Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Behavioral Ecology Vol. 7 No 3: 254-263 Sociality and asociality in white-nosed coatis {Nasua narica): foraging costs and benefits Matthew E. Gompper Department of Zoology and Program in Life Sciences (Ethology), University of Tennessee, Knoxville, TN 37996-0810, USA White-nosed coatis maintain a social structure of female-bonded groups (called bands) and solitary males. I examined the foraging success of social and solitary individuals and the possible importance of intraspecific foraging competition in maintaining the social system, particularly in the associated context of sexual dimorphism. The study population was almost entirely frugivorous-insectivorous. Invertebrate foraging success did not differ between solitary males and band members, although solitary adult females were more successful than those in bands. Fruit foraging success of solitary adult males was generally greater than that of band members, although this result varied with patch size and depended on the age class of examined band members. Small food patches showed the greatest differential between the foraging success of solitary males and band members. Agonistic interactions between males and bands often occurred at fruiting trees, and foraging group size was important in determining the outcome of these events. Larger males were able to displace solitary females and small foraging groups from fruit patches. In turn, larger groups of smaller females displaced solitary males. Male-male agonism at fruit patches was also common, with larger, older males usually winning agonistic interactions. These findings suggest that coati social structure directly influences foraging, and I therefore hypothesize that the coati social system is maintained (in part) by body size sexual dimorphism interacting with reliance on patchy defendable foods. Female group living allows increased access to patchy resources that are otherwise unavailable due to small body size relative to competing males. In contrast, larger males are able to access food patches without living in groups that might increase foraging competition. Key words: coati, Dipteryx, foraging, group living, Nasua narica, Panama, Procyonidae, Scheelea, social structure. [Behav Ecol 7:254-263 (1996)] G roup living and asociality represent social structure extremes maintained by a variety of ecological costs (e.g., increased competition and disease transmission) and benefits (e.g., avoidance of predators or enhanced success in resource acquisition) (Alexander, 1974; Clark and Mangel, 1986; Hamilton, 1971; Pulliam and Caraco, 1984). For species in which the social system varies within populations, ecological factors selecting for or against group living may act on different individuals in different ways. For example, male African lions (Panlhera leo) form coalitions to gain access to female prides, while females form prides to gain the benefits of cooperative hunting and decreased infanticide (Packer et al., 1990). In lions, ecological factors primarily operate in the same direction: toward increased sociality. The case where ecological factors select in different directions for different segments of the same population is more unusual. For example, male cheetahs (Adnonyx jubatus) form coalitions to gain access to females, while females remain solitary due to the costs of prey sharing (Caro, 1989, 1994). Here I examine the role of a single ecological factor, food availability, in maintaining the social structure of the white-nosed coati (Nasua narica: Procyonidae). The coati is unique within the order Carnivora in maintaining a dichotomous social structure of group-living females and solitary males (Gittleman, 1989; Gompper, 1995; Kaufmann, 1962). Groups (called bands) contain from 6 to over 30 related and unrelated females and their immature offspring (Gompper, 1994; Gompper and Wayne, 1996). Coati bands are not harem groups; all adult males remain solitary with the M. Gompper is now at the Department of Environmental and Resource Sciences and Program in Ecology, Evolution, and Conservation Biology, University of Nevada, 1000 Valley Road, Reno, NV 895120013, USA. Received 28 December 1994; revised 30 August 1995; accepted 30 August 1995. 1045-2249/96/S5.00 O 1996 International Society for Behavioral Ecology exception of a brief (approximately 2 week) synchronous breeding season. Coatis are the only truly social procyonids, and bands display a variety of cooperative behaviors not found in the solitary raccoons, ringtails, olingos, and kinkajous. These behaviors (e.g., coalition formation) may influence the ability of the individual to gain access to limited resources such as food (Gompper, 1994). Several researchers have postulated various costs and benefits of coati sociality and mechanisms acting in the maintenance of the coati social structure. Hypotheses for coati band gregariousness that are not resource-based include decreased parasite loads, increased protection from predation and infanticide, and population regulation (Burger and Gochfeld, 1992; Kaufmann, 1962; Russell, 1979, 1981, 1983). With regards to foraging, researchers working on Barro Colorado Island, Panama (BCI), have come to varied conclusions. Kaufmann (1962, p. 161) suggested that while cooperative hunting does not occur, foraging for vertebrates such as lizards is more effective in a group since each individual "benefits from food chased its way by other members of the band, and more such prey is obtained by the coati population as a whole than would be the case if each coati hunted separately." Smythe (1970a) suggested that female sociality and male asociality act to reduce feeding niche overlap. When fruit is rare, Smythe observed bands to concentrate foraging on leaf-litter fauna, while males foraged more on larger prey such as small and mid-sized vertebrates. Russell (1981, 1982, 1983) indicated that coati sociality was primarily due to predator and infanticide avoidance. However, Russell also suggested that fluctuations in food availability (leaf-litter fauna and fruit) combined with the predation-induced social structure strongly influence reproduction and population structure. Thus, most previous studies did not explicitly invoke foraging costs and benefits as underlying coati social structure, although Smythe did view the maintenance of coati social organization as resulting from differentiation of feeding niches. Gompper • Foraging costs and benefits of coati social structure Several unstudied aspects of the relationship between coati social structure and coati foraging costs and benefits, as well as conflicts between the above mentioned studies, suggest that additional analyses are needed. First, foraging by solitary adult males was only generally described by Smythe; the actual use of fruit, leaf-litter fauna, and vertebrates by males was not quantified. Smythe's suggestion that males on BCI are vertebrate predators contrasts with Kaufmann's and Russell's more detailed, longer-term observations suggesting that predation