THE NIGHT SHIFT: LIGHTING AND NOCTURNAL STREPSIRRHINE CARE IN ZOOS GRACE FULLER

Transcription

1 THE NIGHT SHIFT: LIGHTING AND NOCTURNAL STREPSIRRHINE CARE IN ZOOS by GRACE FULLER 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 January 2014

2 CASE WESTERN RESERVE UNIVERSITY SCHOOL OF GRADUATE STUDIES We hereby approve the dissertation of Grace Fuller, candidate for the Doctor of Philosophy degree*. Signed: Kristen E. Lukas, Ph.D. (chair of the committee) Mark Willis, Ph.D. Mary Ann Raghanti, Ph.D. Patricia M. Dennis, D.V.M., Ph.D. Christopher W. Kuhar, Ph.D. Date: 10 October 2013 * We also certify that written approval has been obtained for any proprietary material contained within.

3 Dedicated to: Rene Culler, my mother and my role model; my father Charles Fuller, who would have been so proud; and all the big- eyed friends I made along the way.

4 TABLE OF CONTENTS Acknowledgments vii Abstract xii Chapter 1. Introduction: Light, Activity Patterns, and the Exhibition of 1 Nocturnal Primates in Zoos Chapter 2. A Survey of Husbandry Practices for Lorisid Primates in 30 North American Zoos and Related Facilities Chapter 3. A Retrospective Review of Mortality in Lorises and Pottos 57 in North American Zoos, Chapter 4. Validating Actigraphy for Circadian Monitoring of Behavior 90 in the Pygmy Loris (Nycticebus pygmaeus) and Potto (Perodicticus potto) Chapter 5. Methods for Measuring Salivary Melatonin in the Potto, 113 Perodicticus potto, and Pygmy Loris, Nycticebus pygmaeus Chapter 6. A Case Study Comparing Hormonal and Behavioral 135 Responses to Red and Blue Exhibit Lighting in the Aye- Aye, Daubentonia madagascariensis Chapter 7. A Comparison of Nocturnal Strepsirrhine Behavior in 148 Exhibits Illuminated with Red and Blue Light Chapter 8. Endocrine Responses to Exhibit Lighting in the Potto, 175 Perodicticus potto Chapter 9. General Discussion 195 Appendix I. Questions for Multi- Institutional Husbandry Survey 200 Appendix II. Detailed Tables for Cause of Death in Lorisid Primates by 202 Species and Age Group References 208 i

5 LIST OF TABLES Chapter Two Table 1. Group compositions of lorisid primates in North American facilities. 37 Table 2. Enclosure design features of lorisid primates in North American 40 facilities. Table 3. Lighting design of the primary enclosure. 43 Table 4. Animal care practices. 45 Table 5. Estimated reproductive success of lorisid primates. 49 Chapter Three Table 1. Recorded type of death for lorises and pottos in North American 63 facilities Table 2. Distribution of loris and potto deaths by age class in North American 64 facilities Table 3. Primary cause of death or reason for euthanasia in lorises and pottos 67 housed in North American facilities Table 4. Percent of lorises and pottos with pathology diagnosed by organ 69 system upon postmortem examination Table 5. Neoplasia reported for lorises and pottos in North American facilities Table 6. Circumstances surrounding traumas related to death in lorises and 80 pottos in North American zoos ii

6 Chapter Four Table 1. Ethogram for behavioral data collection for actigraph study. 99 Chapter Six Table 1. Ethogram for behavioral data collection on the aye- aye. 138 Chapter Seven Table 1. Nocturnal strepsirrhine subjects and housing conditions for the 154 multi- zoo study. Table 2. Ethogram for behavioral data collection. 158 Chapter Eight Table 1. Nocturnal strepsirrhine subjects and housing conditions for the 181 multi- zoo study. iii

7 LIST OF FIGURES Chapter Two Figure 1. Age pyramids for lorisid primates in North American institutions. 38 Figure 2. Primary enclosure size for lorisid primates in North American 41 facilities. Chapter Four Figure 1 a- b. a) Attaching the actigraph harness to the pygmy loris subject; b) 95 the potto wears the actigraph harness on exhibit. Figure 2. General activity budget for the potto subject with and without the 101 actigraph harness in place. Figure 3. General activity budget for the pygmy loris subject with and without 103 the actigraph harness in place. Figure 4. Actigraph data for the potto subject. 105 Figure 5. Actigraph data for the pygmy loris subject. 106 Figure 6. Mean activity counts associated with behaviors observed in a potto 107 and a pygmy loris. Chapter Five Figure 1. Collection of saliva from the female potto. 118 Figures 2 a- c. Melatonin concentrations measured following nocturnal 127 exposure to test lights in a potto (a), and pygmy lorises PSL1 (b) and iv

8 PSL2 (c). Figure hour patterns of salivary melatonin expression in a potto and 130 three pygmy lorises (PSL). Chapter Six Figure 1. Daily time of emergence from the nest box by the aye- aye subject 140 based on keeper reports. Figure 2. Time spent performing active behaviors (move, feed, self- directed, or 141 object examination) by the aye- aye subject during the baseline red and experimental blue lighting conditions. Figure 3. Dark phase activity budget ( hrs) for the aye- aye subject 142 during the baseline red and experimental blue lighting conditions. Figure 4. Dark phase salivary melatonin concentrations in the aye- aye during 144 the baseline red and experimental blue lighting conditions. Figure 5. Salivary cortisol rhythms in the aye- aye during the baseline red and 146 experimental blue conditions. Chapter Seven Figure 1 a- b. Percent of time spent performing active behaviors during the dark 162 phase by animals at Cleveland Metroparks Zoo (a) and Cincinnati Zoo and Botanical Garden (b). Figure 2 a- d. Changes in behavior across the three study conditions (C1-3) 164 comparing red and blue light for pottos (a) and pygmy slow lorises v

9 (b) at Cleveland Metroparks Zoo (CMZ) and pottos (c) and a bamboo lemur (d) at Cincinnati Zoo and Botanical Garden (CZBG). Figure 3 a- c. 24- hour activity level data for potto PP2 (a) and pygmy slow 167 lorises NP1 (b) and NP2 (c) (see Table 1 for subject IDs) at Cleveland Metroparks Zoo. Chapter Eight Figure 1 a- c. Melatonin concentrations measured in potto saliva six hours after 185 dark phase onset in red and blue light. Figure 2. Scatterplot of light intensity compared to salivary melatonin 186 concentrations measured in pottos at six hours after dark phase onset. Figure 3. Salivary cortisol concentrations measured in pottos living in blue and 188 red dark phase lighting conditions. vi

10 Acknowledgments It is humbling to look back over these many years and contemplate how many people contributed to my development as a scientist as well as to the body of work I am offering here. First and foremost, I would like to thank my dissertation advisor, Dr. Kristen Lukas. Meeting Kristen inspired me to pursue a path of study examining the care of animals in captivity. Over the years she has been a mentor and a friend, and I am a better person and scientist for having known her. All of my committee members have had a profound influence on how I see the world and have also been a source of great strength and support throughout the process of completing my dissertation research. I am grateful to Dr. Mary Ann Raghanti for her years of friendship and encouragement, for believing in me, and always reminding me to think like an anthropologist. I am grateful to Dr. Pam Dennis for pushing me to ask hard questions and for encouraging me to believe that I could acquire the skills to answer them, and for always reminding me to stand up for the little guys. I am grateful to Dr. Chris Kuhar for pushing me to strive for both scientific rigor and practicality in my research endeavors. Not only did Dr. Mark Willis teach me to think like a scientist, he also demonstrated what it means to be a great teacher both in and outside the classroom. Finally, Dr. Mandi Vick selflessly gave me so much guidance, personally and professionally, that I will always consider her an honorary part of my committee. I look forward to many years of friendship and collaboration with you all. I am also grateful to my academic peers and friends for their support and encouragement over these many years. Dr. Elena Less was a source of inspiration to vii

11 me for both her work ethic and her kindness, and I am so grateful for all the experiences we had together as students. Jason Wark made me laugh every day, and pushed me to think critically and have fun doing it. Christine Cassella was also a great source of friendship and support, and an inspiration for how to live a better life. I am incredibly grateful to Austin Leeds for his assistance in collecting saliva samples for this research, as I truly could not have done it without his help. Bonnie Baird is going to do great things and I will remember that I knew her when. I look forward to many years of friendship and collaboration with you all as well, and I cannot wait to see all the amazing contributions that you make to our field. Working in the Conservation and Science area at Cleveland Metroparks Zoo was amazing because literally everyone I worked with was wonderful. Laura Amendolagine is a top- notch laboratory manager and therapist and has the biggest heart of anyone I know. I will always remember fondly our time working in the lab together. I am also grateful to Kym Gopp for her dedication to loris conservation and for offering me opportunities to contribute, in some small way, to this goal. Finally, I am grateful that I met Sonia DiFiore there too. Of course there are many other staff members at the zoo that I would like to thank for their assistance with my research, especially Becky Johnson, Dawn Stone, and Andrew Smyser, who put in many, many hours of work to assist in my research efforts all while providing exceptional care to the lorises and many other animals. I am also grateful to Terri Rhyner, Heather Mock Strawn, and the other animal keepers in PCA for their dedication and hard work. At Cleveland Metroparks Zoo, I would also like to thank Linda DeHoff, Sharon Gehri, Andi Kornak, Pam Krentz, Dr. viii

12 Albert Lewandowski, Tad Schoffner, and Dr. Mike Selig. Apologies to any one I have omitted what really makes Cleveland Metroparks Zoo great is that all the staff are dedicated, good people. I had a wonderful experience collecting data at Cincinnati Zoo and Botanical Garden, in large part because the staff members were so welcoming and helpful. I am extremely grateful to Michael Guilfoyle for his support of my research project and for sharing with me his many years of knowledge about caring for pottos and other nocturnal mammals. I would also like to thank Matt Miller for his assistance with lighting changes and for creating a little lab for me in Jungle Trails. Finally, I would also like to thank Patrick Callahan, Mike Dulaney, Ron Evans, Valerie Haft, Janet Hutson, Michael Land, Mike Maciariello, Kate MacKinnon, Dr. Terri Roth, Stephanie Schuler, Vicki Ulrich, Amanda Weisel, and any other staff that assisted with this project. I would like to thank Nicole Smith and Jeannine Jackle for coordinating sample collection from animals at Franklin Park Zoo. I had a great experience visiting the zoo and really enjoyed working with you both. Many thanks are also due to the other Tropical Forest staff members who assisted with this project. There are many others who have contributed to my professional development and I would like to thank them here: Dr. Colleen McCann, Helena Fitch- Snyder, Dr. Dean Gibson, Dr. Anna Nekaris, and Dr. Tara Stoinski. All of these brilliant women have been a great source of inspiration to me. Also a great supporter and inspiration was my master s thesis advisor, Dr. Richard Feinberg. ix

13 I would like to express my gratitude to the following institutions for completing the husbandry survey: Akron Zoological Park, Albuquerque Biological Park, Aquarium & Rainforest at Moody Gardens, The Calgary Zoo, Botanical Garden & Prehistoric Park, Capron Park Zoo, Chicago Zoological Society- Brookfield Zoo, Cincinnati Zoo & Botanical Garden, Cleveland Metroparks Zoo, Duke Lemur Center, El Paso Zoo, Franklin Park Zoo, Houston Zoo, Inc., Lee Richardson Zoo, Lincoln Park Zoo, Little Rock Zoo, Los Angeles Zoo and Botanical Gardens, Louisiana Purchase Gardens & Zoo, Memphis Zoo, Mesker Park Zoo & Botanic Garden, Minnesota Zoological Garden, Omaha s Henry Doorly Zoo, The Philadelphia Zoo, Prospect Park Zoo, Pueblo Zoo, San Diego Zoo, Trevor Zoo, Wildlife Conservation Society- Bronx Zoo, Woodland Park Zoo, and Zoo de Granby. I wish also to thank the following facilities for contributing medical records to this study: Albuquerque Biological Park, Aquarium & Rainforest at Moody Gardens, The Calgary Zoo, Botanical Garden & Prehistoric Park, Chicago Zoological Society- Brookfield Zoo, Cincinnati Zoo & Botanical Garden, Cleveland Metroparks Zoo, Columbus Zoo and Aquarium, Denver Zoo, Detroit Zoo, Duke Lemur Center, El Paso Zoo, Franklin Park Zoo, Houston Zoo, Inc., Lake Superior Zoological Gardens, Lee Richardson Zoo, Lincoln Park Zoo, Los Angeles Zoo and Botanical Gardens, The Maryland Zoo in Baltimore, Memphis Zoo, Mesker Park Zoo & Botanic Garden, Minnesota Zoological Garden, Oglebay s Good Zoo, Omaha s Henry Doorly Zoo, The Philadelphia Zoo, San Antonio Zoo, San Diego Zoo and San Diego Wild Animal Park, San Francisco Zoo, Santa Ana Zoo, Trevor Zoo, Virginia Zoological Park, Wildlife Conservation Society- Bronx Zoo, and Woodland Park Zoo. Several individuals went x

14 above and beyond to accommodate this request, including Julie Parks Taylor, Jenn Harrison, Jeannine Jackle, Andrea Katz, and Mark Campbell. I am also grateful to the many volunteers who assisted with entering medical data, including Gail Simpson, Jessica Taylor, Alyssa Mills, and Lauren Starkey. I am grateful to all the lorises and pottos, the big- eyed friends we met and sometimes lost along the way. I will never forget you: Vijay, Sweetie Pie, Fozzy, Moe, Nox, Harry, Hermione, Sing, Tai Sanlo, Philbert, Nicole, Ringo, Jahzira, Caliban, Mojo, Chachi, Rendille, Lil Bit, Jabari, Tiombe, Lucy, Jabari, Amare, and Gabriel. You touched more lives than you could ever know. Finally, to friends and family, both those lost and those whose company I still enjoy, I offer my love and gratitude. Peace. xi

15 The Night Shift: Lighting and Nocturnal Strepsirrhine Care in Zoos by GRACE FULLER Abstract Over billions of years of evolution, light from the sun, moon, and stars has provided organisms with reliable information about the passage of time. Photic cues entrain the circadian system, allowing animals to perform behaviors critical for survival and reproduction at optimal times. Modern artificial lighting has drastically altered environmental light cues. Evidence is accumulating that exposure to light at night (particularly blue wavelengths) from computer screens, urban light pollution, or as an occupational hazard of night- shift work has major implications for human health. Nocturnal animals are the shift workers of zoos; they are generally housed on reversed light cycles so that daytime visitors can observe their active behaviors. As a result, they are exposed to artificial light throughout their subjective night. The goal of this investigation was to examine critically the care of nocturnal strepsirrhine primates in North American zoos, focusing on lorises (Loris and Nycticebus spp.) and pottos (Perodicticus potto). The general hypothesis was that exhibit lighting design affects activity patterns and circadian physiology in nocturnal strepsirrhines. The first specific aim was to assess the status of these populations. A multi- institutional husbandry survey revealed little consensus among zoos in lighting design, with both red and blue light commonly used for nocturnal illumination. A review of medical xii

16 records also revealed high rates of neonate mortality. The second aim was to develop methods for measuring the effects of exhibit lighting on behavior and health. The use of actigraphy for automated activity monitoring was explored. Methods were also developed for measuring salivary melatonin and cortisol as indicators of circadian disruption. Finally, a multi- institutional study was conducted comparing behavioral and endocrine responses to red and blue dark phase lighting. These results showed greater activity levels in strepsirrhines housed under red light than blue. Salivary melatonin concentrations in pottos suggested that blue light suppressed nocturnal melatonin production at higher intensities, but evidence for circadian disruption was equivocal. These results add to the growing body of evidence on the detrimental effects of blue light at night and are a step towards empirical recommendations for nocturnal lighting design in zoos. xiii

17 Chapter One Introduction: Light, Activity Patterns, and the Exhibition of Nocturnal Primates in Zoos Nocturnal species make up more than half of the world s land- dwelling vertebrates but often are cryptic and difficult to observe in the wild (Conway, 1969). Zoos provide an opportunity for close encounters with nocturnal animals, but their care in captivity poses unique challenges. Nocturnal species are generally housed on reversed light cycles in zoos. Artificial lighting is designed to simulate nighttime conditions in exhibits during the day so that nocturnal animals are active when zoo visitors and staff can observe them. Lighting in nocturnal exhibits must be carefully designed to entrain the animal s circadian system to the reversed light cycle, providing enough light for zoo visitors to view nocturnal species without overwhelming the animals sensitive visual systems or suppressing their activity (Erkert, 1989). This perceptual tug- of- war may have negative effects on the behavior, health, and reproduction of nocturnal species in zoo environments. Prior to the advent of artificial lighting, light from the environment provided reliable information about the time of day or year to physiological systems regulating animal behavior, allowing animals to maximize reproductive success by timing behaviors like foraging and sleeping to appropriate environmental conditions (Ashby, 1972; Halle and Stenseth, 2000b). It is now becoming clear that artificial lighting can disrupt internal timekeeping systems, adversely affecting health, reproduction, and behavior in humans and other animals (Navara and Nelson, 2007; Rea et al., 2008). These effects are dramatic enough that the World 1

18 Health Organization categorizes night- shift work as a possible carcinogen (Straif et al., 2007). Nocturnal animals are the shift workers of zoos, and the means and effects of altering their activity patterns in the captive setting are in need of systematic study. The goal of these studies was to examine current husbandry practices for nocturnal strepsirrhines in zoos, focusing on members of the primate family Lorisidae (lorises and pottos), and the behavioral and physiological impacts of exhibit lighting. The general hypothesis was that lighting designs for nocturnal exhibits vary among North American zoos, and these differences have implications for the behavior, circadian rhythms, and health of captive nocturnal strepsirrhines. NOCTURNALITY IN THE PRIMATE ORDER The majority of primates are diurnal, with only one nocturnal haplorhine genus (Aotus) and a preponderance of nocturnal forms among the strepsirrhines (Ankel- Simons and Rasmussen, 2008). However, this simple dichotomy between nocturnal and diurnal forms based along taxonomic lines has been repeatedly challenged by field observations of primates engaged in active behaviors outside their presumed temporal niche (Ankel- Simons and Rasmussen, 2008; Curtis and Rasmussen, 2006). Many lemur species may be instead termed cathemeral, meaning that they are active opportunistically during both day and night (Curtis and Rasmussen, 2006). Many now believe that primate activity patterns and their associated visual adaptations are evolutionarily flexible, allowing species to exploit different temporal niches depending on local conditions (Ankel- Simons and 2

19 Rasmussen, 2008; Pariente, 1979). This new understanding of primate activity patterns casts doubt on the assumption based on fossil evidence that the earliest primates were nocturnal (Ankel- Simons and Rasmussen, 2008). Activity in darkness exposes animals to predators and other dangers (Bearder et al., 2002) and there are several hypotheses to explain the evolutionary benefits of a nocturnal activity pattern (Curtis and Rasmussen, 2006). Nocturnality allows animals to partition environments temporally, minimizing competition for resources. Charles- Dominique (1975) suggests that in tropical environments, nocturnal primates benefit by reduced feeding competition from birds at fruiting trees. Nocturnal species may also benefit by using the cover of night to surreptitiously hunt prey, or as a means of avoiding predation (Bearder et al., 2002). Restricting activity to nighttime can also aid in thermoregulation by avoiding hotter times of the day (Curtis and Rasmussen, 2006; Fernandez- Duque, 2003). Decreased visual information available in darkness has led to a suite of perceptual adaptations that allow nocturnal primates to gather information from the environment using other sensory systems, leading to a great emphasis on chemical (Delbarco- Trillo et al., 2011) and auditory signals (Charles- Dominique, 1975). Visual adaptations that make use of the little light available at night are also common. Nocturnal Primate Visual Systems Given the degree of flexibility in primate activity patterns, it is perhaps not surprising that primate visual systems are highly variable as well and do not 3

20 separate clearly into diurnal and nocturnal groups. Many nocturnal primates possess morphological features that confer improved vision at night, such as large eyes (Ross and Kirk, 2007). Nocturnal species often have greater numbers of rods (sensitive to brightness) relative to cones (sensitive to color) in the retina relative to diurnal species (Silveira, 2004). For example, humans have an estimated 5 or 6 cones for every 100 rods in the retina, while in the potto this ratio is 1 to 300 (Goffart et al., 1976). Many species such as the potto (Goffart et al., 1976) also possess a tapetum lucidum, a reflective layer of cells behind the retina that serves to amplify ambient light. However, a more nuanced examination reveals that variation is the norm for these features. Eye size does not correlate reliably with orbital aperture and among the nocturnal primates, only predatory species (such as Loris and Tarsius) have exceptionally large eyes compared to diurnal species of the same body size (Kirk, 2006). Even the presence of a tapetum lucidum does not correlate reliably with activity pattern among primates (Ankel- Simons and Rasmussen, 2008). Thus, one must take care in making assumptions about species activity patterns based on morphological evidence alone. Color vision is also highly variable in the primate lineage, as Ankel- Simons and Rasumussen (2008) explain. Photosensitive cells in the retina are divided into two classes: rods, which are specialized for vision under dark or scotopic conditions; and cones which are active under bright or photopic conditions and are specialized for visual acuity and color perception. Rod cells contain a single (RH1) opsin (protein that absorbs light), meaning that vision under scotopic conditions is essentially monochromatic and color blind. The extent to which members of a given 4

21 species perceive color depends on which genes that code for cone opsins are present. Most mammals are functionally dichromatic, possessing the short wavelength sensitive type one (SWS1) cone opsin and the middle to long wavelength (M/LWS) cone opsin. Additionally, many haplorhine primates have evolved trichromatic vision by diversification of the M/LWS gene (Kawamura and Kubotera, 2004). Each type of cone responds maximally to different wavelengths of light, and the peak sensitivity of a given opsin varies among different species as well (Melin et al., 2012). The extent and nature of color vision varies among strepsirrhine primates. Many nocturnal lemurs, including mouse lemurs (Microcebus spp.) and aye- ayes (Daubentonia madagascariensis) retain the SWS1 pigment and are functionally dichromatic. Melin et al. (2012) propose that this locus is under active selection in the aye- aye, which is often active during the blue- hued twilight and for whom the color blue may play an important role in foraging and social communication. The cathemeral brown lemur (Eulemur fulvus) retains the SWS1 gene as well (Kawamura and Kubotera, 2004) and trichomatic color vision has evolved among several diurnal lemurs (Tan and Li, 1999). Lorisiform primates express only a single M/LWS opsin and are therefore monochromatic (Deegan and Jacobs, 1996). It appears that the common ancestor of lorises and galagos lost functionality of the SWS1 opsin gene and as a result these species are truly color- blind (Kawamura and Kubotera, 2004). So how do nocturnal primates see the world? All primates have high visual acuity, even nocturnal species that lack a fovea, the central cluster of photoreceptors 5

22 where visual acuity is greatest (Pariente, 1979). Vision also plays an important role in nocturnal primate orientation and communication behaviors (Bearder et al., 2006; Pariente, 1979). Nocturnal primates are able to navigate complex forest environments easily in what humans perceive as almost complete darkness. Yet, many nocturnal species have the visual flexibility to engage in daytime activity, and a surprising number of nocturnal species appear to be capable of at least some degree of color vision (Melin et al., 2012). This high level of flexibility and variability among species has important implications for the care of nocturnal primates in zoos. Lighting designs in enclosures are likely to be perceived very differently based on the visual capabilities of the species in question, and some species may adapt more readily to artificial changes in the light environment than others. Natural History of Lorisid Primates The infraorder Lorisiformes includes two families, the Lorisidae and the Galagonidae (Groves, 2001). The Galagonidae consists of bushbabies or galago species, nocturnal primates which may be found living sympatrically with African lorisines but who exhibit a vertical clinging and leaping locomotor style that contrasts sharply with the slow, quadrupedal climbing of the Lorisidae (Bearder, 1999). The taxonomic diversity of galagos is relatively well described in part due to the conspicuous, species- specific vocalization patterns they exhibit - unlike the quiet, cryptic lorises (Grubb et al., 2003). Groves (2001) further divides the Lorisidae into two subfamilies: the Asiatic Lorisinae (slow and slender lorises) and the African Perodicticinae (pottos and angwantibos). 6

23 The taxonomic diversity of lorisines has not yet been fully described. With Nycticebus menagensis, N. bancanus, and N. borneanus, the newly named N. kayan makes four slow loris species recognized in Borneo alone (Munds et al., 2012). In addition to the pygmy loris, N. pygmaeus, three additional slow loris species are recognized: N. coucang, N. bengalensis, and N. javanicus (Nekaris et al., 2008). The pygmy loris s range extends into southern China, and the other slow lorises are found across continental Asia. Slender lorises are found only in India and Sri Lanka and include at least two species, the red (Loris tardigradus) and grey (L. lydekkerianus) slender lorises (Brandon- Jones et al., 2004). The taxonomy of the perodicticines is even less developed than that of lorises. As Grubb et al. (2003) explain, only two genera are recognized currently: Perodicticus, the potto, and the golden potto or angwantibo, genus Arctocebus. Two species of angwantibo are recognized, the Calabar (A. calabarensis) and the golden anwangtibo (A. aureus). Although only a single potto species (P. potto) is currently recognized, the true diversity of this group has yet to be described. Pottos are found from the west coast of Africa east into Kenya, and the three known subspecies are divided geographically into the western (P. p. potto), central (P. p. edwarsi) and eastern (P. p. ibeanus) pottos. In discussing the natural history of lorises and pottos, it is important to keep in mind that their true behavioral variation, like their true taxonomic breadth, has yet to be fully delineated. Lorises and pottos are found in a variety of forested habitats ranging from primary rainforest, to evergreen and bamboo forests, and agroforests (Streicher, 2004). All lorisines are nocturnal foragers that traverse their environments by 7

