Ambassador Animal Welfare: Bonnie Ann Baird

Home , Peoria Zoo, Rancho (monkey)

AMBASSADOR ANIMAL WELFARE: USING BEHAVIORAL AND PHYSIOLOGICAL INDICATORS TO ASSESS THE WELL-BEING OF ANIMALS USED FOR EDUCATION PROGRAMS IN ZOOS

BONNIE ANN BAIRD

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Department of Biology CASE WESTERN RESERVE UNIVERSITY

May 2018 CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the dissertation of Bonnie Ann Baird, candidate for the degree of Doctor of Philosophy*

Mandi W. Schook, Ph.D. Committee Chair

Kristen E. Lukas, Ph.D. Committee Member

Mark A. Willis, Ph.D. Committee Member

Nadja C. Wielebnowski, Ph.D. Committee Member

Christopher W. Kuhar, Ph.D. Committee Member

Date of Defense: 19 March 2018 *We also certify that written approval has been obtained for any proprietary material contained therein.

To my dad, who taught me to love and respect all creatures great and small, and raised me to believe that I could be anything. I miss you every day, but I know you’ll be with me

in all my future adventures.

iii

TABLE OF CONTENTS

List of Tables. …………………………………………………………………………… ii

List of Figures. ………………………………………………………………………….. iii

Acknowledgements. …………………………………………………………………….. iv

Abstract. …………………………………………………………………………………. 1

Chapter One: Introduction. …………………………………………………………….. 3

Chapter Two: The influence of management role and handling on indicators of welfare

in armadillos, hedgehogs, and red-tailed hawks. ………………………………. 37

Chapter Three: Personality assessment in zoo-housed cheetahs. …………………….. 75

Chapter Four: Assessing welfare in zoo-housed cheetahs in different management roles. … 93

Chapter Five: General discussion and future directions. ……………………………. 133

Appendix I: Husbandry and demographic survey used for cheetahs. ……………….. 141

Appendix II: Cheetah personality survey. …………………………………………… 143

Appendix III: Daily log form used to record human interaction, exercise, and enrichment

in cheetahs. ……………………………………………………………………. 145

References. …………………………………………………………………………… 146

LIST OF TABLES

Chapter Two: Table 2.1: Husbandry and demographic factors for armadillos in Experiment One. …. 43 Table 2.2: Husbandry and demographic factors for species in Experiment Two. …….. 44 Table 2.3: A) Ethogram used for behavioral data collection on armadillos, hedgehogs, and red-tailed hawks for both Experiment One and Experiment Two. B) Modifications to the “Rest” category used when collecting behavioral data in hawks. ………………………………………………………………………….. 49

Chapter Three: Table 3.1: Subjects, raters, and institutional reliability used in the cheetah personality assessment. ……………………………………………………………………... 79 Table 3.2: Behavioral definitions of adjectives used in personality questionnaire. …… 80 Table 3.3: Factor loadings for four components of cheetah personality identified through Principal Components Analysis. ……………………………………………….. 86

Chapter Four: Table 4.1: Husbandry and demographic factors measures in 73 cheetahs. ……………. 97 Table 4.2: Broad categories used to classify human interaction, exercise, and enrichment for comparison across institutions. …………………………………………… 100 Table 4.3: Ethogram used for behavioral data collection in cheetahs. ………………. 104 Table 4.4: Fixed factors analyzed in relation to FGM and behavior using linear mixed models. ………………………………………………………………………... 108

LIST OF FIGURES

Chapter Two: Figure 2.1: FGM concentrations in response to an ACTH injection in armadillos (A-C) and a red-tailed hawk (D) and FGM concentrations in response to a veterinary exam in a hedgehog (E). ………………………………………………………………………. 55 Figure 2.2: Box plots representing average corticosterone concentrations of education, exhibit, and off exhibit armadillos. ………………………………………………… 56 Figure 2.3: Comparison of individual behaviors between education, exhibit, and off exhibit armadillos. ……………………………………………………………… 59 Figure 2.4: Mean (+SEM) FGM concentrations for four La Plata armadillos (A, B, C, D) during the three phases of Experiment Two. ………………………………………. 61 Figure 2.5: Individual differences in mean FGM concentrations for six representative African hedgehogs in each phase of Experiment Two. …………………………………………. 63 Figure 2.6: Average (+SEM) FGM levels in two representative red-tailed hawks during each three-week phase of Experiment Two. ………………………………………………… 64

Chapter Four: Figure 4.1: Least-square mean (± 95% CI) FGM concentrations for cheetahs in different management roles. ……………………………………………………………. 114 Figure 4.2: Interaction between dominant personality and on-exhibit housing on FGM concentrations. ………………………………………………………………... 116 Figure 4.3: Box Plots representing the relationship between primary management role and behavioral diversity scores. ………………………………………………. 118 Figure 4.4: Scatter plots of behavioral diversity scores and A) average and B) baseline FGM concentrations for 39 cheetahs, separated by sex. ……………………………………. 120

iii

ACKNOWLEDGEMENTS

I will always be grateful to all those that guided and supported me along this journey that has been full of the highest of highs and the lowest of lows. First and foremost, I need to thank my wonderful husband, Tommy Klypchak for your unconditional love and patience for the last 15(!) years. Having you as my solid rock at home has enabled me to fly and reach for my dreams. You are truly the most caring and selfless person I have ever met, and I am so lucky to have you in my corner. There’s no one I would rather partner with on this crazy journey of life, and I can’t wait to see what the future has in store for us!

I have to acknowledge my advisor, Dr. Kristen Lukas. You have played such a major role in who I am today, both personally and professionally. I am eternally grateful for your guidance and friendship, and for all of the heart to heart talks. Thanks to you, I was able to do something I never thought I would be able to – travel to Africa (three times!). Part of me still can’t believe that this girl from North Ridgeville, Ohio was able to have such an amazing opportunity and the other part can’t wait to go back. I still have to see wild elephants! Thank you for helping to make my dreams a reality and enabling me to develop my skills and use them to make a difference in the world.

I have also been so fortunate to have a committee full of supportive and brilliant scientists. I’m grateful for the advice and expertise of Dr. Nadja Wielebnowski throughout this process. Despite being on the other side of the country, she has always been ready and willing to answer my endless questions, and I truly appreciate that. Dr.

Mark Willis, thanks for all the great conversations and advice and for recognizing

iv something in me that I didn’t even see in myself yet. Your guidance and advice has meant more to me than you know. Finally, despite technically only serving in an

“advisory role” on my committee for the last 6 years, Dr. Chris Kuhar has never missed a committee meeting or failed to provide feedback on anything I’ve sent him, even with all the responsibilities of a zoo director. Thank you all for taking the time to help develop me into a professional in this field. I hope I can make you proud!

None of this would have been possible without my friends and colleagues at

Cleveland Metroparks Zoo. A very special thank you to Laura Amendolagine, the

Endocrinology Lab Manager for your endless patience, expertise, and for always finding a place in the freezer for thousands of poop samples and making sacrifices to the Assay

Gods with me. Thank you to Ingrid Rinker and Elaine Leickly for volunteering to spend your retirement years crushing fecal samples, and always smiling about it. To all the keepers and curators, especially Andi Kornak, Travis Vineyard, Tad Schoffner, Elena

Less and the PCA team, and Chris Peterson and the rest of the elephant team, your leadership, passion, and dedication to your work is inspiring. I have thoroughly enjoyed working with and learning from each of you. You have all helped me to become a better scientist and a better zoo professional. I look forward to continuing to work with you all in the future. To my academic brothers and sisters that I’ve shared an office with, Dr.

Grace Fuller, Dr. Jason Wark, Austin Leeds, and Laura Bernstein-Kurtycz, thank you for your personal and professional support, for the intellectual and not so intellectual conversations, and most importantly, for your friendship.

To my primary advisor, Dr. Mandi Schook, I don’t think there are enough words to express how grateful I am for all you’ve done to help me grow. Thank you for taking a

v chance on me and for never letting me give up. Thanks to you, I have transformed from someone who likes animals and behavior to a multidisciplinary scientist. Your guidance has helped me to accomplish things that I never thought I could, and I can’t wait to see what the future holds for me. I’m grateful to call you a colleague and a friend. Thank you for being you.

Finally, I want to thank all the staff, students, and volunteers that collected data at each participating institution to make this research possible: Cleveland Metroparks Zoo,

Akron Zoo, Bergan County Zoo, Binder Park Zoo, Busch Gardens Tampa, Cincinnati

Zoo and Botanical Garden, Columbus Zoo and Aquarium, Denver Zoo, Florida

Aquarium, Fresno Chaffee Zoo, The Good Zoo at Oglebay, Happy Hollow Park and Zoo,

Jacksonville Zoo, Lincoln Park Zoo, Tampa’s Lowry Park Zoo, Memphis Zoo, Oregon

Zoo, Peoria Zoo, Philadelphia Zoo, Point Defiance Zoo and Aquarium, San Antonio Zoo,

Seneca Park Zoo, Smithsonian’s National Zoo and Conservation Biology Institute,

Toronto Zoo, Wildlife Safari, The Wilds, Zoo Atlanta, and Zoo Miami. This research was also funded in part by the Association of Zoos and Aquariums Conservation Grants Fund.

Ambassador Animal Welfare: Using Behavioral and Physiological Indicators to Assess

the Well-Being of Animals Used for Education Programs in Zoos.

BONNIE ANN BAIRD

Abstract

Modern accredited zoos strive to adhere to the highest standards of animal care while simultaneously providing meaningful educational experiences for their guests. One avenue where these goals intersect is the widely employed practice of using ambassador animals in education and outreach programs. However, there are currently only limited and anecdotal data on potential welfare impacts, positive or negative, of this increasingly popular practice on the animals themselves. The overall objective of this thesis is to evaluate welfare in ambassador animals with the goal of informing management recommendations that enable these animals to experience optimal welfare while helping to advance the educational mission of zoos and aquariums. A comprehensive approach that includes multiple indicators of welfare, several taxa, and a large sample size that spans multiple institutions is necessary for drawing meaningful conclusions that can then be used to inform management recommendations. Using both behavior and fecal glucocorticoid metabolites (FGM), armadillos, hedgehogs, red-tailed hawks, and cheetahs were evaluated in a series of multi-institutional studies. Mixed model analysis revealed

FGM and undesirable behaviors did not differ between ambassador, exhibit, and off

1 exhibit management roles in armadillos and cheetahs. There was also no effect of handling specifically for education programs on measures of welfare in all species.

However, the overall amount of handling that an animal experienced was positively correlated with FGM and associated with several behaviors in armadillos, hedgehogs, and hawks. In addition, environmental factors such as substrate depth and enclosure size were consistently related to FGM and behavior in these species. For cheetahs, individual personality was measured using established methodologies and included in the larger analysis. Factors such as on-exhibit housing and protected contact management were associated with FGM, but there is evidence that these effects vary based on individual personality and management role. These results indicate that management role is not the primary contributor to welfare in these four species. Rather, husbandry practices and the housing environment are more important factors, and individual differences should be considered when making management recommendations designed to improve well-being in zoo-housed animals.

CHAPTER ONE: Introduction

Today’s accredited modern zoos have evolved from menagerie-style collections of rare and exotic animals into centers for conservation, education, and research. Part of this evolution has occurred because of a shift in the way society views animal welfare, leading to an increased demand to justify the existence of zoos beyond pure entertainment. Although making meaningful contributions to field conservation is consistently named as the top expectation the public has for zoos, providing excellent animal care and educational experiences follow closely behind. In fact, the mission statement for the Association of Zoos and Aquariums (AZA) encourages member institutions “to be leaders in animal welfare, public engagement, and the conservation of species” (aza.org). In the pursuit of these modern goals, accredited zoos have increasingly become centers for research, using scientific evidence to determine best practices for animal care and the best educational methods for connecting zoo visitors with wildlife while fostering pro-environmental attitudes and actions.

The dramatic increase in the amount of zoo-based research over the last 30 years has certainly resulted in improvements on both the animal care and guest education fronts. Advances in husbandry practices, nutrition, and veterinary science have led to zoo-housed animals living longer, healthier lives, to the point that geriatric care is now a major topic of discussion throughout the zoo community (Föllmi et al., 2007). Informed by applied research, zoo enclosures are increasingly designed to promote positive welfare by housing species-appropriate social groups, providing opportunities for animals to exhibit choice and control over their environment, and allowing them to engage in a full repertoire of natural behaviors (Coe, 2004; Owen, Swaisgood, Czekala, & Lindburg,

2005; White, Houser, Fuller, Taylor, & Elliott, 2003). By demonstrating a commitment to the welfare of animals in their care, zoos can accomplish their educational goals more readily. Zoo education programs aim to foster a connection to wildlife and nature by increasing knowledge, changing attitudes, and inspiring visitors to take action for the environment. One of the most popular methods of creating this connection to wildlife is the use of ambassador animals in education programs. Having an up-close interaction with a live animal has been shown to increase knowledge retention and change attitudes by creating empathy for the animal, the species, and the natural world as a whole

(Morgan & Gramann, 1989; Povey, 2002; Povey & Rios, 2002; Sherwood, Rallis, &

Stone, 1989). As a result, ambassador animal programs in zoos are growing rapidly, both in number of animals and in diversity of species utilized.

Although the goals of education and animal welfare are equally important to a modern zoo, there has been very little intersection of the two in the scientific literature.

Often, evidence-based best practices recommendations for zoo-housed species do not apply to ambassador animals, because of the varied housing, husbandry, and human interaction conditions that these individuals experience compared to exhibit animals.

With the increasing popularity and use of ambassador animals, there is a need to understand how these experiences relate to welfare. Some have said that ambassador animals likely have good welfare because the increased personal attention, training, human interaction, and varied routines might be enriching. However, the same reasoning has been used to argue that ambassador animals have compromised welfare and that these experiences are more stressful than enriching (Brando & Buchanan-Smith, 2017). There is unfortunately little scientific evidence to support one perspective over the other.

Understanding how ambassador animals compare to their counterparts housed on exhibit and which aspects of the environment and/or husbandry practices are associated with indicators of welfare is critical for achieving both effective guest education and the highest standards of animal care in zoos.

Animal Welfare

Definition of Welfare

Any discussion of animal welfare must begin with a complete and thorough definition of the term. This task proves to be more difficult than it sounds, as the concept of what “animal welfare” truly means has been debated for decades (Broom & Johnson,

1993). Early ideas about welfare focused solely on biological fitness – if the animal is surviving and reproducing, welfare must be adequate (Barnett & Hemsworth, 1990); or the prevention of suffering– if an animal is not suffering, then it must be experiencing good, or at least adequate, welfare (Maple & Perdue, 2013). The problem with these concepts is that they don’t explain what is meant by “suffering,” and they rely on the assumption that the absence of a negative equals a positive. Later definitions of welfare began to include the affective state of the animal (Dawkins, 1990; Mason & Mendl, 1993;

Mason & Veasey, 2010). Good welfare is associated with the experience of positive emotions (pleasure, joy), and poor welfare is associated with negative emotional states

(pain, fear, suffering; Yeates & Main, 2008). These definitions are an improvement because they define both positive and negative indicators of welfare, and refer to the psychological, rather than just the physical state of the animal. However, psychological definitions alone would prove problematic when trying to objectively assess the welfare

5 state of an individual, as measuring emotional states of non-human animals is challenging at best.

Other conceptions of welfare have focused more on the interaction of the individual and its environment. (Broom, 1986) states “the welfare of an individual is its state as regards to its attempt to cope with its environment.” All animals, including humans, regularly experience various environmental challenges that exert some type of demand on the body’s response systems (Broom, 2001). These challenges can be physical

(pathogens, injury, attack by a predator) or psychological (excessive or deficient mental stimulation, inability to control interactions with the environment). Coping refers to the ability of the animal to maintain control of mental and bodily stability in the face of these challenges (Broom & Johnson, 1993). Welfare, therefore, is related to the ease or difficulty the animal has when attempting to cope with its environment, and to what degree the positive or negative aspects of the environment affect the individual (Broom,

2001). Failure to cope can occur when the animal is experiencing pain, fear, or difficulty controlling environmental interactions due to frustration, under or overstimulation, or too much unpredictability – all signs of poor welfare (Broom, 1991).

This definition also has several implications that help to make it more applicable to the practical assessment of animal welfare (described in Broom, 1991; Broom &

Johnson, 1993). First, welfare is a characteristic of the individual animal, not something that can be given to it. Any resources or interventions provided by humans are not themselves “good welfare;” they are attempts to improve the welfare state of the individual animal. Second, welfare exists on a continuum from very poor to very good

(Broom, 1986, 1991). It is not sufficient to speak of “ensuring welfare” in animals; we

6 want to “improve” or “ensure good welfare” (Broom & Johnson, 1993). The most important implication of Broom’s (1986) definition is that welfare can be measured in a scientific way, independent of moral considerations, providing the foundation for animal welfare as a scientific discipline.

The current AZA definition of welfare incorporates all of these concepts, emphasizing the individual, measurable nature of welfare, and the idea that welfare exists on a continuum over time: “Animal welfare refers to an animal’s collective physical, mental, and emotional states over a period of time, and is measured on a continuum from good to poor.” (AZA Animal Welfare Committee, 2012). This is the definition that will be utilized throughout this project.

Measuring Animal Welfare

The search for reliable indicators of a given welfare state has been a major focus of welfare science for the last few decades (Dawkins, 2001; Mason & Mendl, 1993).

Improvements to welfare have typically centered around three main philosophies: 1) ensure good physical health, 2) minimize negative and maximize positive affective states, and 3) allow for expression of natural behaviors (Fraser, 2009). These three objectives have led to the major avenues of animal welfare measurement: health, affective state

(assessed via physiological measures), and behavior. The most basic aspects of good welfare are related to physical health, and are in fact specifically addressed by two of the five freedoms: Freedom from hunger, thirst, or malnutrition and Freedom from pain, injury, and disease (FAWC, 1992). Although physical health is the basis of all good

7 welfare, the importance of psychological and emotional health is now being realized and new indicators of these aspects of welfare are being actively developed (Dawkins, 2001).

Animal behavior is probably the most common measure of welfare, and it is often the most obvious indicator that an animal is having difficulty coping with environmental challenges (Hill & Broom, 2009). The presence of stereotypic behaviors (pacing, rocking, hair plucking) has long been associated with compromised welfare (Broom, 1986, 1991;

Dawkins, 1988), but this may not always be accurate. More recent research indicates that understanding the factors contributing to and motivations behind stereotypic behaviors will help us to better understand their welfare implications. Many stereotypies seem to arise from a frustrated behavioral motivation – the animal is unable to perform a desired behavior that is essential for physical or psychological health and thus expresses a surrogate behavior in an attempt to cope (Dawkins, 1988; Mason, Clubb, Latham, &

Vickery, 2007). These behaviors are often reduced or eliminated when the animal is provided with an opportunity to display the preferred behavior (reviewed in Mason et al.,

2007). Thus, environmental factors may have a significant influence on the prevalence of stereotypies, especially in zoo-housed species. A meta-analysis by Mason & Latham

(2004) demonstrated that environments that induce or increase stereotypies are likely to be associated with other measures of poor welfare. However, the relationship between stereotypic behavior and welfare is not always so straightforward. The same meta- analysis revealed that stereotypies can occur in environments that appear neutral, even beneficial, in regards to welfare, and some highly aversive situations do not elicit any stereotypic behavior (Mason & Latham, 2004). Thus, stereotypic behavior should never

8 be used as a sole indicator of welfare state, and understanding underlying motivations is likely more useful for addressing welfare concerns than simply preventing stereotypies.

It is important to consider an animal’s full behavioral repertoire when assessing welfare (Hill & Broom, 2009). Animals vary in their behavioral responses to stimuli, and thus do not have a “universal indicator” of poor welfare (Dawkins, 2001). Comparing the behavior and activity budgets of zoo-housed animals to that of wild conspecifics is one method of determining what is “normal” for the species, and what factors could indicate compromised welfare. However, the needs and conditions of wild and captive animals are quite different and may not serve as a valid comparison (Dawkins, 1988; Veasey, Waran,

& Young, 1996). Further, many animals, especially those living in the wild, have developed mechanisms to hide any overt signs of welfare problems to avoid predation or successfully compete with conspecifics (Hill & Broom, 2009), which could prove problematic for accurately using behavior to indicate welfare state. It has also been suggested that behavioral diversity, or the richness and frequency of observed behaviors, may be a positive indicator of welfare ( Clark & Melfi, 2012; Miller, Pisacane, & Vicino,

2016; Vickery & Mason, 2004). The theory behind this measure is that animals housed in environments that are meeting all of their needs would be motivated to perform a full and diverse behavioral repertoire (high behavioral diversity), while animals with low behavioral diversity are “likely stereotyping or completely lethargic”, indicating poor welfare (Miller et al., 2016). There is still a need to further validate this measure, but it could prove useful, especially in species with a wide range of behaviors, such as primates or marine mammals. However, careful consideration of study design would likely be necessary for species that naturally spend a large proportion of their time engaged in a

9 few behaviors, such as felids, reptiles, and some hoofstock. Further, animals housed in a social group have more opportunity to engage in a wide variety of behaviors compared to animals housed alone. Clearly, this and any other behavioral measure of welfare needs to be validated and interpreted carefully, ideally in the context of other welfare indicators

(Hill & Broom, 2009). But with careful validation and interpretation, behavior can serve as a simple, yet robust tool for assessing welfare in zoo-housed animals.

Given the limitations of relating animal behavior to welfare, it is often helpful to include physiological measures as another means of assessing an individual’s welfare state. One method is to use glucocorticoids (GC) and their metabolites to measure activity of the hypothalamic-pituitary-adrenal (HPA) axis, also known as the stress pathway. In this context, “stress” refers to any stimulus that threatens or appears to threaten the homeostasis of an individual (Moberg & Mench, 2000). This physiologic stress response allows the individual to cope with unpredictable events and prepare for a rapid response if necessary. Thus, stress is a normal, adaptive part of life and is not inherently bad, as the everyday usage of the word would suggest. When a stressor is detected, the hypothalamus releases corticotropin-releasing factor which stimulates production of adrenocorticotropic hormone (ACTH) in the pituitary gland that then activates the adrenal glands. The adrenal cortex secretes GC hormones, typically cortisol or corticosterone, which serve to release stored energy and prepare the body for a “flight or fight response.” Metabolites of these hormones are later excreted from the body through urine and feces, allowing for non-invasive measurement of HPA axis activity (Möstl &

Palme, 2002). Although short-term increases in GC are an adaptive coping response, repeated stimulation of the HPA axis without time for rest or recovery can lead to muscle

10 wasting, gastrointestinal dysfunction, reduction in growth and brain function, decreased immune response, and poor reproductive success (Brann & Mahesh, 1991; Moberg &

Mench, 2000; Sapolsky, 1994). Thus, GC concentrations as a measure of the physiological stress response can be an indicator of both an individual’s acute, and long- term welfare state.

Unfortunately, using HPA axis activity as an indicator of welfare is not without challenges (Barnett & Hemsworth, 1990; Rushen, 1991). A major drawback to the use of

HPA axis activation as a measure of welfare is that changes in GC concentrations occur in response to situations that can be either aversive or pleasurable for the animal

(Colborn, Thompson, Roth, Capehart, & White, 1991; Dawkins, 2001; Wiepkema &

Koolhaas, 1993). Glucocorticoid concentrations can also be affected by circadian rhythms, seasonality, diet, age, sex, and social status (Millspaugh & Washburn, 2004;

Touma & Palme, 2005), making careful validation necessary in every species studied.

Therefore, elevated GC concentrations do not automatically indicate a state of distress or compromised welfare, and must be interpreted with caution. Despite these limitations,

GC monitoring is useful in establishing baseline stress levels and identifying factors that lead to chronic, detrimental stress, especially in the poorly understood and diverse species that are housed in zoos. With careful validation and interpretation, longitudinal and cross- institutional GC monitoring of zoo-housed species can become a powerful tool for objectively assessing the affective experience of individual animals in a variety of environments and housing conditions (Shepherdson, Carlstead, & Wielebnowski, 2004).

The single pursuit of any one of these methods will not necessarily improve welfare as judged by the others (Fraser, 2009; Mason & Mendl, 1993; Möstl & Palme,

2002). Different measurement techniques may give conflicting answers, so the use of multiple measures is overwhelmingly preferred (Broom & Johnson, 1993; Dawkins,

2001; Wielebnowski, 2003). Understanding GC response in the context of behavior and vice versa provides a more accurate picture of an animal’s true welfare state. Further, there has recently been a shift toward identifying and utilizing behavioral and physiological indicators of welfare on the positive end of the scale instead of consistently focusing on the indicators of compromised welfare (Boissy et al., 2007; Yeates & Main,

2008). Capitalizing on this new area of research to maximize positive indicators of welfare while simultaneously minimizing negative indicators offers the best chance of ensuring optimal well-being in zoo-housed animals.

Ambassador Animal Education Programs

In much the same way zoos as whole have undergone a shift in the last 30 years, zoo education departments themselves have also evolved – from the formal education of school groups in classroom settings to involvement in exhibit design, interpretive programs, and theatrical shows that serve to facilitate informal education (Andersen,

2003; Packer & Ballantyne, 2010). For a majority of the world’s human population, zoos and aquariums are the only places where people can experience diverse live animals

(Grajal et al., 2017). Zoos and aquariums can capitalize on this large audience in a time when environmental issues are more important than ever. It is no longer enough for zoos to simply impart knowledge of animals and their environments onto visitors; the ultimate goal is now to inspire behavior change that will have far-reaching impacts.

How can zoos accomplish such a lofty goal? To promote learning that leads to behavioral change in this informal setting, educators need to provide both a cognitive and an affective connection to the material presented (Ballantyne & Packer, 2005; Ballantyne,

Packer, Hughes, & Dierking, 2007; MacMillen, 1994). Cognitive learning involves the basic information retention that is common in formal education situations. A recent, large-scale, international study has shown that biodiversity knowledge increases after a visit to a zoo or aquarium along with knowledge of actions that can be taken to protect biodiversity (Moss, Jensen, & Gusset, 2015), so zoos in general seem to be accomplishing this part of their educational goals. Affective learning, however, targets the emotions of the learner to change attitudes, values, and behaviors, and can be more difficult to achieve (Pooley & O’Connor, 2000). Affective engagement helps to foster a personal connection to wildlife and the environment which can be an important aspect of environmental education (Ballantyne & Packer, 2005; Ballantyne et al., 2007). Packer &

Ballantyne (2010) found that after visiting a wildlife encounter (such as a zoo), visitors that reported an emotional connection to the animals retained information about threats to wildlife more than anything else, which lead to more feelings of personal responsibility and direct conservation action. Similarly, positive affect has been shown to be a mediator between observable animal behavior and zoo visitor meaning-making – active animals evoked positive emotions, which facilitated a memorable and meaningful experience for the visitor (Luebke, Watters, Packer, Miller, & Powell, 2016). Grajal et al. (2017) echoed this relationship in their finding that zoo visitors who reported a stronger emotional connection to animals were also more likely to report engaging in pro-environmental behaviors (relating to climate change in this study). However, the authors emphasized

13 that this relationship is complex, and mediated by concern about climate change and perceived effectiveness of personal actions to address climate change. This complex relationship highlights the importance of incorporating both cognitive (knowledge-based) and affective (emotion-based) learning objectives when designing zoo education programs.

Impact of Animals on Visitors

Using animals in education programs has been shown to be an effective means of engaging audiences and nurturing personal connections to wildlife (Ballantyne et al.,

2007; Heinrich & Birney, 1992; Povey & Rios, 2002). According to the AZA, ambassador animals include any individual “whose role includes handling and/or training by staff or volunteers for interaction with the public and in support of institutional education and conservation goals” (AZA Ambassador Animal Policy, 2011). These live animal education programs occur in various forms, but they all appear to have a beneficial effect on visitor learning, more so than that of programs that only feature animal specimens or “BioFacts” (Sherwood et al., 1989). Theater-inspired education shows featuring live animal actors are hugely popular in zoos around the world

(Andersen, 2003). Though these shows are not designed to facilitate up-close animal experiences, they nonetheless are effective at producing cognitive and affective changes in visitors that can lead to pro-environmental behaviors (Davison, McMahon, Skinner,

Horton, & Parks, 1993; Yerke & Burns, 1991). For example, zoo visitors that attended an animal show not only showed increased knowledge of the messages presented in the show, but retention of that knowledge remained high six weeks later (Heinrich & Birney,

1992). Visitors of a different zoo who viewed an elephant show and accompanying

BioFact program were more likely to take action in favor of elephant conservation

(Swanagan, 2000). Zoo guests that attended a dolphin show or participated in a dolphin interaction program demonstrated immediate increases in knowledge, attitudes toward marine conservation, and intentions to engage in conservation-related behaviors. Three months later, this knowledge was retained, and participants reported engaging in more pro-environmental behaviors (Miller et al., 2013). It also seems as though zoo visitors have begun to embrace the educational aspects of animal shows. Analysis of two types of dolphin shows at a zoo in South Africa reveled that visitors generally preferred the educational show that included strong conservation messaging and personal introductions of the animals and trainers over a more theatrical, story-based show (Mann-Lang,

Ballantyne, & Packer, 2016). The authors suggested that zoo visitors are now expecting to be educated, rather than just be entertained, and the personal introduction of trainers and animals fosters a connection that inspires guests to learn more and to care more about the environment.

Close encounter and outreach programs provide the most direct contact with education animals, and can also be highly effective in changing visitor attitudes.

MacMillen (1994) found that when ZooMobile outreach programs focused an information-only curriculum, a high degree of short and long term cognitive learning is achieved, but there was no evidence of affective learning, and therefore no attitude change. But when this type of classroom program supplements traditional animal facts with affective statements about personal responsibility to the environment, both attitude and behavioral changes are observed (Yerke & Burns, 1993). Elementary school students

15 that were given the opportunity to touch either live or preserved crab and sea star specimens during an education program demonstrated differential learning effects.

Although short and long term cognitive learning occurred in both conditions, affective learning only occurred in students exposed to live animals (Sherwood et al., 1989).

Programs involving direct contact with snakes were more effective at changing visitor attitudes toward them than programs that presented information alone (Morgan &

Gramann, 1989). Similar effects have been seen in interpretive programs featuring radiated tortoises and ravens. Close encounters with these animals increased audience engagement, cognitive and affective learning, and positive attitudes toward zoo animals compared to interpretive programs conducted in front of their usual exhibit (Povey,

2002). Visitors who attended a live animal interpretive program featuring a clouded leopard demonstrated longer viewing times, more information seeking behavior, and more positive feelings about the animal’s quality of life compared to visitors who attended a similar interpretive program featuring the same animal in an exhibit (Povey &

Rios, 2002). Visitors frequently commented that the handled leopard appeared to be mentally and physically stimulated, valued emotionally by the zoo, and have its needs met, while the exhibit leopard was perceived to be bored, confined, and lacking in exercise and opportunities to roam. Although this may be the perception of some zoo guests, there is still considerable debate about the ethics and welfare implications of live animal education programs (Ballantyne et al., 2007; Brando & Buchanan-Smith, 2017;

Yerke & Burns, 1991).