on vertebrates is very rare. Thus, whether differentiation of feeding niches (Smythe, 1970a) does exist has not been shown definitively. Second, observations of the importance of antipredator cooperation within coati bands do not explain why all males are solitary. Third, BCI coatis depend heavily on fruits of only a few plant species, and especially on two plant species [Scheelea (= AUaiea, Henderson, 1995) zonensis and Dipteryx panamensis] during an extended portion of the year (Gompper, 1994; Kaufmann, 1962; Russell, 1982). The patchiness of fruit distribution from these trees may influence grouping patterns by limiting access to food and increasing interference competition, as found for other species (S. Altmann, 1974; Leighton and Leighton, 1982; Terborgh, 1983). Fourth, field observations of often intense agonism among individuals over fruit, led me to hypothesize that the influence of body size on direct foraging competition may be important in maintaining coati social structure, independent of additional selective mechanisms. In this article I examine the impact on foraging of sociality and asociality in the BCI coati population, the same population studied by Kaufmann, Russell, and Smythe. First, I indirecdy examine the role of foraging competition by comparing band members and solitary males foraging on leaf-litter invertebrates and on fruit I predict that the quasi-random distribution of leaf-litter invertebrates will lead to no difference in the foraging success of solitary and group-living individuals. In contrast, I predict that the clumped, limited, and defendable distribution of fruit patches will lead to solitary males having increased foraging success relative to group members. Second, I examine foraging success of females that temporarily leave bands and forage solitarily during the birthing season. In this case, it is predicted that foraging success of solitary males and females will not differ. Finally, observations of direct agonistic interactions under fruiting trees between band members and solitary males, and among solitary males are reported. Specifically, I examine the hypothesis that body size and the number of band members are important predictors of access to fruit patches. METHODS Study population Barro Colorado, Panama (9°9' N, 79°51' W) is a 1500-ha island in Gatun Lake, an artificial body of water created in 1912 by die damming of the Chagres River to create the Panama Canal. The island is separated from the mainland by 200 to 1000 m. Many mammals including coatis have been observed swimming in the lake, such that population isolation from the mainland is incomplete. Wet and dry seasons produce contrasting seasons of food availability, which may limit BCI's mammal populations (Foster, 1982; Leigh and Windsor, 1982; Smythe, 1970b; Smythe et al., 1982). BCI is covered by tropical moist forest and has two major habitats that are distinguished by forest age. Older forest has escaped human intervention for 400 to 600 years and was never cleared for agriculture (Piperno, 1990). Younger forest was cleared 100 to 200 years ago (Foster and Brokaw, 1982). Both habitats are included in this study, with individual coatis and bands ranging in both 255 forest types. Reviews of die ecology, history, and mammal fauna of BCI are given by Croat (1978), Enders (1935), Gentry (1990), Leigh et al. (1982), and Wright et al. (1994). Field work occurred on BCI for 23 months between August 1989 and November 1993 (8/89-12/89; 6/90-8/90; 9/91-7/ 92; 9/92-10/92; 7/93; 11/93). During this study the population density of coatis on BCI was 48.2 to 55.6 individuals/km2, mean (± SD) group size was 15.3 ±6.1 individuals, and mean foraging group size was 7.3 individuals (Gompper, 1994; Wright et al., 1994). While the population density was high relative to most other neotropical sites studied, group size and foraging group size were similar to values found at other sites (Gompper, 1994, 1995; Wright et al., 1994). The population was sexually dimorphic in body size; adult female body weight (kg) (mean ± SD = 3.7 ± 0.3; range = 3.1-4.8; n = 37) was approximately 73% that of adult males (mean = 5.1 ± 0.8; range = 3.7-6.8; n = 51). Total length (cm; head-body + tail) of adult females (mean = 103.7 ± 9.8; range = 66.0-113.7; n = 32) was 91% that of adult males (mean = 114.2 ± 4.6; range = 106.0-127.0; n = 42), and mean tail length of adult females (mean = 46.7 ± 8.0; range = 12.2-53.8; n = 33) was 87% diat of adult males (mean = 53.4 ± 2.6; range = 45.860.1; n = 43). These comparisons probably underestimate dimorphism of adults of breeding status due to the inclusion of males that were recent emigrants from natal bands and that had not attained full body size (Gompper, 1994). Additional details on the population biology of the BCI coati during this study as well as methods of trapping and marking are given by Gompper (1994). Behavioral sampling Behavioral data were collected on known (usually marked) individuals that were habituated to my presence. Five bands (6 to 26 individuals per band) were habituated, allowing me to follow at a distance of a few meters. Several of these bands inhabited regions studied by Russell (1976-1978; Russell, 1979, 1982, 1983) and Kaufmann (1959-1960; Kaufmann, 1962). Behavioral data were occasionally collected from a sixth, partially habituated band when band members could be observed from a distance. A total of 34 solitary adult males were habituated to my presence; this included 18 males initially habituated as juvenile band members who remained habituated following transition to a solitary lifestyle. Identities, ages, and diets of males and band members were recorded when first encountered. Habituated animals were followed, and behaviors and food sources were recorded using focal animal and ad libitum methods (J. Altmann, 1974). Focal animal samples lasted 10 min, during which all behaviors were recorded. The resulting continuous data could then be analyzed as both event information (i.e., the number of occurrences of a behavior) and state information (i.e., die time spent performing a behavior). Following the 10 min samples, no data were collected for 5 min to minimize observer fatigue and allow individuals to be relocated, after which the next individual in the band was sampled or the solitary male was resampled. Due to die density of the forest, focal animals were often lost or obscured from sight before completion of a 10 min focal sample. If the animal could not be immediately relocated, samples of <5 min were not analyzed. Samples >5 min but <10 min were used only if major activity patterns did not change during the sample (e.g., continued foraging or continued resting, but no change from foraging to resting behavior), and only for behaviors analyzed as rates. Separate analyses were performed for adult females (>24 months of age) and subadult (12 to 24 months of age) band members. No data were collected on foraging by juveniles (<12 months of age). Behavioral Ecology Vol. 7 No. 3 256 Definitions for feeding and searching for food are as