24 climbing along tree branches and lianas (Bearder, 1987). Their muscular limbs and strong grip allow for careful exploration of the environment, and their slow, deliberate movements may be an adaptation to avoid predator detection (Charles- Dominique, 1977). However, it should be noted that rapid arboreal locomotion has been observed in pygmy (Duckworth, 1994) and slender lorises (Nekaris and Stevens, 2007). Charles- Dominique (1977) likens the niche of the potto in Africa to the slow loris in Asia, and that of the more faunivorous angwantibo to the slender loris. Pottos are known to consume exudates, fruit, and insects (Oates, 1984). Fruit appears to play a larger role in the potto diet than that of slow lorises (Gonzalezkirchner, 1995), which are dedicated exudativores that actively gouge trees to elicit gums and saps. In Cambodia, the pygmy loris primarily consumes exudates, followed by fruits and arthropods (Starr and Nekaris, 2013). The Bengal slow loris (N. bengalensis) feeds almost exclusively on gums during the winter, suggesting that exudates may be a particularly important resources for lorises occupying highly seasonal environments (Swapna et al., 2010). The diet of the slender loris is more faunivorous, and the Mysore slender loris (L. lydekkerianus lydekkerianus) has been observed consuming animal prey at 96% of feeding events (Nekaris and Rasmussen, 2003). When pottos do consume insects, they generally concentrate on slow- moving or noxious species, much like the slow loris (Charles- Dominique, 1977; Wiens et al., 2006). Slow lorises may additionally benefit from this diet by sequestering secondary compounds from arthropod prey for incorporation into venom (Wiens et 8

25 al., 2006). The venomous bite of the slow loris is unique among primates and results from mixing saliva with secretions from a brachial gland (Krane et al., 2003). Toxic bites have led to anaphylactic shock in human caretakers (Kalimullah et al., 2008) as well as necrotic wounds in sanctuary- housed conspecifics (Streicher, 2004), and venom likely plays a role in both interspecific and intraspecific defense. The most striking defensive adaptation of the potto is a series of spines on the cervical vertebrae, which are thrust at attackers when pottos are threatened (Charles- Dominique, 1977). Lorisids are often found foraging alone, but they are far from asocial and their social structure may be better described as dispersed rather than solitary (Pimley et al., 2005). Pottos forage alone but share delayed communication via olfactory signals (Charles- Dominique, 1977). Male and female home ranges overlap and pairs seem to associate with exclusivity (Pimley et al., 2005). A spacing system wherein male home ranges overlap with the ranges of several females is also typical of slender (Nekaris, 2003b) and slow lorises (Wiens and Zitzmann, 2003). Individuals tend to be more social at dawn or dusk. Female slender lorises (L.t. lydekkerianus) sleep in groups with their offspring or other relatives, and adult males are often present in sleeping groups as well (Nekaris, 2003b; Radhakrishna and Singh, 2002). Thus, social networks are maintained through overlapping home ranges, sleeping contact, olfactory and auditory signals, and interactions surrounding mating and infant care. Pottos are relatively common throughout their range, and their populations are thought to be stable (Oates et al., 2008). However, it is clear that all lorises are 9

26 declining in the wild (Nekaris & Jayewardene 2004, Nekaris & Nijman 2007, Iseborn et al. 2011). The greatest threats to Asiatic lorisines are deforestation (for palm oil and other cash crops); and human exploitation for use in traditional Asian medicine, the tourist photo prop trade, and the illegal pet trade (Nekaris et al., 2010). In recent years the pet trade has had an increasingly devastating impact on wild slow loris populations, in part due to the nouveau celebrity of lorises in cute web videos (Nekaris et al., 2013). Many animals rescued from this illegal trade have suffered physical injuries that make reintroduction to the wild impossible, and the need for sanctuary housing grows along with the popularity of lorises as pets (Nekaris & Jaffe 2007). Information gleaned from decades of caring for lorises and pottos in zoos may take on new relevance to in situ conservation efforts as sanctuaries play a more critical role in managing wild populations. STATUS AND CARE OF NOCTURNAL STREPSIRRHINES IN NORTH AMERICAN ZOOS North American zoos and related facilities accredited by the Association of Zoos and Aquariums (AZA) currently house nine genera of nocturnal strepsirrhines according to the most recent population analysis and breeding/transfer plan (Kuhar et al., 2011). This group includes five species from continental Africa: the potto, (P. potto), two lesser bushbabies (G. senegalensis and G. moholi), and two species of greater bushbaby (Otolemur garnettii and O. crassicaudatus). Malagasy strepsirrhines housed in North American zoos include the aye- aye (D. madagascariensis), gray mouse lemur (Microcebus murinus), giant mouse lemur 10

27 (Mirza zaza), and the fat- tailed dwarf lemur (Cheirogaleus medius). Asian lorisines in zoos include representatives of both the genera Nycticebus and Loris. With one exception, these populations all are classified as red Species Survival Plans under AZA s current sustainability framework, meaning that their populations contain fewer than 50 individuals and do not meet minimum standards for genetic diversity (Kuhar et al., 2011). The pygmy loris population, which contains 71 animals at 22 North American facilities, is classified as a yellow SSP (Gibson et al., 2013) and likely shows the greatest potential for future exhibition of nocturnal strepsirrhines in zoos. Caring for Lorisid Primates in Zoos Lorisid primates in AZA zoos include the potto, pygmy loris, red slender loris (L. tardigradus tardigradus) and a hybridized population of N. coucang and N. bengalensis managed under the coucang moniker. Although AZA zoos historically housed an additional subspecies of red slender loris, L.t. nordicus, this population is no longer represented in North America. According to Fitch- Snyder and Schulze (2001), the history of slow and slender lorises in North America dates back to late 19 th /early 20 th century exhibits at the Philadelphia and Bronx zoos, while a breeding population of pygmy slow lorises was established in North America in Current lorisid populations are mostly in decline. At the most recent published population analysis, there were fifteen pottos, thirteen slow lorises, and twelve slender lorises remaining in AZA accredited facilities (Kuhar et al., 2011). 11

28 High infant mortality and traumatic death appear to be major impediments to population growth in lorises and pottos. Debyser (1995) analyzed mortality trends for strepsirrhine juveniles in zoos and primate centers and found that more offspring died prior to weaning in N. coucang, L.t. tardigradus, and P. potto than the thirteen lemur and galago species that were also surveyed. Sutherland- Smith and Stalis (2001) reviewed mortality in lorises from the San Diego Zoo ( ) compared to Duke University Lemur Center ( ). These data show that in addition to trauma, significant contributors to captive loris morbidity and mortality include dental, renal, and respiratory diseases. A comprehensive review of current knowledge of Asian loris captive care is provided in Fitch- Snyder and Schulze s (2001) husbandry manual and is also available at conservation.org. Guidelines for habitat design are briefly summarized from the manual as follows. The minimum cage size should be no less than 15.6 m 3. Relative humidity should be maintained between 40-60% and the temperature between F. Frequent cage cleaning is not recommended as lorises may be stressed by the procedure, but accumulated urine marks should be regularly removed. Finally, an enriched environment should contain nest boxes, leafy branches that provide cover for the animals, other hiding places, and floor substrates that can be used for sleeping and olfactory stimulation. Social and reproductive requirements are also outlined in detail in the husbandry manual (Fitch- Snyder and Schulze, 2001). A challenging aspect of lorisid care in captivity is balancing their naturally dispersed social system with the need for social opportunities in a setting where space is restricted. Lorises should be 12

29 housed as breeding or mother- offspring pairs. Even more ideal would be separate cages that share a common area where lorises can interact; this design mimics the most common natural ranging system in which one male s territory overlaps with several distinct female territories. Aggression among individuals should be monitored and is especially common surrounding breeding activity and parturition. Mixed- species housing may also provide a source of enrichment, and has the additional benefits of utilizing exhibit space more efficiently and creating public education opportunities (Ferrie et al., 2011; Fitch- Snyder and Schulze, 2001). Despite the fact that the spectral composition of moonlight is equivalent to that of sunlight (Erkert, 1989), nocturnal primates are often housed under light shifted toward either red or blue wavelengths (Davis, 1961; Frederick and Fernandes, 1994). The first successful zoo exhibits of nocturnal animals, such as the Bronx Zoo s World of Darkness, were illuminated by dim red light (Conway, 1969). Because the spectral sensitivity of rods is shifted toward the blue end of the spectrum, electrical activity in the potto retina appears similar in response to blue and white light but is attenuated in response to red light (Goffart et al., 1976). This difference between the sensitivity of differing retinal cell types is the rationale for exhibiting nocturnal species under red light, which would arguably be perceived as less bright by them (Davis, 1961). Current guidelines for illumination of nocturnal primate exhibits are limited. The husbandry manual for Asian lorisines (Fitch- Snyder and Schulze, 2001) recommends full- spectrum light during the light phase and full- spectrum or red light during the dark phase, but light intensity is not specified for the dark phase. 13

30 For the aye- aye, the Duke Lemur Center s (DLC) housing guidelines call for red light of less than one lux for dark phase illumination (Williams et al., 2013). The extent to which these husbandry guidelines are followed in practice is unknown, but anecdotal evidence suggests that zoos are using a variety of approaches to create night lighting in the absence of more concrete standards. Given the pivotal role of light in regulating the circadian system, and the potential consequences for behavior and health when this signal is disrupted, lighting design for nocturnal animals in the captive environment should be empirically examined. LIGHT AND THE REGULATION OF PRIMATE ACTIVITY PATTERNS Chronoecology is the study of factors that shape the distribution of animal activities over time (Halle and Stenseth, 2000a). Animals have evolved internal time keeping systems, known as biological clocks, that allow them to predict and respond behaviorally and physiologically to regular patterns of environmental change that occur on a daily or seasonal basis (Rietveld et al., 1993). Located in the suprachiasmatic nucleus (SCN) of the hypothalamus, the internal timekeeping system calibrates rhythmic outputs of behavior, such as sleep- wake cycles, to match environmental conditions (Challet, 2007). Light is the primary signal that entrains the mammalian circadian system. The mammalian retina contains intrinsically photosensitive retinal ganglion cells (IpRGCs) that project to the SCN, as well as neural areas involved in mood and cognition (Bailes and Lucas, 2010). IpRGCs contain a photosensitive opsin called melanopsin that responds strongly to short wavelengths but little to red light (Bailes 14

31 and Lucas, 2010). Consequently, blue light has a greater effect on circadian entrainment than other wavelengths, an effect that has been demonstrated in hamsters (Boulos, 1995), nocturnal mouse lemurs (Perret et al., 2010), and humans (Brainard et al., 2008). Internal clocks also have an inherent flexibility so that they can be synchronized with environmental inputs by entraining agents known as zeitgebers (timekeepers). Clocks are also subject to direct effects of environmental inputs that acutely alter the outputs of internal oscillators, known as masking effects (Rietveld et al., 1993). Masking agents serve the purpose of allowing short- term adaptation to environmental change (Erkert, 2008; Rietveld et al., 1993). The interactions between internal clocks, zeitgebers, and masking agents produce rhythmic outputs of behavior in the form of activity patterns (Fernandez- Duque, 2003). Primatologists are only beginning to examine activity patterns, and there is a great need for research on the chronoecology of primate behavior (Curtis and Rasmussen, 2006). Environmental conditions, including illuminance levels, interact with other factors such as temperature and humidity, behavioral patterns of predator species, morphological characteristics, and patterns of locomotion and food acquisition to structure primate activity patterns in a complex, species- specific manner (Bearder et al., 2002; Fernandez- Duque, 2003). In the captive environment, these variables are under anthropogenic control and represent both artificial zeitgebers and masking agents that interact to shape activity patterns in zoos (Richter, 2006). 15

32 Proximate Factors Shaping Activity Patterns in Lorisids Light is the major proximate factor that entrains activity patterns in nocturnal mammals (Erkert and Cramer, 2006). When housed in constant darkness, the slow loris (N. coucang) has an internal pacemaker that creates an endogenous activity rhythm with a period close to 24 hours (Kavanau and Peters, 1974; Redman, 1979). When housed in artificial lighting, the lorises active period becomes entrained to the dark phase of the light- dark (LD) cycle (Ehrlich, 1968; Redman, 1979; Tenaza et al., 1969). An outdoor colony of slow lorises at a university reliably timed the onset of their behavior to the twilight period despite seasonal changes in day length, leading Kavanau (1976) to argue that lorises have a strong endogenous activity rhythm that is primarily entrained by light levels at twilight. He suggests that twilight may be an important zeitgeber because becoming active as soon as lighting conditions are favorable confers a survival advantage (Kavanau, 1976). Surprisingly, the majority of nocturnal primate species that have been studied are more active under the brighter light of the full moon than on darker nights (Ankel- Simons and Rasmussen, 2008; Fernandez- Duque, 2003; Gursky, 2003). Galagos are known to travel and vocalize more on bright moonlit nights, perhaps because light aids in finding food or navigating the forest (Nash, 1986). However, lorisids may be the exception to this trend. The pottos and angwantibos in Charles- Dominique s (1977) study site limited their active period to that of total darkness. Both species also emerged from their sleeping sites later and returned to them earlier than sympatric galagos, a pattern which Charles- Dominique (1977) attributed to the emphasis on cryptic behavior in the perodicticinae. 16

33 Although some studies have found that Asian lorisines do not alter resting or travel patterns in relation to moonlight levels (Bearder et al., 2002; Bearder et al., 2006), in a seasonal environment the pygmy loris is less likely to forage on cold and bright moonlit nights (Starr et al., 2012). These findings highlight the complexity of activity patterns among species, which are shaped not only by a multitude of variables but also by interactions between them. Husbandry practices in captivity can shape the intensity and patterning of loris and potto activities as well. Daschbach et al. (1982/83) found that slow lorises were more active in larger, more enriched cages. Similarly, Frederick and Fernandes (1996) found that naturalizing an exhibit for two pottos, by adding elements such as leaves and a cricket dispenser, led to an increase in activity and an expansion of the pottos behavioral repertoire to include more exploratory and sexual behaviors. However, the causal relationship between the emergence of sexual behavior and other activity changes is unclear. Furthermore, some of these behaviors declined after an initial peak, suggesting the increase in activity may have been at least partially a result of novelty (Frederick and Fernandes, 1996). Other factors documented to affect captive slow loris activity include zoo visitor presence (Oswald and Kuyk, 1978) and extreme temperatures (Kavanau, 1976). A collection of small- scale studies supports the claim that loris activity is inhibited under high illuminance levels. A reduction in activity with increasing light intensity has been documented in captive galagos (Randolph, 1971), as well as three slow lorises at the Woodland Park Zoo (Trent et al., 1977). Frederick and Fernandes (1994) simultaneously increased the intensity of simulated day lighting, decreased 17

34 the intensity of night lighting, and changed night lighting from blue to full- spectrum lighting for two pottos at Franklin Park Zoo. They found increased activity and behavioral diversity following these changes, but the causal relationship between these findings and the increase in reproductive behavior they also observed is unclear (Frederick and Fernandes, 1994). Also, because so many changes were simultaneously made to the lighting conditions in this study, it is difficult to draw any firm conclusions about their efficacy. Other elements of nocturnal primate lighting design have received less scientific scrutiny than intensity. Frederick et al. (1995) examined the effect of adding simulated dawn and dusk periods to the lighting regimen for two captive pottos, but it is difficult to draw any conclusions from their data because they simultaneously changed the length of the photoperiod in the exhibit as well. Despite the prevalence of differed colored night lighting in zoos, at present there are no systematic studies examining the effects of light wavelength on nocturnal primate behavior. The effects of light wavelength on locomotor rhythms have been examined in Drosophila (Subramanian et al., 2009) and migratory birds (Malik et al., 2004) but have been little studied in mammals. Boulos (1995) demonstrated that the effectiveness of light pulses for altering rhythms of wheel- running in Syrian hamsters (Misocricetus auratus) was dependent on wavelength, with blue light exerting a strong effect compared to green and red. Tests with humans have not shown differential effects of light color on nighttime alertness. Shifts workers chronically exposed to bright light filtered to block short wavelengths reported few 18

35 differences in fatigue and concentration compared to an unfiltered light condition (Schobersberger et al., 2007). Exposure to both red and blue light at night reportedly arouses alertness in human subjects experimentally asked to stay awake and perform work at night, as measured by heart rate and electroencephalogram (EEG) activity (Figueiro et al., 2009). As these studies show, the complexities of circadian regulation in relation to the light environment are far from being understood. This complexity, along with the natural diversity in activity patterns and visual systems among nocturnal primates, complicates the task of providing recommendations for nocturnal primate lighting design. To this aim, Erkert (1989) proposed that artificial lighting regimens for nocturnal primates are effective when animals show evidence of stable circadian rhythms without any apparent masking of activity during the dark phase. However, these criteria may be difficult to judge in lorisid primates for the same reason that it is hard to make inferences about welfare states in nocturnal strepsirrhines that have naturally inactive behavioral profiles (Wright et al., 1989). Examining the effects of lighting on endocrine functioning and health may provide further insight for determining optimal lighting conditions. HEALTH IMPLICATIONS OF EXPOSURE TO LIGHT AT NIGHT Lighting levels that render lorises visible to the zoo- going public may essentially amount to continuous light exposure, which human epidemiological studies and laboratory animal research suggests can have wide- ranging health effects. Constant exposure to light can be a source of stress for mice (Van der Meer 19

36 et al., 2004), and disrupts circadian activity patterns (Albers et al., 1981) and sleep- wake cycles (Ikeda et al., 2000) in laboratory rats. The deleterious effects of light at night appear to be largely endocrine- mediated. Melatonin The principal hormone responsible for conveying information about light to the internal timekeeping system is melatonin. Information about lighting conditions is transmitted to the SCN, the neural seat of the master biological clock, via the retino- hypothalamic track, and ultimately to the pineal gland (Altun and Ugur- Altun, 2007). In response to photic input, pinealocytes synthesize the amine melatonin (N- acetyl- 5- methoxy- tryptamine) by way of a four- step process for which tryptophan is the precursor molecule. Melatonin is generated by the activity of AANAT (arylalkrylamine- N- acetyltransferase), which converts serotonin (5- hydroxytryptamine) into melatonin and is the rate- limiting enzyme for melatonin synthesis (Altun and Ugur- Altun, 2007). This pathway produces serotonin during the day and melatonin in darkness (Murch et al., 2000) In both nocturnal and diurnal species, circulating melatonin levels are much higher during the dark phase of the LD cycle (Altun and Ugur- Altun, 2007). Variation in the timing and duration of melatonin excretion encodes information about day length and therefore time of year, and melatonin s principle timekeeping role appears to be regulating seasonal changes in behavior, especially as those behaviors relate to reproduction (Arendt, 2005; Malpaux et al., 1999). However, 20

37 melatonin also plays a role in regulating circadian patterns of behavior such as core temperature rhythms and sleep- wake cycles (Arendt, 2005). The suppression of nocturnal melatonin production due to light exposure is well documented in humans and many other species. Light at night suppresses pineal melatonin production in a variety of laboratory rodents (Brainard et al., 1984; Depres- Brummer et al., 1995) as well as rhesus macaques (Reppert et al., 1981) and the squirrel monkey (Hoban et al., 1990). The effect of light exposure is dose- responsive, meaning that higher light intensities lead to greater suppression of melatonin (Stevens and Rea, 2001). A greater intensity of light is needed to upset the circadian rhythm of melatonin production than to temporarily suppress or mask its expression (Hashimoto et al., 1996). Suppression of melatonin by light exposure in human subjects has been repeatedly demonstrated in controlled experiments as well as applied studies with shift workers (Khalsa et al., 2003; Laakso et al., 1993; Lowden et al., 2004; Navara and Nelson, 2007; Reiter, 1991; Zeitzer et al., 2000). Even exposure to ordinary room light before bed can delay the onset of nocturnal melatonin production and reduce its duration (Gooley et al., 2010; Zeitzer et al., 2000). As a result of their retinal morphology, nocturnal animals are much more sensitive to light- induced melatonin suppression than diurnal species (Reiter, 1991). Laboratory studies have focused on acute exposure to intense light at night, and the chronic effects of exposure to dimmer illumination are in need of further study (Stevens and Rea, 2001). Chronic dim lighting is probably the most comparable laboratory paradigm to current zoo conditions, in which animals 21

38 housed on reversed light cycles are exposed to light of some intensity all hours of the day. The degree of melatonin suppression that occurs as a result of light exposure depends on light wavelength. Brainard et al. (1999) reviews mammalian variation in action spectra, defined as the degree of biological effect as a function of wavelength. In humans, peak melatonin suppression occurs by exposure to light in the range of visible blue light ( nm), but peak sensitivities in nocturnal rodents may occur in the ultraviolet range (as low as 330 nm) (Brainard et al., 1999). Even relatively dim blue light can suppress melatonin production; the threshold for intensity for blue light suppressing melatonin in horses has been measured at three to ten lux (Walsh et al., 2013). Selectively filtering short wavelengths successfully prevents light- induced melatonin suppression in rats (Rahman et al., 2008) and humans (Schobersberger et al., 2007). Falchi et al. (2011) compare action spectra for circadian regulation to the luminous output of common artificial lights. They recommend a total ban on outdoor lighting that emits wavelengths less than 540 nm in cities to prevent deleterious effects of light on circadian rhythmicity and melatonin production in humans and other animals. Melatonin s role as an endocrine timekeeper means that it has important effects on a wide array of physiological systems that undergo daily or seasonal change including the immune (Skwarlo- Sonta, 2002), metabolic, and reproductive systems (Pandi- Perumal et al., 2006; Reiter, 1991). Melatonin has general immunostimulatory and antioxidant properties, and relationships between illness 22

39 and melatonin suppression are likely mediated by the immune system (Navara and Nelson, 2007). A growing body of evidence implicates exposure to light at night and melatonin suppression in oncogenesis (Reiter et al., 2007). In laboratory studies, animals exposed to constant illumination develop mammary and other tumors at increased rates (Dauchy et al., 1997; Stevens and Rea, 2001). Epidemiological studies show increased rates of ovarian (Bhatti et al., 2013), breast, prostate, endometrial, and colorectal cancer in shift workers and other individuals who routinely experience circadian disruption (Stevens and Rea, 2001). Blask (2009) even suggests that uninterrupted darkness may be a natural form of cancer prevention. Other major diseases associated with shift work that may be related to circadian disruption and the effects of light include cardiovascular disease, major depression, metabolic syndrome, and infertility (Arendt, 2005; Navara and Nelson, 2007). Melatonin regulates reproductive activity through multiple channels and can exert its effects through receptors on each level of the hypothalamic- pituitary- gonadal (HPG) axis (Luboshitzky and Lavie, 1999). Melatonin can alter the normally pulsatile release of hypothalamic gonadotropin- releasing hormone (GnRH), leading to acyclicity (Abbott et al., 2004; Luboshitzky and Lavie, 1999). Melatonin ultimately downregulates gonadal estrogens, and suppression of melatonin is associated with elevated estrogen levels and infertility (Navara and Nelson, 2007). There are thus multiple endocrine- mediated routes through which chronically elevated light 23

40 exposure can have negative impacts on health and reproduction in captive mammals. Melatonin is considered a primary biomarker for circadian dysregulation (Mirick and Davis, 2008) and can be reliably measured in various matrices (Middleton, 2006). Pineal melatonin is not stored but instead immediately enters circulation, so melatonin levels in plasma or serum are a reliable indicator of pineal activity (Altun and Ugur- Altun, 2007). Circulating melatonin passively diffuses into saliva through capillary beds in salivary glands, and salivary melatonin reliably reflects circulating melatonin concentrations in humans (Voultsios et al., 1997). Finally, melatonin or its major metabolite, 6- sulphatoxymelatonin, can also be measured and assessed for rhythmicity in urine (Klante et al., 1997). Because the expression of melatonin varies temporally, repeated sampling is necessary throughout the day (Middleton, 2006). Measurement of salivary melatonin, which is a reliable point measure yet also relatively non- invasive, has shown great utility in epidemiological studies. Nonhuman primates can be easily trained to provide saliva samples (Lutz et al., 2000; Tiefenbacher et al., 2003), and free- ranging primates will readily chew on saliva collection devices without training (Higham et al., 2010). Salivary analysis therefore has great promise for assessment of circadian function in zoo- dwelling and free- ranging wildlife. Cortisol The steroid hormone cortisol exhibits a robust circadian rhythm and is therefore also a useful biomarker for circadian activity (Buckley and Schatzberg, 24