Impact of Visitors on Animals

Although the effect of animals on zoo visitors has been increasingly well documented, the inverse relationship is also worthy of attention. What is the influence of zoo guests on animal welfare? Hosey (2000) proposed that visitors can have three potential effects on the welfare of zoo animals: a relatively neutral effect, an enriching effect, or a stressful effect. There are a few published studies that appear to demonstrate the neutral effect, where visitor number or intensity was not associated with changes in behavior (Choo, Todd, & Li, 2011; Sherwen, Magrath, Butler, Phillips, & Hemsworth,

2014), and a few others that discuss visitor presence as a potential form of enrichment

(Claxton, 2011; Nimon & Dalziel, 1992). However, the majority of published visitor effect studies support the stressful hypothesis. Primates of many species have been shown to exhibit increased agnostic and stereotypic behavior and decreased affiliative behavior in the presence of visitors (Hosey, 2000; Mallapur, Sinha, & Waran, 2005; Sherwen,

Harvey, et al., 2015; Wells, 2005). Similarly, glucocorticoid concentrations were positively correlated with crowd size in capuchins (Sherwen, Harvey, et al., 2015), spider monkeys (Davis, Schaffner, & Smith, 2005), and black rhino (Carlstead & Brown, 2005).

Fernandez, Tamborski, Pickens, & Timberlake (2009) further reviewed visitor effect studies in a variety of species and also concluded that the majority of visitor interactions have a stressful effect on zoo animals.

However, there were considerable inconsistencies among the behavioral responses measured in the reviewed studies, suggesting that there may be differential effects among species, audiences, housing conditions, groups, and individuals (Hosey,

2008; Kuhar, 2008; Stoinski, Jaicks, & Drayton, 2012). Hosey (2000) further noted that

17 all of the studies examined demonstrated an association between zoo animal behavior and the presence of visitors, but none demonstrated an unequivocal direction of causality, a sentiment later echoed by Margulis, Hoyos, & Anderson (2003). The final drawback of these reviews is that a majority of studies focused on primate species, which make up only a portion of zoo populations and could arguably be more sensitive to human interactions (Fernandez et al., 2009; Hosey, 2008). Recently, the number of visitor effect studies focusing on non-primate species has grown (koalas: Larsen, Sherwen, & Rault,

2014; kangaroos: Sherwen, Hemsworth, Butler, Fanson, & Magrath, 2015; penguins:

Sherwen, Magrath, Butler, & Hemsworth, 2015). Although some of these have noted species (felids: Suárez, Recuerda, & Arias-de-Reyna, 2017) and individual differences

(jaguars: Sellinger & Ha, 2005) in response to zoo visitors, all have concluded that visitors can have potentially negative effects on animal welfare.

It is likely that the relationship between zoo visitors and animal welfare is much more complex than originally thought. Hosey refined his theory in 2008, developing a model of Human- Animal Relationships (HAR) in zoo-housed animals. The model states that zoo animal reactions to unfamiliar people (such as visitors) can be predicted by the types of interactions they have had with both familiar (keepers) and unfamiliar humans and there likely isn’t a universal “visitor effect” (Hosey, 2008). A history of positive interactions with both familiar and unfamiliar people predicts that visitors would have an enriching effect on zoo animals. Likewise, a history of negative interactions predicts that visitors are more likely to elicit a stressful response in zoo animals. These are the extremes on the scale of possible visitor effects, and various audience and environmental factors can contribute to more moderate responses. Because there are so many extraneous

18 factors that affect zoo animal responses to visitors, Fernandez et al. (2009) recommends focused research on HAR in zoos to determine which species (or individuals) tend to benefit from visitor interactions, and which ones experience welfare deficits as a result of visitor interaction. This approach may help to clear up the inconsistencies in the current literature and be more in line with the notion that welfare is an individual experience.

The aforementioned studies all examined the effects of passive visitors on exhibited animals. There is very little information on how ambassador animals respond to zoo visitors. Although exhibit animals are simply being viewed in their regular environment, ambassador animals are often handled, transported, and trained while in the presence of unfamiliar people, and sometimes asked to perform in theatrical shows or interact with zoo visitors directly. The effects of this very different husbandry regime have not been examined thoroughly. Anderson, Benne, Bloomsmith, & Maple (2002) analyzed the behavior of sheep and goats in a petting zoo situation and found that undesirable behaviors decreased when the animals were given access to a retreat space, indicating that visitors are potentially having a negative effect on the welfare of these animals, but the effect can be mitigated by providing opportunities for escape. However, a more recent study in petting zoo animals concluded that visitors had relatively little effect, positive or negative, on the behavior of goats, llama, and pigs (Farrand, Hosey, &

Buchanan-Smith, 2014). Salivary cortisol was measured in camels that provide rides to zoo guests, and no differences were found between days where the animals provided no rides, a low number of rides (~50), and a high number of rides (~150), indicating a potentially neutral visitor effect (Majchrzak, Mastromonaco, Korver, & Burness, 2015).

However, the influence of visitors on the welfare of domesticated animals such as these

19 may be quite different from that of the non-domestic species often used in zoo education programs. Miller, Mellen, Greer, & Kuczaj (2011) examined the effects of educational shows and interaction programs on bottlenose dolphin behavior and determined that these programs had an enriching effect on the animals. Affiliative, aggressive, repetitive, and social behaviors were unrelated to shows, and dolphins displayed increased behavioral diversity and play behavior immediately following performances. Although dolphin shows are popular worldwide, this highly social, intelligent, and aquatic species is also not representative of animals typically found in North American zoo education programs.

Small, solitary, easily handled species are much more common and there is a need for investigation into the effects of education programs on these species.

Cheetahs

Although nearly every zoo that has an ambassador animal program employs small mammals, birds, and reptiles, there is an increasing trend for utilizing larger and more dynamic animals for educational purposes. A few North American institutions have distinguished themselves by incorporating a large, iconic carnivore into their ambassador programs. The cheetah (Acinonyx jubatus) is an iconic zoo species that serves in multiple roles across AZA zoos – as exhibit animals, in off-exhibit breeding facilities, and as ambassador animals. However, the current North American population is not sustainable, and this species tends to experience unique reproductive and health challenges in human care. Although a large amount of resources and research attention have been devoted to understanding the causes of and solutions to these challenges, no investigations to date have included ambassador cheetahs.

Natural History

Cheetahs are well known as the fastest land animal, having evolved a suite of physiological and morphological adaptations that allow them to reach speeds up to 70 miles per hour (Hildebrand, 1961). Cheetahs also have a unique social structure and mating system among felids. Males tend to defend small territories and live solitarily or in groups of 2-4, termed coalitions (Caro, 1994). Females, however, travel alone or with their dependent cubs in large ranges that encompass the territories of several males (Caro,

1994). Recent genetic analysis has revealed evidence for multiple paternity in up to 40% of litters, and it is hypothesized that females may gain fitness benefits from mating with multiple males (Gottelli, Wang, Bashir, & Durant, 2007). Litters generally consist of 2-6 cubs and offspring remain with their mothers for nearly two years before becoming independent. Females will disperse to establish separate, but overlapping ranges close to their natal range. Male siblings will remain together as a coalition for life, while single males of a litter will remain solitary or join other males to form new coalitions (Caro,

1994). The unique social flexibility and mating system of the cheetah presents both benefits and challenges to captive management and conservation.

Conservation Status

Cheetahs have been closely associated with humans for millennia, as evidenced by the depiction of this distinct cat in multiple ancient cultures throughout Africa, Asia, and Europe (Pang, Van Valkenburgh, Kitchell, Jr., Dickman, & Marker, 2017). Although cheetahs were widely popular as hunting companions, pets, and status symbols, reports of any captive breeding are essentially non-existent, suggesting these animals were likely

21 taken from the wild as cubs (Marker & O’Brien, 1989; Pang et al., 2017). This demand for cubs coupled with growing human population and increased persecution put considerable pressure on wild cheetah populations, eventually leading to complete extinction in India (Marker, Grisham, & Brewer, 2017). Although cheetah populations have likely been in decline much longer, the need for earnest conservation action wasn’t formally acknowledged until the 1970’s. At that time, the Convention on International

Trade in Endangered Species (CITES) ended the export of wild cheetahs and captive breeding/rehabilitation centers such as the De Wildt Cheetah and Wildlife Centre began to appear in range countries (Marker et al., 2017). Since then, organizations such as

Cheetah Conservation Fund (Namibia), Action for Cheetahs (Kenya) and Cheetah

Outreach (South Africa), have been established with the goal of understanding threats to cheetah populations and identifying strategies to preserve the species. Primary threats to wild populations center around human-wildlife conflict, habitat loss and fragmentation, decline in prey species populations, and illegal wildlife trade, both for parts and live animals (Marker et al., 2017). The most recent estimates of cheetah populations in Africa indicate approximately 7100 individuals occupying 9% of their historical range (Durant et al., 2017). Most concerning is the estimate that 77% of these animals are living outside of protected areas, where threats to extinction are markedly higher (Durant et al., 2017).

Current strategies to mitigate threats to cheetah populations focus on reducing human- wildlife conflict (through the use of livestock guarding dogs and development of alternative livelihoods), establishing larger protected areas with increased enforcement, and, most critically, developing wide-reaching education programs that raise awareness

22 of issues affecting cheetah populations and foster empathy and human-cheetah coexistence (Marker et al., 2017).

Cheetahs in Human Care

Cheetahs in human care are known for experiencing unique health issues and poor reproductive success, and there is a large body of literature devoted to understanding the mechanisms behind these issues and developing strategies to overcome them. Cheetahs are known to frequently develop severe systemic diseases that can affect multiple organ systems including veno-occlusive disease (liver), glomerulosclerosis (kidney), chronic gastritis (GI tract), and feline infectious peritonitis (immune disorder; Munson, 1993). It was originally thought that this disease susceptibility was due to low genetic diversity in the species, a result of a genetic bottleneck 10,000 years ago (O’Brien et al., 1985;

O’Brien, Wildt, Goldman, Merril, & Bush, 1983). However, comparative analysis revealed that these disease processes are only prevalent in captive populations, and are rare or non-clinical in wild populations (Munson et al., 1999, 2005), suggesting an environmental, rather than a genetic, etiology.

In addition to these health concerns, cheetahs have been historically difficult to breed in human care (Marker & O’Brien, 1989; Marker-Kraus & Grisham, 1993). Large- scale reproductive assessments of the species have been undertaken to determine potential causes of infertility, and we now have a better understanding of reproductive biology in both sexes. Males generally have low ejaculate quality and volume as well as a large proportion of structurally abnormal sperm, but are still able to produce pregnancies

(Koester et al., 2015; Lindburg, Durrant, Millard, & Oosterhuis, 1993; Wildt et al., 1993).

Analyses of females have shown a large proportion of individuals do not cycle, and it has been determined that infertility in females is likely due to gonadal inactivity (Munson,

1993; Wildt et al., 1993). The question, then, is what leads to this ovarian inactivity? The aforementioned health diseases don’t appear to be causing infertility (Munson, 1993) and these reproductive issues don’t appear to occur in free ranging populations (Munson et al., 1999). Brown et al. (1996) measured estradiol in 26 female cheetahs for more than a year and found that all individuals experienced periods of anestrus that were sometimes several months in duration. These periods of acyclicity were not related to season and were not synchronous among individuals. One hypothesis is that environmental factors may be leading to reproductive inactivity. Unrelated females housed in close proximity to each other have suppressed ovarian activity, but regular cycling resumes when animals are separated (Wielebnowski, Ziegler, Wildt, Lukas, & Brown, 2002). This finding highlights the importance of mimicking natural social structure in captive environments, especially in cheetahs.

It has also been suggested that many of the issues observed in captive cheetahs are related to the stress response. Cheetahs in human care have lower testosterone concentrations, higher GC concentrations, and larger adrenal cortices (indicating a chronic stress response) compared to free-ranging individuals (Terio, Marker, & Munson,

2004). In addition, Jurke, Czekala, Lindburg, & Millard (1997) measured GC concentrations and ovarian activity and found that females with the highest GC concentrations also did not cycle, while females that were proven breeders fell into the low and medium GC category, providing further support for the idea of stress-mediated reproductive suppression. However, recent investigations have not found any association

24 between GC concentrations and reproductive activity (Koester et al., 2015; Koester,

Wildt, Brown, Meeks, & Crosier, 2017). The volume of research on zoo-housed cheetahs has led to an increase in our understanding of the species, as well as improvements in management practices, reproductive success, and welfare. However, clearly there are still unanswered questions and conflicting information, so cheetahs remain an attractive avenue for investigation.

Management Roles

In addition to the unique health, reproductive, and welfare challenges associated with the species, cheetahs are also managed in various roles across AZA institutions.

Cheetahs are one of the most popular exhibit animals in zoos and help to communicate conservation and educational messages about the plight of wild populations. However, there is some evidence that on-exhibit housing may be associated with stress and reproductive issues in some individuals. Wells, Terio, Ziccardi, & Munson (2004) studied

GC in cheetahs transferred between on- and off-exhibit housing and found that GC concentrations increased more and remained elevated for a longer period of time (up to

30 days) in animals moved from off- to on-exhibit enclosures. Though the authors emphasized individual differences, they found that cheetahs transferring to an exhibit had an 18 times greater risk of a prolonged physiological stress response than animals transferring off-exhibit. Though no associations with GC were observed, Koester et al.,

(2015) found that male cheetahs housed on exhibit produce less sperm. However, there were no differences in ovarian function of females housed on- vs. off-exhibit (Koester et

25 al., 2017). If on-exhibit housing does have a negative effect on welfare in cheetahs, it is likely not the case for all individuals and situations.

In an attempt to increase reproductive success and combat the numerous known issues in the species, cheetahs with breeding recommendations are housed in facilities specifically devoted to this goal. Breeding centers are typically not open to the public and can house multiple cheetahs in individual spacious enclosures. Females are housed alone or with their cubs, and males are housed alone or in coalitions away from the female enclosures. Animal managers utilize behavioral and hormonal signals to determine the best time for pairing males and females and monitor interactions closely. This design has proven successful, with > 90% of cheetah litters produced since 2003 occurring in these off-exhibit facilities (Koester et al., 2015).

Cheetahs are also an increasingly popular ambassador animal due to their

“nonaggressive temperament and ease of trainability and handling when raised by experienced handlers” (Rapp et al., 2017). In the United States, this practice began in

1976, when Laurie Marker, then curator at Wildlife Safari in Oregon, hand reared a female cub that later became the world’s first ambassador cheetah, travelling throughout the United States raising awareness of the threats faced by wild cheetahs (Rapp et al.,

2017). Additional ambassadors raised at different zoological institutions followed, and there are currently nine AZA facilities with ambassador cheetah programs. Ambassador cheetahs are hand reared (usually beginning before 3 weeks of age), so that habituation to humans and imprinting on caregivers can begin as early as possible (Rapp et al., 2017).

Throughout the labor-intensive hand rearing process, ambassador cheetahs receive desensitization and socialization training as they travel everywhere with their caregivers

(Rapp et al., 2017). It is critical that ambassador cheetahs become comfortable and habituated with as many unfamiliar situations and stimuli as possible to ensure success in this role. Education programs often require travel to novel locations, exposure to unpredictable audiences, and close handling and interaction with trainers, so this intensive training process is important for ensuring calm, safe animals during programs.

Ambassador cheetahs are also excluded from the breeding population, as their rearing history and the time and human labor invested into their training often makes it impractical to transfer these individuals to a breeding center situation. There are currently

59 ambassador cheetahs (16%) in the North American population that are excluded from breeding (Crosier, Lombardi, Maloney, & Andrews, 2018), which may be further contributing to population sustainability challenges. Furthermore, none of the other previous investigations of cheetah health, reproduction, or welfare have included ambassador animals. Given the previous work and the questions that remain about the relationships between human exposure, stress, and welfare in this species, there is a need for a large-scale investigation that compares all of these disparate management roles, including ambassadors. Understanding the factors related to welfare in all of these situations will help to inform management recommendations that allow cheetahs to thrive in any of these roles.

Personality in Animals

Although anyone who has ever worked or lived closely with animals would not hesitate to assert that individual animals have different personalities, the scientific community has been hesitant to agree, for fear of anthropomorphism. However, there is a

27 growing body of literature suggesting that individual animal personalities exist, can be measured, and can be used to improve animal management and welfare (Gosling, 2008;

Watters & Powell, 2012). Research on zoo housed species alone has shown that individual personalities are associated with differences in behavior, FGM, and reproductive success (Carlstead, Mellen, & Kleiman, 1999; DeCaluwe, Wielebnowski,

Howard, Pelican, & Ottinger, 2013; Gartner & Powell, 2012; Kuhar, Stoinski, Lukas, &

Maple, 2006; Powell & Svoke, 2008; Razal, Pisacane, & Miller, 2016; Wielebnowski,

1999; Wielebnowski, Ziegler, et al., 2002), and therefore may be important to consider when investigating the relationship between management practices and welfare.

Defining Personality

As mentioned previously, several different definitions have historically been used to describe the term “welfare”. In animal personality, however, several different terms have been used to describe the same concept. Historically, the terms ‘personality’,

‘temperament’, ‘behavioral types or syndromes’, ‘individual differences’ and ‘coping styles’ have all been used to describe the same construct: individual behavioral differences that are relatively consistent across time and context (Gartner & Weiss,

2013a; Powell & Gartner, 2011). The choice of terminology is often specific to the field of study, with human psychologists using the label ‘personality’, and behavioral ecologists and other animal researchers tending toward ‘temperament’ or ‘behavioral syndrome’. Gosling (2008) suggests that this difference in terminology is partly due to a desire by animal researchers to avoid any anthropomorphic associations with the human term ‘personality’. Box (1999) argues that ‘personality’ should refer to a behavioral

28 phenotype that accounts for differences due to genetics and experience while

‘temperament’ should only refer to behavioral differences that have a genetic basis, or a behavioral genotype. Similarly, human developmental psychologists refer to

‘temperament’ as inherited behavioral tendencies that appear early and persist throughout life, serving as the foundations of ‘personality’ that develops through childhood experiences (McCrae et al., 2000). However, defining individual behavioral differences by the process through which they develop (i.e. nature vs. nurture) may prove problematic in the practice of observational animal behavior research.

A number of more recent articles have called for the universal use of the term

‘personality’ across all research discussing the construct of individual behavioral differences (Gartner & Weiss, 2013a; Gosling, 2008; Watters & Powell, 2012).

Weinstein, Capitanio, & Gosling (2008) present three very compelling reasons for only utilizing the term ‘personality’ across disciplines moving forward. First, restricting the term ‘personality’ to humans and ‘temperament’ to animals requires making assumptions about the appearance and development of traits that are likely incorrect. It is becoming apparent that individual differences in adult animal behavior are the result of both genetics and experience, as is the case in adult humans (Weinstein et al., 2008). Second, using the term ‘personality’ exclusively facilitates large-scale meta-analyses across multiple disciplines and comparative research between human and animal studies.

Finally, it is just plain confusing to have multiple terms to describe the same phenomenon; especially if there doesn’t seem to be a consistent conceptual reason for doing so. For these reasons, I will use the term ‘personality’ throughout this thesis when referring to individual behavioral differences.

Measuring Personality

Once a consistent terminology is agreed upon, the next challenge for researchers is how one goes about measuring personality in non-human animals. There are two preferred methods for personality assessment in zoos: coding animal behavior and rating general behavioral tendencies (Gosling, 2001; Watters & Powell, 2012), each with inherent advantages and disadvantages. The coding method is generally considered to be more objective, as it involves direct observation and recording of an animal’s behavior.

Often, behavior is rated when an animal is placed in a situation that encourages them to display their individual personalities, such as exposure to a novel object, new environment, or a mirror. Though this method has been used extensively in studies of animal personality in the field of psychology, it presents some challenges in an applied setting. First, it requires considerable expertise and investment of time from the observer in order to accurately code meaningful behaviors. In addition, these types of behavioral tests are designed to measure an individual’s behavioral response to a threatening situation, and so any aspects of personality that would not be expressed in these situations

(e.g. sociality, playfulness, attitude toward caretakers) cannot be measured (Powell &

Gartner, 2011). These types of behavioral tests can be useful, however, in helping to validate the other method of personality assessment: observer ratings.

In the rating method, observers that are familiar with the subject use a survey tool to quantify their impressions on a number of personality traits. This method has historically been criticized as being too subjective for proper scientific measurement, but several authors have argued that aggregating trait ratings from several independent observers is no more subjective than any other widely used instrument in psychology or

30 ethology (Gosling, 2001). The rating method has the advantage of requiring considerably less time investment from individual observer, which is ideal in a zoo setting. In addition, because the ratings are made by observers familiar with the subjects, traits can be assessed over a variety of contexts and situations to provide a more complete picture of the animal’s personality. The difficult task in this method is choosing and defining traits that can be clearly understood by raters and applied to animal subjects consistently

(Powell & Gartner, 2011). Many animal surveys have utilized human personality assessments as a basis for listing and defining traits (J. E. King & Figueredo, 1997), while others have created trait lists by asking observers to organically provide adjectives to describe the individuals in question (Stevenson-Hinde & Zunz, 1978). Once survey tools have been developed, it is critical to ensure consistent understanding of the traits and definitions across observers (Watters & Powell, 2012). Conducting reliability analyses can determine if individual raters tend to be inconsistent with other raters of the same subject across all traits (perhaps due to lack of experience with the subject) or if individual traits are inconsistently rated across all subjects (perhaps due to an unclear definition). If reliable, observer ratings of traits can be an effective means of quickly measuring personality in large samples of individuals across a variety of contexts in order to gain a full understanding of the range of traits that may be expressed in a species and how these traits influence the individual experience of welfare.

Previous multi-institutional zoo studies have set out to do just that. The Gorilla

Behavior Index (GBI; Gold & Maple, 1994) was developed to quantify individual differences in gorilla behavior and preferences so that these differences can be taken into account when making decisions related to social group housing, breeding

31 recommendations, and transfers between facilities. Four personality factors have been identified using the GBI, three of which roughly correlate to the Five Factor model of human personality proposed by (Gosling & John, 1999): Extroverted, Fearful, and

Understanding. In addition, a Dominance dimension was identified, which appears in several animal personality models, possibly due to the larger role dominance plays in animal society compared with human society (Gosling & John, 1999; J. E. King &

Figueredo, 1997; Kuhar et al., 2006). Relationships between personality factors and rates of social behaviors suggest that the GBI could be used to predict behavior patterns and form successful gorilla groups (Kuhar et al., 2006). There has also been some evidence to suggest a relationship between individual personality and reaction to zoo visitors, though the statistical significance was low (Stoinski et al., 2012). All of these investigations have stressed the importance of considering personality when assessing welfare and called for more research to further clarify the influence of individual personality on an animal’s response to aspects of the zoo environment.

Personality in Felids

The body of literature on animal personality is rapidly growing. However, an overwhelming majority of these studies focus on non-human primates and domestic species, such as dogs and cats. Though the potential benefits of incorporating personality information when making management decisions in zoos are great, there is a need to expand the scope of research to include more non-primate species. One taxon that has received some amount of attention in the personality literature is the felids. However, recent reviews have only identified a handful of studies focused on zoo-housed felid

32 species (Gartner & Weiss, 2013a; Tetley & O’Hara, 2012), and when including investigations published since these reviews, the total number of non-domestic felid studies is still less than a dozen. These studies also span a wide range of species, including clouded leopards (DeCaluwe et al., 2013; Wielebnowski, Fletchall, Carlstead,

Busso, & Brown, 2002), snow leopards (Gartner & Powell, 2012), and Bengal tigers

(Phillips & Peck, 2007). Cheetahs have been the most extensively studied, with four separate investigations of this species including some measure of personality (Baker &

Pullen, 2013; Chadwick, 2014; Razal et al., 2016; Wielebnowski, 1999).

Though these studies often used different methods of assessment, there appears to be some consistency in personality across the taxon. A recent comparative study simultaneously assessed personality in domestic cats, Scottish wildcats, snow leopards,

African lions, and clouded leopards and repeatedly identified factors that could be labelled as “Neuroticism”, “Impulsiveness”, and “Dominance” across species (Gartner,

Powell, & Weiss, 2014). Within cheetahs, a personality dimension comparable to

“Neuroticism” was identified by three of the four studies (Chadwick, 2014; Razal et al.,

2016; Wielebnowski, 1999), and an “Aggression” dimension was described by two of the four (Razal et al., 2016; Wielebnowski, 1999). Although variation in methodology and sampling has likely prevented the exact same personality structure from being identified in all species, there is enough similarity to warrant the development of an instrument similar to the GBI for felids.

Despite the consistency in personality structure across studies, the relationships between personality and indicators of welfare has been less clear. For example,

Wielebnowski (1999) used keeper surveys and a mirror-image stimulation to identify

33 three main personality components (“tense-fearful”, “excitable-vocal”, and “aggressive”) in 44 cheetahs and found that non-breeding cats scored significantly higher on the component “tense-fearful” than those that had successfully produced cubs. However,

Chadwick (2014) used a similar instrument to assess cheetahs in European zoos and found no differences between 10 successful and 14 unsuccessful breeding pairs on the component “fearful-insecure”. Razal et al. (2016) used the same personality survey to assess 17 cheetahs and identified five components (“insecure”, “aggressive”,

“interactive”, “active”, and “unsociable”), but only the component “unsociable” was related to breeding success, with successful breeders scoring higher. A separate study defined three personality dimensions in cheetahs: “dominance”, “sociability”, and

“keeper-directed sociability” (Baker & Pullen, 2013). However, these components were not directly comparable with those found in the previous works because a different survey instrument was utilized, and the authors did not measure reproductive success.

This research did conclude that the “dominance” dimension was significantly lower if keepers regularly entered the enclosure with the cheetah, suggesting that specific husbandry practices may also be related to personality differences in this species. It is unclear if these apparent discrepancies between studies are due to the sample of cheetahs assessed or to slight differences in survey development and analysis.

Clearly, more research is needed to fully understand the relationship between personality, welfare, and the zoo environment, but it is a promising avenue. One practical application would be the use of personality scores (openness or extroversion, for example) to select individuals that would be best suited for ambassador animal roles in education programs vs. individuals that may have improved welfare in a secluded, off-

34 exhibit area. A better grasp of personality in general has the potential to inform personalized management plans that can maximize population sustainability, institutional needs, and individual welfare simultaneously (Watters & Powell, 2012).

Overall Objectives

The overall objective of this thesis is to evaluate the well-being of ambassador animals with the goal of using the results to inform management that ensures these animals are able to experience optimal welfare while helping to advance the educational mission of zoos and aquariums. A comprehensive approach that includes multiple indicators of welfare, several taxa, and a large sample size that spans multiple institutions is necessary for drawing meaningful conclusions that can then be used to inform management recommendations. Because there is currently very little research on this topic, a first step is determining if there are any differences in indicators of welfare between ambassadors and animals in other management roles. The general hypothesis was that there would be differences in behavioral and physiological indicators of welfare between education, exhibit, and off-exhibit animals of all species. I predicted that ambassador animals would have slightly higher GC concentrations and demonstrate increased undesirable behavior compared to their exhibit and off-exhibit counterparts.

However, this project was largely exploratory in nature. Capitalizing on a large, multi-institutional study design, information about various housing, husbandry, and human interaction variables was collected from all animals sampled to achieve my second objective of determining what, if any, specific aspects of the zoo environment are associated with welfare in these species. With large sample sizes and the inherent

35 variability of a multi-institutional study, I was able to analyze these variables in association with behavioral and physiological indicators of welfare with the goal of informing management recommendations that can be applied across zoos.

Chapter Two compares behavior and fecal glucocorticoids in armadillos managed in education, exhibit, and off-exhibit roles. A second experiment uses these same indicators of welfare to evaluate the effect of handling specifically for education programs in armadillos, hedgehogs, and red-tailed hawks. Techniques for measuring fecal glucocorticoids in all three species were validated as part of this study. Both experiments also evaluated effects of housing variables (such as enclosure size and substrate depth) as well as the amount of handling each animal experienced during the study on these indicators of welfare.

In Chapter Three, I measured personality in cheetahs serving in multiple roles across AZA institutions. This investigation builds upon previous cheetah personality research with the inclusion of ambassador animals. The personality structure identified here is compared to these previous works, and individual factors that might predict particular personality types (such as sex, rearing history, and social grouping) are examined.

Chapter Four expands on the design of the armadillo study from Chapter Two, this time evaluating cheetahs in ambassador, exhibit, and off-exhibit roles. Again, various housing, husbandry, and human interaction variables were analyzed in relation to behavior and fecal glucocorticoids. This study also included the measure of personality discussed in Chapter Three, and the influence of individual differences on behavioral and physiological measures of welfare is discussed.

CHAPTER TWO: The influence of management role and handling on indicators of welfare in armadillos, hedgehogs, and red-tailed hawks.

Published in part in Applied Animal Behavior Science:

Baird, B. A., Kuhar, C. W., Lukas, K. E., Amendolagine, L. A., Fuller, G. A., Nemet, J., Willis, M. A., Schook, M. W. (2016). Program animal welfare: Using behavioral and physiological measures to assess the well-being of animals used for education programs in zoos. Applied Animal Behaviour Science, 176, 150–162.

Introduction

In the past several decades, zoos and aquariums have shifted from sources of entertainment to centers for conservation education (Andersen, 2003; Reade & Waran,

1996). Although zoos employ a variety of educational techniques to convey messages, one of the most popular is the use of live animals in interpretive programs (Andersen,

2003). The Association of Zoos and Aquariums (AZA) recognizes program, or ambassador, animals as “an important and powerful educational tool that provides a variety of benefits to zoo and aquarium educators” (AZA Program Animal Position

Statement, 2003). Indeed, several studies have demonstrated that using live animals not only increases knowledge retention, but can be effective in changing visitor attitudes about wildlife, conservation, and personal responsibility to the environment (Heinrich &

Birney, 1992; Miller et al., 2013; Morgan & Gramann, 1989; Povey, 2002; Sherwood et al., 1989; Yerke & Burns, 1991). Povey & Rios (2002) measured zoo visitor interest and empathy toward a single clouded leopard when participating in an interpretive program and while on exhibit. The live demonstration resulted in longer viewing times, more information seeking behavior, and more positive feelings about the animal’s quality of

37 life. Although this may be the perception of the average zoo visitor, debate continues regarding the ethical and welfare implications of using live animals in education programs (Ballantyne et al., 2007; Brando & Buchanan-Smith, 2017; Shani & Pizam,

2008).

There has been some investigation into the effects of zoo visitors on exhibited animals. Hosey (2000) defined three types of effects that visitors may have on zoo animals: a source of stress, a source of enrichment, or a relatively neutral effect. Though some studies have suggested that visitors might have an enriching effect (Miller et al.,

2011; Nimon & Dalziel, 1992), other studies show the opposite—that zoo visitors are a source of stress (reviewed in Davey, 2007; Fernandez, Tamborski, Pickens, &

Timberlake, 2009; Hosey, 2000). It is also important to note that group (Kuhar, 2008) and individual differences (Sellinger & Ha, 2005; Stoinski et al., 2012) in behavioral responses to visitor presence have been reported, suggesting there may not be a single, generalized “visitor effect”.