follows: • Feeding—the act of handling, chewing, and swallowing a food item. A feeding event is the handling, chewing, and swallowing of a single food item (a fruit, an invertebrate). The state of feeding is the time spent handling, chewing, and swallowing a single food item. • Searching—the act of seeking food items when the food items are in the immediate vicinity (within several meters). While this definition is somewhat arbitrary, behaviors indicative of searching are distinct and include smelling of leaf litter or fruit with a head-down posture; digging and scraping soil and leaf litter; and movement within a fruiting tree's seed shadow with a head-down posture. Field and captive studies have suggested that coatis find food primarily via odor cues (Chapman, 1938; Kaufmann, 1962) or by color contrast (Chausseil, 1992). Head-down posture allows searching to be discerned from the act of traveling toward a frequently used food patch, during which the head is usually oriented forward. Data on agonistic interactions were collected ad libitum (J. Altmann, 1974) and included identification of the individuals involved, behaviors displayed, outcome, and context (e.g., fruit foraging, leaf-litter foraging, resting). The following definitions pertaining to agonistic conflicts are based partly or in whole on ethograms of Kaufmann(1962) and Krinsley (1989): • Agonism—conflict involving directed aggression (e.g., fighting, chasing, vocalizing) by one or more individuals toward one or more individuals. Agonistic individuals are those involved in the conflict, without reference to whether they were an aggressor, recipient, winner, or loser. • Aggressor—the individual(s) commencing the agonistic interaction. • Recipient—the individual(s) receiving the directed aggression from the aggressor. Identifying the aggressor and the recipient was usually straightforward since (1) the contexts of behaviors and vocalizations are distinct (Kaufmann, 1962; Krinsley, 1989) and (2) a charge by one or more individuals often began the agonistic encounter. • Winner—the individual (s) that during or after an observed agonistic encounter (1) gained or maintained resource access, (2) chased other agonistic individuals away from the conflict site, or (3) displayed dominant vocal and visual behaviors (Kaufmann, 1962; Krinsley, 1989). • Loser—the individual (s) that during or after the conflict (1) lost or did not gain resource access, (2) was chased from the conflict site, or (3) displayed submissive vocal and visual behaviors (Kaufmann, 1962; Krinsley, 1989). Diet Although foraging on invertebrates could be directly observed, it was not generally possible to identify particular species being consumed. Foraging on fruit was readily' observable, and fruits eaten by coatis were identified using a key to BCI's flora (Croat, 1978) or through consultation with botanists working on BCI. Scats of known individuals were examined for evidence (seeds, undigested fruit, exoskeletons, bones, hair) of recent foraging on fruits, invertebrates, and vertebrate prey. BCI coatis have been observed to eat fruit from at least 48 plant species (Gompper, 1994; Kaufmann, 1962; Russell, 1982). Nonetheless, these studies have all stressed the importance of only a few tree species. Two of these species, Scheelea zonensis and Dipteryx panamensis, were selected for detailed analysis of the impact of fruit availabilit)' on coati foraging patterns. These two species were selected due to (1) the contrast in their relative patch sizes (small and large as measured in the quantity of fruits available to coatis; see below), (2) the similar sizes (approximately 6 cm long and 3 cm wide; Croat, 1978) of the fruits, and, (3) the fact that they form the principle diet of coatis on BCI from approximately May to September and December to February, respectively. The seasonal and daily availability and predictability of fruits of diese species may thus represent extreme but common situations that may influence foraging success and play roles in the maintenance of the coati social system. Scheelea zonensis is a common palm on BCI, with a density of approximately 1.8 trees/ha (Croat, 1978; Milton, 1980). Individual plants bear fruit for an average of 6 weeks and may produce several large infructescences per year, each of which has approximately 100 to 1000 fruits (De Steven et al., 1987; Wright 1990). Most foraging by coatis on Scheeleafruitis terrestrial. While coatis occasionally forage arboreally on Scheelea, die tree is difficult for diem to climb and thus canopy access must be gained through neighboring trees, which is often not possible (Gompper ME, personal observation). In addition, perch space in Scheelea is limited, allowing only a single individual access to fruit clusters (Gompper ME, personal observation; Klein and Klein, 1973). When a band member does climb a tree, the remaining band members forage under the tree on dropped fruits. Dipteryx panamensis is a common BCI canopy tree with a density of approximately 1 tree/ha (Croat, 1978; Milton, 1980). Dipteryx differs from Scheelea in its abundance of fruits; an individual may drop 104 to 105 fruits/year (Bonaccorso et al., 1980). Although coatis will enter the Dipteryx canopy to feed, most foraging is terrestrial. To assess daily variation in the availability of fruits from individual trees, I examined the quantity of fallen fruits. Several trees known to be used by coatis were arbitrarily chosen for study. Trees were censused twice daily during the fruiting season. Fruits were counted, then removed (>50 m) from die fruit shadow of die tree. I assessed the rate of fruit fall by direct observation. Since feeding by arboreal frugivores (primates, toucans, guans) influences rates of fruitfall, observation periods were subdivided based on their presence or absence in the tree. Data analyses All data used in statistical analyses of foraging were corrected per unit time. Nonparametric Kruskal-Wallis ANOVAs (corrected for ties; Siegel, 1956) were used to identify significant differences in die foraging of different age, sex, and social classes. Where significant differences were found, post-hoc Mann-Whitney {/tests (corrected for ties; Siegel, 1956) were used to compare the foraging of solitary adult males relative to social adult females, solitary adult males relative to social subadults, and social adult females relative to social subadults. A Mann-Whitney {/test was also used to compare foraging on invertebrates by solitary and social adult females. Logistic regression was used to examine the relationship between band size and the results of agonistic interactions between males and band members. Dependent variables (results of encounters expressed as percentages) were transformed using die formula logit (y) = log |(y/100)/[l - (y/ 100)]}. Because logistic regression requires values to fall between 0 and 1 prior to transformation, values of 0% were set at 0.1%. A linear regression line was fitted between logit (y) and group size to identify correlation coefficients and significance values. Gompper • Foraging costs and benefits of coati social structure 257 on invertebrates (Kruskal-Wallis: df = 2; tied H = 1.45; tied p = .485; Table 3). When adult females temporarily left the band and foraged solitarily, their consumption of invertebrates increased significandy (tied z = —2.361; tied p = .018; Foraging Table 3). Nonetheless, foraging of solitary females did not on Foraging differ significandy (tied z = 0.092; tied p = .926; Table 3) Foraging leaf-litter on Sample from the foraging of solitary males. on invertevertebrates size A total of 41.4 h of focal animal samples were collected Month and year fruit (%) brates (%) (%) (n) while individuals were foraging on fruit. Foraging could be subdivided into actual eating and manipulation of fruit, and September 1991 50.0 0 16 50.0 time spent in the seed shadow of the tree searching for fruit. 