Chapter Six: A pilot investigation on the use of startle reactivity as a measure of ...... displeasing sounds to humans (e.g. classical music vs. metal music; ...... exhibit and holding cages (iPhone 5s, Apple Inc., Cupertino, CA; SPL Pro ...... (Panasonic HCV700K and Samsung HMX-H200N) were stationed on tripods in front of.
THE INFLUENCE OF THE SOUND ENVIRONMENT ON THE WELFARE OF ZOO-HOUSED CALLITRICHINE MONKEYS
by
JASON D. WARK
Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy
Dissertation Advisor: Dr. Kristen E. Lukas
Department of Biology CASE WESTERN RESERVE UNIVERSITY
August 2015
CASE WESTERN RESERVE UNIVERSITY SCHOOL OF GRADUATE STUDIES
We hereby approve the dissertation of Jason D. Wark, candidate for the degree of Doctor of Philosophy*.
Signed: _____________________________________ Mark A. Willis, Ph.D. (chair of the committee)
_____________________________________ Kristen E. Lukas, Ph.D.
_____________________________________ Mandi W. Schook, Ph.D.
_____________________________________ Christopher W. Kuhar, Ph.D.
_____________________________________ Charles T. Snowdon, Ph.D.
Date: 21 May 2015 * We also certify that written approval has been obtained for any proprietary material contained within.
This work is dedicated to: Frances Chung, my partner and best friend; my parents, Brenda and Lee Wark, for instilling their love of animals; and my friends, both human and animal, that have enriched my life.
i
Table of Contents Acknowledgements ........................................................................................................... vii Abstract ............................................................................................................................... x Chapter One: Introduction: Sound and its significance ................................................... 1 Chapter Two: Fecal glucocorticoid metabolite responses to management stressors and social change in four species of callitrichine monkeys ............................... 32 Chapter Three: The influence of waterfall sounds and access to off-exhibit areas on the behavior and exhibit use of three species of callitrichine monkeys ....................... 58 Chapter Four: Do zoo animals use off-exhibit areas to avoid noise? A case study exploring the influence of sound on the behavior, physiology, and exhibit use of two pied tamarins (Saguinus bicolor) .................................................................... 81 Chapter Five: The response of callitrichine monkeys to auditory stimuli and the importance of exhibit choices ........................................................................................... 98 Chapter Six: A pilot investigation on the use of startle reactivity as a measure of affect in callitrichine monkeys ........................................................................................ 114 Chapter Seven: General discussion and future outlook ................................................ 134 References ....................................................................................................................... 143
ii LIST OF TABLES Chapter Two: Table 1. Demographic background of study subjects. ..................................................... 35 Table 2. Assessment of fecal markers fed to four species of callitrichine monkeys: golden lion tamarin (GLT); callimico (CA), pied tamarin (PT); white-fronted marmoset (WFM).............................................................................................................. 41 Table 3. Summary of fecal glucocorticoid metabolite response of four callitrichine species after veterinary exams. ..................................................................... 43
Chapter Three: Table 1. Demographic background of study subjects. ..................................................... 61 Table 2. Overview of experimental design. ..................................................................... 62 Table 3. Ethogram of behaviors considered in this study. ............................................... 63 Table 4. Sound levels (median ± interquartile range) in the study exhibits with the waterfall feature on and off. ........................................................................................ 69 Table 5. The median (± IQR) number of visitor groupsa per observation at exhibits across study conditions........................................................................................ 70
iii Chapter Four: Table 1. Ethogram of select behaviors considered in this study. ..................................... 85 Table 2. The equivalent continuous (Leq) and median (L50) sound level indices (median ± IQR) for the exhibit area when a waterfall feature was on and off and for the off-exhibit area when a speaker broadcast white noise volume matched to the waterfall noise on exhibit. ........................................................................................... 90
Chapter Five: Table 1. Description of sound conditions. ..................................................................... 101 Table 2. The duration, frequency, and tempo of playback stimuli. ............................... 102
Chapter Six: Table 1. Description of sound conditions. ..................................................................... 118 Table 2. Ethogram of general behaviors considered in this study. ................................ 120 Table 3. Behavior of tamarins and marmosets before, during, and after exposure to sound conditions and acoustic startle event. ............................................................... 124 Table 4. Summary of startle response measures after an acoustic startle event following sound conditions. ............................................................................................ 129
iv LIST OF FIGURES Chapter Two: Figure 1. Fecal glucocorticoid levels of four golden lion tamarins (A: LF1; B: LF2; C: LM1; D: LM2) after capture and anesthesia during a routine veterinary exam (time 0 h). ................................................................................................................ 45 Figure 2. Fecal glucocorticoid metabolite concentrations in a female (A) and male (B) callimico before and after a routine veterinary exam (time 0 h). ...................... 47 Figure 3. Fecal glucocorticoid concentrations of a female pied tamarin (A) and her adult male offspring (B) before and after the death of the breeding male and veterinary exam (time 0 h). ............................................................................................... 48 Figure 4. Fecal glucocorticoid concentrations of two male white-fronted marmosets (A: MM1; B: MM2) before and after the death of the breeding female, temporary removal of a neonate, and subsequent veterinary exam (arrow at time 0 h). ...................................................................................................................................... 50
Chapter Three: Figure 1. The least squares means from GLMM behavior models for each species showing the effects of the waterfall (a-f) and access (g-l) conditions. ............................. 72 Figure 2. The influence of waterfall noise on monitoring the visitor area when monkeys had no access off-exhibit and when they had access off-exhibit....................... 73 Figure 3. The least squares means from GLMM exhibit use models for each species showing the effects of the waterfall. .................................................................... 74
v Chapter Four: Figure 1. The percent of time spent off exhibit during sound conditions involving experimental modifications of a waterfall feature and white noise played from a speaker in the off-exhibit area........................................................................................... 91 Figure 2. Fecal glucocorticoid metabolite concentrations (ng/g) for a male (grey circles) and female (black Xs) pied tamarin across experimental sound conditions. ....... 93
Chapter Five: Figure 1. Spectrograms of playback stimuli used in this study (A,B: Mozart; C,D: Rainforest Sounds; E,F: Affiliation-based Tamarin Music; G,H: Fear/Threatbased Tamarin Music)..................................................................................................... 103 Figure 2. Diagram of experimental playback procedure. .............................................. 105 Figure 3. The mean percent of time three callitrichine groups (pied tamarin, golden lion tamarin, and callimico) spent in the location a speaker was placed during auditory conditions. ............................................................................................. 107 Figure 4. The mean group rate (occurrences/min visible) of behavior for three callitrichine species (A: pied tamarin; B: golden lion tamarin; C: callimico) during auditory conditions.......................................................................................................... 109 Figure 5. The mean group rate (occurrences/min visible) of behavior for three callitrichine species (A: pied tamarin; B: golden lion tamarin; C: callimico) during periods with access to off-exhibit areas (Phase 1) and no access (Phase 2). .................. 110
vi Chapter Six: Figure 1. The percent of visible time marmosets spent inactive and locomoting during baseline and playback periods for each of the seven sound conditions. ............. 125 Figure 2. The percent of visible time pied tamarins spent engaged in self-directed behavior during the 80 dB White Noise startle trial. ...................................................... 127
vii Acknowledgements
I am eternally grateful for the love, support, and guidance I have received from so many during this journey. First and foremost, I want to thank Franke Chung for not breaking up with me when I moved away to Cleveland and, even better, joining me on this adventure. From our first apartment, where we found a cat companion, to our current one, she has made the beautiful and relaxing retreat I am fortunate to call home. She has shown endless patience and support during my periods of endless stress for that I am thankful. Her attention to detail and standards of excellence for her own work have pushed me to be the most rigorous scientist I can be. This work would not be possible without her love and support. I would not be here if it were not for the passion for animals and science that my parents, Lee and Brenda Wark, instilled in me. Our frequent trips to the Philadelphia Zoo and their enrolling me in summer zoo camp, as well as our trips to the San Diego Zoo and Chicago Field Museum, inspired a curiosity of the natural world that will stay with me forever. I have to thank Dr. Kristen Lukas, my academic advisor, for shaping that curiosity into a career. From our first meeting, Kristen has always been much more than an advisor and I cannot thank her enough for her friendship and guidance through these years. She has taught me to keep having fun, no matter how tough life gets. Also, I am especially grateful for her support of my frequent and diverse research interests, which surprisingly have yet to include gorillas. Kristen, I will never forget that you made this all possible for me and will always look to honor this opportunity by paying it forward.
viii My entire dissertation committee has had a profound impact on my academic and professional development and I thank you all for sharing your time to make this effort possible. I am grateful for the personal mentorship Dr. Mandi Schook has given me and for always pushing me to do the best science I can, even though it did mean learning advanced statistics. The experimental nature of this project would not have been possible without the strong support from Dr. Chris Kuhar. Chris has always challenged me to endeavor for the highest standards of scientific rigor and practicality in my research designs and for that I am thankful. I am grateful for Dr. Mark Willis’s support of the zoo program, guidance through my initial years, and never letting my research thoughts wander too far from a hypothesis and predictions. My dissertation project has benefited greatly from Dr. Charles Snowdon’s extensive experience and invaluable insight into callitrichine species—I truly could not have found a better addition to my dissertation committee. Also, I am very appreciative of Dr. Snowdon’s near immediate email responsiveness and patience with our technology conferencing challenges. Finally, Dr. Pam Dennis, although a casualty of a last minute dissertation rule decree, has been a part of this process from day one. Her cheerful attitude and passionate care for all species, big and small, will always be an inspiration to me. I have been especially lucky to have shared an office with a fantastic team of people during my time at Cleveland. To my academic big sisters, Drs. Elena Less and Grace Fuller: I thank you both for all the help and support you’ve offered me over the years and am grateful for the many laughs we’ve shared. I would like to thank Christine Cassella for her passionate care for all life and those vegetarian nudges along the way. I am thankful for Jason Wagman’s fellowship and easygoing personality. As for my final
ix office buddies, the awesome team of Austin Leeds and Bonnie Baird, I am very grateful for all your help and, of course, our friendship. I am also extremely grateful for all the zoo staff and volunteers that helped make this research possible. First, I have to thank Laura Amendolagine, Cleveland’s Wildlife Endocrinology Lab manager, for her patience and always helping my assays do a good job. I am also grateful for all the support of Cleveland Metroparks Zoo’s animal managers for this project, particularly Andi Kornak, Tad Schoffner, and Lynn Koscielny. They have welcomed my views throughout and their leadership has left an indelible mark on my professional development. I thank the RainForest keepers Joe Ropelewski, Scott Parish, Terri Rhyner, and Chris Gertiser for their facilitation of this research. This project would not have been possible without the help of many research volunteers. In particular, I would like to thank Christine Olle, Devyn Riley, Amanda Barabas, Ingrid Rinker, Elaine Leickly, Suzy Peoples and Jen Avondet for their help in data collection and entry. I would like to extend a special thank you to David Teie. A talented composer, Mr. Teie’s interest in the origins of music led to his collaboration with Dr. Snowdon and their development of tamarin-specific music. I am grateful for his permission to have incorporated this unique music into my research project. Lastly, I am fortunate to have known and had the opportunity to work closely with a very special person, Joan Rog. Joan was simply an amazing woman. She had a fervent love of family and critters and tenacity for her research that left its mark on all around her. For Joan and the furry friends I lost along this journey, I will cherish our memories and am forever grateful for our time together.
x The Influence of the Sound Environment on the Welfare of Zoo-housed Callitrichine Monkeys By JASON WARK Abstract
Animals in the zoo environment are exposed to a multitude of sounds. Some sounds, such as those from the visiting public, are inexorable components of the sound environment of a zoo and may, in some cases, have a negative impact on the behavior of animals. Auditory masking of visitor noise, such as from waterfall features, may alleviate adverse effects of noise but this has not yet been evaluated. Other sounds, such as music or habitat sounds, may be introduced in an attempt to enrich the animals but their utility is questionable. This project investigated the influence of the sound environment on four species of callitrichine monkeys: pied tamarin, white-fronted marmoset, golden lion tamarin, and callimico. The goal of this research was to identify enriching and adverse sound environments and, in the case of the latter, evaluate strategies to ameliorate this effect and improve welfare. The general hypothesis was that sounds experienced in the zoo setting can affect the welfare of zoo-housed callitrichine monkeys. This project focused on sound from a nearby waterfall feature as well as music and other sounds that may be viewed by some as auditory enrichment. The aims were fourfold. First, hormonal and cognitive methods for assessing welfare were evaluated. Second, the potential of sound from a nearby waterfall feature to provide beneficial
xi auditory masking was assessed. Third, rainforest sounds and music were assessed as potential forms of auditory enrichment. Lastly, the influence of providing access to quiet off-exhibit areas was examined. Experimental manipulations of the waterfall feature did not identify beneficial auditory masking effects and indicated waterfall noise may instead be aversive. Playing rainforest sounds or music did not elicit behavioral benefits. Providing access to off-exhibit areas was beneficial and appeared to ameliorate some stressful effects of the waterfall noise. These results indicate the sound environment did influence the welfare of callitrichine monkeys, albeit minimally, and suggest quieter sound environments may be beneficial for these species. Providing access to quiet areas is an important strategy for promoting environmental choices for zoo animals and opportunities to cope with noise.