41 2005). Expression of glucocorticoids (GCs, including cortisol) is regulated by the hypothalamic- pituitary- adrenal (HPA) axis. Endocrine cells in the paraventricular nuclei (PVN) of the hypothalamus secrete corticotropin- releasing- hormone (CRH), stimulating release of pituitary adrenocorticotropin (ACTH) and GCs from the adrenal medulla (Van Reeth et al., 2000). The PVN receive input from the SCN and release CRH in a rhythmic manner (Blask, 2009). Like melatonin, cortisol can be easily measured in saliva (Lac, 2001), and this method has been successfully used to document the circadian rhythmicity of cortisol expression in several nonhuman primates (Cross and Rogers, 2004; Heintz et al., 2011; Quabbe et al., 1982). Cortisol concentration, like melatonin, is strongly related to nocturnal light exposure and its health effects. In humans, rhythms of cortisol expression can be experimentally altered with light (Boivin and Czeisler, 1998). In rats, constant light exposure eliminates rhythms in the expression of corticosterone and 6- sulphatoxymelatonin (Claustrat et al., 2008). In humans, cortisol rhythms are altered in sleep disorders (Blask, 2009; Van Reeth et al., 2000) and in depression (Germain and Kupfer, 2008). Rhythmic expression of melatonin and cortisol are also dampened in women with metabolic syndrome (Corbalan- Tutau et al., 2012). Causal relationships between circadian disruptions, altered rhythmicity of hormones, and illness are difficult to untangle and are made more complex by the role of cortisol in the stress response. Cortisol is often described as a stress hormone, although it plays many roles in regulating the metabolic system during homesostasis (Sapolsky et al., 2000). The relationship between cortisol excretion and stress has been widely employed 25

42 with some controversy to investigate the welfare of animals in captivity (Mormede et al., 2007; Rushen, 1991). Stress is the biological response elicited by perceived threats to an individual s homeostasis (Moberg, 2000). Even abiotic factors like lighting and sound can be a source of stress in the captive environment, leading to activation of the HPA axis (Morgan and Tromborg, 2007). The effects of GC release in response to stress are wide- ranging and include augmented cardiovascular activity, rapid activation of the immune system, increased circulating glucose, and inhibition of reproductive behavior (Sapolsky et al., 2000). Chronic stress can result in prolonged elevated GC levels and is associated with a variety of behavioral and health problems (Morgan and Tromborg, 2007). The chronic stress experienced by some animals as a result of captivity (Morgan and Tromborg, 2007) may therefore complicate attempts to utilize cortisol as a biomarker for circadian regulation. Welfare Implications of Circadian Studies for Zoo Animals Exploring captive behavior on a wider temporal scale may have additional benefits beyond understanding proximate factors that precipitate activity. Despite the clear relevance of nighttime behaviors, such as sleep disturbances, as indicators of stress (Abou- Ismail et al., 2007; Van Reeth et al., 2000) or sickness (Millman, 2007), few studies in zoos have incorporated the full daily pattern of activity in behavioral assessments of welfare. Richter (2006) examined activity patterns in zoo- housed koalas and found that evidence that the largely nocturnal animals were likely disturbed by early- morning husbandry routines. Laws et al. (2007) found that both fecal stress hormone metabolites and sleep disturbances increased in an 26

43 African elephant following relocation to a new herd. Yet, these approaches are rare and researchers have instead focused on daytime time budgets (the proportion of time observed devoted to different behaviors) as indicators of welfare in captive settings (McCann et al., 2007). As an alternative approach, studies that examine activity patterns in captivity offer the dual benefit of providing information both on behavioral rhythms as well as revealing specific after- hours behaviors, such as sleep disturbances, that are often neglected in zoo research but are likely indicative of animal welfare states (Anderson, 1998). THESIS OBJECTIVES The general hypothesis was that exhibit lighting design affects activity patterns and circadian physiology in nocturnal strepsirrhine primates. Specifically, the aims were threefold. The first aim was to evaluate the status of the current captive population of lorisid primates in North American zoos and related facilities, including common husbandry practices and major health concerns. The second aim was to develop methods for measuring circadian behavior and physiology in nocturnal strepsirrhines. The final aim was to document behavioral and endocrine patterns associated with lighting design in the captive environment. Chapter Two asks how lorisid primates in North American zoos are actually housed in practice in comparison to current husbandry standards. To answer this question, 29 AZA accredited zoos and related facilities were surveyed about current husbandry practices, with a focus on lighting design. 27

44 Chapter Three examines health trends in captive lorisids. Death records for lorises and pottos housed in AZA zoos over the last thirty years were reviewed. These data showed important trends in age- specific mortality and also identified the major causes of death for these species in captivity. Chapter Four assesses the utility of an automatic behavioral tracking technique, actigraphy, for use in monitoring circadian activity rhythms in the pygmy loris and potto. Although this method was not pursued for the remaining studies here, the data indicate that actigraphy has great potential for future captive and field- based studies examining lorisid activity patterns despite their slow- moving locomotor style. Chapter Five describes the development of techniques for collecting saliva from nocturnal strepsirrhines for hormone analysis. The results of chemical and biological validation experiments used to develop an assay for salivary melatonin in lorises and pottos are also reported. For these studies, 24- hour rhythms of salivary melatonin concentrations were examined as well as acute suppression of salivary melatonin concentrations by exposure to different wavelengths and intensities of light. Chapters Six, Seven, and Eight report on the results of a multi- institutional experiment comparing the behavioral and hormonal effects of red and blue exhibit lighting in zoos. Chapter Six describes the dramatic effects of this manipulation on a single aye- aye subject at Cleveland Metroparks Zoo, while Chapter Seven delineates the behavioral effects of red and blue exhibit lighting in nocturnal strepsirrhines at Cleveland Metroparks Zoo and Cincinnati Zoo and Botanical Gardens. Finally, 28

45 Chapter Eight reports on salivary melatonin and cortisol concentrations measured in pottos under blue and red light at three different zoos. General results of these studies are discussed in Chapter Nine. The future and challenges of research using salivary biomarkers to understand circadian (dys)regulation in applied settings are discussed. In closing, the implications of these studies for the welfare and population sustainability of nocturnal strepsirrhines in zoos are considered and recommendations are put forth for lighting design in nocturnal zoo exhibits. 29

46 Chapter Two A Survey of Husbandry Practices for Lorisid Primates in North American Zoos and Related Facilities INTRODUCTION According to Groves (2001), the primate family Lorisidae (Loridae) consists of small, nocturnal prosimians in two groups: the Asiatic lorises and the African perodicticinae. Asian forms include the grey (Loris lydekkerianus) and red (Loris tardigradus) slender lorises, and the greater (Nycticebus coucang), Bengal (N. bengalensis), and pygmy (N. pygmaeus) slow lorises. A single species of potto (Perodicticus potto) is recognized along with the golden (Arctocebus aureus) and Calabar (A. calabarensis) angwantibos; however, it is considered likely that future investigations will describe several new species in the genus Perodicticus (Grubb et al., 2003). North American zoos and related facilities currently house five lorisid species: the pygmy loris, red slender loris, greater and Bengal slow lorises, and potto. According to the current sustainability framework adopted by the Association of Zoos and Aquariums (AZA), none of these populations qualify as a green Species Survival Plan (SSP), the highest sustainability category defined by population size and genetic diversity. Only the pygmy loris received a yellow (intermediate) designation, while the remaining populations are all considered red programs because they contain fewer than fifty individuals (Kuhar et al., 2011). The effects of premature death or reproductive failure can be profound for such small populations (Schulze, 1998), and indeed, several lorisid species generally 30

47 exhibit poor reproductive success in zoos (Debyser, 1995; Fitch- Snyder and Jurke, 2003). Debyser (1995) reviewed the sources of mortality for strepsirrhine juveniles across zoos and primate centers and found that of all the strepsirrhines studied, the lorisids (N. coucang, L. tardigradus, and P. potto) exhibited the highest frequency of juvenile deaths as measured by the percentage of total animals born that die prior to weaning (the cumulative mortality incidence). Information about basic husbandry is critical for mitigating problems with health and stress in captive animals that are related to inappropriate environments (Schulze, 1998), and population trends of captive lorisids suggest a need for investigation into their care. Reproduction is perhaps the most studied aspect of lorisid husbandry to date. Basic reproductive parameters have been described in captive colonies of pygmy and slow lorises (Izard et al., 1988; Weisenseel et al., 1998), slender lorises (Izard and Rasmussen, 1985), and pottos (Cowgill et al., 1989; Frederick and Campbell, 1995). Hormonal correlates of reproduction have been examined in the slow loris (Perez et al., 1988) and pygmy loris (Jurke et al., 1997; Jurke et al., 1998), in which Fitch- Snyder and Jurke (2003) also examined behavioral correlates of reproductive endocrinology. Some information is also known about reproductive behavior in the slender loris (Goonan, 1993) and mate preference in the pygmy loris (Fisher et al., 2003a). In general there is more literature on captive reproduction in lorises than pottos. Only a few aspects of enclosure design have been empirically evaluated for lorisid species. Daschbach et al. (1982/83) examined behavior in relation to cage size for N. coucang, while Frederick and Fernandes (1996) focused on exhibit 31

48 complexity rather than size to study the effects of naturalizing an exhibit occupied by a potto breeding pair. Trent et al. (1977) found an inverse relationship between the intensity of activity and lighting levels in three slow lorises at the Woodland Park Zoo. Frederick and Fernandes (1994) also found greater activity and behavioral diversity in a potto pair after altering both light and dark phase illuminances and changing night lighting from blue to full- spectrum light. Together these studies suggest that a dark, naturalistic environment promotes active, natural behaviors in lorisid primates. A comprehensive review of current knowledge of Asian loris captive care is provided in Fitch- Snyder and Schulze s (2001) husbandry manual and is also available at conservation.org. Guidelines for habitat design are briefly summarized from the manual as follows. The minimum cage size should be no less than 15.6 m 3. Relative humidity should be maintained between 40-60% and the temperature between F. The animals require approximately 12 hours of daily light, with full- spectrum illuminance preferred during both light phases, and red light as a nighttime alternative. Frequent cage cleaning is not recommended as lorises may be stressed by the procedure, but accumulated urine marks should be regularly removed. Finally, an enriched environment should contain nest boxes, leafy branches that provide cover for the animals, other hiding places, and floor substrates which can be used for sleeping and olfactory stimulation. Social and reproductive requirements are also outlined in detail in the husbandry manual (Fitch- Snyder and Schulze, 2001). Central to understanding lorisid social needs is that their largely solitary ranging system does not indicate 32

49 that animals are asocial; as for most nonhuman primates, opportunities for social interaction are an important aspect of the enriched environment. Lorises should be housed as breeding or mother- offspring pairs. Even more ideal would be separate cages that share a common area where lorises can interact; this design mimics the most common natural ranging system in which one male s territory overlaps with several distinct female territories. Aggression among individuals should be monitored and is especially common surrounding breeding activity and parturition. Mixed species housing may also provide a source of enrichment, and has the additional benefits of utilizing exhibit space more efficiently and creating public education opportunities (Ferrie et al., 2011; Fitch- Snyder and Schulze, 2001). Although the current lorisid husbandry manual (Fitch- Snyder and Schulze, 2001) is thorough, the extent to which husbandry guidelines have been implemented across different institutions is largely unknown. Information about how animals are actually housed in practice is important for understanding the dynamics of health and reproduction in captive populations and for identifying risk factors for individual welfare (Barber, 2009). Thus, the goal of this study was to survey North American facilities housing lorisid taxa and to describe their current husbandry practices. We aimed to determine the extent to which current husbandry practices are consistent with established guidelines, as well as to identify issues of concern and gaps in knowledge for future investigation. 33

50 MATERIALS AND METHODS Survey We solicited all zoos and related facilities currently housing lorisid primates listed in the North American regional studbooks for participation in the survey. We first contacted each institution by sending a postcard to their institutional representative (IR) to the Prosimian Taxon Advisory Group. Some institutions did not have an IR listed; in these cases we contacted a primate curator or someone in an equivalent position listed in the AZA directory. We contacted non- AZA facilities using addresses and phone numbers available on their websites. We then sent each individual a link to the online survey and a research proposal. We made multiple attempts via to contact participants and followed up by phone when necessary as recommended by Plowman et al. (2006). We collected data using an online survey created using SurveyMonkey (Portland, OR, USA). We collected responses between January and May of We asked respondents to complete a survey for each group of lorisid primates housed in their institutional collection, and we defined a group as one or more animals that share the same exhibit or living space the majority of the time. The survey consisted of five sections: basic group information; primary and secondary enclosure design; lighting conditions in the primary enclosure; animal care practices; and a series of demographic questions about each group member. All multiple choice questions included an optional other category and a text field to explain responses. A list of survey questions is included in Appendix 1; a few questions were modified from Tarou et al. (2005). 34

51 We instructed respondents to utilize the following definitions in completing the survey. The primary enclosure was defined as the main space where the group is exhibited. In the case of groups that are not exhibited, this refers to the place where the animals are housed for the majority of the day. The secondary enclosure was defined as an alternate exhibit or holding space which may be an enclosure occupied seasonally or an off- exhibit holding area. The light phase was defined as daylight or part of the day during which bright lights are used to simulate daylight in the enclosure. The dark phase was defined as nighttime or the part of the day during which the enclosure lights are off, dimmed, or shifted in wavelength to simulate nighttime conditions. Data Analysis We asked survey participants to submit a survey for each group as a matter of convenience, because many aspects of physical housing and animal care are the same for animals sharing an exhibit. However, our results revealed that a high number of groups were actually solitary individuals. For this reason, we chose to analyze survey questions based on the percent of individuals for whom a particular response was chosen. All subjects were thus treated as independent data points for the purpose of data analysis. For some questions, multiple response options were permitted; for this reason, total values for categorical data do not total 100% where indicated. 35

52 RESULTS Population Demography and Group Compositions We received surveys from 29 institutions: 100% (n = 27) of those accredited by AZA; one certified related facility; and one of three other non- AZA facilities that currently house lorisid primates in North America. Our sample included 104 lorisid primates representing five species. The sample included eleven greater slow lorises (n = 3.8.0, male.female.unknown) and four Bengal slow lorises (n = 3.1.0). Based on the uncertain hybrid status of most of these individuals (Kuhar et al., 2011), we combined them into a single category called slow loris for data analysis. We received data for 90% of the North American population of pygmy lorises (n = ) representing 42 groups at 23 facilities. We received surveys for 100% of slender lorises (n = individuals, 8 groups at 4 facilities), slow lorises (n = 6.9.0, 12 groups at 10 facilities), and pottos (n= 8.6.1, 12 groups at 3 facilities). Across sexes, the mean age in years (SD) for each species was: 8.3 (4.5) for pygmy lorises, 7.8 (5.3) for slender lorises, 14.7 (3.9) for slow lorises, and 10.9 (9.8) for pottos. The pygmy loris evinced the only potentially stable age pyramid (Figure 1), while the other three populations were much smaller and constrictive in nature. For all species except the slow loris, the majority of individuals were living in breeding groups: 71% of pygmy lorises, 67% of slender lorises, and 53% of pottos resided in breeding groups compared to only 27% of slow lorises. 36

53 Table 1. Group compositions of lorisid primates in North American facilities. Number of Groups Pygmy loris (n = 42 groups) Slender loris (n = 8 groups) Slow loris (n = 12 groups) Potto (n = 12 groups) 1.0 [one male] [one female] 1.1 [one pair] Other 3 [one group of 0.2.0*, one group of 1.2.0, and one mixed species group of pygmy loris and slender loris] 2 [1.0.0 mixed with pygmy loris; and one group of 1.2.0] 0 Mean 1.25 group size 1.50 (0.09) 1.57 (0.3) (0.13) (SE) 1 [one group of 0.1.1] 1.25 (0.13) Group size range * Notation indicates: (# males. # females. # unknown individuals) Our sample included 73 lorisid groups with a mean of (SE) groups per zoo, half of which (49%) were identified as breeding groups. Across species, group sizes ranged from one to three individuals (Table 1). The majority of groups (61%) reported in the survey actually consisted of solitary individuals, including: 23 pygmy lorises (37% of individuals of this species), 4 slender lorises (33%), 9 slow lorises (60%), and 9 pottos (60%). In most other cases, animals were housed in male- female pairs, which consisted of breeding groups, non/post- reproductive pairs, or parent- offspring pairs. Only one group included more than one lorisid species and consisted of pygmy lorises and slender loris. 37

54 Figure 1. Age pyramids for lorisid primates in North American institutions. Ages are calculated from the date of birth listed in the species studbook to the date the survey was distributed, January 19, At this time, the population consisted of a total of 104 lorisids: 62 (31 male: 31 female) pygmy lorises, 12 slender lorises (8.4), 15 slow lorises (6.9), and 15 (8.6.1 unknown) pottos. For the potto, the one unknown individual (a neonate) is not included in the age pyramid above but is included in the population total. Twenty groups containing 28 individuals (27%) were identified as mixed- species groups, and 53 groups containing 76 individuals (73%) were described as non- mixed groups. Pygmy lorises were housed with northern tree shrews (Tupaia belangeri), aye- ayes (Daubentonia madagascariensis), Malagasy jumping rats (Hypogeomys antimena), greater Malayan chevrotains (Tragulus napu), and mouse 38

55 lemurs (Mirza zaza). Pottos were housed with African hedgehogs (Atelerix spp.) and black and rufous elephant shrews (Rhynchocyon petersi). Slender lorises were housed with dwarf lemurs (Cheirogaleus medius) and greater Malayan chevrotains. Slow lorises were found with the widest array of different species, including three- banded armadillos (Tolypeutes matacus), brush- tailed porcupines (Atherurus africanus), chevrotains, aye- ayes, Asian small- clawed otters (Aonyx cinerea), Prevost s squirrels (Callosciurus prevostii); and the only avian species currently housed with a lorisid primate, the blue- bellied roller (Coracias cyanogaster). Enclosure Design The majority of facilities maintained lorisid groups in a dedicated nocturnal building. The second most common housing arrangement consisted of indoor exhibits adjacent to diurnal exhibits rather than a specialized nocturnal area (Table 2). Only one facility housed any of these species (slender loris) outdoors. Overall, 26 lorisids (25%) had access to some secondary space; the remaining animals were all confined to a single primary exhibit which was usually indoors. Only pygmy lorises (n = 9, 38%) had access to alternate enclosures on exhibit to the public. Off- exhibit holding spaces were slightly more common and were available to a single slender loris, a single slow loris, and 15 (63%) pygmy lorises. One slender loris and 13 pygmy lorises (54% of those with secondary enclosures) were allowed to move between the primary and secondary enclosure at will; otherwise secondary enclosures were utilized less frequently or only as needed. 39

56 Table 2. Enclosure design features of lorisid primates in North American facilities. Question Response options Pygmy loris % (#) of n = 62 Where is the primary enclosure located? What best describes the ventilation system for the enclosure? What best describes the substrate in the enclosure?^ Does the enclosure contain any hiding or sleeping sites?* indoors, in a dedicated nocturnal house indoors, in a nocturnal- only area (a separate wing) indoors, in an area with both diurnal and nocturnal exhibits outdoors forced air circulation running throughout the building separate ventilation system from public/keeper areas natural ventilation through windows or outdoor access bare concrete mulch dirt variable, combination, or other nest boxes baskets, tubes, or other covered structures 45% (28) 27% (17) 27% (17) 0% (0) 76% (47) 24% (15) 0% (0) 39% (24) 47% (29) 0% (0) 15% (9) 71% (44) 84% (52) Slender loris % (#) of n = 12 92% (11) 8% (1) 0% (0) 0% (0) 100% (12) 0% (0) 0% (0) 0% (0) 33% (4) 0% (0) 67% (8) 100% (12) 100% (12) Slow loris % (#) of n = 15 47% (7) 20% (3) 27% (4) 7% (1) 87% (13) 7% (1) 7% (1) 40% (6) 40% (6) 7% (1) 13% (2) 87% (13) 93% (14) Potto % (#) of n = 15 47% (7) 0% (0) 53% (8) 0% (0) 93% (14) 7% (1) 0% (0) 60% (9) 40% (6) 0% (0) 0% (0) 80% (12) 80% (12) 56% 8% 67% 13% areas of dense foliage (35) (1) (10) (2) 21% 17% 13% 7% other hiding spots (13) (2) (2) (1) ^ Response option for grass substrate was eliminated because it was never selected. * Multiple responses to this question were permitted. All lorisids % (n = 104) 51% 20% 28% 1% 83% 16% 1% 38% 43% 1% 18% 78% 87% 46% 17% 40

57 Figure 2. Primary enclosure size for lorisid primates in North American facilities. Values are for n =104 animals: 62 pygmy lorises, 12 slender lorises, 15 slow lorises, and 15 pottos. Physical conditions were fairly consistent across species. Enclosure sizes were comparable across species with a tendency toward larger enclosures for slow lorises, although exhibit sizes were highly variable within species (Figure 2). Only one facility, which housed a slow loris outdoors, did not control the exhibit temperature. Other lorisids were maintained at similar mean temperatures, given as F (SE): 75.7 (0.3) for n = 60 pygmy lorises; 77.3 (0.3) for n = 12 slender lorises; 76.3 (0.9) for n = 13 slow lorises; and 75.3 (0.5) for n = 15 pottos. In most cases, exhibits shared a forced air ventilation system that ran throughout an entire building (Table 2). Relative humidity (RH) levels in exhibits were actively controlled only for nine (15%) pygmy lorises and one slender loris. Mean (SE) reported values for exhibits were 49.7 (1.7) %RH for n = 56 pygmy lorises, 44.6 (2.2) % RH for n = 12 slender lorises, 46.9 (4.3) %RH for n = 11 slow lorises, and 46.8 (1.6) %RH for n = 14 pottos. 41

58 Exhibit furnishings are also outlined in Table 2. Most lorisids were housed in exhibits with either bare concrete or mulch on the ground. Other flooring was most often described as wire (presumably cage floors). Every individual received some type of hiding spot on exhibit; the most common refuges were nest boxes, followed by tubes or other covered structures, and areas of dense foliage. Lighting Conditions The most common lighting regimen was the same for all species and consisted of an artificially reversed light cycle (Table 3). The majority of animals in each species except the slender loris were kept on fixed lighting regimens, in which phase lengths remained constant throughout the year. The average length (SE) of the fixed dark phase was 11.6 (0.2) hrs across species. Other animals were exposed to seasonal changes in day length, either as a result of exposure to natural light (16%) or intentional manipulation of artificial lighting (23%). If these two cases are grouped together and natural day lengths are calculated based on zoo latitude, the upper and lower bounds of variable dark phase lengths are 9.7 (0.2) hrs and 12.8 (0.3) hrs. Artificial lighting was used to mimic twilight periods for only eight (13%) pygmy lorises and a single slow loris. 42

59 Table 3. Lighting design of the primary enclosure. Question Response options Pygmy loris % (#) of n = 62 How are light and dark phases created in the exhibit? Do the lengths of the light and dark phases change seasonally? What sources of light are used to illuminate the enclosure during the light phase?* What sources of light are used to illuminate the enclosure during the dark phase?* natural day/night cycles with sunlight access artificially created day/night cycles (un- reversed) artificially created reversed light cycle no, phase lengths are fixed yes, phase lengths vary naturally or artificially mimic seasonal change sunlight fluorescent lights incandescent lights LED lamps high intensity discharge lights no light at all moon or starlight fluorescent lights with filters incandescent lights with filters LED lamps high intensity discharge lights with filters colored bulbs of any type other light source 11% (7) 16% (10) 73% (45) 63% (39) 37% (23) 11% (7) 77% (48) 26% (16) 3% (2) 8% (5) 3% (2) 8% (5) 31% (19) 23% (14) 3% (2) 3% (2) 37% (23) 6% (4) Slender loris % (#) of n = 12 67% (8) 0% (0) 33% (4) 17% (2) 83% (10) 67% (8) 100% (12) 0% (0) 0% (0) 0% (0) 0% (0) 67% (8) 17% (2) 25% (3) 0% (0) 8% (1) 17% (2) 0% (0) Slow loris % (#) of n =15 13% (2) 33% (5) 53% (8) 60% (9) 40% (6) 20% (3) 80% (12) 27% (4) 0% (0) 7% (1) 0% (0) 20% (3) 20% (3) 27% (4) 0% (0) 0% (0) 33% (5) 13% (2) Potto % (#) of n = 15 0% (0) 33% (5) 67% (10) 67% (10) 33% (5) 0% (0) 100% (15) 0% (0) 0% (0) 13% (2) 0% (0) 0% (0) 60% (9) 7% (1) 0% (0) 0% (0) 0% (0) 33% (5) All lorisids % (n = 104) 16% 19% 64% 58% 42% 17% 84% 19% 2% 8% 2% 15% 32% 21% 2% 3% 29% 11% 43

