L-triiodothyronine 125I- RIAs (06B254216; MP Biomedicals, Orangeburg,

NY), following Wasser et al. (2010) and manufacturer instructions. Plasma progesterone was measured with an 125I RIA kit (ImmuChemTM, MP Biomedicals Orangeburg, NY), following manufacturer’s instructions. Body weight was measured with a Big Game digital game scale (150

± 0.09 kg) and body length measured from nose to tail tip. From this, I estimated a body mass index (BMI) as the residuals of a linear model of log-transformed body weight by log- transformed body length.

69 Statistical analyses

To determine if there were differences in GC concentrations and in fitness-relevant downstream measures between habitats, I used multivariate analysis of variance (MANOVA).

This allowed me to correct for the inflation of Type I errors by preforming multiple comparisons.

I conducted one MANOVA using all GC measurements and downstream effects combined, with year and habitat as fixed factors, and examined their interaction. I tested the assumptions of normality and homogeneity of variances with Shapiro’s and Levene’s tests, respectively, for each measurement within each habitat type. Data were normalized with a Box-Cox transformation if needed.

Given the evidence for differences in physiological measurements between habitats (see

Results), I then determined if relationships between GC measurements and fitness-relevant downstream effects were consistent between the two habitats. For this question, I used multiple regression models (function lm in R) with GC measurements and downstream effects as dependent variables, and additive and interactive effects of habitat and year as independent variables. I also included the amount of time between when I was recognized by vicuñas as a potential threat (when they began to respond behaviorally) and the time I collected each sample

(hereafter time lapse), and Julian Date as covariates. I used backwards stepwise regression to eliminate variables when their significance to the overall model was p > 0.10, except the independent GC measurement and the additive or interactive effects of habitat which were always retained. All analyses were performed in R Studio (v. 3.4.1; R Core Team 2017).

Animal ethics statement

All protocols were approved by the The Pennsylvania State University Institutional

Animal Care and Use Committee under protocol #45139. Samples were imported under U.S. Fish

70 and Wildlife Service permitting for threatened animals, with Federal Fish and Wildlife Permit

#MA70993B-2.

Results There was no effect of year (F1,9 = 1.623, p = 0.230, Pillai’s Trace = 0.594) in the

MANOVA, and thus year was omitted from the final model. There was, however, a significant effect of habitat (F1,9 = 3.484, p = 0.032, Pillai’s Trace = 0.758), and the interaction between year and habitat (F1,9 = 4.527, p = 0.014, Pillai’s Trace = 0.803).

Plasma CORT: Neither fecal glucocorticoid metabolites (FGCMs; t = -1.985, p = 0.057;

Figure 3-1) nor neck hair CORT (t = 1.381, p = 0.175) predicted plasma CORT, although there was a negative trend with FGCMs. When FGCMs were the predictor, plasma CORT was significantly higher in animals captured in canyons than meadows (t = -3.000, p = 0.006), and in

2014 than 2015 (t = -2.053, p = 0.050). Lastly, the relationship between plasma GCs and FGCMs was different between habitats (e.g. there were interactive effects of habitat; t = 2.795, p = 0.009), where it was negative in meadows, but there was no relationship in canyons. When neck hair

CORT was the predictor, there were no differences in plasma CORT between habitats (t = 0.472, p = 0.640), and the relationships between plasma and neck hair CORT did not vary by habitat (t =

-0.751, p = 0.457). There were no significant covariates in the models when neck hair CORT was the predictor.

Fecal glucocorticoid metabolites (FGCMs): Neither plasma CORT (t = -1.707, p = 0.098) nor neck hair CORT (t = -1.051, p = 0.302) predicted FGCMs. When plasma CORT was the predictor, FGCMs were significantly higher in canyons than meadows. The relationship between

FGCMs and plasma CORT also differed by habitat; in meadows the relationship was negative, and in canyons there was no relationship (t = 2.608, p = 0.014). When neck hair CORT was the predictor, FGCMs did not vary by habitat (t = -0.861, p = 0.397), nor did their relationship with

71 neck hair CORT (t = 0.666, p = 0.511). There were no other significant covariates in the model when neck hair CORT was the predictor.

Neck hair CORT: Neither plasma CORT (t = 1.005, p = 0.322) nor FGCMs (t = -1.153, p

= 0.260) predicted neck hair CORT. When FGCMs were the independent variable, neck hair

CORT tended to be higher in individuals captured in canyons than meadows (t = -1.733, p =

0.095). The relationships between neck hair CORT and plasma CORT, and between neck hair

CORT and FGCMs, was not significantly different between habitats (both t < 0.478, p > 0.355).

In both models, neck hair CORT was higher in 2014 than 2015 (both t >2.228, p < 0.035).

Glucose: No GC measurements predicted glucose: plasma CORT (t = 0.126, p = 0.901;

Figure 3-2), FGCMs (t = 0.469, p = 0.642), neck hair CORT (t = -0.116, p = 0.908). When neck hair CORT was the dependent variable, glucose was higher in individuals captured in meadows than canyons. The relationships between glucose and GC measurements by habitat did not differ in any model (all t < 0.173, all p > 0.864). When both plasma and neck hair CORT were dependent variables, glucose levels tended to increase with glucose time lapse (both t > 1.878, both p < 0.068). In all models, glucose levels decreased with Julian Date (all |t| > 2.786, all p <

0.009).

Hematocrit: No GC measurements predicted hematocrit; plasma (t = 0.147, p = 0.884),

FGCMs (t = 1.963, p = 0.061), neck hair CORT (t = -0.877, p = 0.386). When neck hair CORT was the dependent variable, hematocrit tended to be greater in canyons than meadows (t = -1.970, p = 0.057). When FGCMs were the independent variable, the relationship between hematocrit and FGCMs was positive in the meadow, and negative in the canyon (t = -2.249, p = 0.034).

When FGCMs were the dependent variable, there was also a trend for hematocrit to decrease with blood time lapse (t = -1.858, p = 0.076), and to increase with Julian date (t = 2.028, p = 0.054).

C.N ratios: No GC measurements in any matrix predicted C:N ratios: plasma (t = -0.020, p = 0.984), FGCMs (t = -0.472, p = 0.640), neck hair CORT (t = 1.557, p = 0.128). There was no

72 difference in C:N ratios between habitats (all |t| < 1.202, all p > 0.239). There was no difference in the relationships between C:N ratios and GC measurements in any matrix (all |t| < 1.241, all p

>0.223). There were no significant other covariates in any model.

Plasma progesterone (PROG): Fecal GCMs predicted PROG (t = 2.211, p = 0.035;

Figure 3-3), but plasma (t = -0.452, p = 0.653) and neck hair CORT (t = -1.558, p = 0.128) did not. When FGCMs were the independent variable, PROG was higher in meadows than canyons (t

= 2.171, p = 0.038), and the relationship between PROG and FGCMs was positive in meadows, but negative in canyons (t = -2.600, p = 0.015). When plasma and neck hair CORT were independent variables, PROG was higher in 2015 than 2014 (both t > 2.096, both p < 0.039).

When neck hair CORT was the dependent variable, PROG levels tended to increase with Julian date (t = -1.895, p = 0.066).

Body mass index (BMI): Plasma CORT predicted BMI (t = -3.051, p = 0.004), but

FGCMs (t = 1.298, p = 0.204) and neck hair CORT (t = 0.327, p = 0.745) did not. When plasma

CORT was the dependent variable, BMI was higher in canyons than meadows (t = -3.173, p =

0.003), and the relationship between the two variables was negative in the meadows, but positive in the canyon (t = 3.496, p = 0.001). There were no other significant covariates in any model.

Triiodothyronine (T3): No GC measurements in any matrix predicted T3 levels: plasma

CORT (t = -0.198, p = 0.825), FGCMs (t = 11.52, p = 0.068), neck hair CORT (t = -0.283, p =

0.779). There were no differences in T3 between habitats (all t < 1.217, all p > 0.232). The relationships between T3 and GC measurements did not vary by habitat with any matrix (all t <

0.186, all p > 0.854). There were no other significant covariates in any model.

Discussion

Habitat is widely accepted to cause differences in fitness-relevant traits, providing insight into effects of habitat characteristics on population health across habitats. In this study, I

73 predicted that glucocorticoid (GC) measures in different biological matrices would be related to each other and would predict fitness-relevant downstream effects in a wild, free-ranging mammal.

However, few downstream effects were predicted by GC measures, suggesting that GCs may not be an informative predictor of these important measures in wild animals. Further, these relationships were inconsistent between difference matrices. Importantly, in a few of these relationships, there was an interactive effect of habitat. This study highlights that interpreting

GCs to predict condition or fitness in wild animals should be approached cautiously. I address several possible explanations for the differences between habitats, below, and suggest ways future research should incorporate multiple metrics of stress.

The potential for habitat to affect relationships between physiological measurements is increasingly being documented in a variety of organisms and contexts. For instance, elemental signatures are used to determine natal sites in fish, but environmental parameters interact to influence these ratios: differences in the concentration ratios of Sr:Ca, and Ba:Ca, and 13C and

180 in juvenile black bream (Acanthopagrus butcheri) otoliths were caused by the interactions of temperature and salinity (Elsdon and Gillanders, 2002). Without accounting for these abiotic parameters and their effects on elemental signatures, management implications would be incorrect. In an example of an even more complex relationship, the stress response in pigeons was not only influenced by habitat, but this was further influenced by plumage color (Corbel et al.,

2016). Corticosterone responses were positively correlated with the degree of dark coloration in pigeons from rural habitats, but not from urbanized habitats (Corbel et al., 2016).

Habitat may alter the relationships between GC measurements and downstream effects for numerous reasons. Firstly, more frequent stressors can lead to chronic stress and higher baseline GC levels due to downregulation of the HPA axis (Cyr and Romero, 2007; Rich and

Romero, 2005) and increased stress responses to additional stressors (Graham et al., 2012; Mills et al., 2015). Alternatively, frequent disturbance can lead to acclimation (Hood et al., 1998; Owen

74 et al., 2014; Romero et al., 2009). In this study, individuals may perceive capture and restraint stress differently, affecting GC levels (Ellenberg et al., 2007; Lodjak et al., 2015); individuals in risky meadows may be more sensitized to disturbances given the higher predation risk, relative to safer canyons. Here, FGCMs were positively related to plasma PROG in canyons, but negatively related in canyons. If the HPA axis suppresses the HPG axis or the production of precursor hormones, which can be measurable in some species’ plasma within minutes to a few hours after the onset of an acute stressor, (Deviche et al., 2016, 2014; Lynn et al., 2010; Newman et al.,

2008; Wingfield et al., 1984), then I may be observing sensitization, where stressed individuals in meadows respond weakly to capture and restraint, followed by an inability to respond physiologically, and thus do not decrease plasma PROG levels. If plasma progesterone levels do not response to acute stressors rapidly in vicuñas, then the negative relationship between FGCMs and plasma progesterone levels in meadows is following the general paradigm that stress inhibits reproduction. Interestingly, despite higher average FGCMs levels in canyons relative to meadows, I did not see the same patterns. In canyons, the relationship between FGCMs and plasma progesterone levels was positive and needs further exploration. There are a variety of reasons that GCs may not inhibit reproduction in some animals (reviewed in Wingfield and

Sapolsky, 2003), including animals with short breeding windows, semelparous species, animals where social hierarchy is volatile, seasonal breeders, and aged individuals. However, none of these reasons make sense in this context. Regardless, it is interesting that the same patterns between GC levels and a fitness-relevant downstream effect is not observed in animals in two different habitats.

Secondly, (epi)genetic differences may exist between my populations, although they are considered of the same subspecies. High vs. low CORT responses are heritable in other animals

(Almasi et al., 2010; Evans et al., 2006; Satterlee and Johnson, 1988; Wada et al., 2009). In my study, animals in the meadows should have low CORT responses to frequent stressors, such as

75 predation attempts, which may be advantageous, as the stress response is considered to be energetically costly. In addition, the stress response can decrease reproduction and fitness

(Bonier et al., 2009; Breuner et al., 2008; Crespi et al., 2013), and thus animals in highly stressful conditions may be more successful when they respond to stress with blunted activation of the

HPA axis (Rich and Romero, 2005; Romero et al., 2009).

Thirdly, energetic demands may influence responses to CORT (Madliger et al., 2015), and negative relationships may only be apparent when energy is limited (French et al., 2007).

Further, negative feedback efficiency has also been positively correlated with food availability

(Heath and Dufty, 1998; Romero and Wikelski, 2010). For instance, in king penguins

(Aptendoytes patagonicus), the HPA response is modulated by nutritional condition. Chicks at the end of a fasting and molting stage had a diminished CORT response to capture and restraint than chicks at the beginning of the phase when energy stores had not yet been depleted (Corbel et al.,

2010). In black howler monkeys, there was an interactive effect of habitat on the seasonality of

FGCMs: they displayed seasonality in protected areas, but not in unprotected areas (Rangel-

Negrín et al., 2014). In my system, previous work indicates that animals in meadows are in worse condition (via bone marrow content) at the time of death, relative to animals in canyons (Donadio et al., 2012). Up to 90% of mortality in this system is contributed by predation by pumas, and thus not directly condition, such as starvation (Donadio et al., 2012). Therefore, I would expect animals in meadows to also have lower nutritional condition. However, I see this only with one matrix, glucose, suggesting that the disparity in condition between these two populations may not be sufficient to detect the negative relationships.

Lastly, the downstream responses of CORT may be related to the importance of this trait within each context. For instance, in eastern fence lizards (Sceloporus undulatus) from sites invaded with predatory invasive fire ants, the immune system is bolstered after lifetime experimental exposure to GCs, but in S. undulatus from fire ant free populations, the immune

76 system is weakened by exposure to CORT (McCormick et al., submitted). Indeed, in areas where lizards are frequently stung by fire ants, which break the skin, a strong immune system is critical, despite likely energetic costs of maintaining that resource under “stress”.

Conclusions

This study revealed that, in general, no CORT matrix was a good predictor of downstream effects. Plasma, fecal and neck CORT were correlated with only two of the downstream measures. Relationships between CORT and downstream effects also varied between the two habitats tested. Therefore, although CORT may predict downstream effects in some systems or situations, this should not be assumed and should be tested for specific species and environments.

Although laboratory studies are essential to identify responses to a single or few stimuli, so too is it crucial to expand our observations into wild animals, especially if we begin to use physiology to characterize the population health of wild animals’ physiology (Madliger et al.,

2016). It is critical to acknowledge that 1) responses to multi-stressors can be very different than single stressors (Todgham and Stillman, 2013), and different multi-stressors often characterize different habitats; 2) populations may be genetically different in the way that they respond to stressors; and 3) populations may be habituated, or sensitized to stressors, resulting in downregulation or sensitization to stressors, including capture and restraint. Without examining the relationships between CORT and fitness-relevant downstream effects under different contexts

(habitats), we are likely missing critical effects of habitat. Rather than inferring fitness-relevant downstream effects by measuring CORT, we may be able to better get a holistic view of habitat effects on individuals by measuring both (French et al., 2008).