1 Chapter One: Introduction: Sound and its significance
Sound plays a central role in the lives of many animals, from signaling early warning of danger, allowing detection of prey, or communication between individuals. However, not all sounds provide reliable unambiguous information. A rustle of leaves could be a squirrel searching for food or a predator ready to pounce. Noise, generally defined as unwanted sounds, may elicit a maladaptive response or potentially hamper an animal’s ability to detect reliable sounds (Wright et al., 2007). Thus, an animal’s ability to distinguish meaningful sounds from noise and promote appropriate responses will likely enhance the fitness of that individual. For animals living in a captive setting—with no risk of predation and no need to hunt or locate mates—many sounds no longer convey biological meaning and much of the sound environment may be perceived as noise. However, maladaptive responses to noise may persist and represent an important concern for animal caretakers. These responses may be short-term and have little consequence, but they may also affect psychological and physiological processes. Sounds can have a strong impact on human emotions and may similarly affect wild animals, both positively and negatively. Research on multiple species have identified direct links from auditory pathways to the amygdala, a brain region associated with emotion (LeDoux, 2000). As sound is pervasive, managing noise, unlike other aspects of the physical environment, is particularly challenging. On top of that, how we experience the sound environment is shaped by our hearing range and perception of sounds. Many species can hear sounds we cannot detect and our knowledge of how animals perceive different sounds—whether as
2 potential threats, distractions, or as a sources of interest—is limited. These concerns highlight the importance of carefully considering the effects of the sound environment on the welfare of zoo-housed animals. Animal Welfare Defining welfare The Animal Welfare Act of 1966 and its subsequent amendments established a legal mandate in the United States for providing proper care and management of animals. Although a clear definition of welfare has proven challenging, a growing number of studies have concluded that welfare is a multifaceted concept and likely varies along a continuum (Boissy et al., 2007; Mason and Mendl, 1993; Yeates and Main, 2008). This project utilizes the definition of animal welfare proposed by the Association of Zoos and Aquariums Animal Welfare Committee (2015): “Animal welfare refers to an animal’s collective physical, mental, and emotional states over a period of time, and is measured on a continuum of good to poor.” Welfare, emotion, and affective states Concern for animal welfare is often based in the assumption that animals are sentient beings capable of experiencing pain and suffering (Mendl et al., 2009). However, this acknowledgement does not signify that animals experience humanlike emotions, which also feature a strong cognitive component. To differentiate between these concepts, I will refer to “affective states” when discussing the positive and negative subjective experiences of animals (i.e. valence) that may or may not involve a cognitive and conscious component, and “emotion” when specifically referring to those states
3 experienced by humans. However, as some authors do attribute emotional states to animals (e.g. fear-potentiated startle response), some overlap is inevitable. Two models of emotion have been debated: the dimensional and discrete models. Briefly, the dimensional model developed by Russell (1980) posits that our emotional experience is built from two primary dimensions: valence (i.e. how attractive or aversive an experience is) and arousal (i.e. how stimulating an experience is). These dimensions are typically modeled as a two-dimensional affective space, with four possible affective endpoints. Proponents for discrete models, on the other hand, argue that discrete emotions form the basis of our subjective experience. For instance, Eckman and Friesen’s (1971) cross-cultural study of facial expressions identified six discrete emotions: happiness, anger, sadness, disgust, surprise, and fear. Although these models may be complementary (Mendl et al., 2010), this project will draw from the dimensional model of emotion for several reasons. First, the measurement of specific dimensions thought to underlie affective states is more practical for zoo research than the approach often employed in studies of discrete emotions, in which individuals are often experimentally subjected to events thought to induce particular emotional states (e.g. forced swim test and depression; Porsolt et al., 1977). Second, one of the measures in the current project, startle reactivity, was developed to reveal underlying moods, or background emotional states. The concept of mood is an important feature of dimensional models of emotion, in which mood states are thought to combine with emotion-inducing events to form the current affective state of an individual.
4 Behavioral and physiological indicators of psychological well-being: Philosophical and historical constraints have hampered discussion of animal affect (Fraser, 2009). The main challenge to studying the subjective experience of animals was famously articulated by Nagel (1974), specifically that we can never truly know what an animal experiences. Although Nagel’s point is valid, indirect measures may be available that can offer us a glimpse of the subjective experience of animals. Changes in behavior or the expression of specific behaviors may be one possible indirect indicator of animal affective state. Some examples include: self-directed behavior in primates (Maestripieri et al., 1992), types of vocalizations (Boinski et al., 1999; Burman et al., 2007), social behaviors (Videan et al., 2007), general activity levels (Markowitz et al., 1995; Shepherdson et al., 1989), stereotypies (Wells and Irwin, 2008), or self-injurious behavior (Lutz et al., 2003). These measures are often ideal choices for zoo research as they can be typically collected non-invasively and with relatively simple equipment. Although identifying affective states are not always the direct purpose of these behavior studies, some authors have suggested that positing affective states can offer a more parsimonious explanation of behavior (Fraser, 2009). Preference tests, although not directly a measure of welfare, may offer additional insight into aspects of the environment or husbandry that can influence welfare. Although other behavior measures could only offer a correlative means of assessing welfare, preference tests allow researchers to more directly “ask” the animal what it perceives as best. In preference testing, the animal is presented with alternatives and the motivational difference for one alternative over the other is used to determine their preference (Kirkden and Pajor, 2006). Motivation describes the strength of the animal’s
5 willingness to either obtain a reward (positive motivation) or avoid an aversive stimulus (negative motivation). The two main forms of preference tests are choice tests, in which the animal is provided with various options and their response is measured, and operant tests, whereby the animal is given control over some stimulus through the use of an operant device (e.g. lever/button pressing). However, it should be noted that although preference tests offers an additional means of assessing welfare, its underlying assumption, that animals prefer things beneficial to them, may not always be the case as is evident in animal models of drug addiction or obesity. Relying on behavior to assess animal welfare has limitations as animals may not always express their affective state overtly. Assessment of physiological measures may provide additional insight of into an animal’s current welfare state (Möstl and Palme, 2002). One common approach is to measure activity of the hypothalamic-pituitaryadrenal (HPA) axis, also known as the stress pathway. After recognition of a stressor, defined as any real or perceived threat to the behavioral or physiological homeostasis of an individual (Moberg and Mench, 2000), corticotropin-releasing hormone is produced by the hypothalamus which stimulates adrenocoticotropic hormone synthesis in the anterior pituitary and promotes the release of glucocorticoids by the adrenal glands. Glucocorticoids are a class of steroid hormones that serve to liberate stored energy and, in the case of a stressor, adaptively respond through fight or flight behaviors. These hormones, typically cortisol or corticosterone, are excreted from the body through urine and feces, allowing non-invasive measurement of HPA activity (Möstl and Palme, 2002). Unfortunately, there are several key limitations to the use of cortisol as a measure of welfare (Mormède et al., 2007). First, HPA activity may follow a circadian rhythm for
6 some species, making the time of cortisol measurement important (Sousa and Ziegler, 1998). Also, cortisol is not only secreted in response to negative situations but also in pleasurable situations as well (Colborn et al., 1991). Under chronic stress, the reactivity of the HPA axis can be reduced, making low glucocorticoid levels hard to interpret (Mizoguchi et al., 2003). Finally, there may be differences of HPA activity based on individual factors such as sex, species, and social rank (Touma and Palme, 2005). These challenges highlight the importance of validating the measurement of glucocorticoids for use in a given species (Touma and Palme, 2005). Choice and control It has been argued that providing animals with a degree of choice or control may be important for promoting positive welfare (Dawkins, 2004). Although there may be some inherent psychological benefit of new choices, additional options also allow animals to seek features that are beneficial and avoid ones that are not. As mentioned previously, one solution has been to incorporate operant devices, allowing the animals to directly control the stimulus (Hanson et al., 1976; Novak and Drewsen, 1989). However, providing active control over the environment may not always be practical in zoos (e.g. visual presence and noise of zoo visitors). In these situations, it may be more feasible to provide zoo animals with a passive form of control by allowing access to off-exhibit areas (Owen et al., 2005; Ross, 2006). When giant pandas were given access to off-exhibit areas, they exhibited lower levels of behaviors indicative of agitation and had lower urinary cortisol concentrations, compared to baseline periods with no access (Owen et al., 2005). When allowed access, pandas spent a mean of 32.8% of their time in off-exhibit areas, although this varied
7 considerably between individuals (5-63%). As the authors note, levels of agitationrelated behaviors were relatively low during baseline conditions (3.5%) compared to other studies (Swaisgood et al., 2001), suggesting that their exhibits met most of their needs. Although the authors discuss many possible motivations individuals may have had for using off-exhibit areas (e.g. avoid noise and proximity to visitors, regulate thermal stress, etc.), the actual reasons individuals used these areas remains unclear. Also, given the large individual differences in the amount of time spent off exhibit, these motivations may have been greater for some individuals than others. Access to off-exhibit areas also appeared to improve the welfare of a pair of zoohoused polar bears (Ross, 2006). Both bears reduced levels of pacing and other stereotypic behaviors. Unlike the giant pandas discussed previously, pacing and other stereotypic behaviors under baseline conditions accounted for a moderate portion of the activity budget of these polar bears. Surprisingly, the bears did not spend much time offexhibit, with the percentage of time not visible increasing from 2.1% in the baseline condition to only 4.3% when given access to off-exhibit areas. The number of visitors at the exhibit and ambient temperatures were consistent across conditions, suggesting these factors did not motivate the behavioral change. These results suggest that, for these animals, the choice to access off-exhibit areas may have been more important than spending time in those areas. As Ross (2006) points out, off-exhibit areas were less enriched than the exhibit space but may have still provided retreat opportunities that benefitted the polar bears. Although no study has yet investigated the effect of off-exhibit access for zoohoused callitrichines, Badihi (2006) did find that when laboratory-housed common
8 marmosets were given free access to an outdoor space their use of this space was dependent on the temperature and weather outside. These results suggest that, if provided with a choice, features of the physical environment, may play a role in motivating animals to use specific areas. Affective startle modification Modulation of the startle reflex may represent an additional but largely unexplored measure of welfare (Paul et al., 2005). In both people and animals, emotional processes have been shown to be involved in the startle response. For example, it is well documented that fear can increase the startle response of people and animals (“fearpotentiated startle effect; Lang et al., 2000). Environmental stimuli may also be able to modulate the affective state and thus the startle responsiveness of people and animals. Nature sounds and music have been shown to modify the startle response of people, with unpleasant sounds causing an increased startle response to a loud startle sound compared to pleasant sounds (Bradley and Lang, 2000; Roy et al., 2009). Furthermore, affective startle modification has been shown to depend on the valence of the environmental stimuli, a phenomenon called “affect matching” (Lang et al., 1990). For instance, participants that viewed a gunshot victim had an increased startle response compared to subjects that viewed an image of an attractive nude (Vrana et al., 1988). Much less work has been done on the modification of the startle response of animals to environmental stimuli. Hoffman and Fleshler (1963) found that when they introduced a continuous white noise background to their startle experiments, rats showed a heightened startle response. Subsequent research has shown that startle sensitivity follows an inverted-U pattern with the amplitude of the background sound, an effect that
9 has been attributed to increasing arousal from increasing background sound intensities eventually being countered by masking/ attenuation of the startle stimulus under very loud background sound intensities (Ison and Hammond, 1971). Although it has not been used directly as an indicator of animal welfare, startle reactions are measured in studies of predator recognition and anti-predator behavior. For example, Searcy and Caine (2003) found that white-fronted marmosets were more startled by the playback of hawk calls compared to the playback of raven calls and the sound of a power drill. In another study, the duration of freezing in response to hawk calls was compared between right- and left-handed marmosets (Braccini and Caine, 2009). These authors found that left-handed marmosets displayed a prolonged freezing response compared to right-handed individuals, suggesting that hemispheric specialization of the brain may influence aspects of temperament. Taken together, these studies suggest that if exposure to sounds or other stimuli in the zoo environment influenced an animal’s affective state, this should modify their startle response. Specifically, individuals experiencing a negative affective state would be expected to have an exaggerated startle reaction whereas individuals experiencing a positive affective state would have a blunted startle reaction and quicker return to baseline behaviors, although this remains to be tested. Sound as Enrichment Overview of environmental enrichment Environmental enrichment is an important method for improving the welfare of captive animals, and is legally mandated for the psychological well-being of captive