Most visitor effect studies have been restricted to animals (often primates) housed on exhibit and the effects of onlookers. Unlike exhibit animals, ambassador animals are frequently handled, transported, and come into close contact with humans. Thus, education animals could have a different response to human interaction than most exhibit animals. Several studies have demonstrated that frequent positive human interactions, including repeated handling, are associated with a reduction in the fear response of various species in their subsequent encounters with humans (sheep: Hargreaves &

Hutson, 1990; pigs: Hemsworth, Barnett, & Hansen, 1986; rabbits: Podberscek,

Blackshaw, & Beattie, 1991). However, these studies focused on domestic animals and

38 similar efforts in wild species have yielded mixed results (wombats: Hogan et al., 2011; koalas: Narayan, Webster, Nicolson, Mucci, & Hero, 2013). Further, there has been little investigation into the specific effects—positive or negative—of education program use on animal welfare.

The purpose of this study was to empirically evaluate ambassador animal welfare using established physiological [fecal glucocorticoid metabolite (FGM)] and behavioral measures of welfare. The use of multiple measures of welfare is generally preferred, as different measurement techniques can give conflicting results if considered independently

(Broom & Johnson, 1993; Hill & Broom, 2009; Wielebnowski, 2003). Because there has been little research on this topic, we wanted to evaluate potential effects of education program animal use in broad terms with the intention of guiding future, specific studies.

Therefore, the aim of Experiment One was to determine if there are any physiological or behavioral differences between animals managed for either education or exhibit.

Armadillos were chosen for Experiment One due to their common presence both in education programs and as exhibit animals in multiple institutions. For Experiment One, we hypothesized that behavioral and physiological measures of welfare would differ between education and exhibit animals. Specifically, we predicted that education animals would have higher FGM levels and demonstrate increased undesirable behavior compared to exhibit armadillos. Because handling is a central component of program animal use, the aim of Experiment Two was to examine the specific effect of handling on measures of welfare in animals used for education programs. Because we wanted to evaluate the effect of handling on the variety of taxa frequently utilized in education programs, we chose to include hedgehogs and red-tailed hawks in addition to armadillos

39 in Experiment Two. In this Experiment, we hypothesized that handling during programs would have no effect on the welfare of education animals, with behavior and FGM concentrations of education animals being similar during periods of regular program use and periods of no program use.

Methods

Experimental Design

All methods and animal use were reviewed and approved by each participating institution and the Animal Care and Use Committee at Cleveland Metroparks Zoo.

Experiment One

To compare physiological and behavioral measures of welfare, La Plata, or

Southern three-banded, (Tolypeutes matcus; n=51), screaming (Chaetophractus vellerosus; n=5), nine-banded (Dasypus novemcinctus; n=2), and six-banded (Euphractus sexcinctus; n=1) armadillos housed at 17 AZA accredited institutions were sampled for

60 days between April and September 2012. Individuals were classified as education, exhibit, or off-exhibit based on where the animal was housed and whether it participated in live animal education programs. Education animals were used in at least one educational program during the study. Animals that were not used for education programs, but housed on exhibit for public viewing were classified as exhibit. Armadillos that were neither used for education programs nor housed on exhibit were classified as off-exhibit; typically maintained in holding areas for breeding purposes with no exposure to the public.

Fecal samples were collected approximately every other day from 57 armadillos

(28 education, 17 exhibit, 12 off-exhibit; see “Fecal Collection and Hormone

Extraction”). Institutions that were able to commit time and resources also collected behavioral data from a subset of armadillos (7 education, 13 exhibit, 3 off-exhibit; see

“Behavioral Data Collection”). A minimum of ten morning (0700-1000) and ten afternoon (1500-1800) observations were collected for each animal. For two La Plata armadillos (1 education, 1 exhibit) from the same institution, behavioral data, but not fecal samples, were collected.

Experiment Two

To determine the specific effect of handling on education program animals, La

Plata armadillos (n=9), a screaming armadillo (n=1), African hedgehogs (Atelerix albiventris; n=12), and red-tailed hawks (Buteo jamaicensis, n=6) at 11 AZA institutions were sampled for 9 weeks between December 2011 and July 2012. All subjects were

“education animals”, defined as being used in live animal demonstrations with the public at least two times during the study. The study period was divided into three, 3-week

“phases”, with each animal used as its own control. In the “Baseline” phase, animals were handled as usual for education programs. In the subsequent “No Handle” phase, animals were not used in any programs or demonstrations. However, for some animals, handling for routine husbandry (picked up for cage cleaning, weighing, etc.) continued throughout this period. During the final “Post” phase, animals were returned to normal education program use. Fecal samples were collected approximately every other day from all subjects (see “Fecal Collection and Hormone Extraction”). Behavioral data were also

41 collected from 7 armadillos, 2 hedgehogs, and 6 hawks (see “Behavioral Data

Collection”). A minimum of five morning (0700-1000) and five afternoon (1500-1800) observations were conducted during each phase of the study.

Demographic, Husbandry, and Handling Data Collection

We chose to examine the potential influence of direct and controllable environmental and husbandry variables in this first investigation, detailed in Table 2.1

(Experiment One) and Table 2.2 (Experiment Two). Demographic and husbandry data included sex and age (if known), enclosure size, type and average depth of substrate provided, and whether the animal was housed under a reversed light cycle (dark during daytime hours). Institutions reported whether the animal is handled regularly, if the animal was handled during the first year of life, and the number of years participating in education programs (education animals only). Finally, each institution recorded information about every instance of handling that the animal experienced during the study period. The date, start and end times (or duration), type (hold, pet, transport, taken off grounds) and purpose (husbandry, education program, veterinary exam, etc.) of each handling event were recorded. Every instance of handling, regardless of type or duration, was counted as a single event and calculated as a weekly total. The duration of each event was used to calculate the total minutes of handling per week. If the duration information of a given event was missing (Experiment One: 25 events, Experiment Two: 9 events), the average duration of similarly labeled events for that individual was used for analyses.

Table 2.1: Husbandry and demographic factors for armadillos in Experiment One.

Factor FGM Analysis Behavior Analysis Mean (Range) Mean (Range) Age (y) 10.8 (1.2 – 33.2) 9.6 (1.2 – 33.2) Substrate depth (cm) 10.5 (0 – 30.5) 6.4 (0 – 27.9) Enclosure size (m2) 4.6 (0.4 – 169.5) 11.1 (0.4 – 169.5) Handling 4.7 (0 – 24) 4.6 (0 – 14) events/week Handling min/week 148.1 (0 – 830) 68.6 (0 – 830)

N N Management Role Education 28 7 Exhibit 17 13 Off Exhibit 12 3 Sex Male 31 11 Female 26 12 Reversed light cycle? Yes 7 1 No 50 22 Substrate Type Wood Shavings 20 10 Straw 15 6 Cardboard 8 5 Multiple Types 7 2 Mulch 3 - Dirt 2 - None 2 -

Table 2.2: Husbandry and demographic factors for species in Experiment Two.

Armadillos Hedgehogs Hawks

Factor Mean (Range) Mean (Range) Mean (Range) Age (y) 6.8 (1.3 – 18.1) 1.8 (0.7 – 3.1) 16.1 (6.2 – 25.8) Substrate depth (cm) 9.2 (2.5 – 22.9) 2.9 (0 – 7) 2.3 (0 – 3.8) Enclosure size (m2) 1.7 (0.4 – 4.5) 0.4 (0.3 – 0.6) 7.9 (3.5 – 11.1) Handling events/week 4.3 (0 – 11) 4.3 (0 – 14) 5.3 (0 – 15) Handling min/week 279.3 (0 – 1500) 145.7 (0 – 767) 139 (0 – 903)

N N N Sex Male 7 4 2 Female 3 8 4 Substrate Type Wood Shavings 4 3 - Hay 3 - - Shredded Paper 1 - - Paper Litter 1 6 - Multiple Types 1 0 0 Wood Chips - 2 - Cloth - 1 - Gravel - - 4 Dirt - - 1 Asphalt - - 1

Fecal Collection and Hormone Extraction For both Experiments, naturally voided feces (< 12 hours old) were opportunistically collected once during operating hours (validation testing indicated no diurnal variation in glucocorticoids) and placed in Whirlpak bags labeled with institution, animal ID and date. Samples were stored at -20oC immediately after collection until project completion and then shipped overnight on ice to Cleveland Metroparks Zoo where they remained frozen until extraction.

A wet fecal extraction method adapted from Brown (2008) was used for armadillos. In brief, 5 mL of 80% methanol: water was added to ~0.5 g wet feces in 20 mL glass tubes. For samples weighing <0.5 g, methanol was added in a 1 mL: 0.1 g feces ratio. Samples < 0.2 g were not used. Tubes were loaded onto a mixer (099A LC1012,

Glas-Col, Terre Haute, IN, USA) and agitated for 1 hour. Samples were then centrifuged

(2500 x g, 20 min), the supernatants collected into a new set of 5 mL plastic tubes and stored at -20oC until assayed.

Due to variability in fecal water content, hedgehog fecal samples were lyophilized

(FreeZone, Model: #7751020, Labconco Corporation, Kansas City, MO, USA) and then manually crushed prior to extraction using a technique adapted from Brown (2008).

Briefly, ~0.2 g fecal powder was placed in glass tubes with 5 mL 80% methanol. For samples weighing < 0.2 g, methanol was added in 1 mL: 0.04 g feces ratio and processed as described for armadillos.

A dry fecal extraction method adapted from Hayward, Booth, & Wasser (2010) was used for hawks. Samples were lyophilized and crushed, taking care to avoid dried urates in the sample. Approximately 0.05 g fecal powder was added to 7 mL 80% ethanol: water in glass tubes and agitated for 1 hour. Samples were then centrifuged

(3000 x g, 15 min), the supernatant transferred into clean glass tubes, and dried with air using an evaporac (Cole Parmer, #EW-01610-25, Vernon Hills, IL, USA). The dry fecal residue was re-suspended in 1 mL 80% ethanol, manually vortexed, sonicated (FS220,

Fisher Scientific, Hanover Park, IL, USA) to concentrate the hormone residues, and dried again. Dried samples were stored at -20oC until assayed. Before assaying, samples were

45 reconstituted in 1 mL provided assay buffer (#K014; Arbor Assays; Ann Arbor, MI,

USA). Samples from all species were assayed within three months of fecal extraction.

Hormone Analysis

Laboratory Validation

For all species and assays, laboratory validation was achieved by demonstrating a parallelism between binding of serial dilutions of fecal extracts and the standard curve.

Intra- and inter-assay coefficients of variation (CVs) were less than 11%. Recovery of fecal extracts spiked with known amounts of corticosterone standard averaged >84% and

FGM concentrations were calculated on an ng/g feces basis (wet feces in armadillos and dry feces in hedgehogs and hawks).

FGM concentrations in armadillos and hedgehogs were determined using a commercial corticosterone radioimmunoassay (RIA) 125I kit (#07-120103; MP

Biomedicals, Solon, OH, USA). Fecal extracts were diluted 1:25 in provided steroid diluent. FGM concentrations were measured using a solid phase 125I gamma counter

(Genii; Genesys, Maple Park, IL, USA). For hawks, FGM were measured using an enzyme immunoassay (EIA; #K014; Arbor Assays). Extracts were diluted at 1:8 in provided assay buffer. Samples that were less than 20% binding and higher than 80% binding were re-assayed at higher or lower dilutions (1:2 to 1:25).

Biological Assay Validation

An adrenocorticotrophic hormone (ACTH) challenge is one standard method for biologically validating methods for FGM measurement in a given species and assay

(Touma & Palme, 2005). Therefore, ACTH (Cortrosyn, Amphastar Pharmaceuticals, Inc.,

Rancho Cucamonga, CA, USA) was administered to one male screaming armadillo, one male and one female La Plata armadillo (all armadillos: 5 IU/kg, IM), and one female red-tailed hawk (25 IU/kg, IM). Fecal samples were collected daily from each individual for one week prior to injection to measure baseline FGM concentrations. For 2-4 days following injection, samples were collected as frequently as possible, and then daily for the following 10 days to observe a return to baseline FGM concentrations. One female hedgehog underwent a veterinary exam during the study period that served as an opportunistic biological validation.

Baseline FGM concentrations for each individual were calculated using an iterative process described in Brown, Schmitt, Bellem, Graham, & Lehnhardt (1999).

Briefly, FGM concentrations exceeding the mean +2SD were removed. Means and SDs were then recalculated, and the process repeated until no values exceeded the mean +

2SD. The mean of the final iteration served as the baseline FGM concentration. FGM were considered elevated if a sample exceeded the baseline + 2SD.

Behavioral Data Collection

Focal animal behavioral observations were conducted in 20-min sessions using an instantaneous point sampling technique at one-minute intervals (Martin, Bateson, &

Bateson, 1993). During each scan, the focal animal’s behavior was recorded using a mutually exclusive and exhaustive ethogram (Table 2.3A). A video ethogram and behavioral data collection training video was developed and mailed to each participating institution. Inter-observer reliability was tested using one of two videotaped 20-min

47 observations that potential observers scored. Data sheets from this reliability test were scored by the primary investigators (PIs). Observers achieved greater than 90% inter-rater reliability with the PIs to collect data for the study. For hawks, the “Rest” behavioral category in the ethogram (Table 2.3A) was divided into two categories: “Rest Alert” and

“Rest Passive” (Table 2.3B). This division allowed differentiation between actual resting and the common vigilance/hunting behavior observed in raptors. A second video ethogram was used for institutions observing hawk behavior.

Hedgehogs are primarily nocturnal animals, and preliminary observations confirmed that the hedgehogs were largely inactive during daylight hours. To accurately sample the behavior of these animals, a video camera (DNV900HD, Bell & Howell,

Wheeling, IL, USA) with an infrared attachment (IR Illuminator, Sima, Hauppauge, NY,

USA) was set to record after business hours. The camera was placed in front of the hedgehog enclosure at the end of the work day so that recording began when the room lights were turned off for the night (between 1630 and 1800) and continued for approximately four hours. Recordings were made Monday-Friday for the entire study period. Digital video was reviewed and scored using the same instantaneous scan sampling technique and ethogram (Table 2.3A) described above. Because observations could not be balanced for time of day as in the other species, 20 minute sections of video were randomly selected for scoring from either the first two hours of the recording or the second two hours of recording, and observations were balanced across these two time periods. Data were collected from two female hedgehogs housed in the same enclosure.

The two individuals were distinguished on video by applying a 1 cm piece of white reflective tape to quills of one of the animals. Limitations on the availability of video

48 equipment for other institutions prevented collection of behavior data in additional animals.

Table 2.3: A) Ethogram used for behavioral data collection on armadillos, hedgehogs, and red- tailed hawks for both Experiment One and Experiment Two. B) Modifications to the “Rest” category used when collecting behavioral data in hawks.

Behavior Undesirable Behavior Pacing (must be observed retracing exact path at least three times (UND) for pacing to be scored – otherwise, score as LOCO), rocking back and forth (> 3 times), plucking hair/feathers, or picking skin. Feeding/Drinking Consuming, handling, or investigating diet items, browse, or (FEED) water. Object Examination Touching, sniffing, or otherwise actively investigating a non-food (OBJX) item in its enclosure, including arm bands or jesses attached to the animal. Includes digging through or shredding newspaper or other litter/hay/mulch. Self-Directed Behavior Scratching or licking itself, auto grooming, and/or rubbing body (SDB) part on object. Locomoting Propelling body in movement resulting in a transfer of physical (LOCO) position in space that is at least or more than its own body length in distance. If the subject is passing behind an object at the instant the timer beeps, but is clearly locomoting from one end of the object to the other, the behavior should be scored as locomoting, rather than not visible. Other Active Engaging in an active behavior not otherwise defined, includes (OACT) social behavior directed toward people or other animals such as beak clapping or vocalizing. Resting (REST) Passive and not engaged in any active behavior. Eyes may be open or closed. Includes sleeping. Not Visible (NV) The behavior of the subject is not visible or discernible. If the animal is situated in such a way (ex. facing away from you or nuzzled in a corner) that either its nose or one limb cannot be seen to discern if it is engaged in an activity the behavior should be scored as not visible.

Not Visible, Presumed The subject is not visible but is known to be in a resting spot such Resting as a nest box; there is no visible or auditory evidence of movement (NV-PR) and the subject is presumed to be resting.

B. Rest Alert Subject is staring intently or vigilantly with head tilting or movement; subject is engaged with surroundings.

Rest Passive Subject is staring with no head movement; eyes may be open or closed. Includes sleeping.

Statistical Analysis

FGM Analysis

For Experiment One, effects of management role (education, exhibit, or off- exhibit), age, sex, species, enclosure size, substrate type and depth, light cycle, regular handling, handling during the first year of life, years in education program, and frequency or duration of handling per week on FGM concentrations were analyzed as fixed factors in a random intercept and slope (using week of study) general linear mixed model (PROC

MIXED; SAS version 9.3; SAS Institute, Cary, NC, USA). FGM concentrations were log transformed to approximate a normal distribution. Random factors (individual animal nested within institution) were added to the model to control for similarity between samples from the same individual or institution. An unstructured covariance structure was chosen based on Akaike’s information criterion (AIC) and Bayesian information criterion

(BIC), which are values indicating model fit (lower value is better). Model building was conducted using maximum likelihood approximation and by removing non-significant factors until the lowest -2 log likelihood (-2LL), AIC, and BIC were achieved. Fixed factors were removed at p > 0.10 unless contributing to a significantly better model fit.

Final models were run using restricted maximum likelihood estimation. Degrees of

50 freedom (df) were calculated using Satterthwaite’s approximation. The management role variable remained in the model at all times because it related to our primary question.

Cook’s and Restricted Likelihood Distance tests were performed to identify potential outliers. Based on these results, five data points from a single animal (of 1686 data points) were removed to obtain the final model for Experiment One. In addition, two armadillos were housed in a very large enclosure (169.5 m2) for a small portion of the study (10 days each). As the next largest enclosure size in the study was 30.75 m2, these

12 data points were considered outliers and removed from analysis. However, any changes in results from this action are reported below. Residuals and random intercept estimates were assessed for normality in each final model. When significant effects (p <

0.05) were determined, post hoc tests for multiple pair-wise comparisons were conducted using the Tukey-Kramer adjustment.

In Experiment Two, FGM were analyzed as above with the following exceptions.

Armadillos, hedgehogs and hawks were analyzed in separate models. To achieve model convergence for armadillos, the model contained a random slope, but not a random intercept. For hedgehogs, the model contained both a random intercept and slope, while in hawks the model contained no random statement. The factor of phase remained in all models, because it was related to our primary hypothesis. Cook’s and Relative Likelihood

Distance tests indicated no significant outliers except seven data points from a single armadillo (of 1035 total). To test the effect of phase of study within individual, a general linear mixed model analysis was conducted using phase of study as the only predictor variable by individual.

Statistically analyzed hormone data are presented as back-transformed mean ±

SEM. For all analyses, p < 0.05 was considered statistically significant, and results are presented as mean ± SEM unless otherwise noted. Although not relevant to our hypotheses, effects of age and sex were tested for and when significant, controlled for in the model, though results are not presented.

Behavior Analysis

For all analyses, Rest and Not Visible, Presumed Rest (NV-PR) behavior categories were combined into a single Rest category. For all behaviors, the effects of management role or phase, age, sex, species, time of day (AM vs. PM), enclosure size, substrate type, substrate depth, reversed light cycle, regular handling, handling during the first year of life, years in education program, and frequency or duration of handling per week were analyzed using a generalized linear mixed model (PROC GLIMMIX; SAS

Institute) as described above with the following exceptions. A negative binomial distribution with a log link function was used for all behaviors except Rest because data were not normally distributed. For Rest, a logistic regression was employed using the scan data in an “events out of trials” format using a binomial distribution and a logit link function. In Experiment One, off-exhibit animals were not observed to perform undesirable behavior, and were excluded from the analysis of that behavior to achieve model convergence. SDB were rare in armadillos during the study (<1% of observations), preventing analysis with a mixed model. Therefore, the scan data were converted to presence/absence data (0= SDB was not exhibited during a 20-min observation, 1= SDB was exhibited) and analyzed using a logistic regression modeling the probability that an

52 animal would exhibit SDB (PROC LOGISTIC; SAS). Effects of age, sex, and time of observation (AM vs. PM) were tested for and when significant, they were controlled for, though specific results are not presented here.

Results

ACTH Challenge

Following the ACTH challenge, FGM concentrations increased by more than 2

SD above baseline concentrations in all armadillos (Fig 2.1A-C). Concentrations of FGM in the male La Plata armadillo increased by 171% compared to baseline within 25 hours of ACTH injection (baseline mean ± SD =17.16 ± 4.83 μg/g feces; Fig 2.1A), and returned to baseline levels ~74 hours after injection. In the female La Plata armadillo,

FGM concentrations were generally higher than those found in the male (baseline mean ±

SD =31.80 ± 17.97 μg/g feces; Fig 2.1B). Approximately 40 hours after ACTH injection,

FGM concentrations increased by 446% compared to baseline. Concentrations of FGM returned to pre-injection levels within ~88 hours. The male screaming armadillo demonstrated a 119% increase in FGM concentration compared to baseline levels

(baseline mean ± SD = 32.81 ± 10.14 μg/g feces; Figure 2.1C) ~39 hours post-injection.

Concentrations of FGM returned to pre-injection levels within ~114 hours after exposure to synthetic ACTH.

An ACTH challenge was also performed on a female red-tailed hawk. In this individual, FGM concentrations increased by >2000% compared to baseline levels

(baseline mean ± SD = 20.84 ± 4.41 μg/g feces; Fig 2.1D) ~19 hours after treatment.

Concentrations returned to pre-injection levels after ~86 hours. In the female African

53 hedgehog (Fig 2.1E), mean baseline FGM concentrations were 95.70 ± 43.84 μg/g feces.

Forty-six hours after receiving a veterinary exam, FGM levels increased by 182% compared to baseline. Hormone concentrations returned to pre-event levels after ~143 hours.

Experiment One

FGM Results

There was substantial inter-animal variation among the 58 armadillos with average FGM concentrations for each individual over the 60-day study period ranging from 10.96 ± 0.87 to 234.66 ± 30.84 μg/g feces. This inter-animal variability was independent of the management role of the individual; overall, there was no difference in the FGM concentrations of education, exhibit, and off-exhibit armadillos (F(2, 46) = 0.55, p

= 0.58; Fig 2.2). However, regardless of management role, armadillos that were handled more frequently had higher FGM concentrations (F(1, 979) = 9.35, p = 0.002). The type of substrate provided was associated with significant differences in FGM levels across all subjects (F(6, 47) = 3.10, p = 0.01). Animals that were provided with no substrate (n = 2) had significantly higher FGM concentrations than animals that were provided with any type of substrate (p < 0.05). Finally, enclosure size was negatively correlated with FGM

(F(1, 45) = 3.46, p = 0.05), with larger enclosures being associated with lower FGM.

Figure 2.1: FGM concentrations in response to an ACTH injection in A) a two year-old male La Plata armadillo, B) a nine year-old female La Plata armadillo, C) a twelve year-old male screaming armadillo, and D) a seven year-old female red-tailed hawk. FGM concentrations in response to a veterinary exam in a two year-old female African hedgehog (E). Solid horizontal lines represent mean baseline concentrations. Dashed horizontal lines represent baseline concentration ±2SD. Arrows denote time of injection. Note that Y axes are scaled to individual peak FGM concentrations.

The factors of age, sex, species, light cycle, regular handling, handling during the first year of life, and years of education program use were not significantly associated with FGM concentrations (p > 0.15). Specific types of handling (hold, pet, taken off grounds, education program, etc.) were also recorded for this study, and some did show significant relationships to FGM. However, since multiple types of handling often occurred simultaneously, it was not possible to tease apart the effects of each specific handling type with the current study design and this is an area that warrants further investigation. When the 12 data points (out of 1686 data points total) from the two animals who spent a short amount of time in a very large enclosure were included in the analysis (see Statistical Analysis), enclosure size was no longer a significant factor in the model (p = 0.07), but all other factors retained similar significance levels.

Figure 2.2: Box plots representing average corticosterone concentrations of education, exhibit, and off exhibit armadillos. There was no difference in FGM concentrations between the three groups (F test, p = 0.32).

Behavior Results

Overall, education animals spent less time resting (F(2, 17) = 7.93, p = 0.004; Fig

2.3A) and performed more Other Active (OACT) behavior (F(2, 20) = 4.30, p = 0.03; Fig

2.3B), than either exhibit or off-exhibit animals. Both education and off-exhibit animals demonstrated more Object Examination (OBJX) behavior than exhibit animals (F(2, 14) =

8.98, p = 0.003; Fig 2.3C). Locomotion also differed among groups (F(2, 633) = 3.07, p =

0.05; Fig 2.3D), with education armadillos locomoting more than off-exhibit (p < 0.05), but not more than exhibit animals. Finally, education animals were observed to engage in

2 SDB more than the other groups (Wald Χ (2) = 9.01, p = 0.01, Fig 2.3E). Based on odds ratios, education armadillos were 3.8 times more likely to exhibit SDB than exhibit (95%

CI: 1.77 – 8.18) and 22.8 times more likely than off-exhibit armadillos (95% CI: 2.91 –

179.01) in a given observation. Exhibit animals were also 6 times more likely to perform

SDB than off-exhibit animals (95% CI: 0.72 – 50.37). There was no significant difference between groups in the amount of undesirable behavior observed (F(1, 17) = 1.86, p = 0.19,

Fig 2.3F).

The same factors that were significantly associated with FGM concentrations in this experiment were also correlated with behavior. Across all management roles, as the number of handling events per week increased, undesirable behavior increased (F(1, 262) =

6.09, p = 0.01). In addition, as handling events increased, rest decreased, but only in exhibit animals (F(1, 642) = 5.42, p = 0.02). The type of substrate provided was also related to both rest (F(6, 642) = 33.24, p < 0.001) and OBJX behavior (F(6, 160) = 5.24, p < 0.001).

When armadillos were provided with mulch substrate, they rested less (p < 0.001) and performed more OBJX behavior (p < 0.05) than when provided with any other type of

57 substrate. Substrate depth was also positively correlated with rest (F(1, 642) = 219.47, p <

0.001) and OBJX behaviors (F(1, 85) = 14.62, p < 0.001), and negatively correlated with

2 SDB (Wald Χ (1) = 6.06, p = 0.01) and undesirable behavior (F(1, 132) = 16.74, p <0.001).

The factors of age, sex, species, light cycle, regular handling, handling during the first year of life, and years of education program use were not significantly associated with any behaviors (p > 0.15). None of the factors analyzed (including management type) were related to feeding and Not Visible (NV) behaviors (Fig 2.3G and 2.3H, respectively). These behaviors were observed infrequently; armadillos displayed feeding behavior an average of 2.5% of the time and were scored as NV 2.6% of the time.

Figure 2.3: Mean (+SEM) percent of time displaying A) Rest, B) other active (OACT), C) object examination (OBJX), D) locomotion (LOCO), E) self-directed behavior (SDB), F) undesirable behavior (UND), G) feeding (FEED), and H) not visible (NV), in education (n=7), exhibit (n=13), and off-exhibit (n=3) armadillos. Within each behavior, different superscripts denote significant differences between groups. Note that the y-axes are on different scales in panels A (Rest) and E (SDB).

Experiment Two

FGM Results

In armadillos (n = 10), phase of study (Baseline, No Handle, Post) did not have a significant effect on FGM concentrations (F(2, 52) = 0.69, p = 0.50). However overall, the duration of handling was positively correlated with FGM (min/week; F(1, 286) = 5.69, p =

0.02). Individually, three of the ten armadillos showed a difference between the phases of the study (Fig 2.4A-C). These three individuals also experienced longer handling durations (> 900 min/week) during one or more of the phases than any of the other animals experienced during the entire study. In two animals, FGM levels were significantly higher in the phase with the greatest amount of handling (Fig 2.4A, 2.4B).

The third animal (Fig 2.4C) was handled similarly in the Baseline and Post periods, but only the Baseline period showed significant increases in FGM (p < 0.05). The remaining seven armadillos showed no significant differences in FGM between phases (Fig 2.4D).

Similar to Experiment One, substrate type (n = 4; F(4, 69) = 56.66, p <0.001) and depth had significant influences on FGM concentrations. However, the relatively large variety of substrates compared to overall small sample size limited our ability to definitively separate substrate effects from inherent individual variability. Although substrate type was controlled for in the statistical model, specific results are not presented here.

Substrate depth and enclosure size were also correlated to FGM concentrations in this experiment (F(1, 76) = 14.80, p < 0.001; F(1, 80) = 17.38, p < 0.001, respectively), with deeper substrates and larger enclosures being associated with lower FGM.

Figure 2.4: Mean (+SEM) FGM concentrations for four La Plata armadillos (A, B, C, D) during the three phases of Experiment Two. Average number of handling events (frequency) and minutes (duration) per week during each phase is reported within each column. Of ten total armadillos in Experiment Two, the individuals in A and B displayed significantly higher FGM concentrations during the phase with the most handling. The individual in C was handled similarly in the Baseline and Post periods, but a significant FGM increase only occurred in the Baseline phase. The armadillo in D showed no difference in FGM concentrations between the three phases and is representative of the remaining six individuals included in the study. Different superscripts denote significant differences between phases within an individual.

Phase of study also had no overall effect on FGM in hedgehogs (F(2, 150) = 0.19, p

= 0.83). There were, however, considerable individual differences in FGM response across the phases (Fig 2.5). Overall, handling frequency and FGM concentrations were positively correlated (F(1, 448) = 4.92, p = 0.03). However, each hedgehog appeared to have a different FGM response to both handling frequency and duration (Fig 2.5).

Substrate type had a significant relationship with FGM in hedgehogs (F(3, 7) = 14.66, p =

0.003). As with armadillos, the variety of substrates coupled with low sample size made determining definitive substrate effects difficult, and specific results are not reported here. The number of years of education program experience was also significantly related to FGM in hedgehogs, with more years (F(1, 7) = 11.52, p = 0.01) being associated with higher FGM.

There was much less individual variation in red-tailed hawks than in the other species; only one animal showed any difference in FGM levels across phases (Fig 2.6A).

This hawk had significantly higher FGM levels during the Baseline period and was also handled three times longer during this period than any other hawk during any phase of the study (821 min/week on average). Concurrently, handling duration (minutes/week) was positively related to FGM (F(1, 215) = 4.68, p = 0.03). In addition, hawks that were handled in the first year of life (n = 2) had significantly lower FGM than those that were not (n = 4), F(1, 215) = 16.55, p < 0.001). The type of substrate provided was also significantly associated with FGM levels (F(2, 215) = 42.22, p < 0.001). As with armadillos and hedgehogs, the variety of substrates coupled with low sample size prevented analysis of definitive substrate effects, and specific results are not reported here.

