American Journal of Primatology 21:17-29 (1990) Choice of Food Patches by Japanese Monkeys (Macaca fuscata) NAOFUMI NAKAGAWA Primate Research Institute, Kyoto University Kanrin, Znuyama, Aichi 484,Japan Food patch choice was investigated in the A-troop of wild Japanese monkeys (macaca fuscata) of Kinkazan Island, Japan. The monkeys visited a Zelkova serrata tree 16 times and fed on its seeds for 43.5 min on average during the 36 day study period. The proportion of fallen seeds to total fallen seeds, the feeding speed, and the number of monkeys feeding on seeds in the crown and on the ground were recorded to clarify the decision factors involved in food-patch choice: when the monkeys visit the tree, do they feed on seeds in the crown or on the ground? The monkeys appeared to the patch quality by feeding speed; they chose a high-quality patch and fed there. As a result, they seemed to maximize seed intake. However, some troop members (in many cases, low-ranking ones) avoided feeding in a high-quality but crowded patch. When the quality of the two patches was equal, the monkeys separated into two groups (in the crown and on the ground) in a ratio of 1:l to feed. As the difference in patch quality became larger, more monkeys fed in the high-quality patch. The frequency of agonistic interactions per individual increased on the ground as the number of monkeys feeding there increased. When the cost of agonistic interactions is considered, the net benefit (energy intake minus energy consumption from agonistic interactions) per individual in a higher-quality but crowded patch may be equal to that in a lower-quality but uncrowded patch. Key words: food patch choice, feeding speed, food patch quality, dominance r a n k INTRODUCTION Food patch choice is one of the important problems to which optimal foraging theory has been applied [Pyke et al., 19771. The problem is to ascertain how an individual chooses and utilizes high-quality patches among the many food patches available, and how it maximizes energy intake. Such choices are complicated by competition among group members for food in group-living animals [for details, see Krebs & Davies, 19871. Several authors have discussed this problem in primates [Alexander, 1974; Janson, 1988a; Janson & Schaik, 19881, and have presented Received for publication April 12, 1989; revision accepted January 11, 1990. Address reprint requests to Naofumi Nakagawa, Primate Research Institute, Kyoto University, Kanrin, Inuyama, Aichi 484, Japan. 0 1990 Wiley-Liss, Inc. 18 I Nakagawa relevant quantitative data [e.g., Slatkin & Hausfater, 1976; Dittus, 1977; Post et al., 1980; Whitten, 1983,1988; Schaik et al., 1983; Janson, 1985,1988b; Iwamoto, 1987; Chapman, 1988; Isabirye-Basuta, 1988; Symington, 19881. Japanese monkeys maintain long inter-individual distances during ranging and feeding, and usually feed alone, thereby avoiding feeding competition [Mori, 1977; Maruhashi, 19861. These studies suggest that both ecological (patch quality) and social factors (presence of other group members competing for food in the patch, and their social status) affect a monkey's choice of food patch. Ihobe [19891concluded that monkeys choose a food patch independent of social relationships because they can feed without entering the intolerance feeding space of other individuals when food is abundant. However, few studies have quantified food-patch quality and related such measures to choice of patch. This study relates food-patch quality to a monkey's choice of patch, an addresses the influence of social factors on choice of patch. This kind of study is difficult in primates under natural conditions for the following reasons. 1) It is technically difficult to quantify the quality of a food patch. 2) It is difficult to judge whether a monkey knows of the existence of a patch when he does not choose it. 3) It is difficult to assess the costs of travelling to each patch when the quality of patches located far from each other is compared. 