10 primates by the Animal Welfare Act. Environmental enrichment can be defined as modifications that increase environmental complexity and confer psychological benefits (Carlstead and Shepherdson, 2000). This complexity can occur in the physical, temporal, or social environment (Carlstead and Shepherdson, 2000). Thus, a well-provisioned environment should allow animals various opportunities, be it interacting with another animal, investigating objects, or just by offering a complex exhibit space to explore. Sensory enrichment One method of increasing the physical complexity of the environment is through sensory enrichment. This form of enrichment often involves incorporating auditory, olfactory, or visual stimuli. Lutz and Novak (2005) term this “passive” enrichment, to differentiate the level of interaction from the “active” enrichment of toys and foraging devices that require manipulation. As this difference in interaction levels suggest, behavioral changes may be more difficult to observe. Thus, Schapiro and Bloomsmith (1995) compared physical, feeding, and sensory (video) enrichment and concluded that sensory enrichment had little impact on the behavior of singly-housed rhesus macaques. Although no behavioral effect of sensory enrichment was observed, it is hard to conclude that this enrichment has little utility as it may effect a more subtle psychological appraisal of the environment that might not be detected through overt behavioral sampling. Animal welfare is complex and evaluating the influence of sensory enrichment on welfare may benefit from psychological indicators of efficacy. In a review of sensory enrichment, Wells (2009) found generally positive behavioral effects of auditory, olfactory, and visual stimulation. Common methods of auditory stimulation involve music or natural sounds from ambient habitats or animal
11 vocalizations. Olfactory stimuli have included biologically-relevant scents such as from a predator, prey, or pheromones from a conspecific animal. Less relevant scents have also been employed, such as lavender, essential oils, or spices. Finally, visual enrichment studies have also investigated biologically-relevant images such as that of conspecifics in addition to less relevant images such as heterospecific animals that may not be encountered in the wild. As positive results have been documented in both biologically-relevant and nonrelevant sensory enrichment studies, it may be unlikely that biological relevance is important for enrichment success (Wells, 2009). Although biological relevance may not be critical, some evidence does suggest that sensory enrichment that stimulates the dominant senses that an animal uses has greater potential for positive welfare benefits. For instance, olfactory stimulation of primates, whose dominant senses are vision and audition, has shown little success in enhancing animal well-being (Wells et al., 2007). Finally, most studies have only been over short time periods and the long-term benefits of sensory enrichment, specifically auditory and olfactory enrichment, remain unresolved (Wells, 2009). Humans Before discussing how auditory enrichment can benefit captive animals, it is important to briefly consider the effect music and other sounds can have on us. Music is known to have existed at least 40,000 years ago and is shared throughout all human cultures (Fitch, 2006). Although the evolutionary origins for music are unclear, its ability to profoundly affect us is certain. Upon hearing music, the emotional response is almost immediate, with music excerpts of a second or less sufficient to cause an emotional
12 response (Bigand et al., 2005; Peretz, 2001). Although some studies have found effects of music experience or culture, the reflexive effect of music on our emotions suggests some processes may be innate. Research on infants supports this claim, with studies showing that infants are able to perceive some aspects of music similar to adults (Trehub and Hannon, 2006). In Western cultures, differential effects of music genre have been noted, with heavy metal and rock music typically increasing anxiety and arousal and classical music lowering arousal (Kemper and Danhauer, 2005). These genre differences have been correlated to some properties of the sounds. Specifically, tempo and rhythm have been implicated in modulating arousal levels (Gomez and Danuser, 2007). Some of the beneficial effects of positive music genres include reductions in subjective measures of anxiety and decreased sympathetic nervous system activation (Kemper and Danhauer, 2005). Knight and Rickard (2001) explored the ability of music to mediate the anxiety of undergraduates preparing for an oral presentation. In their study, exposure to Pachelbel’s Canon in D major reduced heart rate, blood pressure, and salivary cortisol measures in addition to decreasing anxiety as measured through the State-Trait Anxiety Inventory. Khalfa et al. (2003) also documented a decrease in salivary cortisol during a similar procedure. The beneficial effects of music are so widely recognized that they now serve an important role in medical therapy programs (Evans, 2002; Kemper and Danhauer, 2005). Music therapy has been used to treat a variety of both acute and chronic medical illnesses (e.g. Updike and Charles, 1987; Palakanis et al., 1994; Koger et al, 1999). Especially relevant to the present project is the use of music in laboratory-based mood induction procedures (Gerrards‐Hesse et al., 1994; Westermann et
13 al., 1996). Although music is not as successful as other mood induction procedures, it is one of the most simple, as most procedures require some type of instruction with some featuring complex cognitive appraisals (Westermann et al., 1996). Compared to music, our knowledge of the effect of nature sounds is limited. Gomez and Danuser (2004) documented that exposure to the sounds of a stream with birds elicited similar levels of arousal as classical music pieces. Furthermore, nature sounds have also demonstrated utility in therapy applications (Diette et al., 2003). Beneficial effects of aesthetically pleasing nature scenes have also been documented (Kweon et al., 2008; Ulrich et al., 1991). Music as auditory enrichment for animals Given the benefits to humans, it comes as no surprise that there has been considerable interest in whether music can be used to benefit captive animals. Indeed, numerous studies have documented positive behavioral and physiological effects of music in a diverse array of taxa. Classical music exposure, in particular, has shown potential for auditory enrichment. For instance, Wells et al. have documented positive effects of classical music stimulation for diverse species. Classical music correlated with decreased barking and increased resting in kennel dogs (Wells et al., 2002), a nonsignificant trend to increased resting and decreased abnormal behaviors in zoo-housed western lowland gorillas (Wells et al., 2006), and a decrease in stereotypic behavior of zoo-housed Asian elephants (Wells and Irwin, 2008). A decrease in abnormal behaviors was also observed in rhesus macaques exposed to classical music (O'Neill, 1989). Also, Howell et al. (2003) reported an anecdotal decrease in aggressive displays in one laboratory housed chimpanzee during classical music exposure. These results support an
14 overall calming effect of classical music. Classical music has also been shown to stimulate positive physiological effects on fish and rats (Papoutsoglou et al., 2007; 2008; 2010; Akiyama and Sutoo, 2011). Interestingly, several studies have documented greater benefits from classical music of the common practice period, including the Baroque, Classical, and Romantic eras, as compared to music of the modern and contemporary era (Lemmer, 2008; Watanabe and Nemoto, 1998). This suggests that musical characteristics of different eras may be important and classical music should not be viewed as a single monotypic genre. For simplicity, I will use “classical music” throughout to refer to music from the common practice period. Other musical genres have also occasionally shown potential for positively influencing captive animals. Country music has shown some success for enriching dairy cows (Uetake et al., 1997). Radio music has also shown some potential for enrichment. Brent and Weaver (1996) reported that a radio tuned to an oldies station lowered the heart rate of laboratory housed olive baboons but showed no behavioral effect. In two studies on laboratory housed rhesus macaques, animals were given active control over the playback of musical stimuli. In a study by Markowitz and Line (1989), five female rhesus macaques housed in colony rooms were given a lever that could activate or deactivate radio music (no genre specified). The rhesus continued to play radio music throughout the 20 month study duration. Also, the macaques played the radio music for an average of 50% of the day. Novak and Drewsen (1989) reported similar results with laboratory-housed rhesus macaques given control over jazz music. Over the course of the 10 week study, the monkeys played the jazz music for roughly
15 50% of the time provided (two hour periods). Furthermore, Novak and Drewsen documented an increase in affiliative behavior but no effect on urinary cortisol measures. These studies have highlighted behavioral and physiological benefits of music exposure for animals, similar to what has been reported for humans. However, a note of caution needs to be made regarding the musical preferences of animals. As may be expected from this discussion, some of the mechanisms underlying music perception are not unique to humans and shared with a variety of animals. For instance, rhesus macaques, but not songbirds, have demonstrated the ability to generalize octaves (Wright et al., 2000). In addition, a wide variety of species have demonstrated the ability to distinguish between consonance and dissonance, or harmonically pleasing and displeasing sounds to humans (e.g. classical music vs. metal music; McDermott and Hauser, 2005). Humans show clear preferences for consonant sounds over dissonant ones. However, in a y-maze preference task, cotton-top tamarins showed no preference for consonance, although presumably they could distinguish consonance and dissonance (McDermott and Hauser, 2004). This questions the degree to which musical preferences, but not necessarily perception, are shared among humans and animals. An additional feature of music that has been implicated in animal preferences is the tempo of the music. Specifically, Videan et al. (2007) reported a greater decrease in agonism in laboratoryhoused chimpanzees exposed to slower tempo music (easy-listening) as opposed to faster tempo music (classical). McDermott and Hauser (2007) reported a similar preference in tamarins and marmosets for slower tempo music (lullabies) over fast tempo music (techno) using a y-maze preference task. In addition, the authors also observed this preference using artificial audio streams of clicks, thus reducing the possibility of the
16 monkeys responding to other musical features. However, when humans, tamarins, and marmosets were given the choice of slow tempo music (flute lullaby, sung lullaby, Mozart) or silence, monkeys, unlike the human subjects, preferred silence over all three musical pieces. A similar preference for silence over music has been reported for rats (Krohn et al., 2011). Furthermore, music has in some cases been shown to cause stressful reactions. For example, laying hens were more fearful when played classical music compared to hens exposed to normal barn noises, as indicated by an increase in tonic immobility for the classical music condition (Campo et al., 2005). Loud radio music (70 – 80 dB) has been shown to increase the salivary cortisol of marmosets (Pines et al., 2004). Although it is unclear if the animals in these studies found the specific sounds stressful or were reacting to a louder environment, these results do urge caution and highlight the need for additional research before applying auditory enrichment in captive settings (PattersonKane and Farnworth, 2006). In summary, music has been shown to benefit the well-being of animals as evidenced by behavioral and physiological markers. For many species, musical preferences and positive effects of music on animals appears similar to that seen in humans. Generally, classical music has stimulated beneficial influences on welfare. However, our understanding of the mechanisms underlying musical preferences and effects are incomplete. As species-specific differences have been observed in response to music, future research is needed to explore additional sounds for the potential of auditory enrichment. In addition, the effects of music are not always straight forward, as some studies have reported unforeseen negative effects. Music may represent a potential form
17 of auditory enrichment but research should be conducted before its application into the captive environment (Patterson-Kane and Farnworth, 2006). The sounds of nature In contrast to music, the effects of ambient habitat sounds have not been commonly explored (Lutz and Novak, 2005). Three studies have investigated the effect of African rainforest sounds on western lowland gorillas (Ogden et al., 1994; Robbins and Margulis, 2014; Wells et al., 2006). Ogden et al. (1994) observed increased activity in adult gorillas when exposed to the audio playback of rainforest sounds recorded in Cameroon. As the level was greater than expected from studies of wild gorillas, the authors interpreted this response as being indicative of agitation. Surprisingly, infants demonstrated lower levels of clinging, suggesting that they may have been calmed by the rainforest sounds. However, these results are difficult to interpret as decreased clinging by the infants may have been a reflection of the increased activity of adults. Alternatively, the change in the infant’s behavior may have spurred the noted behavioral changes in the adults. Wells et al. (2006) assessed the effects of African rainforest sounds, along with classical music (various artists), and a no audio control condition on gorillas. No behavioral changes were observed under each sound treatment although the authors anecdotally reported the gorillas first reacted to the rainforest sounds with a brief fear response. Most recently, Robbins and Margulis (2014) exposed three gorillas at the Buffalo Zoo to rainforest sounds from a commercial CD, classical music (Chopin), rock music