77 Literature Cited

Ahlering, M.A., Maldonado, J.E., Eggert, L.S., Fleischer, R.C., Western, D., Brown, J.L., 2013.

Conservation outside protected areas and the effect of human-dominated landscapes on

stress hormones in Savannah elephants. Conserv. Biol. 27, 569–575.

Almasi, B., Jenni, L., Jenni-Eiermann, S., Roulin, A., 2010. Regulation of stress response is

heritable and functionally linked to melanin-based coloration. J. Evol. Biol. 23, 987–996.

Avitsur, R., Stark, J.L., Sheridan, J.F., 2001. Social stress induces glucocorticoid resistance in

subordinate animals. Horm. Behav. 39, 247–257.

Bauer, C.M., Skaff, N.K., Bernard, A.B., Trevino, J.M., Ho, J.M., Romero, L.M., Ebensperger,

L.A., Hayes, L.D., 2013. Habitat type influences endocrine stress response in the degu

(Octodon degus). Gen. Comp. Endocrinol. 186, 136–144.

Bonier, F., Martin, P.R., Moore, I.T., Wingfield, J.C., 2009. Do baseline glucocorticoids predict

fitness? Trends Ecol. Evol. 24, 634–642.

Boonstra, R., 2013. The ecology of stress: a marriage of disciplines. Funct. Ecol. 27, 7–10.

Boonstra, R., Hik, D., Singleton, G.R., Tinnikov, A., 1998. The impact of predator-induced stress

on the snowshoe hare cycle. Ecol. Monogr. 79, 371–394.

Boonstra, R., Singleton, G.R., 1993. Population declines in the snowshoe hare and the role of

stress. Gen. Comp. Endocrinol. 91, 126–143.

Breuner, C.W., Patterson, S.H., Hahn, T.P., 2008. In search of relationships between the acute

adrenocortical response and fitness. Gen. Comp. Endocrinol. 157, 288–295.

Corbel, H., Geiger, S., Groscolas, R., 2010. Preparing to fledge: the adrenocortical and metabolic

responses to stress in king penguin chicks. Funct. Ecol. 24, 82–92.

Corbel, H., Legros, A., Haussy, C., Jacquin, L., Gasparini, J., Karimi, B., Frantz, A., 2016. Stress

response varies with plumage colour and local habitat in feral pigeons. J. Ornithol. 157,

825–837.

78 Creel, S.R., Christianson, D., Liley, S., Winnie, J.A., 2007. Predation risk affects reproductive

physiology and demography of elk. Science 315, 960.

Creel, S.R., Christianson, D., Winnie, J.A., 2011. A survey of the effects of wolf predation risk on

pregnancy rates and calf recruitment in elk. Ecol. Appl. 21, 2847–2853.

Crespi, E.J., Williams, T.D., Jessop, T.S., 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.

Cyr, N.E., Michael Romero, L., 2007. Chronic stress in free-living European starlings reduces

corticosterone concentrations and reproductive success. Gen. Comp. Endocrinol. 151,

82–89.

Deviche, P., Beouche-Helias, B., Davies, S., Gao, S., Lane, S., Valle, S., 2014. Regulation of

plasma testosterone, corticosterone, and metabolites in response to stress, reproductive

stage, and social challenges in a desert male songbird. Gen. Comp. Endocrinol. 203, 120–

131.

Deviche, P., Valle, S., Gao, S., Davies, S., Bittner, S., Carpentier, E., 2016. The seasonal

glucocorticoid response of male Rufous-winged sparrows to acute stress correlates with

changes in plasma uric acid, but neither glucose nor testosterone. Gen. Comp.

Endocrinol. 235, 78–88.

Dickens, M.J., Romero, L.M., 2013. A consensus endocrine profile for chronically stressed wild

animals does not exist. Gen. Comp. Endocrinol. 191, 177–189.

Donadio, E., Buskirk, S.W., Novaro, A.J., 2012. Juvenile and adult mortality patterns in a vicuña

(Vicugna vicugna) population. J. Mammal. 93, 1–9.

Ellenberg, U., Setiawan, A.N., Cree, A., Houston, D.M., Seddon, P.J., 2007. Elevated hormonal

stress response and reduced reproductive output in yellow-eyed penguins exposed to

unregulated tourism. Gen. Comp. Endocrinol. 152, 54–63.

79 Elsdon, T.S., Gillanders, B.M., 2002. Interactive effects of temperature and salinity on otolith

chemistry: challenges for determining environmental histories of fish. Can J Fish Aquat

Sci 59, 1796–1808.

Escribano-Avila, G., Pettorelli, N., Virgós, E., Lara-Romero, C., Lozano, J., Barja, I., Cuadra,

F.S., Puerta, M., 2013. Testing Cort-Fitness and Cort-Adaptation hypotheses in a habitat

suitability gradient for roe deer. Acta Oecologica 53, 38–48.

Evans, M.R., Roberts, M.L., Buchanan, K.L., Goldsmith, A.R., 2006. Heritability of

corticosterone response and changes in life history traits during selection in the zebra

finch. J. Evol. Biol. 19, 343–352.

French, S.S., Denardo, D.F., Greives, T.J., Strand, C.R., Demas, G.E., 2010. Human disturbance

alters endocrine and immune responses in the Galapagos marine iguana (Amblyrhynchus

cristatus). Horm. Behav. 58, 792–799.

French, S.S., Fokidis, H.B., Moore, M.C., 2008. Variation in stress and innate immunity in the

tree lizard (Urosaurus ornatus) across an urban-rural gradient. J. Comp. Physiol. B

Biochem. Syst. Environ. Physiol. 178, 997–1005.

French, S.S., McLemore, R., Vernon, B., Johnston, G.I.H., Moore, M.C., 2007. Corticosterone

modulation of reproductive and immune systems trade-offs in female tree lizards: long-

term corticosterone manipulations via injectable gelling material. J. Exp. Biol. 210,

2859–2865.

Goymann, W., Möstl, E., Van’t Hof, T., East, M.L., Hofer, H., 1999. Noninvasive fecal

monitoring of glucocorticoids in spotted hyenas, Crocuta crocuta. Gen. Comp.

Endocrinol. 114, 340–348.

Graham, S.P., Freidenfelds, N.A., McCormick, G.L., Langkilde, T., 2012. The impacts of

invaders: basal and acute stress glucocorticoid profiles and immune function in native

lizards threatened by invasive ants. Gen. Comp. Endocrinol. 176, 400–408.

80 Hackländer, K., Möstl, E., Arnold, W., 2003. Reproductive suppression in female Alpine

marmots, Marmota marmota. Anim. Behav. 65, 1–8.

Hawlena, D., Schmitz, O.J., 2010. Herbivore physiological response to predation risk and

implications for ecosystem nutrient dynamics. Proc. Natl. Acad. Sci. U. S. A. 107,

15503–15507.

Hik, D., McColl, C.J., Boonstra, R., 2001. Why are Arctic ground squirrels more stressed in the

boreal forest than in alpine meadows? Ecoscience 8, 275–288.

Hood, L.C., Boersma, P.D., Wingfield, J.C., 1998. The adrenocortical response to stress in

incubating Magellanic penguins (Spheniscus magellanicus). Auk 115, 76–84.

Kalantaridou, S.N., Makrigiannakis, A., Zoumakis, E., Chrousos, G.P., 2004. Stress and the

female reproductive system. J. Reprod. Immunol. 62, 61–68.

Lavergne, S.G., McGowan, P.O., Krebs, C.J., Boonstra, R., 2014. Impact of high predation risk

on genome-wide hippocampal gene expression in snowshoe hares. Oecologia 176, 613–

624.

Lodjak, J., Mägi, M., Rooni, U., Tilgar, V., 2015. Context-dependent effects of feather

corticosterone on growth rate and fledging success of wild passerine nestlings in

heterogeneous habitat. Oecologia 179, 937–946.

Love, O.P., Chin, E.H., Wynne‐Edwards, K.E., Williams, T.D., 2005. Stress hormones: a link

between maternal condition and sex‐biased reproductive investment. Am. Nat. 166, 751–

766.

Lynn, S.E., Stamplis, T.B., Barrington, W.T., Weida, N., Hudak, C.A., 2010. Food, stress, and

reproduction: short-term fasting alters endocrine physiology and reproductive behavior in

the zebra finch. Horm. Behav. 58, 214–222.

81 Madliger, C.L., Cooke, S.J., Crespi, E.J., Funk, J.L., Hultine, K.R., Hunt, K.E., Rohr, J.R.,

Sinclair, B.J., Suski, C.D., Willis, C.K.R., Love, O.P., 2016. Success stories and

emerging themes in conservation physiology. Conserv. Physiol. 4, 1–17.

Madliger, C.L., Semeniuk, C.A.D., Harris, C.M., Love, O.P., 2015. Assessing baseline stress

physiology as an integrator of environmental quality in a wild avian population:

implications for use as a conservation biomarker. Biol. Conserv. 192, 409–417.

Martin, L.B., 2009. Stress and immunity in wild vertebrates: timing is everything. Gen. Comp.

Endocrinol. 163, 70–76.

McEwen, B.S., Wingfield, J.C., 2003. The concept of allostasis in biology and biomedicine.

Horm. Behav. 43, 2–15.

Mills, S.C., Beldade, R., Chabanet, P., Bigot, L., O’Donnell, J.L., Bernardi, G., 2015. Ghosts of

thermal past: reef fish exposed to historic high temperatures have heightened stress

response to further stressors. Coral Reefs 34, 1255–1260.

Newman, A.E.M., Pradhan, D.S., Soma, K.K., 2008. Dehydroepiandrosterone and corticosterone

are regulated by season and acute stress in a wild songbird: jugular versus brachial

plasma. Endocrinology 149, 2537–2545.

Novak, M.A., Hamel, A.F., Kelly, B.J., Dettmer, A.M., Meyer, J.S., 2013. Stress, the HPA axis,

and nonhuman primate well-being: a review. Appl. Anim. Behav. Sci. 143, 135–149.

Owen, D.A.S., Carter, E.T., Holding, M.L., Islam, K., Moore, I.T., 2014. Roads are associated

with a blunted stress response in a North American pit viper. Gen. Comp. Endocrinol.

202, 87–92.

R Core Team (2017). R: A language and environment for statistical computing. R Foundation for

Statistical Computing, Vienna, Austria. URL

82 Rangel-Negrín, A., Coyohua-Fuentes, A., Chavira, R., Canales-Espinosa, D., Dias, P.A.D., 2014.

Primates living outside protected habitats are more stressed: the case of black howler

monkeys in the Yucatán peninsula. PLoS One 9. e112329

Rich, E.L., Romero, L.M., 2005. Exposure to chronic stress downregulates corticosterone

responses to acute stressors. Am. J. Physiol. Regul. Integr. Comp. Physiol. 288, R1628–

R1636.

Romero, L.M., 2004. Physiological stress in ecology: lessons from biomedical research. Trends

Ecol. Evol. 19, 249–255.

Romero, L.M., Dickens, M.J., Cyr, N.E., 2009. The reactive scope model - a new model

integrating homeostasis, allostasis, and stress. Horm. Behav. 55, 375–389.

Sapolsky, R.M., 2000. Stress hormones: good and bad. Neurobiol.Dis. 7, 540–542.

Sapolsky, R.M., Romero, L.M., Munck, A.U., 2000. How do glucocorticoids influence stress

responses? Integrating permissive, suppressive, stimulatory and preparative actions.

Endocr. Rev. 21, 55–89.

Satterlee, D.G., Johnson, W.A., 1988. Selection of Japanese quail for contrasting blood

corticosterone response to immobilization. Poult. Sci. 67, 25–32.

Schreck, C.B., Contreras-Sanchez, W., Fitzpatrick, M.S., 2001. Effects of stress on fish

reproduction, gamete quality, and progeny. Aquaculture 197, 3–24.

Silverin, B., 1986. Corticosterone-binding proteins and behavioral effects of high plasma levels of

corticosterone during the breeding period in the pied flycatcher. Gen. Comp. Endocrinol.

64, 67–74.

Todgham, A.E., Stillman, J.H., 2013. Physiological responses to shifts in multiple environmental

stressors: relevance in a changing world. Integr. Comp. Biol. 53, 539–544.

Tsigos, C., Chrousos, G.P., 2002. Hypothalamic-pituitary-adrenal axis, neuroendocrine factors

and stress. J. Psychosom. Res. 53, 865–871.

83 Vineis, C., Aluigi, A., Tonin, C., 2010. Outstanding traits and thermal behaviour for the

identification of specialty animal fibres. Text. Res. J. 81, 264–272.

Wada, H., Salvante, K.G., Wagner, E., Williams, T.D., Breuner, C.W., 2009. Ontogeny and

individual variation in the adrenocortical response of zebra finch (Taeniopygia guttata)

nestlings. Physiol. Biochem. Zool. 82, 325–331.

Wasser, S.K., Azkarate, J.C., Booth, R.K., Hayward, L.S., Hunt, K., Ayres, K.L., Vynne, C.,

Gobush, K., Canales-Espinosa, D., Rodríguez-Luna, E., 2010. Non-invasive

measurement of thyroid hormone in feces of a diverse array of avian and mammalian

species. Gen. Comp. Endocrinol. 168, 1–7.

Wingfield, J.C., O'Reilly, M., Astheimer, L.B., 1994. Modulation of the adrenocortical response

to stress in birds. Amer. Zool. 35, 285-294

Wingfield, J.C., Sapolsky, R.M., 2003. Reproduction and resistance to stress: when and how. J.

Neuroendocrinol. 15, 711–724.

Wingfield, J.C., Smith, J.P., Farner, D.S., 1982. Endocrine responses of white-crowned sparrows

to environmental stress. Condor 84, 399–409.

84

3 3

)

L

d

/ g

u 2 2

(

T

R

O

C

a

m 1 1

s

a

l P

0 0 0 20 40 0.0 0.5 1.0 Fecal GCM (ng/g) Neck hair CORT (pg/ng)

40 40

)

g

/

g

n

(

s

M 20 20

C

G

l

a

c

e F

0 0 0.5 1.5 2.5 0.0 0.5 1.0 Plasma CORT (ug/dL) Neck hair CORT (pg/ng)

1.0 1.0

)

g

n

/

g

p

(

T R

O 0.5 0.5

C

r

i

a

h

k

c

e N

0.0 0.0 0.5 1.5 2.5 0 20 40 Plasma CORT (ug/dL) Fecal GCM (ng/g)

Figure 3-1. Multiple linear regressions of glucocorticoid concentrations in different matrices plotted with 95% confidence intervals. Single solid black lines indicate models where there were

23 no effects of habitat on the relationships between the two glucocorticoid measurements. When there were additive or interactive effects of habitat, each habitat has been graphed separately.

Solid blue lines indicate canyons, and red dotted lines indicate meadows.

