Behavioral Research at the San Diego Zoo

Dr. Ron Swaisgood, C.R.E.S.

The Center for Reproduction of Endangered Species of the Zoological Society of San Diego initiated a behavioral research program for the giant panda in the mid-1990's. Building upon earlier work by Devra Kleiman and colleagues at the National Zoological Park, the aim of our research is to (1) provide baseline characterization of behavioral patterns in the panda; (2) test various hypotheses related to the regulation of social and mating behavior; (3) promote well-being in captivity; (4) facilitate captive breeding; and (5) ultimately achieve a better understanding of the panda's behavioral biology that will also assist in conservation efforts in the wild. Given there readily apparent public appeal, it is surprising how little is known about the giant panda's behavior and biology, although recent efforts on the part of several western and Chinese institutions have began to make significant advances. Although to date most of our research has focused on captive animals, we have framed our research questions in functional terms, and have sought guidance from the research of George Schaller, Pan Wenshi and others studying giant pandas in nature. Of course, due to the elusive nature of the panda in the wild, many of the questions we have addressed would be difficult or impossible to research with wild-living pandas. We view efforts in both the field and captivity to be essential and complimentary aspects of any program to remove some of the mysteries of giant panda biology. Such mystery may promote the mystic of the species, but they seriously hinder conservation efforts.

The desirability of working together with other institutions can not be overstated, as only a sustained collaborative effort provides hope that these rather lofty goals will be realized. We have entered into numerous collaborative partnerships with a number of institutions around the world, including the Chengdu Giant Panda Breeding Base and Chengdu Zoo, Beijing Zoo, Ueno Zoo, Adventure World, Mexico City Zoo, National Zoological Park, and Zoo Atlanta. Our most important partnership is with the Wolong Breeding Center in Sichuan, China, where we spend many months each year working together with the research, animal care, keeper and veterinary staff. The information outlined below is based predominantly upon these collaborative efforts at Wolong, as well as intensive studies of the pandas residing at the San Diego Zoo. Wolong, which currently houses more than 40 pandas, provides a unique opportunity conduct experimental investigations on an adequate number of individuals to allow some generalizability to the population as a whole.

Many individuals have contributed to this research program. In San Diego panda team leader Donald Lindburg and research assistants Suzanne Hall, Valerie Hare, Edy MacDonald, Megan Owen and Erin Tepper have made major contributions, as has Angela White, Laura McGeehan of the Endocrinology Division, and Allison Alberts of the Ecology Division. In Wolong director Zhang Hemin, behaviorist Zhou Xiaoping, and assistant director and animal care manager Zhang Guiquan have been our primary collaborators. Other contributors from Wolong include Hongyan Han, Daming Hu, Dasheng Li, Pengyan Wang, Rongping Wei, and Lifeng Wu. In a study of the wild giant pandas we collaborated with Wenshi Pan, Xiaojian Zhu, and Dajun Wang of Peking University.

Chemical Communication

Giant pandas possess a specialized anogenital gland that secretes a waxy substance used in communication with conspecifics. Urine is also employed as a chemical signal, and female pandas urine mark much more frequently to signal incipient estrus. In the wild pandas communicate with one another at communal scent mark stations, often a cluster of trees on a ridge top where neighboring pandas come to leave scent marks and investigate the marks left by other pandas. Such a system might be usefully compared to a community bulletin board, where a person posts a message, accompanied by their name and date the message was left. Similarly, pandas visiting scent mark stations should include analogous components in the chemical signal deposited on trees. First, the mark must contain a "message" such as the sex, reproductive status, or competitive ability of the signaler. Second, the mark should convey the identity of the signaler via an individually distinctive chemical signature. Third, the signal may change over time to afford information regarding the timing of deposition. A visitor to such a scent station should be able to assess and integrate these sources of information to guide an appropriate response. For example, a male investigating a scent might determine that the signaler was a female in estrus, recognize her as an individual that inhabits a range just south of the station, but decide not to pursue her because the mark is a week old and she will have already ovulated. We devised a series of experiments to test if pandas can communicate these kinds of information with their scent marks and urine. Below I outline some of the findings from these studies.