18 (Muse), and a no audio control period. Playback of rainforest sounds did cause a trend of decreased regurgitation and reingestion but overall was generally similar to the no audio control period. Interestingly, several individuals reacted aversively to classical and rock music exposure, with increased hair plucking and regurgitation and reingestion. Taken together, these three studies on gorillas paint a confusing picture and highlight the inconsistent responses to sounds intended as auditory enrichment. Rainforest sounds have also been evaluated as enrichment for several nocturnal species housed together (Clark and Melfi, 2012). Clark and Melfi observed decreased species-typical foraging behaviors in bush babies and sloths. Alarmingly, the bush babies spent more than 90% of their time in nest boxes when the rainforest sounds were playing, which the authors interpreted as a hiding response possibly reflecting neophobia. Clark and Melfi suggest that the sounds may not have been representative of nocturnal periods in the wild and defend the use of auditory enrichment to potentially mask visitor noise. However, the biological relevance of nature sounds to animals born in captivity is questionable as these sounds are likely novel (Newberry, 1995). Habitat sounds are often incorporated into zoo buildings to enhance the immersion experience of guests and additional research on the impact of these sounds on zoo animals is warranted. In addition to habitat sounds, the playback of animal vocalizations may represent an additional form of auditory enrichment. Positive effects have been documented for both conspecific (Rukstalis and French, 2005; Shepherdson et al., 1989) and heterospecific (Markowitz et al., 1995) vocalizations. For example, playing temporarily isolated marmosets the vocalizations of their social partner correlated with lower urinary cortisol values (Rukstalis and French, 2005). However, although animal vocalizations
19 are likely to cause a behavioral change, repeated exposure to the same vocalizations will likely cause the sounds to lose biological meaning for the animals and reduce its potential effectiveness as auditory enrichment. Species-specific music These inconsistencies of music and habitat sounds as auditory enrichment highlight the need for identifying more meaningful sounds for animals. One promising form of auditory enrichment that has recently been developed is species-specific music (Snowdon and Teie, 2010; Snowdon et al., 2015). Species-specific music is composed using characteristics of the species such as heart rate, hearing range, and features of vocalizations thought to convey affect. Snowdon and Teie first demonstrated the potential of species-specific music by developing music of cotton-top tamarins. Spectral and temporal features of affiliative and fear or threat vocalizations were used to compose two types of musical pieces. For example, long, tonal, pure-tone notes observed in affiliation vocalizations and rapid, staccato notes observed in fear/threat vocalizations were used in the composing the tamarin music. The final pieces were then performed using a cello and singing by David Teie and the final recordings were then shifted to higher frequencies to reflect the hearing sensitivity of tamarins. Seven pairs of cotton-top tamarins were observed before and after presentation of a 30 s excerpt of affiliation- and fear/threat-based tamarin music and equivalent human music. In the five min after exposure to the fear/threat-based tamarin music, tamarins moved more and displayed more anxious behaviors (e.g. piloerection, urination, scent marking, head shaking, and stretching) compared to the affiliation-based music, which the authors interpreted as a negative reaction. Snowdon and Teie also observed increased
20 social behaviors (e.g. grooming, huddling, and sex) following fear/threat-based tamarin music and, although this was not predicted, the authors suggest this may have indicated social comforting following a threatening stimulus. Compared to baseline measures, tamarins exposed to the affiliation-based tamarin music moved less, foraged more, and showed a trend to decreased social behavior. Tamarins oriented to the speakers more after exposure to the fear/threat-based tamarin music compared to baseline conditions, possibly reflecting heightened vigilance. Tamarins moved less following the human fear/threat-based music and displayed less anxious behavior following the human affiliation-based music. These results suggest that tamarin-specific music had greater effects than human music and that specific components of vocalizations can be used in species-specific music to modulate the affective response of the animals. More recently, Snowdon et al. (2015) demonstrated that domestic cats were more attentive to catspecific music compared to human music. Mechanism of auditory enrichment It is important to consider the mechanisms by which auditory enrichment, natural or otherwise, exert an enriching effect on captive animals. The two common mechanisms discussed in the animal literature are masking effects (Ogden et al., 1994; Tromborg, 1999) and neurophysiological effects (Wells, 2009). The masking hypothesis suggests that auditory enrichment may be beneficial by masking, or decreasing the perceptibility, of startling sounds arising in the animals’ environments, such as from zoo visitors or machinery (Ogden et al., 1994; Patterson-Kane and Farnworth, 2006; Wells, 2009). This suggestion likely stems from the successful application of masking sounds for humans. For example, white noise and water sounds have been used in open-plan office settings to
21 reduce distracting sounds (Haapakangas et al., 2011; Loewen and Suedfeld, 1992). Water sounds (e.g. fountains, waterfalls) have also been shown to be successful at masking traffic noise (De Coensel et al., 2011; Galbrun and Ali, 2013; Jeon et al., 2010; Nilsson et al., 2010). Interestingly, De Coensel et al. (2011) found that participants rated fountain sounds better at masking steady traffic noise but bird sounds were better for variable traffic noise. The authors suggest that variable sounds may provide a better masking sound as they may draw attention away from the noise. This finding suggests an alternative hypothesis – that auditory enrichment may be beneficial by simply distracting attention from variable sources of noise. Currently, no study has demonstrated positive effects of auditory masking for animals. The study by Ogden et al. (1994) evaluated the potential of the rainforest sounds to mask sounds of caretakers or bonobo vocalizations, with masking sounds causing agitation in the gorillas. Tromborg (1999) explored the effects of various background sounds on laboratory-housed California ground squirrels as part of his dissertation research. He identified a longer latency to emerge from nest boxes when exposed to thunderstorm sounds, a response he argued was the result of auditory masking promoting cautious behavior in the squirrels. As discussed above, when researchers studying the startle response of animals applied white noise in the background of their experiments in an attempt to mask exogenous sounds, they found the startle response was enhanced and auditory masking was only achieved under very loud background sounds (>80 dB; Hoffman and Flesher, 1963, Ison and Hammond, 1971). Additionally, as Wells (2009) notes, the masking hypothesis cannot explain the differential effects of music genres that have been reported, as one genre is not more clearly masking than another.
22 Additional research is needed to determine whether auditory masking can benefit animals. An additional hypothesis on the benefits of auditory enrichment is that of a specific neurophysiological effect of audio exposure (Wells, 2009). The underlying assumption of this hypothesis is that some general perceptual mechanisms may be shared between animals and humans and lead to the positive effects of music found across species. Neurophysiological effects of music have been described in mice. After exposure to music, brain-derived neurotrophic factor (BDNF) increases and this has been linked to the anxiolytic effects of music observed in mice with a BDNF polymorphism (Angelucci et al., 2007; Li et al., 2010). Importantly, these changes were not observed when mice were exposed to white noise, suggesting the anxiolytic effect is based on characteristics of the sound and not just on auditory stimulation (Li et al., 2010). Neurophysiological reactions to specific features of sound may be important for identifying both danger and safety and have an innate neurological basis. The brain stem is important for rapidly responding to specific features of sound that may signal danger, such as loudness, suddenness, low pitch, and high levels of dissonance and this has been implicated as one potential primal mechanism underlying the emotional response of humans to music (Juslin and Västfjäll, 2008). For example, in developing warning sounds for use in various applications, researchers have found fast temporal patterns are important for signaling urgency to listeners (Suied et al., 2008). The innate response to these features may have an ultimate explanation in how affect is conveyed through conspecific vocalizations, which is the mechanistic basis for species-specific music (Juslin and Laukka, 2003; Snowdon and Teie, 2013). Aspects of predator vocalizations
23 may also trigger an innate response, although some degree of learning may be necessary to correctly identify potential predators (Friant et al., 2008; Searcy and Caine, 2003). Although Lenti Boero and Bottoni (2008) speculated that there may be an evolutionary preference for certain habitat sounds, response to these sounds may also require experience and thus, not carry biological meaning for captive-reared individuals. The response of animals to auditory enrichment is complex and beneficial effects will likely depend on acoustic features of the sounds and/ or their ability to mask aversive noise. Noise and its impact on animals How anthropogenic noise affects free-ranging wildlife With an ever-expanding human population, the impact of noise from humanrelated activities, or anthropogenic noise, on wildlife has become an emerging conservation concern (Barber et al., 2010; Francis and Barber, 2013). For example, extensive research has shown that noise may interfere with an animal’s ability to detect the vocalizations of conspecifics (Bee and Swanson, 2007), causing many species to shift aspects of their calls to adapt (Slabbekoorn and Ripmeester, 2008). However, some species may be limited in their ability to adapt their communication systems to noise (Hu and Cardoso, 2010). Although the long-term fitness consequences of vocal adjustments have not been fully explored, some evidence suggests these changes may influence female mate choice and male-male competition (for review of this topic for avian species, see Patricelli and Blickley, 2006). In addition, noise may interfere with the ability to detect predators, promoting increased vigilance and anti-predator behaviors (Quinn et al., 2006). For example,
24 ground squirrels living near a wind turbine exhibited elevated levels of vigilance and increased likelihood of returning to their burrows in response to the playback of conspecific alarm calls compared to ground squirrels living in a quiet area (Rabin et al., 2006). Noise may also adversely affect the hunting ability of predators to locate prey (Siemers and Schaub, 2011). One response to these potential challenges is to avoid noisy areas. For example, greater sage grouse have been shown to avoid breeding lek areas during the experimental playback of natural gas drilling noise, and the animals that remained demonstrated elevated fecal glucocorticoid levels, an effect the authors attributed to masking of communication or increase perception of predation risk (Blickley et al., 2012; Blickley and Patricelli, 2012). Duarte et al. (2011) observed marmosets living in an urban park to avoid the noisy periphery of the park, including areas with high food availability. This finding also raises questions about the degree to which animals habituate to frequent noise exposure. Francis and Barber (2013) outlined a framework for understanding the effects of noise, the disturbance-interference continuum. In their view, the impact of noise varies along a temporal gradient, with sudden, infrequent, and unpredictable sounds being perceived as a threat and causing a startle response and anti-predation behaviors (riskdisturbance hypothesis, see Frid and Dill, 2002), whereas chronic noise is suggested to cause interference of cues through auditory masking. They propose the severity of the response may depend on temporal, intensity, and frequency characteristics of the sound. Alternatively, Chan et al. (2010) have shown evidence that noise may just act as a distraction (distracted prey hypothesis). Although these hypotheses are based on the
25 responses of free-ranging wildlife, they may also provide a valuable framework for understanding the impacts of noise on wildlife in a zoo setting. Comparing the sound environment in the wild and captivity The potentially negative effects of noise on animals is concerning for zoos, which may be louder than wild habitats. In a study of sound levels in natural habitats, Waser and Brown (1986) found that rainforest habitats (Kakamega Reserve, Kenya and Kibale Reserve, Uganda) ranged from 27 dB to 40 dB. Sound levels recorded in zoos are typically much greater than those wild estimates. Ogden et al. (1994) reported background sound levels near a gorilla exhibit was approximately 40 dB, however when the ventilation system was activated sounds exceeded 58 dB. Sound levels ranging 62 dB to 72 dB, with an average of 70 dB, were recorded at two California zoos, with the higher values relating to visitor activity (San Francisco Zoo and Sacramento Zoo, Tromborg and Coss, 1995). More recent studies of noise in zoos have generally documented sound levels ranging between 55 – 70 dB (Cooke and Schillaci, 2007; Owen et al., 2004; Powell et al., 2006; Quadros et al., 2014). Noise in zoos A growing body of research has identified adverse effects of noise in the zoo environment. Owen et al. (2004) were the first to quantitatively describe the impact of noise on zoo animals. They measured the behavioral and hormonal responses to ambient noise by two zoo-housed giant pandas housed at the San Diego Zoo. The pandas displayed increased door scratching, locomotion, stress-related vocalizations, and urinary cortisol levels on days categorized as loud compared to quiet days. Powell et al. (2006)
26 documented a similar response to construction noise by pandas housed at the National Zoo. Most recently, Quadros et al. (2014) performed a detailed observational study of noise at the Belo Horizonte Zoo in Brazil. They found sound levels in exhibits were influenced by increasing numbers of visitors, the popularity of the species, and the style of the exhibit with open circular exhibits that permitted the greatest number of visitors being the loudest. Although no behavioral differences were observed in relation to noise, the authors did document increased vigilance in response to visitors for multiple species. Several studies have evaluated visitor noise experimentally. Birke (2002) manipulated visitor noise by asking volunteers standing in front of an orangutan exhibit to either watch quietly or talk loudly. In the loud visitor condition, the orangutans looked at the public more and infants increased the time spent holding onto parents, suggesting that loud visitors were a stressor. Larsen et al. (2014) investigated the effect of visitors on koalas under standard conditions as well as during audio playback of visitor noise. The koalas responded to both increasing visitor numbers and louder noise treatments with heightened vigilance. Noise is an inevitable part of the zoo environment and an unavoidable stimulus for the animals housed within. These results suggest that noise can negatively impact the lives of zoo animals and solutions to this dilemma are unclear. Noise, at suitable levels, may beneficially increase the complexity of the zoo environment. Although technically challenging, control over noise may ameliorate its negative influence. Using sounds to mask zoo noise may be attractive but has not yet been documented. A better understanding of the effect of the sound environment on the welfare of zoo animals is needed, particularly for sensitive species such as callitrichine monkeys.