All figures 85

300

)

L

/

l o

m 200

m

(

e

s

o

c

u l G 100

50 0 1 2 3 0 20 40 0.0 0.5 1.0 1.5 Plasma CORT (ug/dL) Fecal GCMs (ng/g) Neck hair CORT (pg/ng)

45

)

%

(

t 40

i

r

c

o

t a

m 35

e

h

e

g

a r

e 30

v A

25 0 1 3 0 20 40 0.0 0.5 1.0 Plasma CORT (ug/dL) Fecal GCMs (ng/g) Neck hair CORT (pg/ng)

50

n e

g 40

o

r

t

i

N

:

n

o b

r 30

a C

20 0 1 2 3 0 20 40 0.0 0.5 1.0 1.5 Plasma CORT (ug/dL) Fecal GCMs (ng/g) Neck hair CORT (pg/ng)

Figure 3-2. Multiple linear regressions of glucocorticoid concentrations and downstream effects plotted with 95% confidence intervals. Single solid black lines indicate models where there were 46 no effects of habitat on the relationships between the two glucocorticoid measurements. When there were additive or interactive effects of habitat, each habitat has been graphed separately.

Solid blue lines indicate canyons, and red dotted lines indicate meadows.

86

8

e

n 6

o

r

e

t

)

s

L

e

g

m /

o 4

r

g

p

n

(

a

m s

a 2

l P

0 0 1 2 3 0 20 40 0.0 0.5 1.0 1.5 Plasma cortisol (ug/dL) Fecal GCMs (ng/g) Neck hair CORT (pg/ng)

0.2 s

l 0.1

a

u

d i

s 0.0

e

r

I W

B −0.1

−0.2

0 1 2 3 0 20 40 0.0 0.5 1.0 1.5 Plasma CORT (ug/dL) Fecal GCMs (ng/g) Neck hair CORT (pg/ng)

5500

)

g

/

g

n

(

3 5000 T

4500 0 1 2 3 0 20 40 0.0 0.5 1.0 1.5 Plasma CORT (ug/dL) Fecal GCMs (ng/g) Neck hair CORT (pg/ng)

Figure 3-3. Results from multiple linear regressions plotted with 95% confidence intervals. Single solid black lines indicate models where there were no effects of habitat on the relationships 47 between the two glucocorticoid measurements. When there were additive or interactive effects of habitat, each habitat has been graphed separately. Solid blue lines indicate canyons, and red dotted lines indicate meadows.

87

Table 3-1 Expected and observed relationships between plasma cortisol concentrations (CORTc), fecal glucocorticoid metabolites (FGCMs), and neck hair cortisol concentrations and downstream effects. A “+” indicates a positive relationship, “-" indicates a negative relationship, and “=” indicates no relationship.

Dependent Variable

Glucose Average Hematocrit C:N Plasma Progesterone BMI T3

Expected Observed Expected Observed Expected Observed Expected Observed Expected Observed Expected Observed

Plasma + + + = - = + = + + Habitat - = CORTc * Habitat

FGCMs - + - = + = - - = + = * Habitat Neck - + - = + + - = - = + = hair CORTc

88 Chapter 4

Glucocorticoids and Triiodothyronine do not Correlate with Behavior in a Wild Mammal

Abstract

State-dependent foraging theory states that animals should make decisions about foraging based on their internal state, where animals with fewer energetic reserves should prioritize foraging over other behavior, including anti-predatory behaviors. However, few studies have investigated these trade-offs at an individual level in wild animals. I used vicuñas (Vicugna vicugna) in the high

Andes Mountains of Argentina as a study species. I observed and sampled vicuñas in two habitats, which differ in predation risk and forage quality, and predicted that animals in riskier meadows with higher food quality would prioritize vigilance (an antipredator response) over foraging, whereas animals in safer canyons with lower food quality would prioritize foraging over vigilance. Using non-invasively collected fecal samples from individuals, I measured glucocorticoids (GCs) and thyroid hormones (THs) as indicators of fear and nutritional stress, respectively. I recorded 20-minute behavioral observations, and investigated the relationships between internal condition (GCs and THs), foraging, and vigilance. I found no relationships between internal condition and behavior, although I found differences in behavior between habitats, where vigilance was approximately 10% higher in the “safer” canyons than the meadows. I suggest that state-dependent foraging may be difficult to observe in large mammals under baseline conditions.

Introduction

Food consumption provides the energy essential for reproduction, influencing subsequent individual fitness and population dynamics (Parker et al., 2009; Pedersen and Greives, 2008;

Taylor et al., 2005). Simultaneously, animals must balance other needs, such as avoiding

89 predation, which are necessary to stay alive to achieve those goals but can themselves be energetically expensive (Bourdeau et al., 2016; Buskirk, 2000; Persons et al., 2002). For example, animals are particularly susceptible to predation when foraging (Stephens et al. 2007), leading to trade-offs between foraging and anti-predator behaviors (Houston et al., 1993; Lima, 1998; Lima and Dill, 1990). Therefore, animals in poor energetic condition or that have high energetic demands, may be required to take risks to reach a minimum energy requirement for daily maintenance and future reproduction (Beale and Monaghan, 2004; Lima and Bednekoff, 1999;

Wirsing et al., 2007; Ydenberg et al., 2007).

Previous work with animals has focused on behavioral responses to external conditions including predation risk, food availability, and their interactions (Brown, 1999; Searle et al.,

2008; Sinclair and Arcese, 1995). However, internal condition can also strongly influence behavioral trade-offs (Nonacs, 2001) with important fitness consequences (Gauthier-Clerc et al.,

2001; Olsson, 1997). Specifically, the theory of state-dependent foraging addresses interactions between internal condition and behavior, suggesting that when food is scarce, animals should accept more risk in order to obtain the daily required energy at the cost of other behaviors, including vigilance and refuge use (Mangel and Clark, 1986; McNamara and Houston, 1996).

This theory was largely established with small animals with high metabolic rates (Caraco et al.,

1990; Carter et al., 2016; Croy and Hughes, 1991), with one assumption being that the animal would starve if it did not obtain sufficient forage daily (Gilby and Wrangham, 2007; Stephens,

1981). This holds for several small species, such as willow tits (Parus montanus), for which 90% of their field metabolic rate (kJ/day) is attributed to maintenance (thermoregulation and maintenance metabolism; Moreno et al., 1988). However, this theory may not extend well to animals who are unlikely to reach starvation within shorter periods of times (Mirza et al., 2006;

Stephens, 1981). For example, larger animals may not adhere as strictly to state-dependent foraging as smaller animals due to generally lower metabolic rates, which reduce energy

90 requirements and increase time to starvation (Brown et al., 2004). Thus, state-dependent behavior may be less prevalent in large animals under baseline conditions.

Internal state, or condition, is challenging to quantify, especially in wild animals. The field of physiological ecology has benefited tremendously from the efforts of others who have manipulated internal condition in wild animals (Angelier et al., 2007; Astheimer et al., 1992;

Crossin et al., 2012; Gray et al., 1990; Krause et al., 2017; Lynn et al., 2010). However, these approaches may force animals into extended periods of relatively extreme conditions to which animals may exhibit exaggerated behavioral and physiological responses (Fokidis et al., 2012).

The relationships between condition and behavior have also been examined at the population level in wild animals (Christianson and Creel, 2010; Rehnus et al., 2014; Tarjuelo et al., 2015).

This approach is valuable but may miss individual-level variation, and individuals at far ends of the condition spectrum, where behavioral trade-offs are most likely to occur (but see Chmura et al., 2016; Ellis et al., 2011; Mateo, 2007).

One particularly important potential indicator of an animal’s state is the internal concentrations of hormones, specifically, glucocorticoids (GCs). In the laboratory, experimentally elevated GCs can increase foraging (Astheimer et al., 1992; Hamelink et al., 1994; King et al.,

1992). In wild animals, foraging increases with GCs at baseline levels (Chmura et al., 2016;

Dallman et al., 1993; Landys-Ciannelli et al., 2006), and CORT decreases as individuals find food (Angelier et al., 2007). Conversely, elevated GCs have also been positively linked to anti- predator behaviors such as alarm-calling (Blumstein et al., 2006), anxiety and fearful behavior

(Korte, 2001; Thaker et al., 2009) and increased vigilance (Thaker et al., 2009; Voellmy et al.,

2014). Thus, evidence from experimental manipulations suggests that GCs may increase both foraging and anti-predatory behavior.

In observational studies of wild animals, the relative impacts of food and predation cannot be distinguished based on GC values alone. For example, in arctic ground squirrels, alpine

91 meadows were determined to be higher quality habitats than boreal forests because ground squirrels had lower GC levels within meadows compared to forests (Hik et al. 2001). However, it is impossible to determine whether these lower GC levels are due to higher food abundance, lower predation risk, or lower disturbance (among other possibilities). Recent work has highlighted the need for physiological measures capable of distinguishing between such environmental characteristics, and thyroid hormones have been offered as one possibility (Wasser et al., 2010). Specifically, triiodothyronine (T3) has recently been used effectively to represent energetic condition (Ayres et al., 2012; Vynne et al., 2014; Wasser et al., 2010). In addition to elevating GC levels, nutritional deficits generally decrease T3 levels in humans and some animals

(Douyon and Schteingart, 2002; Samuels and McDaniel, 1997; Schew et al., 1996).

Triiodothyronine is important in regulating thermoregulation and metabolism and is secreted via the hypothalamic-pituitary-thyroidal (HPT) axis (Douyon and Schteingart, 2002). This hormone decreases during food restriction, lowering basal metabolic rate and resting energy expenditure

(Harvey and Klandorf, 1982; Kitaysky et al., 2005; Klandorf et al., 1981; Rosen and Kumagai,

2008). By using GCs in tandem with T3, we may be able to estimate the relative importance of multiple stressors simultaneously. For example, limited food resources should decrease T3 and increase GCs, whereas fear should increase GC levels but not alter T3 levels (Ayres et al., 2012;

Wasser et al., 2011, 2010).

Medical literature, however, suggests a close neurological link between GC and T3 levels, such that elevated GCs may decrease T3 levels without a concomitant change in caloric intake or food deprivation (Degroot & Hoye 1976, Engler 1997, Samuels & McDaniel 1997,

Shimokaze et al. 2012). For instance, in captive animal studies, experimentally elevated GC levels by injection or induced stress (such as injury) also decreased TH levels without changes in food intake (Servatius et al. 1994, Walpita et al. 2007, Kilburn-Watt et al. 2010). In wild animals, too, T3 and GCs can be inversely related, but the interpretation of this result needs attention. For

92 instance, Wasser et al., (2011) interpreted decreased T3 and increased GCs as representing that both nutritional and psychological stress was increased. This result may have been influenced by the interactions between the HPG and HPA axis, however, such that no psychological stress was increased, but, similarly to other stressors, elevated GC levels reflected nutritional stress.

Including behavioral observations in concert with the measurement of GCs and T3 may help illuminate their functional interactions. For example, if animal A with high GC and low T3 levels foraged more than animal B with high T3 and low GC levels, this may indicate that nutritional stress is being perceived as a stressor, resulting in high GC levels, and that psychological stress

(or fear) is not driving high GC levels. In this case, perhaps elevated GCs promote foraging.

Alternatively, if animal A is more vigilant than animal B, this may indicate that psychological stress is causing both elevated GCs, and via the interaction between the HPA and HPT axis, T3 is decreased. In this case, perhaps decreased metabolism facilitates energy savings while the animal engages in anti-predatory, escape, or hiding behaviors (Krams et al., 2013).

In this study, I examined how internal condition (concentrations of GCs and T3) may influence behaviors (foraging and vigilance) in a wild mammal. I also examined how these two commonly used biomarkers (GCs and T3) were related to one other. I sampled two metapopulations which are naturally confined to two different habitats that vary in forage availability and quality, and predation risk. I predicted that 1) an animal with elevated T3 would forage less relative to an individual with low T3, 2) an animal’s GCs would be positively related to the time it spent being vigilant, and 3) that GCs and T3 would be negatively related within an individual, in concordance with the biomedical literature. My study thus links unmanipulated physiology and behavior at an individual level in a wild animal, with important implications for conservation physiology.

93 Methods

Study animal and study site

Vicuñas (Vicugna vicugna) are medium-sized ungulates (45-55 kg in this region) who inhabit high altitudes in South America. The smallest members of the Camelidae family, they are

South American relatives of llamas, alpacas, and guanacos. Harem family groups consist of a single dominant and territorial male, females, and yearling offspring. Family groups are commonly isolated (>300 m between groups), although they occasionally mix into groups of over

50 (personal obs.). Sexual dimorphism is not present in this species, with the exception of genitalia which cannot be observed from afar, and behavior, where males tend to be more alert or are defending territory. Therefore, this study was only conducted on known females where individuals were identified via numbered and colored ear tags, colored GPS and VHF collars, and

VHF frequency.

This study was conducted in San Guillermo National Park (SGNP), Argentina. At

~3,400 m above sea level, this region is considered a desert, receiving less than 240 mm of precipitation annually (Donadio and Buskirk, 2016). The park is pristine, with few anthropogenic disturbances and stressors, such as roads, novel noise, tourism, and habitat degradation or modification to account for. Here, puma (Puma concolor) predation accounts for approximately

90% of vicuña mortality and pumas are the only predator of adults (Donadio et al., 2012).

Vicuñas are also the main prey item of pumas in this system (Donadio and Buskirk, 2016).

For this study, I conducted behavioral observations on vicuñas in two habitat types: canyons, and meadows. Canyons (Figure 4-1) are valleys (10-300 m wide) in between hills, and are edged by steep, loose rubble and rocky outcroppings. They have low forage quality largely consisting of sporadic shrubby cover, and low predation risk to vicuñas compared to meadows.

Vicuñas in canyons experience approximately 90% more predation than expected than given the canyon’s spatial extent (approximately 15% of the park; Donadio and Buskirk, 2016). Meadows

94 are large, flat areas where vegetation mainly consists of tall, dense grasses, and thus puma predation is more common here than in canyons (Donadio and Buskirk, 2016). Vicuñas in meadows experience approximately 480% more predation than expected given their spatial extent

(approximately 4% of the park; Donadio and Buskirk, 2016). This high predation rate appears to be a function of both vicuña density (~6 times higher in meadows than canyons) and perhaps the ambush-style hunting mode of pumas, which is successful in this habitat due to dense vegetation cover. Thus, these two habitats may differ in risk and meadows offer more rewards than canyons.

The remaining 81% of habitat in SGNP is plains, visually similar to flat, gravel parking lots. In these areas, puma predation occurs 30% less than would be expected given their spatial extent.

For more details on these study sites see Donadio and Buskirk (2016). Crucially, GPS collar data indicate that site fidelity is very high in both canyons and meadows, and individual vicuñas do not move between these two particular canyon and meadow sites (Pritchard, unpublished data).

Thus, individuals occupying these sites experience relatively consistent conditions across years

(2014-2016).

