Orangutan (Pongo pygmaeus abelii) Spatial Memory and Behavior in a Foraging Task

Transcription

1 Journal of Comparative Psychology 1999, Vol. 113, No. 2, Copyright 1999 by the American Psychological Association, Inc /99/S3.00 BRIEF COMMUNICATIONS Orangutan (Pongo pygmaeus abelii) Spatial Memory and Behavior in a Foraging Task Suzanne E. MacDonald and Maria M. Agnes York University Orangutan (Pongo pygmaeus abelii) spatial memory was explored by using an arboreal foraging paradigm in a zoo environment. The experiment consisted of 4 tasks: walking maze, win-stay, food density, and competition. All 3 orangutans made very few revisits to previously depleted sites in the walking maze task. In the win-stay task, 2 orangutans were more accurate than chance at remembering locations of specific baited sites. Food density affected the behavior of 1 orangutan. Foraging with a competitor did not increase the orangutans' accuracy at recovering food from baited locations. The orangutans did not compete directly for access to the food sites but did avoid visiting sites depleted by a competitor. The orangutans' foraging patterns suggest that they minimized energy expended by reducing the distance traveled to retrieve food items. The ability to remember locations in the environment is an important cognitive skill that can allow animals to search efficiently for food, mates, and other resources. Not surprisingly, a wide variety of organisms have been documented to have well-developed spatial memory, including primates. Chimpanzees (e.g., Menzel, 1973) and gorillas (MacDonald, 1994), as well as several monkey species (e.g., De Lillo, Visalberghi, & Aversano, 1997; Garber, 1989; MacDonald, Pang, & Gibeault, 1994; MacDonald & Wilkie, 1990), have demonstrated excellent memory for spatial location in a wide variety of testing paradigms, ranging from laboratory to field studies. However, in contrast to the findings that suggest that nonhuman primates have highly developed spatial abilities, Andrews (1988) found that titi monkeys (Callicebus moloch) and squirrel monkeys (Saimiri sciureus) did not avoid revisiting previously visited food sites in an eight-arm maze. Similarly, Roberts, Mitchell, and Phelps (1993) did not find evidence of accurate spatial memory in squirrel monkeys foraging in laboratory trees. Although the monkeys could distinguish between baited and empty trees, they had difficulty finding specific baited sites on individual trees. Suzanne E. MacDonald, Department of Psychology, York University, Toronto, Ontario, Canada; Maria M. Agnes, Department of Biology, York University, Toronto, Ontario, Canada. This research was supported by an operating grant from the Natural Sciences and Engineering Research Council of Canada. Portions of this research were presented at the annual Conference on Comparative Cognition, Melbourne, Florida, March The encouragement, cooperation, and assistance by the staff at the Metropolitan Toronto Zoo were greatly appreciated. Special thanks to Bev Carter. Correspondence concerning this article should be addressed to Suzanne E. MacDonald, Department of Psychology, York University, Atkinson College, 4700 Keele Street, Toronto, Ontario M3J 1P3 Canada. Electronic mail may be sent to suzmac@yorku.ca. Spatial memory, then, does not appear to be equally relied upon by all primate species. In fact, there is some evidence that the ability to remember locations in the environment may be related to the distribution of resources and the foraging problems a species faces. Milton (1981, 1993) suggested that the extreme diversity of plant foods in tropical rain forests and their distribution have been selective forces in the cognitive development of primates. Herbivores do not have to expend energy or time pursuing prey, because their food sources are sessile, but they are under selection pressure to increase food search efficiency, which lowers time and energy costs expended in foraging and also reduces predator exposure. All else being equal, Milton hypothesized that primate species dependent on the most dispersed and patchy foods, such as fruit, should show greater evidence of cognitive development than primates eating more uniform dietary resources, such as leaves. Orangutans are predominantly frugivores (MacKinnon, 1974; Rodman, 1977) and spend most of their time foraging (Galdikas, 1988). In spite of the irregular temporal and spatial distribution of resources and poor visibility because of dense foliage, researchers have long reported that orangutans are highly selective and adept at finding fruit in the rain forest (e.g., Galdikas & Vasey, 1992). The purpose of the present study was to experimentally explore spatial memory in orangutans (Pongo pygmaeus abelii) by using a naturalistic foraging task. Eight food sites were placed throughout a large enclosure. To incorporate the arboreal nature of orangutan foraging, we designed the task so that all food sites were elevated and the orangutans had to climb or swing from site to site. The study consisted of several tasks: walking maze, win-stay, food density, and competition. The walking maze task was an analogue of a radial-arm maze, with individual orangutans foraging freely among the eight baited sites. The primary purpose of the walking maze task 213