27 Callitrichines Natural history of callitrichine monkeys Callitrichine monkeys are a group of small-bodied New World primates native to Central and South America. Although their taxonomy has been disputed, the subfamily Callitrichinae is currently recognized to include six genera: the tamarins (Saguinus), pygmy marmosets (Cebuella), Amazonian marmosets (Mico), and monophyletic callimico (Callimico) residing in the Amazonian basin; and the lion tamarins (Leontopithecus) and Atlantic marmosets (Callithrix) native to the Atlantic forests (Schneider and Sampaio, 2015). These species are distinguished for a number of key characteristics. Most notable is their small body size, being the smallest anthropoids in the world. Although as a group callitrichines are diminutive, body sizes can range six-fold within Callitrichinae – from the pygmy marmoset (Cebuella pygmaea) at a little over 100 g to the golden-headed lion tamarin at almost 700 g (Leontopithecus chrysomelas; Garber, 1992). In conjunction with small body size, callitrichines have clawlike nails (tengulae) on all digits except the hallux – adaptations that reflect their arboreal lifestyle. Callitrichines are also remarkable for their obligate birth of twins, occurring in all species except callimico, and cooperative rearing of young, the only primate beyond humans to exhibit this trait (Fernandez-Duque et al., 2012). Social groups in the wild are generally small (3 to 12 animals) and have a flexible social organization typically involving one female with one or more males and offspring. Mating systems are variable and monogamy, polyandry, and polygamy have all been
28 reported (see Fernandez-Duque et al., 2012 for review). In captivity, callitrichines are housed as monogamous breeding pairs (Anzenberger and Falk, 2012). Why study callitrichines? There are a number of reasons to study the effects of sounds on callitrichine monkeys. First, these species generally live in dense rainforest habitats. As vision may be limited in these environments, this may make these species more reliant on sounds. Second, callitrichines are the smallest anthropoids and thus, prey to numerous predators (Caine, 1993). This predation pressure may make them particularly sensitive to humans. Research in zoos has suggested a negative effect of zoo visitors on these monkeys. Glaston et al. (1984) compared cotton-top tamarins on display and off exhibit. They found reduced social behaviors in tamarins housed on display, including after the two groups were switched housing arrangements. Armstrong and Santymire (2013) similarly compared a pair of pied tamarins housed on display with an off-exhibit pair. They observed decreased grooming and foraging and increased locomotion, use of nest boxes, and fecal glucocorticoid levels in the pair housed on display. This group was also noted to perform stereotypic behaviors. Wormell et al. (1996) also investigated the visitor effect on pied tamarins and found a group that experienced frequent visitation to perform more threat displays, piloerection, and approaches to the front of their cage than groups with lower levels of visitation. A similar pattern has also been observed in laboratory conditions comparing weekdays with frequent human activity to quiet weekend periods (Barbosa and Mota, 2009). Common marmosets were more frequently scent marking and moving, and performed autogrooming for longer periods of time during the busy weekdays vs. weekends. Although Barbosa and Mota (2009) did not
29 identify an effect of human activity on fecal glucocorticoid measures, Cross et al. (2004) has reported increased saliva cortisol in laboratory-housed common marmosets following a disturbance period. The presence of a human observer has even been utilized as a procedure to study fear and anxiety in marmosets (Human Threat test; Costall et al., 1992). Third, callitrichine species have shown a unique response to sounds. This has included an indifference to classical music (McDermott and Hauser, 2007), no preference for consonant (i.e. harmonious) over dissonant sounds (McDermott and Hauser, 2004), and no avoidance of screeching sounds described as being similar to nails on a chalk board (McDermott and Hauser, 2004). In addition, species-specific music has been developed for the cotton-top tamarin (Snowdon and Teie, 2010) and may represent a potential form of auditory enrichment for other callitrichines. Lastly, it is important to consider that callitrichines, not unlike other species, are living in increasingly close proximity to humans in the wild and, consequentially, greater exposure to anthropogenic noise. For example, the critically endangered pied tamarin has one of the most restricted ranges of any primate and is only found within and near Manaus, Brazil. A greater understanding of how a species responds to sounds in a zoo setting may have insights for their conservation in the wild. Thesis Objectives The goal of this research was to identify enriching and adverse sound environments for zoo-housed callitrichine monkeys, and in the case of the latter, evaluate methods to ameliorate this effect and improve welfare. The general hypothesis was that
30 sounds commonly experienced in the zoo setting can affect the welfare of zoo-housed callitrichine monkeys. My aims were fourfold. First, I sought to develop and validate additional methods for assessing the welfare of these species through hormonal and cognitive assessments.
Second, I evaluated whether sound from a nearby waterfall
feature provided beneficial auditory masking.
Third, I investigated the potential of
sounds as auditory enrichment for these monkeys. Fourth, I explored the effect of exhibit choices on the response to the sound environment. Chapter Two validates fecal glucocorticoid measures as an indicator of stress in pied tamarins, white-fronted marmosets, golden lion tamarins, and callimico. Fecal samples were assessed before and after a routine veterinary exam to document increased cortisol production from handling and anesthesia. In addition, we evaluated the impact of social changes on fecal glucocorticoid levels. Chapters Three and Four investigate the effect of the waterfall feature on callitrichine monkeys. In Chapter Three, the waterfall feature and access to off-exhibit areas was experimentally manipulated in multiple phases for pied tamarins, white-fronted marmosets, and callimicos. Chapter Four expands on this investigation by carefully manipulating the sound levels in both the exhibit and off-exhibit areas for the pied tamarin group. Chapters Five and Six explore the potential of sounds as auditory enrichment for callitrichines. Chapter Five observed the behavioral response to 30 min exposure of sounds during a quiet period for the pied tamarin, golden lion tamarin, and callimico groups. In addition, access to off-exhibit areas was manipulated during the study to provide a measure of preference and evaluate the effect of exhibit choices on the response
31 to sounds. Chapter Six investigates whether exposure to sounds can modify the startle response of pied tamarins and white-fronted marmosets. This study exposed animals to a brief (5 min) playback of sounds that was immediately followed by an acoustic startle event. This methodology was validated through exposure to loud white noise, a stimuli known to be aversive. This investigation represents a novel method for assessing the affective state of zoo-housed animals.
32 Chapter Two: Fecal glucocorticoid metabolite responses to management stressors and social change in four species of callitrichine monkeys
Introduction Measuring glucocorticoids can provide valuable insight into the health and welfare of animals. Metabolites of these hormones, typically cortisol or corticosterone, can be measured non-invasively through feces using enzyme immunoassay (EIA), providing a valuable tool for monitoring sensitive species with minimal disruption (Keay et al., 2006; Schwarzenberger, 2007). However, as significant variation exists in the metabolism and excretion time of glucocorticoids between species, as well as possible individual differences in hormone production within a species, it is necessary to validate the EIA assay using an event known to activate the HPA axis (Touma and Palme, 2005). One common method involves the administration of adrenocorticotropic hormone (ACTH), a precursor in the HPA axis known to stimulate glucocorticoid production (Heistermann et al., 2006). Although this represents a gold standard, ACTH challenges are not always possible or ideal in some situations. Alternatively, assessing glucocorticoids before and after stressful events, such as capture, restraint, and anesthesia during veterinary exams (Armstrong and Santymire, 2013) or translocation between institutions (Heistermann et al., 2006), can indicate the biological validity of an assay. Also, demonstrating circadian variation of glucocorticoids (Sousa and Ziegler, 1998) or increases in relation to parturition (Murray et al., 2013) may provide an additional means to validate an assay.
33 Fecal glucocorticoid metabolites (FGM) have been previously assessed in some species of the family Callitrichinae, a group of arboreal, small-bodied New World monkeys (common marmoset, Callithrix jacchus: Heistermann et al., 2006; Raminelli et al., 2001; Sousa and Ziegler, 1998; golden lion tamarin, Leontopithecus rosalia: Bales et al., 2005, 2006; pied tamarin, Saguinus bicolor: Armstrong and Santymire, 2013; cottontop tamarin, Saguinus oedipus: Fontani et al., 2014; Ziegler and Sousa, 2002; mustached tamarin, Saguinus mystax: Huck et al., 2005). Although previous reports largely describe similar patterns of cortisol metabolite excretion, some discrepancies have been noted. For example, in a study of female common marmosets, Sousa and Ziegler (1998) demonstrated diurnal variation of cortisol, with higher concentrations detected in feces during the afternoon versus morning. Previous studies on marmosets have documented diurnal variation in both urine (Wied’s tufted ear marmoset; Smith and French, 1997) and saliva (common marmoset; Cross and Rogers, 2004). However, Raminelli et al. (2001) identified elevated FGM during the afternoon in female common marmosets, but not in males. Huck et al. (2005) also observed no diurnal variation of FGM in mustached tamarins. Similarly, elevated FGM concentrations during pregnancy have been reported in golden lion tamarins (Bales et al., 2005) and common marmosets (Ziegler and Sousa, 2002) but not for mustached tamarins (Huck et al., 2005). These differences highlight the need for careful validation of FGM measures for each species, even when dealing with closely related species. Social changes are a common occurrence in these monkeys and these events influence HPA activity. For example, periods of social instability, such as during pair bond formation or separation from a natal group, have been shown to increase cortisol
34 levels (Johnson et al., 1996; Smith et al., 2011; Ziegler et al., 1995). In the wild, these periods are typically brief and resolved through immigration and emigration between groups (Lazaro-Perea et al., 2000). However, in a zoo setting, replacement of individuals within a group after deaths and/or separations is not always immediately possible. Understanding the impact of these social changes is important for animal managers as social instability may impact welfare and the formation of new social bonds (Johnson et al., 1996; Smith et al., 2011). The aims of this investigation were two-fold. The first aim was to validate methods of individual fecal identification and measurement of FGM concentrations in four species of callitrichine monkeys: golden lion tamarin (L. rosalia), callimico (Callimico goeldii), pied tamarin (S. bicolor), and white-fronted marmoset (Callithrix geoffroyi). This is the first study to measure FGM concentrations in white-fronted marmosets and callimicos and the first to compare FGM responses among callitrichine species. Biological validations were conducted using the stress response to veterinary exams. Second, we evaluated the impact of deaths of group members on FGM levels in the pied tamarin and white-fronted marmoset groups. We hypothesized that the death of breeding animals would lead to social instability and thus, alter FGM levels. Methods Subjects and housing This study involved 16 monkeys from four species: golden lion tamarin (n=7), callimico (n=2), pied tamarin (n=3), and white-fronted marmoset (n=4; Table 1). Animals were housed in similar single-species exhibits in the RainForest building of
35 Cleveland Metroparks Zoo. The monkeys were fed in the morning (10 AM) and late afternoon (4:30 PM) a diet consisting of canned marmoset diet (Mazuri, St. Louis, MO), New World primate biscuit (Mazuri, St. Louis, MO), vegetables, fruit, and mealworms. The pied tamarins also received a calcium supplement and the white-fronted marmoset diet included gum arabic. Water was available ad-libitum. Table 1. Demographic background of study subjects. Age (yr)a
Species
ID
Sex
Golden lion tamarin (L. rosalia)
LF1 LF2 LF3 LM1 LM2 LM3 LM4
Female Female Female Male Male Male Male
6.2 1.2 0.3 12.0 3.1 1.2 0.3
Callimico (C. goeldii)
CF1 CM1
Female Male
3.7 10.5
Pied tamarin (S. bicolor)
TF1 TM1b TM2
Female Male Male
6.2 17.5 3.6
MF1b Female 6.3 MM1 Male 8.3 MM2 Male 3.2 b MM3 Male 2.9 a Age was calculated at the time of the veterinary exam for all individuals except MM3, who died prior to the exam. For this individual, age at death is provided. White-fronted marmoset (C. geoffroyi)
b
Individuals that died prior to or during a veterinary exam.