60 Question Response options Pygmy loris % (#) of n = 62 What color(s) of 39% red artificial light (24) illuminate the 35% blue enclosure during the (22) dark phase?* 23% white (14) other color/not 21% applicable (13) * Multiple responses to this question were permitted. Slender loris % (#) of n = 12 17% (2) 17% (2) 0% (0) 67% (8) Slow loris % (#) of n = 15 40% (6) 40% (6) 20% (3) 47% (7) Potto % (#) of n = 15 47% (7) 33% (5) 33% (5) 0% (0) All lorisids % (n = 104) 38% 34% 21% 26% Facilities reported a great deal of variation in exhibit lighting fixtures and design (Table 3). Many facilities used more than one light source during dark and light phases. Fluorescent lights were most commonly used during the light phase, and filtered fluorescent lights were often used to brighten the exhibit for visitor viewing during the dark phase. Many institutions reported using more than one light color to illuminate the exhibit during the dark phase. The majority of animals were exposed to some kind of artificial light for all 24 hours of the day. Animal Care Practices Facilities also reported on basic husbandry practices for their groups (Table 4). All lorisids were cared for by at least two animal keepers. In most cases exhibits were cleaned (or at least spot- cleaned) daily. For all species, the majority of individuals received their daily diet in a single meal. However, many institutions noted that diets were scatter- fed or otherwise presented to promote foraging, and daily diets were often supplemented with food used for enrichment. Although the 44

61 vast majority of animals were provided with daily enrichment, training was rarely or never conducted with most individuals. Table 4. Animal care practices. Question How many keepers work directly with this group during a given week?* How often is the primary enclosure cleaned? In how many feedings per day is the main diet presented (not including food used for training)? How often is enrichment presented for this group? Response options two three more than three daily (including spot cleans) 2-3 times per week weekly other frequency one two three daily weekly monthly Pygmy loris % (#) of n = 62 53% (33) 24% (15) 23% (14) 89% (55) 3% (2) 8% (5) 0% (0) 56% (35) 40% (25) 3% (2) 79% (49) 8% (5) 8% (5) Slender loris % (#) of n = 12 67% (8) 17% (2) 17% (2) 92% (11) 0% (0) 8% (1) 0% (0) 92% (11) 8% (1) 0% (0) 100% (12) 0% (0) 0% (0) Slow loris % (#) of n = 15 53% (8) 13% (2) 33% (5) 100% (15) 0% (0) 0% (0) 0% (0) 60% (9) 40% (6) 0% (0) 67% (10) 33% (5) 0% (0) Potto % (#) of n = 15 53% (8) 47% (7) 0% (0) 87% (13) 0% (0) 0% (0) 13% (2) 100% (15) 0% (0) 0% (0) 87% (13) 0% (0) 13% (2) All lorisids % (n = 104) 55% 25% 20% 90% 2% 6% 2% 67% 31% 2% 81% 10% 7% rarely # 5% (3) 0% (0) 0% (0) 0% (0) 3% 45

62 Question How often are training sessions conducted with this group? Are specific members of the group generally removed during parturition and/or infant rearing? Response options daily weekly monthly rarely never # pairs are usually separated # pairs are not usually separated # the zoo has tried both of the above strategies # Pygmy loris % (#) of n = 62 5% (3) 18% (11) 10% (6) 55% (34) 13% (8) 23% (14) 6% (4) 15% (9) Slender loris % (#) of n = 12 0% (0) 0% (0) 0% (0) 83% (10) 17% (2) 17% (2) 0% (0) 0% (0) Slow loris % (#) of n = 15 7% (1) 20% (3) 0% (0) 60% (9) 13% (2) 0% (0) 0% (0) 0% (0) Potto % (#) of n = 15 0% (0) 13% (2) 0% (0) 87% (13) 0% (0) 53% (8) 0% (0) 0% (0) All lorisids % (n = 104) 4% 15% 6% 63% 11% 23% not applicable # 37% 42% 100% 47% (23) (5) (15) (7) 48% no response # 19% 42% 0% 0% (12) (5) (0) (0) 16% How frequently do 37% 67% 60% 53% yearly members of this group (23) (8) (9) (8) 46% receive physical exams every other 31% 17% 20% 47% by a veterinarian?^ year (19) (2) (3) (7) 30% only as needed 31% 8% 13% 0% (19) (1) (2) (0) 21% other 2% 8% 7% 0% frequency (1) (1) (1) (0) 3% * Response option for one keeper was eliminated because it was never selected. # Response option category was added in data analysis based on write- in responses. ^ Response option for twice a year was eliminated because it was never selected on the survey. 4% 9% For all species, most facilities reported giving individuals annual veterinary exams. Finally, breeding pairs were most commonly separated for parturition 46

63 and/or infant rearing; however, most institutions responded that this question was not applicable as their groups were non- reproductive. Individual Characteristics Finally, we asked respondents to estimate each individual s reproductive success (Table 5). The only species reported to include a high percentage of reliable breeders was the potto, and only two slender lorises (a pair) were characterized as reliable breeders. A particularly large number of slow lorises were described as never having bred despite repeated attempts. This question was reportedly not applicable to most animals, although the reasons given for this response varied among species. Most slender lorises were not housed in breeding situations. One slow loris pair was used for education purposes, one female previously underwent a hysterectomy, and the others were housed alone or were not in breeding situations. One aged female potto was presumed to be post- reproductive, and two males did not have available breeding partners other than direct relatives. Pygmy lorises appeared to show a different pattern. Most pygmy lorises were described as potential (but not proven) breeders that had recently been moved among institutions based on breeding recommendations, had only recently been paired with a new breeding partner, or were potentially pregnant at the time the survey was completed. 47

64 DISCUSSION We surveyed all AZA zoos and related facilities (and one non- AZA facility) in North America that currently house lorisid primates. Our data indicate that the typical lorisid is housed solitarily or in a pair; pair housing was more common for slender and pygmy lorises than other taxa. Most lorisids do not reside in mixed- species exhibits. Most animals are housed in buildings dedicated to exhibiting nocturnal animals and have access to only one living space. Animals are most commonly housed under a reversed light cycle, presumably so they are visible to the public during their active period. Most occupy complex environments and receive additional enrichment frequently. Animals are always cared for by at least two keepers, who feed animals and clean exhibits once a day, but rarely conduct training sessions. Reported practices were thus fairly consistent with existing husbandry recommendations for lorises (Fitch- Snyder and Schulze, 2001). The Physical Environment Institutional housing practices are consistent with most current standards for abiotic exhibit design (Fitch- Snyder and Schulze, 2001). Reported temperature and humidity ranges are consistent with those listed in the husbandry manual, and most animals receive at least 12 hours of light a day. Pygmy and slow lorises commonly occupy spaces larger than the recommended minimum while pottos and slender lorises occupy slightly smaller spaces. Daschbach et al. (1982/83) reported that increased space stimulates activity 48

65 in the slow loris, although it should be noted that both cages tested in this study were smaller than current space recommendations. It may be that the quality of space, rather than quantity, has more dramatic effects on lorisid behavior. Frederick and Fernandes (1996) found that naturalizing an exhibit by increasing the amount of vegetative covering and hiding spots produced an increase in activity and social behaviors in two pottos. Almost every institution in our survey provided nest boxes, other hiding places, mulch or other substrates, and daily enrichment to their lorisid groups. These results suggest that many facilities are taking advantage of the positive behavioral benefits of natural, enriched environments (Clark and Melfi, 2011; Lutz and Novak, 2005; Markowitz and Spinelli, 1986). Table 5. Estimated reproductive success of lorisid primates. Reproductive Success Sex Pygmy loris # (%) of n = 62 ( )* Reliable breeder (When given the opportunity, an offspring is almost always produced.) Successful breeder (Does not necessarily produce an offspring after each opportunity.) Slender loris # (%) of n = 12 (8.4.0) 17% (2) 12.5% Slow loris # (%) of n = 15 (6.9.0) 0% (0) 0% Potto # (%) of n = 15 (8.6.1) All lorisids % of n = 96 Combined 16% 40% sexes (10) (6) 17% Male 16% 37.5% (5) (1) (0) (3) - Female 16% 25% 0% 50% (5) (1) (0) (3) - Unknown % (0) - Combined 6% 0% 0% 7% sexes (4) (0) (0) (1) 5% Male 6% 0% 0% 12.5% (2) (0) (0) (1) - Female 6% 0% 0% 0% (2) (0) (0) (0) - Unknown % (0) - 49

66 Reproductive Success Sex Pygmy loris # (%) of n = 62 ( )* 50/50 breeder (Has bred successfully but only produces an offspring about half the time following an opportunity.) Rare breeder (Has only reproduced one time or rarely done so.) Never bred (Despite repeated attempts.) Not applicable or other Combined sexes Male Female 2% (1) 0% (0) 3% (1) Slender loris # (%) of n = 12 (8.4.0) 0% (0) 0% (0) 0% (0) Slow loris # (%) of n = 15 (6.9.0) 0% (0) 0% (0) 0% (0) Unknown Combined sexes Male Female 11% (7) 13% (4) 10% (3) 0% (0) 0% (0) 0% (0) 7% (1) 0% (0) 11% (1) Unknown Combined sexes Male Female 11% (7) 10% (3) 13% (4) 17% (2) 12.5% (1) 25% (1) 40% (6) 50% (3) 33% (3) Unknown Combined sexes Male Female 53% (33) 55% (17) 52% (16) 67% (8) 75% (6) 50% (2) 53% (8) 50% (3) 56% (5) Unknown * Notation indicates: (# males. # females. # unknown individuals). Potto # (%) of n = 15 (8.6.1) 0% (0) 0% (0) 0% (0) 0% (0) 0% (0) 0% (0) 0% (0) 0% (0) 20% (3) 12.5% (1) 33% (2) 0% (0) 33% (5) 37.5% (3) 17% (1) 100% (1) All lorisids % of n = 96 1% % % % One aspect of the abiotic environment that appears to be less consistent among facilities is lighting design. Day lighting is fairly consistent among facilities, but a wide range of light sources and colors are employed to simulate night lighting. 50

67 Strepsirrhines may perceive blue light as brighter than red (Frederick and Fernandes, 1994; Goffart et al., 1976), and therefore red or neutral- density filters are thought to be preferable for night lighting in exhibits (Fitch- Snyder and Schulze, 2001). Despite this, slightly more than half of captive lorisid groups are currently housed under blue light. Furthermore, most facilities we contacted were unable to provide light intensities for their exhibits, despite research suggesting that lorises are less active in brighter exhibits (Frederick and Fernandes, 1994; Trent et al., 1977). Light color and intensity may also have implications for animal health and reproduction. Exposure to light at night is known to suppress production of the timekeeping hormone melatonin in humans (Mirick and Davis, 2008) and other primates (Hoban et al., 1990; Reppert et al., 1981). The suppressive effects of nocturnal light exposure are more pronounced in nocturnal species (Reiter, 1991), as well as under greater light intensities and shorter (blue) wavelengths (Aral et al., 2006; Rahman et al., 2008). In humans, nocturnal melatonin suppression is associated with elevated cancer risk, cardiovascular disease, major depression, metabolic syndrome, and decreased fertility (Arendt, 2005; Navara and Nelson, 2007). In many mammals, melatonin also plays an important regulatory role in seasonal reproduction (Malpaux et al., 1999). It is not clear if lorisids are seasonal breeders (cf. Fitch- Snyder and Jurke, 2003; Izard et al., 1988; Radhakrishna and Singh, 2004) or if reproduction is under photoperiodic control, but captive lighting could conceivably impact reproductive success in these species through this hormonal pathway. Clearly there is a need for further research on the behavioral and physiological 51

68 impacts of lighting on nocturnal primates, and evidence- based guidelines should be developed for nocturnal lighting intensity and color. The Social Environment Our survey results suggest that the social lives of lorisids present management challenges. Nearly half of the animals in our survey were housed solitarily. Although our survey did not directly address aggression, respondents often cited aggression during introductions as the reason why potential breeding pairs had not successfully reproduced. Anecdotal reports of groups being separated due to aggression are also common. Yet, wild pottos may be found in breeding pairs (Pimley et al., 2005), and slender lorises at several field sites have shown high levels of gregariousness (Nekaris, 2001; Radhakrishna and Singh, 2002). It may be that many institutions lack the flexibility in design needed to replicate the dispersed, community style social structure of these species. Few institutions in our survey reported having secondary exhibits or holding areas for animals. Daschbach et al. (1982/83) did not find a relationship between cage size and frequency of agonistic behavior in two slow loris pairs. However, perceived social density appears to moderate aggression in many primate species (Honess and Marin, 2006). It would be fascinating to see the results of research exploring how creative manipulation of exhibit structure and size affects the ability of zoos to manage lorisids in pairs or other groups, as well as if greater social familiarity promotes breeding success (Fisher et al., 2003b). 52

69 Mixed- species exhibits may provide another source of social enrichment (Leonardi et al., 2010). Only one group in our survey contained multiple lorisid species (pygmy and slender loris), although historically pygmy lorises have been housed with both slow and slender lorises (Ferrie et al., 2011). Overall, less than a third of the animals surveyed shared exhibits with non- lorisid taxa. Adequate off- exhibit holding space is considered important for creating successful mixed- species groups (Ferrie et al., 2011), so the lack of secondary spaces identified in our survey may explain why mixed groups are not more common. Our survey does confirm that lorisids tend to mix well with other strepsirrhines such as aye- ayes (Ferrie et al., 2011); although, the success of any particular mixed group will no doubt depend on individual characteristics as well as environmental factors. Perhaps more facilities will take advantage of mixed species opportunities as other nocturnal primates become more common in zoos. When animals cannot be housed socially, olfactory enrichment may provide social stimulation. Scents from other animals or even predator feces can stimulate activity in lorises (Fitch- Snyder and Schulze, 2001). Frequent cage cleaning may also interfere with olfactory communication, which may play an important role in the expression of reproductive behaviors. When experimentally given a choice between two males, female pygmy lorises were more likely to show preference toward a male whose scent was familiar (Fisher et al., 2003b), and males will compete with one another by countermarking over the scents of other males when given the opportunity (Fisher et al., 2003a). Our survey respondents indicated that cages were generally cleaned daily. However, we did not distinguish between intensive cleaning 53

70 measures, which would likely interfere with chemosensory signals, and spot cleaning. Nevertheless, it may be worthwhile for facilities to consider the adequacy of their habitat design for meeting olfactory needs, especially for lorisids in breeding situations. Animal training and other types of keeper interaction may be another source of social stimulation for animals (Pizzutto et al., 2007), providing mental stimulation and acting as a means of environmental enrichment (Laule and Desmond, 1998). The vast majority of institutions in our survey rarely or never conduct training with lorisids. Information on training is notably absent from the current husbandry manual (Fitch- Snyder and Schulze, 2001). Perhaps caretakers do not think that lorises respond well to training; however, our recent success with training pottos and pygmy lorises for saliva collection suggests this may not be the case. Training can have many positive benefits for animal care such as desensitizing animals to fear- invoking stimuli (which could be particularly helpful in shy loris species), improving social management, and promoting voluntary cooperation in husbandry and medical procedures (Laule and Whittaker, 2007; Young and Cipreste, 2004). It seems that institutions may be missing a valuable opportunity to use positive reinforcement training as a tool to promote the welfare of captive lorisids. Population Status Reproductive and demographic data collected in our survey indicate an uncertain future for most lorisid species in North American zoos. The age structures of these populations illustrate that slender lorises, slow lorises, and pottos are 54

71 essentially in demographic crisis. In response, the current nocturnal strepsirrhine breeding and transfer plan has recommended breeding for all available animals in these species (Kuhar et al., 2011). Our survey included questions intended to assess the reproductive potential of these populations. Few slender and slow lorises were described as reliable breeders. Many of these animals are not housed in situations to promote breeding, most likely due to their age or reproductive history. Exacerbating breeding problems is the naturally low reproductive rate of slender (Izard and Rasmussen, 1985) and slow (Izard et al., 1988) lorises. More pottos were identified as reliable breeders in the survey, a designation which is somewhat at odds with the reproductive history of this species in captivity. Given how limited reproductive opportunities are for these species, it is important that institutions follow breeding recommendations to ensure that no prospects are missed. Research may also be beneficial to these populations, as improved knowledge of husbandry in these species may enable caretakers to maximize rare reproductive opportunities. The demographic status of the captive pygmy loris situation is more stable, and pygmy lorises are currently the most common lorisid primate exhibited in North American zoos. Pygmy lorises have qualities which promote population growth such as a higher twinning rates, faster maturation, and shorter lactation lengths than other Nycticebus species (Weisenseel et al., 1998). Although our survey indicated that many animals have not successfully bred at this time, it appears that management changes are in progress and that there is actually great reproductive potential in the current pygmy loris population. 55

72 We conducted a comprehensive survey of animal care and housing practices for slender, slow, and pygmy lorises, and pottos, in North American facilities. Although our survey generally showed that institutional housing practices are consistent with current husbandry guidelines, there are aspects of the physical environment, such as lighting design, that show room for improvement or where additional knowledge is needed. Our survey also highlighted the need for further information on reproduction in captive lorisids, particularly on reproductive seasonality and its environmental triggers. The urgency of this need is underlined by the population declines revealed by our demographic and reproductive data. Even if management of these species in AZA collections is eventually discontinued, lessons learned from their husbandry may inform future efforts to manage the pygmy loris population. As commercial demand for pet lorises grows (Nekaris and Campbell, 2012), it is also likely that husbandry information will be of increasing importance for managing animals in situ in sanctuaries as they are recovered from the wildlife trade. In zoos, preservation of the captive pygmy loris population will ensure that lorisid primates remain as captive ambassadors for educating visitors about the dire threats faced by all wild lorisids, as well as the ecological niche occupied by lorises and their fellow creatures of the night. 56

73 Chapter Three A Retrospective Review of Mortality in Lorises and Pottos in North American Zoos, INTRODUCTION Lorises (genus Loris and Nycticebus) are strepsirrhine primates that dwell in forested habitats across Southeast Asia. Slender lorises include at least two species in India and Sri Lanka (Brandon- Jones et al., 2004), and the slow lorises are a radiation of species extending through Indonesia into southern China. Ongoing investigations are revealing increased diversity within the genus Nycticebus (Munds et al., 2012), just as it is becoming clear that all loris species are declining in the wild (Iseborn et al., 2011; Nekaris and Jayewardene, 2004; Nekaris and Nijman, 2007). In recent years the pet trade has had an increasingly devastating impact on wild slow loris populations (Nekaris and Campbell, 2012; Nekaris et al., 2010). Many animals rescued from this illegal trade have suffered physical injuries that make reintroduction to the wild impossible, and the need for sanctuary housing grows along with the popularity of lorises as pets (Nekaris and Jaffe, 2007). Zoos that have historically housed lorises thus may have the opportunity to contribute to in situ conservation efforts by providing sanctuaries with information on captive loris husbandry and veterinary care. Zoo populations of slender and slow lorises are found in Europe, Asia, and North America. North American zoos have historically housed the pygmy slow loris N. pygmaeus, and a hybridized population of N. coucang and N. bengalensis managed under the coucang moniker. North American zoos have also housed two subspecies 57

74 of red slender loris: L. tardigradus tardigradus and L.t. nordicus, although the North American population of the latter is now extinct. According to Fitch- Snyder and Schulze (2001), the history of slow and slender lorises in North America dates back to late 19 th /early 20 th century exhibits at the Philadelphia and Bronx zoos, while a breeding population of pygmy slow lorises was established in North America in In addition to slender and slow lorises, North American zoos are home to a small population of an African lorisid: the potto, Perodicticus potto. Pottos are thought to be widespread throughout sub- Saharan Africa but their true taxonomic and conservation status is largely unknown (Grubb et al., 2003). Pygmy loris numbers are increasing in zoos, and currently they are the only lorisid species nearing sustainability in North American collections (Kuhar et al., 2011). Populations of other lorisid species are small (fewer than 50 individuals) and declining. Although L.t. nordicus is no longer found in captivity in North America, there are still some individuals managed in Europe. Research into reproduction, as well as veterinary and social needs, may play an important role in maintaining viable populations of these species in captivity (Schulze, 1998). High infant mortality and traumatic death appear to be major impediments to population growth in lorises and pottos. Debyser (1995) analyzed mortality trends for strepsirrhine juveniles in zoos and primate centers and found that more offspring died prior to weaning in N. coucang, L.t. tardigradus, and P. potto than the thirteen lemur and galago species that were also surveyed. Sutherland Smith and Stalis (2001) reviewed mortality in lorises from the San Diego Zoo ( ) compared to Duke University Lemur Center ( ). These data show that in 58

75 addition to trauma, significant contributors to captive loris morbidity and mortality include dental, renal, and respiratory disease. This study expands on this research (with some sampling overlap) and seeks to describe more comprehensively the factors contributing to mortality in captive lorises and pottos. We reviewed postmortem records for L.t. nordicus, L.t. tardigradus, N. coucang, N. pygmaeus, and P. potto born between 1980 and 2010 that lived in North American zoos. Our primary aim was to describe major sources of mortality in these species. In a companion to this paper, we also surveyed facilities that are currently housing lorisid primates to determine how animals are actually cared for in practice (Fuller et al., 2013). This study sets the stage for future hypothesis- driven research examining associations between environmental design, diet, health, and animal behavior. Ultimately, our goal is to identify husbandry concerns and areas of research most urgently needed to address the health, welfare, and population sustainability of lorises and pottos in the care of sanctuaries and zoos. MATERIALS AND METHODS Study Population We requested full medical records from all North American zoos and related facilities for animals born between January 1, 1980 and December 31, 2010 representing five species: Loris tardigradus nordicus, L.t. tardigradus, Nycticebus coucang, N. pygmaeus, and Perodicticus potto. Animals and facilities included in these populations were identified by consulting North American Regional Studbooks for each species as of February Species groups are based on 59

76 studbook data as well. L.t. nordicus and L.t. tardigradus have separate studbooks and were treated as separate populations here. In contrast, N. coucang and N. bengalensis have historically been grouped as one species in North American facilities and the population s hybrid status is uncertain; for these reasons they are treated as a single population (referred to as N. coucang) here. In total, one unaccredited zoo and 32 of 35 zoos (91%) accredited by the Association of Zoos and Aquariums (AZA) where a lorisid primate died during this time period contributed medical records to this study, representing a total of 367 animals. The total number of records received for each species were: 20 (10.9.1) (male.female.unknown) out of 25 L.t. nordicus (80%), 72 ( ) of 80 L.t. tardigradus (90%), 109 ( ) of 152 N. coucang (72%), 133 ( ) out of 173 total N. pygmaeus (77%), and 33 ( ) of 37 P. potto (79%). Across species, this sample included 167 males (45.5%), 158 females (43.0%), and 42 animals of unknown sex (11.4%). The majority of records (241, 66%) were formatted using MedARKS or ARKS (International Species Information Systems, Eagan, MN, USA) but the sample included other electronic formats and hand- written records. Additionally, one facility elected to provide summaries of necropsy and histopathology findings rather than submitting full records. Medical records included in this study varied greatly in detail and scope. Necropsy reports were available in 247 (67%) of the records received, and 214 (58%) included a histopathology report. 60

77 Organ System Classification and Cause of Death Data for each animal were entered into a Microsoft Access database. Age at death, sex, and death type (spontaneous, euthanasia, or unknown) were recorded for each animal. Based on their age at death, the animals within each species were categorized into one of four age groups: neonate, x 30 days old; juvenile, 30 days < x 1 year; and adult, 1 year < x geriatric; and geriatric. Age at which an animal was considered geriatric varied by species and was calculated as 75% of the mean lifespan of all animals in our study population that reached adulthood (i.e. excluding neonates and juveniles). L.t. nordicus were considered geriatric at 9 years old, L.t. tardigradus at 10y, N. coucang at 9y, N. pygmaeus at 8y, and P. potto at 12 years of age. One year was chosen as a common cutoff point between juveniles and adults based on published reports of age at first reproduction in these species (Charles- Dominique, 1977; Fitch- Snyder and Schulze, 2001). Diagnoses from necropsy and histopathology reports were used to determine which organ systems showed evidence of pathology at the time of the animal s death. Organ system classifications were based on Hope and Deem (2006) and included the following categories: Cardiovascular and Hemolymphatic, Central and Peripheral Nervous System (CNS), Dental, Endocrine & Metabolic, Ear, Nose & Throat (ENT), Gastrointestinal, Hepatic & Biliary, Immunologic (including lymph nodes), Integumentary, Musculoskeletal, Ocular, Renal (including the urinary bladder), Reproductive, Respiratory, and Whole Body. Records were also reviewed to determine the cause of each animal s death or primary reason for being euthanized. If a cause of death was not clearly indicated in 61

78 the medical report, one was identified based on organ system pathology described in the record. In addition to the organ systems listed above, cause of death included the following categories specific to neonates: Maternal Neglect, Stillborn & Abortion, and Unexplained Neonate Death. Deaths for neonates were considered unexplained when no diagnosis was given, whether or not a necropsy and/or histopathology exam was performed (and available in the record). For all other age classes, the cause of death was only listed as Unknown if the record contained a necropsy and histopathology report but a primary cause of death was still unclear. When a histopathology report was not available and the cause of death was not clearly stated in the record, the cause of death was undeterminable and was listed as Incomplete Record. For all age classes, in clear cases of environmental or social injury, the cause of death was listed as Trauma; even if death was more immediately attributable to septicemia or infection subsequent to the initial insult. RESULTS Age- Specific Patterns of Mortality For all animals in this study, mean age of death, regardless of death type, was 7.7y (7.1 SD). The mean age at death for each species was: 8.9y (6.6 SD, range ) for L.t. nordicus, 8.0y (7.3, range ) for L.t. tardigradus, 9.5y (6.8, range ) N. coucang, 6.4y (6.3, range ) for N. pygmaeus, and 6.2y (9.0, range ) for P. potto. Taking into account only animals that reached adulthood, the mean age at death for each species was: 11.8y (4.7 SD, N=15) for L.t. nordicus, 12.7y (5.1, N=45) for L.t. tardigradus, 11.8y (5.6, N=88) N. coucang, 10.1y (5.1, N=84) for 62