Figure 2.5: Individual differences in mean FGM concentrations for six representative African hedgehogs in each phase of Experiment Two. Average number of handling events (frequency) and minutes (duration) per week during each phase is reported within each column. Of twelve total individuals in the study, some (n=3) exhibited significantly higher FGM levels during the phase with the most handling (A, B). Others (n=2) showed the opposite pattern, with significantly lower FGM levels in the phase with the most handling (C). Two animals showed no apparent association between FGM and handling (D), and others (n=4) exhibited no significant difference in FGM between phases at all (E). Only one animal had significantly lower FGM levels during the No Handle period (F). Error bars represent SEM, and different superscripts indicate significant differences (p < 0.05) in FGM between phases within each individual.

Figure 2.6: Average (+SEM) FGM levels in two representative red-tailed hawks during each three-week phase of Experiment Two. Handling frequency (# of events/week) and duration (# of minutes/week) were averaged for each phase and are reported within each column. Individual A had significantly higher FGM during the Baseline phase, when it experienced the most handling. Individual B showed no significant differences in FGM concentrations between the phases and is representative of the remaining four hawks studied. Different superscripts represent significant differences in FGM concentrations between phases within each individual.

Behavior Results

For armadillos, none of the husbandry factors analyzed were related to rest, locomotion, feeding, or NV behaviors. Phase of study also did not affect any of the behaviors overall. However, enclosure size was negatively correlated with SDB (F(1, 284) =

4.02, p = 0.05). Red-tailed hawks performed significantly less SDB as handling frequency or duration increased (Events: F(1, 221) = 17.09, p < 0.001; Minutes: F(1, 221) =

9.97, p = 0.002). Hawks exhibited significantly less Rest Alert behavior during the Post phase than either the Baseline or No Handle periods (Mean % of observations (± SE):

Baseline = 48.11 ± 3.04, No Handle = 52.03 ± 3.03, Post = 45.75 ± 3.12; F(2, 218) = 4.19, p

= 0.02). Rest Alert behavior was also positively correlated with handling frequency (F(1,

152) = 4.83, p = 0.03). In addition, there was a significant interaction between substrate type and phase of study in relation to both Rest Alert and Rest Passive behavior.

However, due to the low animal numbers housed on each type of substrate, specific substrate effects could not be determined. Therefore, substrate type was controlled for in the statistical models, but results are not reported here. In hedgehogs, none of the factors analyzed were related to LOCO, FEED, NV, OBJX, or REST behaviors. Hedgehogs were not observed to perform UND or SDB, and OACT behavior was excluded from analysis because only two occurrences of this behavior were observed during the entire study.

Discussion

This study utilized a multimethod approach to measure welfare in education program animals and resulted in several novel findings. First, methods for non-invasive measurement of glucocorticoids were validated for several species. Second, the results clearly demonstrate that there was no difference in the FGM concentrations of education, exhibit, and off-exhibit armadillos and no effect of handling specifically for education programs on FGM in any species studied. The groups differed in their expression of some types of behavior, but not in the rate of undesirable behavior—a common measure of welfare. However, in both experiments, the amount of handling that each animal experienced, the type and depth of substrate provided in the enclosure, and enclosure size were strongly correlated to both physiological and behavioral measures of welfare.

FGM Validation

Methods of determining FGM were validated in four species with ≥ two-fold increases in FGM within 19-46 hours in every species and a return to baseline ranging

65 from 74-143 hours. Peak FGM (cortisol) responses have previously been reported 30-94 hours following biological events and an ACTH challenge in La Plata armadillos

(Howell-Stephens, Brown, Bernier, Mulkerin, & Santymire, 2012). Our results were similar (peak at 25-40 hours), using an anti-corticosterone antibody. Plasma glucocorticoid levels have been reported in nine-banded armadillos (Czekala, Hodges,

Gause, & Lasley, 1980) as well as FGM concentrations in the pichi, Zaedyus pichiy

(Superina, Carreño, & Jahn, 2009), but to our knowledge, this is the first FGM validation in the screaming armadillo. While FGM response has been successfully measured in other raptors including owls (Wasser et al., 2000; Wasser & Hunt, 2005), American kestrel (Pereira, Granzinolli, & Duarte, 2010), golden eagle, and peregrine falcon (Staley,

Blanco, Dufty, Wildt, & Monfort, 2007), this is the first FGM validation in red-tailed hawks. Similarly, the veterinary exam utilized in the current study is the first reported

FGM validation in hedgehogs or related species.

Differences in Management Role

There were no differences between education, exhibit and off-exhibit armadillos in FGM concentrations or the rate of undesirable behavior, providing strong evidence that management role is not a primary contributor to welfare in armadillos. In addition, the lack of overall relationship between FGM or undesirable behavior and being handled specifically for education programs in Experiment Two suggests that some level of program use does not negatively impact welfare in the species studied. Other recent investigations have concluded that interactions with zoo visitors do not negatively impact animal welfare (goats, llama, and pigs in a petting zoo: Farrand, Hosey, & Buchanan-

Smith, 2014) and may even be a source of enrichment (camels: Majchrzak,

Mastromonaco, Korver, & Burness, 2015; dolphins: Miller et al., 2011). Although there were no differences in FGM or undesirable behavior between groups, education animals exhibited more variety in their behavioral repertoires than exhibit or off-exhibit armadillos. They spent more time locomoting and less time resting, suggesting that education armadillos were more active than other groups when observations took place.

Nine-banded armadillos are reported to spend 18-21 hours/day sleeping (McNab, 1980;

Prudom & Klemm, 1973), which more closely resembles the 80% of time non-education animals spent resting in the current study. However, the majority (86%) of animals in our study were La Plata armadillos and comprehensive activity budgets have not been published for this species in the wild. The 55% of time that education animals spent sleeping could be an indicator of reduced rest overall, or a shift in resting patterns due to education program use. Abou-Ismail et al. (2008) found that laboratory rats with sleeping patterns disrupted by cage cleaning, weighing, and handling not only slept less, but displayed higher levels of behavioral, physiological, and pathological indicators of reduced welfare than rats experiencing these procedures during their normal active period. The very short active period (4-6 hours/day) of wild La Plata and nine-banded armadillos has been described as occurring primarily between 1800 and 0200 (Cuéllar,

2002; McDonough & Loughry, 1997) which does not coincide with the typical times education animals are used for programs. However, education animals may not be sleeping less than exhibit or off-exhibit animals; but instead altering activity patterns to coincide with periods of human activity and sleeping at different times. Behavioral data were collected from 0700-1000 and 1500-1800 in the current study, when armadillos are

67 expected to be largely inactive (as seen in the exhibit and off-exhibit groups), but when education animals are most often handled for programs. Determining which of these explanations is more likely requires further investigation, perhaps through 24-hour activity monitoring using an activity monitoring device.

The fact that education armadillos spent more time engaged in OACT behavior, which was most often scored as behavior oriented toward humans outside of the enclosure, is not surprising given that education animals likely experience more human interaction, and may anticipate contact with humans. Tanida et al. (1995) reported that regularly handled piglets demonstrated increased approach behavior and initiated physical contact with both familiar and unfamiliar humans more than infrequently handled individuals. In regularly handled armadillos, OACT behavior may be a similar attempt to approach or initiate contact. Education animals also performed more SDB than the other groups. Self-directed grooming is a well-established indicator of underlying anxiety in primates and birds (Carder & Semple, 2008; Delius, 1988; Maestripieri,

Schino, Aureli, & Troisi, 1992). Descovich et al. (2012) described a tentative link between self-grooming and anxiety in captive wombats, and suggested the behavior may be a potential indicator of welfare. Similarly, there may be a link between SDB and welfare in armadillos and when considered with our other findings, this warrants further investigation.

Unexpectedly, exhibit animals spent less time performing OBJX behavior than other groups, which was often scored when the animal was digging or had its snout in substrate. This may suggest that either exhibit animals had fewer opportunities for investigation than the other groups, or their regular foraging times did not coincide with

68 our observation periods. The former explanation warrants further attention to ensure that exhibit animals are provided with adequate opportunity to display natural feeding and foraging behavior. However, investigations of wild nine-banded armadillo behavior have determined that the short active period (1800-0200) is dominated by feeding/foraging behavior, which was defined similarly to our OBJX behavior (Ancona & Loughry, 2009;

2010). Thus, exhibit animals may actually be behaving most similarly to wild armadillos if investigative behavior is mainly occurring overnight.

Effect of Handling

Although management type and handling specifically for education programs were not related to our measures of welfare, several husbandry factors were strongly associated with increased FGM and undesirable behavior in both experiments. The amount of handling (either for programs or husbandry) was consistently positively correlated with FGM in armadillos, hedgehogs and hawks, and with undesirable behavior in armadillos in Experiment One, and negatively correlated with SDB in red-tailed hawks. The effects of handing on glucocorticoids and behavior have been documented in several taxa, but studies were focused on animals not accustomed to regular handling.

For example, wild impala kept in a boma had higher cortisol after capture and restraint than unrestrained animals (Hattingh, Pitts, & Ganhao, 1988); cortisol levels in fruit bats rose 800% from baseline after restraint (Widmaier & Kunz, 1993); and handling was associated with increased plasma corticosterone in wild American kestrels (Sockman &

Schwabl, 2001). Domesticated animals also show increases in physiological and behavioral measures of stress in response to handling and restraint (Grandin, 1997;

Hargreaves & Hutson, 1990). The few studies of zoo-housed animals that frequently interact with visitors yielded different results than the former studies. Farrand et al.

(2014) investigated the effect of visitor presence and density on the behavior of goats, llama, and pigs in a petting zoo, concluding that all were relatively unaffected by visitor contact. Similarly, Majchrzak et al. (2015) found no differences in salivary cortisol in camels on days they were providing rides to zoo visitors compared to days standing unsaddled in a pasture. The current study supports these recent investigations, where some handling and human interaction for education programs is not associated with changes in behavior or physiology. However, our results demonstrate that the amount of handling experienced by each animal may have significant welfare implications.

The use of phases in Experiment Two allowed us to compare handling frequency and duration with FGM responses and behavior in three-week periods and determine the response of each animal to various amounts of handling. There was considerable individual variation in FGM response to handling. Hedgehogs were the most variable, with nearly every animal displaying a different relationship between FGM and handling.

It has been suggested that genetic factors (such as personality) interact with previous experience to determine how an individual animal will respond to a particular handling event (Grandin, 1997). The individuality of the response to handling could be evidence for individual personality in hedgehogs, and some animals of this species may be better suited than others for regular handling (Watters & Powell, 2012). Thus, developing measures to assess personality and determining which traits are associated with success as an ambassador animal are important areas for future study. In Experiment Two, only three of ten armadillos and one of six hawks exhibited significantly higher FGM

70 concentrations during one of the phases, and it occurred only when animals were handled more than 800 min/week. There were no other instances of this amount of handling in

Experiment Two. While anecdotal, it does suggest that setting weekly limits to handling in these species may be beneficial. It is clear from Experiment Two that some amount of handling for education programs occurred without eliciting a change in behavior or FGM in many of the animals in the current study. However, the study design did not allow exact thresholds to be determined and further research is necessary.

There is a body of work in laboratory and farm animals demonstrating that handling early in life reduces behavioral and physiological stress responses in adults

(Rats: Meaney et al., 1991; Núñez, Ferré, Escorihuela, Tobeña, & Fernández-Teruel,

1996; Pigs: Hemsworth, Barnett, Hansen, & Gonyou, 1986; Silver Fox: Pedersen, 1994).

In support of this, we observed that red-tailed hawks handled in the first year of life had lower FGM than those that were not. However, this was not observed in armadillos or hedgehogs, so any benefit to early handling could be species-dependent. Habituation to handling later in life has also been shown to reduce behavioral reactivity and avoidance in a variety of species (sheep: Hargreaves & Hutson, 1990; rabbits: Podberscek et al.,

1991; primates: Clay, Bloomsmith, Marr, & Maple, 2009). However, the number of years an animal participated in education programs had no association with FGM or behavior in any species with one exception. More program years were associated with higher FGM in hedgehogs, even after controlling for age, which appears to contradict the idea of habituation. Hogan et al. (2011) concluded that captive wombats did not habituate to handling, but rather developed a ‘learned helplessness’ response. Although avoidance behavior in the wombats decreased over time, stress related behaviors (defecation and

71 stereotypies) increased and there was no reduction in cortisol levels with continued handling. Similarly, FGM and potential stress related behaviors (undesirable behavior in armadillos and SDB in hawks) were associated with handling in the current study. It has been suggested that habituation may depend on the predictability or controllability of the experience (Hargreaves & Hutson, 1990), and the mechanisms of learned helplessness may be related to inescapable contact (Hogan et al., 2011). Therefore, offering more choice and control to all wild animals when being handled may help to facilitate habituation and a reduction in the stress response. The quality and type of early handling may also play a role in reducing fear of humans. Cloutier, Panksepp, & Newberry (2012) demonstrated that playful handling in juvenile rats (i.e. mimicking rat social play) was more effective at reducing the fear response to humans than comparable amounts of handling for routine husbandry or forceful restraint. Therefore, there are many aspects of early and repeated handling that could potentially affect an individual animal’s response to human interaction that deserve future investigation.

Substrate and Enclosure Size Effects

For armadillos and hedgehogs, substrate type and depth were strongly associated with measures of welfare in Experiments One and Two. Both taxa are adapted to digging for food or shelter (Clark, 1951; Santana, Jantz, & Best, 2010; Smith, 2007), so it may not be surprising that opportunities to perform this natural behavior would influence welfare. In line with these results, other studies have demonstrated that providing substrate improves welfare in burrowing and digging animals (Pigs: Tuyttens, 2005,

Bolhuis, Schouten, Schrama, & Wiegant, 2005; Mice: Sherwin, Haug, Terkelsen, &

Vadgama, 2004). Deep substrates were associated with increased OBJX (often scored as digging or rooting behavior) and resting in armadillos, both of which are similar to natural burrowing and foraging behaviors (Ancona & Loughry, 2009, 2010; Smith,

2007). Deep substrates were also negatively correlated with undesirable behavior and

SDB in armadillos, further demonstrating the importance of adequate digging material in these animals. Gerbils have a similarly strong motivation to dig, and providing opportunities to “achieve a goal” (i.e. create a burrow) was more effective at satisfying this motivation and reducing stereotypies than simply providing opportunities to

“perform” the behavior (i.e. providing substrate; Wiedenmayer, 1997). Thus, understanding the motivations behind digging behavior in armadillos and hedgehogs and providing opportunities to achieve the goals of these motivations may also be beneficial.

Several studies have documented a negative correlation between enclosure size and glucocorticoids in zoo-housed mammals (e.g. lynx: Fanson & Wielebnowski, 2013; deer: Li, Jiang, Tang, & Zeng, 2007; gibbons: Pirovino et al., 2011). Similarly, armadillos in larger enclosures had lower FGM concentrations in both experiments and performed less SDB. Education armadillos in this study were generally housed in smaller enclosures (~2 m2 on average in both experiments) than their exhibit counterparts (~7.5 m2 on average) while off-exhibit armadillos averaged 1.5 m2 of enclosure space. These results highlight a general discrepancy between on and off exhibit housing and the importance of enclosure size in promoting good welfare for this taxon.

Conclusions

Handling frequency and duration, enclosure size, and substrate were significantly associated with behavioral and physiological measures of welfare in the current study.

The individuals’ management role (i.e., education or exhibit) or being handled specifically for education programs had no significant effect on any behavioral or physiological indicator of welfare used here. The fact that these results were consistent across both experiments and in all three species indicates potentially strong relationships between these management factors and welfare in zoo–housed armadillos, hedgehogs, and hawks. This study was the first approach to evaluating welfare in ambassador animals and has led to questions worthy of future investigation, including evaluating thresholds or weekly limits to handling, individual animal personalities in relation to handling and management type, factors contributing to the tolerance of handling including crowd size and offering choice and control, and the effect of diet and enrichment on measures of welfare in education and exhibit species. It is clear that providing adequate substrate is a simple and practical step to ensuring better welfare in burrowing animals. Future investigations will hopefully clarify specific space and environmental requirements and handling thresholds for individual species which will lead to husbandry recommendations that allow zoo-housed animals to thrive in any role.

CHAPTER THREE: Personality assessment in zoo-housed cheetahs

Introduction

Although anyone that has ever worked or lived closely with animals would not hesitate to assert that individual animals have different personalities, the scientific community at large has been hesitant to agree. However, it is becoming increasingly clear that individual behavioral patterns exist and can be measured (Gosling, 2008; Watters &

Powell, 2012). In the literature, these individual behavioral patterns have been termed personality, temperament, and behavioral syndromes, among other names (Gosling,

2001; Tetley & O’Hara, 2012; Watters & Powell, 2012), but have all generally referred to behaviors that are relatively consistent within an individual across time and context. I will use the term ‘personality’ here, in order to be consistent with recommendations from other animal personality researchers (Gartner & Weiss, 2013a; Gosling, 2001; Watters &

Powell, 2012).

The majority of published studies on animal personality have focused on nonhuman primates in laboratory settings and domestic species, such as cats and dogs.

However, there is great potential for utilizing personality information to inform management decisions in zoos. Previous research on zoo-housed species has shown that personality can play an important role in reproductive success (Carlstead, Mellen, et al.,

1999; Powell & Svoke, 2008; Wielebnowski, 1999), social group formation (Gold &

Maple, 1994), effectiveness of environmental enrichment (Gartner & Powell, 2012), and behavioral response to zoo visitors (Barlow, Caldwell, & Lee, 2007). Clearly,

75 understanding the measurement and application of animal personality would be a useful tool for animal managers striving to ensure optimal welfare.

In addition to primates, felids have been a popular taxon for personality research in zoos (Gartner et al., 2014; Gartner & Weiss, 2013a), often in an effort to utilize individual personalities to predict reproductive success or responses to the captive environment. Cheetahs have been the most extensively studied, with four separate investigations of this species including some measure of personality (Baker & Pullen,

2013; Chadwick, 2014; Razal et al., 2016; Wielebnowski, 1999). Though these studies had similar research goals, the findings have not always been consistent. For example,

Wielebnowski (1999) found that non-breeding cheetahs scored significantly higher on the component “tense-fearful” than those that had successfully produced cubs. Using a similar personality instrument, Chadwick (2014) found no differences between successful and unsuccessful breeding pairs on the component “fearful-insecure”. Razal et al. (2016) used the same survey and found that only the component “unsociable” was related to breeding success, not their component labeled “insecure”, and Baker & Pullen (2013) did not include a measure of reproductive success in their investigation. It is unclear if these inconsistencies are due to the sample of cheetahs assessed or to slight differences in survey development and analysis.

There is enough overlap between these investigations, however, to suggest that a stable personality structure across all cheetahs may exist. What remains uncertain is how personality may be related to stress or reproductive success in the cheetah. A larger, more diverse sample of individuals is needed to fully understand how personality types may be related to zoo-based management and welfare in this species. This chapter focuses on the

76 measurement of personality in cheetahs as part of a larger study investigating welfare across a variety of management roles, including exhibit, off-exhibit, and ambassador animals. The goals of this aspect of the project were to use previously established methodologies to identify personality structure in this sample of cheetahs and then determine what, if any, individual characteristics of the animal (sex, rearing history, social housing, management role, etc.) might predict personality types. The personality information established here will then be incorporated into the larger analysis of behavioral and physiological indicators of welfare.

Methods

Subjects

All methods and animal use were reviewed and approved by each participating institution and the Animal Care and Use Committee at the Cleveland Metroparks Zoo. As part of a larger study, 73 (41.32) cheetahs were sampled for the personality assessment during one of two data collection years (2014 & 2015). Seven institutions housing 40 cheetahs (24.16) were evaluated in 2014 and an additional seven institutions housing 33 cheetahs (17.16) were evaluated in 2015.

Survey Data Collection

Husbandry and Demographic Survey

Several surveys were completed for each cheetah by primary caretakers to gather information on potential factors that may be associated with indicators of welfare.

Demographic and husbandry questionnaires included information such as age, sex,

77 rearing type (hand- or mother-reared), lifetime reproductive history, and current housing and social conditions for each cheetah (Appendix I).

Based on the information provided, individuals were classified as ambassadors, exhibit, or off-exhibit based on their primary management role. Cheetahs were considered “ambassadors” (n=17) if their “role includes handling and/or training by staff or volunteers for interaction with the public and in support of institutional education and conservation goals” (AZA Ambassador Animal Policy, 2011), and they participated in at least one public program during the study period. Animals that were primarily housed on exhibit for public viewing, but not trained for education programs were classified as

Personality Assessment

A personality assessment survey was also completed for each cheetah by a minimum of two primary caretakers. Raters were advised not to discuss the ratings of any individual cheetah prior to completing the questionnaire. A total of 38 raters completed surveys for the cheetahs in this study, with 2-5 raters assessing any individual cheetah.

Raters had an average of 7.08 ± 5.38 (range: 0.25-25) years of experience working with cheetahs in general, and an average of 3.24 ± 2.13 (range: 0.25-10) years of experience with the cheetahs they rated (Table 3.1).

Table 3.1: Subjects and raters included in personality analysis. Four institutions did not meet the 75% agreement criteria for within-institution reliability, and these institutions also were the only ones to include raters with one year or less of experience working with cheetahs. These raters were therefore removed from further analysis to improve institutional reliability. Data for these four institutions are shown before and after rater exclusion.

% Agreement Years experience N Raters Total Across Traits Institution N Cheetahs w/cheetahs (per cheetah) within institution Mean (Range) Mean (SEM) 1 4 2 (2) 3.75 (3-4.5) 80.52 (6.9) 2 1 2 (2) 4.5 (3-6) 80.33 (3.0) 3 2 2 (2) 4 (3-5) 86.38 (7.2) 4 5 4 (2) 11.5 (6-25) 78.29 (8.0) 5 6 3 (3) 6.3 (1-10) 72.53 (7.5)* 6 2 3 (3) 8.5 (2-12) 79.19 (8.3) 7 4 2 (2) 8.75 (3.5-14) 83.40 (6.4) 8 2 2 (2) 2.75 (1-4.5) 69.43 (10.0)* 9 2 2 (2) 15 (15) 76.79 (12.5) 10 12 2 (2) 11 (10-12) 77.21 (5.6) 11 8 5 (2) 4.5 (2-8) 77.50 (6.4) 12 7 2 (2) 3.6 (0.25-7) 73.52 (8.5)* 13 16 2 (2) 5.5 (2-9) 77.92 (4.6) 14 2 5 (5) 7.2 (1-17) 63.48 (7.8)*

*After removing raters with 1 year or less of experience with cheetahs: 5 6 2 (2) 9 (8-10) 80.90 (6.4) 8 2 1 (1) 4.5 -- 12 7 1 (1) 7 -- 14 2 4 (4) 8.75 (3-17) 75.86 (7.6)

The survey was based on previously published felid personality surveys (Fazio,

2016; Feaver, Mendl, & Bateson, 1986; Wielebnowski, 1999) from which 21 traits were taken (Table 3.2). Traits were assessed using a continuous scale methodology described in (Feaver et al., 1986) where a 100 mm calibrated horizontal line was placed next to each item on the survey, with the minimum score on the left (this cheetah NEVER exhibits this trait) and the maximum score on the right (this cheetah ALWAYS exhibits

79 this trait). Raters placed an “X” on the line for each item indicating their impression of the individual compared to all cheetahs they have ever known. The distance from the left side of the line to the “X” represents the numerical score for each individual on each item on a scale of 0-100 (Appendix II).

Table 3.2: Behavioral definitions of adjectives used in personality questionnaire.

Active Moves frequently (i.e.., patrols, runs, paces, stalks a lot) Aggressive to Conspecifics Frequently reacts with hostility (i.e., attacks, growls) toward other cheetahs Aggressive to Familiar Frequently reacts with hostility and/or threatening People behavior toward primary staff Aggressive to Strangers Frequently reacts with hostility and/or threatening behavior toward unfamiliar people Calm Not easily disturbed by changes in the environment Curious Readily approaches and explores changes in the environment Eccentric Shows stereotypic or unusual behaviors Excitable Overreacts to changes in the environment Friendly to Conspecifics Social; initiates and seems to seek out close proximity to other cheetahs Friendly to Familiar Initiates close proximity to primary staff; approaches People readily and in a friendly manner (i.e. rubs on fence, purrs) Friendly to Strangers Initiates close proximity to unfamiliar people; approaches readily and in a friendly manner (i.e. rubs on fence, purrs) Fearful of Conspecifics Retreats and hides readily from other cheetahs Fearful of Familiar People Retreats and hides readily from primary staff Fearful of Strangers Retreats and hides readily from unfamiliar people Insecure Seems scared easily; “jumpy”, and fearful in general Playful Initiates and engages in play behavior (seemingly meaningless, but nonaggressive behavior with objects, people, and/or other cheetahs Confident Moves in a seemingly confident, well-coordinated, and relaxed manner

Smart Learns to quickly associate certain events and appears to remember for a long time Solitary Spends time alone, seems to avoid company Tense Shows restraint in movement and posture Vocal Frequently and readily vocalizes

Statistical Analysis

Personality analysis

Inter-rater reliability was assessed both within institutions and within each trait

(across all institutions). Due to the variability in the number of animals at each institution and the number of raters for each animal, a single statistic could not be applied in all situations to assess inter-rater reliability within each institution (the degree to which different raters were in agreement when assessing the same animal). Therefore, a simple percent agreement method for determining reliability was employed. The mean difference between the two raters’ scores for each trait were calculated and then averaged across all cheetahs within the institution. When three or more raters scored the same cat

(n=10), the highest and lowest scores were used to calculate difference values. These difference scores were then averaged across all 21 traits to determine the degree to which raters were consistent when assessing the same animal. Because a 100-point scale was utilized on the surveys, an average difference of 25 translates to a 75% agreement within an institution, which was the cutoff value used to deem an institution reliable.

To assess the inter-rater reliability across institutions (the degree to which all raters could consistently assess any one trait), intra-class correlation coefficients ICC(3,1) and ICC(3,k) were calculated (Shrout & Fleiss, 1979). As in previous studies (Gartner et al., 2014; Gartner & Weiss, 2013b; Lee & Moss, 2012), items with an ICC(3,1) and/or an

ICC(3,k) less than or equal to zero were deemed unreliable and excluded from further analysis. Intra-class correlation analysis was conducted using IBM SPSS v. 23 for

Windows.

Items deemed reliable were then subjected to principal components analysis

(PCA) using SAS University Edition v. 3.6 (SAS Institute, Cary, NC, USA). The Kaiser criterion (Eigenvalues >1) and examination of the scree plots were used to determine the number of components to retain (O’Rourke & Hatcher, 2013). Components were rotated via the varimax procedure and factor loadings ≥|0.4| were considered relevant as in previous studies (Gartner et al., 2014; Gartner & Weiss, 2013b; Lee & Moss, 2012).

Based on these loadings, component scores were calculated for each cheetah using a least squares regression approach (DiStefano, Zhu, & Mîndrilă, 2009; O’Rourke & Hatcher,

2013). In addition to this numeric score for each of the retained components, each cheetah was also assigned to a “dominant personality” category based on the component with the highest positive score. The numeric personality values and the categorical

“dominant personality” variable never appeared in the same statistical model, as they are representations of the same construct.

To understand factors that might predict cheetah personality, the numeric factor scores for each of the components identified through PCA were analyzed in relation to age, sex, hand rearing, social group type, management role, and breeding success.

Shapiro-Wilk tests of normality confirmed that none of the components were normally distributed, so non-parametric tests were used. For categorical variables, Wilcoxon Rank

Sum or Kruskall-Wallis tests were used and Spearman correlation was used for the

82 continuous variable of age. For breeding success, only cheetahs that have had reproductive opportunity in their lifetime were included (n = 41; 19 males, 21 females).

Results

Inter-rater Reliabilities

Four institutions did not meet the established within-institution reliability criteria

(≥ 75% agreement). These institutions were also the only facilities with raters that had one year or less of experience working with cheetahs (one rater at each institution). These four raters were excluded from analysis, which allowed all 14 institutions to meet reliability criteria (mean reliability = 82.45; SD = 7.94). However, this removal resulted in two institutions (representing nine cheetahs) only having one rater assessment per cheetah. Within-institution agreement before and after the exclusion of raters is described in Table 3.1. As the removal of these raters greatly improved reliability, and there was a commonality between the raters that were excluded, we felt this decision was justified.

For reliability of individual items, the ICC(3,1) ranged from 0.03 (tense) to 0.57

(playful), with a mean reliability of 0.34. ICC(3,k) ranged from 0.04 (calm) to 0.71

(friendly to familiar people), with a mean reliability of 0.41. The ICC values for four items: active, aggressive to conspecifics, eccentric, and fearful of familiar people, were ≤

0 and were excluded from further analysis.

Principal Components Analysis

Based on the eigenvalues and scree plot, four components accounting for 69.13% of variance were retained (Table 3.3). Each component was labeled according to the

83 items with the highest loadings: Tense-Insecure, Aggressive (toward people), Playful-

Friendly (toward people), and Social (with other cheetahs). The fourth component was reverse coded (all factor scores multiplied by -1) for ease of interpretation. The reliabilities of individual ratings, ICC(3,1), were .26 for Tense-Insecure, .43 for

Aggressive, .53 for Playful-Friendly, and .33 for Social, with a mean reliability of .39.

The reliabilities of mean ratings, ICC(3,k), were .33 for Tense-Insecure, .53 for

Aggressive, .64 for Playful-Friendly, and .37 for Social, with a mean reliability of .47.

For each cheetah, the dominant personality type was assigned according to the component with the highest positive factor score, resulting in 18 cheetahs (25%) categorized as Tense-Insecure, 17 (23%) as Aggressive, 21 (29%) as Playful-Friendly, and 17 (23%) as Social. All four dominant personality types were represented in each of the three management roles, though most Playful-Friendly cheetahs (n=10) were ambassadors, and most Aggressive (n=11) and Tense-Insecure (n=13) cheetahs were housed off exhibit.

Factors Related to Personality

Age was negatively correlated with Component 3, Playful-Friendly (rs = -0.458, p

< 0.001). For Component 4, Social, males scored higher than females (median: males =

0.68, females = -0.80; Z = -5.53, p < 0.001) and cheetahs housed in social groups scored higher than those housed alone (median: social = 0.43, alone = -0.91; Z = -5.61, p <

0.001). In addition, successful breeders (n = 18) scored significantly lower on Component

1, Tense-Insecure, than unsuccessful breeders (n= 23; median: breeders = -0.34, non- breeders = 0.20; Z = -2.06, p = 0.04).

Hand reared cheetahs scored significantly lower on Component 2, Aggression

(median: Yes = -0.59, No = 0.16; Z = -3.13, p = 0.002) and significantly higher on

Component 3, Playful-Friendly (median: Yes = 0.56, No = -0.39; Z = 4.57, p <0.001).

There was also a significant effect of management role on Component 3, Playful-Friendly

2 (Χ (2) = 16.63, p < 0.001). Ambassador and exhibit cheetahs had significantly higher scores (medians: ambassador = 0.51, exhibit = 0.44) than off exhibit cheetahs (median = -

0.39). However, all ambassador animals (n=17), half of exhibit cheetahs (n=10), and only two off exhibit cheetahs were hand reared, indicating a possible confound. There were no other significant associations between management role and personality.