Scent and reproductive strategies. In one of our first studies, we systematically exposed males, estrous and non-estrous females to each other s odors by swapping them into each other's pens. The response was dramatic: both scent marking activity and olfactory investigation increased more than tenfold, suggesting that these odors are very important to pandas. The most exciting result from these studies is the relationship between these odors and reproductive behavior. For example, males strongly preferred female odors over male odors, investigating, licking, scent marking and vocalizing more in response to female than male odors. This demonstrates that they can indeed tell the difference between male and female odors. Male odors encouraged aggressive behaviors, such as foot-scraping, but female odors caused a significant increase in bleating by males. Because bleats are sexually motivated vocalizations, this study suggests that female odors promote sexual motivation in males. Males also bleated more in response to estrous than non-estrous female odors, suggesting that males can assess when a female is in estrus by smelling her scent, and become more sexually excited when they smell estrous female odors.

In a more controlled study in which female urine was presented to pandas on wooden blocks, we found that males but not females displayed a significant investigatory preference for estrous over nonestrous female urine, as indicated by sniffing and flehmen. These findings indicate that female urine contains chemical cues permitting discrimination of estrous status, and probably serves a reproductive advertisement function. In addition, the prevalence of flehmen and licking by males during chemosensory investigation of female urine suggests that males may rely on vomeronasal organ (VNO), a secondary chemosensory apparatus located in the roof of the mouth, to assess female reproductive condition. Efficient use of the VNO requires that males be able to contact the scent directly, not just smell it through the cage bars. If this hypothesis is correct, it would behoove managers at captive breeding facilities to allow pandas direct access to one another's scent prior to mating introductions. Indeed, Wolong staff utilizes knowledge of chemical communication in their breeding management by giving pandas ample opportunities to investigate one another's odors prior to attempting a mating introduction. This strategy may be a major factor explaining the remarkable success rate for natural mating in Wolong, where most females mate naturally every year.

Scent and assessment of competitive ability. In another series of investigations we examined the role that chemical cues play in assessment of reproductive maturity and competitive ability (dominance, territorial status). Escalated fights in pandas can have severe consequences, and avoidance appears to be the primary mechanism for preventing them. Undoubtedly, confrontations with adults or high status individuals carry higher risks than those with subadults or low status individuals. Therefore, benefits should accrue to individuals that possess the ability to assess maturity and fighting ability on the basis of signals, enabling them to avoid risky encounters with adult high status residents. Assessment of chemical signals might also play an important role in governing reproductive behavior. With several female ranges overlapping a given male's range, it is important that the male identify which females should be monitored prior to the onset of the mating season, allowing him to concentrate his efforts on reproductively mature females. Similarly, females might use chemical cues to identify potential mates and assess male quality as it relates to age, dominance, and territory ownership.

Our findings are consistent with these hypotheses. In the first study, 25 pandas of all age-sex categories were presented with scent stimuli (male urine, male scent mark, female urine) collected from adults and subadults. The results clearly show that pandas from all age-sex categories prefer to investigate adult more than subadult conspecific odors. The fact that males and females of all ages were capable of and motivated to perform this discrimination suggests that this represents an important olfactory assessment task fundamental to many giant panda social functions. Adult/subadult status appears important both between and within the sexes, suggesting that it plays a role in governing both competitive and sexual interactions.

A second factor that may be correlated with risk and/or mate quality is competitive ability or status. Competitive ability could be conveyed through inherent chemical properties of the scent and/or through the pattern or frequency of scent marking, as well as deposition posture. For example, the height of a scent mark may afford reliable information regarding the signaler's body size, a known determinant of competitive ability. Clearly, larger animals can deposit scents higher than smaller animals. Once assessors begin using scent deposition height as a cue of competitive ability, selection should favor signalers that exaggerate this cue by adopting postures that allow higher placement of the scent, subject to physical constraints of body size. Since such a posture may be interpreted as a threat or claim to dominance, selection may subsequently favor individuals that employ this strategy selectively, for example, only when they are a territory owner or dominant animal.