2 214 BRIEF COMMUNICATIONS was to examine the orangutans' ability to avoid revisits to depleted sites. The remaining tasks were variations on the win-stay task used in prior research with other primate species (MacDonald, 1994; MacDonald et al., 1994; MacDonald & Wilkie, 1990). In the win-stay task, only four randomly determined food sites (of the eight possible sites) were baited on each trial. During the search phase, an orangutan freely foraged among the sites and obtained the food. After visiting all eight sites, the orangutan was removed from the enclosure for a delay period. During the delay, the same four sites that had previously contained food were rebaited. Then, during the retrieval phase, the animal was allowed to forage again. This basic win-stay paradigm allowed us to explore the orangutans' ability to recall specific locations in their environment. The food density and competition tasks were variations on the win-stay task. In the food density task, one of the four baited food sites was randomly chosen to be baited with five times more food than was contained in the other sites. Of interest was whether the orangutans would alter their foraging pattern to preferentially visit the overbaited site once they had discovered it. In the competition task, the orangutans initially searched the baited sites alone, but each participated in the retrieval phase with a competitor. We were interested in the effect that foraging for limited food with a social partner would have on the strategies used by the orangutans. Subjects Method Three experimentally naive orangutans (Pongo pygmaeus abelii), 1 adult female and 2 subadult males, served as subjects. The adult female had been wild bom but had lived in captivity for 24 years; the 2 subadult males had been born in captivity. All 3 animals were housed at the Metropolitan Toronto Zoo (Toronto, Ontario) for the duration of the experiments. At the time of testing, Puppe (adult female) was approximately 30 years old, Juara (subadult male) was 11 years old, and Dinar (subadult male) was 9 years old. Testing Area Testing was conducted at the Metro Toronto Zoo in a large (8.7 m X 4.6 m X 7.0 m) indoor orangutan enclosure, shown in Figure 1. The enclosure was part of the holding area where animals 8 catwalk N. windows orangutan door V feeding platform keeper door orangutan \door Figure 1. Schematic side view of the orangutan testing area.

3 BRIEF COMMUNICATIONS 215 resided overnight or when off exhibit; only zoo staff and the experimenters had access to this area. The enclosure contained several steel climbing structures, ropes, elevated platforms, and steel-mesh nests. The floor, ceiling, and walls of the enclosure were concrete. Natural light illuminated the enclosure via skylights. Lights in the general holding area came on at 7:30 a.m. and went off at 5:30 p.m. Three Plexiglas windows allowed visual access to the enclosure. The largest of these windows was situated above a platform containing a feed box. Observations were made from one of the two smaller side windows shown in Figure 1. The orangutans could be remotely transferred in and out of the testing area to adjacent holding pens via three Plexiglas doors. In addition, two keeper doors allowed access to the ground level of the enclosure and the highest platform. Eight white opaque plastic containers, similar to bleach containers, with a hole cut in the side to accommodate an orangutan's fist served as food sites. The containers were attached with heavy string to climbing structures at a variety of heights within the enclosure, as shown in Figure 1. The containers remained in the same locations throughout the experiment and were baited with bananas (a portion of the orangutans' daily rations). Because the containers hung from the climbing structures, they remained in the same positions once depleted and dropped. This arrangement eliminated the possibility that the orangutans could use the positions of the containers as cues to avoid revisiting depleted sites. Testing took place at 2:00 p.m., 3 to 5 days per week. Individual orangutans received only one trial per testing day. The orangutans were fed twice daily, first at 8:00 a.m. Their second daily meal was usually delivered at 12 noon, but on testing days, delivery of this meal was delayed until after testing was completed (approximately 3:30 p.m.). Procedure The experiment consisted of four tasks: walking maze, win-stay, food density, and competition. Juara and Dinar participated in all four tasks. Puppe participated in the walking maze and win-stay tasks only; she became pregnant during the win-stay trials and was no longer available for testing. Walking maze task. On each walking maze trial, all eight food sites were baited with one (approximately 10 g) piece of banana. The orangutans were released individually into the enclosure and allowed to explore, removing the banana from each food site. The order in which sites were visited was recorded. We considered a site "visited" if an orangutan put his or her hand into the container or turned the container to look into its opening. After an animal had visited all eight