79 N. pygmaeus and 15.6y (7.3, N=13) for P. potto. Mean age at death for individuals that reached adulthood was nearly equal for males and females in every species tested, and there were no significant differences in lifespan based on sex. For all the species in this study, death most commonly occurred spontaneously (Table 1) rather than as a result of euthanasia. Neonate mortality (death within the first 30 days of life) was high for each species, most strikingly the potto, in which 60% of the sample population died by 30 days of age (Table 2). Of all the animals in this study, 30.2% (111/367) died as neonates. For all species, mortality during the juvenile period was relatively low. Generally, mortality was greatest in neonate and geriatric groups (Table 2). Table 1. Recorded type of death for lorises and pottos in North American facilities Species Death Type Loris tardigradus nordicus N= 20, % (#) Loris tardigradus tardigradus N= 72, % (#) Nycticebus coucang N=109 % (#) Nycticebus pygmaeus N=133, %(#) Perodicticus potto N=33, %(#) Euthanasia 45.0 (9) 29.2 (21) 30.3 (33) 30.1 (40) 21.2 (7) Spontaneous 40.0 (8) 69.4 (50) 63.3 (69) 67.7 (90) 69.7 (23) Unknown 15.0 (3) 1.4 (1) 6.4 (7) 2.2 (3) 9.1 (3) All Animals, N=367, % (#) 30.0 (110) 65.4 (240) 4.6 (17) 63

80 Organ System Pathology and Cause of Death For all the lorises and pottos in this study, multi- systemic disease was the most frequent cause of death or euthanasia, followed by renal disease and trauma (Table 3). Many animals were stillborn or died as neonates due to maternal neglect or other unexplained reasons. Only 2-6% of deaths were attributed to diseases in each of the cardiovascular & hemolymphatic, gastrointestinal, hepatic & biliary, musculoskeletal, and respiratory systems. Less than 2% of deaths were attributed to each of the remaining organ systems. Table 2. Distribution of loris and potto deaths by age class in North American facilities Species Age Class Neonate (x 30 days) Juvenile (30 days < x 1) Adult (1 year < x geriatric*) Loris tardigradus nordicus N= 20, % (#) Loris tardigradus tardigradus N= 72, % (#) Nycticebus coucang N=109 % (#) Nycticebus pygmaeus N=133, %(#) Perodicticus potto N=33, %(#) 25.0 (5) 31.9 (23) 15.6 (17) 34.6 (46) 60.6 (20) (4) 3.7 (4) 2.3 (3) (4) 15.3 (11) 24.8 (27) 21.1 (28) 9.1 (3) 55.0 (11) 47.2 (34) 56.0 (61) 42.1 (56) 30.3 (10) All Animals, N=367, % (#) 30.2 (111) 5.2 (19) 17.7 (65) 46.9 (172) Geriatric *Cutoff age for geriatric animals was based on 75% of the mean adult lifespan in this population: L.t. nordicus > 9y, L.t. tardigradus > 10y, N. coucang > 9y, N. pygmaeus > 8y, and P. potto > 12y. 64

81 For all animals (Table 4), the most common organ system showing lesions or abnormalities at death was the renal system, with 50% of all animals in the study affected. Over 40% of all lorisids sampled also showed pathologies in the respiratory and hepatic & biliary systems. Across species, organ systems in which more than 20% of animals showed signs of disease included the cardiovascular & hemolymphatic, endocrine & metabolic, gastrointestinal, and immunologic systems. Specific diseases affecting more than 20% of animals in each species are detailed below. Neoplasia was responsible for the deaths of 10.6% (39/367) of animals; all of these cases occurred in pygmy and slow lorises, with the exception of one potto (Table 5). Medical records included 80 reports of distinct neoplastic growths in 66 animals; 39 of these were fatal. All neoplasms occurred in geriatric animals with the exception of two adult N. coucang and four adult N. pygmaeus. Both neoplasias reported for L.t. nordicus occurred in males as did five of six cases in L.t. tardigradus. Neoplasia occurred equally between the sexes in N. pygmaeus (13 cases in males and 14 in females) and P. potto (1 male and 1 female), but was more common in female N. coucang (20 cases) than male (9 cases). Most animals that experienced traumas were neonates that died as a result of bite wounds (Table 6). Only cases in which cannibalism clearly occurred prior to death were counted as traumas. Often cases were classified as unexplained neonate deaths because the carcass was scavenged but it was not possible to determine if this occurred pre- or postmortem. Several cases in which conspecific bite wounds led to necrosis, cellulitis, or other infection occurred, most commonly in Nycticebus 65

82 spp. All traumas thought to be related to mortality, even if cause of death was listed under another organ system, are detailed in Table 6. Loris tardigradus nordicus The primary cause of death was attributed to renal pathology in the majority of the L.t. nordicus in this study (Table 3). Records refer to renal failure (2/5) or glomerulonephritis, nephritis, and general glomerulopathy. There were no deaths attributed to neoplasia in this species, although some neoplastic growths were discovered in necropsy (Table 5). The renal and cardiovascular systems were most commonly affected by disease in L.t. nordicus (Table 4). Animals suffering from renal disease had a group of diagnoses including nephritis, glomerulonephritis, fibrosis, renal infarcts, and distention. One case of cystitis was also included in the renal category. Cardiovascular & hemolymphatic diagnoses included cardiomyopathy, leukocytosis, fibrosis, myocarditis, endocardiosis, and epicarditis. CNS diseases included two cases of hemorrhage and notes of age- related histological changes in the brain. Ocular changes included synechiae, cataracts, blindness, and nerve fiber degeneration. Respiratory illnesses included pneumonia, serositis, mineralization, metaplasia, and one case of a benign bronchioalveolar adenocarcinoma. Whole Body changes of note included anorexia, serous atrophy of adipose tissue, and dehydration. 66

83 Table 3. Primary cause of death or reason for euthanasia in lorises and pottos housed in North American facilities Species Cause of Death Cardiovascular & Hemolymphatic Loris tardigradus nordicus N=20 (10.9.1) % (#) Loris tardigradus tardigradus N=72 ( ) % (#) Nycticebus coucang N=109 ( ) % (#) Nycticebus pygmaeus N=133 ( ) % (#) Perodicticus potto N=33 ( ) % (#) All Animals N=367 ( ) % (#) (3) 3.7 (4) 4.5 (6) 6.1 (2) 4.1 (15) Central Nervous System (2) (3) 3.0 (1) 1.6 (6) Endocrine & Metabolic (3) 0.8 (1) (4) Ear, Nose, & Throat (3) 0.8 (1) (4) 67 Gastrointestinal 5.0 (1) (7) 1.5 (2) (10) Hepatic & Biliary 5.0 (1) 5.6 (4) 5.5 (6) 3.0 (4) (15) Immunologic (1) (1) Integumentary (1) (2) (3) Musculoskeletal (2) 2.8 (3) 5.3 (7) (12) Ocular 5.0 (1) (1) (2) Renal 25.0 (5) 18.1 (13) 19.3 (21) 15.0 (20) 6.1 (2) 16.6 (61) Reproductive (4) 0.8 (1) 3.0 (1) 1.6 (6) Respiratory 10.0 (2) 4.2 (3) 2.8 (3) 5.3 (7) 15.2 (5) 5.4 (20) Multi- systemic 5.0 (1) 19.4 (14) 22.9 (25) 15.8 (21) 15.2 (5) 18.0 (66) Trauma 5.0 (1) 13.9 (10) 10.1 (11) 10.5 (14) 9.1 (3) 10.6 (39)

84 Species Cause of Death Loris tardigradus nordicus N=20 (10.9.1) % (#) Loris tardigradus tardigradus N=72 ( ) % (#) Nycticebus coucang N=109 ( ) % (#) Nycticebus pygmaeus N=133 ( ) % (#) Perodicticus potto N=33 ( ) % (#) All Animals N=367 ( ) % (#) Unknown (2) 0.8 (1) (3) Incomplete Record 20.0 (4) 4.2 (3) 5.5 (6) 7.5 (10) 6.1 (2) 6.8 (25) Maternal Neglect (neonates) Stillborn/Abortion (neonates) Unexplained Neonate Death (neonates) 10.0 (2) 1.4 (1) 0.9 (1) 6.8 (9) 6.1 (2) 4.1 (15) (3) 5.5 (6) 9.8 (13) 12.1 (4) 7.1 (26) 10.0 (2) 18.1 (13) 3.7 (4) 6.8 (9) 18.1 (6) 9.3 (34) 68

85 Table 4. Percent of lorises and pottos with pathology diagnosed by organ system upon postmortem examination Species Organ System Cardiovascular & Hemolymphatic Loris tardigradus nordicus N=20 % (#) Loris tardigradus tardigradus N=72 % (#) Nycticebus coucang N=109 % (#) Nycticebus pygmaeus N=133 % (#) Perodicticus potto N=33 % (#) 30.0 (6) 26.4 (19) 33.0 (36) 32.3 (43) 18.2 (6) Central Nervous System 20.0 (4) 16.7 (12) 15.6 (17) 16.5 (22) 18.2 (6) Dental 5.0 (1) 4.2 (3) 4.6 (5) 3.0 (4) 0.0 Endocrine & Metabolic 15.0 (3) 12.5 (9) 35.8 (39) 19.5 (26) 18.2 (6) Ear, Nose, & Throat 15.0 (3) 4.2 (3) 14.7 (16) 6.0 (8) 12.1 (4) Gastrointestinal 15.0 (3) 8.3 (6) 36.7 (40) 22.6 (30) 21.2 (7) Hepatic & Biliary 15.0 (3) 47.2 (34) 51.4 (56) 42.9 (57) 27.3 (9) Immunologic 5.0 (1) 9.7 (7) 35.8 (39) 23.3 (31) 9.1 (3) Integumentary (5) 5.5 (6) 7.5 (10) 0.0 Multi- systemic 5.0 (1) 19.4 (14) 22.0 (24) 14.3 (19) 12.1 (4) Musculoskeletal 10.0 (2) 11.1 (8) 14.7 (16) 15.0 (20) 6.1 (2) Ocular 25.0 (5) 19.4 (14) 4.6 (5) 6.8 (9) 3.0 (1) Renal 50.0 (10) 41.7 (30) 57.8 (63) 51.9 (69) 33.3 (11) All Animals N=367 % (#) 30.0 (110) 16.6 (61) 3.5 (13) 22.6 (83) 9.3 (34) 23.4 (86) 43.3 (159) 22.1 (81) 5.7 (21) 16.9 (62) 13.1 (48) 9.3 (34) 49.9 (183)

86 Species Organ System Loris tardigradus nordicus N=20 % (#) Loris tardigradus tardigradus N=72 % (#) Nycticebus coucang N=109 % (#) Nycticebus pygmaeus N=133 % (#) Perodicticus potto N=33 % (#) Reproductive 15.0 (3) 9.7 (7) 22.9 (25) 12.8 (17) 9.1 (3) Respiratory 25.0 (5) 47.2 (34) 46.8 (51) 46.6 (62) 30.3 (10) All Animals N=367 % (#) 15.0 (55) 44.1 (162) Whole Body 20.0 (4) 12.5 (9) 11.0 (12) 17.3 (23) 18.2 (6) 14.7 (54) 70

87 Loris tardigradus tardigradus Cause of death in the red slender loris was most commonly attributed to multi- systemic disease (Table 3). Many of these deaths (5/14) were attributed to septicemia, one to hepatitis secondary to bacterial infection, and one to infection- associated granulomatous lesions attacking multiple organ systems. In the remaining cases, animals were geriatric or had problems affecting multiple organ systems with no obviously primary condition. Deaths involving the renal system were also common and were attributed generally to renal failure (6/13), fibrosis (1/13), or glomerulonephritis or nephritis (6/13). Trauma and neonatal mortality were significant in this group as well (Table 6). Of the ten deaths attributed to trauma, two were bite wounds (with subsequent infection in one case) that lead to juvenile deaths, and six others were neonate deaths. Traumatic deaths in neonates occurred due to traumatic bone injury (1/9), aggression from a neighboring animal (1/9), and within- group aggression leading to bite wounds and death (5/9). Additionally, one neonate death was attributed to maternal neglect and 14 were unexplained. Like L.t. nordicus, there were no deaths attributed primarily to neoplasia in L.t tardigradus. Overall, the most common pathologies affecting L.t tardigradus occurred in the hepatic & biliary, and respiratory systems (Table 4). Liver disease was common and included cirrhosis, hepatitis, hyperplasia, congestion, fibrosis, lipidosis, vacuolar change and degeneration, and one presumptive case of hemosiderosis. Reports of liver neoplasias included one hepatoma and a hepatocellular adenoma. One animal reportedly suffered from an obstruction in the gall bladder and another from 71

88 cholangitis. In the respiratory system, pneumonia was commonly reported. Other respiratory changes included atelectasis, hemorrhage, congestion, hemosiderosis, edema, fibrosis, and bronchitis. Cardiovascular issues included congestive heart failure, myocardial atrophy, fibrosis, myocarditis, and congestion. Vascular problems were also common including vasculitis, histiocytosis, arteriopathy, atherosclerosis, and thrombosis. Renal disease was a common finding, and in addition to the common renal changes previously listed for L.t. nordicus, there was one report of a renal cyst and one of a renal cell carcinoma (Table 5). Nycticebus coucang Like the red slender loris, the most common causes of death for slow lorises were multi- systemic and renal diseases (Table 3). Neoplasia (Table 5) affecting multiple organ systems accounted for several (6/23) cases and included two adenocarcinomas and two lymphosarcomas. Septicemia was responsible for three deaths and infections for five. One animal reportedly died from toxicosis but a toxin was never identified, and the remaining cases were all attributed to non- infectious diseases processes (often age- related) affecting multiple organ systems. Renal diseases included two neoplasias, cases described generally as renal failure (7/21), and nephritis, including pyelonephritis and glomerulonephritis (12/21). After renal and multi- systemic disease, trauma was the most frequent cause of death in slow lorises (Table 6). Most victims of trauma (6/11) were neonates. One animal was euthanized due to pathological maternal over- grooming, and the others died following bite wounds and neglect from parents. Two juveniles died of trauma 72

89 as a result of parental aggression, one from multiple skull fractures and the other from necrosis and cellulitis resulting from bite wounds. Three adults died as a result of traumatic events. One animal developed cellulitis secondary to wounds received from a conspecific, while another escaped an enclosure and was attacked by an ocelot. A four- year old male succumbed to acute heart failure following an immobilization made necessary because the loris had become entangled in a cargo net used as exhibit perching. 73

90 Table 5. Neoplasia reported for lorises and pottos in North American facilities Cases marked * were cited as the primary cause of death or reason for euthanizing the animal. 74 Species Organ System Cardiovascular & Hemolymphatic Central Nervous System Endocrine and Metabolic Loris tardigradus nordicus 4 cases Loris tardigradus tardigradus 6 cases Nycticebus coucang 34 cases Nycticebus pygmaeus 34 cases sarcoma (heart) ependymal (brain) - adenoma (parathyroid) - myelolipoma (adrenal) Ear, Nose & Throat - - Gastrointestinal - - Hepatic & Biliary adenoma (1 adrenal, 1 thyroid) - adenoma (liver) - hepatoma (liver) Immunologic Integumentary adenoma (adrenal) - carcinoma (pancreas*) - squamous cell carcinoma (larynx/pharynx*) - sarcoma (large intestine*) - 2 adenoma (both liver) - 4 carcinoma (all liver***) - unspecified mass (liver*) - astrocytoma (brain stem*) - ependymoma (cerebrum*) - adenoma (thyroid) - sarcoma (pancreas) - squamous cell carcinoma (tongue*) - lymphosarcoma (small intestine*) - adenocarcinoma (liver) - adenoma (liver*) - lymphosarcoma, leukemic (liver*) - lymphoma (spleen) - 2 lymphosarcoma (both spleen*) - spindle cell sarcoma (dorsum*) Perodicticus potto 2 cases

91 75 Species Organ System Loris tardigradus nordicus 4 cases Musculoskeletal - - Ocular Renal Reproductive granulosa cell tumor (ovary) Loris tardigradus tardigradus 6 cases - melanoma (iris) - carcinoma (kidney) - Nycticebus coucang 34 cases - adenocarcinoma (skeletal muscle*) - spinal mass Nycticebus pygmaeus 34 cases - fibroma (tarsal region) - 3 fibrosarcoma (2 lumbar**, 1 elbow*) osteosarcoma (rib*) adenocarcinoma (kidney*) - carcinoma (bladder*) - 2 adenocarcinoma (1 mammary, 1 uterus*) - carcinosarcoma (uterus*) - 2 leiomyoma (both uterus*) - multiocular mass (mammary) - prostate mass - uterine mass* - leiomyosarcoma (bladder*) - spindle cell sarcoma (kidney*) - adenoma (seminal vesicle) - 2 carcinoma (1 testis, 1 uterus) - 2 leiomyoma (both uterus*) - granulosa cell tumor (ovary) - Leydig cell tumor (testis) - 2 unspecified testicular mass Perodicticus potto 2 cases - - adenocarcinoma (kidney) - carcinoma (endometrium*) Respiratory - carcinoma (bronchioalveolar) adenoma (both lung) - carcinoma (lung) - carcinoma (lung*) -

92 Species Organ System Multi- systemic Loris tardigradus nordicus 4 cases - Loris tardigradus tardigradus 6 cases - Nycticebus coucang 34 cases - unspecified mass (abdominal) - 2 adenocarcinoma ** - 2 lymphosarcoma** - sarcoma* - transitional cell carcinoma Nycticebus pygmaeus 34 cases - unspecified mass (abdominal*) - lymphoma* - lymphosarcoma (leukemic) - metastatic histiocytic round cell tumor* Perodicticus potto 2 cases - 76

93 Neoplasia was common in slow lorises; including the cases above, 17.4% of animals (19/109) died or were euthanized due to neoplasia, meaning it was one of the leading causes of death in captive slow lorises (Tables 3,5). All three reproductive neoplasia affected females and were uterine in origin. Other notable causes of death were two probable cases of diabetes mellitus. Four animals died from focal infections: myocarditis, otitis, rhinitis, and a chronic abscess of the thigh muscle. Overall, more than 45% of slow lorises showed some lesions or abnormality in their hepatic & biliary, respiratory, and renal systems. Common liver diseases reported were hemosiderosis, hepatic lipidosis, hepatitis, and vacuolar change. The most common renal lesions were due to nephritis, glomerular or pyelonephritis. Diseases affecting the bladder occurred in eight animals, including cystitis (six cases), serositis, and two bladder cancers. Pneumonia was the most common respiratory finding, although several neoplasias were reported in this organ system as well (Table 5). Several other organ systems showed more than a 20% prevalence of disease. Cardiovascular disease noted at death included cardiomyopathy, fibrosis, endocardiosis, and myocarditis. Diseases of the endocrine & metabolic system targeted the adrenal glands (adrenalitis, amyloidosis, hemosiderosis), pancreas (mainly hyperplasia), and thyroid (goiters, cysts). The most common gastrointestinal diseases were enteritis and gastritis, in addition to parasitism in the GI tract. Immunological changes noted at death included histiocytosis, hyperplasia, and neoplasia of the lymph nodes, and splenitis and congestion affecting the spleen. 77

94 Nine animals showed evidence of hemosiderosis affecting the spleen and/or lymph nodes. Reproductive diseases affected females more than males and included cervicitis, pyometra, plancentitis, and ovarian cysts. Reproductive neoplasms occurred in several animals (Table 5), including cancers of the uterus, ovary, and prostate. Two geriatric animals- one male and one female- had mammary neoplasms. In general, cases of neoplasia were largely limited to geriatric animals. Nycticebus pygmaeus Neonatal mortality had a significant impact on the study group, accounting for 34.6% of deaths in our sample of pygmy slow lorises (Tables 2, 3). Trauma (11/46) and maternal neglect (9/46) caused neonate deaths in 43.3% (20/46) of cases. When cases of unexplained neonate death that reference carcass scavenging are included, this total increases to 50%. Traumas experienced by neonates included head trauma (4/11), inter- group aggression (1/11) and intragroup aggression (6/11). In several cases neonates were cannibalized, and one infant was euthanized after losing several limbs. Four mature pygmy lorises died as a result of trauma. Two adults died from what was thought to be an acute toxicosis, but a toxin was never identified. One geriatric male died of septicemia following infection from an armadillo bite, and a 3y- old gravid female was killed by an ocelot after escaping her home enclosure (Table 6). The most common causes of death in pygmy slow lorises were renal and multi- systemic diseases. Renal diseases cited as the cause of death were renal failure (6/20), nephritis (10/20), and one case in which a kidney abscess led to 78

95 peritonitis and eventual death. Two animals died from complications related to cystitis, which led to bladder necrosis in one case. Renal neoplasms were considered the primary cause of death for two animals (Table 5): one geriatric male (14y) developed a spindle cell sarcoma in the kidney, while an adult (7y) female was euthanized due to a leiomyosarcoma of the bladder. Three additional neoplasias were listed as multi- systemic deaths (3/19), all for geriatric animals. Other deaths coded as multi- systemic included cases of septicemia (7/19), other infectious processes (5/19), and non- infectious disease processes and aging (6/19). Neoplasias reported as the primary cause of death were found in nearly every organ system and accounted for 14.3% (19/133) of pygmy loris deaths overall (Table 5). The organ systems most frequently affected at death for the pygmy loris are described in Table 4. Organ systems which showed pathology for more than 20% of animals were the cardiovascular & hemolymphatic, gastrointestinal, hepatic & biliary, immunologic, renal, and respiratory systems. Neoplasia was common throughout all organ systems (Table 5). 79

96 Table 6. Circumstances surrounding traumas related to death in lorises and pottos in North American zoos Species Trauma type Loris tardigradus nordicus Loris tardigradus tardigradus Nycticebus coucang Nycticebus pygmaeus Perodicticus potto 80 Bite wounds and cannibalism Infection/disease secondary to bite wound - 1 neonate (housed with dam and other (?) cagemates, bite wound and fall during intragroup aggression) neonates (one from conspecific neighbor, others could have been dam or sire) - 2 juveniles (one from cagemate, other could have been dam or sire) - 1 adult (septicemia) - 2 geriatric (1 septicemia from month- old bite wound; 1 lung edema possibly related to bite wound thought to have healed) - 5 neonates (1 killed by dam, 2 suspect sire or other male; 2 could have been dam or sire) - 1 juvenile (suspect male cagemate) - 1 juvenile (cellulitis and necrosis on hands and feet, bite from dam or male neighbor (?)) - 3 adults (1 severe necrotizing myositis of bite wound from cagemate; 1 cellulitis from cagemate bite; 1 had chronic sinusitis resulting from bite received as an infant from dam); - 1 geriatric (euthanized for neoplasia, also had other recent problems including bite wounds) - 6 neonates (1 suspect sire; 1 conspecific neighbor; housing situation unclear for others) - 1 neonate (septicemia from hand mutilated by mother) - 1 juvenile (cellulitis, history of chronic, poorly healing bite wounds) - 1 geriatric (purulent abscess from bite wound) - 1 geriatric (infected bite from armadillo) - 1 neonate -

97 81 Species Trauma type Cases of scavenging presumed to have occurred postmortem Head trauma/ traumatic fall Loris tardigradus nordicus Other - - Loris tardigradus tardigradus Nycticebus coucang Nycticebus pygmaeus Perodicticus potto - 1 neonate - 10 neonates - 1 neonate - 6 neonates - 2 neonates neonate - 1 adult (related to metabolic condition?) neonate (maternal over- grooming) - 1 adult (interspecific predation) - 1 adult (accident involving cage furnishings) - 4 neonates (3 incurred during intragroup aggression) - 1 juvenile (possible fall related to maternal neglect?) - 1 geriatric (spinal fracture, age- related?) - 1 adult with fetus (interspecific predation) - 1 geriatric (bite wounds contributed to decision to euthanized, cause of death listed as renal failure) - 2 neonates -

98 Perodicticus potto Most of the pottos in this study died as neonates (Tables 2,3). All reported cases of trauma involved neonates, and two of seven unexplained neonate deaths involved scavenged bodies. Two additional neonate deaths were attributed to maternal neglect and one to a systemic infection. Additionally, three of five respiratory deaths occurred in neonates. The majority of other deaths occurred in geriatric animals. Multi- systemic causes of geriatric death were all related to multiple organ failure with associated shock. Other geriatric deaths were attributed to cardiomyopathy, endocarditis, renal failure, and one endometrial carcinoma representing the only death attributed to neoplasia in all pottos (Table 5). Lesions occurring in more than 20% of animals were found in the gastrointestinal, hepatic & biliary, renal, and respiratory systems (Table 4). Gastrointestinal pathologies included esophagitis, enteritis, and colitis. Liver conditions reported were cirrhosis, hepatitis, congestion, and necrosis. Most renal pathologies were varieties of nephritis; however, a geriatric male and female were both diagnosed with polycystic renal disease. The majority of respiratory disease was pneumonia. Other interesting lesions discovered at death included four animals with rhinitis, and extensive arteriosclerosis in three animals. DISCUSSION We reviewed medical records for 367 lorises and pottos that died in North American zoos over a 30- year period. Our results clearly show that poor neonate 82