4)It is impossible to assess patch quality for different food items by feeding speed alone because each food item differs in its nutritional content. Wild Japanese monkeys in the A-troop that inhabits Kinkazan Island in Japan were chosen as subjects for this study. A big Zelkova serrata tree that produced a large amount of fruit and that the A-troop often visited was chosen and regarded as one food patch. Whether a monkey fed on zelkova seeds in the crown or on the ground was examined when it visited this zelkova-seed patch (named All. The patch of seed in the crown (Crown patch) and the patch on the ground (Ground patch) are regarded as different patches when reference is made to the choice of food patch in Al. In this study, the four problems mentioned above were resolved as follows. 1)Patch quality was assessed by the density of seeds and the speed of feeding on seeds in the Crown and Ground patches. 2) All members of the A-troop knew the location of Al. 3) When a monkey visited Al, the cost of travelling t o the Crown patch was considered to be negligible. 4) Patch quality could be assessed simply by feeding speed because the monkeys fed on the same food items in both the Crown and Ground patches. While the A-troop utilized A1 repeatedly during the study period, the quality of the Crown and Ground patches gradually changed over days because the seeds fell to the ground. Thus, data were obtained on change in patch quality and on the monkey's choice of patches in A l . In the present case, optimal foraging theory predicts that under competition-free conditions all troop members choose and utilize the higher-quality one of the Ground and Crown patches. However, if there is feeding competition among troop members which reduces feeding speed, then an alternative patch allowing for the same feeding speed without competition becomes equally attractive to other members. At equilibrium, troop members separate and feed so that the net benefit per individual in a feeding bout is equalized, as predicted by an ideal free distribution [Fretwell & Lucas, 1970; Fretwell, 19721. METHOD Kinkazan Island (Fig. 1)lies about 700 m offshore a t its shortest distance from the Oshika Peninsula, Miyagi Prefecture, Japan (38" 16", 141" 35'E). The island is 5.1 km long and 3.7 km wide. The total area is about 10 km2. The highest peak is 445 m above sea level. The mean annual temperature is 11°C and the mean Food Patch Choice by Japanese Monkeys / 19 I 0 1000 M Fig. 1. A map of Kinkazan Island and its location. The area enclosed by dotted lines shows the home range of A-t.roopfrom October 16 to December 1, 1985 (see also Fig. 2). 20 I Nakagawa annual rainfall is about 1,500 mm. This island is seldom covered with snow, although snow sometimes falls in winter. The vertical distribution of vegetation on the island can be divided roughly into three zones: a Fagus crenata zone a t more than 200 m above sea level; a n A bies firma zone at less than 200 m above sea level; and a Pinus thunbergii zone around the coast. Besides these zones, several patches of grassland are distributed on the island, and these are dominated by Zoysia japonica, Miscanthus sinensis, and Pteridium aquilinum [Yoshii & Yishioka, 1949; Takatsuki, 19801. The 2. serrata trees considered in this study grow in the flat and moist areas of the valleys or alluvial fans belonging to the Abies firma zone a t 100-200 m above sea level or to a transition zone between the Fagus crenata zone and the Abies firma zone. Therefore, the region occupied by the Zelkova serrata community is restricted to a small area of the island. The A-troop habituated by Sato, Izawa, and others was chosen as the subject of this study. In December 1985, the A-troop consisted of 25 animals: 3 adult males, 10 adult females, 2 young females, 3 juveniles, and 7 infants. The study period extended from October 16 to December 1, 1985. The method of data collection was t h a t of focal-animal sampling [Altmann, 19741. The total focal-animal sampling time was 346 hours 4 min (36 days). As a rule, one individual was followed all day long. Five adult females were chosen as focal animals. The focal animal was changed daily. Data were recorded a s follows. Activities were divided into 4 categories: feeding, moving, resting, and social activities, and were recorded in seconds. The food items were also recorded during feeding. When the duration of a feeding bout (as described below) on zelkova seeds exceeded 5 min, the feeding