Table 3.3: Four major components of cheetah personality derived from observer ratings of 71 cheetahs housed at 14 AZA-accredited zoos and obtained through principal components analysis. Traits with factor loadings >|0.4| define each component and are in boldface. Components were labelled according to the traits with the highest positive loadings. Prior to statistical analysis, the component “Social” was reverse coded (component scores for each cheetah multiplied by -1) for ease of interpretation. Principal components analysis: Varimax rotation Tense- Aggressive Playful/Friendly Social Item Insecure (to people) (to People) (w/cheetahs) Aggressive to Familiar .16 .88 -.16 .11 People Aggressive to Strangers .23 .91 .00 .00 Calm -.75 -.06 .12 -.17 Curious -.72 .02 .30 -.04 Excitable .73 .14 -.11 -.13 Friendly to Conspecifics -.15 .04 .00 -.85 Friendly to Familiar People -.18 -.50 .72 -.06 Friendly to Strangers -.18 -.50 .70 .06 Fearful of Conspecifics .53 -.01 .30 .51 Fearful of Strangers .72 .45 .13 -.01 Insecure .87 .23 .02 .08 Playful -.21 .22 .79 -.06 Confident -.80 -.01 .16 -.29 Smart -.59 -.18 .10 -.07 Solitary .06 .08 -.13 .86 Tense .73 .10 -.16 .14 Vocal -.04 -.10 .50 -.48

Discussion

This study aimed to quantify personality in this sample of cheetahs using previously established methodologies with the eventual goal of relating personality structure to behavioral and physiological indicators of welfare. After establishing reliability both within and between institutions, four components of cheetah personality were identified that are comparable to those found in previous assessments in this species. There is some evidence that these personality types may be associated with

86 rearing history and social group housing, though potential confounds warrant interpreting these results with caution.

Personality Assessment

The personality assessment used in this study was shown to be generally reliable across 14 institutions and 73 cheetahs, similar to other animal personality surveys utilizing an observer rating method (Gartner et al., 2014; Gartner & Weiss, 2013b; Weiss,

King, & Figueredo, 2000). Though some previous studies have found that raters with limited experience with the animals they are rating can be reliable (Feaver et al., 1986;

Wielebnowski, 1999), this study found that raters with less than one year of experience working with cheetahs differed significantly in their assessments from those with more experience. The degree of familiarity between observers and subjects is a logical factor that may affect inter-observer agreement. However, previous reviews of animal personality methodology have yet to determine exactly how much experience is

“enough” (Gosling, 2001; Watters & Powell, 2012). For this reason, it is critical to continue to measure observer familiarity with the animals rated in personality studies and evaluate any potential reliability issues on a case-by-case basis.

The personality survey utilized here defined four distinct components: Tense-

Insecure, Aggressive, Playful-Friendly (toward people), and Social (with conspecifics).

These components are similar to those found in various other personality assessments of cheetahs and other felid species. Tense-Insecure or its equivalent has previously been identified as a personality component in cheetahs (Chadwick, 2014; Razal et al., 2016;

Wielebnowski, 1999), domestic cats, clouded leopards, African lions, and snow leopards

(“Neuroticism”, Gartner et al., 2014; “Timid/Anxious”, Gartner & Powell, 2012).

Aggression has been previously identified in cheetahs (Razal et al., 2016; Wielebnowski,

1999) and similar traits comprised a component labeled “Dominance” in the five felid species assessed in Gartner et al. (2014). Friendliness toward People was also described in cheetahs (Baker & Pullen, 2013; Chadwick, 2014) and a similarly structured component was found in both Scottish wildcats (Gartner & Weiss, 2013b) and domestic cats (Feaver et al., 1986; Gartner et al., 2014). Finally, the component of Sociability has been previously identified in studies of cheetahs (Baker & Pullen, 2013; Chadwick, 2014;

Razal et al., 2016) but a clear equivalent has not been noted in other felid species studied.

The consistency across studies indicates a potential for the existence of stable personality structure across all zoo-housed cheetahs.

In a larger context, there are similarities between the personality structure of the five felid species described in Gartner et al. (2014) and that of cheetahs described here.

Although variation in methodology and sampling has likely prevented the exact same personality structure from being identified in all species, there is enough similarity to support the idea that there is a consistent personality structure across the taxon and the development of a single, universal felid personality survey is possible and warranted, as advocated by previous researchers (Tetley & O’Hara, 2012; Watters & Powell, 2012).

Assessing a wide range of felid species using the same survey instrument would not only increase understanding of how personality structures may have evolved throughout the

Felidae clade, but also may help with the development of management plans that can be tailored to individual needs.

Factors Related to Personality

The factors that are related to personality component scores were similar to those found in previous investigations. Males and cheetahs housed in a social group scored higher on the Social component, which aligns with the findings of Baker & Pullen (2013) and Chadwick (2014). In the wild, males live in groups, termed coalitions, and females are generally solitary unless they are with cubs (Caro, 1994), so it is not surprising that male cheetahs would be rated as more social. It is possible that group-housed cheetahs were rated higher on the “friendly to conspecifics” trait and singly-housed cheetahs were rated as more “solitary” by their caretakers simply due to their current housing situation.

However, housing individuals in groups does not ensure positive interactions. Likewise, housing individuals alone, especially in breeding center situations, does not prevent social interactions with other individuals. Kuhar, Stoinski, Lukas, & Maple (2006) found that male gorillas scoring highly on the personality factor “understanding” were more likely to be housed in social groups. This component was comprised of several traits related to calm, positive social behavior, and it was argued that high understanding scores may be predictive of successful social group housing, rather than a consequence.

Similarly, cheetahs with low Social scores may be more suited to solitary housing. In a species with a fluid social structure such as the cheetah, Sociability scores may prove to be a useful management tool when deciding to form or dissolve social groups, especially in males.

After limiting analysis to cheetahs with reproductive opportunity, non-breeders scored higher on the component Tense-Insecure than those that had successfully produced cubs. This finding is consistent with that of Wielebnowski (1999) who also

89 found that reproductively successful cheetahs scored lower on the component “tense- fearful” compared with unsuccessful breeders. However, Chadwick (2014) did not find a relationship between breeding success and her “tense-fearful” component. That study classified successful and unsuccessful breeders in pairs, rather than individuals, which may not have been specific enough to detect individual personality differences. Another major finding in her study was that the personality component scores of unsuccessful breeding pairs were more similar to each other than scores from successful pairs. If successful breeding pairs have dissimilar personalities, it may be more useful to examine personality at the individual level, especially in relation to breeding success. In addition, the Chadwick study was conducted exclusively in European institutions, so the difference in management strategies and cheetah populations between North America and Europe may also be playing a role in this discrepancy.

Ambassador and exhibit cheetahs were scored as more Playful-Friendly than cheetahs housed off exhibit. However, there is a large degree of overlap between management role and the other factor associated with this component, hand rearing.

Though causality cannot be determined from this study alone, it is more plausible that early rearing history would have a larger effect on personality than management role, which may change throughout an animal’s life. Hand-reared cheetahs were observed to have higher scores on the Playful-Friendly component and lower scores on the

Aggression component. Although these traits are highly desirable in successful ambassador animals, in some species or individuals, hand rearing may result in behavioral and reproductive challenges that would not be ideal for individuals in non- ambassador roles. Much of the literature focusing on hand rearing in zoos has focused on

90 primate species. These studies stress the importance of early exposure to conspecifics and an appropriate social environment; without which, individuals may exhibit increased stereotypic behavior (Lutz, Well, & Novak, 2003; Meder, 1989) and inappropriate reproductive, maternal, and social behavior (Beck & Power, 1988; Freeman & Ross,

2014;. King & Mellen, 1994; Ryan, Thompson, Roth, & Gold, 2002). There is considerably less information about the long-term behavioral consequences of hand rearing in felid species. Mellen (1992) found that human reared domestic cat females copulated significantly less and were more aggressive to male partners and human caretakers than parent reared individuals, leading to reduced reproductive success.

Analysis of international studbook data for Siberian tigers, clouded leopards, snow leopards, and cheetahs also revealed significant differences in reproductive success between hand reared and parent reared individuals of all species (Hampson & Schwitzer,

2016). Fewer offspring were produced by hand reared tigers (females), cheetahs (males), and snow leopards (both sexes) compared to parent reared individuals. In addition, hand reared mothers were less likely to subsequently rear their own offspring in all species.

In cheetahs, the potential for reproductive success issues related to personality may present a challenge for population sustainability. The AZA cheetah SSP is currently listed as a Yellow program, meaning that under the current rate of breeding, the population cannot maintain 90% genetic diversity for 100 years. There are also currently

59 individuals (16% of the population) excluded from breeding because they are ambassador animals, which further contributes to sustainability challenges (Crosier et al.,

2018). In an effort to increase reproduction, some institutions have begun to consider strategies for incorporating ambassadors, which are always hand reared, into the breeding

91 population. Providing these animals with the opportunity to breed may become a viable way to meet the growing demand for ambassador cheetahs without removing animals from the breeding population, thus meeting both the sustainability and educational goals of the program. If reproductive success is additionally related to the Tense-Insecure component, developing strategies for increasing population sustainability in animals with this personality type may prove more challenging. None of the other factors measured here were related to this component, so additional research is necessary to understand what may contribute to the development of this personality type and if there are any potentially negative welfare consequences associated with this personality type under certain management conditions.

The personality structures identified here will be included in analysis of factors related to behavioral and physiological indicators of welfare in this sample of cheetahs in

Chapter Four. As welfare is an individual experience, it is reasonable to assume that animals with different personalities may respond differently behaviorally or physiologically to the same housing, husbandry, and management practices. Having a means to quantify these individual differences is beneficial for understanding the whole picture when evaluating welfare and ensuring management recommendations can be customized for individual needs.

CHAPTER FOUR: Assessing Welfare in Zoo-Housed Cheetahs in Different

Management Roles

Introduction

Cheetahs (Acinonyx jubatus) are an iconic zoo species that serves in multiple roles across institutions accredited by the Association of Zoos and Aquariums (AZA). Though popular exhibit animals, the North American cheetah population is not self-sustaining, largely due to poor reproductive success (Marker-Kraus & Grisham, 1993) and a high prevalence of diseases such as chronic gastritis and veno-occlusive disease (Munson,

1993; Munson et al., 1999). These issues occur at a much higher rate in the captive population compared to free-ranging animals (Evermann, Laurenson, McKeirnan, &

Caro, 1993; Munson et al., 2005) and it has been suggested that environmental factors

(Wielebnowski, Ziegler, et al., 2002) and chronic ‘stress’ may be contributing to these observed differences (Terio et al., 2004). In an attempt to limit potential effects of environmental stressors on cheetah reproduction, many genetically valuable adult cheetahs are housed in breeding centers which typically consist of large enclosures with little exposure to the public. In fact, more than 90% of cheetah cubs produced in the last decade have come from these off-exhibit breeding centers as opposed to traditional zoos

(Koester et al., 2015).

In addition to housing on exhibit at zoos and off exhibit in breeding centers, cheetahs are also an increasingly popular ambassador animal species. There are currently nine AZA-accredited facilities that utilize cheetahs in their animal ambassador programs, and there is a growing interest in this species from other zoo education programs due to

93 their “nonaggressive temperament and ease of trainability and handling when raised by experienced handlers” (Rapp et al., 2017). The housing and management requirements of a large carnivore, such as a cheetah, are quite different from that of many small mammals and birds in ambassador programs. Ambassador cheetahs are hand reared from an early age and spend a large amount of their time in close proximity to their trainers as they mature. A great deal of positive reinforcement and desensitization training is required to raise a successful ambassador cheetah that is able to routinely participate in outreach programs both on and off grounds (Rapp et al., 2017). As a result, ambassador cheetahs have a different life experience compared with those in other roles. There has been a substantial amount of research focusing on cheetah reproduction (Brown et al., 1996;

Brown & Wildt, 1997; Crosier et al., 2011; Terrell et al., 2011; Wildt et al., 1988), behavior (Quirke & O’Riordan, 2015; Quirke, O’Riordan, & Zuur, 2012; Wielebnowski,

Ziegler, et al., 2002), and aspects of zoo management (Koester et al., 2015; Wells et al.,

2004; Williams, Waran, Carruthers, & Young, 1996), but none of these studies have included ambassador cheetahs. Though our previous study (Baird et al., 2016) found no differences in fecal glucocorticoid metabolite (FGM) concentrations between ambassador, exhibit, and off-exhibit armadillos, there is a need to understand if this finding holds true across taxa. Further, because cheetahs are managed in such disparate roles across AZA and it is thought that this species may be especially sensitive to aspects of living in a zoo environment, it is important to understand how the varied housing and management conditions inherent in each of these roles are influencing cheetah welfare.

The goals of the current study were to build upon previous research on zoo- housed cheetahs by sampling cats housed under a variety of management conditions,

94 including ambassador animals. Specifically, we aimed to determine the extent to which housing and husbandry practices, personality, human interaction, exercise, and enrichment were related to FGM and behavior in cheetahs. Based on the findings of our previous ambassador animal study (Baird et al., 2016), we predicted that there would be no differences in FGM or undesirable behavior between cheetahs in different management roles, but that housing and husbandry variables, such as amount of human interaction, would be correlated with FGM and behavior. Understanding the associations between these varied practices and indicators of welfare in this species will help inform best practices for cheetah care across a variety of management roles.

Methods

Experimental Design

All methods and animal use were reviewed and approved by each participating institution and the Animal Care and Use Committee at the Cleveland Metroparks Zoo.

To assess physiological and behavioral measures of welfare, 73 (40.33) cheetahs were sampled for a 90-day period in one of two data collection years (2014 & 2015).

Seven institutions housing 41 cheetahs (24.17) were sampled in 2014 and an additional seven institutions housing 32 cheetahs (16.16) were sampled in 2015. To help control for any seasonal or crowd size effects, northern climate institutions were limited to sampling during peak visitor season (May-early October), while southern-climate institutions were able to collect data through December. Ten institutions were able to dedicate the additional time and resources required to collect behavioral data from a subset of these cheetahs (n = 42: 27.15; see Behavioral Data Collection). For two cheetahs (1.1) at two

95 separate institutions, behavioral data, but not fecal samples, were collected, leaving 71

(39.32) cheetahs included in hormone analysis.

As described in the previous chapter, individuals were classified as ambassadors, exhibit, or off-exhibit based on their primary management role. Cheetahs were considered “ambassadors” (n=18) if their “role includes handling and/or training by staff or volunteers for interaction with the public and in support of institutional education and conservation goals” (AZA Ambassador Animal Policy, 2011), and they participated in at least one public program during the study period. Animals that were primarily housed on exhibit for public viewing, but not trained for education programs were classified as

“exhibit” (n=21). Cheetahs that were primarily housed in off-exhibit facilities (though not necessarily actively breeding) with little contact with the public were considered “off- exhibit” (n=34). However, many cheetahs filled multiple roles throughout the study. For example, many “ambassador” cheetahs were housed on public exhibit when not participating in education programs and many “off-exhibit” cheetahs were regularly rotated between spaces that may or may not be in view of the public. To account for this, each management role was also examined separately by comparing cheetahs that had spent some time in that role during the study to those that had spent no time in that role during the study (Table 4.1).

Table 4.1: Husbandry and demographic factors measured in 73 cheetahs housed at 14 institutions.

Factor N Males N Females Sex 40 33 Year of Study Year 1 (2014) 24 17 Year 2 (2015) 16 16 Primary Management Role Ambassador 8 9 Exhibit 15 6 Off Exhibit 17 18 Ambassador animal? Yes 11 10 No 29 23 Housed on exhibit? Yes 21 14 No 19 19 Housed off exhibit? Yes 17 18 No 23 15 Contact Type Free 31 28 Protected 9 5 Hand Reared?a Yes 15 15 No 23 18

Mean Range Age (y) 6.85 1.70-15.85 Space Experience (ft2) 9923.76 1170.06-23876.75 Human Interaction Frequency per week 8.94 0-59 Duration per week (min) 78.05 0-690

Exercise Frequency per week 0.79 0-9 Duration per week (min) 6.49 0-190 Enrichment Frequency per week 2.19 0-9 a Two male cheetahs were considered “partially hand-reared” as they were hand reared for a short time before being cross-fostered and subsequently raised by another female cheetah. These two individuals were removed from all analysis containing the hand-rearing variable.

Survey Data Collection

Husbandry and Demographic Survey

Several surveys were completed for each cheetah by primary caretakers to gather information on potential factors that may be associated with indicators of welfare.

Demographic and husbandry questionnaires included information such as age, sex, rearing type (hand- or mother-reared), lifetime reproductive history, and current housing and social conditions for each cheetah (Appendix II). As many cheetahs in the study were routinely given access to several different enclosures each day, we chose to use a “space experience” metric (described in Meehan, Hogan, Bonaparte-Saller, & Mench, 2016) to assess enclosure size. In brief, the size (in ft2) of each enclosure was multiplied by the estimated proportion of time each cheetah spent in that enclosure over the study period.

These weighted enclosure sizes were then averaged to calculate the final space experience for each cheetah in the study.

Daily Logs

Each institution also recorded information about every instance of human interaction, keeper-directed exercise, and enrichment that the animal experienced during

98 the study period (Appendix III). For human interactions and exercise, the date, time, duration, and a brief description of the event were recorded. It was also noted if the event occurred in free (humans sharing space with the animal) or protected contact (a physical barrier separating human space from animal space), whether unfamiliar people were present or not, and a subjective rating of the animal’s interest level in the event (1-5 scale).

Types of interactions and exercise were grouped into broad, non-mutually exclusive categories for comparison across institutions (described in Table 4.2A and B).

For enrichment, only frequency and type were recorded. Enrichment types were similarly grouped into broad, non-mutually exclusive categories (Table 4.2C). Every instance of human interaction, regardless of type or duration, was counted as a single event and calculated as a weekly total (“interaction frequency”). The duration of each event was used to calculate the total minutes of interaction per week (“interaction duration”). If the duration for a given event was missing, (95 out of 7628 total events), the average duration for similarly labeled events for that individual was used for analysis. The same method was used to calculate “exercise frequency”, “exercise duration”, and “enrichment frequency”. Human interactions, exercise, and enrichment were analyzed separately and never appeared in the same statistical model, as a single event may fall into several of these categories.

Table 4.2: Broad categories used to classify A.) human interaction, B.) exercise, and C.) enrichment events for comparison across institutions. Categories were not mutually exclusive, and therefore no two categories ever appeared in the same statistical model.

Human Interaction Description Category Training Positive reinforcement training sessions that typically follow the pattern of cue-behavior-reward. Training can take place in a variety of contexts with a wide range of behavioral goals. Examples: weighing, participation in medical procedures (blood draws, ultrasounds), “show” behaviors (station, lure running, A to B, “posing” for photo opportunities). Hand Feed Manually delivering food items in small amounts to the animal by hand or feed pole. Training sessions typically come with the expectation of hand feeding a reward, so this category was only scored independently of “training” where food was delivered without a cue or behavior being performed. Shift Moving animal from one location to another. Typically occurs more frequently than full training sessions and in a less structured manner. Relationship Human-animal interaction for the sole purpose of relationship building. Examples: “scratches” or “pets” occurring in free contact or through mesh, “grooming sessions”, time spent in the enclosure with the animal outside of structured training sessions and husbandry routines. Free Contact Husbandry services performed while sharing space with the Husbandry animal (cleaning, feeding, providing enrichment, shifting). Whether the human directly interacts with the animal or not, the act of sharing space can be considered a change of the animal’s normal routine, and therefore an interaction. Outreach/ Public encounter or demonstration involving a cheetah. Can Program/Show include programs with ambassador animals such as running demonstrations, educational presentations or “shows”, media appearances, and up-close encounters such as photo opportunities or painting sessions. Can also include presentations with non- ambassador cheetahs such as keeper talks or behind-the-scenes tours if part of the presentation involves direct interaction between the cheetah and a person (typically a training session or hand feeding demonstration in front of guests).

100

Off Grounds Animal is transported off zoo grounds for a period of time, typically for outreach programs.

B. Exercise Category Lure Course Mechanical system of lures and pulleys that simulate the movement of live prey. Cheetahs are trained to chase the lure object through the course for exercise and public demonstration and have reached speeds of up to 60 mph. Leash Walk Cheetah walking outside of usual enclosure with a caretaker for the purpose of exercise. Other Exercise Other keeper-directed physical activity. Examples: chasing a ball or other object, chasing keepers located outside of the enclosure, running with passing vehicles.

C. Enrichment Category Food Enrichment that contains some type of edible item. Also includes bones and small whole prey (rabbits, quails) given as part of the diet once a week or less frequently. Object Novel physical object (ball, box, barrel, etc.). May be used in conjunction with food or sensory categories Sensory Novel olfactory or tactile stimulation (bedding/fur from other species or conspecifics, spices, perfumes, bubbles, snow) Housing Change to housing environment (new furniture, bedding, climbing/scratching structures). Can also include rotation to a novel yard for a short time, if it does not happen daily.

Personality Assessment

A personality assessment was also completed for each cheetah, and the complete methodology and results have been described in the previous chapter. The resulting factor scores for each of the four personality components as well as the categorical “dominant personality type” (Tense-Insecure, Aggressive (to people), Playful-Friendly (to people),

101 and Social (with conspecifics)) for each individual cheetah were included in the current analyses.

Fecal Collection and Hormone Extraction

Naturally voided feces were opportunistically collected approximately every other day and placed in Whirlpak bags labeled with institution, animal ID, and date. Samples were stored at -20oC immediately after collection until project completion and then shipped overnight on ice to Cleveland Metroparks Zoo (CMZ) where they remained frozen until extraction.

Due to variability in fecal water content, fecal samples were lyophilized

(FreeZone, Model: #7751020, Labconco Corporation, Kansas City, MO, USA). Samples from four institutions were lyophilized prior to shipment to CMZ. Dried fecal samples were manually crushed prior to extraction using a technique adapted from Brown (2008).

Briefly, ~0.2g fecal powder was placed in 20 mL glass tubes with 5 mL 80% methanol.

Tubes were loaded onto a mixer (099A LC10102, Glas-Col, Terre Haute, IN, USA) and agitated for 1 hour. Samples were then centrifuged (2500 rpm, 20 min), the supernatants collected into a new set of 5 mL plastic tubes and stored at until assayed. A subset of samples were extracted several times and tested for loss of sample integrity, demonstrating that sample extracts remained viable for up to a year when stored at -80oC.

Thus, all samples were assayed within one year of extraction.

102

Fecal Glucocorticoid Metabolite (FGM) Analysis

Laboratory validation was achieved by demonstrating a parallelism between binding of serial dilutions of fecal extracts and the standard curve. Average recovery of serum samples spiked with high and low corticosterone controls was 108% and concentrations were calculated on an ng/g dry feces basis. Baseline concentrations for each individual were calculated using an iterative process described in (Brown et al.,

1999).

FGM were measured using a corticosterone-2-CMO enzyme immunoassay (EIA;

Leibniz Institute for Zoo and Wildlife Research, Berlin, Germany) that has been previously validated in cheetahs (Ludwig et al., 2013). Fecal extracts were diluted 1:10 with 30% methanol: water. Inter-and intra- coefficients of variation (CVs) were less than

10%.

Behavioral Data Collection

Focal animal behavioral observations were conducted in 30-min sessions using an instantaneous point sampling technique at one-minute intervals (Martin et al., 1993).

During each scan, the focal animal’s activity level and behavior were recorded using a mutually exclusive and exhaustive ethogram (Table 4.3). A minimum of ten morning

(0700-1000), ten midday (1000-1300), and ten afternoon (1300-1600) observations were conducted for each animal, for a total of at least thirty, 30-minute observations throughout the 90-day study period. A visual ethogram and behavioral data collection training video was developed and mailed to each participating institution, using a similar protocol to that described in Baird et al. (2016). Inter-observer reliability was tested using

103 one of two recorded 30-min observations that potential observers scored. Data sheets from this reliability test were scored by the principal investigator (PI), and observers that achieved greater than 90% reliability with the PI were permitted to collect data for this study.

Table 4.3: Ethogram used for behavioral data collection in cheetahs. During each one-minute scan, one behavioral category was recorded from the Activity Level and Behavior groupings.

Activity Level Mobile Subject is moving (walking or running) resulting in a transfer of physical position in space that is at least or more than its own body length in distance. Immobile Subject is maintaining a position in space that consists of – or is less than double in area than – its own body length. Activity Not The activity level of the subject is not visible or discernible. Visible

Behavior Undesirable Any behavior that is repetitive, and could be considered abnormal compared to the behavior of a wild cheetah. Often scored when the subject is pacing, defined as repetitive locomotory movement along a given route (up/down fence line, around enclosure, or object in enclosure) uninterrupted by other behaviors. Subject must trace the same exact path three times for UND to be scored, otherwise score as LOCO. Other examples of UND behavior include repetitive fence/object licking/biting and head rolling. Exploratory* Subject is interacting with items in the environment including trees, rocks, substrate, enrichment items, etc. Subject can be engaging in various behaviors directed at the environment, including scratching, climbing, rubbing on, sniffing (including Flehmen response), or scent marking. Also includes solitary play, defined as non-serious behavior directed at self or objects. Social Affiliative* Subject is positively interacting with a conspecific. Includes social grooming, head-rubbing (subject rubbing partner’s face with own),

104

following (one subject following partner in close proximity), and social play (engaging in non-serious behavior with partner). Social Agonistic* Subject is interacting with a conspecific in an aggressive or threatening manner. Behaviors exhibited can include growling, hissing, striking with paws, biting, or attacking. Human-Directed Subject is actively attending to or interacting with staff, visitors, or observer. This behavior can be affiliative (purring, rubbing, play, etc.), neutral (staring at or engaging in vigilance behavior obviously directed at humans), or agonistic (growl, hiss, threaten, etc.). Locomotion* Subject is moving (walking or running) resulting in a transfer of physical position in space that is at least or more than its own body length in distance, and engaged in no other behaviors simultaneously. Predatory* Subject is crouching, stalking, chasing, or engaging in hunting behavior. Maintenance* Subject is engaged in general maintenance behaviors including autogroom, feed/drink, or urinate/defecate. Alert* Subject disengages from all other activities with eyes open and aware/vigilant of surroundings. Subject can be sitting or lying down. Other Active* Subject is engaged in some active behavior not otherwise described. Rest* Subject is lying down, not vigilant. Eyes can be open or closed.

Behavior Not The behavior of the subject is not visible or discernible. Visible *Behaviors that were included in calculations of behavioral diversity.

105

Statistical Analysis

FGM analysis

Three female cheetahs from a single institution were pregnant during the 90-day study period and gave birth during or shortly after data collection. These three females also had considerably higher FGM concentrations (more than four times) than all other individuals. Therefore, these three individuals were not included in FGM analysis, resulting in a final sample size of 68 (39.29) cheetahs in all FGM models. Frequency of unfamiliar people present during human interactions was often confounded with management role, as ambassador animals interact with unfamiliar people more often than cheetahs in other roles. For cheetahs in non-ambassador roles, presence of unfamiliar people during interactions occurred too infrequently for meaningful conclusions, and so this variable was removed from analysis. The keeper rating of cheetah interest level recorded for each instance of interaction, exercise, and enrichment was also removed from analysis, as there was little variation in ratings across all 7628 events (nearly all events were rated as a 4 or 5 on the 5-point scale).

FGM concentrations were log transformed to approximate a normal distribution and then analyzed using a general linear mixed model (PROC MIXED; SAS University v. 3.6). Institution and animal ID were included as random intercepts in each model to control for similarity between samples from the same animal and from the same institution. Month of data collection was also included as a random slope in each model.

Based on -2 log likelihood (-2LL), Akaike’s information criterion (AIC), and Bayesian information criterion (BIC), an unstructured covariance structure was chosen for all models.

106

All model building occurred using maximum likelihood estimation utilizing a step-wise process where non-significant fixed factors (p > 0.10) were removed until the lowest -2LL, AIC, and BIC were achieved. Due to the number of variables measured in this study, it was necessary to first include fixed factors in small groups so that each variable and interaction could be analyzed with sufficient power. Fixed factors analyzed and initial groupings are described in Table 4.4. Age and sex were controlled for as fixed factors in all models and variables that were confounding were never included in the same model (e.g. the numeric and categorical representations of social grouping and personality, or frequency and duration of human interaction).

Final models were obtained using a restricted maximum likelihood estimation and chosen based on the lowest -2LL, AIC, and BIC values. Residuals were assessed for normality in each final model and Cook’s Distance tests indicated no significant outliers.

Degrees of freedom were estimated using Satterthwaite’s approximation, and when significant effects were determined (p < 0.05), post hoc tests for multiple comparisons were conducted using the Tukey-Kramer adjustment. Statistically analyzed FGM results are presented as back-transformed means and 95% confidence interval (CI) unless otherwise noted.

107

Table 4.4: Fixed factors analyzed in relation to FGM and behavior using linear mixed models. Factors were initially included in the model in small groups, and non-significant variables (p>0.10) were removed from further analysis. Factors that remained from each grouping were then included together and the step-wise process was repeated to obtain the final model with the lowest -2 log likelihood, AIC, and BIC. Variables that were confounding were never included in the same statistical model (i.e. different representations of the same construct).

Fixed Factor Description Individual Variables Year* Year of Study (1 = 2014, 2 = 2015) Age* Age of animal (years) Sex* Sex of animal (male, female) Hand Rearing* Was the animal hand reared? (Yes, No) Breeding* Is the animal currently in a breeding situation? (Yes, No)

Housing & Husbandry Variables Space Experience* Amount of space animal has access to during study. Size of each space (in sq. ft) weighed by proportion of time given access to that space during the 90-day study period. Number of Spaces* Number of different housing spaces the animal is regularly moved between Contact Type* Do humans ever share space with the animal (free contact) or is there always a barrier between humans and cheetahs (protected contact)? Social Group Type* How is the animal managed socially? (social, solitary) Social Group Size* Size of animal’s social group (1, 2, 3 or more) Visual Access to Does animal have visual access to other cheetahs that it is not Cheetahs* housed with? (Yes/No) Visual Access to Other What type of species does the animal have visual access to? Species* Prey (hoofstock, birds, suidae, etc.), Carnivore (predator or competitor species such as felids or canids), Both (both potential prey or potential predator/competition species), None (no visual access to other species). Feeding Schedule* Daily feeding schedule (predictable, unpredictable)

Human Interaction/Enrichment Interaction Frequency Number of human interaction events (per week) Interaction Duration Total duration of human interaction (per week) Training Frequency Number of training sessions (per week) Program/Outreach Number of programs/outreach/keeper talks/shows (per week) Frequency Exercise Frequency Number of human-directed exercise events (per week)

108

Exercise Duration Total duration of human-directed exercise (per week) Lure Course Does the animal ever have the opportunity to run on a lure Participation* course? (Yes, No) Lure Frequency Number of lure course events (per week) Enrichment Frequency Number of times enrichment was provided (per week)

Management Category Primary Management Management role that the animal primarily participates in Role* (Ambassador, Exhibit, Off-exhibit) Ambassador Is this animal ever utilized as an ambassador? (Yes, No) Exhibit Is this animal ever housed on public exhibit? (Yes, No) Off-Exhibit Is this animal ever housed in an off-exhibit breeding facility? (Yes, No)

Personality Dominant Personality* Personality component with the highest positive numerical factor score. (Tense-Insecure, Aggressive, Playful-Friendly, Social) Personality Component 1* Component 1 from PCA analysis – Tense-Insecure. Range: -3 – 3 Personality Component 2* Component 2 from PCA analysis – Aggressive (to people). Range: -3 – 3 Personality Component 3* Component 3 from PCA analysis – Playful-Friendly (to people). Range: -3– 3 Personality Component 4* Component 4 from PCA analysis – Social (with other cheetahs). Range: -3 - 3 *Factors that were examined in relation to behavioral diversity scores using a general linear model.