0 12 October 1991 16.7 83.3 My ability to define searching time-allows foraging success to 0 9 November 1991 44.4 55.6 be quantified in terms of (1) die number of fruit eaten per 36 December 1991 61.1 33.3 5.6* 0 34 January 1992 64.7 35.3 minute, (2) die time spent eating per minute, (3) the time 0 37 February 1992 73.0 27.0 spent searching per minute, and (4) die time spent eating 0 14 March 1992 14.3 85.7 fruit per minute relative to the time spent searching for fruit — — — — April 1992 per minute. ANOVAs revealed significant differences in the 0 29 May 1992 20.7 79.3 foraging success of solitary adult males, adult female band 0 32 June 1992 53.1 46.9 members, and subadult band members for all four measures 0 43 July 1992 74.4 25.6 (Table 4). For all species of fruiting trees combined, foraging — — — — August 1992 success of males was significandy greater than diat of band — — — — September 1992 members (adult females and subadults; column 2 in Table 4). 0 31 October 1992 51.6 48.4 I also examined subsets of the data for foraging only on ScheeJuly 1993 28.1 71.9 0 32 lea or only on Dipteryx. For the smaller patches (fruits not November 1993 0 21.9 78.1 32 abundant under trees) of Scheelea fruit, ANOVAs were again significant for all four measures. In each case, foraging success 1 Includes possible scavenging on an agouti (Dasyprocta pundaia) of solitary males feeding on Scheelea was greater dian diat of carcass. band members. For die larger patches (fruits are abundant under trees) of Dipteryx fruits, ANOVAs were significant for RESULTS all measures except time searching per minute (p = .366; Table 4). Diet and fruit availability Post-hoc analyses of differences among adults and subadults The coatis of BCI are predominately insectivorous and frugivrevealed that, for all fruit foraging data combined, solitary orous. I rarely observed coatis to forage on vertebrates. Of 86 adult males and adult females differed significandy only in scats (24 from solitary males, 66 from band members) exthe number of fruits eaten per minute [p < .0001), although amined throughout the year, 27.9% contained solely invertedie time spent searching per minute approached significance brate matter, 58.1% contained solely fruit matter, and 14.0% (p = .073). For all four measures, both adult males and adult contained both fruit and insect matter. Vertebrate remains females had greater foraging success dian subadults (Table 4). were not found in scats. Direct observations of foraging coatis Thus results of die ANOVAs were primarily due to poor for(Table 1) revealed similar patterns; on average 44.2% were aging success by subadults relative to adults. foraging on leaf-litter invertebrates, 55.4% on fruit, and 0.4% For analyses of only Scheelea fruits, solitary adult males alon vertebrate matter. These proportions varied greatly beways had significandy greater foraging success than both adult tween months (Table 1). No successful predation on mamfemales and subadults. Comparisons of adult females and submals or birds was observed. Both solitary males and band members occasionally ate caecilians (OscaecUia ochrocephala, adults indicated that Scheelea foraging success was greater for n = 3 occasions), snakes (small Spilotes pullalus, n = 2), and adult females, widi the exception of die number of fruits gained per minute, for which tiiere was no significant differfrogs (Bufo marinvs, n = 2). Males also feed on iguana eggs ence (p = .074; Table 4). Thus solitary males feeding on Schee{Iguana iguana, n = 3 clutches). On several occasions band members chased agoutis (Dasyprocta punclata) briefly, but lea have greater foraging success dian adult females, who in turn have greater foraging success than subadults. these presumed predation attempts were unsuccessful. Like die patterns of foraging success on Scheeleafruits,analObservations of individual Scheelea trees showed that on a yses of foraging on Dipteryx indicate that solitary adult males given day the number of ripe fruits available to a terrestrially have significandy greater foraging success dian adult females foraging coati was small, typically less than 100 (Figure 1). for all measures except time searching per minute. However, Peak periods of fruit availability occurred for each tree. Howdifferences between adult males and subadults were only sigever, there was considerable variation and asynchrony in the nificant for die number of fruit eaten per minute (Table 4). period of fruitfall and variation in the quantity of fruit availIn addition, subadults did not differ significandy from adult able among trees during these peak periods (Figure 1). For Dipteryx, I did not collect data on daily variation in terrestrial females in die number of fruits eaten per minute or in die time spent searching per minute, aldiough the ratio of time fruit availability. However, variance similar to that seen in Scheelea (Figure 1) was expected due to similar patterns of eating to time searching and the quantity of time eating per minute were significant (Table 4). animal-induced fruitfall (Table 2). That is, for both tree species, fruit did not fall as it ripened; rather, it was directly or indirectly dropped as a result of the activity of arboreal fruAgonistic interactions givores (Table 2; see also Bonaccorso et al., 1980). Forty agonistic interactions between solitary males and band members were observed. Of these, 55% occurred during forForaging behavior aging on fruit and 12.5% occurred during foraging on invertebrates. Agonistic interactions often occurred when foraging An ANOVA revealed no significant differences in the foraging groups approached and started feeding at fruiting trees, and success of adult males, adult females, and subadults feeding Table 1 Sightings of Nasua narica foraging by food type and by month of observation 258 Behavioral Ecology Vol. 7 No. 3 15 Figure 1 Daily availability of fruits under six ScheeUa zonensis trees dur- ing the period 27 July 1990 to 21 August 1990. Trees were checked twice daily for ripe fruits in the tree shadow that were removed from the area so as not to bias subsequent checks. 