speed (as described below) was recorded. However, when monkeys feeding both in the Crown patch and in the Ground patch were observed during the feeding of the focal animal, the feeding speed of the monkeys that were feeding in a different patch from the patch where the focal animal was feeding was also recorded. The zelkova seed patches in which the mean duration of feeding bouts exceeded 5 min were marked with vinyl numbering tape and their locations were recorded accurately on a map. In addition, while the focal animal fed in A l , the names of all the monkeys, except the infants, and their locations in A1 (Crown patch or Ground patch) were recorded by the 5 min interval scanning method [Altmann, 19741. The times at which agonistic interactions occurred and the participants were also recorded. The terms used in this paper are defined as follows. Feeding bout: Continuous feeding on the same food item for more than 1 min. However, when the interval between 2 continuous feeding periods, each on the same food item and each lasting for more than 1 min, was less than 2 rnin and when the total moving time in this interval was less than 1min, these 2 continuous feedings were regarded a s one feeding bout [Nakagawa, 1989al. Food patch: One tree where the focal animal was observed feeding was usually regarded as one food patch. However, when zelkova seeds were consumed on the ground and a specific food patch could not be identified because of the continuous distribution of seeds, the site within a radius of 15 m from the trunk was regarded as one food patch, for convenience. Feeding speed: One minute units were randomly set up in a feeding bout, and mean feeding speed was estimated by averaging the number of seeds eaten per 1 min unit. The number of fallen seeds under A1 was measured by using a seed trap (100 x 80 cm2 in area). It was installed on October 21 a t about 5 m from the trunk of A l , a tree that the A-troop had visited twice during the previous 5 days. The trap Food Patch Choice by Japanese Monkeys 1 21 was made of fine-mesh nylon net, with vinyl chloride pipes supporting the net at the 4 corners. It was covered with 2 cm mesh wire-netting on 4 sides except for an opening extending 5 cm in height above ground level. The net was covered with 4 cm mesh wire-netting on the top. The wire-netting covered the trap on the top and the sides to prevent the monkeys from eating the seeds on the net and beneath the trap, respectively. The 5 cm opening allowed wild mice to eat the seeds beneath the trap. When the seed trap was installed, the seeds beneath the trap were counted. Thereafter, the seeds in the trap were counted and replaced beneath the trap another six times, a t intervals of about a week, during the study period. On December 2, when the observations were finished, five small quadrats (20 x 20 cm2 in area) were set up at a distance of 7 m, on average, from the trunk of Al, and the number of remaining seeds in them was counted. Four hundred seeds (in the crown of Al, on October 161, 400 seeds (on the ground under A95, on December lo), and 300 seeds (on the ground under Al, on December 10) were collected to examine the percentage of sound zelkova seeds. RESULTS Nutritional Value and Feeding Technique of Zelkova Seeds Zelkova seeds are kidney-shaped and are about 4 mm long. Two to 4 seeds are found in a single axilla. All the zelkova seeds in both Crown and Ground patches are considered to be mature during the study period, since zelkova seeds usually mature in late October [Hashizume & Aikawa, 19771. The percentages of sound seeds (i.e., seeds whose embryos developed normally and were not damaged by insects) were 47% (in the samples collected in the crown of Al, on October 16),41% (in those on the ground under Al, on December lo), and 43% (in those on the ground under A95, on December 10). The percentage of sound seeds, thus, was 40-50%, but the monkeys apparently did not eat only sound seeds. The average dry weights of a sound zelkova seed and a non-sound zelkova seed were, respectively, 0.0121 g and 0.0067 g. Their protein contents were, respectively, 20.25% and 4.73% on a dry weight (dw) basis, and their caloric contents were 4.87 kcallg dw and 4.00 kcalig dw (Nakagawa, unpublished