Behavioral Analyses

I video recorded behavior for 20 minutes on known individuals (individual equipped with

GPS collars and numbered ear tags) in fall through early winter (April-June) of 2014, 2015, and

2016. The same individuals were observed through multiple years, until they died of natural causes. Behavioral videos were analyzed continuously using JWatcher software (v1.0) and vigilance was defined as an animal having its head up and looking around, and was not mutually exclusive of chewing, but the individual was not actively engaged in foraging. Foraging was defined as an animal having its head down, either actively foraging or searching for forage. I measured the total proportion of time spent being vigilant and foraging, engaging in other behaviors, or out of sight. I measured group size by determining the number of animals that

95 stayed in close proximity to each other (~50 m) and moved together. I also recorded wind speed and time of day at the point of each observation, Julian date, and year. None of these covariates explained significant variation in behavior and thus all were omitted from final models (all p <

0.10; see Statistical Analyses, below).

Animals spent between 0 and 93% of the time foraging (mean 59%, range 0–94%) or being vigilant (mean 30%, range 0–98%). The remaining 7% of the time was spent walking

(mean 2%, range 0–14%), laying down (mean 4%, range 0–100%), running (mean 0.2%, range

0–2%), nursing (mean 0.3%, range 0–10%), and engaging in other behaviors such as scratching

(mean 0.3%, range 0–3%). Since these behaviors were rare and my research question was focused on understanding the role of external and internal conditions in the trade-offs between foraging and vigilance, I did not examine these other behaviors.

Fecal sample analyses

Once each behavioral observation was completed, I waited for a fecal sample to be dropped from that individual. The sample was collected immediately, placed on ice in the field, and stored at -20C upon returning to the field station. For comparisons to behavior, I created means of FGCMs and T3 levels collected from multiple samples. Others have collected a single fecal sample, and used it for comparison to behavioral observations conducted over two subsequent weeks (Chmura et al., 2016; Mateo, 2007). Here, I analyzed multiple (up to 9) fecal samples from the same individual collected over time (during initial captures and subsequent behavioral observations over three years) to determine if hormone concentrations were consistent through time. I used linear mixed effects models, where hormone concentration was the dependent variable, and the time difference between samples (in days) was used as the independent variable, with vicuña identity (ID) as a random effect. I found that the time between samplings had no effect on T3 concentrations (t =0.515, df = 56, p = 0.608) or FGCM levels (t = -

96 1.651, df = 65, p = 0.103), indicating that hormone concentrations remained relatively consistent within an individual over my sampling period. Therefore, I “completed” (using the ‘complete’ function in R package tidyr) the dataset for behavioral observations that did not have matching fecal samples for that day, using the hormone concentrations from the most recent previous sample (minimum 0.92 days apart, mean 10.4 days apart, 3rd quantile 15 days, maximum of 38 days apart).

Fecal GCM levels were evaluated with an EIA validated for this species (Arias et al.

2013), using 11-oxoaetiocholanolone (Goymann et al., 1999). Briefly, pellets were lyophilized for

48 hours and ground to a fine powder with mortar and pestle. 0.500 ± 0.05 g were extracted for 5 minutes on a vortexer with 5 mL of 80% methanol. Pellets were then centrifuged for 2,500 gs for

30 minutes. The supernatant was then diluted (1:10) with assay buffer. Fecal T3 was measured with L-triiodothyronine 125I- RIAs (06B254216; MP Biomedicals, Orangeburg, NY), following

Wasser et al. (2010) and manufacturer instructions.

Statistical analyses

I used the penalized quasilikelihood (PQL; function ‘glmmPQL’ in package MASS) method of analysis due to my non-normal data and unbalanced design (more observations in the

Meadow). All analyses were performed in R Studio (v. 3.4.1; R Core Team 2017). Model parameters were selected using backwards step selection and were retained where p < 0.10. The proportion of time spent vigilant and feeding were response variables, while FGCMs and T3 were independent variables. Fixed factors included habitat and year. Covariates included wind speed,

Julian date, time of day, and group size. Individual ID was included as a random effect in all models. I retained variables of interest (GCs and T3) in all models. I checked the residuals of the models visually, to ensure these met the assumptions of normal distribution. I examined the relationships between GCs and T3 using non-parametric Spearman correlation coefficients. Data

97 are represented as mean  1 standard error (SE). I examined the relationships between GCs and habitat, and T3 and habitat using glmmPQL, with the physiological measurement as the dependent variable, habitat as fixed factor, and animal ID as a random effect.

Animal ethics statement

All protocols were approved by The Pennsylvania State University Institutional Animal

Care and Use Committee under protocol #45139. Samples were imported under U.S. Fish and

Wildlife Service permitting for threatened animals, with Federal Fish and Wildlife Permit

#MA70993B-2.

Results

There was no relationship between the proportion of time spent vicuñas foraging and their fecal glucocorticoid metabolite (FGCM) levels (estimate < -0.0001; SE 0.001, t = -0.067, p

= 0.947) or T3 levels (estimate < -0.0001; SE < 0.001, t = - 0.817, p = 0.415; Figure 4-2).

Vicuñas spent approximately 11% more time foraging in meadows than canyons (61  2% vs 51

 3%, respectively; estimate = 0.112, SE = 0.049, t = 2.289, p = 0.024).

There was also no relationship between the proportion of time vicuñas spent vigilant and their FGCM (estimate = -0.0007, SE = 0.001, t = -0.589, p = 0.557) or T3 levels (estimate =

0.086, SE = 0.066, t = -1.919, p = 0.199; Figure 4-3). There was a trend for vicuñas to be more vigilant in canyons than in meadows by approximately 11% (39  3% and 28  1%, respectively; estimate = -0.088, SE = 0.046, t = -1.9199, p = 0.057; Figure 4-4).

98 There was no correlation between GCs and T3 (rs = 0.026, p = 0.734). There were no differences in GC (estimate = -3.99, SE = 3.779, t= -1.056, p = 0.293) or T3 levels (estimate =

89.923, SE = 77.04, t = 1.167, p = 0.235; Figure 4-5) between habitats.

Discussion

In this study, I found that neither glucocorticoids (GCs) nor triiodothyronine (T3) were significantly correlated with vicuña behavior. Vicuñas tended to be more vigilant and spend less time foraging in canyons than meadows but there were no significant differences in either GC or

T3 levels between the two habitats. Glucocorticoids (GCs) and triiodothyronine (T3) were not correlated within an individual. There was also no effect of group size on behavior. The effect of group size is relatively weak (Cohen f2 = 0.02), with R2 = 0.005, and a power analysis suggests that I would need to increase my sample size from 166 to 545 to detect a statistically significant effect.

Together, this evidence suggests that neither GC or T3 biomarkers may be reliable in predicting behavior in wild, undisturbed populations of vicuñas, that may be exhibiting baseline physiological conditions. This lack of relationship between hormones and behavior could be driven by vicuñas’ energetic condition at the time of sampling. For instance, state-dependent behavioral switches from antipredator behaviors to foraging are often observed only under energetically constrained conditions (Astheimer et al., 1992; Gray et al., 1990; Landys-Ciannelli et al., 2006; Santana et al., 1995; Tempel et al., 1992). Furthermore, under baseline conditions,

GCs may not be sufficiently responsive to detect subtle changes in body condition (Sorenson et al., 2017).

Other studies of state-dependent decision-making in large, wild animals generally occur at long temporal scales, or address state-dependent decisions across a broad spatial landscape

99 (Long et al., 2014; Montgomery et al., 2013). For instance, Monteith et al. (2011) describe the effects of nutritional condition of mule deer (Odocolileus hemionus) on biannual migration, reporting that deer in good nutritional condition delayed their autumn migration compared to animals in poor nutritional condition. Studies taken at these large scales provide a foundation on which to base the context of state-dependent decision making, but crucial too, is to understand if and how these decisions are made at short-timescales, which may influence a more persistent nutritional condition. Although I found no relationships between short-term activity budgets and hormone levels in this study, vicuñas may be making state-dependent decisions at larger spatial and temporal scales, including habitat selection. The recent deployment of GPS collars in this system will help me address this possibility.

Contrary to my expectations that glucocorticoids and thyroid hormones concentrations would be negatively correlated (Douyon and Schteingart, 2002; Hunt et al., 2012; Welcker et al.,

2015), I found no relationship between these two measurements. There is growing evidence that these relationships are not well-understood in many wild animals (Jeanniard du Dot et al., 2009;

Jesmer et al., 2017). For instance, strong positive correlations have been reported (Jesmer et al.,

2017), and in Hawaiian monk seals (Monachus schauinslandi), GCs and T3 were positively correlated at four study sites, but two additional study sites were characterized by having the highest and lowest GCs, and but both sites had relatively low T3 (Gobush et al. 2014). Others have found even more complex relationships between these biomarkers (Ayres et al., 2012;

Hayward et al., 2011). For example, Keogh et al. (2013) found variable relationships between

GCs and T3 among years, but found that both biomarkers decreased as the elapsed time from a stressor increased. Our study adds to a growing body of evidence that the interactions between these two hormones are not well-understood in wild animals.

100 Conclusion

Both behavior (Carney and Sydeman, 1999; Donadio and Buskirk, 2006; Satterthwaite and Mangel, 2012) and physiology (Ahlering et al., 2013; Madliger and Love, 2016) have been used in conservation to understand population or environmental health. Wildlife and conservation managers are increasingly searching for techniques to monitor population health within a single generation, as opposed to previous approaches that track changes in demography (Sorenson et al.,

2017). Biomonitoring tools including glucocorticoids and thyroid hormones have been suggested as a potentially strong approach to identify changes in conservation-related variables, such as anthropogenic disturbance (Busch and Hayward, 2009; Cooke and O’Connor, 2010; Madliger et al., 2016; Wikelski and Cooke, 2006). The relationship between endocrine biomarkers and behavior has been suggested. This work adds to a growing body of evidence suggesting that these markers are not simple to interpret and so are unlikely to provide an easy monitoring tool for conservation biologists. Field studies across taxa are needed to understand the information provided by GCs, thyroid hormones, and their interactions in wild animals, before they can be accurately used as biomarkers of psychological and nutritional stress (Boonstra, 2013; Otovic and

Hutchinson, 2015).

101 Literature Cited

Ahlering MA, Maldonado JE, Eggert LS, Fleischer RC, Western D, Brown JL (2013)

Conservation outside protected areas and the effect of human-dominated landscapes on

stress hormones in Savannah elephants. Conserv Biol 27: 569–575.

Angelier F, Shaffer SA, Weimerskirch H, Trouvé C, Chastel O (2007) Corticosterone and

foraging behavior in a pelagic seabird. Physiol Biochem Zool 80: 283–292.

Astheimer LB, Buttemer WA, Wingfield JC (1992) Interactions of corticosterone with feeding,

activity and metabolism in passerine birds. Ornis Scand 23: 355–365.

Ayres KL, Booth RK, Hempelmann JA, Koski KL, Emmons CK, Baird RW, Balcomb-Bartok K,

Hanson MB, Ford MJ, Wasser SK (2012) Distinguishing the impacts of inadequate prey

and vessel traffic on an endangered killer whale (Orcinus orca) population. PLoS One 7:

e36842.

Beale CM, Monaghan P (2004) Behavioural responses to human disturbance: A matter of choice?

Anim Behav 68: 1065–1069.

Blumstein DT, Patton ML, Saltzman W (2006) Faecal glucocorticoid metabolites and alarm

calling in free-living yellow-bellied marmots. Biol Lett 2: 29–32.

Boonstra R (2013) Reality as the leading cause of stress: rethinking the impact of chronic stress in

nature. Funct Ecol 27: 11–23.

Bourdeau PE, Bach MT, Peacor SD (2016) Predator presence dramatically reduces copepod

abundance through condition-mediated non-consumptive effects. Freshw Biol 61: 1020–

1031.

Brown JH, Gillooly JF, Allen AP, Savage VM, West GB (2004) Toward a metabolic theory of

ecology. Ecology 85: 1771–1789.

Brown JS (1999) Vigilance, patch use and habitat selection: foraging under predation risk. Evol

Ecol Res 1: 49–71.

102 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.

Buskirk J Van (2000) The costs of an inducible defense in anuran larvae. Ecology 81: 2813–

2821.

Caraco T, Blanckenhorn WU, Gregory GM, Newman JA, Recer GM, Zwicker SM (1990) Risk-

sensitivity: ambient temperature affects foraging choice. Anim Behav 39: 338–345.

Carney KM, Sydeman WJ (1999) A review of human disturbance effects on nesting colonial

waterbirds. Waterbirds: Int J Waterbird Ecol 22: 68–79.

Carter AW, Paitz RT, McGhee KE, Bowden RM (2016) Turtle hatchlings show behavioral types

that are robust to developmental manipulations. Physiol Behav 155: 46–55.

Chmura HE, Wey TW, Blumstein DT (2016) Assessing the sensitivity of foraging and vigilance

to internal state and environmental variables in yellow-bellied marmots (Marmota

flaviventris). Behav Ecol Sociobiol 70: 1901–1910.

Christianson D, Creel SR (2010) A nutritionally mediated risk effect of wolves on elk. Ecology

91: 1184–1191.

Cooke SJ, O’Connor CM (2010) Making conservation physiology relevant to policy makers and

conservation practitioners. Conserv Lett 3: 159–166.

Crossin GT, Trathan PN, Phillips RA, Gorman KB, Dawson A, Sakamoto KQ, Williams TD

(2012) Corticosterone predicts foraging behavior and parental care in Macaroni penguins.

Am Nat 180: E31–E41.

Croy MI, Hughes RN (1991) Effects of food supply, hunger, danger and competition on choice of

foraging location by the fifteen-spined stickleback, Spinachia spinachia L. Anim Behav

42: 131–139.

103 Dallman MF, Strack AM, Akana SF, Bradbury MJ, Hanson ES, Scribner KA, Smith M (1993)

Feast and famine: critical role of glucocorticoids with insulin in daily energy flow. Front

Neuroendocrinol 14: 303–347.

De Groef B, Goris N, Arckens L, Kühn ER, Darras VM (2003) Corticotropin-releasing hormone

(CRH)-induced thyrotropin release is directly mediated through CRH receptor type 2 on

thyrotropes. Endocrinology 144: 5537–5544.

Donadio E, Buskirk SW (2006) Flight behavior in guanacos and vicuñas in areas with and

without poaching in western Argentina. Biol Conserv 127: 139–145.

Donadio E, Buskirk SW (2016) Linking predation risk, ungulate antipredator responses, and

patterns of vegetation in the high Andes. J Mammal 97: 966–977.

Donadio E, Buskirk SW, Novaro AJ (2012) Juvenile and adult mortality patterns in a vicuña

(Vicugna vicugna) population. J Mammal 93: 1–9.

Douyon L, Schteingart DE (2002) Effect of obesity and starvation on thyroid hormone, growth

hormone, and cortisol secretion. Endocrinol Metab Clin North Am 31: 173–189.

Ellis JJ, MacLarnon AM, Heistermann M, Semple S (2011) The social correlates of self-directed

behaviour and faecal glucocorticoid levels among adult male Olive baboons (Papio

hamadryas anubis) in Gashaka-Gumti National Park, Nigeria. African Zool 46: 302–308.