food sites, the session was terminated and the orangutan was taken from the testing area. No time limit was placed on the sessions, although most ended within 10 min. Walking maze testing continued until each orangutan had visited all eight sites with no revisits on five consecutive trials. Juara met this criterion in five sessions. Dinar and Puppe met the criterion in 7 and 10 days, respectively. Win-stay task. In the win-stay task, each trial consisted of two phases: search and retrieval. As in the walking maze task, individual orangutans foraged among eight food sites. However, only four randomly determined food sites were baited (with a single piece of banana) during each trial. In the search phase, an animal visited all eight food sites and retrieved the food from the four baited sites. The animal was removed from the test enclosure for a short delay (ranging from 5 to 10 min), during which an experimenter rebaited the four food sites that had been baited during the search phase. The animal was then reintroduced into the test enclosure and allowed to search through the food sites again, in what was referred to as the retrieval phase of the trial. For both the search and retrieval phases, the order of visits and the number of choices taken to recover the food from all four baited sites were recorded. After an animal had visited all eight food sites, the session was terminated and the orangutan was taken from the testing area. No time limit was placed on the sessions, although most ended within 10 min. Juara participated in 19 win-stay trials, Dinar participated in 17 trials, and Puppe participated in 18 trials. Food density task. As in the win-stay task, four randomly chosen food sites were baited on each food density trial. In this task, one of the four sites was randomly chosen to be "overbaited" with five pieces of banana. The other three sites were baited as before, with a single piece of banana. The testing procedure (including search and retrieval phases) was identical to that of the win-stay task. Dinar and Juara each participated in 12 food density trials. Competition task. In the competition task, orangutans foraged individually during the search phase of each trial but foraged with a competitor during the retrieval phase of each trial. As in the win-stay task, four randomly chosen food sites were baited with one piece of banana. The orangutans were released individually into the testing area and allowed to explore all eight food sites. Once an orangutan had visited all eight sites and recovered the food from the four baited sites, he was removed from the test area. For both subjects, the four baited sites in the search phase were identical. After a 5-min delay, during which the same four sites were rebaited, both subjects were reintroduced into the test area together and permitted to search through the food sites. The competition task was terminated after 10 trials, when Juara was transferred to another zoo. Results Behavioral observations of the orangutans during all four tasks suggest that the orangutans were highly motivated to find the hidden food items. As soon as the experimenters) entered the orangutan holding area, the animals began vocalizing loudly, each trying to gain access to the connecting tunnel to the test area. The animals always depleted the food from all baited sites before leaving the test area. Walking Maze Task All 3 orangutans were accurate at the walking maze task; they visited all eight food sites and made few revisits to depleted sites. Juara made no revisits in five trials (0 out of 40 total choices). Dinar made four total revisits in 7 trials (4 out of 60 total choices), and Puppe made 9 total revisits in 10 trials (9 out of 89 total choices). The task appeared to be extremely easy for the orangutans, probably because they tended to visit the sites in a consistent pattern. All 3 orangutans depleted the containers on the ground level (Sites 1, 2, 3, and 4) first. As shown in Figure 1, these sites were closest to the animals' point of entry into the test area. After these sites were depleted, they climbed up to the sites at the intermediate level (Sites 5 and 6), and then they climbed to the sites at the highest level (Sites 7 and 8). The arrangement of the sites in the enclosure was such that sites on the same level were less than 2 m apart, which corresponded roughly to the arm span of the orangutans. Distances between sites on different levels ranged from 3.5 m to 6 m, which meant that the animals had to climb and swing to move from one level to the next. All 3 orangutans exhaustively searched adjacent sites on one level before moving to the sites on the