data). As judged from the nutritional content per unit weight, sound zelkova seeds are higher-quality food items than many other food items in the diet of these monkeys [cf. Nakagawa, 1989bl. In the crown, the monkeys pulled twigs towards themselves and ate seeds one by one directly by mouth without handling them. On the ground, they sat down, pushing the fallen leaves aside, and picked up and ate seeds one by one using both hands in turn. Percentage of Time Spent Feeding on Zelkova Seeds and Distribution of Zelkova-Seed Patches Zelkova seeds were one of the main food items during the study period and accounted for 25% of the total time spent feeding. The average duration of feeding bouts on zelkova seeds was 24.1 min (s.d. = 33.0, n = 109). Figure 2 shows the distribution of the main zelkova-seed patches (those in which the mean duration of feeding bouts exceeded 5 min) and the percentage of time spent feeding in each patch. There were 18 main zelkova-seed patches during the study period. The total time spent feeding in these 18 main patches occupied 88%of the total time spent feeding on zelkova seeds. In particular, A1 was visited 16 times repeatedly by the A-troop and the proportion of time spent feeding in A1 occupied 41% of the total time spent feeding on zelkova seeds. The duration of feeding bouts in A1 was 43.5 min on average (s.d. = 43.2, n = 23). A1 was the most 22 I Nakagawa I 0 Fig. 2. Distribution of main zelkova-seed patches (solid circles) and the percentage of time spent feeding in each patch (vertical bars in the map). A main food patch is a patch where the duration of feeding bouts exceeded 5 min. Two of 18 main food patches are not located on the map. The area enclosed by solid lines shows the home range of the A-troop from October 16 to December 1, 1985. important zelkova-seed patch, although the focal animals did not visit A1 but intensively utilized torreya-seed patches from October 27 to November 8. Changes Over Time in the Percentage of Fallen Zelkova Seeds Some 2,773 seeds fell in the seed trap (100 x 80 cm2 in area) by December 2, including the 41 seeds found beneath the trap on October 21. On March 24, 1986, when I visited the island again, only 17 additional seeds had fallen into the trap. Thus, 2,790 (2,773 plus 17) was considered to be the total number of fallen seeds. Figure 3 shows the change over time in the number of fallen seeds. This estimate does not reflect total amount of seed production because it neglects seeds eaten in the Crown patch. However, it shows roughly the change over days in the quality of the Crown and Ground patches making comprehensible the change of feeding speed shown in Figure 5. Although few seeds had fallen at the beginning of the study, seeds gradually fell to the ground with the passage of days. Around November 9, the percentage of fallen seeds was about 50%of the total fallen seeds. Almost all the seeds had fallen (the percentage of fallen seeds was 99.4%) by December 2 at the end of the study. Although 2,682 seeds were expected beneath the trap on December 2, there were only 1,890 seeds. As the portion beneath the trap was enclosed with 2 cm mesh wire-netting, it is impossible for the monkeys to have eaten these seeds. It is improbable that rain washed the seeds out since A1 is in a flat area. Therefore, this decrease in seeds was probably the result of feeding by Japanese wood mice (Apo- Food Patch Choice by Japanese Monkeys / 23 (1599) i501 Oct.16 / / 20 25 Nov.1 5 10 15 20 25 Dec.1 Mar.24 Fig. 3. Changes in the proportion of fallen seeds to total seeds and of the number of fallen seeds (in parentheses) in a zelkova-seed patch (Al). The values are based on the number of fallen seeds in a seed trap (100 x 80 cm2 in area). : Proportion of fallen seeds to total seeds. - - - - -: Proportion of fallen seeds eaten by wild mice. : Proportion of fallen seeds eaten by monkeys and wild mice. demus argenteus) which inhabit this island [Ohta, 19671 and entered beneath the trap through the 5 cm opening. On December 2, the numbers of remaining seeds in 5 small quadrats (20 x 20 cm2) were 87, 71, 70, 109, and 85 with an average of 84.4 When this value was converted to that for an area of 100 x 80 cm2,it amounted to 