Fokidis HB, des Roziers MB, Sparr R, Rogowski C, Sweazea K, Deviche P (2012) Unpredictable

food availability induces metabolic and hormonal changes independent of food intake in

a sedentary songbird. J Exp Biol 215: 2920–2930.

Gauthier-Clerc M, Le Maho Y, Gendner JP, Durant J, Handrich Y (2001) State-dependent

decisions in long-term fasting king penguins, Aptenodytes patagonicus, during courtship

and incubation. Anim Behav 62: 661–669.

104 Geris KL, Berghman LR, Kühn ER, Darras VM (1999) The dropin plasma thyrotropin

concentrations in fasted chickens is caused by an action at the level of the hypothalamus:

role of corticoserone. Domest Anim Endocrinol 16: 231–237.

Geris KL, Kotanen SP, Berghman LR, Kühn ER, Darras VM (1996) Evidence of a thyrotropin-

releasing activity of ovine corticotropin-releasing factor in the domestic fowl (Gallus

domesticus). Gen Comp Endocrinol 104: 139–146.

Gilby IC, Wrangham RW (2007) Risk-prone hunting by chimpanzees (Pan troglodytes

schweinfurthii) increases during periods of high diet quality. Behav Ecol Sociobiol 61:

1771–1779.

Goymann W, Möstl E, Van’t Hof T, East ML, Hofer H (1999) Noninvasive fecal monitoring of

glucocorticoids in spotted hyenas, Crocuta crocuta. Gen Comp Endocrinol 114: 340–

348.

Gray M, Yarian D, Ramenofsky M, Gray JM, Yarian D, Ramenofsky M (1990) Corticosterone,

foraging behavior, and metabolism in dark-eyed Juncos, Junco hyemalis. Gen Comp

Endocrinol 79: 375–384.

Hamelink CR, Currie PJ, Chambers JW, Castonguay TW, Coscina D V (1994) Corticosterone-

responsive and -unresponsive metabolic characteristics of adrenalectomized rats. Am J

Physiol 267: R799–R804.

Harvey S, Klandorf H (1982) Reduced adrenocortical function and increased thyroid function in

fasted and refed chickens. J Endocrinol 98: 129–135.

Hayward LS, Bowles AE, Ha JC, Wasser SK (2011) Impacts of acute and long-term vehicle

exposure on physiology and reproductive success of the Northern spotted owl. Ecosphere

2: Art65.

105 Houston AI, McNamara JM, Hutchinson JMC (1993) General results concerning the trade-off

between gaining energy and avoiding predation. Philos Trans R Soc B Biol Sci 341: 375–

397.

Hunt KE, Innis C, Rolland RM (2012) Corticosterone and thyroxine in cold-stunned Kemp’s

ridley sea turtles (Lepidochelys kempii). J Zoo Wildl Med 43: 479–493.

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.

Jesmer BR, Goheen JR, Monteith KL, Kauffman MJ (2017) State-dependent behavior alters

endocrine-energy relationship: implications for conservation and management. Ecol Appl

27: 2303–2312.

Keogh MJ, Atkinson S, Maniscalco JM (2013) Body condition and endocrine profiles of Steller

sea lion (Eumetopias jubatus) pups during the early postnatal period. Gen Comp

Endocrinol 184: 42–50.

King BM, Zansler CA, Richard SM, Gutierrez C, Dallman MF (1992) Paraventricular

hypothalamic obesity in rats: role of corticosterone. Physiol Behav. 51:1207-1212.

Kitaysky AS, Romano MD, Piatt JF, Wingfield JC, Kikuchi M (2005) The adrenocortical

response of tufted puffin chicks to nutritional deficits. Horm Behav 47: 606–619.

Klandorf H, Sharp PJ, Macleod MG (1981) The relationship between heat production and

concentrations of plasma thyroid hormones in the domestic hen. Gen Comp Endocrinol

45: 513–520.

Korte SM (2001) Corticosteroids in relation to fear, anxiety and psychopathology. Neurosci

Biobehav Rev 25: 117–142.

106 Krams I, Kivleniece I, Kuusik A, Krama T, Freeberg TM, Mänd R, Vrublevska J, Rantala MJ,

Mänd M (2013) Predation selects for low resting metabolic rate and consistent individual

differences in anti-predator behavior in a beetle. Acta Ethol 16: 163–172.

Krause JS, Pérez JH, Meddle SL, Wingfield JC (2017) Effects of short-term fasting on stress

physiology, body condition, and locomotor activity in wintering male white-crowned

sparrows. Physiol Behav 177: 282–290.

Landys-Ciannelli 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.

Lima SL (1998) Stress and decision making under the risk of predation: recent developments

from behavioral, reproductive and ecological perspectives. Adv Study Behav 27: 215–

290.

Lima SL, Bednekoff PA (1999) Temporal variation in danger drives antipredator behavior: the

predation risk allocation hypothesis. Am Nat 153: 649–659.

Lima SL, Dill LM (1990) Behavioral decisions made under the risk of predation: a review and

prospectus. Can J Zool 68: 619–640.

Long RA, Terry Bowyer R, Porter WP, Mathewson P, Monteith KL, Kie JG (2014) Behavior and

nutritional condition buffer a large-bodied endotherm against direct and indirect effects

of climate. Ecol Monogr 84: 513–532.

Lynn SE, Stamplis TB, Barrington WT, Weida N, Hudak CA (2010) Food, stress, and

reproduction: short-term fasting alters endocrine physiology and reproductive behavior in

the zebra finch. Horm Behav 58: 214–222.

Madliger CL, Cooke SJ, Crespi EJ, Funk JL, Hultine KR, Hunt KE, Rohr JR, Sinclair BJ, Suski

CD, Willis CKR, Love OP. (2016) Success stories and emerging themes in conservation

physiology. Conserv Physiol 4: 1–17.

107 Madliger CL, Love OP (2016) Conservation implications of a lack of relationship between

baseline glucocorticoids and fitness in a wild passerine. Ecol Appl 26: 2730–2743.

Mangel M, Clark CW (1986) Towards a unified foraging theory. Ecology 67: 1127–1138.

Mateo JM (2007) Ecological and hormonal correlates of antipredator behavior in adult Belding’s

ground squirrels (Spermophilus beldingi). Behav Ecol Sociobiol 62: 37–49.

McNamara JM, Houston AI (1996) State-dependent life histories. Nature 380: 215–221.

Mirza RS, Mathis A, Chivers DP (2006) Does temporal variation in predation risk influence the

intensity of antipredator responses? A test of the risk allocation hypothesis. Ethology

112: 44–51.

Monteith KL, Bleich VC, Stephenson TR, Pierce BM, Conner MM, Klaver RW, Bowyer T

(2011) Timing of seasonal migration in mule deer: effects of climate, plant phenology,

and life-history characteristics. Ecosphere 2: art47

Montgomery RA, Vucetich JA, Peterson RO, Roloff GJ, Millenbah KF (2013) The influence of

winter severity, predation and senescence on moose habitat use. J Anim Ecol 82: 301–

309.

Moreno J, Carlson A, Alatalo R V (1988) Winter energetics of coniferous forest tits Paridae in the

North: the implications of body size. Funct Ecol 2: 163–170.

Nonacs P (2001) State dependent behavior and the marginal value theorem. Behav Ecol 12: 71–

83.

Norman AW, Litwack G (1997) Thyroid hormones. In: Hormones. Academic Press, California,

pp. 221-262.

Okada R, Miller MF, Yamamoto K, Groef B De, Denver RJ, Kikuyama S (2007) Involvement of

the corticotropin-releasing factor (CRF) type 2 receptor in CRF-induced thyrotropin

release by the amphibian pituitary gland. Gen Comp Endocrinol 150: 437–444.

108 Olsson O (1997) Clutch abandonment: a state-dependent decision in king penguins. J Avian Biol

28: 264–267.

Otovic P, Hutchinson E (2015) Limits to using HPA axis activity as an indication of animal

welfare. Altex 32: 41–50.

Parker KL, Barboza PS, Gillingham MP (2009) Nutrition integrates environmental responses of

ungulates. Funct Ecol. 23: 57–69.

Pedersen AB, Greives TJ (2008) The interaction of parasites and resources cause crashes in a

wild mouse population. J Anim Ecol 77: 370–377.

Persons MH, Walker S, Rypstra A (2002) Fitness costs and benefits of antipredator behavior

mediated by chemotactile cues in the wolf spider Pardosa milvina (Araneae: Lycosidae).

Behav Ecol 13: 386–392.

Rehnus M, Wehrle M, Palme R (2014) Mountain hares Lepus timidus and tourism: Stress events

and reactions. J Appl Ecol 51: 6–12.

Rosen DAS, Kumagai S (2008) Hormone changes indicate that winter is a critical period for food

shortages in Steller sea lions. J Comp Physiol B Biochem Syst Environ Physiol 178: 573–

583.

Samuels M, McDaniel P (1997) Thyrotropin levels during hydrocorticosterone infusions that

mimic fasting-induced cortisol elevations: a clinical research study. J Clin Endocrinol

Metab 82: 3700–3704.

Santana P, Akana SF, Hanson ES, Strack A, Sebastian R, Dallman MF (1995) Aldosterone and

dexamethasone both stimulate energy acquisition whereas only the glucocorticoid alters

energy storage. Endocrinology 136: 2214–2222.

Satterthwaite WH, Mangel M (2012) Behavioral models as a common framework to predict

impacts of environmental change on seabirds and fur seals. Deep Res Part II Top Stud

Oceanogr 65–70: 304–315.

109 Schew WA, McNabb FM, Scanes CG (1996) Comparison of the ontogenesis of thyroid

hormones, growth hormone, and insulin-like growth factor-I in ad libitum and food-

restricted (altricial) European starlings and (precocial) Japanese quail. Gen Comp

Endocrinol 101: 304–316.

Searle KR, Stokes CJ, Gordon IJ (2008) When foraging and fear meet: using foraging hierarchies

to inform assessments of landscapes of fear. Behav Ecol 19: 475–482.

Sinclair ARE, Arcese P (1995) Population consequences of predation sensitive foraging: the

Serengeti wildebeest. Ecology 76: 882–891.

Sorenson GH, Dey CJ, Madliger CL, Love OP (2017) Effectiveness of baseline corticosterone as

a monitoring tool for fitness: a meta-analysis in seabirds. Oecologia 183: 353–365.

Stephens DW (1981) The logic of risk-sensitive foraging preferences. Anim Behav 29: 628–629.

Stephens DW, Brown JS, Ydenberg RC (2007) Foraging: behavior and ecology. University of

Chicago Press, Chicago

Tarjuelo R, Barja I, Morales MB, Traba J, Benítez-López A, Casas F, Arroyo B, Delgado MP,

Mougeot F (2015) Effects of human activity on physiological and behavioral responses of

an endangered steppe bird. Behav Ecol 26: 828–838.

Taylor EN, Malawy MA, Browning D, Lemar S V, DeNardo DF (2005) Effects of food

supplementation on the physiological ecology of female Western diamond-backed

rattlesnakes (Crotalus atrox). Oecologia 144: 206–213.

Tempel DL, McEwen BS, Leibowitz SF (1992) Effects of adrenal steroid agonists on food intake

and macronutrient selection. Physiol Behav 52: 1161–1166.

Thaker M, Lima SL, Hews DK (2009) Acute corticosterone elevation enhances antipredator

behaviors in male tree lizard morphs. Horm Behav 56: 51–57.

Tomasi TE (1991) Utilization rates of thyroid hormones in mammals. Comp Biochem Physiol -

Part A Physiol 100: 503–516.

110 Voellmy IK, Goncalves IB, Barrette MF, Monfort SL, Manser MB (2014) Mean fecal

glucocorticoid metabolites are associated with vigilance, whereas immediate cortisol

levels better reflect acute anti-predator responses in meerkats. Horm Behav 66: 759–765.

Vynne C, Booth RK, Wasser SK (2014) Physiological implications of landscape use by free-

ranging maned wolves (Chrysocyon brachyurus) in Brazil. J Mammal 95: 696–706.

Walpita CN, Grommen SVH, Darras VM, Van der Geyten S (2007) The influence of stress on

thyroid hormone production and peripheral deiodination in the Nile tilapia (Oreochromis

niloticus). Gen Comp Endocrinol 150: 18–25.

Wasser SK, Azkarate JC, Booth RK, Hayward LS, Hunt K, Ayres KL, Vynne C, Gobush K,

Canales-Espinosa D, Rodríguez-Luna E (2010) Non-invasive measurement of thyroid

hormone in feces of a diverse array of avian and mammalian species. Gen Comp

Endocrinol 168: 1–7.

Wasser SK, Keim JL, Taper ML, Lele SR (2011) The influences of wolf predation, habitat loss,

and human activity on caribou and moose in the Alberta oil sands. Front Ecol Environ 1:

376–382.

Welcker J, Speakman JR, Elliott KH, Hatch SA, Kitaysky AS (2015) Resting and daily energy

expenditures during reproduction are adjusted in opposite directions in free-living birds.

Funct Ecol 29: 250–258.

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

Wirsing AJ, Heithaus MR, Dill LM (2007) Fear factor: Do dugongs (Dugong dugon) trade food

for safety from tiger sharks (Galeocerdo cuvier)? Oecologia 153: 1031–1040.

Ydenberg RC, Butler RW, Lank DB (2007) Effects of predator landscapes on the evolutionary

ecology of routing, timing and molt by long-distance migrants. J Avian Biol 38: 523–529.

Ivlev, VS (1961) Experimental ecology of the feeding of fishes. Yale University Press, New

Haven.

111 Robbins J (1981) Factors altering thyroid hormone metabolism. Environmental Health

Perspectives 38:65- 70.

112

Figure 4-1. Pumas in tall grass of the meadows (left) and open habitat of the canyons (right). Both the high vegetation density in meadows, and the rocky outcroppings and steep slopes in canyon conceal pumas well and facilitate successful hunting.

T 3 and Foraging, dataset 1

Plot.T3.Foraging1<-ggplot(behavior1, aes(x=T3, y=Foraging)) + geom_point(aes(colour = Vicuna.ID), size = 5, alpha=0.7) + geom_smooth(method="lm", color="darkgrey") + theme(panel.grid = element_blank(), panel.background = element_rect(fill = "white", colour = "grey50"), plot.title = element_text(hjust = 0.5, size=20), legend.position = "none", axis.title = element_text(size = 16), axis.text = element_text(size=14 )) + 113 scale_colour_manual(values=pal2(50)) + labs(x="Triiodothyronine (ng/g)", y="") + scale_y_continuous(limits = c(0,1))

1.00 1.00

) %

( 0.75 0.75

g

n

i

g

a

r

o f

0.50 0.50

t

n

e

p

s

e m

i 0.25 0.25 T

0.00 0.00 0 20 40 60 5000 5500 Fecal glucocorticoid Triiodothyronine (ng/g) metabolites (ng/g)

Figure 4-2. Relationships between physiological measurements and the percentage of time spent foraging. There was no relationship between an individual’s fecal glucocorticoid metabolite

(GCs) or triiodothyronine (T3) concentrations and the percent time it spent foraging. Colors indicate individual animals.

