A Message From The SSP Coordinator, Don Lindburg
Research on the Foraging Strategies, Perceptual, and Cognitive Abilities of the Giant Panda
Principal Investigator: Loraine R. Tarou
In collaboration with: Rebecca Snyder, Zoo Atlanta, and Ron Swaisgood, C.R.E.S. San Diego
Optimal foraging theory postulates that animals use efficient foraging strategies when searching for and processing food. Efficient strategies are those that maximize the energetic gains and minimize the energetic costs of foraging. One of the goals of research on optimal foraging theory is to determine what type of information a forager uses to make decisions such as where to forage, when to forage, which food items to consume and how long to stay in a particular food site. In locating sources of food, animals can use vision, olfaction, spatial memory, or any combination of these cues. The giant panda is somewhat of an anomaly in the animal kingdom in that it is a carnivore that subsists almost exclusively on an herbivorous diet. Bamboo comprises approximately 99% of their diet (Schaller, et al., 1985). Pandas have many morphological, physiological and behavioral adaptations that presumably help them increase their foraging efficiency on a low-quality food source like bamboo. For example, they have a unique pseudo-thumb which helps decrease handling time. They also are large in size which gives them a lower metabolic rate and allows them to subsist on a poor diet (Schaller, et al., 1985). Behaviorally, Schaller et al., (1985) state that pandas remain inactive for much of their waking hours to conserve energy. However, it is unknown what role perceptual/cognitive abilities may play in giant panda foraging. We designed a series of studies to examine the foraging strategies used by the giant pandas. The study was initially conducted with the pandas at Zoo Atlanta. It was then expanded to include the pandas at both the Smithsonian National Zoological Park and the San Diego Zoo. The specific goal of the study was to determine the type of information pandas use to make foraging decisions.
Experimental Procedure
This study includes one male (Yang Yang) and one female (Lun Lun) giant panda at Zoo Atlanta, one male (Tian Tian) and 
one female (Mei Xiang) giant panda at the Smithsonian National Zoological Park, and one male (Shi Shi), one female (Bai Yun) and one juvenile female (Hua Mei) giant panda at the San Diego Zoo. The project is divided into six foraging tasks (listed below). Feeders were constructed of PVC pipe. Each feeder was topped with an opaque piece of plexiglass, which was connected to the feeder by a spring-loaded hinge to form a lid. The pandas are able to obtain a food reinforcer, specifically leafeater biscuits, by lifting the lid, using either their muzzles or paw. The pandas at Zoo Atlanta and the San Diego Zoo were tested separately in an indoor room. The pandas at the Smithsonian National Zoological Park were tested separately in their outdoor yards. In the test areas at each of the three institutions, eight of feeders were arranged approximately three feet apart to form a circle. One trial was conducted each day.
Task 1- Exploratory:
All eight feeders were baited with food for fifteen trials. Of interest was the pattern of movement between the feeders used by the pandas during foraging and the frequency with which the pandas visited previously depleted sites. Efficient foragers often visit adjacent food sites to decrease travel time between sites and avoid revisiting previously depleted sites.
Task 2 – Spatial:
The pandas were tested to determine whether they could use spatial cues alone to locate viable food sites. Four of the eight feeders were baited with food. Of interest in this task was the ability of the pandas to learn to visit the baited sites and ignore unbaited sites. Efficient foraging required the panda to learn position of the baited feeders and visit those feeders before any others. Trials were conducted until the pandas visited at least 3 of the 4 baited feeders in their first 4 visits across 5 consecutive trials or until a maximum of 30 trials was reached.
Task 3 – Reversal:
The locations of the four baited feeders were reversed. This let us test whether the pandas were using extraneous cues to locate the food in the feeders. Trials were conducted until the pandas visited at least 3 of the 4 baited feeders in their first 4 visits across 5 consecutive trials or until a maximum of 30 trials was reached.
Task 4 – Visual:
This task was designed to test the ability of the pandas to use visual cues alone to locate viable food sources. Four feeders were randomly baited at the beginning of each session of the study. These four feeders were distinguishable from the unbaited feeders by a difference in the appearance of the feeder itself. To forage efficiently, the pandas had to learn the association between the appearance of the feeders and the presence of food. Trials were conducted until the pandas visited at least 3 of the 4 baited feeders in their first 4 visits across 5 consecutive trials or until a maximum of 30 trials was reached.
Task 5 – Olfactory:
This task was designed to determine if pandas were capable of using olfactory cues alone to locate viable food sites. This task was similar to that of Foraging Task 4, however, baited food sites were distinguishable only by a particular scent associated with the feeder. To forage efficiently, the pandas had to learn the association between the scent and the presence of food. Trials were conducted until the pandas visited at least 3 of the 4 baited feeders in their first 4 visits across 5 consecutive trials or until a maximum of 30 trials was reached.
Task 6 – Importance of Cues:
In this task, two trials were conducted each day. In the search trial, two randomly chosen feeders were baited and signaled by spatial, visual, and olfactory cues. In the re-search trial, the visual and olfactory cues were each moved randomly so that two of the eight feeders were signaled by visual cues, two by olfactory cues, and two by spatial cues. None of the feeders were baited in the re-search phase to encourage the pandas to open all of the feeders. Reliance on each of the cues to locate food will be determined by the order in which the pandas visit unbaited sites to search for food. There were 15 search/re-search trials and 45 “filler” trials in which a search trial was followed by another search trial. These filler trials were interspersed randomly among the test trials to ensure that the pandas did not learn that the second trial of the day never contained food.