1,688; 202 seeds per 100 x 80 cm2 (1,890 minus 1,688) seemed to have been eaten by the monkeys in the Ground patch. This number may include some seeds eaten by small birds, such as Oriental Greenfinch (Carduelis sinica). Changes in Feeding Speed Over Days Figure 4 shows the decrease over time in feeding speed in the Crown and Ground patches. At the beginning of the observations, feeding speed in the Crown patch was very high, since few seeds had yet fallen from the tree. However, the speed decreased as the seeds fell (see also Table Ia). The speed decreased to the same value as that in the Ground patch when about 50%of seeds had fallen and the number of seeds in both patches was equal a t the beginning of November, even if the decrease in seeds as a result of the monkeys’ and mice’s feeding is taken into account. After that, feeding speed in the Crown patch decreased progressively as the seeds fell. By contrast, the feeding speed in the Ground patch remained almost constant after it became equal to that in the Crown patch, despite the increasing number of fallen seeds (see also Table Ib). This result seems to be due to the following factors. The density of fallen seeds was so high that feeding speed converged to a maximum in the Ground patch. This result also implies that the monkeys must spend a fixed amount of time in picking up seeds one by one, with both hands, even if they do not spend any time searching for a seed. They could feed at higher speeds in the Crown patch as the density of seeds in the Crown patch was initially higher. The time required for handling a seed is short in the Crown patch because here monkeys take the seeds directly by mouth without using their hands. Changes Over Time in the Number of Monkeys Feeding in the Crown Patch and the Ground Patch When the A-troop visited A l , 15.0 monkeys on average (s.d. = 3.9, n = 16) (excluding infants) fed on the seeds during the focal animal’s feeding there. The 24 I Nakagawa 40 1 0 0 0 -0 0 m v: 25 a 0 , Oct.16 I I 20 25 I Nov.1 1 1 5 10 20 15 1 4 25 Dec. 1 Fig. 4. Changes in feeding speed in the Crown patch (solid circles) and in the Ground patch (open circles) of a zelkova-seed patch (All. Feeding speed in the Crown patch showed a significant correlation over time (r = -0.918, n = 8, P = 0.0013)and the regression equation is as follows: Y = -0.33X + 38.67 (X = number of days passed Y = feeding speed). This correlation was not significant in the Ground patch (r = -0.369, n = 7, P = 0.4158). a a a a 7 3 Oci.Ih 20 25 No". 1 5 10 15 25 Dec. 1 Fig. 5. Changes in the percentage of monkeys feeding in the Crown patch (individual-minutes: if one monkey feeds for 1 min, this value is 1 individual-min. average number of monkeys feeding in A1 at the same time was 12.7 individuals1 scan (s.d. = 4.0, n = 252) (excluding infants). These values include the numbers of non-troop males who often fed in A1 with the members of A-troop. Figure 5 , which is based on scan samples during the focal animal's feeding in Al, shows the decrease over time in the percentage of monkeys feeding in the Crown patch. When feeding speed in the Crown patch was very high because of the low proportion of fallen seeds a t the beginning of the study, almost all the monkeys fed in the Crown patch. However, when the percentage of fallen seeds increased to about 50% and the feeding speed in the Crown patch decreased to the same level as that in the Ground patch, the number of monkeys feeding in the Crown patch and that in the Ground patch approached a ratio of 1: l. After that, the proportion of monkeys feeding in the Crown patch gradually decreased with feeding speed in the Crown patch. Finally, almost all the monkeys fed in the Ground patch, when the percentage of fallen seeds exceeded 95% (see also Table Ic-f). Food Patch Choice by Japanese Monkeys I 25 TABLE I. Correlation Among Various Variables (the Number of Seeds, Feeding Speed, Proportion of Monkeys Feeding in Each Patch) Dependent variables a) FSC b) FSG c) %MC d) %MG e ) %MC f) %MG r 0.904* -0.508 0.802** 0.805** 0.727** -0.366 P 0.0021 0.2439 0.0003 0.0003 0.0021 0.4196 Independent variables" #SCb #SGb #SCb #SGb FSC" FSG "#SC: Number of seeds in the Crown patch. #SG number of seeds in the ground patch. FSC: Feeding speed in the Crown patch. FSG: Feeding speed in the Ground patch. %MC: % of monkeys feeding in the Crown patch. %MG % of monkeys feeding in the Ground patch. bThis number was estimated from Figure 3. 