4

T 3 and vigilance, dataset1

Plot.T3.Vigilance1<-ggplot(behavior1, aes(x=T3, y=Vigilance)) + geom_point(aes(colour = Vicuna.ID), size = 5, alpha=0.7) + geom_smooth(method="lm", color="darkgrey") + theme(panel.grid = element_blank(), panel.background = element_rect(fill = "white", colour = "grey50"), plot.title = element_text(hjust = 0.5, size=20), legend.position = "none", axis.title = element_text(size = 16), axis.text = element_text(size=14)) + scale_colour_manual(values=pal2(50)) + labs(x="Triiodothyronine (ng/g)", y="") + scale_y_continuous(limits = c(0,1)) 114

1.00 1.00 )

% 0.75 0.75

(

t

n

a

l

i

g

i

v

t 0.50 0.50

n

e

p

s

e m

i 0.25 0.25 T

0.00 0.00 0 20 40 60 5000 5500 Fecal glucocorticoid Triiodothyronine (ng/g) metabolites (ng/g)

Boxplot for foraging between habitat s. Figure# Basic 4-3.boxplot Relationships between physiological measurements and the percentage of time spent boxplot(Foraging ~ Habitat, data=behavior1, vigilant. There waspar( nomar relationship= c(2, 5, between2, 2)+ fecal0.1), glucocorticoid metabolites (GCs) or names=c("Canyon", "Meadow"), triiodothyroninenotch (T3)= TRUE and) the percent time an individual exhibited vigilance behavior. Colors text(1.5,0.9,"*", cex = 3) indicatemtext( individual"Proportion animals.of time spent foraging", side=2, line=2.2, cex=1.1, font=2)

# Add data points mylevels<-levels(behavior1$Habitat) levelProportions<-summary(behavior1 $Habitat)/nrow(behavior1)

6

115

Figure 4-4. Proportion of time spent foraging and vigilant between the canyon and meadow habitats. Vicuñas spent approximately 11% more time foraging in meadows than canyons (p =

0.027). There was a trend for vicuñas to be more vigilant in canyons than in meadows by approximately 11% (p = 0.067).

116

Figure 4-5. Physiological measurements in canyon and meadow habitats. There were no differences between fecal glucocorticoid metabolite or triiodothyronine concentrations in canyons and meadows once individual ID was accounted for. Colors indicate individual animals.

117 Chapter 5

Direct and Indirect Estimates of Risk and Associated Antipredator Responses

Abstract

Predation risk can be assessed based on both direct and indirect cues, and these cues may produce different antipredator behaviors in prey. Direct cues of risk may include the actual presence of a predator, to which prey may respond reactively, for instance by moving away from the predator.

Indirect cues of risk may include time, where some times of day are riskier than others, and space, where some areas are riskier than others. Prey may anticipate risk based on indirect cues, and respond proactively, for instance by leaving a risky area at risky times of day. I investigated temporal and spatial predator-prey dynamics in vicuñas (Vicugna vicugna) and pumas (Puma concolor) in the high Andes Mountains of Argentina, using GPS collared animals in two habitats.

I found that vicuñas largely did not spatially avoid pumas within 400 m and 90 minutes of a puma’s location, but these species encountered each other rarely within these shared spaces. I also found differences in indirect risk, measured as minimum convex polygon overlap, between the two habitats, indicating higher risk in canyons where spatial overlap was greater. Lastly, I found evidence that by engaging in diel migrations, vicuñas reduced their spatial overlap with pumas and avoided both canyons and meadows during night. My results have implications for the nature of interactions between species of concern.

Introduction

The strength of predator-prey interactions can vary with habitat, where some landscapes are considered riskier or safer than others (Creel et al., 2005; Latombe et al., 2014). These

“landscapes of fear” are represented by hills and valleys of background levels of predation risk,

118 which are characterized by habitat features that promote safety or risk for prey (Laundré et al.,

2010). Prey can respond reactively to these varying levels of predation risk. For instance, when wolves were present in the area within the previous 24 hours, elk selected habitats with a higher density of conifers (Creel et al., 2005). Or prey can react anticipatorily, by moving out of areas before risky times are likely to occur in that area. For instance, elk that migrate annually from their summer ranges where wolf density was highest were exposed to 70% less predation than non-migrant animals who stayed in their summer ranges (Hebblewhite and Merrill, 2007).

The spatial and temporal overlap of predators and prey can have strong implications for population dynamics of both predators and prey. Prey can respond to predator presence with antipredator behaviors, which can have strong impacts on foraging and carry associated costs of predator avoidance (Lima and Bednekoff, 1999). These dynamics can also have strong effects on predator performance (Boyd 1966), as predators must successfully track their prey as prey respond to perceived predation risk (Russell et al., 1992). Importantly, prey can respond to both direct and indirect indicators of predation risk, where a direct indicator of risk is the actual detection of a predator, and indirect indicators may include indicators of the predator’s presence such as olfactory cues (urine or scat; Eccard et al., 2017; Kuijper et al., 2014; Orrock, 2010), evidence of conspecific mortality (Moll et al., 2016), time when risk is high (Packer et al., 2011;

Palmer et al., 2017; Zaret and Suffern, 1976), and habitat characteristics that may facilitate successful hunting (Davies et al., 2016; Thaker et al., 2015; Winnie and Creel, 2007). Prey use this information to make decisions, including about whether or how to change behavior. By combining temporal (risk factors over time) and spatial (risk factors over space) components of predator risk, quantifying dynamic interactions provides a more integrative view of prey responses to predators. For instance, elk modified group sizes when wolves were within 5 km of the elk in the past 8 hours, but not in the past 24 or 48 hours (Proffitt et al., 2009). Looking only at distance or only at time would not give the full picture of these interactions, which may hold

119 important information for understanding animal perception or trade-offs in anti-predator responses. In the above example, elk may not be able to track changes to the perceived risk over periods longer than 8 hours.

Ambush predators, such as felines, may elicit particularly strong antipredator behaviors

(Schmitz, 2008; Thaker et al., 2010), as prey may behaviorally respond to cues of risk based on habitat characteristics (ambush sites). Here, I examined indirect and direct predation risk between vicuñas (Vicugna vicugna) and their ambush predators, pumas (Puma concolor) between two habitats, canyons and meadows. Specifically, I combined quantification of temporal and spatial overlap of these species to examine direct and indirect risk. Short-term temporal overlap and small-scale spatial overlap between pumas and vicuñas (i.e., these species occurring in a similar place at a similar time) was used as an indicator of direct risk. As a result, certain times of day and certain places where overlap is greatest could provide indirect cues of risk, where some times and habitats are riskier than others.

In the high Andes Mountains of Argentina, vicuñas have the potential to strongly influence vegetation dynamics (Donadio and Buskirk, 2016), which differ among habitats and may be a function of varying predation risk. Previous work in this region suggests that puma predation risk to vicuñas may differ between canyon and meadow habitats (Donadio and Buskirk,

2016). For instance, higher densities of puma scat were found in canyons than meadows, indicating that pumas may use canyons more intensively than meadows, and vicuñas in canyons may be at higher risk (Donadio and Buskirk, 2016). However, carcass abundance was higher in meadows than canyons, given meadows’ spatial extent, which may indicate higher risk for vicuñas in meadows (Donadio and Buskirk, 2016). Vicuña density in meadows was also higher, however, which could indicate a dilution effect of predation risk for vicuñas in meadows, reducing risk.

120 I deployed GPS collars on both vicuñas and pumas both of these habitat locations to track the overlap of this predator and prey through time. First, based on lower vicuña abundance and higher puma scat density, I predict that canyons are riskier than meadows, which should be reflected in more direct interactions between this predator and prey. Second, I predicted that vicuñas may engage in diel migrations to avoid risky places during risky times. The results indicate that more direct interactions were observed in canyons than meadows and vicuñas from both habitats reduced risk by avoiding risky areas during risky times.

Methods

Study animals

I used vicuñas (Vicugna vicugna) and pumas (Puma concolor) to test for dynamic interactions between prey and predators. In this system, vicuñas are the primary prey of pumas, and pumas account for approximately 90% of vicuña mortality in most years (Donadio et al.,

2012). Vicuñas are medium-sized ungulates (45-55 kg in this system) that are South American members of the Camelidae family, which also contains llamas, alpacas, and guanacos. Vicuñas naturally inhabit regions in Perú, Boliva, Chile, and Argentina, between 3,000-4,700 m above sea level (Koford 1957), and have been introduced in Ecuador (IUCN 2008). Prized for centuries for their valuable wool, vicuñas were listed as Vulnerable from 1982-1994 (IUCN 2008) and as

Lower Risk/Conservation Dependent in 1996, but are now listed as Least Concern.

Vicuñas occur in harem family groups composed of a territorial male, females, and their young. Family groups are generally smaller than 20 group members, although occasionally groups will occur close proximity to one another, resulting in large (> 100 individuals) herds.

Importantly, vicuñas make diel migrations whereby they move to foraging grounds during the day and retreat to high elevation plains with little vegetation at night (Donadio and Buskirk, 2016;

Renaudeau D’Arc et al., 2000). Presumably this migration incurs a cost of reduced foraging that

121 is outweighed by the reduction in predation risk from pumas, which are largely crepuscular ambush predators. Further, these plains lack the habitat characteristics that conceal pumas, such as rocky outcroppings and dense vegetation, which generally facilitate successful puma hunting.

Study site

This study was conducted in San Guillermo National Park (SGNP), Argentina, in two habitats: a meadow and a canyon. These study locations have been described in detail elsewhere

(Donadio et al., 2012; Donadio and Buskirk, 2016) but, briefly, these locations differ in a number of habitat characteristics that provide indirect indications that meadows may be riskier than canyons but have better foraging opportunities (Donadio and Buskirk, 2016). First, forage availability and quality may be higher in meadows than canyons (Donadio and Buskirk, 2016). In meadows, grass height and cover, which may be more easily digestible, is greater than in canyons, whereas canyons have more sporadic shrubs (Donadio and Buskirk, 2016). However, plants from both habitats contain approximately equal nitrogen content (Donadio and Buskirk,

2016) and vicuñas seem to be generalist herbivores, consuming a wide variety of plants with little preference based on nutritional quality (Borgnia et al., 2010), suggesting vicuñas in canyons may not be nutrient or energy limited. Second, puma kill density is relatively high in meadows than canyons (480% higher in meadows vs 90% higher in canyons) given their spatial extent (Donadio and Buskirk, 2016). Third, vicuñas are more abundant in meadows than shrublands both in SGNP

(based on dung pile transects) and in other regions of their distribution (reviewed in Arzamendia et al., 2006). This indicates that the dilution effect could result in approximately equal risk between habitats. Furthermore, canyons have approximately four times higher densities of puma scat than meadows (Donadio and Buskirk, 2016), suggesting that direct indicators of risk may be greater in this habitat. The overall risk posed in these different habitats is thus difficult to determine without actually measuring encounter rates.

122

Vicuña and puma capture

Between 2014 and 2015, I equipped 24 female vicuñas (12 in canyons, 12 in meadows) and nine pumas (five females, four males; ranges overlap both habitats) with GPS collars. Vicuña were equipped with Lotek GPS 6000SD (Lotek, New Market, Ontario, Canada) collars with three-hour fixes, resulting in between 946-7741 fixes per individual, with fixes recorded up to three years (2014–2016) depending on mortality. Pumas were equipped with Lotek Globalstar collars (Lotek, New Market, Ontario, Canada) with fixes at one- to three- hour intervals for up to four years (2014–2017), resulting in between 762 and 14,970 fixes per individual, depending on mortality. Pumas and vicuñas were captured with The Pennsylvania State University Institutional

Care and Use Committee permit #45139, Foreign CITES permit # 041071, U.S. CITES permit #

MA70993B-2. See Appendix 1 for details.

Statistical analyses

Home ranges

This is the first time that vicuñas have been GPS collared, and the first time that pumas have been collared in this region. Therefore, I report basic spatial analyses for both animals, including home ranges and mean distances between GPS fixes. For both vicuñas and pumas, I used a movement-based kernel density estimator (MKDE) that uses biased random bridges

(BRBs) by incorporating specific movement and activity information (Benhamou and Cornelis

2010). I set the maximum acceptable interval between fixes (tmax) to 6 hours, excluded locations that were >50 meters apart as indicating the animal was inactive (lmin), and the smoothing parameter (hmin) to 30m.

To determine if female vicuña home range sizes differed by year or habitat, I used linear mixed effects models (package lme4) with home range area as the dependent factor, year and

123 habitat as the independent factor, and ID as random variable. Summaries are reported as mean  1 standard error (SE). Puma home range overlapped both types of habitats, and therefore I did not test for differences in puma home ranges between habitats. For pumas, I used a linear mixed effect model with home range area as the dependent factor, year and sex as the independent factors, and individual identity (ID) as the random effect.

Utilization distribution overlap

Although home ranges describe well the space used by individuals and give some basic biology of an animal, activity is not evenly distributed within home ranges. When comparing spatial use by two animals, it may be more informative to describe areas of shared activity, compared to areas where one of the animals may just be passing through or resting (Benhamou et al., 2014). For comparisons between pumas and vicuñas, I created annual utilization distribution overlap indices (UDs; Fieberg and Kochanny (2005); function kerneloverlaphr) from package adehabitatHR in Program R (Calenge 2015; R Foundation for Statistical Computing, Vienna,

Austria) with 95% isopleths, and the href method parameter. I used Bhattacharyya’s affinity, which assumes independence of two animal movements, where no overlap where utilization distribution is equal to 0, and increases as overlap increases (Fieberg and Kochanny 2005). I used linear mixed effect models test the null hypothesis that vicuñas in canyons and meadows did not differ in their UD overlap with pumas. Here, UD was the dependent variable, year and habitat were dependent variables, and ID was the random effect.

Dynamic interactions

To understand whether vicuñas in canyons or meadows were exposed to greater direct risk, I created a subset of data to create pairs of pumas and vicuñas (dyads) with temporal (TO) and spatial (SO) overlap that was greater than 0. On this subset of data, I used the IAB function

124 from R package wildlifeDI (Long 2016), which accounts for serial locations in statistical testing, rather than treating consecutive locations as independent observations, or subsamples (Benhamou et al., 2014). I determined the threshold for determining simultaneous fixes as 90 minutes, equating to the recommended half the time interval of vicuña fixes. I defined the critical distance as 400m, to reflect immediate risk, as pumas can successfully ambush prey from 400m (Pierce et al., 2004).

To determine if predation risk was reduced by daily migrations, I created trajectories of each animal annually based on whether the GPS fix occurred during the night or day, using longitude, latitude, and Julian date. I then tested if spatial overlap between pumas and vicuñas was lower at night than during the day using annual minimum convex polygons for each animal. I used linear mixed models, with spatial overlap (proportion) as the dependent variable, and year, habitat, and time (day or night), and an interaction between habitat and time, as independent factors, and individual ID as the random effect.