2222 2 2 A 2 2 523 55532 ?22522£2555553 DATE then continued when participants chased one another away from these trees. Therefore, the actual proportion of agonistic interactions related to foraging on fruit may be higher, since some of these interactions may have been misclassified. Only two interactions (5%) were observed when a band was not foraging or traveling; these occurred during resting and nursing activity. Of the 22 interactions at fruiting trees, 10 occurred at ScheeUa trees and five at Dipteryx trees, with the remaining interactions occurring under Faramea, Ficus, Eugenia, and Quararibea trees. Foraging group size was important in determining the outcome of agonistic interactions between solitary males and bands (Table 5). Males were more likely to leave an area when foraging group size was larger (logistic regression: r = .89; p = .044). If the male did not immediately leave the area, it was also more likely to be chased from the area by one or more group members as foraging group size increased. Foraging groups of seven or more individuals always chased males that did not leave (Table 5). The probability of a male remaining in the area, or continuing to feed at the fruiting tree, declined with increasing size of the foraging group (r = —.91; p = .032; Table 5). This may be due to the increased chances of a male receiving aggression from band members, and to the possibility of less fruit remaining at the tree due to an increased number of foraging individuals. The latter possibility is suggested by the decreasing relationship between foraging group size and the sum (expressed proportionally) of the number of males remaining plus the number of males returning to the tree after the foraging group has departed (r = —.88; p = .055; Table 5). Males chased band members (Table 6). The chance of diis occurring decreased slighdy with increasing foraging group size, although the regression is not significant (r = —.76; p = .162). However, this relationship may be misleading due to categorization of foraging group size. Of die five chases diat occurred under fruiting trees, four involved only a single for- Gompper • Foraging costs and benefits of coati social structure Table 2 Mean number of fruits dropped per minute observation before, during, and after the presence of large arboreal frugivores (e.g., primates, toucans) for two BCI tree species Table 3 Leaf-litter invertebrate foraging of Nasua narica subdivided by age class and sociality during time of data collection Age and social category Mean Number number focal of items samples eaten/min 3.5 ± 3.5 (0-13)' 0.04 ± 0.2 (0-1) Adult males (solitary) Adult females (social) Adult females (solitary) Subadults (social) 29 32 23 25 57 101 Subadults include individuals 12 to 24 mondis old. 0.6 ± 0.7 (0-2) 0.0 36 9 may be related, at least in part, to body size sexual dimorphism interacting with a reliance on patchy and defendable food resources. Adult male coatis are significantly larger than adult females (Gompper, 1994). Several lines of evidence suggest that this dimorphism acts direcdy and indirectly in the maintenance of the coati social system. First, larger and older individuals generally were winners of conflicts which usually occurred over food patches. Larger males were able to displace solitary females and small foraging groups. In turn, larger groups of smaller individuals (females) displaced large solitary males. Second, smaller individuals may be more vulnerable to predation pressure than large individuals, who may be able to aggressively defend against predators such as ocelots, tayras, Cebus monkeys, raptors, and large snakes (all reported to feed on juvenile and subadult coatis). While no detailed study has been made of body size-based selection of coatis by predators (but see relevant discussion by Karanth and Sunquist, 1995), this supposition does imply that smaller individuals are under increased selection to minimize predau'on pressure via group living which would increase the number of vigilant, detecting eyes and dilute the chances of being chosen for attack (Gitdeman, 1989; Hamilton, 1971; Treisman, 1975). Third, if larger individuals require more food than small individuals, they may also be more detrimentally influenced by living with individuals that are simultaneously competing at food patches. My qualitative observations of the ontogeny of male solitary foraging support this third line of evidence. Subadult males of age 20 to 24 months increase foraging time away from the band, arrive at fruiting trees earlier than other band members, and leave later than other band members, often running to catch up to the traveling band. Yet band members do not appear to increase aggression toward these males as they age. This suggests that males become solitary not because they are driven from the band, but rather for some other reason, such as to increase foraging success. Before vertebrate arrives During vertebrate After vertebrate at tree arboreal feeding leaves tree Dipteryx panamensis Fruit dropped/min 0.0 Observation nme (min) 90 259 0.833 0.565 0.743 0.592 ±SE 0.118 0.047 0.057 0.076 Scheelea zonensis Fruit dropped/min 0.0 Observation nme (min) 60 * Values are means ± SD with range in parentheses. aging group member that remained at the food patch after the remainder of the group departed (Table 6). This suggests that a solitary female may have difficulty gaining access to food patches when faced with competition from larger solitary males. Of 41 agonistic interactions between solitary adult males, 22 occurred in the context of foraging on fruit. In two cases individuals sustained injuries during the conflict; a winner suffered a cut shoulder and a loser a cut ear. I may have missed recording additional injuries, as conflicts were often obscured from observation by forest undergrowth. In 18 conflicts the age or relative age of both males was known. In all cases with an age differential between the males (n = 15), the older male was the winner. To examine the characteristics associated with the outcome of agonistic encounters further, I compared the morphometrics of winners and losers (Table 7). The average winner had a marginally higher (p = .055) mean weight (measured within 6 months of the encounter) than the average loser. A paired comparison of nine interactions in which the weights of both the winner and the loser were known confirmed this pattern (p = .035). However, body weight of an individual may vary significantly within a year(Gompper, 1994; Russell, 1982), and there was no significant difference in body weights of winners and losers when weights were measured within 6 weeks of the conflict For less variable measures, winners were significantly larger than losers for head-body length, but not for testes length (Table 7). DISCUSSION These results suggest that (1) coati social structure directly influences foraging success, and (2) that intraspecific foraging competition may play an important role in maintaining this sexually dimorphic social system. Solitary adult males have greater foraging success than band members when feeding on fruit (Table 4), suggesting that males may