'As these values were not measured on 7 observation days, they were estimated by the regression equation shown in Figure 4. *: P < 0.05; **: P < 0.01. Figure 6 shows the changes during feeding bouts in the percentage of feeding monkeys that were in the Crown patch. Three long feeding bouts were chosen as examples during the period when monkeys fed in both the Crown and Ground patches. In every case, the proportion of monkeys feeding in the Crown patch remained almost constant during a feeding bout. Frequency of Agonistic Interactions The total numbers of monkey-hours spent feeding in Al, measured by the 5 min-interval scanning method, were 181 and 84 individual-hours (excluding infants) in the Crown patch and the Ground patch, respectively. Twelve and 40 agonistic interactions occurred in the Crown patch and the Ground patch, respectively, during this period including 4 interactions between adults and infants in the Ground patch. The frequency of agonistic interactions in which the monkeys, excluding infants, participated was 0.13 and 0.90 per individual-hour in the Crown patch and the Ground patch respectively. Agonistic interactions include only those that involved both physical contact and chasing. Difficulties in observing the interactions in the Crown patch may have influenced the low frequency of agonistic interactions recorded there. DISCUSSION Assessment of the Quality of the Food Patch The monkeys could probably not assess visually the difference in quality between the Crown patch and the Ground patch because the zelkova seeds were covered with fallen leaves on the ground. Only twice did one focal animal feed for more than 1 min in both the Crown and Ground patches during one visit to Al. However, monkeys other than the focal animals often fed for more than 1 min in both the Crown and Ground patches during the focal animals' feeding in Al. Such cases were observed 46 times in total (in 17 troop members excluding a-female, "Hera"). Monkeys other than the focal animals might have fed in both patches when the focal animal was not feeding in Al. The A-troop might have done so when the monkeys were not observed during the study period. Incidentally, the A-troop is sure to have fed in A1 on November 17,25, and 26 out of the 11days when the monkeys were not observed during the study period. Given these data, it appears that the monkeys assess the quality of these patches in terms of their feeding speed 26 I Nakagawa . , 0- 50 ( m i n ) 0 0 , 0 3 I I I , , 50 L a p s e of t i m e 0 4 , L a p s e of t i m e I I 50 L a p s e of t i m e I I I , , , , 100 (min) I 100 (min) Fig. 6. Change in the percentage of the number of monkeys feeding in the Crown patch i n each scan during a feeding bout. a: Date and time: 12 November, 1985;8:04:32-82455 and 8:36:40-912:ZO. b Date and time: 12 November, 1985; 9:21:08-11:0446. c: Date and time: 1 5 November, 1985; 6:22:02-6:28:15 and 6:31:15-8:07:55. while they feed in each patch. Consequently, they choose the higher-quality patch (i.e., where the feeding speed is higher), and they seem to maximize seed intake. Choice of Food Patch There are various costs and benefits of group-living. One of the important potential costs of group-living is competition for food [for details, see Krebs & Davies, 19871. In group-living primates, feeding competition has been demonstrated by the following facts: 1) The larger the group size, the shorter is the duration of feeding bout [e.g., Slatkin & Hausfater, 1976; Janson, 1988bl. 2) The larger the group size, the greater is the foraging effort le.g., Schaik et al., 1983; Chapman, 1988; Isabirye-Basuta, 1988; Janson, 198813; Schaik & Noordwijk, 1988; Symington, 19881. 