It is well known that pandas use four different postures to deposit scents at varying heights, but the functions of these postures remain unknown. Reverse and leg cock postures, in which the panda backs into a vertical surface to deposit scent, have been seen in both males and females. Squats, where scent is deposited on the ground, are seen in all individuals, but mostly in subadults and females. Handstands have only been observed in adult males and usually only as a urine mark. Advertisement of competitive status via selection to exaggerate apparent body size may offer the most plausible explanation for the efforts that males go to when depositing urine in this handstand position. Indeed, males are less likely to use the handstand position when "trespassing" on another panda's home area, perhaps to avoid assertion of dominance in an area that they do not normally occupy.

In the second study we examined the effect of the height of the chemical signal on behavioral responsiveness. Adult male scent mark was placed at 0 m and 0.5 m to mimic squat and reverse/legcock postures frequently employed by adult males. Adult male urine was placed at 0 m and 1 m, reflecting the common placement of urine by males in either the squat or handstand position. Females usually scent mark in either the squat or reverse positions, and thus we placed female scent mark at 0 m and 0.5 m. Females rarely urinate in any posture other than the squat posture and therefore female urine was not used in this experiment.

The height of scent stimuli had a pronounced and highly significant effect on several measures of behavioral responsiveness. For example, males and females of all ages spent more time investigating scent stimuli placed high than those placed low, regardless of whether the scent was male scent mark, male urine, or female scent mark. Thus, pandas may perceive conspecific odors placed high on vertical surfaces as indicating the presence of a more "important" animal, and therefore invest more time in memorizing the scent in order to recognize it again in the future. As hypothesized earlier, one reason for this perceived importance might be the relationship between height of scent mark deposition and competitive ability. Also consistent with this hypothesis is the finding that pandas showed a significant tendency to subsequently avoid the area where male urine was deposited at a height mimicking the handstand, presumably governed by the perceiver's attempts to avoid confrontation with a high status male. This apparent fear-inducing effect was most pronounced in subadult males, who are most at risk for injurious aggression from dominant adult males. Thus, the conclusion most readily drawn from these results is that scent mark postures function in part to convey some aspects of competitive ability.

Scent and mother-infant recognition. The female panda often leaves her cub for long periods of time while she ventures off to forage on bamboo. In relocating the cub, particularly once it is mobile, the mother needs a way to recognize her cub and distinguish it from other cubs. As a species that relies heavily on scent for so many aspects of its social behavior, scent might also be expected to play a prominent role in mother-infant recognition. Our studies have shown that females recently separated from their cubs prefer to investigate the odors of their own cubs over those of a strange cub of the same age and sex. Interestingly, females also engaged in significantly more investigation of the environment in response to their own cub's odor, suggesting that olfactory detection of their cub may have stimulated further searching for the cub's whereabouts. When presented with their mother's and a strange female's odors on a stuffed burlap sack, cubs sniffed and engaged in more comfort-related behaviors with the sack when the sack was saturated with their own mother's odors. They also displayed fewer signs of agitation, suggesting that maternal odors have a calming effect.

Chemical signatures. Recognition of individually distinctive cues in chemical signals is another important aspect of olfactory communication systems. To test for this ability, we employed a habituation-discrimination paradigm in which each subject was exposed to the scent of one individual on five habituation trials, followed by the test presentation of scent from the same and a different individual. Pandas showed a significant preference for the novel scent, as indicated by time spent smelling the stimuli. These results provide strong evidence that giant panda scent marks contain individually distinctive chemical cues, which enables them to recognize each other solely on the basis of odors.