Behavior Analysis

The behaviors Social Affiliative, Social Agonistic, Predatory, Maintenance, and

Other Active were recorded in less than 5% of observation sessions (out of 1400) and were not central to our current questions, so were not included in this analysis.

Undesirable behavior was also rare, occurring in less than 1% of all scans (out of 41933), but it was included because it is a common indicator of welfare. In addition, Mobile, but not Locomotion, was included in this analysis as the former variable is a more inclusive

109 measure of activity level than the latter. All behaviors were analyzed using a generalized linear mixed model (PROC GLIMMIX, SAS University v. 3.6) following the same model building procedure described above with the following exceptions. A negative binomial distribution with a log link function was specified for the behaviors Mobile,

Exploratory, Human-directed, and Alert because data were not normally distributed. For the behaviors Undesirable (a rare behavior) and Rest (an extremely common behavior), a logistic regression was utilized, specifying the count data in an “events out of trials” format using a binomial distribution and a logit link function. Analysis of random effects indicated no variance due to institution, and the inclusion of this factor as a second random intercept did not improve model fit. Therefore, in all models, only animal ID was included as a random intercept with a variance components covariance structure. All models also included an offset term to adjust for number of scans visible in a given observation and degrees of freedom were estimated using a between-within method.

Age, sex, and observation time (morning, midday, and afternoon) were included in all models, even when non-significant, unless their removal resulted in a significantly better model fit. Fixed factors analyzed were the same as those used in the FGM models

(Table 4.4) with the following exceptions. Only the primary management role variable was used to examine differences between roles. In the subset of cheetahs where behavioral data was collected (n=42 at 10 institutions), there were no animals that spent time in both exhibit and off-exhibit roles, and all ambassador cheetahs also spent time on exhibit, making it unnecessary to examine these roles separately. Further, primary management role and hand rearing were somewhat, but not completely, confounded. All ambassador cheetahs (n=10) were hand reared, and none of the off-exhibit cheetahs

110

(n=19) were hand reared. The exhibit category (n=13) contained both hand reared (n=5) and mother reared (n=8) individuals. As both of these factors are important to our central questions, both factors were included together in all models to tease out effects of rearing history and current management role. All results reported for either primary management role or hand rearing are significant in the model after controlling for the other.

Behavioral Diversity

For each 30-minute observation, behavioral diversity was calculated using the

Shannon-Wiener diversity index (Shannon & Weaver, 1949). The index (H) was calculated using the following equation where p is the proportion of time engaged in the ith behavior:

(퐻) = − ∑(푝푖 ln 푝푖)

With this equation, a greater number of behaviors observed or a more even distribution among all possible behaviors acts to increase H, indicating higher levels of behavioral diversity. Behaviors chosen for inclusion in the index were all species-appropriate behaviors, and thus did not include Undesirable and Human-Directed behaviors (Table

4.3). Behavioral diversity scores were averaged across the entire study period, creating a single H score for each cheetah.

A General Linear Model (GLM) was used to test the effects of various fixed factors on behavioral diversity scores. As the behavioral diversity index provides a single measurement per cheetah, only factors with constant values throughout the study period were analyzed. Averaging human interaction, exercise, or enrichment variables across the

111 entire study period would have removed all variation that an animal experiences from day to day, and thus would not have provided a good representation of these constructs for analysis. Further, only primary management role was utilized to examine differences between roles in this subset of cheetahs, as described above. Variables included as fixed factors in these GLM models are noted in Table 4.4.

To investigate the relationship between behavioral diversity and FGM concentrations, a separate GLM was constructed. In this case, FGM values were averaged across the entire study period so that results could be compared to a previous study relating FGM values and behavioral diversity scores in cheetahs (Miller et al., 2016).

However, unlike the previous investigation, age and sex were controlled for in the model and baseline FGM concentrations were also calculated for each individual using an iterative method described in (Brown et al., 1999). Baseline concentrations are an established method of characterizing FGM over a period of time in many species

(Koester et al., 2015; Young et al., 2004), as this iterative technique helps to “normalize” fluctuating hormone data by systematically removing extreme values within an individual.

Both average and baseline FGM concentration variables were log transformed prior to analysis in order to approximate a normal distribution. Outlier analysis indicated a single female cheetah with significantly higher FGM concentrations than the rest of the individuals (average FGM 15% higher than next highest cat, and baseline FGM 26% higher than next highest cat). This animal was removed from this particular analysis. In addition, behavioral data, but not fecal samples were collected from two other

112 individuals, resulting in the inclusion of 39 (26.13) cheetahs in the analysis of FGM and behavioral diversity.

Results

Factors Associated with FGM

A majority of factors analyzed were not significantly associated with FGM concentrations, including primary management role. There were no differences between cheetahs primarily managed in ambassador, exhibit, or off-exhibit roles (F(2, 52) = 0.32, p

= 0.73; Figure 4.1A). Likewise, there were no differences in FGM between cheetahs that spent time in an ambassador role during the study and those that did not (F(1,34) = 0.38, p

= 0.54; Figure 4.1B) and between cheetahs that were in an off-exhibit role at any time during the study and those that were not (F(1,55) = 0.12, p = 0.73; Figure 4.1D). However, there was a significant difference between cheetahs that were housed on public exhibit during the study (Mean = 211.69, 95% CI: 181.13-247.34) and those that never were

(Mean = 139.32, 95% CI: 114.95-168.85; F(1,26) = 24.20, p <0.001; Figure 4.1C).

In addition, cheetahs managed in free contact (n = 54) had higher FGM than those managed in protected contact (n = 14; F(1, 9) = 12.58, p = 0.006). Though this factor was highly significant in the full model, only cheetahs that were regularly housed on exhibit were managed in protected contact. When comparing the 21 cheetahs housed exclusively in an exhibit role, those managed in free contact (n = 11) still had higher FGM than those in protected contact (n = 10; F(1, 4) = 9.86, p = 0.04). All cheetahs primarily in ambassador and off-exhibit roles are managed in free contact, so no comparisons could be made within those roles.

113

Figure 4.1: Least-square mean (± 95% CI) FGM concentrations for cheetahs in different management roles. A) compares cheetahs that primarily fill ambassador (n=17), exhibit (n=21), and off-exhibit (n=30) roles. Because many cheetahs filled multiple roles during the study period, each role was analyzed separately by comparing cheetahs that B) participated in an ambassador animal role (n=20) to those that did not (n=48), C) cheetahs that spent any time housed on public exhibit (n=35) to those that never did (n=33), and D) cheetahs that were housed in an off-exhibit area during the study (n=30) to those that never were (n=38). Within comparisons, different superscripts demote significant differences between groups. A significant difference between groups was only found in the on-exhibit condition (F test, p<0.001).

Though there was no overall association between FGM and dominant personality type, there was a significant interaction between dominant personality and whether the cheetah was housed on exhibit or not (F(3,32) = 7.10, p = 0.009, Figure 4.2). Cheetahs that are primarily Tense-Insecure and housed on exhibit had significantly higher FGM than

114 nearly every other combination of personality and exhibit housing (on exhibit:

Aggressive, off-exhibit: Tense-Insecure, Aggressive, Social; p < 0.05). There was also a trend for Tense-Insecure cheetahs housed on exhibit to have higher FGM than Playful-

Friendly cheetahs housed on exhibit (p < 0.1). Finally, Social cheetahs housed on exhibit had higher FGM than Social cheetahs that were not housed on exhibit (p = 0.004).

115

Figure 4.2: Interaction between dominant personality and on-exhibit housing. Least-square mean (± 95% CI) FGM concentrations for cheetahs that were housed on exhibit during the study period and those that were not, separated by dominant personality type: Tense=-Insecure (T/I), Aggressive (Agg), Playful-Friendly (P/F), and Social (Soc). Sample sizes for each group are presented and different superscripts indicate significant differences between groups across all conditions (F test; p<0.05). Cheetahs that were predominantly T/I and housed on exhibit had significantly higher FGM concentrations than Agg cheetahs on exhibit and T/I, Agg, and Soc cheetahs that were never on exhibit. In addition, Social cheetahs housed on exhibit had significantly higher FGM concentrations than Soc cheetahs that were not on exhibit.

116

Factors Related to Behavior

In general, cheetahs were more active during morning observations (0700-1000).

More Mobile, Exploratory, Alert, and less Rest behavior were observed in the morning (F tests, p < 0.001 in all models). Exhibit cheetahs were observed to Rest less than both ambassador and off-exhibit animals (F(2,78) = 3.32, p = 0.037). Cheetahs housed off- exhibit and those that were mother-reared engaged in less Alert behavior compared to other management roles and hand-reared cheetahs (management role: F(2,76) = 3.41, p =

0.033; rearing: F(1,34) = 5.27, p = 0.028). After controlling for management role, hand- reared cheetahs engaged in more Human-Directed behavior (F(1,32) = 7.61, p = 0.006).

Hand-reared cheetahs were also 18 times more likely to perform Undesirable behavior than those that were mother-reared (F(1,35) = 7.57, p = 0.009).

Cheetahs that had visual access to other cheetahs that they were not housed with spent more time Mobile (F(1,38) = 21.25, p < 0.001) and less time engaging in Alert behavior (F(1,34) = 9.12, p = 0.005). Finally, cheetahs with a predictable feeding schedule were less Mobile (F(1,38) = 5.65, p = 0.023) and performed more Alert behavior (F(1,34) =

12.35, p = 0.001).

Behavioral Diversity

Behavioral diversity scores ranged from 0.43 – 0.89 (Mean = 0.64 ± 0.11). There was a significant negative correlation between behavioral diversity scores and space experience; cheetahs in larger enclosures demonstrated decreased behavioral diversity

(F(1,33) = 4.65, p = 0.04). However, it should be noted that this effect may be driven by institutional differences. All cheetahs within an institution have similar space experience

117 scores, and the institution with the largest enclosures also had the lowest behavioral diversity scores. When this institution is removed from analysis, space experience is no longer a significant factor, but the sample size is also reduced by nearly a third.

Therefore, space experience was controlled for in the remaining models but this result should be interpreted with caution. Primary management role was significantly associated with behavioral diversity scores (F(2,33) = 5.95, p = 0.006), with cheetahs in an off-exhibit role (n=19) exhibiting lower behavioral diversity than those in exhibit (n=13) or ambassador (n = 10) roles (Fig 4.3).

Figure 4.3: Box plots representing the relationship between primary management role and behavioral diversity. Cheetahs housed off-exhibit (n=19) had significantly lower scores on the Shannon-Wiener diversity index than those in ambassador (n=10) or exhibit (n=13) roles (GLM, p = 0.006).

118

Average FGM concentrations ranged from 93.34 – 698.93 (Mean = 240.18 ±

148.84). Age and sex were significantly associated with average FGM concentrations, with older cats and females having higher FGM (age: F(1,35) = 10.14, p = 0.003; sex: F(1,35)

= 6.40, p = 0.016). There was a positive correlation between average FGM concentrations and behavioral diversity scores (F(1,35) = 11.59, p = 0.002, Fig 4.4A). Baseline FGM concentrations ranged from 70.50 – 512.65 (Mean = 196.66 ± 122.42). The same significant associations were observed between baseline FGM concentrations and age, sex, and behavioral diversity (p < 0.01 in all models; Fig 4.4B).

119

Figure 4.4: Scatter plots of behavioral diversity scores and A) average and B) baseline FGM concentrations for 39 cheetahs, separated by sex. There was a significant positive correlation between behavioral diversity and both average and baseline FGM concentrations after controlling for age and sex (GLM, p < 0.01 in all models).

120

Discussion

This study aimed to evaluate how environmental factors and individual personality are related to FGM and behavior in zoo-housed cheetahs. Though several previous studies have investigated factors related to glucocorticoids (Koester et al., 2015,

2017; A. Wells et al., 2004; Wielebnowski, Ziegler, et al., 2002) and behavior

(Chadwick, 2014; Quirke & O’Riordan, 2015; Quirke et al., 2012; Wielebnowski, 1999) in zoo-housed cheetahs, this is the first to include ambassador animals. Although we did not observe any differences in FGM or behavior between ambassador and non- ambassador cheetahs, we did find cheetahs housed on public exhibit had significantly higher FGM concentrations and rested less than those that were never housed on exhibit.

Additionally, there was an interaction between personality and on-exhibit housing, and a significant effect of free vs. protected contact management on FGM concentrations.

Protected contact management only occurred in exhibit animals here, so these results taken together suggest that cheetahs in an exhibit role may be an important area of focus when developing management recommendations to ensure optimal animal welfare.

Differences between Management Roles

The primary finding that there was no difference in FGM or undesirable behavior between cheetahs in ambassador, exhibit, or off exhibit roles echoes the results of our previous study in armadillos (Baird et al., 2016). Even when considering the fact that cheetahs are often rotated between roles, there were still no differences in FGM between ambassadors and non-ambassadors. In fact, the only differences related to management role were that cheetahs who spent time on public exhibit had significantly higher FGM

121 concentrations and rested less than cheetahs that were never on exhibit. This finding contradicts that of Koester et al. (2015, 2017) who did not find any differences in FGM concentrations or behavior between cheetahs housed on- and off-exhibit. However, those studies utilized a cortisol-3-CMO EIA to evaluate glucocorticoids, which has been shown to be less sensitive to adrenal output in cheetahs than the assay used here (Ludwig et al.,

2013). The difference in assay sensitivity and behavioral methodology combined with a smaller sample size may have inhibited detection of on- and off-exhibit differences in the previous investigations.

In the wild, cheetahs have large home ranges and avoid human contact (Caro,

1994), so it may not be surprising that a typical zoo exhibit environment may be related to physiological stress in this species. Wells et al. (2004) demonstrated an increased risk for prolonged FGM elevation in cheetahs that were transferred from off-exhibit to on- exhibit facilities, which they suggested may be indicative of a chronic stress response in these individuals. Housing on public exhibit has also been associated with elevated FGM concentrations in clouded leopards, another solitary and elusive felid species

(Wielebnowski, Fletchall, et al., 2002). If being housed on public display is associated with physiological stress in cheetahs and other elusive cats, then it follows that this effect may be compounded by a free-contact management style where caregivers share space with the animal. Our study did find that cheetahs managed in free contact had significantly higher FGM than those managed in protected contact. However, there were several important caveats to this result. First, only 14 of the 71 cheetahs in this study were managed in protected contact, and 10 of those cats were managed exclusively on exhibit. The other four protected contact cheetahs were listed as primarily “off- exhibit”

122 but were not breeding and were repeatedly rotated to the public exhibit area. All of the cheetahs housed at breeding facilities or utilized as ambassador animals are managed exclusively in free contact, so no FGM comparisons could be made within those roles.

However, this difference in FGM was still observed when the analysis was limited to the

21 cheetahs housed exclusively on exhibit.

Taken together, these results seem to lead to more questions than answers. If exposure to people is associated with increased physiological stress in cheetahs, then ambassador animals would be expected to have extremely high FGM values, because they experience very close, often hands-on, contact with a wide variety of people on a daily basis. Likewise, if free contact management is associated with increased stress, then both ambassador and off-exhibit cheetahs would be expected to have higher FGM than our sample of exhibit animals, since half of these are managed in protected contact.

However, there were no differences in FGM between these three primary roles and no differences were seen when ambassador and off-exhibit roles were examined separately.

Perhaps the answer lies in the degree of exposure to humans and the daily experiences of cheetahs in each role. Ambassadors experience considerably more human interaction than cheetahs in other roles; an average of 23 events per week compared with

6 per week for exhibit animals and 2 per week for off exhibit animals. However, skilled animal trainers have worked to prepare these ambassadors for this role by beginning desensitization at an early age. Guidelines for hand rearing and training ambassador cheetahs recommend forming a bond with the cub and beginning the habituation process before 3 weeks of age (Rapp et al., 2017). In our personality analysis (Chapter Three), we observed hand reared cheetahs to be generally more Playful-Friendly toward people and

123 less Aggressive, and hand-reared cheetahs engaged in more human-directed behavior in the current analysis. There is evidence from other species that early habituation and desensitization can reduce the stress response to novel stimuli (Hargreaves & Hutson,

1990; Meaney et al., 1991; Meerlo, Horvath, Nagy, Bohus, & Koolhaas, 1999; Núñez et al., 1996; Pedersen, 1994; Podberscek et al., 1991). The intensive training and hand rearing protocol employed for ambassador cheetahs may similarly mitigate their FGM response to humans and other environmental stimuli. However, after controlling for management role, hand-reared cheetahs were also observed to perform more undesirable behavior than those that were mother-reared, though the behavior occurred less than 2% of time in this group on average. Increased stereotypic behavior has also been reported in hand-reared primates (Lutz et al., 2003; Meder, 1989) and some hand-reared felids have been noted to be extremely aggressive (Mellen, 1992; Sellinger & Ha, 2005). Animal managers caring for hand-reared cheetahs in any role should be aware of the increased risk for undesirable behavior in these individuals and be prepared with strategies for mitigating the behavior should it arise.

Cheetahs housed in off-exhibit breeding facilities generally do not receive any intensive habituation or desensitization training, and management occurs in free contact.

However, direct human-cheetah interactions occur infrequently because routine feeding and enrichment provision often does not require caretakers to enter the enclosure with the animal. In livestock, human avoidance distance is routinely used to measure an animal’s fear of humans which can, in turn, relate to their welfare (Mattiello et al., 2010;

Waiblinger et al., 2006; Windschnurer, Boivin, & Waiblinger, 2009). Providing opportunities for animals to control their degree of exposure to humans has been shown

124 to be an effective means of reducing indicators of negative welfare in a variety of species.

Petting zoo animals perform less undesirable behavior when given access to a retreat space (Anderson et al., 2002) and visual barriers to the public have been shown to reduce undesirable behavior in zoo-housed primate species (Kuhar, 2008; Sherwen, Harvey, et al., 2015; Stoinski et al., 2012), though individual differences in response were noted in all studies. The larger enclosures at cheetah breeding facilities may allow for animals that are not accustomed to regular human interactions the ability to regulate their own avoidance distance when keepers do share their space. More intensive free contact management of off-exhibit cheetahs typically only occurs during periods of active breeding, when keepers will take advantage of this avoidance distance by using their bodies to encourage cats to move between enclosures for pairing. However, this type of management occurs infrequently, maybe only for a few weeks a year if the animal has a breeding recommendation.

Exhibit cheetahs also do not typically receive the intensive desensitization training of ambassadors, and they are in view of the public daily. No differences in FGM or behavior between the three primary management roles were observed, likely due to the fact that many cheetahs serve in multiple roles. However, when examined separately, cheetahs that spent time housed on public exhibit had higher FGM concentrations and spent less time resting. Free contact management may be an additional source of physiological stress in these exhibit animals, as evidenced by significantly lower FGM in the ten cats managed in protected contact. It is possible that these results have to do with the degree of human exposure, specifically whether the animal can adequately control their human avoidance distance, both in relation to the public and to caretakers that enter

125 their enclosures for husbandry purposes. Carlstead, Fraser, Bennett, & Kleiman (1999) found that a higher percentage of public access around the perimeter of the enclosure was associated with increased fear behaviors and mortality in zoo-housed black rhinos. There may be a similar effect happening here, and investigating specific aspects of cheetah exhibits, such as the percentage of public access, number of hiding spaces, and human avoidance distance in relation to indicators of welfare would be an important next step for research. Although there seems to be a general pattern of association emerging between welfare indicators and on-exhibit housing in cheetahs and other felids, it is reasonable to question whether this pattern applies to all individuals of the species or if individual differences in adrenal activity are indicators of a more complex relationship.

When individual personality is taken into consideration, another layer of explanation emerges. Cheetahs that were predominantly Tense-Insecure and housed on exhibit had significantly higher FGM than nearly every other combination of personality type and on-exhibit housing. Wielebnowski (1999) suggested that cheetahs that score high on the “tense-fearful” dimension may have more difficulty coping with the captive environment than those with lower scores and recommended housing these individuals in enclosures that are more secluded. Our findings appear to confirm that recommendation.

Compounding a tense/insecure/fearful personality type with additional human factors such as public exposure and/or free contact management may exacerbate the physiological stress response in these individuals. Unfortunately, there were too few cheetahs managed in protected contact in the current study to relate personality type to this specific factor, but it would be an appropriate avenue for future investigation.

126

Other Factors Related to Behavior

Behavioral analysis also revealed relationships between activity levels and the husbandry factors of visual access to other cheetahs and predictability of feeding schedule. Cheetahs that could see other cheetahs in adjacent enclosures were more

Mobile and less Alert, while cheetahs that were fed on a predictable schedule exhibited the opposite pattern. Quirke et al. (2012) found that cheetahs with visual access to conspecifics were more likely to pace, defined as “repetitive locomotory movement along a given route”. Our definition of undesirable behavior was more conservative than the one used in the previous study because we required three full repetitions of the pattern to be observed before pacing could be scored (characteristics validated by Cless, Voss-

Hoynes, Ritzmann, & Lukas, 2015). However, our Mobile variable included all forms of locomotor activity, both pacing and normal locomotion, so regardless of the specific definition of pacing, it appears that both studies support the idea that visual access to other cheetahs is associated with increased locomotor activity of some kind. The mechanisms behind this behavior are still unclear, but they may be related to the need to delineate territories in the presence of unfamiliar individuals (Caro, 1994) or the social compatibility (or lack thereof) of individuals housed in adjacent enclosures (Quirke et al.,

2012; Wielebnowski, Ziegler, et al., 2002).

A more puzzling result is that cheetahs with a predictable feeding schedule are less active than those fed on an unpredictable schedule. Quirke et al. (2012) found that cheetahs with a predictable feeding schedule spent more time pacing, which contradicts the current findings, but is more in line with other discussions of predictable feeding schedules. Additional studies in cheetahs and other carnivores have found that temporal

127 variation in feeding is an effective form of enrichment for reducing stereotypic pacing

(tiger: Jenny & Schmid, 2002; red fox: Kistler, Hegglin, Würbel, & König, 2009; cheetah: Quirke & O’Riordan, 2011a, 2011b). In contrast, Wagman et al. (2018) investigated feeding enrichment in four species of bears and found that adding temporal variation did not reduce abnormal locomotor behavior any more than the enrichment itself did. However, the unpredictable enrichment schedule did increase exploratory behavior. Routines associated with husbandry practices can lead to the development of anticipatory behavior in zoo-housed animals, which often manifests as increased locomotor or route-tracing behavior leading up to a desired event, such as feeding

(Watters, 2014). It is possible that without a reliable signal indicating feeding time, zoo- housed animals spend more of their time engaging in exploratory and searching behaviors throughout the day, and thus are more active (Wagman et al., 2018). With a reliable and predictable signal, activity only increases in the short period of time immediately prior to feeding, resulting in less activity overall. As decreased activity is not necessarily an indicator of poor welfare, the behavioral benefits of temporal variation in feeding likely exceed any potential costs, and this practice should still be considered an effective form of enrichment.

Behavioral Diversity

The Shannon-Wiener diversity index has been increasingly utilized in zoo studies as an indicator of positive welfare ( Clark & Melfi, 2012; Collins, Quirke, Overy,

Flannery, & O’Riordan, 2016; Razal et al., 2016). It is logical to assume that an increase in species-typical behaviors is a good thing (Swaisgood, 2007), which makes this

128 quantitative measure of diversity attractive to researchers. However, previous studies have found mixed results rendering the implications of behavioral diversity scores unclear. Some investigations have noted a negative correlation between behavioral diversity and stereotypic behavior (Dantzer, 1986; Gunn & Morton, 1995), but others have found no such relationship (Vickery & Mason, 2004). One study focusing specifically on environmental enrichment in zoo-housed animals found behavioral diversity scores to increase following enrichment presentation (Shepherdson, Carlstead,

Mellen, & Seidensticker, 1993), while another found no change in diversity scores in several enrichment conditions ( Clark & Melfi, 2012). However, both studies concluded that the enrichment presented was generally effective. A previous study in cheetahs found a negative correlation between FGM and behavioral diversity scores (Miller et al., 2016), but the current study found a positive correlation. These mixed results make it difficult to understand the meaning of diversity scores in the larger context of animal welfare.

It may be that behavioral diversity scores themselves are best applied within a particular investigation, rather than comparing across studies, species, or situations. The

Shannon-Wiener diversity index is not only a measure of the number of different behaviors performed, but also the amount of time devoted them (Vickery & Mason,

2004). Thus, similar scores would be calculated when many different behaviors are observed in uneven frequencies and when a few different behaviors are observed in a more even distribution. Diversity calculations are also based on the ethogram and sampling methodology chosen, and thus possible minimum and maximum values will vary for each study (Collins et al., 2016). Miller et al. (2016) utilized an all-occurrence sampling method in short observation sessions and an ethogram consisting of very

129 specific social and exploratory behaviors (such as chase, object play, and tail flicking), while the current study utilized a point sampling method in longer observation periods and a more general ethogram that included resting, locomotion, and vigilance. Though both are valid approaches to studying behavior, diversity scores cannot be compared between studies due to the differences in methodology.

Focusing, then, on relationships within the current study, off exhibit animals exhibited lower behavioral diversity, even after controlling for space experience.

Cheetahs and other carnivores are known to spend a large portion of their activity budget resting (Siegel, 2005), so low behavioral diversity may not be unusual, depending on when observations took place. Frézard & Pape (2003) compared the behavior of six wolf packs at different zoological parks and found that the wolves in the largest enclosures spent more time resting than those in smaller enclosures. They suggested that these animals are better able to choose their resting periods because they are less disturbed by visitors and keepers. Though it is difficult to separate space experience and management role in the current sample, a similar explanation may apply to the off-exhibit cheetahs here. Cheetahs in this role also spent less time engaged in vigilance (Alert) behavior than those in ambassador or exhibit roles. Whether these animals have lower behavioral diversity due to enclosure size or to lack of disturbance by the public, it does not appear that this result indicates a welfare issue, especially given the observed relationship between FGM and behavioral diversity scores. In species that are typically inactive for large portions of the day or in individuals that may be more sensitive to the zoo environment, high behavioral diversity scores may actually be more of a concern when it comes to well-being. But, like any other indicator of welfare, careful validation of this

130 measure is needed to fully understand the implications of behavioral diversity scores and the extent to which they can be generalized across individuals, species, and situations.

Conclusions

Based on these results, we may want to first focus on the experience of exhibit cheetahs when making management recommendations. As most cheetahs in AZA zoos are now born and reared in off-exhibit facilities, nearly all exhibit cheetahs have little to no experience with human interaction and public exposure when they are first transferred to on-exhibit housing. It may be beneficial to implement a desensitization training program in off-exhibit facilities for individuals that are recommended to move to an exhibit role so they are better prepared to cope with their new environment. In addition, if there is not a specific need for free contact management (as there is for ambassador and breeding cheetahs), it may be better to utilize protected contact as much as possible, to avoid contributing to a chronic state of adrenal activation. When human-cheetah interaction is necessary, consciously utilizing techniques aimed at increasing choice and control for the animal would be advisable. Positive reinforcement training has been shown to reduce potential behavioral and physiological indicators of reduced welfare in a variety of species (Coleman, 2012; Dembiec, Snider, & Zanella, 2004; Laule &

Whittaker, 2007; Leeds, Elsner, & Lukas, 2016; Pomerantz & Terkel, 2009;

Shepherdson, Lewis, Carlstead, Bauman, & Perrin, 2013; Shyne & Block, 2010) and may similarly help to mitigate the physiological stress response when managing zoo-housed cheetahs. Recognizing and responding to individual differences in personality may also be a tool for improving welfare of this species. Cheetahs that appear to be more tense,

131 fearful, insecure, or excitable may not be thriving in an exhibit environment. If transfer to off-exhibit housing is not a feasible option, providing opportunities for these animals to choose to avoid or escape human interaction (e.g. hiding places, visual barriers, indoor access) should be considered (Carlstead, Brown, & Seidensticker, 1993; Wielebnowski,

Fletchall, et al., 2002). Developing a standardized personality assessment tool in cheetahs may prove to be a valuable population management tool for the SSP.

These findings add to the body of work supporting the idea that welfare is an individual experience. Information that is garnered from comprehensive multi- institutional studies, such as this one, can help us to further refine management practices so that recommendations can be tailored to the individual. Cheetahs are a species in which management practices are highly varied, there are large amounts of individual differences, and significant challenges to reproductive success and population sustainability exist. In addition, cheetahs are one of the few SSP managed carnivore populations that routinely allocates individuals into disparate exhibit, breeding, and ambassador roles. In this unique situation, having a deeper understanding of how all of these factors might be intertwined allows us to focus on matching individuals with environments that are the best fit for their well-being, which can lead to increased reproductive success, more impactful educational messaging, and better experiences for zoo visitors.

132

CHAPTER FIVE: General Discussion and Future Directions

Ambassador Animals

This project aimed to provide the first comprehensive evaluation of welfare in zoo-housed ambassador animals using a multi-institutional and multi-method approach.

Because only limited and anecdotal data existed on this topic previously, evaluating whether any group level differences exist between ambassador and non-ambassador animals was a first step in addressing this question. In both armadillos and cheetahs, there were no observed differences in behavioral or physiological indicators of welfare between management roles. Further, handling or human interaction specifically for educational programming had no association with the same indicators of welfare across all four taxa studied. Generally, these findings imply that ambassador animals, as a group, are not experiencing compromised welfare compared to their on-exhibit counterparts. For these taxa, serving in an ambassador animal role is not inherently better or worse for welfare than serving in an exhibit or off-exhibit role would be. Although we could not evaluate every aspect of the environment or incorporate every possible indicator of welfare, these results provide a solid foundation for future ambassador animal welfare research.

It is important to note that these findings should not necessarily be generalized across all species and all individuals. For example, though there were no differences between the ambassador and non-ambassador cheetahs studied here, independent investigations should be conducted to evaluate welfare in closely related species, such as servals. One of the main findings of both studies was the degree of individual differences

133 within species. Each of the 12 ambassador hedgehogs studied exhibited a different physiological response to human handling, and personality type played a significant role in cheetahs’ response to on-exhibit housing. Welfare is an individual experience, and it is important to remember that management recommendations are often not one-size-fits-all.