remain solitary for foraging benefits rather than due to an inability to join bands. In contrast, although adult females had higher foraging success when feeding solitarily, they were occasionally unable to gain access to valuable food patches without assistance due to direct competition and aggression from solitary males. These observations of agonism also emphasize the importance of body size in structuring the outcome of contested food patch acquisition. In essence, the coati sociality-asociality dichotomy Variance in foraging relating to sociality and direct conflict over food sources suggests that food access plays a greater role in coati social structure than previously stated (Kaufrnann, 1962; Russell, 1982) and differendy than implied (Smythe, 1970a). The paucity of my observations of coati predation on vertebrates is very similar to findings of Kaufrnann (1962) and Russell (1982) who reported only rare predation on mice and lizards. Thus Smythe's hypothesis (1970a) that solitary adult males are more likely to hunt vertebrates than are females or band members and that the hunting of vertebrates is more common in the dry season when fruit availability is low is unsupported. Nonetheless, all studies of the coati on BCI stress the importance of only a few plant species, including Dipleryx and Scheelea. The variance in foraging success between males and band members, and within band age classes, is especially great when Behavioral Ecology Vol. 7 No. 3 260 Table 4 Fruit foraging of solitary males relative to band members Tied x, tied p Bands Males All fruiting tree species (41.43 h) A 1.059 ± 0.479; 132 0.777 0.717; 148 B 32.493 ± 12.835; 128 23.743 13.807; 124 28.251 ± 14.601; 52 C 23.144 ± 12.238; 128 30.313 11.830; 120 27.508 ± 12.846; 52 D 2.690 ± 3.407; 128 1.196 1.296; 120 1.635 ± 1.619; 52 Scheelea (24.07 h) A 1.058 ± 0.467; 88 0.652 ± 0.415; 80 H.p Subadults Adult females 0.764 ± 0.379; 68 1 0.591 ± 0.493; 68 50.603; <.0O01 20.228 ± 12.988; 64 33.105; <.0001 32.612 ± 11.261; 60 24.850; <.0O01 0.879 ± 0.904; 60 30.962; <.0O01 0.681 ± 0.328; 52 0.600 ± 0.544; 28 B 38.915 ± 8.742; 84 29.519 12.856; 64 34.279 ± 10.323; 40 21.586 ± 12.921; 24 C 17.151 ± 8.952; 84 24.737 10.213; 60 22.588 ± 9.195; 40 29.036 ± 11.007; 20 D 3.733 ± 3.799; 84 1.717 1.523; 60 2.064 ± 1.614; 40 1.022 ± 1.047; 20 0.673 ± 0.136; 8 0.594 ± 0.484; 32 25.585; <.0001 31.279; <.0001 18.910; <.0001 26.158; •C.0001 -4.354; <.0001 -1.515; .1297 -1.793; .0730 -1.566; .1174 -6.427; <.0001 -5.687; <.0001 -4.694; <.0001 -5.430; <.0001 2.9697 .0030 -2.844; .0045 -2.195; .0281 -2.755; .0059 -4.426; <.0001 -2.225; .0261 -2.696; .0070 -2.396; .0166 -3.515; .0004 -5.442; •C.OOOl -3.896; <.0001 -4.820; •C.OOOl -1.788; .0737 -3.557; .0004 -2.011; .0443 -3.269; .0011 -2.215; .0268 -2.933; .0034 -1.466; .1426 -2.933; .0034 -3.950; <.0001 -0.591; .5547 -0.394; .6937 -0.591; .5547 -1.631; .1030 -2.174; .0297 -1.087; .2770 -2.174; .0297 Dipteryx (13.69 h) 0.633 ± 0.414; 48 A 0.942 ± 0.350; 36 B 22.627 ± 9.472; 36 C 32.039 ± 7.706; 36 D 0.806 ± 0.491; 36 19.693 33.044 12.281; 48 11.041 ± 5.622; 8 8.990; 48 37.764 ± 1.536; 8 0.761; 48 0.287 ± 0.137; 8 0.787 17.718; .0001 21.141 ± 13.966; 32 7.780; .0204 32.301 ± 10.685; 32 2.010; .3660 0.927 ± 0.884; 32 7.780; .0204 Values for sex and social classes are mean ± SD; sample size. Four measures of foraging are given: (A) number of fruits/min foraging time; (B) time eating/min; (C) time searching/min; (D) time eating per min/time searching per min. H and Rvalues are for a Kruskal-Wallis ANOVA comparing solitary adult males, adult female band members, and subadult band members. Tied z and tied p values are for MannWhitney (/tests comparing (1) solitary adult males and adult female band members, (2) solitary adult males and subadult band members, and (3) adult female and subadult band members. Table 5 Results of encounters between solitary adult males and bands or foraging groups of varying numbers of individuals Group size n 1-3 4-6 7-9 10-12 Mean ± SD Male leaves' Male chased6 Male remains' Male returns'1 0 (0/3) 10.0 (1/10) 27.3 (3/11) 22.2 (2/9) 66.7 (2/3) 33.3 (1/3) 77.8 (7/9) 100 (8/8) 100 (7/7) 100 (1/1) 66.7 50.0 27.3 11.1 (2/3) (5/10) (3/11) (1/9) 0 (0/3) 66.7 60.0 54.5 11.1 0 25.2 ± 25.5 82.2 ± 29.0 31.0 ± 27.4 (2/3) (6/10) (6/11) (1/9) (0/3) the population depends on Scheelea zonensis fruits. While this species is common on BCI, each tree has only a small number of ripe fruits available at any given momenL Solitary individuals are able to monopolize and eventually eat virtually all fruits under a given tree, with foraging success limited primarily by patch size. In contrast, band members are limited Table 6 Characteristics of aggression directed by solitary adult males toward band members during male-band encounters, subdivided by foraging group size 38.5 ± 30.6 Values are percentages, with actual counts in parendieses. " Male leaves area without being chased by band member(s). b Male chased from area by band member (s). Excludes males that left without being chased. ' Male remains in area or is not chased away despite band presence and chase attempts by band members. d Male returns to the site after the band leaves the area. Includes individuals in column five. Foraging group size 1-3 4-6 7-9 10-12 >12 Mean ± SD Time males chased band members during encounters 33.3 20.0 18.2 22.2 0 18.7 (1/3) (2/10) (2/11) (2/9) (0/3) ± 12.0 Band members remaining during aggression Foraging context 1 5; 6 1; 1 1; 12 n/a Dipteryx Leaf litter, Scheelea Ficus, Ficus Scheelea; Leaf litter n/a Gompper • Foraging costs and benefits of coati social structure 261 Table 7 Morphometric measures of males involved in male-male conflicts Winner Loser Unpaired t; p Paired difference Paired t, p Weight' ± 6 weeks Weight ± 6 months Head-body length ± 6 months Testes length ± 6 mondis 5.5 ± 0.33; 15" 5.41 ± 0.67; 9 0.256; .8003 0.38 ± 0.67; 5 1.27; .2738 5.7 ± 0.6; 21 5.2 ± 0.7; 11 1.993; .0554 0.45 ± 0.53; 9 2.54; .0349 119.72 ± 4.42; 16 114.27 ± 3.10; 7 2.942; .0078 4.8 ± 2.77; 4 3.469; .0404 14.39 ± 0.72; 15 13.66 ± 1.21; 7 1.802; .0866 0.70 ± 1.61; 4 0.869; .4486 Analyses exclude measurements made when the individual was a subadult Paired difference is the size of the winner minus the size of the loser. * Weight measured in kilograms; length measured in centimeters. b Values given are means ± SD; sample size. in fruit access not only by patch size, but also by direct competition (group size). For example, if a Scheelea has the mean number of fruits available at any given moment (25.94 ± 34.21 fruits when the number of fruits is >0; 6 trees, n = 68 days; Figure 1) and the mean foraging group size is 7.3 individuals (Gompper, 1994), then the average individual will gain only 3.6 fruits per tree visit However, the variance in this value is likely to be high, especially when the number of fruits is low, since individuals do not arrive at a tree simultaneously, and individuals vary in their ability to secure food. Thus, some individuals may gain few or no fruits. Clark and Mangel (1986) have