Age of scent. We also conducted an experiment testing giant pandas' ability to discriminate among scent marks of different ages. The ability to assess the age of a scent mark may have important consequences for regulation of giant panda social behavior. As scent marks left in the environment are left to age, exposed to sun, air and moisture, they can degrade in fairly predictable ways. Volatile constituents are slowly lost, weakening the signal, and the chemical structure of some compounds may be altered. An important source of contextual information extractable from scent marks by pandas, therefore, may be elapsed time since the scent was deposited. Pandas visit communal scent mark stations where they undoubtedly encounter a potentially confusing variety of scents of various ages. Clearly, the age of scent marks potentially conveys useful information such as the recency and/or frequency of the signaler's visit to the area.

We conducted a series of tests and found that adult pandas prefer to investigate fresher scents (e.g., 1 day old) more than older scents (e.g., 5 days old). In contrast, juveniles, while equally interested in investigating conspecific odors, did not show any tendency to prefer fresh odors over older odors. One interpretation of these findings is that young pandas must learn to discriminate between these odors through repeated exposure to conspecific odor. This research suggests that young pandas should be exposed to odors of other pandas so that they can learn how to communicate effectively.

In order to understand chemical communication in the giant panda, we must determine how long these chemical signals remain biologically active. Are scent marks ephemeral chemical signals or can pandas detect them days or months later? Knowledge of this ability to detect aged scents has important implications for the functions of chemical communication in this species, as well as their practical application in the interest of conservation. Based on pandas' behavioral responses to odors aged under natural conditions exposed to sun, rain, etc., anogenital marks may last as long as 3 months, but urine only last 2 weeks to a month.

Conservation implications. To date direct applications of this new understanding of chemical communication in pandas has been limited to captive breeding efforts. Some of these applications are discussed above. However, we envision a future where we can use chemical signals to guide an informed management strategy in conservation programs with the wild population. For example, we now know that pandas can recognize individual scents, and presumably their own, and that scent plays a prominent role in the regulation of many aspects of social behavior. Efforts to reintroduce captive animals to the wild have often failed because the animal fails to remain near the release site or because of excessive aggression between the reintroductee and neighboring conspecifics. A promising strategy to address this problem is the pre-application of the reintroductee's own scent to demarcate a territory prior to release. This may reduce the stress involved in relocation to the new environment, encourage the reintroductee to remain in the area, and reduce aggressive interactions with neighbors by familiarizing them with the reintroductee's scent and "bluffing" the presence of a territory owner.

In another scenario, it has been established that many species use the presence of other conspecifics as and indicator that the area is suitable and safe (e.g., the use of duck decoys relies on this principle). This may be one reason that suitable habitat in new reserves or habitat corridors is not readily utilized, raising the possibility of "planting" olfactory cues from conspecifics to encourage the use of these areas. In the case of pandas, we have shown that some cues may be aversive, while others are attractive. For example, it might be inadvisable to place male urine at a height mimicking the handstand posture of a male of high competitive ability, as other pandas may tend to avoid the area.

Reproductive Behavior

One of our first research projects was to study the course of behavioral events of estrus, documenting the timing of onset and termination of reproductive behaviors. Daily observations of several females have shown a great deal of inter-individual variability in the details of behavioral estrus, but also identified some common behavioral profiles present in most females. Behaviors such as water play, rolling, and backwards walking vary substantially among females. Scent marking, however, follows a similar trend in most females, increasing in rate several days before the fertile period and decreasing to baseline just before ovulation. Coincident with this decline in scent marking, we see dramatic peaks in behaviors such as chirping, bleating, tail up and presentation of the vaginal area to the male. These behaviors offer a reasonably reliable indicator of ovulation, correlate well with physiological measures of ovulation, and have been used successfully to determine the best time for natural mating or artificial insemination