Handling

We did not see any associations between the human interaction variables and

FGM or behavior in cheetahs, though amount of handling was positively correlated with

FGM in armadillos, hedgehogs, and hawks and associated with several behaviors. This discrepancy may be explained by the differences in the way that small animals, such as armadillos, are handled and managed compared with large carnivores, such as cheetahs.

Positive reinforcement training (PRT) is typically utilized when managing larger animals, both for husbandry purposes (shifting, medical procedures) and ambassador programming. PRT has been shown to reduce negative indicators of welfare and may even be enriching (e.g. Coleman & Maier, 2010; Leeds, Elsner, & Lukas, 2016; Melfi,

2013; Pomerantz & Terkel, 2009). A hallmark of PRT is offering an animal the choice to participate, which allows them to exert some control over their environment. Choice and control have been shown to have benefits for welfare (reviewed in Swaisgood, 2007) and many zoo exhibits and husbandry practices are increasingly designed with these concepts in mind, allowing animals to choose between different habitats and/or control their degree of exposure to different environmental factors (sun/shade, conspecifics, humans).

While employing PRT is the most common (and often most practical) means of working with large or dangerous animals on a day to day basis, the same cannot be said for all

134 taxa. Smaller animals are usually just picked up from their enclosures during husbandry procedures or for education programs, removing the opportunity for the animal have a choice and exhibit control over its environment. The differing response of cheetahs and the smaller taxa in relation to human interaction and handling observed here may reinforce the idea that offering choice and control via PRT can have positive influences on welfare. This is certainly an idea that should be further investigated experimentally, especially in small, tractable species used in ambassador programs.

Our findings that armadillos and hawks handled more than 800 minutes per week experience significant increases in FGM concentrations suggest that there may be limits to the amount of handling that an animal can tolerate, beyond which welfare may be negatively affected. Although threshold limits are likely specific to an individual, experimental investigations to determine approximate thresholds within a species would be helpful for informing management practices. As a result of the preliminary findings presented here, several AZA zoos have already begun to implement protocols for tracking and limiting the amount of handling individual ambassador animals experience.

These tracking systems coupled with handlers that are attuned to the behavior of the animals they are working with can help to define thresholds for individual animals.

Understanding individual needs and adjusting practices accordingly can have substantial impacts on well-being.

Husbandry factors

The multi-institutional nature of these investigations allowed for a variety of housing and husbandry factors to be evaluated in relation to welfare indicators by taking

135 advantage of the inherent variation between zoos. Our findings indicate that these factors can have a significant effect on welfare, more so than management role. Armadillos and hedgehogs are burrowing species, so it makes sense that providing deep substrate would improve welfare by encouraging natural behaviors. Having data to confirm that a simple change in husbandry can have a major impact on welfare has already resulted in several institutions updating their substrate guidelines. These results are now informing best practices standards for armadillos and other digging species in all management roles. Our research also highlighted the general discrepancy between on and off-exhibit housing sizes and conditions. AZA accreditation standards are beginning to focus more on encouraging institutions to provide the same standards of care and housing throughout the zoo, whether the animal is on public display or not. Ambassador animal housing specifically is also being examined more closely. Hopefully, the results presented here will continue to be used to inform and improve best practice guidelines for zoo-housed animals across management roles.

Future of Ambassador Animal Research

This research demonstrates that ambassador animals (at least armadillos and cheetahs) do not experience compromised welfare when compared to their non- ambassador counterparts. However, this should not be taken as the end of ambassador animal welfare research, but only the beginning. Although there may not be differences in FGM or undesirable behavior between the ambassador and non-ambassador animals studied here, these are only two possible indicators of welfare. We do not yet have many reliable indicators of positive welfare at our disposal to evaluate if animals in any

136 management role are truly thriving, though this is an active area of investigation.

Oxytocin is one promising avenue of research that could be used to further evaluate the quality of human animal relationships across management roles, and ambassadors in particular. Evaluating oxytocin concentrations in relation to handling and human interaction factors in a large-scale investigation like this one might elucidate the overall effects of these variables on positive welfare. Alternatively, behavioral (activity level, anticipatory behavior), physical (heart/respiration rate), or physiological (FGM or oxytocin measured via saliva) indicators of welfare could be measured before, during, and after human interactions to provide a more specific evaluation of an individual animal’s perception of that experience. Incorporating additional indicators of individual well-being in future studies would help to inform more specific guidelines and recommendations for animals in any management role, including ambassadors.

In addition, the Ambassador animals of many species are often managed in ways that are contrary to their natural behavioral patterns. For example, nocturnal species are often handled and removed from their enclosures during daylight hours, which alters their natural sleep/wake cycles. Our findings indicated that ambassador armadillos were more active during our daytime observations than non-ambassadors. However, we don’t yet understand the implications of this altered sleep pattern or repeated exposure to light on the welfare of nocturnal animals, and additional research is needed. In addition, health should not be overlooked when evaluating an animal’s well-being. Many zoo-housed cheetahs across management roles still experience chronic health issues such as gastritis, and continuing to evaluate environmental factors to understand the root of these issues is critical for ensuring optimal welfare in the species as a whole. Although cheetahs have

137 been extensively studied already, many species utilized in ambassador programs have never been evaluated in any capacity and may also be experiencing chronic health conditions that we aren’t yet aware of. Basic morbidity and mortality analyses may be a good first step in evaluating the health status of these species, and provide a standard for comparing ambassador animals. It is also critical for future investigations to expand beyond mammals, as birds, reptiles, and amphibians comprise a sizeable portion of ambassador animal collections. This project was intended to provide a basis for future investigations and there is much more work to be done.

There is another aspect of ambassador animal programming that is equally important to the welfare question – messaging. I have discussed the small, but growing, body of literature that demonstrates positive cognitive and affective learning outcomes after an encounter with an ambassador animal (Heinrich & Birney, 1992; Miller et al.,

2013; Povey, 2002; Povey & Rios, 2002; Yerke & Burns, 1991, 1993). However, this research often focuses on general knowledge gains and attitude improvements, not on outcomes specific to the messages presented or species encountered. There is an emerging trend in the AZA community to grow their ambassador animal collections by adding new and varied species, which often translates to larger, uncommon, or more

“exciting” animals. The thought is that more dynamic ambassador animals will attract and hold more attention from visitors, which will then translate to more effective messaging. However, no research has been conducted to validate that claim. Is the use of an ambassador cheetah or hyena the most effective way to communicate messages about human-wildlife conflict, or can the same message be effectively conveyed with a domestic dog or sheep? The “wow factor” of an encounter with unusual ambassador

138 animal for a zoo visitor certainly should not be discounted, but more research is needed to understand the species-specific educational impacts of the growing variety of species used in programs.

Similarly, how can we ensure that our visitors are leaving an ambassador animal program with the messages we meant to convey, and that we are not creating unintended consequences? Does handling and close contact with ambassador animals inadvertently imply that these animals make good pets or that wild populations are not threatened, despite what we are saying? Ross, Vreeman, & Lonsdorf (2011) found that when chimpanzees are viewed with a human or in a human setting, respondents were more likely to consider chimpanzees to be an appealing pet and to believe that chimpanzee populations were stable. These findings were based on artificially generated images alone, but it is possible that these effects could be enhanced during a live demonstration where the animal is handled, wearing a leash and collar, or displaying behaviors similar to those seen in domestic cats and dogs (such as playing with a toy). Further, the exotic pet trade is a serious issue for many species, both from a conservation and a welfare perspective. Many ambassador programs incorporate messaging discussing the threat that the pet trade poses to wild populations and reinforcing the fact that exotic animals do not make good pets. However, research in the conservation psychology field suggests that this type of messaging strategy may be counterproductive. Humans tend to conform to social norms: both what other people are doing, and what is socially acceptable to do

(Cialdini, 2003; McKenzie-Mohr & Schultz, 2014). Highlighting the fact that many people have these animals as pets, but it is wrong to do, may be inadvertently pitting these two types of norms against each other (Cialdini, 2003). This dissonance, especially

139 coupled with the added dissonance of a person in close contact with the animal in question, may be reducing the effectiveness of messaging and leading to untended consequences by creating a “do as I say, not as I do” situation (McKenzie-Mohr &

Schultz, 2014). It should be emphasized that there is currently no research on this topic to support or refute these claims in an ambassador animal setting. But given the rapid growth of both the field of conservation psychology and ambassador animal collections, it is a worthy avenue for future investigation.

140

APPENDIX I Cheetah Demographic and Husbandry Survey

Institution: ______Animal ID#:______Animal Name (if applicable):______Studbook#: ______Date of Birth (or Age): ______Sex: ______Animal Information 1. Is this animal currently considered an Education, Exhibit, Off Exhibit breeding animal, or some combination of these management types? ______2. How long has this animal been housed in its current management type (in an education program, on exhibit, in a breeding facility)? ______3. Was this animal hand-reared? ______4. To your knowledge, has this animal ever been given the opportunity to reproduce? ______a. If so, have reproduction attempts resulted in offspring? ______5. Will this animal be in a breeding situation or potentially pregnant during the 90-day study period? ______

Housing and Husbandry Information 1. Does this animal have access to multiple enclosures (periodically shifted between spaces)? ______2. Please list each enclosure below and describe the size (in ft2), whether it is indoor or outdoor, the substrate type available, and how much time (estimated %) this animal typically spends in each area during the study period. Examples are provided. Enclosure Size In/Out/Both Substrate Estimated % Time Exhibit Yard 1700 sq ft Outdoor Grass, Dirt 90% Holding Area 480 sq ft Indoor Concrete 10% ______

141

3. Is this animal managed in a Protected Contact (a physical barrier separates human space from animal space at all times) or a Free Contact (at times, no barrier separates human and animal space) system? ______4. Is this animal housed alone or in a social group? ______5. If in a social group, what are the ages and sexes of the other group members? (can also provide Animal ID# if also completing surveys for other group members) ______6. Are the other group members related to this animal (siblings, mother and cubs, etc.)? ______7. Does this animal have visual access to other cheetahs that it is not housed with? ______a. If so, how many males and females? Males: ______Females: ______8. Does this animal have visual access to other species? ______a. If so, which species? ______9. What is this animal’s current diet? a. Brand ______b. Amount ______c. Number of feedings per day ______d. Are feedings predictable (in the same hour each day) or unpredictable? ______10. In the past year, has this animal received a physical exam that included a gastric biopsy? ______a. Would your institution be willing to share the results of that biopsy for the purposes of this investigation? ______b. If you are unsure, who would be the best person to contact at your institution with this request? ______

Is there any additional information that you would like to note about this animal? ______

142

APPENDIX II

143

144

APPENDIX III

145

REFERENCES

Abou-Ismail, U. A., Burman, O. H. P., Nicol, C. J., & Mendl, M. (2008). Let sleeping

rats lie: Does the timing of husbandry procedures affect laboratory rat behaviour,

physiology and welfare? Applied Animal Behaviour Science, 111(3), 329–341.

Ancona, K. A., & Loughry, W. J. (2009). Time budgets of wild nine-banded armadillos.

Southeastern Naturalist, 8(4), 587–598.

Ancona, K. A., & Loughry, W. J. (2010). Sources of variation in the time budgets of wild

nine-banded armadillos. Mammalia, 74(2), 127–134.

Andersen, L. L. (2003). Zoo education: from formal school programmes to exhibit design

and interpretation. International Zoo Yearbook, 38(1), 75–81.

Anderson, U. S., Benne, M., Bloomsmith, M. A., & Maple, T. L. (2002). Retreat space

and human visitor density moderate undesirable behavior in petting zoo animals.

Journal of Applied Animal Welfare Science, 5(2), 125–137.

AZA Animal Welfare Committee (2015). Definition of animal welfare. Association of

Zoos and Aquariums. USA.

AZA Conservation Education Committee (2003). Ambassador animal position statement.

Association of Zoos and Aquariums. USA.

Baird, B. A., Kuhar, C. W., Lukas, K. E., Amendolagine, L. A., Fuller, G. A., Nemet, J.,

… Schook, M. W. (2016). Program animal welfare: Using behavioral and

physiological measures to assess the well-being of animals used for education

programs in zoos. Applied Animal Behaviour Science, 176, 150–162.

146

Baker, K., & Pullen, P. K. (2013). The impact of housing and husbandry on the

personality of cheetah (Acinonyx jubatus). Journal of Zoo and Aquarium

Research, 1(1), 35–40.

Ballantyne, R., & Packer, J. (2005). Promoting environmentally sustainable attitudes and

behaviour through free‐choice learning experiences: what is the state of the game?

Environmental Education Research, 11(3), 281–295.

Ballantyne, R., Packer, J., Hughes, K., & Dierking, L. (2007). Conservation learning in

wildlife tourism settings: lessons from research in zoos and aquariums.

Environmental Education Research, 13(3), 367–383.

Barlow, C. J. C., Caldwell, C. A., & Lee, P. C. (2007). Individual differences and

response to visitors in zoo-housed Diana monkeys (Cercopithecus diana diana).

Presented at the 8th Annual Symposium on Zoo Research, Colchester Zoo.

Barnett, J. L., & Hemsworth, P. H. (1990). The validity of physiological and behavioural

measures of animal welfare. Applied Animal Behaviour Science, 25(1), 177–187.

Beck, B. B., & Power, M. L. (1988). Correlates of sexual and maternal competence in

captive gorillas. Zoo Biology, 7(4), 339–350.

Boissy, A., Manteuffel, G., Jensen, M. B., Moe, R. O., Spruijt, B., Keeling, L. J., …

Aubert, A. (2007). Assessment of positive emotions in animals to improve their

welfare. Physiology & Behavior, 92(3), 375–397.

Bolhuis, J. E., Schouten, W. G. P., Schrama, J. W., & Wiegant, V. M. (2005).

Behavioural development of pigs with different coping characteristics in barren

147

and substrate-enriched housing conditions. Applied Animal Behaviour Science,

93(3), 213–228.

Box, H. (1999). Studies of temperament in simian primates with implications for socially

mediated learning. International Journal of Comparative Psychology, 12(4).

Brando, S., & Buchanan-Smith, H. M. (2017). The 24/7 approach to promoting optimal

welfare for captive wild animals. Behavioural Processes.

Brann, D. W., & Mahesh, V. B. (1991). Role of corticosteroids in female reproduction.

The FASEB Journal, 5(12), 2691–2698.

Broom, D. M. (1986). Indicators of poor welfare. British Veterinary Journal, 142(6),

524–526.

Broom, D. M. (1991). Animal welfare: concepts and measurement. Journal of Animal

Science, 69(10), 4167–4175.

Broom, D. M. (2001). Coping, stress, and welfare. In Coping with challenge: Welfare in

animals including humans, Proceedings of Dahlem Conference. 1-9. Berlin:

Dahlem University Press.

Broom, D. M., & Johnson, K. G. (1993). Stress and Animal Welfare. Springer Science &

Business Media.

Brown, J. L. (2008). Wildlife Endocrinology Manual. Front Royal, VA: CRC Endocrine

Research Laboratory.

Brown, J. L., Schmitt, D. L., Bellem, A., Graham, L. H., & Lehnhardt, J. (1999).

Hormone secretion in the Asian elephant (Elephas maximus): Characterization of

148

ovulatory and anovulatory luteinizing hormone surges. Biology of Reproduction,

61(5), 1294–1299.

Brown, J. L., & Wildt, D. E. (1997). Assessing reproductive status in wild felids by

noninvasive faecal steroid monitoring. International Zoo Yearbook, 35(1), 173–

191.

Brown, J. L., Wildt, D. E., Wielebnowski, N., Goodrowe, K. L., Graham, L. H., Wells,

S., & Howard, J. G. (1996). Reproductive activity in captive female cheetahs

(Acinonyx jubatus) assessed by faecal steroids. Journal of Reproduction and

Fertility, 106(2), 337–346.

Carder, G., & Semple, S. (2008). Visitor effects on anxiety in two captive groups of

western lowland gorillas. Applied Animal Behaviour Science, 115(3), 211–220.

Carlstead, K., & Brown, J. L. (2005). Relationships between patterns of Fecal corticoid

excretion and behavior, reproduction, and environmental factors in captive black

(Diceros bicornis) and white (Ceratotherium simum) rhinoceros. Zoo Biology,

24(3), 215–232.

Carlstead, K., Brown, J. L., & Seidensticker, J. (1993). Behavioral and adrenocortical

responses to environmental changes in leopard cats (Felis bengalensis). Zoo

Biology, 12(4), 321–331.

Carlstead, K., Fraser, J., Bennett, C., & Kleiman, D. G. (1999). Black rhinoceros (Diceros

bicornis) in U.S. zoos: II. Behavior, breeding success, and mortality in relation to

housing facilities. Zoo Biology, 18(1), 35–52.

149

Carlstead, K., Mellen, J., & Kleiman, D. G. (1999). Black rhinoceros (Diceros bicornis)

in U.S. zoos: I. Individual behavior profiles and their relationship to breeding

success. Zoo Biology, 18(1), 17–34.

Caro, T. (1994). Cheetahs of the Serengeti Plains: Group Living in an Asocial Species.

University of Chicago Press.

Chadwick, C. (2014). Social behaviour and personality assessment as a tool for

improving the management of cheetahs (Acinonyx jubatus) in captivity (PhD).

University of Salford.

Choo, Y., Todd, P. A., & Li, D. (2011). Visitor effects on zoo orangutans in two novel,

naturalistic enclosures. Applied Animal Behaviour Science, 133(1), 78–86.

Cialdini, R. B. (2003). Crafting normative messages to protect the environment. Current

directions in psychological science, 12(4), 105–109.

Clark, F. E., & Melfi, V. A. (2012). Environmental enrichment for a mixed-species

nocturnal mammal exhibit. Zoo Biology, 31(4), 397–413.

Clark, W. K. (1951). Ecological life history of the armadillo in the Eastern Edwards

Plateau region. The American Midland Naturalist, 46(2), 337–358.

Claxton, A. M. (2011). The potential of the human–animal relationship as an

environmental enrichment for the welfare of zoo-housed animals. Applied Animal

Behaviour Science, 133(1), 1–10.

150

Clay, A. W., Bloomsmith, M. A., Marr, M. J., & Maple, T. L. (2009). Habituation and

desensitization as methods for reducing fearful behavior in singly housed rhesus

macaques. American Journal of Primatology, 71(1), 30–39.

Cless, I. T., Voss-Hoynes, H. A., Ritzmann, R. E., & Lukas, K. E. (2015). Defining

pacing quantitatively: A comparison of gait characteristics between pacing and

non-repetitive locomotion in zoo-housed polar bears. Applied Animal Behaviour

Science, 169, 78–85.

Cloutier, S., Panksepp, J., & Newberry, R. C. (2012). Playful handling by caretakers

reduces fear of humans in the laboratory rat. Applied Animal Behaviour Science,

140(3), 161–171.

Coe, J. (2004). Mixed species rotation exhibits. Presented at the ARAZPA Annual

Conference, Australia, Healesville, VIC, Austrailia: Jon Coe Design Pty Ltd.

Colborn, D. R., Thompson, D. L., Roth, T. L., Capehart, J. S., & White, K. L. (1991).

Responses of cortisol and prolactin to sexual excitement and stress in stallions and

geldings. Journal of Animal Science, 69(6), 2556.

Coleman, K. (2012). Individual differences in temperament and behavioral management

practices for nonhuman primates. Applied Animal Behaviour Science, 137(3),

106–113.

Coleman, K., & Maier, A. (2010). The use of positive reinforcement training to reduce

stereotypic behavior in rhesus macaques. Applied Animal Behaviour Science,

124(3), 142–148.

151

Collins, C. K., Quirke, T., Overy, L., Flannery, K., & O’Riordan, R. (2016). The effect of

the zoo setting on the behavioural diversity of captive gentoo penguins and the

implications for their educational potential. Journal of Zoo and Aquarium

Research, (2).

Crosier, A., Comizzoli, P., Baker, T., Davidson, A., Munson, L., Howard, J., … Wildt, D.

E. (2011). Increasing age influences uterine integrity, but not ovarian function or

oocyte quality, in the cheetah (Acinonyx jubatus). Biology of Reproduction, 85(2),

243–253.

Crosier, A., Lombardi, C., Maloney, E., & Andrews, J. (2018). Population Analysis and

Breeding and Transfer Plan for the Cheetah (Acinonyx jubatus) Yellow Species

Survival Program®. Chicago, IL. AZA Population Management Center at Lincoln

Park Zoo.

Cuéllar, E. (2002). Census of the three-banded armadillo Tolypeutes matacus using dogs,

southern Chaco, Bolivia. Mammalia, 66(3), 448–451.

Czekala, N. M., Hodges, J. K., Gause, G. E., & Lasley, B. L. (1980). Annual circulating

testosterone levels in captive and free-ranging male armadillos (Dasypus

novemcinctus). Journal of Reproduction and Fertility, 59(1), 199–204.

Dantzer, R. (1986). Symposium on “Indices to Measure Animal Well-Being.”

Behavioral, physiological and functional aspects of stereotyped behavior: A

review and a re-interpretation. Journal of Animal Science, 62(6), 1776–1786.

Davey, G. (2007). Visitors’ effects on the welfare of animals in the zoo: A review.

Journal of Applied Animal Welfare Science, 10(2), 169–183.

152

Davis, N., Schaffner, C. M., & Smith, T. E. (2005). Evidence that zoo visitors influence

HPA activity in spider monkeys (Ateles geoffroyii rufiventris). Applied Animal

Behaviour Science, 90(2), 131–141.

Davison, V. M., McMahon, L., Skinner, T. L., Horton, C. M., & Parks, B. J. (1993).

Animals as actors: Take 2. In Annual Proceedings of the American Association of

Zoological Parks and Aquariums (pp. 150–155).

Dawkins, M. S. (1988). Behavioural deprivation: A central problem in animal welfare.

Applied Animal Behaviour Science, 20(3), 209–225.

Dawkins, M. S. (1990). From an animal’s point of view: Motivation, fitness, and animal

welfare. Behavioral and Brain Sciences, 13(1), 1–9.

Dawkins, M. S. (2001). How can we recognize and assess good welfare? In Coping with

Challenge: Welfare in Animals Including Humans (pp. 67–76). Berlin: Dahlem

University Press.

DeCaluwe, H. B., Wielebnowski, N. C., Howard, J., Pelican, K. M., & Ottinger, M. A.

(2013). Behavioral reactions relate to adrenal activity and temperament in male

clouded leopards (Neofelis nebulosa). Applied Animal Behaviour Science, 149(1–

4), 63–71.

Delius, J. D. (1988). Preening and associated comfort behavior in birds. Annals of the

New York Academy of Sciences, 525(1), 40–55.

Dembiec, D. P., Snider, R. J., & Zanella, A. J. (2004). The effects of transport stress on

tiger physiology and behavior. Zoo Biology, 23(4), 335–346.

153

Descovich, K. A., Lisle, A. T., Johnston, S., & Phillips, C. J. C. (2012). Space allowance

and the behaviour of captive southern hairy-nosed wombats (Lasiorhinus

latifrons). Applied Animal Behaviour Science, 140(1), 92–98.

DiStefano, C., Zhu, M., & Mîndrilă, D. (2009). Understanding and using factor scores:

Considerations for the applied researcher. Practical Assessment, Research &

Evaluation, 14(20), 1–11.

Durant, S. M., Mitchell, N., Groom, R., Pettorelli, N., Ipavec, A., Jacobson, A. P., …

Young-Overton, K. (2017). The global decline of cheetah Acinonyx jubatus and

what it means for conservation. Proceedings of the National Academy of

Sciences, 114(3), 528–533.

Evermann, J. F., Laurenson, M. K., McKeirnan, A. J., & Caro, T. M. (1993). Infectious

disease surveillance in captive and free-living cheetahs: An integral part of the

species survival plan. Zoo Biology, 12(1), 125–133.

Fanson, K., & Wielebnowski, N. (2013). Effect of housing and husbandry practices on

adrenocortical activity in captive Canada lynx (Lynx canadensis). Animal

Welfare, 22(2), 159–165.

Farrand, A., Hosey, G., & Buchanan-Smith, H. M. (2014). The visitor effect in petting

zoo-housed animals: Aversive or enriching? Applied Animal Behaviour Science,

151, 117–127.

FAWC (1992). The five freedoms. London: Farm Animal Welfare Council.

154

Fazio, J. M. (2016). Assessment of adrenal activity and reproductive cycles during

captive management in the fishing cat (Prionailurus viverrinus) (Ph.D.). George

Mason University, United States -- Virginia.

Feaver, J., Mendl, M., & Bateson, P. (1986). A method for rating the individual

distinctiveness of domestic cats. Animal Behaviour, 34(4), 1016–1025.

Fernandez, E. J., Tamborski, M. A., Pickens, S. R., & Timberlake, W. (2009). Animal–

visitor interactions in the modern zoo: Conflicts and interventions. Applied

Animal Behaviour Science, 120(1), 1–8.

Föllmi, J., Steiger, A., Walzer, C., Robert, N., Geissbühler, U., Doherr, M., & Wenker, C.

(2007). A scoring system to evaluate physical condition and quality of life in

geriatric zoo mammals. Animal Welfare, 16(3), 309–318.

Fraser, D. (2009). Assessing animal welfare: different philosophies, different scientific

approaches. Zoo Biology, 28(6), 507–518.

Freeman, H. D., & Ross, S. R. (2014). The impact of atypical early histories on pet or

performer chimpanzees. PeerJ, 2, e579.

Frézard, A., & Pape, G. L. (2003). Contribution to the welfare of captive wolves (Canis

lupus lupus): A behavioral comparison of six wolf packs. Zoo Biology, 22(1), 33–

44.

Gartner, M. C., & Powell, D. M. (2012). Personality assessment in snow leopards (Uncia

uncia). Zoo Biology, 31(2), 151–165.

155

Gartner, M. C., Powell, D. M., & Weiss, A. (2014). Personality structure in the domestic

cat (Felis silvestris catus), Scottish wildcat (Felis silvestris grampia), clouded

leopard (Neofelis nebulosa), snow leopard (Panthera uncia), and African lion

(Panthera leo): A comparative study. Journal of Comparative Psychology, 128(4),

414–426.

Gartner, M. C., & Weiss, A. (2013a). Personality in felids: A review. Applied Animal

Behaviour Science, 144(1–2), 1–13.

Gartner, M. C., & Weiss, A. (2013b). Scottish wildcat (Felis silvestris grampia)

personality and subjective well-being: Implications for captive management.

Applied Animal Behaviour Science, 147(3–4), 261–267.

Gold, K. C., & Maple, T. L. (1994). Personality assessment in the gorilla and its utility as

a management tool. Zoo Biology, 13(5), 509–522.

Gosling, S. D. (2001). From mice to men: What can we learn about personality from

animal research? Psychological Bulletin, 127(1), 45–86.

Gosling, S. D. (2008). Personality in non-human animals. Social and Personality

Psychology Compass, 2(2), 985–1001.

Gosling, S. D., & John, O. P. (1999). Personality dimensions in nonhuman animals: A

cross-species review. Current Directions in Psychological Science, 8(3), 69–75.

Gottelli, D., Wang, J., Bashir, S., & Durant, S. M. (2007). Genetic analysis reveals

promiscuity among female cheetahs. Proceedings of the Royal Society of London

B: Biological Sciences, 274(1621), 1993–2001.

156

Grajal, A., Luebke, J. F., Kelly, L.-A. D., Matiasek, J., Clayton, S., Karazsia, B. T., …

Stanoss, R. (2017). The complex relationship between personal sense of

connection to animals and self-reported pro-environmental behaviors by zoo

visitors. Conservation Biology, 31(2), 322–330.

Grandin, T. (1997). Assessment of stress during handling and transport. Journal of

Animal Science, 75(1), 249–257.

Gunn, D., & Morton, D. B. (1995). Inventory of the behaviour of New Zealand White

rabbits in laboratory cages. Applied Animal Behaviour Science, 45(3), 277–292.

Hampson, M. C., & Schwitzer, C. (2016). Effects of hand-rearing on reproductive

success in captive large cats Panthera tigris altaica, Uncia uncia, Acinonyx

jubatus and Neofelis nebulosa. PLOS ONE, 11(5), e0155992.

Hargreaves, A. L., & Hutson, G. D. (1990). Some effects of repeated handling on stress

responses in sheep. Applied Animal Behaviour Science, 26(3), 253–265.

Hattingh, J., Pitts, N. I., & Ganhao, M. F. (1988). Immediate response to repeated capture

and handling of wild impala. Journal of Experimental Zoology, 248(1), 109–112.

Hayward, L. S., Booth, R. K., & Wasser, S. K. (2010). Eliminating the artificial effect of

sample mass on avian fecal hormone metabolite concentration. General and

Comparative Endocrinology, 169(2), 117–122.

Heinrich, C. J., & Birney, B. A. (1992). Effects of live animal demonstrations on zoo

visitors’ retention of information. Anthrozoös, 5(2), 113–121.

157

Hemsworth, P. H., Barnett, J. L., & Hansen, C. (1986). The influence of handling by

humans on the behaviour, reproduction and corticosteroids of male and female

pigs. Applied Animal Behaviour Science, 15(4), 303–314.

Hemsworth, P. H., Barnett, J. L., Hansen, C., & Gonyou, H. W. (1986). The influence of

early contact with humans on subsequent behavioural response of pigs to humans.

Applied Animal Behaviour Science, 15(1), 55–63.

Hildebrand, M. (1961). Further studies on locomotion of the cheetah. Journal of

Mammalogy, 42(1), 84–91.

Hill, S. P., & Broom, D. M. (2009). Measuring zoo animal welfare: theory and practice.

Zoo Biology, 28(6), 531–544.

Hogan, L. A., Johnston, S. D., Lisle, A. T., Keeley, T., Wong, P., Nicolson, V., …

Phillips, C. J. C. (2011). Behavioural and physiological responses of captive

wombats (Lasiorhinus latifrons) to regular handling by humans. Applied Animal

Behaviour Science, 134(3), 217–228.

Hosey, G. (2000). Zoo animals and their human audiences: What is the visitor effect?

Animal Welfare, 9(4), 343–357.

Hosey, G. (2008). A preliminary model of human–animal relationships in the zoo.

Applied Animal Behaviour Science, 109(2), 105–127.

Howell-Stephens, J. A., Brown, J. S., Bernier, D., Mulkerin, D., & Santymire, R. M.

(2012). Characterizing adrenocortical activity in zoo-housed southern three-

158

banded armadillos (Tolypeutes matacus). General and Comparative

Endocrinology, 178(1), 64–74.

Jenny, S., & Schmid, H. (2002). Effect of feeding boxes on the behavior of stereotyping

amur tigers (Panthera tigris altaica) in the Zurich Zoo, Zurich, Switzerland. Zoo

Biology, 21(6), 573–584.

Jurke, M. H., Czekala, N. M., Lindburg, D. G., & Millard, S. E. (1997). Fecal corticoid

metabolite measurement in the cheetah (Acinonyx jubatus). Zoo Biology, 16(2),

133–147.

King, J. E., & Figueredo, A. J. (1997). The five-factor model plus dominance in

chimpanzee personality. Journal of Research in Personality, 31(2), 257–271.