Results

Home ranges

Average home ranges were not different between habitats (meadows vs canyons) for the female vicuñas tracked in this study (estimate = 0.9285, SE = 2.50, t = 0.372, p = 0.713; Table 5-

1). In 2015, vicuña home ranges were ~ 2.3 times greater than in 2014 and 2016 (estimate = 8.83,

SE = 2.56, t = 3.453, p = 0.001; mean = 14.15  2.79 vs 6.05  1.28 km2). The mean distances between fixes (180 min) was 579  14 m.

Puma home ranges covered both habitats, and size differed by year (estimate -136  44, t

= -3.09, df = 7, p = 0.018), where home ranges in 2017 were smaller than 2014-2016 (Table 5-2).

Home ranges did not differ between sexes (estimate 97.22  60.25, t = 1.61, df = 7, p = 0.151).

125 Mean distances between fixes (10 minutes to 3 hrs) were approximately twice as long as those for vicuñas, at 902  32 m.

Utilization distributions overlap

I initially focused on all observations between pumas and vicuñas (377 pairs). Overlap between these species overall was low (first quartile = 0, third quartile 0.164; Table 5-3). There was no difference in UD overlap between habitats (estimate = -0.361, df = 337, t = -1.552, p =

0.122). There was a significant effect of year, however, where overlap in 2015 was greater than overlap 2014 and 2016 (estimate=0.705, df = 337, t = 2.558, p = 0.011). There were 16 observations where UD was greater than 1, indicating greater than average home range overlap compared to homogenous space use (Fieberg & Kochanny 2005). All of these overlaps occurred with Puma3 in 2015; 11 occurred in canyons and 5 occurred in meadows.

Dynamic interactions

There was no difference in Benhamou’s interaction index (IAB; the proportion of time

2 moving within a shared home range) between habitats ( 1 = 2.046, p = 0.153). In all pairs

(pumas and vicuñas located within 90 minutes and 400m of each other; n = 111), the expected values of interaction, under the null hypothesis of independent movements, were low (all IAB <

0.077), indicating that when moving within a shared home range, pumas and vicuñas encounter each other rarely. For instance, the maximum IAB value I observed, 0.064, occurred over 140.75 days where Puma6 and Vicuña17 were within 400m of each other within 90 minutes 34 times.

With the highest IAB, but temporal overlap in the lowest quartile, this is one of the highest incidences of contacts in the animal pairs. However, the observed interaction was significantly lower than expected given their overlap in temporal and spatial extent, and so therefore do not meet the criteria to be classified as interacting significantly more than expected. In the canyons,

126 54 pairs of animals interacted based on my criteria, but only two pairs of animals interacted significantly more than expected under the null hypothesis of independent movements. Puma4 and Vicuña31 (p = 0.023), and Puma5 and Vicuña13 (p = 0.008), interacted more than expected.

In the meadows, 55 pairs of animals interacted; Puma12 interacted with Vicuña24, Vicuña25, and

Vicuña35 (all p < 0.047), and Puma4 interacted with Vicuña17 (p = 0.043), more than expected.

Based on MCP overlap, the smallest polygon that can be created around all GPS fixes of an animal, vicuñas in canyons overlapped with pumas more than vicuñas in meadows did (Table

1; estimate = -0.047, df=33, t = -3.160, p = 0.003), and vicuñas from both habitats overlapped more with pumas during the day than during the night (estimate = -0.02, df = 295, t = -2.161, p =

0.031). There was no effect of year on MCP overlap (estimate = -0.002, df = 316, t =-0.575, p =

0.565), nor was there an interaction between habitat and time (estimate = 0.019, df = 294, t =

1.402, p = 0.162).

Discussion

In this study, I measured direct and some measures of indirect predation risk between pumas and vicuñas in two different habitats. I report, for the first time, the home range sizes of vicuñas via GPS collar data. I found that there was little evidence of coordinated movements of attraction or avoidance between vicuñas and pumas in both habitats in response to direct risk, and there was an approximately equal number of pairs of animals that overlapped within 400 m and

90 minutes in each habitat. Lastly, I found that overlap between pairs of animals was lower during the night than day, suggesting that vicuñas daily migrations may result from indirect cues of risky times.

There are a few potential caveats that may have resulted in an under-estimation of indirect risk effects in this system. First, an interaction between pumas and vicuñas could have occurred but not been recorded within my 90-minute observation time ("step bias"; Creel et al.,

127 2013), which would have under-estimated indirect risk. However, this step bias is more likely to be a concern with cursorial predators, which can cover large distances quickly. Secondly, an interaction between an uncollared puma and a collared vicuña could have occurred and would not have been detected by these analyses (“scatter bias”; Creel et al. 2013), which would have under- estimated indirect risk. This scatter bias is likely to be higher in solitary predators, such as pumas, than social predators. Indeed, the density of pumas within SGNP appears to be high compared to other areas in Argentina (Kelly et al., 2008) as I captured six pumas within six trapping nights.

These caveats are difficult, if not impossible, to avoid when studying wild, free-ranging large mammals and the resulting data can nonetheless be informative.

Based on estimates of immediate risk, I found that movements of these two species rarely deviated from random when within the same home range. Although infrequent avoidance of has been observed in a number of other systems (Davies et al., 2016; Eriksen et al., 2009; Sand et al.,

2006; Sivertsen et al., 2016), many other studies suggest prey may respond behaviorally to indirect cues of predation risk that I was not able to address here. For instance, African ungulates tended to aggregate more in the presence of a lion kills than to the actual presence of a lion (Moll et al., 2017). There may also be additive and interactive effects of habitat characteristics and antipredator behaviors, including habitat selection (Latombe et al., 2014; Liley and Creel, 2008;

Martin and Owen-Smith, 2016), and considering these habitat features may increase my ability to detect antipredator behavior of vicuñas towards pumas in the future. For instance, extracting habitat features from locations of vicuña carcasses which were predated upon by pumas may identify particularly risky habitat features, which I could subsequently use for resource selection functions, but the difficulty in gathering GIS layers at the appropriate scale to examine these interactions in this system precluded our use of these analyses. Future work based on predictive maps of risk, including habitat characteristics and locations of kill sites will help me address these potential cues.

128 Here, I report annual home ranges of vicuñas measured using GPS technology. Others have described vicuñas home ranges via observations on foot and by vehicle with an average of

11 fixes per individual (Arzamendia et al., 2018). These home ranges were reported to be much smaller than in this study. Arzamendia et al. (2018), reported that females had home range sizes between 0.19-0.46 km2 versus my estimate of 5-15 km2. These previously-reported home range sizes for a northern region of Argentina were estimated with different parameters, including the estimator and a 50% isopleth (in comparison to my 95% isopleth), which can indicate core areas of activity or home range cores. Future work comparing home ranges of animals in these two areas using similar parameters would identify whether methodological or ecological factors are driving the differences in these estimates.

I found no spatial overlap between vicuñas that occurred within the two habitats (e.g. vicuñas did not move between habitats). This has important implications for population dynamics and genetic movement (Morales et al., 2010). Despite the vicuñas current IUCN status of Least

Concern, the impacts of increased poaching, the unknown role of climate change on vicuñas’ fragile arid environments, and the introduction of livestock to these habitats, leaves this species at risk to declines numbers in the near future (IUCN 2008). It is crucial to provide local, regional, and federal governments with information on habitat use that can be used to design protected areas. Although I found no spatial overlap between vicuñas in each of our habitats, suggesting limited gene flow between our two habitats, future work should focus on dispersing as a potential source of gene flow. Likewise, limited spatial overlap between these two groups may indicate that repopulation efforts may have only localized success.

Vicuña home ranges were larger in 2015 than in 2014 and 2016, possibly indicating lower resource availability in 2015. Vicuñas appear to be water limited. For instance, it is thought that vicuñas select areas containing plants species with high moisture content in response to decreased annual precipitation (Arzamendia et al., 2006), and vicuñas are generally located

129 within 2 km of a water source (Koford 1957; Arzamendia et al., 2018). In my study, decreased precipitation in 2015 may have extended vicuña home ranges to increase access to resources, including plants and water. However, I do not currently have access to these data for this region.

Indirect indicators of risk are divergent in predicting habitat riskiness. Suggesting meadows are riskier, carcass abundance was higher in meadows than canyons, given the relative spatial extent of these habitats but vicuña density was also higher in meadows (Donadio and

Buskirk, 2016) which could indicate a dilution effect of predation risk for vicuñas in this habitat, reducing risk. Suggesting canyons are riskier, puma scat density is higher in canyons than meadows, indicating that pumas may use canyons more intensively than meadows (Donadio and

Buskirk, 2016). Here, one of my indirect measure of risk, spatial overlap, between vicuñas and pumas were higher in canyons than meadows, suggesting canyons may be riskier than meadows.

Lastly, the number of interacting pairs of vicuñas and pumas was approximately equal in these two habitats, but meadow vicuñas may benefit from a dilution effects given their higher abundance.

I found preliminary evidence that daily migrations in vicuñas may reduce spatial overlap with pumas, thus reducing the risk of predation. Pumas are considered crepuscular or nocturnal hunters (Anderson and Lindzey, 2003), and thus dawn, dusk, and night may generally be considered as “risky times” for prey. Therefore, by moving to open areas at night, which are largely devoid of habitat which provides cover for pumas, vicuñas may reduce risk of predation, as pumas likely visit areas where they may be successful, less frequently. Possibly as a result of these diel movements altering densities, days appear to carry the highest risk of predation, as spatial overlap between pumas and vicuñas is highest during the day. Evidence that seasonal migration may reduce predation risk in ungulates has only recently emerged (Bastille-Rousseau et al., 2015; Hebblewhite and Merrill, 2007), and a reduction in risk resulting from diel migrations

130 is sparse (but see Davies et al., 2016). Future research examining the costs and associated benefits would give us a clearer picture of the evolutionary development of this behavior.

Conclusion

Here, I document little avoidance to direct risk by vicuñas, and a reduction in overlap between vicuñas and pumas during the night vs the day, which may represent anticipatory antipredator behaviors by leaving risky areas during risky times. I also present the first estimated home ranges from vicuñas gathered with GPS data. Vicuña home ranges appear to be large and encompass two different habitats daily, which my data suggest may provide them with refuge from predation from pumas. Future work including resource selection function analysis with habitat characteristics will help inform conservation and management plans to plan appropriately for these large home ranges, which may include multiple habitat types that vicuñas use daily. My work provides information on the importance of time and space in examining and predicting antipredatory behaviors.

131 Literature Cited

Anderson CR, Lindzey FG (2003) Estimating cougar predation rates from GPS location clusters.

J Wildl Manage 67: 307–316.

Arzamendia Y, Carbajo AE, Vilá BL (2018) Social group dynamics and composition of managed

wild vicuñas (Vicugna vicugna vicugna) in Jujuy, Argentina. J Ethol 36: 125–134.

Arzamendia Y, Cassini MH, Vilá BL (2006) Habitat use by vicuña Vicugna vicugna in Laguna

Pozuelos Reserve, Jujuy, Argentina. Oryx 40: 198–203.

Bastille-Rousseau G, Potts JR, Schaefer JA, Lewis MA, Ellington EH, Rayl ND, Mahoney SP,

Murray DL (2015) Unveiling trade-offs in resource selection of migratory caribou using a

mechanistic movement model of availability. Ecography 38: 1049–1059.

Benhamou, S. and Cornelis, D. (2010) Incorporating movement behavior and barriers to improve

biological relevance of kernel home range space use estimates. J Wildl Manage 74, 1353-

1360.

Benhamou S, Valeix M, Chamaillé-Jammes S, Macdonald DW, Loveridge AJ (2014) Movement-

based analysis of interactions in African lions. Anim Behav 90: 171–180.

Borgnia M, Vilá BL, Cassini MH (2010) Foraging ecology of vicuña, Vicugna vicugna, in dry

puna of Argentina. Small Rumin Res 88: 44–53.

Boyd II (1996) Temporal scales of foraging in a marine predator. Ecology 77:426-434.

Creel SR, Winnie JA, Christianson D (2013) Underestimating the frequency, strength and cost of

antipredator responses with data from GPS collars: An example with wolves and elk.

Ecol Evol 3: 5189–5200.

Creel SR, Winnie JA, Maxwell B, Hamlin KL, Creel M (2005) Elk alter habitat selection as an

antipredator response to wolves. Ecology 86: 3387–3397.

132 Davies AB, Tambling CJ, Kerley GIH, Asner GP (2016) Limited spatial response to direct

predation risk by African herbivores following predator reintroduction. Ecol Evol 6:

5728–5748.

Donadio E, Buskirk SW (2016) Linking predation risk, ungulate antipredator responses, and

patterns of vegetation in the high Andes. J Mammal gyw020.

Donadio E, Buskirk SW, Novaro AJ (2012) Juvenile and adult mortality patterns in a vicuña

(Vicugna vicugna) population. J Mammal 93: 1–9.

Eccard JA, Mener JK, Heurich M (2017) European roe deer increase vigilance when raced with

immediate predation risk by Eurasian lynx. Ethology 123: 30–40.

Eriksen A, Wabakken P, Zimmermann B, Andreassen HP, Arnemo JM, Gundersen H, Milner JM,

Liberg O, Linnell J, Pedersen HC, et al. (2009) Encounter frequencies between GPS-

collared wolves (Canis lupus) and moose (Alces alces) in a Scandinavian wolf territory.

Ecol Res 24: 547–557.

Fieberg, J. and Kochanny, C.O. (2005) Quantifying home-range overlap: the importance of the

utilization distribution. J Wildl Manage, 69, 1346--1359.

Hebblewhite M, Merrill EH (2007) Multiscale wolf predation risk for elk: does migration reduce

risk? Oecologia 152: 377–387.

The IUCN Red List of Threatened Species. Version 2018-1. . Downloaded

on 11 August 2018.

Kelly MJ, Noss AJ, Di Bitetti MS, Maffei L, Arispe RL, Paviolo A, De Angelo CD, Di Blanco

YE (2008) Estimating puma densities from camera trapping across three study sites:

Bolivia, Argentina, and Belize. J Mammal 89: 408–418.

Kuijper DPJ, Verwijmeren M, Churski M, Zbyryt A, Schmidt K, Jedrzejewska B, Smit C (2014)

What cues do ungulates use to assess predation risk in dense temperate forests? PLoS

One 9: 1–12.

133 Koford CB (1957) The vicuña and the puna. Ecol Monogr 27: 153–219.

Latombe G, Fortin D, Parrott L (2014) Spatio-temporal dynamics in the response of woodland

caribou and moose to the passage of grey wolf. J Anim Ecol 83: 185–198.

Laundré JW, Hernández L, Ripple WJ (2010) The landscape of fear: ecological implications of

being afraid. Open Ecol J 3: 1–7.

Liley S, Creel SR (2008) What best explains vigilance in elk: Characteristics of prey, predators,

or the environment? Behav Ecol 19: 245–254.

Lima SL, Bednekoff PA (1999) Back to the basics of antipredatory vigilance: can nonvigilant

animals detect attack? Anim Behav 58: 537–543.