3) The higher the dominance rank of monkeys, the larger is their food intake [e.g., Whitten, 1983; Janson, 19851,the longer their feeding bouts last [e.g., Post et al., 1980; Iwamoto, 19871, and the higher is their efficiency of foraging [Dittus, 19771. Moreover, the results of feeding competition influence mortality [Dittus, 19771 and birth rate of adult females [Whitten, 19831. I examined whether the monkeys feeding in the Crown patch were lowranking monkeys as a result of feeding competition when the quality in the Crown patch became lower than that in the Ground patch. Of the monkeys feeding in each scan, those in the top half of the dominance rank order were regarded as highranking and those in the lower half were regarded as low-ranking. When there was an odd number of monkeys feeding, the middle-ranking monkey was included with the low-ranking ones. The distribution of monkeys feeding in the Crown patch between low and high rank was tested across all sample days with lower crown patch quality (i.e., on or after November 12). Low-ranking monkeys fed significantly more often in the Crown patch than high-ranking monkeys (x2 = 24.849, P Food Patch Choice by Japanese Monkeys I 27 < 0.01, df = 1).As a result, seed intake of high-ranking monkeys may be larger than that of low-ranking ones. On the other hand, the frequency of agonistic interactions per individual significantly increased in the Ground patch as the number of monkeys feeding there increased (Kendall's rank correlation, tau = 0.50, n = 16, P < 0.011, although the correlation in the Crown patch was not significant (tau = 0.23, n = 16, P > 0.05). Klein  observed that when a subgroup of Ateles belzebuth fed in a tree, all members were dispersed evenly throughout the canopy, and when one member moved t o a different location, the other members tended to maintain an equal spacing. The cost of repeatedly maintaining personal space would increase dramatically in a large subgroup. Waser 119771 found that aggression increased in small patches as the size of groups of Cercocebus albigena increased. If the frequency of agonistic interactions (including moderate interactions such as supplanting) increased, as stated above, because of increases in the number of monkeys feeding in the higher-quality patch, the cost of agonistic interactions may be higher in higher-quality patches. As a result, the net benefit (energy intake minus energy consumption from agonistic interactions) in a feeding bout may decrease as the number of monkeys feeding in a patch increases. Therefore, the net benefit per individual in a higherquality but crowded patch (in many cases, high-ranking) may be equal to that in a lower-quality but uncrowded patch (in many cases, low-ranking). Thus, it is possible that troop members separate and feed so that the net benefit per individual in a feeding bout is equalized, as predicted by an ideal free distribution (Fretwell & Lucas, 1970; Fretwell, 1972). CONCLUSIONS 1. The troop of Japanese monkeys visited a big Zelkova serrata tree 16 times during the study period. They fed on its seeds in the crown or on the ground for 43.5 min on average. It appeared that they assessed patch quality by feeding speed. 2. The monkeys preferred to feed in the Crown patch or in the Ground patch, depending on which patch was of higher quality. As a result, they seemed to maximize seed intake. 3. However, some troop members (in many cases, low-ranking ones) avoided feeding in a high-quality but crowded patch. 4. When the higher cost of agonistic interactions in a crowded patch is considered, the net benefit per individual in a higher-quality but crowded patch may be equal to that in a lower-quality but uncrowded patch. ACKNOWLEDGMENTS I would like to express my deepest thanks to Prof. Y. Sugiyama of the Primate Research Institute, Kyoto University, for invaluable suggestions about this manuscript. I also wish to thank Dr. K. Izawa of Miyagi University of Education and Ms. S. Sato of Yamagata University for offering information given prior to my study and for their invaluable help. Special thanks are accorded to Dr. H. Ohsawa of the Primate Research Institute, Kyoto University, for his critical reading and correction of this manuscript. I am grateful to the members of the socio-ecology seminar of the Primate Research Institute, Kyoto University, for their instructive comments on this study. I was provided with excellent facilities for my field work by the staff of Kinkazan Koganeyama Shrine, especially by the Chief Priest M. Okumi. This study was financed in part by the Cooperative Research Fund of the 28 I Nakagawa Primate Research Institute, Kyoto University, and also by a Grant-in-Aid for Special Project Research on Biological Aspects of Optimal Strategy and Social Structure from the Ministry of Education, Science and Culture, Japan. REFERENCES Alexander, R.D. The evolution of social be- Klein, L.L. The ecology and social organizahavior. ANNUAL REVIEW OF ECOLtion of the spider monkeys (Atetes belzeOGY AND SYSTEMATICS 5:325-383, buth). Ph.D. Dissertation. University of California, Berkeley, 1972. 