99 survivorship is a major concern both for individual health and welfare, as well as population management. However, animals that survive the critical first month of life are likely to reach sexual maturity and live into the geriatric stage. Adult and geriatric animals are most likely to die of renal disease or multi- systemic issues such as systemic infections, neoplasia, or multiple organ failure. High neonate mortality was due to high stillbirth percentages as well as trauma, which was often inflicted by conspecifics. Adults also fall victim to trauma from cagemates, suggesting that greater efforts to address the social management of lorises in captivity would likely have positive impacts on population sustainability and animal health. Other studies have also found exceptionally high infant mortality in lorisiform primates living in both zoo and sanctuary settings. Streicher (2004) summarized health problems experienced by pygmy lorises confiscated from the wildlife trade in Vietnam. Of the 15 deaths reported over an 8- year period, 11 were animals less than 1 year old; overall, 42% of the sanctuary animals died during their first year of life. Debyser (1995) found infant mortality to be higher among lorises and their close relatives compared to other strepsirrhines, and like others (Streicher, 2004; Tartabini, 1991) speculated that captivity- induced stress was likely an underlying cause of infant deaths involving maternal neglect and trauma. Information about social conditions was difficult to infer from our medical records, but it was clear that infants in this group were killed by sires, dams, and other conspecifics. Several different fitness benefits have been hypothesized to explain infanticide among nonhuman primates, including male- male competition and resource competition (Hrdy, 1979). Infanticide risk may be related to social 83

100 density, which is interesting considering that cage sizes for captive pottos and slender lorises in North America are currently smaller than the minimum size recommended by Fitch- Snyder and Schulze s (2001) husbandry manual (Fuller et al. 2013). Deaths involving maternal neglect in this study hint at a pattern similar to that described for captive Galago crassicaudatus umbrosus by Tartabini (1991). Failure to perform maternal behavior, for example due to poor socialization or stress related to the captive setting, leads to infant starvation and death, and the dead infant becomes a resource to consume (Tartabini, 1991). If this is the case, then careful attention to postpartum maternal behavior and if necessary, swift intervention may be important for saving infant lives. Maternal parity and litter sizes are also associated with infant survivorship (Pollock, 1986), and new mothers or multiple births will likely require special attention. Management of group composition around the perinatal period is likely to have a large impact on infant survivorship. Nekaris (2003a) speculates that paternal care may be one factor promoting the relatively greater fecundity of slender than slow lorises, and paternal care has also been observed in captive pottos (Frederick, 1998). These kinds of socialization opportunities are important for the formation and maintenance of social groups, and the 2001 husbandry manual states that it is preferable to keep groups intact through births unless there is cause to suspect an individual needs to be removed (Fitch- Snyder & Schulze 2001). However, it is possible that social management practices surrounding parturition have changed in recent years, perhaps to address traumatic infant death. A 2010 survey of AZA facilities housing lorisids showed that very few males and females were housed 84

101 together during the period surrounding birth, although more facilities indicated that they had attempted this strategy in the past (Fuller et al. 2013). This survey also showed that a large number of animals in each species were solitarily housed on a perpetual basis (Fuller et al., 2013). It may be that efforts to increase the social wellbeing of captive lorisids by providing social partners may be in conflict with strategies needed to minimize infant mortality. Given the tenuous status of these captive populations, isolating gravid females is probably a sound strategy until the proximate causes underlying infanticide are better understood. It was often not possible in our study to distinguish between cases of infanticide, neglect, or trauma, or other possible causes of death; because infant carcasses were cannibalized. Although recovery of nutritional resources is a possible explanation for cannibalism, this behavior has been rarely observed in wild primates; if cannibalism is truly more prevalent in captive and reintroduced animals, then it is likely pathological in nature (Dellatore et al., 2009; Tartabini, 1991). However, lack of tissues for necropsy renders it impossible to determine if infants actually died due to congenital or other disease states, or if infants were neglected by mothers because they were weak or ill. Greater efforts should be made to promptly remove deceased infants so that postmortem exams can clarify these issues. Traumas were a significant contributor of mortality to adult as well as immature animals. In several cases, animals died following bite wounds that were chronically non- healing, leading to necrosis and cellulitis. In one case, a 3y- old adult slow loris died from chronic sinusitis from an infection that originated with a bite 85

102 wound sustained from his mother as a neonate. It has been speculated that a chemical produced by the brachial gland, mixed with slow loris saliva, may be the source of toxicity in bites (Krane et al., 2003). Necrotic bite wounds have also been reported in sanctuary- housed pygmy lorises (Streicher, 2004), and more information on the chemical structure and physiological role of loris venom may aid caretakers in the treatment and management of wounds. Across the species in this study, renal disease was the second most common cause of death, and the renal system exhibited the greatest frequency of lesions at death compared to all other organ systems. Renal changes such as glomerulosclerosis and nephritis are known to be prevalent in captive strepsirrhines and are often age- related (Burkholder 1981, Fitch- Snyder & Schulze 2001, Junge 2003). Renal disease appears to occur more frequently in lorisoids (lorises, pottos, and galagos) than lemur species (Boraski, 1981). Iron storage disease, or hemosiderosis, is a common pathology in lemurs (Benirschke et al., 1985) but was not very prevalent in the lorises and pottos in this study. A review of necropsies of L. tardigradus housed at a German university revealed several cases of cholelithiosis or gallstones, all of which were composed of cholesterol and were speculated to be related to dietary factors like the presence of egg yolk and the reduced insect composition in the captive diet (Plesker and Schulze, 2006). Of the two cases of gallstones in this study, one occurred in L.t. tardigradus and one in L.t. nordicus, but none were reported for slow lorises or pottos. We also recorded one case of diabetes mellitus in L.t. tardigradus, five in N. coucang, and two in N. pygmaeus. Captive lorises are prone to obesity (Ratajszczak 1998, Fitch- Snyder 86

103 & Schulze 2001), but body condition was not reported consistently at death so its role in the development of degenerative diseases is not evident here. It seems likely that diet is a contributing factor to non- infectious and degenerative diseases in captive lorises. The low metabolic rate characteristic of lorises is associated with a natural diet high in toxic secondary compounds from insects and plants (Wiens et al., 2006). It is tempting to speculate that metabolic derangement may occur when a diet this specialized cannot be replicated in the captive setting. The Mysore slender loris (Loris lydekkerianus lydekkerianus) feeds almost exclusively on animal prey, many of which are toxic species (Nekaris and Rasmussen, 2003). New data from the field has highlighted the importance of nectars and exudates in the slow loris diet (Nekaris, 2009; Tan and Drake, 2001); at one study site in India, N. bengalensis fed almost exclusively on exudates during the winter season (Swapna et al., 2010). Lorises often obtain gums by gouging at trees, and the lack of opportunities to perform this behavior in captivity may be associated with periodontal disease (Nekaris, 2009). We were surprised by how rarely dental disease was indicated on histopathology and necropsy reports of the animals we studied. It seems likely that in this study, many necropsy reports did not include notes of missing teeth or gingival disease that may have occurred many years earlier and/or seemed unrelated to the immediate cause of death. Dental disease has previously been reported as a major source of morbidity for Asian lorisines in captivity (Sutherland Smith and Stalis, 2001). In a sample of 25 L. tardigradus at a German university, 7 individuals had missing or loose teeth, or severe calculus at death, and another 4 87

104 animals showed evidence of inflammation and infection secondary to dental disease (Plesker and Schulze, 2013). In this study, several cases were noted in which a retrobulbar or other facial abscess occurred secondary to dental infection. Dental disease may play an under- recognized role in loris health, and more efforts should be made to institute preventative care during routine exams (Plesker & Schulze 2013) and to identify dental abnormalities or infections at necropsy. Around ten percent of the animals in our sample died as a result of neoplasia. A malignant lymphoma that occurred in a slow loris in this sample has been previously described elsewhere and was thought to be caused by a herpes viral infection (Stetter et al., 1995). Our sample also overlaps with neoplasias reported for the Duke Lemur Center (DLC), which contributed records to this study, and at which cases of neoplasia have been thoroughly described (Remick et al., 2009; Zadrozny et al., 2009). However, cases of hepatic neoplasia were reported from four additional facilities in this study, suggesting liver cancer may play an important role in mortality for AZA loris populations as a whole. Our results concur with those reviewed by Remick et al. (2009) that neoplasia in lorises and pottos, like other strepsirrhines, commonly occurs in the digestive (including liver), hematopoietic (here these were generally classified as multi- systemic), and reproductive systems. Little research has addressed the etiology of cancers affecting captive strepsirrhines, and further research into the role of diet (Bingham et al., 1976; Cowgill et al., 1989), exhibit lighting (Navara & Nelson 2007, Fuller et al. 2013), and other factors contributing to neoplasia is needed. 88

105 After reviewing death records spanning 30 years for all lorises and pottos in North American zoos, it is clear that social management remains a challenge in these species. Deaths due to trauma are common and most likely to affect neonates and juveniles before they have the opportunity to reproduce. These trends suggest that targeting efforts to improve infant survival could have major benefits for the sustainability of captive populations. Animals that live to adulthood are likely to die as a result of infectious agents, neoplasia, or degenerative changes. Renal disease remains a major source of pathology for all lorises and pottos, and further investigation is needed to understand its etiology. A greater understanding of the role of diet in the development of non- infectious disease is also needed. 89

106 Chapter Four Validating Actigraphy for Circadian Monitoring of Behavior in the Pygmy Loris (Nycticebus pygmaeus) and Potto (Perodicticus potto) INTRODUCTION Animals have evolved internal time keeping systems, or biological clocks, that allow them to predict and respond to regular patterns of environmental change that occur on a daily or seasonal basis (Rietveld et al., 1993). Biological clocks coordinate changes in physiology and behavior with environmental conditions to which they are synchronized by ambient light levels or other reliable zeitgebers (German for time keeper ) (Challet, 2007). Internal clocks also have the flexibility to adjust to acute, short- term environmental changes termed masking agents (Erkert, 2008; Rietveld et al., 1993). The interactions between internal clocks, zeitgebers, and masking agents produce rhythmic outputs of behavior in the form of activity patterns (Fernandez- Duque, 2003). Investigating how biological clocks interact with ecological factors to shape animal activity patterns has been termed chronoecology (Halle and Stenseth, 2000a). In primatology, this theoretical approach has illuminated how environmental conditions, including light and temperature, interface with predator behavior, morphology, metabolic needs, and other forces to structure primate activity patterns in a complex, species- specific manner (Bearder et al., 2002; Curtis and Rasmussen, 2006; Fernandez- Duque, 2003). All species in the Family Lorisidae (Loridae) (Groves, 2001) are small- bodied, solitary, nocturnal foragers that consume high- energy diets comprised of fruits, 90

107 exudates, and small animal prey (Bearder, 1987). Lorisids are adapted to a slow way of life; they have a low basal metabolic rate, low activity rates, and low levels of energy expenditure (Wiens et al., 2006). Lorisids climb slowly along substrates in a quadrupedal stance, a locomotor pattern which may serve to aid in the detection of food or avoidance of predators, or promote olfactory marking behavior (Oates, 1984). Studies of captive slow lorises (Nycticebus coucang, Kavanau, 1976) and free- ranging pottos (Perodicticus potto, Charles- Dominique, 1977) indicate that patterns of activity in these nocturnal species are largely regulated by photic conditions. Additionally, their nighttime activity periods are often punctuated with periods of inactivity (Charles- Dominique, 1977); for example, Nekaris (2001) found that the Mysore slender loris (Loris tardigradus lydekkerianus), spends up to 46% of its nighttime active period inactive. In short, lorisids are slow, inactive animals that may be sensitive to light. AZA- accredited zoos in North America currently house four lorisid species: the African potto and the Asiatic slender (L. tardigradus), slow, and pygmy (Nycticebus pygmaeus) lorises. Several lines of evidence suggest that husbandry practices alter activity patterns in these species. Potto activity levels are affected by exhibit lighting design (Frederick and Fernandes, 1994) and complexity (Frederick and Fernandes, 1996), and slow lorises are also more active in larger, more enriched cages (Daschbach et al., 1982/83). Oswald and Kuyk (1978) found that activity in three slow lorises peaked during hours that visitors were present in their nocturnal house, and Kavanau (1976) found that the time of onset of nocturnal activity was affected by extreme temperatures in an outdoor colony of slow lorises. From a 91

108 chronoecological perspective, husbandry and housing practices represent both artificial zeitgebers and masking agents which interact to shape the activity patterns of captive animals (Richter, 2006). Because a chronoecological approach requires behavioral monitoring across the 24- hour cycle, it is necessary from a practical standpoint to have an automated means of data collection. It is becoming more common in zoos to utilize video for data collection, and time- lapse video techniques can be used to examine behavior over long time spans, while infrared technology facilitates observations on nocturnal species (London et al., 1998). The utility of video may depend on the quality of the technology employed; Munoz- Delgado et al. (1995) were able to study behavioral sleep in stumptail macaques (Macaca arctoides) by using high sensitivity video equipment to record subtle behaviors such as eye movements and myoclonus. High quality video equipment can be expensive, and in zoo exhibits, it may be difficult to visualize entire exhibits without installing multi- camera systems. Another potential disadvantage of video observation is that a live observer is still needed to score behaviors from video recordings at a later time, so the procedure can be time intensive. In some cases, actigraphy may be a preferable option over video observation. Actigraphs are small accelerometers which can be used to track animal activity patterns when worn by subjects (Mann et al., 2005). Actigraphs record gross movement patterns by means of an omni- directional sensor that records the degree and speed of motion over pre- determined sampling epochs. The average level of activity over a given epoch is calculated by integrating the intensity, number, and 92

109 duration of accelerations and is reported as an activity count (Muller and Schrader, 2003). Actigraphs have been employed for activity monitoring in a variety of species, including African elephants (Loxodonta africana, Rothwell et al., 2012), elk (Cervus elaphus nelsonii, Naylor and Kie, 2004), dairy cows (Muller and Schrader, 2003), and human children (Finn and Specker, 2000). They have been widely used with nonhuman primates such as rhesus monkeys (Macaca mulatta, Papailiou et al., 2008), spider monkeys (Ateles geoffroyi, Munoz- Delgado et al., 2005), callitrichids (Kantha and Suzuki, 2006b), and a single red- ruffed lemur at a zoo (Varecia variegata rubra, Sellers and Crompton, 2004). Fernandez- Duque and Erkert (2006) also measured lunar variation in patterns of activity in free- ranging owl monkeys (Aotus azarai azarai) wearing accelerometers on collars. Actigraphs are also widely used for monitoring sleep patterns in a variety of species (Kantha and Suzuki, 2006b). However, the utility of actigraphy for behavioral monitoring in lorisids may be compromised by the exceptionally slow locomotor patterns of these species. The location where the device is placed on the animal can also affect activity recordings (Muller and Schrader, 2003; Papailiou et al., 2008), so it is important to validate actigraph measures for each new species tested. The goal of this study was to validate the use of actigraphy for monitoring behavior in the pygmy loris and potto. We first sought to assess if wearing the actigraph device compromised animal safety or produced changes in behavior. We then examined whether different intensities of activity recorded by the actigraph device were correlated with specific behaviors, in order to understand the 93

110 resolution with which this technique can be applied for behavioral monitoring. Ultimately we aimed to validate methods that may be widely applicable to investigating topics such as captive animal welfare, mechanisms controlling behavioral rhythmicity, and environmental forces that shape animal behavioral patterns in captive and field settings. MATERIALS AND METHODS Subjects and Housing We worked with several different individuals to conduct initial pilot tests of the actigraph harness. We first tested the harness alone using an 18- yr old female slow loris. We also attempted to use the actigraph harness with two additional male pygmy lorises, one of whom was three yrs old and the other eleven years. However, none of these animals wore the harness for an extended period of time, and these individuals are not included in any data analysis. Our subjects were a single male pygmy loris (nine yrs old) and a male potto (16 yrs), which both wore the actigraph harness for extended periods of time. Both individuals were born in captivity and parent reared. These animals were housed in separate exhibits in the Primate, Cat, and Aquatics building at Cleveland Metroparks Zoo (CMZ) in Cleveland, Ohio, and tests were conducted during October and November,

111 Figure 1. a) Attaching the actigraph harness to the pygmy loris subject; b) the potto wears the actigraph harness on exhibit. The pygmy loris was housed alone in an exhibit measuring 2.6 x 3.0 x 2.9 m. The exhibit contained a wooden nest box, gunnite rock fixtures, wooden perches covered with artificial vegetation, and mulch substrate. The loris was housed on a 12:12 light- dark (LD) cycle with the dark phase from hrs. Light phase illumination was provided by fluorescent and high- intensity discharge lamps, and dark phase lighting with fluorescent fixtures covered by blue gel filters. The potto was housed with a nulliparous female potto (a potential breeding partner) in a similarly furnished exhibit that measured 2.9 x 4.9 x 4.3 m. The exhibit also contained a breeding pair of black and rufous elephant shrews (Rhynchocyon petersi). Lighting conditions for the potto were the same as for the pygmy loris, except that dark phase onset occurred at 2200 hrs in this exhibit and fluorescent lights were covered with both red and blue gel filters. Both the potto and pygmy loris were fed a single meal each morning consisting of LabDiet New World Primate Diet 5040 (PMI Nutrition International, St. Louis, MO), superworms (Zophobas 95

112 morio), and mixed produce including sweet potatoes, apples, carrots, bananas, and endive. Activity Recording We collected activity data using a tri- axial accelerometer, the ActiSleep (ActiGraph, Pensacola, FL). The ActiSleep weighs 18 g and measures 43.2 x 38.1 x mm. The ActiSleep records accelerations ranging from 0.05 to 2.5 G which are digitized and integrated into activity counts over a user- defined epoch. We used a 30- sec sampling epoch, which is slightly more frequent than prior research with small primates (Kantha and Suzuki, 2006b), to ensure adequate sampling of the infrequent, brief bouts of activity characteristic of our study subjects. The study subjects were fitted with a small pet harness (Super Pet, Elk Grove Village, IL) to which the ActiSleep monitor was attached dorsally using heavy- duty Velcro and a fabric strap (Figure 1) (Kantha and Suzuki, 2006b; Sellers and Crompton, 2004). Zookeepers fitted the actigraph harness onto the animals while they were awake using manual restraint. We attached the harness to the animals during the active (dark phase) period to avoid disrupting their sleep- wake cycles. The total weight of the ActiSleep and harness was 31.4 g for the loris and 39.2 g for the potto. This weight represented 5.9% of total body weight for the pygmy loris (532 g) and 2.8% for the potto (1381 g). Although the general consensus among field researchers is that radio collars should represent no more than 5% of animal body weight, Gursky (1998) found that tarsiers (Tarsius spectrum), which are extremely small- bodied (~150 g) tolerated collars up to 7% of 96

113 their body weight without any detectable behavioral deficits. Thus, we considered the harness weights to be within an appropriate range for our animals, which were also continuously monitored for the first day that they wore the harness for signs of discomfort or compromised safety. The animals were also visually inspected to ensure that the harness did not cause abrasions or other irritation. This research was reviewed and approved by the Animal Care and Use Committee at CMZ. Behavioral Data Collection Behavioral data were collected by a live observer while the subjects wore the ActiSleep harness. We compared baseline observations (no harness) to behavioral observations taken while each subject wore the harness. Continuous, focal- animal behavioral data were collected (Altmann, 1974) over 30- min sessions using a hand- held computer and Observer Software (Noldus Information Technology Inc., Leesburg VA). Observations were timed to coincide with the subject s active period and occurred from 1000 to 1700 hrs for the pygmy loris and 2200 to 0800 hrs for the potto; observations were balanced across these time periods. Baseline observations were conducted on the potto over four days in October 2009 and observations with the harness occurred on two subsequent days. Baseline observations on the pygmy loris were conducted over six days in September 2009, and observations with the harness were collected over four days that November. The ethogram (Table 1) was constructed so that behaviors could be grouped into inactive, low activity, and high intensity activities, with the aim of determining 97

114 whether it was possible to differentiate among these classes using the actigraph data (e.g. Muller and Schrader, 2003; Naylor and Kie, 2004). Data Analysis We analyzed data within subjects and compared each subject s behavior while wearing the harness to baseline observations using t- tests for which equal variances were not assumed. For analysis of activity budgets, several behaviors were lumped into the following groups. Sleep and rest were combined into rest. Other consisted of exploration, other (low), solitary play, and other (high) behaviors. Move included move, climb up, climb down, and suspensory behavior. Finally, the category social included the behaviors of social touching, social play, reproductive behavior, and agonistic behavior. Following Naylor and Kie (2004), discriminant function analysis was used to examine the predictive value of actigraph activity counts for identifying behaviors recorded during focal sampling. To do so, instances of behaviors that spanned the entire 30- sec recording epoch were first identified using behavioral data to ensure that only a single activity contributed to each data point. These analyses were conducted on individual behaviors as well as the consolidated behavior classes grouped by intensity in the ethogram (Table 1), and finally on gross differences between activity and inactivity. ActiSleep data were generated in ActiLife v and exported to Microsoft Excel. Statistical analyses were conducted using SPSS 12.0 and p- values were set at

115 Table 1. Ethogram for behavioral data collection for actigraph study Behavior* Operational Definition # Level 0: Inactive Behavior Rest Animal is motionless and may be lying down, standing (upright with all four limbs), rearing (standing on two legs), sitting (body hunched but head erect), or hanging with one or two feet. Animal may be sleeping. Level 1: Low- Intensity Active Behaviors Exploration Sniffing, licking, or manually manipulating objects or stationary enclosure furnishings. Self- Directed Auto- grooming using tooth comb or tongue, scratching the self using grooming claw or nails, facial rubbing (rubbing snout, chin, cheeks, or neck on a substrate or object), or rubbing the head against the arms. Feed Ingesting food, normally by grabbing a food item with one hand and taking it to mouth. Includes drinking from surface or dipping hands in liquid and licking the hand. Social Allogrooming (licking or combing another animal s face or fur) or social Touching exploration (sniffing the body of another animal). Other (Low) Any other behavior which does not result in movement of the animal by a distance of more than one body length. Level 2: High- Intensity Active Behaviors Move Quadrupedal motion in any direction, including climbing and backing up. Social Play Attempted mild bite or attack (partner does not attempt to flee), dangling by feet, wriggling body and wrestling between animals. Reproductive Behavior Agonistic Behavior Other (High) Female hangs while male mounts her while dorsally clasping her sides and making rapid thrusting movements, or attempted mount with no thrusting observed. Attempted or successful bite or attack, lunging with open mouth, aggressive pursuit, or turning away or fleeing in response to received aggression. Any other behavior which results in movement of the animal by a distance of more than one body length. Not Visible Behavior Not Visible The animal or its behavior is not visible. *Behaviors are adapted from (Fitch- Snyder and Schulze, 2001). #All behaviors are defined as states. Classifications of intensity level are based on motor patterns underlying each behavior (e.g Papailiou et al., 2008). Inactive behaviors involve little to no movement, low- intensity behaviors involve head/neck or arm movement only, while high- intensity active behaviors involve whole- body movement. 99

116 RESULTS Animal Harnessing The degree of difficulty in attaching the actigraph harness varied among the individual animals tested. Restraining the slow loris to apply the harness was difficult, and one keeper received a bite wound during the process. Once the harness was in place, the slow loris slipped out of the harness within 30 min, and no further attempts were made to work with this animal due to unrelated health issues. The eleven- yr old pygmy loris wore the actigraph and harness for only about an hour as well. The harness repeatedly slipped out of place on this animal, which also appeared to be agitated by the presence of the harness. While the harness was in place, the loris moved in rapid circles around his enclosure, a behavior that the keeper described as unusual for this individual. No further attempts were made to harness this animal, again, for unrelated health reasons. Finally, the three- yr old pygmy loris was successfully harnessed after two attempts. This animal was harnessed immediately prior to the onset of his light phase. He slept through the light phase with the harness in place, but when the dark activity period began, he immediately removed the harness by using his forelimbs to slide the apparatus over his head. Ultimately, we were only successful in harnessing two of our five potential subjects for an appreciable amount of time, and only these individuals are included in further analyses. The nine- yr old pygmy loris repeatedly wore the harness for extended periods of time, although he only did so after multiple adjustments to the fit were made. In each trial, attaching the harness to this animal did not prove 100

117 difficult (Figure 1a); however, he slipped out of the harness within 24 hours on the first two trials. Figure 2. General activity budget for the potto subject with and without the actigraph harness in place. This individual did not have access to a nest box. Two keepers successfully harnessed the potto using manual restraint immediately prior to the onset of the light phase in his exhibit (Figure 1b). The potto slept through the light phase while wearing the harness and observers did not report any unusual behavior during this time, suggesting that the presence of the harness did not disturb the potto s sleep. Anecdotally, the potto did appear to have 101