Long J (2016) wildlifeDI: Calculate indices of dynamic interaction for wildlife telemetry data. R

package version 0.3. http://jedalong.github.io/wildlifeDI

Martin J, Owen-Smith N (2016) Habitat selectivity influences the reactive responses of African

ungulates to encounters with lions. Anim Behav 116: 163–170.

Moll RJ, Killion AK, Montgomery RA, Tambling CJ, Hayward MW (2016) Spatial patterns of

African ungulate aggregation reveal complex but limited risk effects from reintroduced

carnivores. Ecology 97: 1123–1134.

Moll RJ, Redilla KM, Mudumba T, Muneza AB, Gray SM, Abade L, Hayward MW, Millspaugh

JJ, Montgomery RA (2017) The many faces of fear: a synthesis of the methodological

variation in characterizing predation risk. J Anim Ecol. 86: 749–765.

Morales JM, Moorcroft PR, Matthiopoulos J, Frair JL, Kie JG, Powell RA, Merrill EH, Haydon

DT (2010) Building the bridge between animal movement and population dynamics.

Philos Trans R Soc B Biol Sci. 365, 2289–2301.

Orrock JL (2010) When the ghost of predation has passed: Do rodents from islands with and

without fox predators exhibit aversion to fox cues? Ethology 116: 338–345.

134 Packer C, Swanson A, Ikanda D, Kushnir H (2011) Fear of darkness, the full moon and the

nocturnal ecology of African lions. PLoS One 6: 4–7.

Palmer MS, Fieberg J, Swanson A, Kosmala M, Packer C (2017) A ‘dynamic’ landscape of fear:

prey responses to spatiotemporal variations in predation risk across the lunar cycle. Ecol

Lett. 20:1364–1373.

Pierce BM, Bowyer RT, Bleich VC (2004) Habitat selection by mule deer: forage benefits or risk

of predation? J Wildl Manage 68: 533–541.

Proffitt KM, Grigg JL, Hamlin KL, Garrott RA (2009) Contrasting effects of wolves and human

hunters on elk behavioral responses to predation risk. J Wildl Manage 73: 345–356.

Renaudeau D’Arc N, Cassini MH, Vilá BL (2000) Habitat use by vicuñas Vicugna vicugna in the

Laguna Blanca Reserve (Catamarca, Argentina). J Arid Environ 46: 107–115.

Russell RW, Hunt GL, Coyle KO, Cooney RT (1992) Foraging in a fractal environment: Prey

system spatial patterns in a marine predator. Landscape Ecology 7: 195–209.

Sand H, Wikenros C, Wabakken P, Liberg O (2006) Cross-continental differences in patterns of

predation: will naive moose in Scandinavia ever learn? Proc R Soc B Biol Sci 273: 1421–

1427.

Schmitz OJ (2008) Effects of predator hunting mode on grassland ecosystem function. Science

319: 952–954.

Sivertsen TR, Åhman B, Steyaert SMJG, Rönnegård L, Frank J, Segerström P, Støen OG, Skarin

A (2016) Reindeer habitat selection under the risk of brown bear predation during calving

season. Ecosphere 7: e01583.

Thaker M, Vanak AT, Owen CR, Ogden MB, Niemann SM, Slotow R (2015) Minimizing

predation risk in a landscape of multiple predators: effects on the spatial distribution of

African ungulates. Ecology 92: 398–407.

135 Thaker M, Vanak AT, Owen CR, Ogden MB, Slotow R (2010) Group dynamics of zebra and

wildebeest in a woodland savanna: Effects of predation risk and habitat density. PLoS

One 5: e12758.

Winnie JA, Creel SR (2007) Sex-specific behavioural responses of elk to spatial and temporal

variation in the threat of wolf predation. Anim Behav 73: 215–225.

Zaret TM, Suffern JS (1976) Vertical migration in zooplankton as a predator avoidance

mechanism. Limnol Oceanogr 21: 804–813.

136

Table 5-1. Description of female vicuña home ranges by year and habitat.

Year Habitat Mean Area (km2) Number of vicuñas observed 2014 Canyon 6.25  1.01 6 2014 Meadow 4.92  0.94 10 2015 Canyon 13.77  4.18 10 2015 Meadow 14.58  3.88 9 2016 Canyon 5.32  1.83 6 2016 Meadow 7.71  1.79 8

137 Table 5-2. Description of puma home ranges by year. Puma home ranges covered both habitats.

Year Mean home range Number of pumas area (km2) observed 2014 179.86  43.81 6

2015 219.17  82.64 4

2016 139.05  22.98 4

2017 93.09  22.95 5

138 Table 5-3. Home range overlap between pumas and vicuñas. Overlap was greater in 2015 than in

2014 or 2016.

Year Habitat Mean HR Overlap  Number of SE Observations 2014 Canyon 0.14  0.03 42 2014 Meadow 0.09  0.02 60 2015 Canyon 1.24  0.48 77 2015 Meadow 0.41  0.19 63 2016 Canyon 0.12  0.03 42 2016 Meadow 0.14  0.02 48

139 Table 5-4. Spatial overlap (proportion) between vicuñas and pumas based on minimum convex polygons.

Time Habitat Overlap Number of Mean observations 1 Day Canyon 0.07  0.01 75 2 Day Meadow 0.03  0 87 3 Night Canyon 0.05  0.01 74 4 Night Meadow 0.02  0 86

140 Appendix

Vicuña and Puma Capture and Chemical Immobilization

I searched for vicuñas from a truck at idle speed along the roads in San Guillermo

National Park. A female in a family group consisting of at least one male and one female was identified as a target for collaring and sampling. I then approached the female indirectly on foot, finding that a direct trajectory initiated flight behavior sooner than a wandering approach in most cases. Depending on wind speed, direction, and habitat, it was necessary to move within at least

35 m of the individual.

Vicuña chemical immobilization

Anesthetics were delivered remotely via a CO2 powered Dan-Inject dart equipped with

1–3 cc darts and barbed, collared, or plain 1” needles. Three different chemical immobilization methodologies were tested: 1) a combination of xylazine and ketamine with Yohimbine as an antagonist; 2) Thiafentanil (10 mg/ml, ZooPharm Inc., Laramie, Wyoming, USA) with

Naltrexone (50 mg/mL; name) as an antagonist, and 3) Carfentanil with Naltrexone as an antagonist. In vicuñas, large visual differences in adult masses are not evident, thus weight was estimated as 45 kg for all individuals.

A combination of xylazine and ketamine was chosen for initial immobilization efforts in vicuñas because of their success to chemically immobilize a number of ungulates (Kreeger et al.

2001, Frowler 2011), and drug costs are moderate compared to other options. Ketamine (2.5 mg/kg) and xylazine (0.4 mg/kg) were mixed, and delivered via 2–3 cc darts intramuscularly

(Fowler 2011). However, despite success with these drugs in other ungulates, vicuñas proved to be nearly immune to this drug combination, preventing successful captures. Therefore,

Thiafentanil was used, a drug that has been used successfully in other species where xylazine and ketamine were ineffective (Kreeger et al. 2001).

141 Thiafentanil oxalate (also known as A3080) has not been previously been used in wild vicuñas. Therefore, to maximize the likelihood of success and to maintain a large margin of safety for the animals, captures began with relatively high doses (10 mg per dart) of thiafentanil.

With the resulting rapid induction times, doses were subsequently decreased to 5 mg (0.1 mg/kg) and again to 3 mg (0.6 mg/kg) using 1-3 cc darts. If the entire dose was not injected (generally due to dart malfunction), and the animal did not become immobilized, another 3 mg dart was deployed. Alternatively, if the individual was only lightly immobilized (characterized by early attempts to stand once in a sternally recumbent position), an additional 1–3 mgs were injected IM by hand into the hindquarters. Naltrexone served as the antagonist and was delivered IM by hand at 10 mg Naltrexone/1 mg thiafentanil. In late 2014, thiafentanil oxalate became highly regulated in the United States and was no longer available for field use in early 2015. Carfentanil, a drug similar to thiafentanil, became the next best option because of its similarities to thiafentanil

(Fowler 2011).

Carfentanil has been the primary anesthetic used to successfully chemically immobilize various ungulates (Cushing et al., 2011; Kilgallon et al., 2010; Napier et al., 2011) including guanacos (Lama guanicoe), a sister taxon of vicuña (Karesh et al., 2011). Again, relatively high dosage of 3 mg (suggested for guanacos with 2X the body weight of vicuñas) were initially used, but doses decreased 1.5 mg. If the animal was not immobilized, another dart or an injection of 1.5 mg was delivered IM. Naltrexone served as the antagonist and was delivered IM by hand at 100 mg/1mg carfentanil.

142 Vicuña sampling and collaring

After immobilization by either thiafentanil or carfentanil, vicuñas were held in a sternally recumbent position, and immediately blindfolded with their neck supported in straight, upright position. No hobbling or further restraint was necessary with either drug. With each capture, I recorded the time of injection, induction time, and time to sternal recumbancy. Throughout sampling and collaring I monitored rectal temperature, respiration rate, heart rate, and arterial oxygen saturation (SpO2) approximately every 10 minutes. Approximate age was recorded based on tooth eruption and wear (Donadio et al., 2012). Body measurements including girth, metatarsus length, head length, ear length, neck circumference, head circumference and weight

[Big Game digital scale (150 ± 0.09 kg)] were recorded.

Each individual was then equipped with a Lotek GPS 6000 (minimum neck size 26 cm) collar and fixed with a numbered, colored ear tag. In 2014, vicuñas were tagged with Destron

Fearing ™ DuFlex ™ Visual Identification Matrix Tags for Livestock, but they appeared agitating to vicuñas in windy conditions. Therefore, in 2015 they were tagged with smaller,

Perma-Flex Medium Cattle Ear Tags, which appeared to disturb vicuñas less. These tags permitted individual visual identification when GPS collars were not transmitting.

Individuals were lifted and turned to face the direction in which their family group had moved before antagonist injection. Once the antagonist was delivered and during late recovery, the blindfold was removed, but animals remained physically restrained in a sternally recumbant position until they were deemed capable of standing and moving.

Puma capture and chemical immobilization

Puma captures were conducted by a large felid specialist, using leg-hold snares. The leg- hold snares were highly modified from their primitive form, including claw-traps, and incorporated all suggestions found in the literature to ameliorate join and musculoskeletal injury,

143 including anchors, springs, and transmitters. The snares were anchored cable nooses that sprung up and around the leg of the animal when it stepped on a hidden trigger. All snares were placed on puma travel routes or in areas of known activity. Snares were anchored to immoveable objects

(e.g. large boulders, solid bed-rock, earth anchors capable of holding >300 lbs of pulling force) with steel chain, contained a shock-absorbing steel spring to minimize joint and muscle strain, and a trap site transmitter to facilitate rapid detection of and response to a capture event. Snares were checked every 2-3 hours when deployed to ensure rapid response to a capture. Once a puma was captured, it’s weight was estimated, and ketamine (2 mg/kg) and xylazine (2 mg/kg) were administered with a single injection from a CO2-powered dart rifle. Pumas were equipped with

Lotek Globalstar collars (Lotek, New Market, Ontario, Canada), and injected Yohimbine (0.125 mg/kg) as the reversal.

Literature Cited

Cushing A, McClean M, Stanford M, Lohe T, Alcantar BE, Chirife AD (2011) Anesthesia of

Tibetan Yak (Bos grunniens) using thiafentanil-xylazine and carfentanil-xylazine. J Zoo

Wildl Med 42: 713–717.

Donadio E, Buskirk SW, Novaro AJ (2012) Juvenile and adult mortality patterns in a vicuña

(Vicugna vicugna) population. J Mammal 93: 1–9.

Fowler, Murray E. Medicine and surgery of South American camelids: llama, alpaca, vicuña,

guanaco. No. Ed. 2. Iowa State University Press., 2011.

Karesh WB, Uhart MM, Dierenfeld ES, Braselton I, Torres A, House C, Puche H, Cook RA,

Dierenfeld S (1998) Health evaluation of free-ranging guanaco (Lama guanicoe). J Zoo

Wildl Med 29: 134–141.

144 Kilgallon CP, Lamberski N, Larsen RS (2010) Comparison of thiafenantil-xylazine and

carfentanil-xylazine for immobilization of gemsbok (Oryx gazella). J Zoo Wildl Med 41:

567–571.

Kreeger, T. J., Cook, W. E., Piché, C. A., & Smith, T. (2001). Anesthesia of pronghorns using

thiafentanil or thiafentanil plus xylazine. J Wildl Mangmt, 25-28.

Napier JE, Loskutoff NM, Simmons LG, Armstrong DL (2011) Comparison of carfentanil-

xylazine and thiafentanil-medetomidine in electroejaculation of captive gaur (Bos

gaurus). J Zoo Wildl Med 42: 430–436.

VITA

Catharine Elizabeth Pritchard

EDUCATION

Ph.D. Wildlife and Fisheries Science. Pennsylvania State University, December 2018 M.S. Biology. University of Oregon. December 2013 B.S. Biology University of Wisconsin, Superior, Wisconsin. August 2009

PUBLICATIONS

Pritchard CE, Rimler RN, Rumrill SS, Emlet RB, Shanks AA (2016). Variation in larval supply and recruitment of Ostrea lurida in the Coos Bay estuary, Oregon, USA. Marine Ecology Progress Series 560: 159-171 Zaneveld JR, Burkepile De, Shantz AA, Pritchard CE, McMinds R, Payet J, Welsh R, Simoes- Correa A, Lemoine NP, Rosales S, Fuchs C, Vega Thurber R (2016). Overfishing, pollution, and thermal stress interact to disrupt coral reefs down to a microbial scale. Nature Communications 7: 11833 Pritchard CE, Shanks A, Rimler R, Oates M, Rumrill S (2015) The Olympia Oyster, Ostrea lurida: recent advances in natural history, ecology and restoration. Journal of Shellfish Research 34: 259-271 Burkepile DE, Allgeier JE, Shantz A, Pritchard C, Lemoine N, Bhatti L, Layman C (2013) Nutrient supply from fishes facilitates macroalgae and suppresses corals in a Caribbean coral reef ecosystem. Scientific Reports 3:1493 Vega R, Burkepile D, Simoes-Correa A, Thurber A, Shantz A, Welsh R, Pritchard C, Rosales S (2012) Macroalgae decrease growth and alter microbial community structure of the reef- building coral, Porites astreoides. PLoS ONE 7: e44246

GRANTS & AWARDS

2014-2018 Graduate Teaching Assistantship, Pennsylvania State University ($184,000) 2016 Charlotte Mangum Student Support. Society for Integrative and Comparative Biology ($100) 2016 College of Agricultural Sciences Travel Grant, Pennsylvania State University ($300) 2016 Summer Tuition Assistance Program Fellowship, Pennsylvania State University ($1,985) 2013 Oregon Society of Conchologists Scholarship for Mollusc Research ($500) 2012 Melbourne R. Carriker Student Research Award, American Malacological Society ($500) 2012 Student Sustainability Fund, University of Oregon ($4,300)