4 216 BRIEF COMMUNICATIONS next level. From the first test session, each of the orangutans stopped searching when the eight food sites were depleted, climbed up to the feeding platform, and waited for the keeper to open the remotely operated exit door. Win-Stay Task Because the minimum number of visits required to recover all of the food items was four, accuracy in the win-stay task was measured by the number of correct choices made in the first four site visits. A correct choice was defined as a visit to a baited site. In the walking maze task, the orangutans had almost exclusively visited the groundlevel Sites (in that order) in their first four choices. In the win-stay task, the orangutans continued to follow that pattern during the search phase (Juara: 16 out of 19 trials; Dinar: 14 out of 17 trials; Puppe: 12 out of 18 trials). However, in the retrieval phase, the behavior of the orangutans changed. Juara visited Sites on only 5 out of 19 trials; Dinar visited those sites on 7 of 17 trials. Puppe actually increased the number of visits to those sites in the retrieval phase, using the pattern on 16 out of 18 trials. All 3 orangutans used the pattern only on trials for which at least two (range: two to four sites) of the ground-level sites were baited. Because the orangutans used a pattern during the search phase, we compared the number of baited sites visited on the first four choices of the search phase (when the animals did not know where the food was located) with the number of baited sites visited on the first four choices of the retrieval phase (when the animals did know where food was located). In this way, our chance probability for the accuracy calculations for each orangutan was adjusted to take into account the orangutan's foraging patterns. The mean number of correct choices in the first four choices was above chance for 2 of the orangutans, Juara: search phase: 1.69 versus retrieval phase: 2.56, /(18) = 3.39, p <.01; Dinar: search phase: 1.43 versus retrieval phase: 2.14, f(16) = 3.14, p <.01. Accuracy for the third orangutan, Puppe, was not significantly different from chance: search phase: 1.94 versus retrieval phase: 2.0, f(17) =.67, p >.05. There was no acquisition curve; accuracy did not improve over trials, F(2,15) =.978,/».05. Food Density Task The presence of the overbaited site in the food density task did not appear to affect Dinar's behavior, but it did alter Juara's foraging pattern. Juara chose the overbaited site first in the retrieval phase of 5 of the 12 trials. On each of those 5 trials, the overbaited site was the one that had never been chosen first in the search phase. As in the win-stay task, accuracy was calculated on the basis of the animals' "chance" performance during the search phase. Juara's accuracy remained above chance, t(ll) = 5.45, p <.001, and in fact improved compared with his win-stay accuracy, f(ll) = 2.15, p < 0.5. Dinar, on the other hand, chose the overbaited site first in the retrieval phase of only 1 of the 12 trials; that site was the same as the site visited first during the search phase of the trial, suggesting that Dinar visited the site first purely by chance. Dinar's accuracy in the food density task did not differ from his win-stay accuracy. Competition Task In the competition task, Dinar was much more accurate than Juara. Juara's accuracy in the competition task was significantly lower than chance, t(9) = 6.09, p <.001; Dinar's accuracy did not differ from his accuracy in the win-stay task. Thus, the presence of a competitor did not increase orangutans' accuracy at recovering food from baited locations. The most interesting result of the competition task was that the behavior of each individual depended on the behavior of his competitor: When the orangutans foraged alone in the search phase, each visited all eight sites; in the retrieval phase, each orangutan visited far fewer than eight sites. The combined number of visits made by both orangutans across trials was not statistically different from eight, ((9) = 1.62, p >.05. This finding suggests the orangutans may have avoided visiting sites they had seen their competitor deplete. This finding also explains Juara's decreased accuracy in this task. Dinar moved much faster in the competition task than in the win-stay task and therefore depleted more sites (mean number visited: 6.2) than Juara (mean number visited: 2.4). Because Juara avoided visiting sites that had already been visited by his competitor, he visited few sites and as a result exhibited low accuracy. Discussion The orangutans in this study made few revisits to depleted sites in the walking maze task, which is evidence that the animals may have remembered which sites they had visited. However, all 3 orangutans visited the sites in a very consistent order, which probably reduced their memory load. The win-stay task was a better estimate of the orangutans' memory for location, because the baited sites differed from trial to trial, forcing the animals to break up their preferred foraging patterns. Two of 3 animals were accurate at recalling the locations of baited sites in the win-stay task and continued to perform above chance in the food density task. Increasing the density of food in one of the four baited sites affected the foraging behavior of only 1 of 2 orangutans. However, because our experimental setup provided little incentive for the orangutans to visit the overbaited site first, it is difficult to draw conclusions from these data. Although our results suggest that orangutans can discriminate food density, further research using more pronounced density differences, perhaps in a forced-choice paradigm, should provide more definitive results. We had expected the presence of a feeding competitor in the competition task to increase accuracy, because incorrect choices would mean that the orangutans risked losing food to their competitor. In fact, the opposite effect occurred: Feeding competition decreased 1 animal's accuracy and had no effect on the accuracy of the other. The finding that the behavior of the individual orangutans depended on the