118 some difficulty moving around his exhibit when he first became active. However, he quickly adjusted his movements to utilize larger supports, avoiding smaller branches on which the harness more easily became entangled. The potto s female cagemate persistently chewed and pulled on the male s harness, and she ultimately assisted the male in removing the harness by pulling it over his head while they were hanging suspended from a perch. All further data analysis is based on one extended time period wearing the harness for each of our two subjects. The data presented here represent 38 continuous hrs wearing the ActiSleep harness for the potto and 97 hrs for the pygmy loris. We did not notice any behavioral signs of discomfort in the animals during this time period, and physical inspections did not reveal any signs of irritation to the animals skin. Animal Activity Budgets A total of 30 hrs of behavioral data were collected on the potto: 20 hrs (10, 30- min samples) of baseline observations and 10 hrs of observation while the potto wore the ActiSleep. For the pygmy loris, 21 hrs of baseline (no harness) observations were conducted and 18 hrs with the harness, for a total of 39 observation hours. The potto showed a few significant changes in behavior with the ActiSleep harness in place compared to baseline data (Figure 2). While wearing the harness, the potto spent significantly less time feeding (n= 60 observations, t (54) = 2.421, p = 0.019) and not visible (n = 60, t (40) = 2.515, p = 0.016). The potto spent 102

119 significantly more time engaged in other (high and low combined) behaviors while wearing the harness compared to baseline (n = 60, t (53) = , p =.035). This difference was not due to solitary play (which was never observed in this subject) or exploration, which was not observed during the actigraph condition. Although the potto appeared to spend more time engaged in social behavior while wearing the harness compared to baseline, this difference was not statistically significant. However, the apparent increase in social behavior was likely a result of the female s attention to the male s harness. Figure 3. General activity budget for the pygmy loris subject with and without the actigraph harness in place. This individual had access to a nest box and was often not visible in this location. 103

120 In all conditions, the only social behaviors observed in these subjects were social touching and play; reproductive and agonistic behaviors were never observed. The pygmy loris subject s behavior (Figure 3) did not differ as greatly as that of the potto between conditions. Like the potto, the pygmy loris spent significantly more time engaged in other behavior while wearing the harness compared to baseline (n = 78 observations, t (35) = , p = 0.002). Time spent moving (n = 78, t (46) = , p = 0.119) and engaging in self- directed behavior (n = 78, t (45) = , p = 0.083) both increased while the pygmy loris was wearing the ActiSleep harness; these differences approached but did not reach significance. The solitarily- housed loris was never observed to engage in social behavior or solitary play, and only a single bout of exploration occurred. Circadian Actigraph Data The potto s actigraph data demonstrate a robust nocturnal activity pattern (Figure 4). The onset of activity occurred almost immediately when the exhibit lights were extinguished each night at 2200 hrs. Bursts of activity during the light phase were brief and of low intensity. The average activity count reported during the dark phase was (SD 233.2) compared to 10.2 (74.3) for the light phase. Actigraph data for the pygmy loris (Figure 5) covers a longer time span than for the potto. The data are not directly comparable, but the pygmy loris activity does not seem to be as clearly confined to the dark phase as that of the potto. On several occasions, the loris exhibited prolonged bouts of activity during the light phase, although these were of lower intensity than activity bouts recorded during 104

121 the dark phase. Activity counts averaged 3.6 (SD, 30.1) for the loris during the light phase and 21.6 (96.7) during the dark phase. The difference between activity count magnitudes recorded during the light versus the dark phase was therefore much smaller for the loris than the potto. Activity counts were generally higher for the potto than the loris, reflecting a greater intensity, frequency, and duration of movement by the potto. Figure 4. Actigraph data for the potto subject. The graph shows activity counts reported at 30- second intervals for 36 of the 38 hours the potto wore the harness. The last two hours are omitted because the pottos were adjusting, and finally removing, the harness. The dark phase in the exhibit lasted from hrs. 105

122 Figure 5. Actigraph data for the pygmy loris subject. The graph shows activity counts reported at 30- second intervals for 96 of the 97 hours the loris wore the harness. The dark phase in the exhibit lasted from hrs. Correspondence between Actigraph and Behavioral Data The magnitude of activity counts recorded during specific behaviors varied widely within behaviors for both the potto and pygmy loris (Figure 6). For all behaviors, activity counts were higher for the potto than the pygmy loris. We used discriminant function analysis to determine whether activity counts predicted concurrent behaviors. The discriminant function correctly identified rest 90% of the time for the potto and 99.1% for the loris, but action (i.e. anything but rest) was classified correctly only 44.9% and 39.1% of the time for the potto and loris, respectively. Significant results were not obtained using more fine- grained activity types such as feeding or self- directed behavior. 106

123 Figure 6. Mean activity counts associated with behaviors observed in a potto and a pygmy loris. Only behaviors that spanned the entire 30- sec recording epoch of the ActiSleep were included in this analysis. The number of times each behavior was observed is indicated by the number above the bar on the histogram. No social behavior was observed in the pygmy loris, which was housed solitarily. DISCUSSION We attempted to validate actigraphy for automated activity monitoring in the pygmy loris and potto. We assessed potential impacts on the welfare of animals by comparing animal activity budgets with and without the harness in place, and monitoring other physical and behavioral signs of discomfort. We also explored whether the occurrence of specific behaviors could be reliably inferred from activity counts recorded by the actigraph. Our results demonstrate that actigraph data may 107

124 provide useful information about the gross activity patterns of lorisid primates despite their slow- moving locomotor style; however, identifying individual behaviors on the basis of activity counts was not possible for our subjects. Although the harness did not cause the animals any obvious physical discomfort, behavioral changes associated with the presence of the harness warrant further investigation. Our experiences attempting to harness lorises for this study also suggest that animal responses to the procedure are highly individual- specific. By far the greatest challenge in using actigraphy for monitoring our study subjects was finding a way to safely and comfortably but also firmly attach the ActiSleep to the animals. Successful techniques are likely to be species- specific. For ungulates, actigraphs are generally placed on anklets (dairy cows, Muller and Schrader, 2003) or collars (Martiskainen et al., 2009; Naylor and Kie, 2004; Piccione et al., 2008; Van Oort et al., 2004). Collars have also been used to conduct actigraphy studies with several nonhuman primates, including common marmosets, (Callithrix jaccus, Mann et al., 2005), rhesus monkeys (Macaca mulatta, Papailiou et al., 2008), spider monkeys (Munoz- Delgado et al., 2004), and owl monkeys in both captive (Kantha and Suzuki, 2006a) and field (Fernandez- Duque and Erkert, 2006) settings. The harness style- design we employed was used previously for a zoo- housed red- ruffed lemur (Sellers and Crompton, 2004) and lab- housed callitrichids (Kantha and Suzuki, 2006b). This approach is less common, but we were concerned about leaving animals unattended while wearing collars. Although we did not observe any safety issues with our harnesses, our subjects were able to remove them fairly easily using their forelimbs. Further testing may show that collars are more effective at 108

125 keeping actigraphs in place. Other potential methods for future studies include the small sweater used with a zoo- housed koala (Phascolarctos cinereus, Takahashi et al., 2009) or even gluing actigraphs onto animals, as Byrnes et al. (2008) did to investigate gliding dynamics in the Malayan colugo (Galeopterus variegatus). It was also difficult to fit our animals into their harnesses as they struggled under manual restraint, suggesting this procedure may be stressful in nature. Studies of animal handling are few and have mixed implications. With repeated handling, wombats (Lasiorhinus latifrons, Hogan et al., 2011), sheep (Hargreaves and Hutson, 1990), and rabbits (Podberscek et al., 1991) all exhibited reductions in flight distance or other behavioral signs of stress associated with handling. Although animals appeared to habituate behaviorally, adrenocortical responses remained at levels indicative of stress despite repeated handling trials in sheep and wombats (Hargreaves and Hutson, 1990; Hogan et al., 2011). An alternative approach may be to apply the harness while animals are under anesthesia, as did Kantha and Suzuki (2006b) with marmosets; although, this approach presents its own challenges and concerns. However the actigraph is worn, its acceptance will likely vary based on individual animal temperament (Juri Suzuki, personal communication). In each case, the means by which the actigraph is attached to the animal presents welfare trade- offs that vary for individuals and species under investigation. Further research is also needed to examine the duration and extent of behavioral impacts associated with actigraph harnesses. In our study, the potto spent less time feeding with the harness in place compared to baseline, which could be of concern over longer time periods. However, Gursky found that free- ranging 109

126 tarsiers did not show deficits in body weight or prey capture rates after six months of wearing radio collars. Thus, it seems unlikely that provisioned, captive animals would experience nutritional deficits associated with harnesses. In our study, the increased time spent performing self- directed behavior by the pygmy loris may also indicate some discomfort caused by the harness. The animal may habituate or the harness could become increasingly uncomfortable over time. It is common in actigraph research to discard the first 24 hours of data collected with the animal wearing the device to allow for habituation time (Juri Suzuki, personal communication), but whether or not behavioral adjustment occurs after 24 hours or at all should be empirically evaluated before utilizing actigraphy for long- term animal monitoring. The circadian data collected during our actigraph trials illustrates both the promise and difficulty of this method for animal monitoring. Rarely is behavioral data collected on such extended time scales in zoos. It was fascinating to get a glimpse into long- term patterns of our lorisids behavior, as well as activity occurring outside normal staff hours. The potential applications for animal monitoring are numerous. However, for research purposes, it is currently recommended that at least 72 hours of continuous data be collected for analysis of circadian rhythms in human actigraphy (Littner et al., 2003). After multiple trials, we were only able to collect data meeting these criteria on a single pygmy loris. We were also unable to link specific behaviors to reported activity counts (cf. Martiskainen et al., 2009; Naylor and Kie, 2004), a result consistent with the finding that accelerometers are useful mainly as a measure of whole body movement in 110

127 rhesus monkeys (Papailiou et al., 2008). For these reasons, and due to concerns about animal handling, we ultimately decided not to pursue this method in future studies examining behavioral impacts of lighting on lorisid primates. Yet, there are many reasons to consider further this approach in future research. Animal welfare is notoriously difficult to define and measure (Mason and Mendl, 1993; Rushen and Depassille, 1992). A common approach to behaviorally evaluating captive welfare is to calculate the percent of time animals spend performing particular behaviors (the activity budget), including stereotypic or abnormal behaviors indicative of negative welfare, and compare the time budget to that of free- ranging conspecifics (McCann et al., 2007; Novak and Suomi, 1988). Perhaps for highly inactive species, such as lorisids, behavioral rhythmicity and its response to environmental change would be a more informative measure. Information about quotidian patterns of behavior can also inform captive managers of the best time of day to observe their animals, which can be important for monitoring relatively inactive strepsirrhine species (Wright et al., 1989). Actigraphy also enables investigators to examine patterns of animal sleep, an often overlooked behavior that may be extremely informative about animal welfare (Abou- Ismail et al., 2007; Anderson, 1998). Circadian patterns of behavior provide insight into the functioning of the hypothalamic- pituitary- adrenal (HPA) axis (Buckley and Schatzberg, 2005; Kant et al., 1995; Van Reeth et al., 2000), a system which is often the target of welfare investigations in which glucocorticoid secretion is measured as a proxy for stress (Rushen, 1991; Shepherdson et al., 2004). Indeed, changes in sleep architecture may be directly indicative of chronic stress (Van Reeth 111

128 et al., 2000). In humans, sleep patterns measured via actigraphy are a reliable measure of circadian disruption (Ancoli- Israel et al., 2003). In turn, poor sleep quality and circadian disruption are associated with a plethora of maladies that may negatively impact welfare, including depression (Germain and Kupfer, 2008), cardiovascular disease (Malhotra and Loscalzo, 2009), and inflammation (Patel et al., 2009). In conclusion, although the method has its limitations, actigraphy presents many opportunities for better understanding the behavior and welfare of lorisid primates and other captive taxa. Future investigators will need to weigh these potential benefits against the welfare concerns presented by fitting animals with monitoring devices. It may well be worth the effort; actigraphy offers the potential to reveal new connections between behavioral outputs and the physiological systems underlying them in an innovative, holistic manner. 112

129 Chapter Five Methods for Measuring Salivary Melatonin in the Potto, Perodicticus potto, and Pygmy Loris, Nycticebus pygmaeus INTRODUCTION Non- invasive methods for monitoring animal hormones have created a myriad of opportunities for understanding animal health, physiology, and hormone- behavior interactions (Whitten et al., 1998). Non- invasive measures are of particular importance for understanding the physiology of animal stress, which can be complicated by standard blood collection procedures that are inherently stressful (Mormede et al., 2007). Non- invasive measures of fecal or urinary glucocorticoid metabolites (i.e., stress hormones) have been widely applied in zoo and other settings for the assessment of animal welfare (Shepherdson et al., 2004). Many of the same steroid hormones that are measured in feces and urine also can be measured in saliva, along with a variety of amines and peptides (Groschl, 2008). Saliva has become an important matrix from which to extract information about endocrine activity in both research (Kirschbaum and Hellhammer, 1994) and clinical settings (Lac, 2001; Papacosta and Nassis, 2011). Saliva has multiple functions that include the facilitation of swallowing and digestion of food (Groschl, 2008). Cytokines in saliva contribute to oral immune defense and regulate oral tumorogenesis, while steroid hormones in saliva may act as pheromones, playing a role in animal social behavior (Groschl, 2009). According to Groschl (2008), human saliva is comprised of secretions from three paired salivary glands in the oral cavity. Steroid and amine hormones circulating in blood 113

130 are introduced into saliva via passive diffusion through capillary beds, whereas peptide hormones like insulin are usually actively transported into saliva. Consequently, salivary levels of steroid and amine hormones more accurately reflect circulating hormone levels in plasma than peptide hormones. For steroids and amines, salivary measures have the additional advantage of providing a relatively instantaneous measure of circulating hormone levels, in contrast to fecal metabolite measurements that provide data on aggregate hormone levels over a period roughly equal to the animal s gut transit time (Whitten et al., 1998). Thus, saliva can be used to quickly and accurately assess hormonal responses to acute stimuli. Methods for collecting saliva have been investigated in several primate species. Lutz (2000) compared two apparatuses for collecting saliva samples for cortisol analysis from laboratory- housed rhesus macaques (Macaca mulatta). The lick technique consisted of monkeys licking a sheet of gauze covered with wire, while for the pole method, the monkeys chewed on a rope attached to a PVC pole. There were no significant differences found in cortisol values between these two methods, and flavoring the ropes with juice to entice chewing also did not affect cortisol values. Cross et al. (2004) examined salivary cortisol levels in marmosets (Callithrix jacchus), which were trained to lick a cotton bud coated with a film of banana. Although measured cortisol concentrations were depressed by the banana flavoring, the effects were linear and sample concentrations could be easily corrected for the flavor s effects. Using these or similar measures, salivary cortisol has been measured in a variety of nonhuman primate species, including squirrel monkeys (Saimiri sciureus) (Fuchs et al., 1997; Tiefenbacher et al., 2003), rhesus 114

131 macaques (Boyce et al., 1995), hamadryas baboons (Papio hamadryas) (Pearson et al., 2008), western lowland gorillas (Gorilla gorilla gorilla) (Kuhar et al., 2005), and chimpanzees (Pan troglodytes) (Heintz et al., 2011). The most commonly measured amine in saliva is melatonin. Melatonin is the primary hormone that conveys information about time of day to bodily systems (Arendt, 2005) and is therefore frequently utilized in circadian rhythm research (Groschl, 2008; Laakso et al., 1993; Voultsios et al., 1997) and as a biomarker of circadian disruption (Mirick and Davis, 2008). Humans (Kennaway and Voultsios, 1998) and other animals exhibit true circadian rhythms in melatonin production, and circulating levels of melatonin are much higher in the dark than the light phase for all species that have been studied, regardless of activity pattern (Arendt, 2005). Human subjects can be trained easily to collect their own saliva samples for melatonin analysis, facilitating large- scale epidemiological studies on the effects of shift work, travel, sleep disorders, and other conditions associated with altered circadian rhythms (Arendt, 2005; Blask, 2009; Grundy et al., 2009) Exposure to light at night has been widely demonstrated to suppress pineal melatonin production in humans (Hashimoto et al., 1996; Reiter et al., 2007; Shanahan et al., 1997; Stevens and Rea, 2001; Zeitzer et al., 2000) and other animals (Dauchy et al., 1997; Depres- Brummer et al., 1995; Hoban et al., 1990; Reppert et al., 1981). Light toward the blue end of the spectrum also has a greater effect on melatonin suppression than light of longer wavelengths (red light) (Boulos, 1995; Rahman et al., 2008; Schobersberger et al., 2007). In turn, melatonin suppression is associated with a wide variety of maladies, including breast and other endocrine- 115

132 related cancers, cardiovascular disease, major depression, metabolic syndrome, and decreased fertility (Arendt, 2005; Navara and Nelson, 2007; Stevens and Rea, 2001). Melatonin or its metabolites can be measured in most bodily fluids including blood, urine, and saliva (de Almeida et al., 2011). Salivary measures rarely have been utilized with nonhuman animals, although Stark et al. (1997) measured salivary melatonin responses to radio signals in dairy cattle. Surprisingly little research has been conducted on melatonin expression in nonhuman primates. Aujuard et al. (2001) examined the effects of photoperiod on cellular aging in gray mouse lemurs (Microcebus murinus) by measuring urinary sulfatoxymelatonin. Nocturnal light exposure has been documented to suppress melatonin concentrations in squirrel monkey plasma (Hoban et al., 1990) and in the cerebrospinal fluid of rhesus macaques (Reppert et al., 1981). However, to our knowledge salivary melatonin has not been investigated in any nonhuman primate species. Our ultimate goal was to develop a biomarker that would be useful to assess the physiological effects of lighting design for zoo- housed pygmy slow lorises (PSL; Nycticebus pygmaeus) and pottos (Perodicticus potto). Saliva sampling also enabled us to take melatonin measurements at several time points throughout the day, which would be impossible in such small animals using blood sampling. Thus, our specific aims were to create a procedure for saliva collection from these species and condition animals to provide samples, and to develop laboratory assays for quantifying melatonin concentrations in loris and potto saliva. To biologically validate the melatonin assay, we also examined acute suppression of melatonin due 116

133 to nocturnal light exposure as well as 24- hour rhythms of melatonin expression in lorises and pottos. MATERIALS AND METHODS Saliva Collection Saliva samples were collected using a pole apparatus similar to that described by Lutz et al. (2000), which consisted of a 1.5 m (1.27 cm diameter) fixed length of PVC (Figure 1). The lorises and pottos were trained for saliva collection using positive reinforcement techniques. The subjects were habituated to the presence of the pole and the researcher using super worms (Zophobas morio) presented on a hook attached to the pole. Although the potto subjects habituated to the pole in only one or two training sessions, it took many weeks of training to get the pygmy lorises to interact with the pole. Once the animals would regularly take worms off the pole, they were presented with the collection swabs. During sample collection, the animals were given 10 min to chew on the swab. Afterwards, the chewed portion of the swab was immediately cut off the pole using scissors and placed in a saliva collection vial (Salimetrics LLC, State College, PA). Saliva was recovered from the swabs by centrifuging the vials for 15 min at 2500 rpm. 117

134 Figure 1. Collection of saliva from the female potto. Various collection media were tested for this study, including cotton rope and salivettes (SARSTEDT AG & Co, Nümbrecht, Germany). The most successful collection swabs were children s swabs (Salimetrics LLC, State College, PA). The children s swabs are composed of an inert polymer that is durable but also soft. The length of the children s swab also allowed the animals to hold it while they chewed, seemingly a preferred behavior (Figure 1). Although we tried many different flavoring agents including various juices, mashed banana, and mashed super worms the animals responded most favorably to honey and sucrose solutions. The final collection media developed for these experiments consisted of children s swabs quickly dipped in a 1:3.5 dilution of honey in water. The swabs were then dried under forced air for 48 hrs to remove excess liquid. The decision to use honey rather 118

135 than sucrose was based on the results of standard laboratory validations such as parallelism of pooled samples to the standard curve and accuracy/recovery values. Laboratory Analysis of Melatonin Concentrations We tested two different laboratory methods for quantifying melatonin in saliva: an enzyme- immunoassay (EIA) for direct measurement of melatonin in human saliva (RE54041, IBL International, Hamburg, Germany) and a radio- immunoassay (RIA) for direct determination of melatonin in human serum, plasma, or saliva (BA R- 3300, Labor Diagnostika Nord GmbH & Co. KG, Nordhorn, Germany). Initial results using both the EIA and RIA indicated that a component of the saliva sample was interfering with melatonin measurement. Possible sources of interference included the saliva matrix, flavoring agent, or collection medium. To identify possible sources of matrix interference, we compared serially collected saliva samples using cotton rope soaked in diluted honey or sucrose in pygmy lorises, pottos, and human volunteers by RIA. An organic solvent extraction was also tested to eliminate matrix interference in the RIA. Samples were mixed with diethyl ether in a 1:11 dilution, vortexed, centrifuged for 10 minutes at 2500 x g, and immediately frozen at - 80 C. After one hour, the ether was decanted and the supernatant was dried under air overnight. The samples were then reconstituted in distilled water equal to the original sample volume. We also tested several different saliva extraction techniques in an attempt to eliminate matrix interference in EIA measurements (Harumi and Matsushima, 2000). Using saliva from the same loris and potto pools, we compared a liquid- liquid 119

136 extraction technique (methanol- chloroform) (de la Puerta et al., 2007) to a methanol- solid phase extraction (SPE) protocol (Iriti et al., 2006; Kolar and Machackova, 2005; Romsing et al., 2006). For the methanol- chloroform extraction, samples were first mixed with methanol in a 1:11 dilution, shaken on a multi- tube vortexer for 30 minutes, centrifuged at 3000 x g, frozen at - 80 C, decanted, and the supernatant was dried under air. Samples were then reconstituted in phosphate buffered saline (PBS) and mixed with chloroform in a 1:3 dilution. These samples were then frozen and decanted, and the supernatant was dried under air. For the methanol- SPE procedure, samples were extracted using methanol following the methods described above and then re- suspended in a 1:10 dilution of methanol to PBS. Samples were processed through C- 18 Sep- Pak cartridges, alternately eluted and washed using methanol and water, and reconstituted in water prior to analysis. To assess the effectiveness of the extraction technique, parallelism to the standard curve was measured and compared to values obtained using un- extracted, dilute saliva. BIOLOGICAL VALIDATION EXPERIMENTS Experiment One: Acute Suppression of Melatonin by Light Exposure The purpose of this experiment was to demonstrate suppression of melatonin production as a result of acute light exposure during the dark phase. This effect was expected to be dose- responsive, meaning that the degree of suppression would increase with light intensity. We also expected short wavelength (blue) light to have a greater suppressive effect than longer (red) wavelengths. 120

137 Subjects and Housing The subjects for this experiment were one female potto (Jahzira, 4 yrs) and 1.1 pygmy lorises. The adult female (PSL1, Hermione, 5 yrs) was housed in an exhibit measuring 3 x 2.7 x 2.9 m. Her male offspring (PSL2, Harry, 2 yrs) was housed alone in a nearby exhibit measuring 3 x 2.6 x 2.9 m. The potto was housed with a male (18 yrs) who did not regularly provide saliva samples and was therefore excluded from the experiment. The potto exhibit measured 2.9 x 4.9 x 4.3 m. The three exhibits were similarly designed, including gunnite rockwork, natural wood perching, and leafy cover. None of the exhibits contained a nest box during this experiment. All the exhibits were located in a nocturnal wing of the Primate, Cat, and Aquatics house at CMZ. The public floor space in this area was illuminated with dim red light, in addition to some diffuse white light from nearby diurnal exhibits. During the experiment, the animals were housed on a 12:12 Light- Dark (LD) cycle with dark phase onset at 1800 hrs. The exhibits were illuminated by white halogen bulbs during the L phase. No direct artificial light was provided during the D phase, except for whatever small amount bled into exhibits from nearby hallways. To minimize uncontrolled light exposure, the public viewing glass at the front of each exhibit was covered with a wooden board and a canvass tarp from 1700 to 0700 hrs every evening during the study. Light intensities were measured using a SPER Scientific light meter (#840020, Scottsdale AZ, USA). L phase values (lux) for each animal were: for PSL1, for PSL2, and for the potto. D phase intensities in the exhibits 121