noted that, in general, the benefits of group foraging for ephemeral patches increase with patch size and with the search time required to locate patches. Foraging on Scheelea and Dipteryx represent opposite ends of the continuum suggested by this relationship. Increased patch size (as seen for Dipteryx relative to Scheelea) leads to decreased foraging differentials between solitary males and band members (Table 4). Nonetheless, some band members continue to forage poorly relative to solitary individuals (Table 4), suggesting one or more nonexclusive phenomena. First, some band members may have increased foraging success relative to other band members due to increased competitive ability (or decreased feeding rates due to satiation or feeding efficiency, see below). For example, adult females have greater fruit foraging success than subadults for three of four Scheelea foraging measures (Table 4). Second, foraging success of some band members may vary due to an additional cost or benefit of group living unrelated to foraging competition. For example, social adult females foraging on invertebrates have foraging success equal to subadults (Table 3). This is contrary to the expectation that older, more experienced individuals should have increased foraging success relative to younger, less experienced individuals. The foraging success of females foraging solitarily is higher dian that of females foraging socially, and not significantly different from that of solitary adult males. Similarly, the foraging success (two of four measures; Table 4) of subadults is greater than that of adult females when feeding on Dipleryxfruits.These results suggest that the lower foraging success of group-foraging adult females is not due to competition from other group members or due to some factor intrinsic to adult females, such as poorer health. Excluding foraging competition, costs of sociality extrinsic to the adult females but intrinsic to the band include increased time for care of juveniles (e.g., vigilance, nursing, grooming) and increased time devoted to social functions (e.g., allogrooming). Russell (1979, 1982) attributed higher dry season foraging success of juveniles (<12 months of age; 2.1 items/min) relative to adult females (1.2 items/min) to less frequent vigilance behavior by juveniles. I have interpreted the differences between sex and age classes as being due to competitive disadvantages. However, an alternative explanation is that the apparent competitive disadvantage is an artifact of some individuals foraging less intensively than other individuals. Thus, adult females may have poorer Scheelea foraging success than solitary males simply because they are able to gain the resources they require with less effort, or because they are more selective in fruit choice. This might also explain patterns of foraging success when feeding on Dipteryx, where differences in the feeding of males and subadults are less extensive than differences between solitary males and adult females. In fact, foraging on Dipteryx by subadults is more efficient than foraging by adult females (Table 4). Dipteryx fruiting overlaps the coati mating period when some solitary adult males join bands and interact nonagonistically (or less aggressively) with band members. They often feed with band members during this time. Female foraging during this time may be more efficient, or females may spend more time foraging solitarily (an observation that may be important in understanding the coati mating system; Gompper, 1994; Gompper and Wayne, 1996). Unfortunately data do not currently exist to allow a comparison of the foraging success of adult males and band members foraging on Dipteryx in the presence of one another. Similarly, solitary males may be more motivated to feed at a patch than females for reasons such as greater body size requiring greater food intake, or perhaps due to lower feeding success between fruit patches. Because invertebrate foraging often occurs while individuals are traveling between fruit patches, and solitary males do not differ from band members in feeding on invertebrates, it is tempting to rule out solitary males having lower feeding success between fruit patches than members. However, information on differences in the daily activity budget of males and females, especially quantification of food obtained within versus between patches, as well as metabolic rates of males and females, is needed. When available, this information would allow assessment of differences in the daily feeding success of solitary and social individuals relative to caloric requirements. I conclude that the size of fruit patches plays a role in structuring the BCI coati societies. Are coati populations at other sites greatly dependent during some portion of the year on species with relatively small fruit crops which might foster intersexual foraging competition? Unfortunately, detailed dietary studies of coatis at other sites have yet to be carried out. However, among primate societies the importance of keystone plant species is well established, and the limited feeding site availability in these plants has been shown to influence social structure (Andelman, 1986; Cant, 1990; Ghiglieri, 1984; Terborgh, 1983). An intriguing extension of these findings is that where reliance on patchy food sources is relaxed, social structure may 262 shift. Observations in Arizona and Panama suggest males may occasionally join bands for brief periods (a few days to a month) outside the breeding season and may also occasionally delay dispersal by up to a year (Gilbert, 1973; Gompper and Krinsley, 1992; Kaufmann et al., 1976; Wallmo and Gallizioli, 1954). These observations suggest that die coati social system is extremely flexible and shaped by an array of ecological and evolutionary factors. I thank J. Gitdeman, J. Kaufmann, M. Mangel, G. Burghardt, C. Boake, A. Hoylman, R. Kays, G. Anderson, and an anonymous reviewer for comments on the manuscript. I also thank S Rand, N. Smythe, and the Smithsonian Tropical Research Institute for generous logistic support in the field. Support for this research came from a National Science Foundation dissertation improvement grant (DEB 9212747), Smithsonian Institution fellowships, a Fulbright Foundation fellowship, Sigma Xi grants in aid of research, an Explorers Club exploration grant, the Theodore Roosevelt Fund of the American Museum of Natural History, and an American Society of Mammalogists grant in aid of research. Additional support came from the Science Alliance, the Department of Zoology of the University of Tennessee, and a predoctoral training grant fellowship from the National Institute of Child Health and Development awarded to the University of Tennessee Graduate Program in Ethology (HD 07303). REFERENCES Alexander RD, 1974. The evolution of social behavior. Annu Rev Ecol Syst 5:325-383. Altmann J, 1974. Observational study of behavior: sampling methods. Behaviour 49:227-267. Altmann SA, 1974. Baboons, space, time, and energy. Am Zool 14: 221-248. Andelman SJ, 1986. Ecological