As part of this research, we have developed a method of recording social interactions to better reveal reciprocal influences and behavioral contingencies in male-female interactions across the female's estrous cycle. We are using both sequential and traditional analysis techniques to tease out the meaning of courtship behaviors. We have identified an interesting pattern wherein male efforts to interact with females are relatively affiliative regardless of the female's estrous phase, whereas female behavior shifts from neutral and aggressive to one of ambivalence, and finally to affiliative/sexual as the peri-ovulatory period commences. We are addressing questions such as, "How do males and females interact prior to successful copulation, and how does this interaction differ between successful and non-successful breeders?" And, "What behaviors by the male stimulate female sexual motivation and what behaviors by the female stimulate male sexual motivation?" These data, in conjunction with the data on female behavioral estrus, have been used to evaluate behavioral deficiencies in non-breeders and suggest remedial action that appears to have met with some success.

We also have videotaped mating introductions in a study aimed at understanding the penultimate stages prior to successful or failed mating, as well as the behavioral form of copulation itself. These videotapes are being analyzed to determine which behaviors lead to successful copulation and why some mating introductions fail.

Mother Infant Relationship and Early Development

This research program has two primary objectives. First, we seek to estimate the range of inter-individual variation in maternal care-giving in captive giant pandas and document species-typical mother-infant relationships. By establishing patterns of successful maternal behavior, we will be better prepared to evaluate individual mothering styles that fall outside the normal range and target specific behavioral problems that contribute to rearing failure. The second goal is to seek causal relationships between environmental variables, mothering style, and rearing success. By observing the interactions between the mother and infant closely, we began to understand what sorts of activities a good panda mother engages in, and how these behaviors change as the cub develops. For example, we were impressed with the amount of care the mother devotes to her cub. At a mere 4 ounces, the mother must provide for all the needs of their extremely undeveloped offspring. During the first few weeks of life, the mother is in constant contact with the cub, and makes an effort to keep the cub covered with her arm, paw or chin. If the mother's attention lapses and the cub becomes exposed to the air, it emits a loud squawk or more quiet croak, and the mother immediately attends to the cubs needs. The mother also invests a great deal in keeping the cub clean, frequently licking the cub, stimulating it to pass urine or feces, and cleaning up the waste afterwards. As the cub matures, it gradually becomes less reliant on the mother, and we observe a concomitant reduction in maternal care by the mother. For example, the mother spends less time in contact with the cub, licks and interacts with it less, and the cub vocalizes less frequently while the mother becomes less responsive to those vocalizations.

This research program has already begun to bear fruits in the management realm. We recently developed a method to facilitate maternal care in a giant panda female that rejected her cub immediately after giving birth. Upon birth the mother appeared fearful of her newborn cub, sniffing it cautiously and jumping back when it vocalized. Following removal of the cub, the female was systematically exposed to a regime of infant-related stimuli. We used a stuffed panda toy as a vehicle to deliver the infant's scent (urine and feces) to the mother, which was accompanied by playbacks of the infant's vocalizations. The female treated the surrogate almost as if it were a real cub; for example, she spent 61% of her time holding the surrogate in positions typical for the species, licked the surrogate to clean off the urine, and responded to playbacks of the cub's croaks, squawks, and cries. She also showed signs of gradual improvement in how she treated the surrogate, encouraging us to begin a process of gradual reintroduction of the cub to her mother. After a transitional period in which we assisted the female in her efforts to nurse and groom the infant, all maternal care-giving responsibilities were returned to the mother. Over the next three months the pattern of maternal care followed the species-typical course of declining mother-infant contact, grooming, overall interaction time, and responsiveness to the cub's vocalizations. This study marks the first successful reunification of a giant panda mother with an infant separated at birth, and hopefully will serve as a model for similar efforts elsewhere.