All of the pandas were tested on the exploratory task first. However, we randomized the order in which we tested the pandas on the spatial memory/reversal, visual and olfactory tasks to control for carryover effects that can occur in studies of learning and memory.
Findings to Date
In the exploratory task, we were interested in the pattern of movement by the pandas between each of the feeders. Traveling to adjacent feeders is indicative of efficient foraging. All of the pandas were very efficient foragers. However, the pandas were more efficient in the last five sessions of the study than they were in the first five sessions of the study, indicating that there was learning across consecutive trials. We were also interested in whether the pandas revisited previously depleted feeders. Once food was obtained from a feeder, the pandas should not return to that particular feeder. The results showed that each of the pandas revisited previously depleted feeders more often in the first five sessions of the study than in the last five sessions. Therefore, all of the pandas learned to forage more efficiently over time with exposure to the task. One interesting difference observed in the pandas across the institutions was that the pandas at the National Zoo and the San Diego Zoo revisited depleted feeders more often than the pandas at Zoo Atlanta. One possible reason for this is that, at the time of the study, the pandas at both the National Zoo and the San Diego Zoo were scatter fed biscuits on a regular basis, which may have taught them to return to sites where they previously found food. The pandas at Zoo Atlanta were hand-fed all biscuits when we started the study with them.
We have not finished analyzing the results, however, the pandas at all three institutions have performed similarly on the tasks. The results of our studies so far indicate that pandas are able to rely on spatial skills alone to locate the presence of viable food sites. Most of the pandas tested on the spatial memory task learned to visit only the four baited feeders within 20 test sessions. Reversing the baited feeders disrupted performance, corroborating our conclusion that pandas are able to use spatial cues alone to located hidden food. Furthermore, the pandas did not learn the position of the baited feeders in the reversal phase as quickly as they had learned the position in the spatial phase, suggesting that once they have learned the location of food they have difficulty adjusting their behavior and learning new locations. Therefore, pandas may not readily adapt to sudden changes in their environment.
It was hypothesized that pandas may learn to find food more quickly when required to use olfactory or visual cues to locate viable food sites, however, we found that this was not the case. To forage efficiently in this task, the pandas needed to learn the association between either the appearance of the feeders in the visual task or the smell of the feeders in the olfactory task and the presence of food. To date, none of the pandas that have been tested on the olfactory or visual phases used the cues provided to locate food. Instead they systematically checked all feeders until they found the four baited feeders. This may suggest that visual and olfactory cues are not important for foraging in the wild. Interestingly, our results from Task 6 (which allows use of each of the types of cueing) indicate that visual cues are most important in this task. This is surprising given the pandas’ poor performance on the visual task in which they did not use visual cues to locate food. More research will be required to explain these results.
We are currently in the process of replicating this research with two other species of bear, spectacled bears and sloth bears, housed at the Smithsonian National Zoological Park. Both of these species have very different foraging habits from the giant panda; spectacled bears eat mostly fruit, whereas sloth bears specialize on termites and ants. It is hypothesized that sloth bears and spectacled bears will perform better on all tasks than the giant pandas because their food sources are more widely dispersed and their locations are predictable in both space and time. We hope to begin data collection on these other bear species in January 2003.
Implications for Conservation
We believe that cognitive studies on endangered animals such as the panda are important for their conservation for several reasons. First, basic research is necessary for fully understanding the species we are trying to save, leading to important findings that may be useful in ways we cannot yet imagine. Second, knowledge of the foraging skills and learning capabilities could be useful in preparing animals for introduction. Animals raised in captivity often have trouble foraging efficiently when released back into the wild. Studies that can improve foraging efficiency in captive animals and stimulate naturalistic foraging behavior may be helpful for increasing the speed with which reintroduced animals can support themselves in the wild. In addition, studies of learning and behavior can provide information on the flexibility or adaptability of animals to sudden or gradual changes in their environment. For example, in this study we are finding that once pandas learn the location of food sources, they do not adapt very quickly to a sudden change in the location of those food sources. This could help us understand and deal with the impact of deforestation or other changes in habitat on the giant panda. Finally, another advantage of conducting cognitive research on endangered animals in captivity is the educational impact it can have on people. According to research, the general public is very interested in animal intelligence. In fact, a recent national survey found that intelligence was the third most important factor influencing public preference for a species, after size and aesthetic appeal. They may be more likely to conserve a species if they can associate it with intelligent behavior. Our research has recently been featured in a video for 6th, 7th, and 8th grade students produced by the Fairfax County Public School System of Virginia. This video was designed to feature scientists using the scientific method to answer important questions in animal behavior. The video was picked up by 39 states and was aired for approximately 7.5 million students on October 10, 2002. In addition, it is being shown on local PBS stations across the country. Information about this video can be found on the Fairfax Network website: www.fcps.edu/FairfaxNetwork. Conducting cognitive research in zoos and other public institutions provides a unique opportunity to reach out to a wide range of people and to educate them about the evolution of particular behavioral traits in endangered species. This in turn may alter their perceptions of animals and encourage them to contribute to and support efforts to save and protect critically endangered animals.