138 were less than one lux. During the D phase, animals were acutely exposed to red, blue, and full- spectrum (FS) light from fluorescent bulbs. The fluorescent lights were covered with gel filters (Rosco Laboratories Inc., Stamford CT, USA) to create the red (filters #2001, 4690, 4660, and 4630), blue (#4290, 4260, and 4230), and FS (#3404, 3403, 3402) conditions. Data Collection Data were collected December 2011 through January For this experiment, the subjects were experimentally exposed to 105 min of FS, red, or blue light during the dark phase; and saliva samples were collected for analysis of salivary melatonin. Saliva samples were collected using the pole technique and honey- flavored children s swabs previously described. Sample collection was staggered among the three subjects. First, a baseline saliva sample was collected at two hrs after D phase onset for the PSL subjects (2000 hrs) and three hrs for the potto (2100 hrs). Next, the fluorescent lights were switched on in the exhibit, and saliva samples were then collected after 60 and 90 min of light exposure. The animals were exposed twice to each of eight conditions. Animals were tested under two control conditions: darkness and high intensity FS light (no filter; lux), in addition to dim (0-1 lux) or bright ( lux) FS, red, or blue light. Only a single test was conducted each night, and the two tests for each condition were conducted serially. An interval of 48 hrs was observed between each test to allow animals to re- entrain to their baseline lighting regimen. This interval was chosen based on research showing that in humans experiencing jetlag or other 122

139 circadian disruptions, a 24- hr interval is needed to adjust sleep patterns by 1 to 2 hrs (Kolla and Auger, 2011). The order in which the lighting conditions were tested was randomized among the exhibits. Samples were immediately refrigerated following collection. The swabs were centrifuged the same day for 15 min at 2500 rpm to collect saliva, and the samples were frozen at - 18 C. Melatonin concentrations were measured in un- extracted saliva in a 10/90 dilution using the EIA methods previously described. Experiment Two: 24- Hour Melatonin Rhythms For this experiment, saliva samples were collected regularly throughout the day to look for evidence of rhythmicity in melatonin expression. Melatonin levels were expected to be significantly higher during the D phase compared to the L phase. This study was conducted over a two- week period in April Saliva samples were collected from subjects using the pole method described for experiment one. Samples were collected every three hours, at: 1000, 1300, 1600, 1900, 2200, 0100, 0400, and 0700 hrs. Samples were stored and analyzed using the same methods described for experiment one, except that sample swabs were immediately frozen after collection at - 18 C. The swabs were thawed and centrifuged to collect saliva immediately prior to EIA analysis. The subjects for this experiment included the same animals from experiment one with the addition of an unrelated adult male pygmy loris (PSL3, Tai, 6 yrs). PSL3 was housed alone in an exhibit next to PSL1 measuring 2.9 x 2.6 x 2.9 m. Exhibit 123

140 conditions for PSL3 were comparable to those described for the other exhibits in experiment one, and this animal did not have access to a nest box. PSL1 and 2 both had access to a nest box in their exhibits during the experiment. During the experiment, the animals were housed on a 12:12 reversed LD cycle with dark phase onset at 1000 hrs. The exhibits were illuminated by white halogen bulbs during the L phase and red- filtered fluorescent lights (#2001, Rosco Laboratories Inc.) during the D phase. Intensities (lux) in the exhibits were as follows: 23.5 L: 0.7 D for the potto; L: 1.1 D for PSL1; L: 1.9 D for PSL2; and L: 0.9 D for PSL3. Data Analysis For both experiments, data were analyzed using Microsoft Excel and SPSS Circadian variables were calculated using Cosinor software available at ( Refinetti 2013; Refinetti et al., 2007). Period length was defined as length of time to complete one oscillation of a rhythmic variable (Halle and Weinert, 2000). RESULTS Comparison of Melatonin EIA and RIA Serial dilutions indicated greater levels of matrix interference using sucrose than honey in potto and pygmy loris samples. Human melatonin values also more closely approached unflavored values using honey- flavored ropes than those soaked in sucrose. Although it was not possible to determine from these experiments 124

141 whether the source of interference was the cotton in the rope or the sugar flavoring, we ultimately decided to use synthetic collection swabs (Salimetrics children s swabs) to eliminate cotton as a potential inhibiting factor. Pooled samples extracted with ether demonstrated the most consistent control values using the RIA. Serially diluted extracts displayed parallelism with the standard curve (t=2.103, p=0.054 for pygmy loris; t=2.046, p=0.063 for potto). Pooled samples were used to measure recovery of high (20 pg/ml) and low (6 pg/ml) melatonin concentrations. Recoveries for high concentrations were 109.2% for PSL and 95.56% for potto, and recoveries for low concentrations were 100.8% for PSL and % for potto. Although the RIA met parallelism and recovery validations, we recorded unexpectedly high melatonin levels on a control (water- soaked) swab. As we had generally been measuring values toward the low end of the standard curve using this assay, we ultimately determined that the assay lacked the sensitivity to accurately measure melatonin concentrations in our subjects. To assess the effectiveness of the different extraction techniques using the EIA, recoveries of hormone from pooled samples spiked with 2.5 pg/ml or 27.5 pg/ml of standards were compared to values generated from un- extracted pools analyzed at a 1:10 dilution. For pottos, recoveries for pools extracted using methanol- chloroform were 130.0% (low) and 66.9% (high). SPE recoveries for pottos were 160.3% (low) and 60.71% (high), and recoveries from un- extracted pools were 68.0% (low) and 100.1% (high). For pygmy loris, methanol- chloroform recoveries were 179.1% (low) and 72.8% (high), SPE recoveries were 141.5% (low) and 60.9% (high), and recoveries from un- extracted saliva were 109.0% (low) and 125

142 103.5% (high). Parallelism to the standard curve was only achieved using un- extracted samples, not with either extraction technique. Ultimately, we determined that analysis of melatonin concentrations in these species was most successful using the EIA with samples analyzed at a 1:10 dilution to eliminate matrix interference. Serially diluted samples displayed parallelism to the standard curve (t= , p = for pygmy loris; t= , p=0.303 for potto). The EIA detected no measureable melatonin in a control sample prepared by soaking a honey- flavored swab in a volume of distilled water equal to that of a typical saliva sample. We also directly assayed the diluted honey solution used to flavor swabs and found no detectable melatonin content using the EIA. Given that the EIA also demonstrated greater cross- reactivity with both species than the RIA, we determined EIA to be the preferred approach to melatonin determination in the potto and pygmy loris. Experiment One: Acute Suppression of Melatonin by Light Exposure Melatonin measurements taken on the potto were comparable under darkness and red light, and higher melatonin values were measured in these conditions compared to the blue and bright control conditions (Figure 2a). Although both intensities of blue light appeared to affect the potto in the same way, higher levels of melatonin were measured under dim red light than following exposure to bright red light. Surprisingly, the highest mean melatonin levels were measured following dim red light exposure, not in the dark (0 lux) control condition. 126

143 Figures 2 a- c. Melatonin concentrations measured following nocturnal exposure to test lights in a potto (a), and pygmy lorises PSL1 (b) and PSL2 (c). The subjects were exposed to 105 minutes of full- spectrum, blue, or red light of varying intensities. Light wavelength is indicated by bar color and intensity is given on the y axis. Light exposure began three hours after dark phase onset for the potto and two hours for each PSL. Samples collected after 60 and 90 minutes of light exposure on two test presentations (three for the 0.0 lux measurements) are combined for each light condition. Values above standard error bars indicate sample number. 127

144 Melatonin levels in the potto were also higher in darkness and in dim FS light compared to both medium and high intensity FS light. Of all three subjects, only the potto demonstrated an overall significant effect of lighting condition in a repeated measures ANOVA (F7,27 = 3.184, P=0.031), although the female loris approached significance (F2,27=2.218, P=.097). Paired t- tests approached significance for the potto with higher melatonin under dim red light compared both to dim blue (t2= , P=0.142) and bright blue (t2= , P=0.076) light. Visual inspection of results for the pygmy lorises suggests opposing trends. The male pygmy loris (Figure 2c) had higher melatonin levels under dim red light than all other conditions, although overall intensity differences were negligible. The female pygmy loris (Figure 2b) consistently demonstrated patterns antithetical to our predictions: her melatonin levels were highest in the high intensity control condition, followed by the bright light conditions, and lower in the dim and dark conditions. Experiment Two: 24- Hour Melatonin Rhythms For this study, between one and four useable saliva samples were collected per subject at each time point. Mean melatonin concentrations varied significantly with lighting phase for the potto ( pg/ml + SE for the D phase (N=16) vs for the L phase (n=13), t29 = , p< 0.001). However, for all three pygmy lorises, L and D phase melatonin values were not significantly different: D (N=17) vs L phase (N=6) for PSL1; D 128

145 (N=16) vs L (N=11) for PSL2; and D phase (N= 16) vs L (N=16) for PSL3. For PSL1 and 2, melatonin measurements did not appear to vary depending on whether the subject was in the nest box at the time of sample collection. Melatonin values varied among subjects and were often quite variable at a given time point within a single subject (Figure 2). Using a repeated measures ANOVA, only the potto demonstrated a significant effect of time of day on melatonin (F2,27= 7.875, p =0.001). The cosinor model identified a period of 22.8 hrs for melatonin expression the potto (F2,27 = 9.169, p= 0.001), with an amplitude of 15.8 pg/ml. A significant period for melatonin expression was not identified in any of the PSL subjects using the cosinor model. 129

146 Figure hour patterns of salivary melatonin expression in a potto and three pygmy lorises (PSL). Between one and four samples were collected per subject at each time point, for a total of N= 29 for the potto, N = 23 for PSL1, N = 27 for PSL2, and N = 32 for PSL3. Animals were housed on a 12:12 LD cycle with dark phase onset at 1000 hrs. DISCUSSION We aimed to develop methods for using the hormone melatonin as a biomarker for the health effects of lighting design on nocturnal strepsirrhines exhibited in zoos. We conditioned several pygmy lorises and a potto for voluntary saliva collection and tested the subjects receptiveness to various flavoring and collection media. We compared laboratory results between a commercial EIA and 130

147 RIA and ultimately determined the best results were obtained analyzing honey- flavored collection swabs by EIA. Chemical validations of the EIA were successful for both the potto and pygmy loris, but biological validation experiments examining acute suppression of melatonin by nocturnal light exposure and 24- hour melatonin rhythms produced unexpected results in the pygmy loris. Ultimately, we concluded that the assay is effective for salivary melatonin analysis in the potto but further testing is required for the pygmy loris. The nature of the saliva matrix can create challenges for interpreting salivary hormone levels, even in the absence of added flavoring. The composition of saliva and the concentration of hormones within it are dependent upon a variety of factors. Saliva composition can be affected by disease states such as diabetes mellitus, kidney dysfunction, and epilepsy, among others (Aps and Martens, 2005). Saliva is made up of secretions from several glands, which may vary in their contributions to mixed saliva at a given time, and gland- specific saliva cannot be easily harvested outside of medical settings (Groschl, 2008). Saliva secretion is largely moderated by the autonomic nervous system, and saliva composition can depend on the specific autonomic receptors that are activated (Papacosta and Nassis, 2011). Furthermore, saliva flow rate exhibits regular circadian variation and is also affected by the type of stimulation used to collect samples (Aps and Martens, 2005). Steroid hormone concentrations in saliva are thought to be largely unaffected by saliva flow rate because they are passively diffused into saliva (Papacosta and Nassis, 2011); however, whether this same argument applies to amines is unclear. 131

148 In humans, melatonin concentrations measured in samples collected by gentle chewing on parafilm were higher compared to samples collected after high saliva flow rate was induced using citric acid (Voultsios et al., 1997). Although we did not directly control for salivary flow rate, during the course of our experiments we limited sample collection time to ten minutes per sample. Future investigators may want to account for saliva dilution by incorporating sample volume into statistical comparisons. Another possible control would be to use a measure of the total protein concentration of saliva as an indicator of saliva dilution, much as creatinine concentrations are used to assess urine dilution in endocrine analyses (Lac, 2001). Non- specific interference may have also varied between the two assays we tested. Andersson et al. (2000) compared two commercial RIAs for measurement of melatonin in porcine plasma, both of which passed standard laboratory validations for parallelism and accuracy/recovery. However, they found only one of the RIAs detected expected low/absent levels of melatonin during the light phase, while the other assay showed comparable melatonin values at midday and midnight. These investigators speculate that non- specific binding of plasma proteins may have influenced results; however, they also found that extraction of samples prior to analysis did not eliminate this effect (Andersson et al., 2000). It may be possible that other substances in the saliva samples (in addition to the flavoring) interfered with melatonin measurement to different extents in the two assays we tested. We also found better accuracy/recovery values with potto rather than loris samples, and it is tempting to speculate that lorises, as the only known venomous primate (Kalimullah 132

149 et al., 2008; Krane et al., 2003), may have substances in their saliva that cause greater assay interference than pottos. The results of our lighting experiments demonstrate the importance of properly validating hormone measures both analytically and physiologically (Touma and Palme, 2005). Results with the potto largely met our expectations, as this subject demonstrated an identifiable rhythm in melatonin concentrations, a clear difference between mean melatonin concentrations between the dark and light phase, and differential suppression of melatonin by acute exposure to lights at different intensities and wavelengths. In contrast, the pygmy loris subjects did not exhibit a melatonin rhythm or suppression resulting from light exposure. There are several explanations for why the biological validation experiments were ultimately unsuccessful in the pygmy loris. In mammals, a relatively narrow band of wavelengths cause melatonin suppression, and the specific wavelengths responsible vary among species (Brainard and Hanifin, 2002). It may be that case that the experimental lighting we designed did not target the action spectra responsible for melatonin suppression in pygmy lorises. However, if these action spectra were known, it might be possible to design custom lighting that filters out only the wavelengths that suppress melatonin, without shifting the overall appearance of the light color (Rahman et al., 2008; Schobersberger et al., 2007). Another possible explanation is that the lorises in our study were experiencing chronic melatonin suppression. When we tested 24- hour melatonin rhythms, the animals were housed on a reversed light cycle, meaning they were exposed to some intensity of light all 24 hours of the day. Housing rats under 133

150 constant dim red light causes circadian rhythms of melatonin, activity, and body temperature to become desynchronized (Aguzzi et al., 2006). However, the potto, whose exhibit had the smallest difference in intensity between the light and dark phase of any exhibit, showed clear differences in melatonin concentrations based on time of day. There are some species that do not show the expected nocturnal rise in melatonin production or for whom this may vary on a seasonal basis (Reiter et al., 1987). Whether these trends represent true species differences will remain unclear until additional pottos can be tested, as well as lorises and pottos under different lighting regimens. The results of our efforts to validate methods for measuring salivary melatonin in the pygmy loris and potto show both the potential and challenges of applying salivary hormone analysis to nonhuman animal experiments. An alternative validation to perform in future experiments may be to orally administer melatonin and then test its subsequent concentrations in saliva (Kovacs et al., 2000). In spite of such challenges, salivary hormone analysis shows great promise for understanding a variety of health and physiological issues in captive animals. 134

151 Chapter Six A Case Study Comparing Hormonal and Behavioral Responses to Red and Blue Exhibit Lighting in the Aye- Aye, Daubentonia madagascariensis INTRODUCTION Nocturnal mammals are often housed on reversed light cycles in zoos so that visitors can observe active behavior. Zoo exhibits historically utilized red light based on the reasoning that this would appear darker to nocturnal species with rod- dominated retinas (Conway, 1969; Davis, 1961). However, a recent survey of facilities housing lorisid primates in North America revealed that institutions exhibit animals under red and blue dark phase lighting equally (Fuller et al., 2013). Currently, little empirical evidence informs the debate over light color for nocturnal exhibits. Light signals play an important role in the mammalian circadian system, and artificial light cues may have dramatic effects on behavior and physiology (Erkert, 1989). Several studies have demonstrated that captive nocturnal strepsirrhines are more active at lower light intensities during the dark phase (Frederick and Fernandes, 1994; Randolph, 1971; Trent et al., 1977). However, there is currently no research examining the effects of wavelength on nocturnal primate behavior. Exposure to light at night disrupts the body s timekeeping system by suppressing production of the hormone melatonin (Hoban et al., 1990; Zeitzer et al., 2000), altering daily rhythms of cortisol and other hormones as well as activity patterns (Depres- Brummer et al., 1995; Mirick and Davis, 2008; Reiter, 1991). In human studies, chronic melatonin suppression is linked to maladies including 135

152 cancer, cardiovascular disease, and immunosuppression (Navara and Nelson, 2007; Reiter et al., 2007). Shorter wavelength (blue) light has a greater suppressive effect on melatonin than longer wavelengths (red) in humans (Brainard and Hanifin, 2002; Schobersberger et al., 2007) and other species (Rahman et al., 2008; Walsh et al., 2013). Color- specific melatonin suppression thus has important husbandry implications for zoos. The aye- aye (Daubentonia madagascariensis) is a nocturnal strepsirrhine endemic to Madagascar (Groves, 2001). The goal of this study was to systematically investigate the effects of nocturnal light color in a single aye- aye subject. We hypothesized that the aye- aye would demonstrate signs of circadian disruption under blue but not red light. MATERIALS AND METHODS The subject was an 18- yr old, captive- born, female aye- aye housed at Cleveland Metroparks Zoo (CMZ) in Cleveland, OH. The enclosure measured 3.6 x 4.0 x 4.4 m with gunnite rockwork and natural log perching. The exhibit contained a cardboard box or bag for nesting, nesting materials, puzzle feeders, and other enrichment items. A smaller off- exhibit room was accessible most days and every evening after 1700 hrs. Data were collected October 2012 February 2013, during three study conditions: (1) baseline red light (5 weeks); (2) experimental blue light (5 weeks); and (3) a second red light baseline (3 weeks). In humans experiencing jetlag or other circadian disruptions, a 24- hr interval is needed to adjust sleep patterns by 1 136

153 to 2 hrs (Kolla and Auger, 2011). To allow for entrainment, there was a two- week break in data collection between the first two conditions. This interval was extended to four weeks between the last two conditions to accommodate another experiment. All aspects of lighting were constant across conditions except for dark phase light color. The aye- aye was housed on a light- dark cycle matched to Madagascar photoperiod, with dark phase onset at 1000 hrs and light phase onset between hrs. The exhibit was illuminated with full- spectrum compact fluorescent lights, half of which were covered with red (#2001) or blue (#4290) gel filters (Rosco Laboratories Inc., Stamford CT, USA) to simulate the darkness. Light intensities were measured using a SPER Scientific light meter (Scottsdale AZ, USA) and were held constant at 4.5 lux during the dark phase (2.7 lux off- exhibit) and 234 lux during the light phase (200 lux off- exhibit). For conditions one and two, behavior was observed all 24 hours of the day, while only dark phase observations were made during condition three. A continuous behavior sampling protocol was employed in ten- minute sessions using a simple ethogram (Table 1) and The Observer 5.0 (Noldus Information Technology, Wageningen, The Netherlands). HOBO data loggers (Onset Computer Corp., Cape Cod MA, USA) were placed in exhibits to monitor light intensity, temperature, and relative humidity. Supplemental information on behavior and health was obtained from daily keeper reports and medical records. 137

154 Table 1. Ethogram for behavioral data collection on the aye- aye. Behavior Social Move Feed Self- Directed Object Examination Rest Other Not Visible (Nest box) Not Visible (Out) Operational Definition Allogrooming, social exploration, social play, reproductive behaviors, or agonistic behavior. Motion in any direction, including climbing and backing up. Ingesting food, normally by grabbing a food item with one hand and taking it to mouth. Includes drinking from surface or dipping hands in liquid and licking the hand. Auto- grooming using tooth comb or tongue, scratching the self using grooming claw or nails, facial rubbing (rubbing snout, chin, cheeks, or neck on a substrate or object), or rubbing the head against the arms. Animal is physically manipulating an object using hands or teeth or is actively sniffing an object. Animal is motionless and is lying down, sitting, or in another sleeping posture. The animal s eyes may be open or closed. The animal may be actively scanning the environment. The animal is exhibiting any other behavior than those defined above. The animal is not visible for an extended period of time (more than one minute) and inspection of the exhibit suggests it is resting in the nest box or another hiding location. The animal or its behavior is not visible because it has briefly gone out of sight but has NOT retreated to the nest box. Saliva samples were collected for hormone analysis at 1000, 1300, 1600, 1900, and 0100 hrs using voluntarily chewed swabs (Salimetrics, State College PA, USA) flavored with diluted honey and dried. Swabs were centrifuged at 2500 rpm for 15- min. Saliva was aliquoted and frozen at - 18 C until analysis. Salivary melatonin concentrations were measured with a commercial enzyme immunoassay (EIA) (IBL International Corp., Toronto ON, Canada). Salivary cortisol was quantified by EIA using methods by Munro and Lasley (1988), and an anti- cortisol antiserum (R4866) and cortisol- horseradish peroxidase (HRP) ligand obtained from Coralie Munro (University of California, Davis, CA). The polyclonal antiserum cross- reacts with cortisol (100%), prednisolone (9.9%), prednisone 138

155 (6.3%), cortisone (5%) and < 1% with androstenedione, androsterone, corticosterone, desoxycorticosterone, 11- desoxycortisol, 21- desoxycortisone, and testosterone (Munro and Lasley, 1988). For both assays, serially diluted extracts were parallel to the standard curve (tmelatonin = , p = 0.727; tcortisol = , p = 0.190). Recoveries of hormone from saliva spiked with 2.4 pg/ml or 15.0 pg/ml of melatonin standard were 79% and 83%, respectively. Recoveries from saliva spiked with 0.4 ng/ml or 8.0 ng/ml of cortisol standard were 98.0% and 109.0%. Inter- and intra- assay coefficients of variation were less than 17% for melatonin and 15% for cortisol, and all samples were analyzed in duplicate. Data were analyzed using Microsoft Excel. Circadian variables were calculated using Cosinor software available at ( Refinetti 2013; Refinetti et al., 2007). Period length was defined as length of time to complete one oscillation of a rhythmic variable (Halle and Weinert, 2000). RESULTS Condition two ended prematurely due to delayed waking time (Figure 1) and other unusual behaviors (squatting, lethargy, lack of interaction with enrichment) beginning 19 days after entering the blue light condition. A veterinary exam was conducted at day 23 of blue light. Blood, fecal, and urine analyses were in normal ranges, but ultrasound revealed a fluid- filled cyst adjacent to the bladder. Two cracked premolars were also extracted, resulting in a ten- day course of Clamavox. The aye- aye returned to the exhibit but continued to demonstrate abnormal behavior, prompting the premature change of the lights back to red. The aye- aye s 139

156 waking time grew gradually earlier, returning to 1000 hrs after ten days under red light. By this point caretakers described her behavior and appetite as back to normal. Figure 1. Daily time of emergence from the nest box by the aye- aye subject based on keeper reports. Dark phase lights in the exhibit were changed from blue to red at 0 days after 31 days of blue light. Dark phase onset occurred at 1000 hrs, which was considered the normal onset of activity for this animal. Open circles represent days that keepers intentionally awoke the aye- aye; on other days she emerged spontaneously from the nest. The triangle indicates the date the aye- aye was anesthetized for a veterinary exam. 140

157 Ultimately, 179 observations (29.8 hrs) were conducted during condition one, in which 10 behavior samples/hr were taken in the dark phase and 5/hr in the light phase. During condition two, 91 observations were conducted (15.2 hrs): 7-9 observations/hr from hrs, and 2-4 observations/hr from and hrs. During condition three 121 observations (20.2 hrs) were conducted, 10/hr of the light phase. We suspended observations for five days following the veterinary exam. Figure 2. Time spent performing active behaviors (move, feed, self- directed, or object examination) by the aye- aye subject during the baseline red and experimental blue lighting conditions. Data for all 24 hours was only available during the first baseline. Dark phase in the exhibit occurred from hrs. Sample size refers to the number of ten- minute behavior observations. 141

158 The aye- aye was much less active during the blue condition than either red condition (Figure 2). Inactivity consisted of time in the nest box or resting, and all other behaviors were considered active. On multiple days during condition two, the aye- aye moved around the exhibit during the light phase, which was never observed during red conditions. The subject spent less time moving and examining objects, and slightly less time feeding, under blue light compared to red (Figure 3). There were no fluctuations in temperature or relative humidity corresponding to behavioral changes under blue light. Figure 3. Dark phase activity budget ( hrs) for the aye- aye subject during the baseline red and experimental blue lighting conditions. Sample size refers to the number of ten- minute behavior observations. 142

159 Hormone concentrations were similar under red conditions but differed appreciably under blue light. Dark phase salivary melatonin concentrations were lower under blue light than red (Figure 4). The mean dark phase melatonin concentration under blue light (N=12, (SE) pg/ml) was more comparable to light phase melatonin values (N=5, ) than to dark phase values measured under red light (N= 75, ). Period length for melatonin expression was 24.0 hr and 24.8 hr under red conditions one and three, respectively; compared to 25.1 hr for blue condition two. Salivary cortisol concentrations varied more with time of day under red than blue light and were higher overall under red (Figure 5). Period lengths for cortisol expression were 21.8 hr and 24.1 hr under red light and 22.6 hr under blue light. The amplitude of cortisol rhythms was lower under blue light (13.00 ng/ml) compared to both red conditions (27.50 ng/ml and ng/ml). 143

160 Figure 4. Dark phase salivary melatonin concentrations in the aye- aye during the baseline red and experimental blue lighting conditions. Sample sizes are indicated above bars. Samples were collected at T0 (1000 hrs), T3 (1300 hrs), T6 (1600 hrs), and T9 (1900 hrs). For condition one, total samples = 10 each at T0, T3, T6, and T9. For condition two, total samples = 2 at T0, 5 at T3, 3 at T6, and 2 at T9. For condition three, total samples = 10 at T0, T3, and T6 and 5 at T9. 144