Studies of the reproductive strategies pertaining to infant growth and development and maternal solicitude provide one of the few detailed glimpses into the behavioral ecology of this species in the wild. By following several radio-collared females throughout the periods of mating, birth, and early cub development, our colleagues at Peking University have provided new insights into the life history strategies in giant pandas. Panda reproduction is characterized by short gestation, delayed implantation if the embryo, extremely altricial young, a high degree of maternal investment, and slow infant growth and development. Although this strategy shares many affinities with the Ursidae, pandas differ from other bears in the larger mother:infant weight ratio at birth and the markedly more active participation in care of the infant. Bear cubs receive little direct assistance from the mother and must seek the mother's warmth and find the teat on their own. By contrast panda dams actively hold the infant and assist in the cub's efforts to locate the teat and suckle. These deviations from the norm for ursids may relate to the panda's adaptation to and reliance on bamboo as the primary food source. Nutritional constraints of this diet may preclude hibernation, and require that the dam give birth to more altricial young and continue to forage throughout the denning phase to support lactation. However, other aspects of panda reproduction might be better explained by the conservatism of phylogenetic inertia. Perhaps pandas continue to give birth to twins, even though they can only rear a single cub, because of their shared evolutionary history with bears, which routinely give birth to more than two cubs. In addition to insights into panda life histories, this study provides valuable data from wild pandas that can serve as a model for appropriate mother-infant interactions in captivity.

Behavioral Development and Environmental Enrichment

Behavioral development is of the utmost importance for captive breeding because the consequences of adverse developmental outcomes can have far-reaching and possibly irreversible effects. We devised an ethogram emphasizing play behavior, social and environmental responsiveness, chemical communication skills and reproductive behavior. We are currently observing several individuals in a long-term longitudinal study of development. The primary objective of this research is to obtain a record detailing the course of developmental events in giant pandas and to examine the effect of varied developmental experiences on behavioral ontogeny. In a related study, we are trying to determine whether style or efficiency of feeding on bamboo varies as a function of origin (wild- or captive-born), with the goal of evaluating competency of candidates for reintroduction. Videotapes are decoded to provide precise quantitative records of preferences, manipulation for consumption, mastication, and other details of feeding.

We have noted that many pandas living in captivity express some form of abnormal behavior to varying degrees. These "stereotypic" behaviors are highly repetitive, invariant in form, and deviate from behavior patterns observed in the wild. These behaviors are believed to indicate suboptimal psychological well-being. To address these concerns, an environmental enrichment program aimed at promoting physical and psychological health was developed. We have integrated enrichment principles into husbandry practices in Wolong based in part on tests of enrichment in San Diego. In Wolong (1) seven large naturalistic exhibits have been built, (2) existing exhibits have been modified to increase structural complexity by adding vegetation, dead wood, and digging pits, (3) keeper interactions are more gentle, (4) better crowd control management has been adopted, (5) smaller meals are fed more frequently to avoid the development of feeding anticipation, (6) high-fiber biscuits have replaced richer breads, and (7) bamboo feeding has increased. These changes provide more opportunities for pandas to interact with their environment, work for food, and have more control over access to important resources such as food and shelter.

We also developed a supplemental enrichment program in which the pandas are regularly provisioned with manipulable objects such as fresh spruce branches and burlap sacks stuffed with straw and feeder devices that make pandas work to obtain food. Initial tests of these enrichment items demonstrated a significant reduction in many stereotypic behaviors when enrichment was present than when it was absent. They also spent more time active and engaged in a greater diversity of more natural behaviors. These effects continued even in the immediate aftermath of an enrichment interaction, suggesting the enrichment had a carry-over motivational effect beyond just occupying more of their time. Although they interact less with these supplemental enrichment items today than when they were first given, they still occupy a significant amount of many pandas' time.

Perhaps because of these diversions that occupy their time with more natural behaviors, we have witnessed a decline in the time that pandas spent engaged in stereotypic behavior in recent years. Another positive change that may be partially attributable to these efforts is the recruitment of three new breeding females that had heretofore failed to breed successfully. These three females are probably the most timid animals at the center, and it seems plausible that this aspect of their temperament interacted negatively with potentially stressful living conditions, and that recent success stems from a more positive living environment.