36 Fecal marker assessment Four indigestible items were qualitatively evaluated as fecal markers (Table 2): food dye (McCormick and Company, Inc., Hunt Valley, MD), food coloring paste (Wilton Industries, Woodridge, IL), 2 mm glass jewelry beads (Darice, Strongsville, OH), and non-toxic glitter (Advantus Corporation, Jacksonville, FL). Fecal markers were hand-fed in banana pieces on Monday and Wednesdays afternoons (1-3PM) and first morning void samples were evaluated for evidence of the marker for 48 h. During fecal marker evaluation, one marker type was fed per day for all species and all individuals in a group received the same marker color. Fecal Sample Collection and Processing First morning fecal voids were collected between 7 AM and 9 AM before and after veterinary exams. Based on the results of the fecal marker study, fecal samples were individually identified using glitter for the golden lion tamarins, pied tamarins, and white-fronted marmosets. No fecal marker was used with the callimicos and individual samples were visually identified by an observer as the animals defecated in the morning. In addition, immediately after exams, all fresh fecal samples observed throughout the day were collected for a period of 48-72 h for all species except the white-fronted marmoset. Fecal collection for this species was extended to 7 days because of small fecal size and difficulties collecting first morning voids.
Fecal markers were not fed to the golden lion
tamarins on the morning of the exam, thus fecal samples collected that afternoon that were not witnessed were not individually identifiable and are shown for reference. In addition, on the morning after the exam, fecal samples from the callimicos were present before the lights were turned on and these individually unidentifiable samples were
37 collected and shown for reference. Fecal samples were stored in Whirl-pak fecal bags at -20°C. Feces were lyophilized (FreeZone 4.5 liter freeze dry system, Model: #7751020, Labconco Corporation, Kansas City, MO, USA), pulverized, and sifted to remove food particles and glitter. Hormonal metabolites were extracted using a method adapted from Brown (2008): 5 ml of methanol were added to 0.2 g dry feces, briefly hand vortexed, mixed for 1 h (large capacity mixer, Model: #099A LC1012, Glas-Col, Terre Haute, IN, USA), and centrifuged at 2500 RPM for 20 min (Sorvall Legend RT+). The supernatant was poured off and stored in a -20°C freezer prior to analysis and assayed within six months. Hormonal analysis Fecal glucocorticoid metabolites were analyzed via the enzyme immunoassay (EIA) protocol of Brown (2008) using polyclonal cortisol antiserum (R4866, 1:300,000 dilution) provided by Coralie Munro (University of California, Davis, CA). The EIA was validated for each species by demonstrating: 1) parallelism of the binding inhibition curve of pooled fecal extract dilutions (1:1 to 1:1024) with the cortisol standard (R2>94%), and 2) significant recovery (> 92.0%) of exogenous cortisol from diluted fecal samples spiked with the quality control high (200 pg/well) and low (5 pg/well). The mean inter-assay coefficient of variation (CV) of standards and high- and low-value quality controls was 7.2%. Intra-assay CVs were less than 10%. Microtiter plates were read using a spectrophotometer (Epoch, Bio-Tek, Winooski, VT, USA) at 450 nm wavelength through the Gen5 software (v2.03, Bio-Tek, Winooski, VT, USA).
38 Biological validation Fecal samples were collected before and after a routine veterinary exam for the golden lion tamarins and callimicos. Samples were collected 6 days prior to and 10 days post exam for the golden lion tamarins and 5 days prior to and 7 days post exam for the callimicos. Prior to the exam, the golden lion tamarins could not be reliably identified by researchers or staff and a pooled fecal sample from the group was collected each morning. During the exam, the golden lion tamarins were marked (Flouro Tell Tail Animal Marker, Fil, Mount Maunganui, New Zealand), allowing for individual hand feeding of fecal markers and identification of samples post exam. In addition, fecal samples were collected from the pied tamarins and whitefronted marmosets opportunistically after veterinary exams prompted by the death of a social partner. For the pied tamarins, the entire group (TF1, TM1, and TM2) was captured and transported to the veterinary hospital for an exam on the breeding male (TM1). During this exam, the male (TM1) was euthanized as a result of an age-related health decline, and the female received an exam and injection of Depo-Provera (Pfizer Inc., New York, NY, USA) for contraception. The male offspring (TM2) did not receive an exam and remained in a transport box for the duration of the exam (≈1 h). As a result of elevated FGM levels in afternoon samples from the pied tamarins immediately after the exam, we assessed the potential diurnal variability of FGM concentrations by collecting an additional week of morning (7-8 AM) and afternoon (3-5 PM) fecal samples. Diurnal variation was observed and, therefore, afternoon samples following the veterinary exam were excluded from analysis.
39 The breeding female white-fronted marmoset (MF1) died unexpectedly and a 2 week-old neonate was temporarily removed from the group for hand rearing. The breeding male (MM1) received a physical exam the day after this event; the male offspring (MM2) did not receive an exam and remained in the marmoset exhibit during MM1’s exam. Two days after this exam, the neonate was returned to the holding area for hand rearing procedures. Baseline data were available from 50 days prior to this event. Fecal samples from the marmosets were collected for 10 days post exam. During this time, the neonate was housed in an incubator within audiovisual contact of the marmoset holding area. Staff entered the holding area between 6 AM and 12 AM to feed the neonate. Additional data collection began 19 days after this initial collection period, at which point the neonate was housed in a wire mesh cage attached to the holding cage which allowed for physical contact between the neonate and adult marmosets (MM1 and MM2). Staff entered the holding area between 6 AM-6 PM for hand rearing procedures. Two additional deaths occurred in the white-fronted marmoset group from unrelated causes. An adult male offspring (MM3) died 89 days prior to the death of the breeding female. The breeding male marmoset (MM1) was euthanized 40 days after the death of the breeding female (MF1). Statistical analysis The baseline FGM concentration was calculated for each individual using an iterative process whereby all baseline samples greater than the mean plus 2 standard deviations (SD) were removed, and this statistic was recalculated until no additional samples were greater than this value, at which point the mean was designated as the individual baseline (Brown et al., 1999). A sample’s FGM concentration was considered
40 elevated if it exceeded 2 SD of the individual’s baseline value. For the female callimico (CF1), one elevated sample that occurred prior to the exam was excluded from baseline calculations to generate a representative baseline for this individual but this value was retained in this individual’s validation graph. Baseline FGM levels were compared between the male and female callimicos and pied tamarins using distribution-free confidence intervals (95% CIdf) around the median. As the male pied tamarin’s (TM2) FGM levels appeared to increase following the death of his father (TM1), the male and female’s FGM levels were again compared using baseline data from a research project that began 98 days after the death of the breeding male. We considered a difference significant if confidence intervals did not overlap. Diurnal variation was assessed in the male (TM2) and female (TF1) pied tamarins using descriptive statistics because of small sample sizes. Data were analyzed using SAS University Edition (SAS Institute Inc., Cary, NC). Results Assessment of Fecal Markers Fecal markers were fed in the afternoon (1-3PM) and all were observable by the following morning voids (Table 2). Although this time frame was successful for most species, when the pied tamarins were fed in early afternoon (1PM), evidence of the food coloring marker was observed prior to first morning voids, indicating the marker was excreted before the tamarins entered their nest box for the evening (5-6PM). In addition, when the tamarins were fed glitter, fecal samples would occasionally contain glitter colors fed to different individuals, even though food sharing did not occur during hand
41 feeding sessions. Although not witnessed during this project, coprophagy has been observed in this group. For the pied tamarins, feeding fecal markers later in the afternoon (3PM) was found to be more successful. Consumption of the fecal markers differed drastically between species. The golden lion tamarins and pied tamarins readily consumed all fecal markers, whereas the callimicos and white-fronted marmosets reacted aversively to the food coloring and food paste, often dropping the banana pieces, and would eject the beads out of the side of the mouth. As glitter was the only fecal marker readily consumed by all species, this marker was chosen and subsequently used for the remainder of the study. Table 2. Assessment of fecal markers fed to four species of callitrichine monkeys: golden lion tamarin (GLT); callimico (CA), pied tamarin (PT); white-fronted marmoset (WFM). Fecal Marker
Amount
Colors
Identificationa
Consumption GLT
CA
PT
WFM
Food coloring
0.1 mL
Red, green, blue, yellow
Yes
No
Yes
No
Yellow not observed, other colors identifiable.
Food coloring paste
0.1 mL
Red, green, blue
Yes
No
Yes
No
All colors identifiable.
Beads
5-10 per White, animal pink, orange, yellow
Yes
No
Yes
No
Roughly 10% (WFM and GM) to 40% (PT and GLT) of beads recovered.
Glitter
0.35 g
Green, Yes Yes Yes Yes All colors gold identifiable. a Fecal markers were fed in the afternoon (1-3PM) and assessed the following morning (7-9AM).
42 Changes in FGM levels in relation to veterinary exams and animal deaths In all four species, we observed a significant increase in FGM concentrations 2448 h after a veterinary exam (Table 3). As the magnitude and pattern of increase differed across species, these results are presented separately for each group.
43
Table 3. Summary of fecal glucocorticoid metabolite response of four callitrichine species after veterinary exams. Speciesa
Peak FGM b, d (ng/g)
ID
Baseline FGM (ng/g)
Golden lion tamarin (L. rosalia)
LF1 LF2 LF3 LM1 LM2 LM3 LM4
481.5 380.6 1041.8 751.3 218.6 397.5 791.2
Callimico (C. goeldii)
GF1 GM1
492.9 189.4
1550.7 [2649.5] 600.4 [2649.5]
Pied tamarin (S. bicolor)
TF1 TM1
123.5 62.9
2627.9 266.7
2383.2 4087.0 3688.8 5748.7 1132.9 9576.5 6837.4
[13639.1] [13639.1] [13639.1] [13639.1] [13639.1] [13639.1] [13639.1]
Magnitude c, d Increase 4.9 10.7 3.5 7.7 5.2 24.1 8.6
Elevated Lag (h)
Peak b Lag (h)
[28.3] [35.8] [13.1] [18.2] [62.4] [34.3] [17.2]
31.7 22.3 25.6 6.0 23.8 4.4 24.4
31.8 26.8 25.6 6.0 23.8 4.4 24.4
3.1 [5.4] 3.2 [14.0]
19.5 19.7
20.3 21.3
20.2 20.2
20.2 20.2
21.3 4.2
White-fronted marmoset MM1 32.7 4822.7 143.1 24.3 238.3 (C. geoffroyi) MM2 40.8 798.3 19.6 48.3 165.3 a Pied tamarin and white-fronted marmoset also experienced social changes that included the death of a breeding individual and for the white-fronted marmoset, hand-rearing of a neonate in a nearby area. b
Data represent the initial FGM peak.
c
Magnitude represents the fold increase between baseline and peak FGM levels following veterinary exam.
d
FGM concentrations of individually unidentifiable fecal samples collected 5-20 hrs following the exam are shown in brackets for reference.
44 Golden lion tamarin After capture and anesthesia during a routine veterinary exam, all golden lion tamarins demonstrated elevated FGM levels within 48 h post-exam (Fig. 1; representative individuals). In addition, all individually unidentifiable samples collected the afternoon of the exam were elevated above all individuals’ baseline FGM values. A secondary elevation in FGM levels, occurring at roughly 48-72 h post-exam, was observed in multiple individuals (LF1, LF2, LM2 and LM3; see Fig. 1). The magnitude of the primary FGM increases varied considerably across individuals (~4 to 24-fold increase compared to baseline), with a mean increase for the group of 9-fold above baseline. Individual FGM had returned to baseline levels by 24-48 h.
45
Figure 1. Fecal glucocorticoid levels of four golden lion tamarins (A: LF1; B: LF2; C: LM1; D: LM2) after capture and anesthesia during a routine veterinary exam (time 0 h). Pooled samples from the group collected before the exam (grey circles) are shown for reference but were not used in individual baseline calculations. Baseline and baseline + 2SD are indicated by solid and dashed lines, respectively. Individually unidentifiable samples collected the afternoon of the exam are shown for reference and denoted with an X (see Methods).
46 Callimico Both the male and female callimico showed a peak FGM level roughly 20 h post exam (Fig. 2). The magnitude of this peak was similar for both animals: 3.2 and 3.1 times greater than baseline for the male and female, respectively. By 24 h post exam, FGM had temporarily returned to baseline concentrations for both individuals. Secondary peaks of comparable magnitude to the primary peak were also observed (Male: 43 h and 68 h post exam; Female: 68 h post exam).
47
Figure 2. Fecal glucocorticoid metabolite concentrations in a female (A) and male (B) callimico before and after a routine veterinary exam (time 0 h). Baseline and baseline + 2SD are indicated by solid and dashed lines, respectively. Individually unidentifiable samples collected the morning after the exam are shown for reference and denoted with an X (see Methods).