King, N. E., & Mellen, J. D. (1994). The effects of early experience on adult copulatory

behavior in zoo-born chimpanzees (Pan troglodytes). Zoo Biology, 13(1), 51–59.

Kistler, C., Hegglin, D., Würbel, H., & König, B. (2009). Feeding enrichment in an

opportunistic carnivore: The red fox. Applied Animal Behaviour Science, 116(2),

260–265.

Koester, D. C., Freeman, E. W., Brown, J. L., Wildt, D. E., Terrell, K. A., Franklin, A.

D., & Crosier, A. E. (2015). Motile sperm output by male cheetahs (Acinonyx

jubatus) managed ex situ is influenced by public exposure and number of care-

givers. PLOS ONE, 10(9).

Koester, D. C., Wildt, D. E., Brown, J. L., Meeks, K., & Crosier, A. E. (2017). Public

exposure and number of conspecifics have no influence on ovarian and adrenal

159

activity in the cheetah (Acinonyx jubatus). General and Comparative

Endocrinology, 243(Supplement C), 120–129.

Kuhar, C. W. (2008). Group differences in captive gorillas’ reaction to large crowds.

Applied Animal Behaviour Science, 110(3), 377–385.

Kuhar, C. W., Stoinski, T. S., Lukas, K. E., & Maple, T. L. (2006). Gorilla Behavior

Index revisited: Age, housing and behavior. Applied Animal Behaviour Science,

96(3), 315–326.

Larsen, M. J., Sherwen, S. L., & Rault, J.-L. (2014). Number of nearby visitors and noise

level affect vigilance in captive koalas. Applied Animal Behaviour Science, 154,

76–82.

Laule, G., & Whittaker, M. (2007). Enhancing nonhuman primate care and welfare

through the use of positive reinforcement training. Journal of Applied Animal

Welfare Science, 10(1), 31–38.

Lee, P. C., & Moss, C. J. (2012). Wild female African elephants (Loxodonta africana)

exhibit personality traits of leadership and social integration. Journal of

Comparative Psychology, 126(3), 224–232.

Leeds, A., Elsner, R., & Lukas, K. (2016). The effect of positive reinforcement training

on an adult female western lowland gorilla’s (Gorilla gorilla gorilla) rate of

abnormal and aggressive behavior. Animal Behavior and Cognition, 3(2), 78–87.

160

Li, C., Jiang, Z., Tang, S., & Zeng, Y. (2007). Influence of enclosure size and animal

density on fecal cortisol concentration and aggression in Père David’s deer stags.

General and Comparative Endocrinology, 151(2), 202–209.

Lindburg, D. G., Durrant, B. S., Millard, S. E., & Oosterhuis, J. E. (1993). Fertility

assessment of cheetah males with poor quality semen. Zoo Biology, 12(1), 97–

103.

Ludwig, C., Wachter, B., Silinski-Mehr, S., Ganswindt, A., Bertschinger, H., Hofer, H.,

& Dehnhard, M. (2013). Characterisation and validation of an enzyme-

immunoassay for the non-invasive assessment of faecal glucocorticoid

metabolites in cheetahs (Acinonyx jubatus). General and Comparative

Endocrinology, 180 (Supplement C), 15–23.

Luebke, J. F., Watters, J. V., Packer, J., Miller, L. J., & Powell, D. M. (2016). Zoo

Visitors’ affective responses to observing animal behaviors. Visitor Studies,

19(1), 60–76.

Lutz, C., Well, A., & Novak, M. (2003). Stereotypic and self-injurious behavior in rhesus

macaques: A survey and retrospective analysis of environment and early

experience. American Journal of Primatology, 60(1), 1–15.

MacMillen, O. (1994). Zoomobile effectiveness: sixth graders learning vertebrate

classification. In Annual Proceedings of the American Association of Zoological

Parks and Aquariums (pp. 181–183).

161

Maestripieri, D., Schino, G., Aureli, F., & Troisi, A. (1992). A modest proposal:

displacement activities as an indicator of emotions in primates. Animal

Behaviour, 44(5), 967–979.

Majchrzak, Y. N., Mastromonaco, G. F., Korver, W., & Burness, G. (2015). Use of

salivary cortisol to evaluate the influence of rides in dromedary camels. General

and Comparative Endocrinology, 211, 123–130

Mallapur, A., Sinha, A., & Waran, N. (2005). Influence of visitor presence on the

behaviour of captive lion-tailed macaques (Macaca silenus) housed in Indian

zoos. Applied Animal Behaviour Science, 94(3), 341–352.

Mann-Lang, J. B., Ballantyne, R., & Packer, J. (2016). Does more education mean less

fun? A comparison of two animal presentations. International Zoo Yearbook,

50(1), 155–164.

Maple, T. L., & Perdue, B. (2013). Zoo animal welfare. New York: Springer.

Margulis, S. W., Hoyos, C., & Anderson, M. (2003). Effect of felid activity on zoo visitor

interest. Zoo Biology, 22(6), 587–599.

Marker, L., Grisham, J., & Brewer, B. (2017). A brief history of cheetah conservation. In

L. Marker, L. Boast, & A. Schmidt-Kuntzel (Eds.), Cheetahs: Biology and

Conservation (1st ed.). Massachusetts, USA: Academic Press.

Marker, L., & O’Brien, S. J. (1989). Captive breeding of the cheetah (Acinonyx jubatus)

in North American zoos (1871–1986). Zoo Biology, 8(1), 3–16.

162

Marker-Kraus, L., & Grisham, J. (1993). Captive breeding of cheetahs in North American

Zoos: 1987–1991. Zoo Biology, 12(1), 5–18.

Martin, P., Bateson, P. P. G., & Bateson, P. (1993). Measuring Behaviour: An

Introductory Guide. Cambridge University Press.

Mason, G., Clubb, R., Latham, N., & Vickery, S. (2007). Why and how should we use

environmental enrichment to tackle stereotypic behaviour? Applied Animal

Behaviour Science, 102(3), 163–188.

Mason, G., & Latham, N. (2004). Can’t stop, won’t stop: Is stereotypy a reliable animal

welfare indicator? Animal Welfare, 13(1), 57–69.

Mason, G., & Mendl, M. (1993). Why is there no simple way of measuring animal

welfare? Animal Welfare, 2(4), 301–319.

Mason, G., & Veasey, J. S. (2010). How should the psychological well-being of zoo

elephants be objectively investigated? Zoo Biology, 29(2), 237–255.

Mattiello, S., Battini, M., Andreoli, E., Minero, M., Barbieri, S., & Canali, E. (2010).

Avoidance distance test in goats: A comparison with its application in cows.

Small Ruminant Research, 91(2), 215–218.

McCrae, R. R., Costa, P. T., Ostendorf, F., Angleitner, A., Hřebíčková, M., Avia, M. D.,

… Smith, P. B. (2000). Nature over nurture: Temperament, personality, and life

span development. Journal of Personality and Social Psychology, 78(1), 173–186.

McDonough, C. M., & Loughry, W. J. (1997). Influences on activity patterns in a

population of nine-banded armadillos. Journal of Mammalogy, 78(3), 932–941.

163

McKenzie-Mohr, D., & Schultz, P. W. (2014). Choosing effective behavior change tools.

Social Marketing Quarterly, 20(1), 35–46.

McNab, B. K. (1980). Energetics and the limits to a temperate distribution in armadillos.

Journal of Mammalogy, 61(4), 606–627.

Meaney, M. J., Mitchell, J. B., Aitken, D. H., Bhatnagar, S., Bodnoff, S. R., Iny, L. J., &

Sarrieau, A. (1991). The effects of neonatal handling on the development of the

adrenocortical response to stress: Implications for neuropathology and cognitive

deficits in later life. Psychoneuroendocrinology, 16(1), 85–103.

Meder, A. (1989). Effects of hand-rearing on the behavioral development of infant and

juvenile gorillas (Gorilla g. gorilla). Developmental Psychobiology, 22(4), 357–

376.

Meehan, C. L., Hogan, J. N., Bonaparte-Saller, M. K., & Mench, J. A. (2016). Housing

and social environments of African (Loxodonta africana) and Asian (Elephas

maximus) elephants in North American zoos. PLOS ONE, 11(7), e0146703.

Meerlo, Horvath, Nagy, Bohus, & Koolhaas. (1999). The Influence of Postnatal Handling

on Adult Neuroendocrine and Behavioural Stress Reactivity. Journal of

Neuroendocrinology, 11(12), 925–933.

Melfi, V. (2013). Is training zoo animals enriching? Applied Animal Behaviour Science,

147(3), 299–305.

164

Mellen, J. D. (1992). Effects of early rearing experience on subsequent adult sexual

behavior using domestic cats (Felis catus) as a model for exotic small felids. Zoo

Biology, 11(1), 17–32.

Miller, L., Mellen, J., Greer, T., & Kuczaj, S. (2011). The effects of education

programmes on Atlantic bottlenose dolphin (Tursiops truncatus) behaviour.

Animal Welfare, 20(2), 159–172.

Miller, L., Pisacane, C., & Vicino, G. (2016). Relationship between behavioural diversity

and faecal glucocorticoid metabolites: a case study with cheetahs (Acinonyx

jubatus). Animal Welfare, 25(3), 325–329.

Miller, L., Zeigler-Hill, V., Mellen, J., Koeppel, J., Greer, T., & Kuczaj, S. (2013).

Dolphin shows and interaction programs: Benefits for conservation education?

Zoo Biology, 32(1), 45–53.

Millspaugh, J. J., & Washburn, B. E. (2004). Use of fecal glucocorticoid metabolite

measures in conservation biology research: considerations for application and

interpretation. General and Comparative Endocrinology, 138(3), 189–199.

Moberg, G. P., & Mench, J. A. (2000). The Biology of Animal Stress: Basic Principles

and Implications for Animal Welfare. CABI.

Morgan, J. M., & Gramann, J. H. (1989). Predicting effectiveness of wildlife education

programs: A study of students’ attitudes and knowledge toward snakes. Wildlife

Society Bulletin, 17(4), 501–509.

165

Moss, A., Jensen, E., & Gusset, M. (2015). Evaluating the contribution of zoos and

aquariums to Aichi Biodiversity Target 1. Conservation Biology, 29(2), 537–544.

Möstl, E., & Palme, R. (2002). Hormones as indicators of stress. Domestic Animal

Endocrinology, 23(1), 67–74.

Munson, L. (1993). Diseases of captive cheetahs (Acinonyx jubatus): Results of the

cheetah research council pathology survey, 1989–1992. Zoo Biology, 12(1), 105–

124.

Munson, L., Nesbit, J. W., Meltzer, D. G. A., Colly, L. P., Bolton, L., & Kriek, N. P. J.

(1999). Diseases of captive cheetahs (Acinonyx jubatus jubatus) in South Africa:

A 20-year retrospective survey. Journal of Zoo and Wildlife Medicine, 30(3),

342–347.

Munson, L., Terio, K. A., Worley, M., Jago, M., Bagot-Smith, A., & Marker, L. (2005).

Extrinsic factors significantly affect patterns of disease in free-ranging and

captive cheetah (Acinonyx jubatus) populations. Journal of Wildlife Diseases,

41(3), 542–548.

Narayan, E. J., Webster, K., Nicolson, V., Mucci, A., & Hero, J.-M. (2013). Non-invasive

evaluation of physiological stress in an iconic Australian marsupial: The Koala

(Phascolarctos cinereus). General and Comparative Endocrinology, 187, 39–47.

Nimon, A. J., & Dalziel, F. R. (1992). Cross-species interaction and communication: a

study method applied to captive siamang (Hylobates syndactylus) and long-billed

corella (Cacatua tenuirostris) contacts with humans. Applied Animal Behaviour

Science, 33(2), 261–272.

166

Núñez, J. F., Ferré, P., Escorihuela, R. M., Tobeña, A., & Fernández-Teruel, A. (1996).

Effects of postnatal handling of rats on emotional, HPA-axis, and prolactin

reactivity to novelty and conflict. Physiology & Behavior, 60(5), 1355–1359.

O’Brien, S. J., Roelke, M. E., Marker, L., Newman, A., Winkler, C. A., Meltzer, D., …

Wildt, D. E. (1985). Genetic basis for species vulnerability in the cheetah.

Science, 227(4693), 1428–1434.

O’Brien, S. J., Wildt, D. E., Goldman, D., Merril, C. R., & Bush, M. (1983). The cheetah

is depauperate in genetic variation. Science, 221(4609), 459–462.

O’Rourke, N., & Hatcher, L. (2013). A Step-by-Step Approach to Using SAS for Factor

Analysis and Structural Equation Modeling, Second Edition. SAS Institute.

Owen, M. A., Swaisgood, R. R., Czekala, N. M., & Lindburg, D. G. (2005). Enclosure

choice and well-being in giant pandas: is it all about control? Zoo Biology, 24(5),

475–481.

Packer, J., & Ballantyne, R. (2010). The role of zoos and aquariums in education for a

sustainable future. New Directions for Adult and Continuing Education,

2010(127), 25–34.

Pang, B., Van Valkenburgh, B., Kitchell, Jr., K. F., Dickman, A., & Marker, L. (2017).

History of the cheetah-human relationship. In A. Schmidt-Kuntzel (Ed.),

Cheetahs: Biology and Conservation (1st ed.). Massachusetts, USA: Academic

Press.

167

Pedersen, V. (1994). Long-term effects of different handling procedures on behavioural,

physiological, and production-related parameters in silver foxes. Applied Animal

Behaviour Science, 40(3), 285–296.

Pereira, R. J. G., Granzinolli, M. A. M., & Duarte, J. M. B. (2010). Annual profile of

fecal androgen and glucocorticoid levels in free-living male American kestrels

from southern mid-latitude areas. General and Comparative Endocrinology,

166(1), 94–103.

Phillips, C., & Peck, D. (2007). The effects of personality of keepers and tigers (Panthera

tigris tigris) on their behaviour in an interactive zoo exhibit. Applied Animal

Behaviour Science, 106(4), 244–258.

Pirovino, M., Heistermann, M., Zimmermann, N., Zingg, R., Clauss, M., Codron, D., …

Steinmetz, H. W. (2011). Fecal glucocorticoid measurements and their relation to

rearing, behavior, and environmental factors in the population of pileated gibbons

(Hylobates pileatus) held in European zoos. International Journal of Primatology,

32(5), 1161.

Podberscek, A. L., Blackshaw, J. K., & Beattie, A. W. (1991). The effects of repeated

handling by familiar and unfamiliar people on rabbits in individual cages and

group pens. Applied Animal Behaviour Science, 28(4), 365–373.

Pomerantz, O., & Terkel, J. (2009). Effects of positive reinforcement training techniques

on the psychological welfare of zoo-housed chimpanzees (Pan troglodytes).

American Journal of Primatology, 71(8), 687–695.

168

Pooley, J. A., & O’Connor, M. (2000). Environmental education and attitudes: Emotions

and beliefs are what is needed. Environment and Behavior, 32(5), 711–723.

Povey, K. (2002). Close encounters: the benefits of using education program animals. In

Annual Proceedings of the Association of Zoos and Aquariums (pp. 117–121).

Povey, K., & Rios, J. (2002). Using interpretive animals to deliver affective messages in

zoos. Journal of Interpretation Research, 7(2), 19–28.

Powell, D. M., & Gartner, M. C. (2011). Applications of personality to the management

and conservation of nonhuman animals. In From Genes to Animal Behavior (pp.

185–199). Springer, Tokyo.

Powell, D. M., & Svoke, J. T. (2008). Novel environmental enrichment may provide a

tool for rapid assessment of animal personality: A case study with giant pandas

(Ailuropoda melanoleuca). Journal of Applied Animal Welfare Science, 11(4),

301–318.

Prudom, A. E., & Klemm, W. R. (1973). Electrographic correlates of sleep behavior in a

primitive mammal, the armadillo Dasypus novemcinctus. Physiology & Behavior,

10(2), 275–282.

Quirke, T., & O’Riordan, R. (2015). An investigation into the prevalence of exploratory

behavior in captive cheetahs (Acinonyx jubatus): Exploratory Behavior in Captive

Cheetahs. Zoo Biology, 34(2), 130–138.

169

Quirke, T., & O’Riordan, R. M. (2011a). The effect of a randomised enrichment

treatment schedule on the behaviour of cheetahs (Acinonyx jubatus). Applied

Animal Behaviour Science, 135(1), 103–109.

Quirke, T., & O’Riordan, R. M. (2011b). The effect of different types of enrichment on

the behaviour of cheetahs (Acinonyx jubatus) in captivity. Applied Animal

Behaviour Science, 133(1), 87–94.

Quirke, T., O’Riordan, R. M., & Zuur, A. (2012). Factors influencing the prevalence of

stereotypical behaviour in captive cheetahs (Acinonyx jubatus). Applied Animal

Behaviour Science, 142(3–4), 189–197.

Rapp, S., Vannelli, K., Castaneda, L., Beckhelling, A., Ekard, S., Hilker, C., … Stanek,

L. (2017). Communicating the conservation message – using ambassador

cheetahs in education. In L. Marker, L. Boast, & A. Schmidt-Kuntzel (Eds.),

Cheetahs: Biology and Conservation (1st ed.). Massachusetts, USA: Academic

Press.

Razal, C. B., Pisacane, C. B., & Miller, L. J. (2016). Multifaceted approach to personality

assessment in cheetahs (Acinonyx jubatus). Animal Behavior and Cognition, 3(1),

22–31.

Reade, L. S., & Waran, N. K. (1996). The modern zoo: How do people perceive zoo

animals? Applied Animal Behaviour Science, 47(1), 109–118.

Ross, S. R., Vreeman, V. M., & Lonsdorf, E. V. (2011). Specific image characteristics

influence attitudes about chimpanzee conservation and use as pets. PLOS ONE,

6(7), e22050.

170

Rushen, J. (1991). Problems associated with the interpretation of physiological data in the

assessment of animal welfare. Applied Animal Behaviour Science, 28(4), 381–

386.

Ryan, S., Thompson, S. D., Roth, A. M., & Gold, K. C. (2002). Effects of hand-rearing

on the reproductive success of western lowland gorillas in North America. Zoo

Biology, 21(4), 389–401.

Santana, E. M., Jantz, H. E., & Best, T. L. (2010). Atelerix albiventris (Erinaceomorpha:

Erinaceidae). Mammalian Species, 99–110.

Sapolsky, R. M. (1994). Individual differences and the stress response. Seminars in

Neuroscience, 6(4), 261–269.

Sellinger, R. L., & Ha, J. C. (2005). The effects of visitor density and intensity on the

behavior of two captive jaguars (Panthera onca). Journal of Applied Animal

Welfare Science, 8(4), 233–244.

Shani, A., & Pizam, A. (2008). Towards an ethical framework for animal‐based

attractions. International Journal of Contemporary Hospitality Management,

20(6), 679–693.

Shannon, C., & Weaver, W. (1949). The mathematical theory of communication. Urbana,

IL: University of Illinois Press.

Shepherdson, D. J., Carlstead, K., Mellen, J. D., & Seidensticker, J. (1993). The influence

of food presentation on the behavior of small cats in confined environments. Zoo

Biology, 12(2), 203–216.

171

Shepherdson, D. J., Carlstead, K., & Wielebnowski, N. C. (2004). Cross-institutional

assessment of stress responses in zoo animals using longitudinal monitoring of

faecal corticoids and behaviour. Animal Welfare, 13(1), 105–113.

Shepherdson, D. J., Lewis, K. D., Carlstead, K., Bauman, J., & Perrin, N. (2013).

Individual and environmental factors associated with stereotypic behavior and

fecal glucocorticoid metabolite levels in zoo housed polar bears. Applied Animal

Behaviour Science, 147(3), 268–277.

Sherwen, S. L., Harvey, T. J., Magrath, M. J. L., Butler, K. L., Fanson, K. V., &

Hemsworth, P. H. (2015). Effects of visual contact with zoo visitors on black-

capped capuchin welfare. Applied Animal Behaviour Science, 167, 65–73.

Sherwen, S. L., Hemsworth, P. H., Butler, K. L., Fanson, K. V., & Magrath, M. J. L.

(2015). Impacts of visitor number on kangaroos housed in free-range exhibits.

Zoo Biology, 34(4), 287–295.

Sherwen, S. L., Magrath, M. J. L., Butler, K. L., & Hemsworth, P. H. (2015). Little

penguins, Eudyptula minor, show increased avoidance, aggression and vigilance

in response to zoo visitors. Applied Animal Behaviour Science, 168, 71–76.

Sherwen, S. L., Magrath, M. J. L., Butler, K. L., Phillips, C. J. C., & Hemsworth, P. H.

(2014). A multi-enclosure study investigating the behavioural response of

meerkats to zoo visitors. Applied Animal Behaviour Science, 156, 70–77.

Sherwin, C. M., Haug, E., Terkelsen, N., & Vadgama, M. (2004). Studies on the

motivation for burrowing by laboratory mice. Applied Animal Behaviour Science,

88(3), 343–358.

172

Sherwood, K. P., Rallis, S. F., & Stone, J. (1989). Effects of live animals vs. preserved

specimens on student learning. Zoo Biology, 8(1), 99–104.

Shrout, P. E., & Fleiss, J. L. (1979). Intra-class correlations: Uses in assessing rater

reliability. Psychological Bulletin, 86(2), 420–428.

Shyne, A., & Block, M. (2010). The effects of husbandry training on stereotypic pacing

in captive African wild dogs (Lycaon pictus). Journal of Applied Animal Welfare

Science, 13(1), 56–65.

Siegel, J. M. (2005). Clues to the functions of mammalian sleep. Nature, 437 (7063),

1264.

Smith, P. (2007). Southern Three-Banded Armadillo Tolypeutes matacus. In FAUNA

Paraguay Handbook of the Mammals of Paraguay (Vol. 7).

Sockman, K. W., & Schwabl, H. (2001). Plasma corticosterone in nestling American

kestrels: effects of age, handling stress, yolk androgens, and body condition.

General and Comparative Endocrinology, 122(2), 205–212.

Staley, A. M., Blanco, J. M., Dufty, A. M., Wildt, D. E., & Monfort, S. L. (2007). Fecal

steroid monitoring for assessing gonadal and adrenal activity in the golden eagle

and peregrine falcon. Journal of Comparative Physiology B, 177(6), 609–622.

Stevenson-Hinde, J., & Zunz, M. (1978). Subjective assessment of individual rhesus

monkeys. Primates, 19(3), 473–482.

173

Stoinski, T. S., Jaicks, H. F., & Drayton, L. A. (2012). Visitor effects on the behavior of

captive western lowland gorillas: The importance of individual differences in

examining welfare. Zoo Biology, 31(5), 586–599.

Suárez, P., Recuerda, P., & Arias-de-Reyna, L. (2017). Behaviour and welfare: The

visitor effect in captive felids. Animal Welfare, 26(1), 25–34.

Superina, M., Carreño, N., & Jahn, G. A. (2009). Characterization of seasonal

reproduction patterns in female pichis Zaedyus pichiy (Xenarthra: Dasypodidae)

estimated by fecal sex steroid metabolites and ovarian histology. Animal

Reproduction Science, 116(3), 358–369.

Swaisgood, R. R. (2007). Current status and future directions of applied behavioral

research for animal welfare and conservation. Applied Animal Behaviour

Science, 102(3), 139–162.

Swanagan, J. S. (2000). Factors influencing zoo visitors’ conservation attitudes and

behavior. The Journal of Environmental Education, 31(4), 26–31.

Tanida, H., Miura, A., Tanaka, T., & Yoshimoto, T. (1995). Behavioral response to

humans in individually handled weanling pigs. Applied Animal Behaviour

Science, 42(4), 249–259.

Terio, K. A., Marker, L., & Munson, L. (2004). Evidence for chronic stress in captive but

not free-ranging cheetahs (Acinonyx jubatus) based on adrenal morphology and

function. Journal of Wildlife Diseases, 40(2), 259–266.

174

Terrell, K. A., Wildt, D. E., Anthony, N. M., Bavister, B. D., Leibo, S. P., Penfold, L. M.,

… Crosier, A. E. (2011). Oxidative phosphorylation is essential for felid sperm

function, but is substantially lower in cheetah (Acinonyx jubatus) compared to

domestic cat (Felis catus) ejaculate. Biology of Reproduction, 85(3), 473–481.

Tetley, C., & O’Hara, S. (2012). Ratings of animal personality as a tool for improving the

breeding, management and welfare of zoo mammals. Animal Welfare, 21(4),

463–476.

Touma, C., & Palme, R. (2005). Measuring fecal glucocorticoid metabolites in mammals

and birds: The importance of validation. Annals of the New York Academy of

Sciences, 1046(1), 54–74.

Tuyttens, F. A. M. (2005). The importance of straw for pig and cattle welfare: A review.

Applied Animal Behaviour Science, 92(3), 261–282.

Veasey, J. S., Waran, N. K., & Young, R. J. (1996). On comparing the behaviour of zoo

housed animals with wild conspecifics as a welfare indicator. Animal Welfare,

5(1), 13–24.

Vickery, S., & Mason, G. (2004). Stereotypic behavior in Asiatic black and Malayan sun

bears. Zoo Biology, 23(5), 409–430.

Wagman, J. D., Lukas, K. E., Dennis, P. M., Willis, M. A., Carroscia, J., Gindlesperger,

C., & Schook, M. W. (2018). A work-for-food enrichment program increases

exploration and decreases stereotypies in four species of bears. Zoo Biology,

37(1), 3–15.

175

Waiblinger, S., Boivin, X., Pedersen, V., Tosi, M.-V., Janczak, A. M., Visser, E. K., &

Jones, R. B. (2006). Assessing the human–animal relationship in farmed species:

A critical review. Applied Animal Behaviour Science, 101(3), 185–242.

Wasser, S. K., & Hunt, K. E. (2005). Noninvasive measures of reproductive function and

disturbance in the barred owl, great horned owl, and Northern spotted owl. Annals

of the New York Academy of Sciences, 1046(1), 109–137.

Wasser, S. K., Hunt, K. E., Brown, J. L., Cooper, K., Crockett, C. M., Bechert, U., …

Monfort, S. L. (2000). A generalized fecal glucocorticoid assay for use in a

diverse array of nondomestic mammalian and avian species. General and

Comparative Endocrinology, 120(3), 260–275.

Watters, J. V. (2014). Searching for behavioral indicators of welfare in zoos: Uncovering

anticipatory behavior: Anticipatory behavior as a welfare tool. Zoo Biology,

33(4), 251–256.

Watters, J. V., & Powell, D. M. (2012). Measuring animal personality for use in

population management in zoos: Suggested methods and rationale. Zoo Biology,

31(1), 1–12.

Weinstein, T. A., Capitanio, J. P., & Gosling, S. D. (2008). Personality in animals. In O.

P. John, R. W. Robins, & L. A. Pervin (Eds.), Handbook of Personality: Theory

and Research (3rd ed., pp. 328–349). New York: The Guilford Press.

Weiss, A., King, J. E., & Figueredo, A. J. (2000). The heritability of personality factors in

chimpanzees (Pan troglodytes). Behavior Genetics, 30(3), 213–221.

176

Wells, A., Terio, K. A., Ziccardi, M. H., & Munson, L. (2004). The stress response to

environmental change in captive cheetahs (Acinonyx jubatus). Journal of Zoo and

Wildlife Medicine, 35(1), 8–14.

Wells, D. L. (2005). A note on the influence of visitors on the behaviour and welfare of

zoo-housed gorillas. Applied Animal Behaviour Science, 93(1), 13–17.

White, B. C., Houser, L. A., Fuller, J. A., Taylor, S., & Elliott, J. L. L. (2003). Activity-

based exhibition of five mammalian species: Evaluation of behavioral changes.

Zoo Biology, 22(3), 269–285.

Widmaier, E. P., & Kunz, T. H. (1993). Basal, diurnal, and stress-induced levels of

glucose and glucocorticoids in captive bats. Journal of Experimental Zoology,

265(5), 533–540.

Wiedenmayer, C. (1997). Causation of the ontogenetic development of stereotypic

digging in gerbils. Animal Behaviour, 53(3), 461–470.

Wielebnowski, N. C. (1999). Behavioral differences as predictors of breeding status in

captive cheetahs. Zoo Biology, 18(4), 335–349.

Wielebnowski, N. C. (2003). Stress and distress: evaluating their impact for the well-

being of zoo animals. Journal of the American Veterinary Medical Association,

223(7), 973–977.

Wielebnowski, N. C., Fletchall, N., Carlstead, K., Busso, J. M., & Brown, J. L. (2002).

Noninvasive assessment of adrenal activity associated with husbandry and

177

behavioral factors in the North American clouded leopard population. Zoo

Biology, 21(1), 77–98.

Wielebnowski, N. C., Ziegler, K., Wildt, D. E., Lukas, J., & Brown, J. L. (2002). Impact

of social management on reproductive, adrenal and behavioural activity in the

cheetah (Acinonyx jubatus) Animal Conservation Forum, 5(4), 291–301.

Wiepkema, P. R., & Koolhaas, J. M. (1993). Stress and Animal Welfare. Animal

Welfare, 2(3), 195–218.

Wildt, D. E., Brown, J. L., Bush, M., Barone, M. A., Cooper, K. A., Grisham, J., &

Howard, J. G. (1993). Reproductive status of cheetahs (Acinonyx jubatus) in

North American Zoos: The benefits of physiological surveys for strategic

planning. Zoo Biology, 12(1), 45–80.

Wildt, D. E., Phillips, L. G., Simmons, L. G., Chakraborty, P. K., Brown, J. L., Howard,

J. G., … Bush, M. (1988). A comparative analysis of ejaculate and hormonal

characteristics of the captive male cheetah, tiger, leopard, and puma. Biology of

Reproduction, 38(2), 245–255.

Williams, B. G., Waran, N. K., Carruthers, J., & Young, R. J. (1996). The effect of a

moving bait on the behaviour of captive cheetahs (Acinonyx Jubatus). Animal

Welfare, 5(3), 271–281.

Windschnurer, I., Boivin, X., & Waiblinger, S. (2009). Reliability of an avoidance

distance test for the assessment of animals’ responsiveness to humans and a

preliminary investigation of its association with farmers’ attitudes on bull

fattening farms. Applied Animal Behaviour Science, 117(3), 117–127.

178

Yeates, J. W., & Main, D. C. J. (2008). Assessment of positive welfare: A review. The

Veterinary Journal, 175(3), 293–300.

Yerke, R., & Burns, A. (1991). Measuring the impact of animal shows on visitor

attitudes. In Annual Proceedings of the American Association of Zoological Parks

and Aquariums (pp. 532–534).

Yerke, R., & Burns, A. (1993). Evaluation of the educational effectiveness of an animal

show outreach program for schools. In Proceedings of the American Association

of Zoological Parks and Aquariums (pp. 366–371).

Young, K. M., Walker, S. L., Lanthier, C., Waddell, W. T., Monfort, S. L., & Brown, J.

L. (2004). Noninvasive monitoring of adrenocortical activity in carnivores by

fecal glucocorticoid analyses. General and Comparative Endocrinology, 137(2),

148–165.

179