5 BRIEF COMMUNICATIONS 217 behavior of their competitor is consistent with the results of a similar study conducted with western lowland gorillas (Gibeault & MacDonald, 1998). Like the gorillas, orangutans rarely visited locations that they had seen visited by their competitor. However, unlike the gorillas, the orangutans in this study did not engage in overt competition no aggression over baited sites was observed. Whether this result was a consequence of the relatively low stakes involved in a zoo environment or related to the relatively solitary lifestyle of orangutans remains to be explored. The level of accuracy exhibited by these orangutans is consistent with that observed in other primates in similar tasks, including gorillas (Gibeault & MacDonald, 1998; MacDonald, 1994), marmosets (MacDonald et al., 1994), and yellow-nosed monkeys (MacDonald & Wilkie, 1990). However, it is important to note that the orangutans' performance was not error free, and that only 2 of the 3 orangutans performed at a level above chance. We believe that the orangutans' errors were primarily the result of their heavy reliance on an adjacency strategy in all four tasks. The use of an adjacency strategy may minimize energy expended by reducing the distance traveled to retrieve food items. This finding is consistent with observations made by researchers studying orangutans in the rain forest. According to MacKinnon (1974) and Galdikas and Vasey (1992), orangutans use direct routes to reach out-of-sight fruit trees. Such "shortcuts" suggest that they minimize foraging costs by choosing the most efficient routes between food sources. This strategy is adaptive for orangutans considering their large body size (Eckhardt, 1975), consequent high energy requirements, and costs of locomotion (Wheatley, 1982). This pattern has also been observed in other primates (De Lillo et al., 1997; Garber, 1989; MacDonald et al., 1994; MacDonald & Wilkie, 1990; Menzel, 1973), suggesting that the use of adjacency or least-distance strategies may be a widespread adaptation. References Andrews, M. (1988). Selection of food sites by Callicebus moloch and Saimiri sciureus under spatially and temporally varying food distribution. Learning and Motivation, 19, De Lillo, C., Visalberghi, E., & Aversano, M. (1997). The organization of exhaustive searches in a patchy space by capuchin monkeys {Cebus apella). Journal of Comparative Psychology, 111, Eckhardt, R. (1975). The relative weights of Borean and Sumatran orangutans. American Journal of Physical Anthropology, 42, Galdikas, B. M. (1988). Orangutan diet, range and activity at Tanjung Puting, Central Borneo. International Journal ofprimatology, 9, Galdikas, B. M., & Vasey, P. (1992). Why are orangutans so smart? Ecological and social hypotheses. In F. Burton (Ed.), Social processes and mental abilities in non-human primates (pp ). New York: Edwin Mellen Press. Garber, P. (1989). Role of spatial memory in primate foraging patterns: Saguinus mystax and Saguinus fuscicollis. American Journal of Primatology, 19, Gibeault, S., & MacDonald, S. E. (1998). Spatial memory and foraging behavior in captive western lowland gorillas (Gorilla gorilla gorilla). Manuscript submitted for publication. MacDonald, S. E. (1994). Gorillas' (Gorilla gorilla gorilla) spatial memory in a foraging task. Journal of Comparative Psychology, 108, MacDonald, S. E., Pang, J., & Gibeault, S. (1994). Marmoset (Callithrix jacchus jacchus) spatial memory in a foraging task: Win-stay versus win-shift strategies. Journal of Comparative Psychology, 108, MacDonald, S. E., & Wilkie, D. M. (1990). Yellow-nosed monkeys' (Cercopithecus ascanius whitesidei) spatial memory in a simulated foraging environment. Journal of Comparative Psychology, 104, MacKinnon, J. (1974). The behavior and ecology of wild orangutans (Pongo pygmaeus). Animal Behaviour, 22, Menzel, E. (1973, November 30). Chimpanzee spatial memory organization. Science, 182, Milton, K. (1981). Distribution patterns of tropical plant foods as an evolutionary stimulus to primate mental development. American Anthropologist, 83, Milton, K. (1993, August). Diet and primate evolution. Scientific American, 269, Roberts, W, Mitchell, S., & Phelps, M. (1993). Foraging in laboratory trees: Spatial memory in squirrel monkeys. In T. Zentall (Ed.), Animal cognition A tribute to Donald Riley (pp ). Hillsdale, NJ: Erlbaum. Rodman, P. (1977). Feeding behavior of orangutans of the Kutai Nature Reserve, East Kalimantan. In T. Clutton-Brock (Ed.), Primate ecology: Studies of feeding and ranging behavior in lemurs, monkeys and apes (pp ). London: Academic Press. Wheatley, B. (1982). Energetics of foraging in Macacafascicularis and Pongo pygmaeus and a selective advantage of large body size in the orangutan. Primates, 23, Received July 29, 1997 Revision received August 18, 1998 Accepted August 25, 1998