Pied tamarin After a veterinary exam on the breeding male (TM1) and female (TF1) pied tamarins, during which the breeding male was euthanized due to age-related health declines, the breeding female pied tamarin and her adult male offspring (TM2) demonstrated elevated FGM concentrations post exam, with peak concentrations
48 occurring at 20 h for the breeding female and her male offspring (Fig. 3). The breeding female’s FGM peak after the exam was 21.3 times greater than baseline. The male offspring, who was transported to the veterinary hospital but not examined, exhibited a lower magnitude FGM increase of 4.3 times greater than baseline. Both individuals demonstrated a return to baseline FGM levels within 48 h.
Figure 3. Fecal glucocorticoid concentrations of a female pied tamarin (A) and her adult male offspring (B) before and after the death of the breeding male and veterinary exam (time 0 h). Baseline and baseline + 2SD are indicated by solid and dashed lines, respectively.
49 White-fronted marmoset Fecal samples from both male marmosets displayed a progressive increase in FGM levels following a veterinary exam on the breeding male (MM1) prompted by the death of the breeding female (MF1) and temporary removal of a neonate (Fig. 4). The neonate was returned to the marmoset holding area 2 days after the exam and housed in an incubator for hand rearing procedures. The progressive FGM increase reached a peak magnitude for individual MM1 that was 143 times greater than baseline (11 days post exam) and 19.6 times greater than baseline for individual MM2 (8 days post exam). Elevated samples began 24 h post exam for the breeding male and 48 h for the male offspring, who did not receive an exam but did experience a capture event. Fecal samples 27 days post exam showed a return to near baseline FGM levels, although a consistent return to baseline levels was not observed for either individual. No FGM increase was observed after the death of an adult male offspring (MM3) or after the death of the breeding male (MM1).
50
Figure 4. Fecal glucocorticoid concentrations of two male white-fronted marmosets (A: MM1; B: MM2) before and after the death of the breeding female, temporary removal of a neonate, and subsequent veterinary exam (arrow at time 0 h). The neonate was returned to the off-exhibit holding area after the veterinary exam (arrow at time 48 h). An adult offspring died before the veterinary exam validation and is shown for reference (arrow at time -2124 h). Individual MM1 died during data collection and this event is noted for MM2 (B; arrow at time 924 h). Baseline and baseline + 2SD are indicated by solid and dashed lines, respectively.
Factors influencing fecal glucocorticoid metabolite levels We observed significantly higher FGM values in the female callimico (Mdn=402.5 ng/g, 95% CIdf=281.3-762.1) compared to the male (Mdn=156.5 ng/g, 95% CIdf=125.8-264.8). In the pied tamarin group, the breeding female exhibited higher FGM levels (Mdn=120.0 ng/g, 95% CIdf=78.9-170.1) than both the breeding male (TM1:
51 Mdn=43.6 ng/g, 95% CIdf=31.0-48.9) and their male offspring (TM2: Mdn=58.7 ng/g, 95% CIdf=51.4-76.4). However, during a subsequent study conducted 98 days after the death of the breeding male, we observed no significant difference between the female (Mdn=148.7 ng/g, 95% CIdf=104.5-189.7) and her male offspring (Mdn=129.6 ng/g, 95% CIdf=104.4-182.3) in samples collected over a period of 11 days. In addition to sex differences, we observed diurnal variation of FGM for the pied tamarins, with elevated levels in the afternoon compared to morning samples for both the female (TF1; AM: n=4, Mdn=74.5 ng/g, Range=44.2-130.9; PM: n=5, Mdn=399.3 ng/g, Range=303.7-2126.1) and her male offspring (TM2; AM: n=3, Mdn=106.1 ng/g, Range=55.4-109.2; PM: n=6, Mdn=520.2 ng/g, Range=312.2-640.0). Discussion To our knowledge, this is the first study to provide comparative information on the fecal glucocorticoid response of callitrichine species. All animals assessed in this study (n=13) demonstrated elevated FGM within 24-48 h of a veterinary exam, highlighting the broad applicability of this cortisol EIA to measure the stress response of callitrichine species. In addition, we demonstrated diurnal variation of FGM in pied tamarins. We observed differences between species in the pattern and magnitude of the response to stressors. Specifically, the marmoset group that experienced an exam in response to the death of a social partner and hand rearing of a neonate showed the greatest response and suggests that FGM measures are sensitive to acute and prolonged stressors. Differences in baseline FGM concentrations between males and females were observed in some species, although this may depend on the social status of individuals.
52 Assessment of fecal markers for use in callitrichine monkeys As first morning fecal samples were collected from multiple individuals each day, identification of a reliable fecal marker was necessary. We observed strong differences across species in their reaction to fecal markers, with glitter being the only marker that was readily consumed by all species. Although all animals within a group reacted similarly, the observed preferences may be group-specific, as other studies have described different preferences (Fuller et al., 2011). As callitrichines have demonstrated rapid gut passage times (2.5-4.5 h; Heymann and Smith, 1999; Power and Oftedal, 1996), fecal markers were fed in the early afternoon for identification of first morning voids the following day. The pied tamarins appeared to have a more rapid excretion time than other species, as fecal markers fed in the early afternoon were identifiable in the late afternoon of the same day. Differences in gut transit times among callitrichines have been noted and linked to dietary differences, with shorter retention times for more frugivorous species such as tamarins compared to marmosets that specialize on gum exudativory (Power and Oftedal, 1996). Effect of veterinary exams and social changes on FGM concentrations Veterinary exams produced a consistent FGM increase across most species that was visible for most individuals within 24 h of the exam, with peak concentrations often observed within 48 h. Following the initial peak, secondary increases were noted in several species before FGM concentrations returned to baseline levels. This pattern has been observed in other species (Steller sea lion, Hunt et al., 2004; greater sage grouse, Jankowski et al., 2009), including primates (yellow baboon, Wasser et al., 2000), and
53 these authors have suggested that a secondary increase may be a byproduct of enterohepatic recirculation of hormones (Roberts et al., 2002) or negative feedback regulation of HPA activity (Jankowski et al., 2009). Although the presence of individually unidentified samples post exam prevents a clear understanding of the absolute magnitude increase of FGM levels for each individual, the callimicos appeared to exhibit a reduced and variable FGM response compared to the golden lion tamarins and pied tamarins. Differences in the excretion or metabolism of cortisol may exist between this species and other callitrichines. The response of the two male marmosets differed from the overall pattern observed in the other study species, with a consistent increase in FGM concentrations observed for 10 days after an exam in response to the death of the breeding female, which for one individual reached a peak FGM concentration 143 times greater than baseline. The increase in this male was considerably larger than other individuals in this study and that reported by other studies, such as the 10.8x increase over baseline Heistermann et al. (2006) described in two common marmosets after translocation to a new institution. Two days after the marmoset’s veterinary exam, a neonate that was born 9 days prior to the death of the breeding female, was returned to the marmoset off-exhibit holding area for hand-rearing in an incubator. The presence of the neonate and its frequent vocalizations may have contributed to this progressive increase in FGM as several studies have found male marmosets are highly motivated by infant vocal cues (e.g. Zahed et al., 2008). The playback of infant distress calls to unrelated male common marmosets increased serum cortisol with no evidence of habituation after repeated exposures (Barbosa and Mota, 2014). As these authors point out, marmosets exhibit a high degree of male parental
54 investment and increased cortisol in response to neonate cues may be an important mechanism to promote care-giving behavior. In addition to the presence of the neonate, increased human activity in the holding area during the hand rearing process may have also negatively influenced the marmosets. Previous studies on laboratory-housed common marmosets have reported increased fecal glucocorticoids (Barbosa and Mota, 2009) and salivary cortisol (Cross et al., 2004) in response to increased human activity. As monogamous pair-bonded primates, the death of the breeding female marmoset may have contributed to this prolonged stress response. Partner-directed affiliative and protective behaviors have been reported in a wild male common marmoset to a dying female mate (Bezerra et al., 2014). However, it is unlikely that the death of the female contributed directly to the dramatic increase in FGM concentrations observed for two reasons. First, elevated samples were first observed 24-48 h after the exam, which corresponds to 48-72 h after the death of the breeding female and removal of the neonate. Based on the time lag to elevated FGM concentrations observed in other animals in this study (24-48 h), this initial increase was more likely to be in response to the exam than to the female’s death. Second, this dramatic increase in FGM concentrations that occurred in both the breeding male and adult offspring was not observed after the death of an adult male offspring or after the death of the breeding male. Similarly, the pied tamarins demonstrated a rapid return to baseline FGM levels after a veterinary exam and death of the breeding male, with no prolonged response to the male’s absence. Huck et al. (2005) also reported no increase in FGM in a wild group of mustached tamarins after the death of a breeding female.
55 Both marmosets in the present study showed a reduction to near baseline levels 27 days after the exam but neither demonstrated a complete return to baseline. During this subsequent collection period, the neonate had been moved to a temporary cage attached to the marmoset holding cages that allowed for physical contact between the neonate and adult marmosets. Human activity from the hand rearing procedure had also decreased by this point, but was still more frequent than baseline periods. Taken together, it is likely that a combination of factors related to the nearby hand rearing of a neonate contributed to the continual FGM response in the marmosets. Although being in audiovisual contact with conspecifics during the initial hand rearing may benefit the neonate and is a recommended practice (Ruivo, 2010), the potential adverse effects on adults should be considered and efforts should be made to minimize this disturbance as much as possible. Demographic and diurnal factors influencing FGM levels Differences in FGM concentrations were observed in the callimicos and pied tamarins, with higher concentrations present in females vs. males, suggesting possible sex differences in these species although the sample size was limited. Higher cortisol concentrations in females compared to males have also been reported for other callitrichine species: pied tamarin (Armstrong and Santymire, 2013), common marmoset (Johnson et al., 1996; Raminelli et al., 2001), and Wied’s black tufted-ear marmoset (Callithrix kuhlii; Smith and French, 1997). In the pied tamarin group this difference appeared to depend on the social status of the males. Although the female had higher baseline FGM levels than both the breeding male and adult offspring during the initial baseline periods, no significant difference was observed between the female and male offspring roughly 3 months after the death of the breeding male. While the female’s
56 FGM level appeared similar during both time periods, the male offspring’s FGM level doubled. This change may reflect the changing social conditions for the male offspring, as Ziegler et al. (1996) reported lower urinary cortisol in experienced breeding males and their male offspring living in the natal group compared to inexperienced males living with or near females. Although FGM differences between sexes may actually be a reflection of possible cross-reactivity of the cortisol EIA antibody with sex-specific hormones as suggested by Goymann (2012), this is unlikely in the present study as previously published cross-reactivity of the cortisol antibody showed strong specificity for cortisol (Young et al., 2004). However, sex differences in the metabolism of cortisol remains possible and requires further assessment. Diurnal variation of FGM concentrations was evident for the pied tamarins, with increased glucocorticoids present in afternoon samples compared to morning samples. A clear circadian pattern of cortisol production has been described in other callitrichine species, with salivary and urinary cortisol measures increasing from the first morning void to a late morning peak followed by a decrease in the late afternoon (Cross and Rogers, 2004; McCallister et al., 2004; Smith and French, 1997). In contrast, FGM levels have been shown to be higher in the afternoon (common marmosets: Raminelli et al., 2001; Sousa and Ziegler, 1998), likely owing to the lag time of excretion through feces. However, detecting a circadian rhythm of cortisol in feces has not been clearly documented across primates (Huck et al., 2005; Setchell et al., 2008; Weingrill et al., 2004). Additional research is needed to assess factors such as diet and metabolism that may contribute to species-specific differences in the excretion of cortisol in feces.
57 This study demonstrated the broad applicability of a fecal cortisol EIA to measure the stress response of callitrichine monkeys through noninvasive methods. We observed differences between species with respect to the magnitude of FGM increase in response to veterinary exams and other potential stressors, although all animals responded to the stress of veterinary handling. Specifically, a progressive increase in FGM levels was observed in two male marmosets during the separation and hand-rearing of a neonate, highlighting the importance of considering the potential impact of hand rearing on group members. Death of conspecifics did not appear to influence FGM levels of the pied tamarins or white-fronted marmosets. Although sample sizes were limited, we did observe some species-specific differences in FGM concentrations between sexes. Similar species-specific sex differences between closely related species have been noted in Ateles (Rangel‐Negrín et al., 2009; Rimbach et al., 2013) and suggest that HPA activity may not be generalizable across even closely-related species. Future research is needed to explore species-specific differences in HPA activity and to uncover what factors may have contributed to these changes.