and social determinants of Cercopithecine mating patterns. In: Ecological aspects of social evoluuon (Rubenstein DI, Wrangham RW, edj). Princeton, New Jersey: Princeton University Press; 201-216. Bonaccorso FJ, Glanz WE, Sanford CM, 1980. Feeding assemblages of mammals at fruiting Dipleryx panamensis (Papilionaceae) trees in Panama: seed predation, dispersal, and parasitism. Rev Biol Trop 28:61-72. Burger J, Gochfeld M, 1992 Effect of group size on vigilance while drinking in die coati, Nasua narica, in Costa Rica. Anim Behav 44: 1053-1057. CantJGH, 1990. Feeding ecology of spider monkeys (Ateies geoffroyi) atTikal, Guatemala. Human Evol 5:269-281. Caro TM, 1989. Determinants of asociality in felids. In: Comparative socioecology (Standen V, Foley RA, eds). Oxford: Blackwell Sciendfic; 41-74. Caro TM, 1994. Cheetahs of the Serengeti plains. Chicago: University of Chicago Press. Chapman FM, 1938. Life in an ajr casde. New York: D. AppletonCentury Co. Chausseil M, 1992. Evidence for color vision in procyonides: comparison between diurnal coaus (Nasua) and nocturnal kinkajous (Polos flauus). Anim Learn Behav 20:259-265. Clark CW, Mangel M, 1986. The evolutionary advantages of group foraging. Theor Popul Biol 30:45—75. Croat TB, 1978. Flora of Barro Colorado Island. Stanford, California: Stanford University Press. De Steven D, Windsor DM, Putz FE, DeLeon B, 1987. Vegetative and reproductive phenologies of a palm assemblage in Panama. Biotropica 19:342-356. Enders RK, 1935. Mammalian life histories from Barro Colorado Island, Panama. Bull Mus Comp Zool 78:385-502. Foster RB, 1982. The seasonal rhythm of fruitfall on Barro Colorado Island. In: Ecology of a tropical forest (Leigh EGJr, Rand AS, Windsor DM, eds). Washington, D.C.: Smithsonian Institution Press; 151-172. Foster RB, Brokaw NVL, 1982. Structure and history of the vegetation of Barro Colorado Island In: Ecology of a tropical forest (Leigh EGJr, Rand AS, Windsor DM, eds). Washington, D.C.: Smithsonian Institution Press; 67-81. Behavioral Ecology Vol. 7 No. 3 Gentry AH, 1990. Four neotropical rainforests. New Haven, Connecticut; Yale University Press. Ghiglieri MP, 1984. Feeding ecology and sociality of chimpanzees in Kibale Forest, Uganda. In: Adaptations for foraging in nonhuman primates (Rodman PS, CantJGH, eds). New York: Columbia University Press; 161-194. Gilbert B, 1973. Chulo. New York: Alfred A Knopf. Gitdeman JL, 1989. Carnivore gTOtip living: comparative trends. In: Carnivore behavior, ecology, and evolution (Gitdeman JL, ed). Ithaca, New York: Cornell University Press; 183-207. Gompper ME, 1994. The importance of ecology, behavior, and genetics in the maintenance of coati (Nasua nanca) social structure (PhD dissertation). Knoxville: University of Tennessee. Gompper ME, 1995. Nasua nanca. Mammal Species 487:1-10. Gompper ME, Krinsley JS, 1992. Variation in social behavior of adult male coatis (Nasua narica) in Panama. Biotropica 24:216-219. Gompper ME, Wayne RK, 1996. Genetic relationships among individuals within carnivore societies. In: Carnivore behavior, ecology, and evolution, vol. 2 (Gitdeman JL, ed). Ithaca, New York: Cornell University Press; 429-452. Hamilton WD, 1971. Geometry for the selfish herd. J Theor Biol 31: 295-311. Henderson A, 1995. The palms of the Amazon. New York: Oxford University Press. Karanth KU, Sunquist ME, 1995. Prey selection by tiger, leopard and dhole in tropical forests. J Anim Ecol 64:439-450. Kaufmann JH, 1962. Ecology and the social behavior of the coati, Nasua nanca, on Barro Colorado Island, Panama. Umv Calif Publ Zool 60:95-222. Kaufmann JH, Lanning DV, Poole SE, 1976. Current status and distribution of the coati in the United States. J Mammal 57:621-637. Klein LL, Klein DJ, 1973. Observations on two types of neotropical primate intertaxa associations. Am J Physiol Anihropol 38:649-654. Krinsley JS, 1989. An ethogram of the white-nosed coati (Nasua nasua nanca) (MS thesis). Madison: University' of Wisconsin. Leigh EGJr, Rand AS, Windsor DM (eds), 1982. The ecology of a tropical forest Washington, D.C.: Smithsonian Institution Press. Leigh EGJr, Windsor DM, 1982. Forest production and regulation of primary consumers on Barro Colorado Island. In: The ecology of a tropical forest (Leigh EGJr, Rand AS, Windsor DM, eds). Washington, D.C.: Smithsonian Institution Press; 111-122. Leighton M, Leighton DR, 1982. The relationship of size of feeding aggregate to size of food patch: howler monkeys (Alouatta paUiata) feeding in Trichilia cipo fruit trees on Barro Colorado Island. Biotropica 14:81-90. Milton K, 1980. The foraging strategy of howler monkeys. New York: Columbia University Press. Packer C, Scheel D, Pusey AE, 1990. Why lions form groups: food is not enough. Am Nat 136:1-19. Piperno DR, 1990. Fitolitos, arqueologia y cambios prehistoricos de la vegetacion en un lote de cincuenta hectareas de la isla de Barro Colorado. In: Ecologia de un bosque tropical (Leigh EGJr, Rand AS, Windsor DM, eds). Washington, D.C.: Smithsonian Institution Press; 153-156. Pulliam HR, Caraco T, 1984. Living in groups: is there an optima] group size? In: Behavioral ecology: an evolutionary approach (KrebsJR, Davies NB, eds). Sunderland: Sinauer; 122-147. Russell JK, 1979. Reciprocity in the social behavior of coatis (Nasua nanca) (PhD dissertation). Chapel Hill: University of North Carolina. Russell JK, 1981. Exclusion of adult male coatis from social groups: protection from predation. J Mammal 62:206-208. Russell JK, 1982. Timing of reproduction by coatis (Nasua narica) in relation to fluctuations in food resources. In: The ecology of a tropical forest (Leigh EGJr, Rand AS, Windsor DM, eds). Washington, D.C.: Smithsonian Institution Press; 413—431. Russell JK, 1983. Altruism in coati bands: nepotism or reciprocity'? In: Social behavior of female vertebrates (Wasser SK, ed). New York: Academic Press; 263-290. Siegel S, 1956. Nonparametric statistics for the behavioral sciences. New York: McGraw-Hill. Smythe N, 1970a. The adaptive value of the social organization of the coati (Nasua narica).] Mammal 51:818-820. Smythe \', 197flh Relationships between fruiting seasons and seed dispersal methods in a neotropical forest. Am Nat 104:25-35. Gompper • Foraging costs and benefits of coati social structure Smythe N, Glanz WE, Leigh EG Jr, 1982. Population regulation in some terrestrial frugivores. In: The ecology of a tropical forest (Leigh EG Jr, Rand AS, Windsor DM, eds). Washington, D.C.: Smithsonian Institution Press; 227-238. Terborgh J, 1983. Five new world primates. Princeton, New Jersey: Princeton University Press. Treisman M, 1975. Predation and the evolution of gregariousness. I. Models for concealment and evasion. Anim Behav 23:779—800. 263 Wallmo OC, Gallizioli S, 1954. Status of the coati in Arizona. J Mammal 35:48-54. Wright SJ, 1990. Cumulative satiation of a seed predator over the fruiting season of its host. Oikos 58:272-276. Wright SJ, Gompper ME, DeLeon B, 1994. Are large predators keystone species in neotropical forests? The evidence from Barro Colorado Island. Oikos 71:279—294.