58 Chapter Three: The influence of waterfall sounds and access to off-exhibit areas on the behavior and exhibit use of three species of callitrichine monkeys
Introduction Noise, or unwanted sound, is a potential source of stress for animals (Morgan and Tromborg, 2007). With zoos often being located in urban settings and attracting large numbers of visitors, the effects of noise may be particularly concerning for zoo-housed animals. Research has identified potential negative consequences of ambient noise in the zoo setting. For examples, in response to visitor noise, studies have documented increased vigilance behavior (orangutan, Birke, 2002; gorilla, Clark et al., 2012; whitehanded gibbon, Cooke and Schillaci, 2007; lion, Farrand, 2007; koala, Larsen et al., 2014), increased activity (giant panda, Owen et al., 2004), altered maternal (sun bear, Owen et al., 2014) and infant (orangutan, Birke, 2002) behavior, increased stress-related behaviors (giant panda, Owen et al., 2004), and elevated cortisol levels (giant panda, Owen et al., 2004). Similar effects have also been described for construction noise (giant panda, Powell et al., 2006; snow leopard, Sulser et al., 2008). Addressing these concerns is challenging and will likely require modifying the sound environment for zoo animals. One approach that has been suggested is to introduce sounds in order to mask disruptive or distracting noises arising in the environment (e.g. Ogden et al., 1994). Auditory masking is the process whereby a masking sound interferes with the perception of other sounds resulting from increased hearing thresholds. For people, masking sounds have been successfully introduced into office settings to reduce
59 distracting sounds (Loewen and Suedfeld, 1992) and a recent study identified water sounds as being particularly beneficial (Haapakangas et al., 2011). Similarly, water features have been found to successfully mask road noise and improve the soundscape quality of parks (De Coensel et al., 2011; Galbrun and Ali, 2013; Jeon et al., 2010; Nilsson et al., 2010). Currently, few studies have investigated the potential of auditory masking for animals, although this has been suggested as a potential explanation for the beneficial effects of auditory enrichment for animals (Patterson-Kane and Farnworth, 2006; Wells, 2009). However, data on free-ranging wildlife have highlighted potential costs of auditory masking from anthropogenic sources of noise (Barber et al., 2010). One obvious cost is the masking of communication between animals (Slabbekoorn and Ripmeester, 2008). This may even carry direct fitness consequences, as sparrows breeding near a generator showed reduced parental success, an effect the authors attributed to masking of parent-offspring communication (Schroeder et al., 2012). For prey species, auditory masking can impact the ability to detect predators. For example, ground squirrels living near a wind turbine exhibited elevated levels of vigilance and increased likelihood of returning to their burrows in response to the playback of conspecific alarm calls compared to ground squirrels living in a quiet area (Rabin et al., 2006). The primary aim of the current study was to determine if loud sound from a waterfall feature provided beneficial masking effects for three species of callitrichine monkeys housed nearby: pied tamarin, white-fronted marmoset, and callimico. Also, we investigated whether access to quiet off-exhibit areas was beneficial and could modify the response to the sound environment. Behavior and exhibit use were recorded during
60 experimental modifications of the waterfall feature and access to off-exhibit areas. In addition, we recorded visitor presence and sound levels to describe their potential relationship and effect on behavior. If the waterfall provided benefits through auditory masking of disruptive or distracting sounds, we expected the waterfall sounds to promote decreased vigilance and other behavioral indicators of anxiety (e.g. self-directed behavior and scent marking), and increased use of the exhibit space when animals had access to off-exhibit areas. Overall, we expected increased choices through access to off-exhibit areas would have beneficial effects on behavior. Materials/Methods Subjects and housing Subjects for this study included nine individuals representing three species of callitrichine monkeys: pied tamarin (Saguinus bicolor), white-fronted marmoset (Callithrix geofroyii), and callimico (Callimico goeldii). Demographic characteristics of the social groups are shown in Table 1. An adult male marmoset died of natural causes during the seventh week of the study and data from this individual were excluded.
61
Table 1. Demographic background of study subjects. Individual
Sex
Age (y.o.)
Pied tamarin TM1 M 15.3 TF1 F 4.1 TM2 M 1.4 TM3 M 0.9 White-fronted marmoset MM1a M 5.6 MF1 F 3.9 MM2 M 0.8 MM3 M 0.8 Callimico CM1 M 20.2 CM2 F 2.5 a Individual died during study and data were not included in analysis.
Animals were housed in single-species exhibits in the front of the RainForest building at Cleveland Metroparks Zoo. Exhibits were approximately 18 m3 and included branches and small trees for climbing. Guests were able to view animals through a mesh barrier in the front of exhibits that was separated from viewing areas by a small planter that included low bushes. The viewing area in front of the pied tamarin exhibit was situated in a gift shop area. The gift shop separated this exhibit from an open lobby area in front of the marmoset and callimico exhibits. A small door at the back of the exhibits connected to off-exhibit overnight holding cages (4 m3). Exhibits and holding areas of each species were visually isolated from other species but auditory contact between species was possible.
62 A waterfall was located in the front entrance lobby area adjacent to exhibits. The waterfall featured a 30 ft. drop and, in addition to the overhead ventilation system and ground fans near the waterfall, created a loud broadband sound environment within exhibits (Fig. 1). Exhibits were oriented parallel to the waterfall, with the marmoset exhibit being the closest (6 m.) and the callimico and tamarin exhibits at roughly 13 m from the edge of the waterfall.. Experimental design This study was conducted during weekdays from February to May 2012. Modifications to the waterfall feature (on vs. off) and access to off-exhibit areas (access vs. no access) were systematically manipulated in six two-week experimental phases in an ABACDA design. Baseline conditions (A) with the waterfall on and no access to offexhibit areas were alternated with experimental conditions that modified the waterfall or access features (B, C, D; Table 2). Baseline conditions were followed during weekends. Table 2. Overview of experimental design. Experimental Modification Waterfall On Waterfall Off Off-exhibit access No off-exhibit access
Two-week experimental phase A B A C
D
A
X
X
X
X X
X
X
X X
X
X
X
After the first two-week baseline phase, minor construction in the gift shop took place that required data collection to be suspended for approximately one week. Additionally, data from one day of the study were excluded as a result of errors manipulating the waterfall feature.
63 Behavioral data collection Data were recorded during 15 min focal observations (n=505; 55-57 observations per animal) using instantaneous point-sampling at 30 s intervals for state behaviors and exhibit use and all-occurrences sampling for event behaviors (Martin and Bateson, 2007; Table 3). Additionally, continuous measurement of the time the focal animal was visible was recorded to adjust all-occurrences data. Observations were balanced between morning (AM: 10AM – 1PM, n=253) and afternoon (PM: 1PM – 4PM, n=252) and the daily order of individual observations was randomized. Table 3. Ethogram of behaviors considered in this study. Behavior
Definition
State Behaviors (instantaneous point-sampling) Monitor Visitor
Animal is stationary and visually focused on an object (including visitors or observer) outside the exhibit or scanning the visitor area for a minimum of 3 seconds.
Locomote
Animal travels greater than one body length (not including the tail).
Inactive
Animal is stationary and either alert and actively monitoring surroundings or resting and not attending to environment, typically with tail curled around body and eyes closed.
Social Affiliation
Animals are engaged in a non-aggressive positive behavior and includes grooming and play behavior.
Event behaviors (all-occurrences sampling) Self-directed
Animal manipulates body using hand, feet, or mouth and typically took the form of scratching or self-grooming.
Scent Marking
Repeated rubbing of scent glands in anogenital, suprapubic, sternal, or facial area across substrate or body part (including tail marking in the callimicos and hand marking in the pied tamarins).
64 Visitor and sound measurement To evaluate for the potential effect of visitors, the total number of visitor groups that stopped at an exhibit to view animals during behavior observations was recorded. A group was defined as a visitor party of any size that approached the exhibit together and viewed the exhibit for a minimum of 2 s. This measure was chosen for ease of recording and based on pilot data that found the majority of groups at this exhibit were small (72% < 3 visitors). In addition, overall zoo attendance was recorded each day. Sound levels were measured in each exhibit using a data logging sound level meter (Model 407760, Extech Instruments, Nashua, NH). The sound level meter was mounted in a protective enclosure and placed on the floor of the front of the exhibit and oriented towards the visitor area to minimize the influence of animal vocalizations on sound level measurements. The sound level meter was capable of measuring sounds between 20 Hz-20 kHz with an accuracy of ± 2dB. The sound level meter was set to record sound levels every 0.5s using a ‘Fast’ (125 ms) time-weighting response and ‘A’ frequency weighting curve (dBA). A-weighting adjusts sound levels based on the frequency sensitivity of human hearing and was chosen because most primate species have been shown to have a broadly similar pattern of hearing (Heffner, 2004). As sound pressure levels fluctuated, equivalent continuous sound levels (Leq) were calculated (Eq. 1). The Leq represents the mean energy level during a sampling period expressed on the decibel scale and has been previously used in studies assessing noise in zoos (Quadros et al., 2014). Equivalent continuous sound levels were calculated for the entire day (10:0016:00; Daily Leq) and during 15 min observations (Observation Leq). Because sound level
65 meters were not available for each exhibit every day, these data were used to describe the sound environment and not included as a variable in the behavior analyses. Eq 1. Formula for calculating the equivalent continuous sound level (Leq) from N equally spaced sound level measurements (see Raichel, 2006): 𝑁𝑁
1 Leq = 10log( � 10𝐿𝐿𝐿𝐿/10 ) 𝑁𝑁 𝑖𝑖=1
Statistical analysis Daily Leq sound levels were compared between days with the waterfall on vs. off for each exhibit using a Wilcoxon rank sum test with t approximation. In addition, pairwise comparisons of daily Leq sound levels between the three exhibits when the waterfall was off vs. on were performed using Wilcoxon rank sum tests. The relationship between visitors and sound levels was evaluated using a Spearman rank correlation between Observation Leq sound levels and the number of visitor groups. As we observed a potential influence of loud animal vocalizations on observation sound levels, we reassessed this relationship using Leq values that excluded sound levels above 85 dBA, the minimum sound level of vocalizations we suspected based on exploratory graphs. To assess differences between exhibits or conditions in the number of visitor groups stopping during an observation (Groups), a generalized linear mixed model (GLMM; Proc GLIMMIX; SAS) was constructed. Fixed effects included Exhibit (tamarin, marmoset, callimico), Waterfall (on or off), Access (access or no access), and overall zoo attendance as a covariate to account for changes during the study. A random effect of Date was
66 included with a Variance Components covariance structure to account for the potential correlation of multiple observations across exhibits on the same day. As the Groups variable was non-normal, the model was constructed using a negative binomial distribution with log link function. The negative binomial distribution is commonly used with overdispersed skewed count data. Parameters were estimated using Laplace approximation and degrees of freedom was calculated using the between-within method. Behavioral data were assessed as counts (number of scans for state behaviors and exhibit use or total number of occurrences for event behaviors) and analyzed using GLMMs. Models were adjusted for the time spent visible using an offset term. The GLMM model for the time spent in the Front exhibit location was adjusted using an offset term of the scans on exhibit to account for changes in the total available space depending on access to off-exhibit areas. All behaviors except Inactive were modeled using a negative binomial distribution and a log link function to account for overdispersion and a skewed data distribution. As the Inactive behavior model residuals were approximately normal using a log link function, a Gaussian distribution was specified. Parameters were estimated using Laplace approximation and degrees of freedom were calculated using the between-within method. To account for the nonindependence of repeated measures data, a random effect of Individual (animal identity) was included in all models using a Variance Components structure. A full model was first constructed that included fixed effects of Species (tamarin, marmoset, and callimico), Waterfall (on or off), Access (off-exhibit access or no access), Groups (number of visitor groups present during observation), Waterfall*Access interaction, and Species interactions with Waterfall, Access, and Groups. To account for
67 variability of behaviors across time of day, a random slope effect of Time (AM or PM) was assessed with the full model using a likelihood ratio test comparing an intercept-only model to an intercept-and-slope model (Bolker et al., 2009). The addition of the random slope effect of Time significantly improved model fit for all behaviors except Inactive and Off-exhibit. For these behaviors, intercept-and-slope models did not converge and intercept-only and slope-only models were compared, resulting in a random interceptonly model for Inactive and random-slope only model for Off-exhibit. As the main question of the study involved the effects of the waterfall and access on the study species, these main effects and their species interactions were retained in all models. The Waterfall*Access, Groups, and Groups*Species effects were removed in a step-wise fashion from the full model if they were non-significant (p>0.1) and their removal did not worsen model fit (change in corrected Akaike information criterion