1974. Altmann, J . Observational study of behav- Krebs, J.R.; Davies, N.B. AN INTRODUCior: sampling method. BEHAVIOUR 49: TION TO BEHAVIOURAL ECOLOGY 227-265, 1974. (Second Edition). Oxford, Blackwell ScienChapman, C. Patch use and patch depletion tific Publications, 1987. by the Spider and Howling monkeys of Maruhashi, T. Feeding ecology of Japanese Santa Rosa National Park, Costa Rica. BEmonkeys in Yakushima Island. Pp. 13-59 HAVIOUR 150:99-116, 1988. in THE WILD JAPANESE MONKEYS Dittus, W. The social regulation of populaON YAKUSHIMA ISLAND. T. Marution density and age-sex distribution in the hashi, J. Yamagiwa, T. Furuichi eds. Totoque monkey. BEHAVIOUR 63:281-322, kyo, Tokai Daigaku Shuppankai, 1986 (in 1977. Japanese). Fretwell, S.D. POPULATIONS IN A SEA- Mori, A. Intra-troop spacing mechanism of SONAL ENVIRONMENT. Princeton, the wild Japanese monkeys of the Koshima Princeton University Press, 1972. troop. PRIMATES 18:331-357, 1977. Fretwell, S.D.; Lucas, Jr., H.L. On territorial Nakagawa, N. Feeding strategies of Japabehavior and other factors influencing nese monkeys against the deterioration of habitat distribution in birds. ACTA BJOhabitat quality PRIMATES 3O:l-16, THEORETICA 19:16-36, 1970. 1989a. Hashizume. H.: Aikawa, T. Flowering and Nakagawa, N. Bioenergetics of Japanese fruit bearing i n Zelkovu serrata. NIHONmonkeys (Macaca fuscata) on Kinkazan IsRINGAKKAI-HAPPY0-RONBUNSHU land during winter. PRIMATES 30: 88:199-200, 1977. (in Japanese) 441-460, 198913. Ihobe, H. How social relationships influence a monkey's choice of feeding sites in the Ohta, K. Small mammal survey on Kinkazan Island, Miyagi Prefecture in 1966. ANtroop of Japanese macaques (Macaca f u s NUAL REPORT OF JIBP-CT 5:184-188, cata) on Koshima Islet. PRIMATES 30:171967 (in Japanese). 25, 1989. Isabirye-Basuta, G. Food competition among Post, D.G.; Hausfater, G.; McCuskey, S.A. Feeding behavior of Yellow baboons (Papio individuals in a free-ranging chimpanzee cynocephatus): relationship to age, gender community in Kibale Forest, Uganda. BEand dominance rank. FOLIA PRIMATOHAVIOUR 105~135-147, 1988. LOICA 34:170-195, 1980. Iwamoto, T. Feeding strategies of primates Pyke, G.H.; Pulliam, H.R.; Charnov, E.L. in relation to social status. Pp. 243-252 in Optimal foraging: a selective review of theANIMAL SOCIETIES: THEORIES AND ory and tests. QUARTERLY REVIEW OF FACTS. Y. Ito, J.L. Brown, J. Kikkawa BIOLOGY 52:137-154, 1977. eds. Tokyo, Japan Scientific Societies Schaik, C.P. van; Noordwijk, M.A. van. Press, 1987. Scramble and contest in feeding competiJanson, C.H. Aggressive competition and intion among female Long-tailed macaques dividual food consumption in wild brown (Macaca fascicularis). BEHAVIOUR 105: capuchin monkeys (Cebus apella). BE77-98, 1988. HAVIORAL ECOLOGY AND SOCIOBISchaik, C.P. van; Noordwijk, M.A. van; OLOGY 18:125-138, 1985. Boer, R.J. de; Tonkelaar, I. den. The effect Janson, C.H. Intra-specific food competition of group size on time budgets and social and primate social structure: a synthesis. behaviour in wild long-tailed macaques BEHAVIOUR 105:1-17,1988a. (Macaca fascicutaris); BEHAVIORAL Janson, C.H. Food competition in brown caECOLOGY AND SOCIOBIOLOGY 13: puchin monkeys (Cebus apella): quantita173-181, 1983. tive effects of group size and tree producSlatkin, M.; Hausfater, G. A note on the activity. BEHAVIOUR 105:53-76, 1988b. tivities of solitary male baboon. PRIJanson, C.H.; Schaik, C.P. van. Recognizing MATES 17:311-322, 1976. the many faces of primate food competiSymington, M. McFarland. Food competition: methods. BEHAVIOUR 105:165-186, tion and foraging party size in the Black 1988. Food I?atch Choice by Japanese Monkeys / 29 spider monkey (Ateles paniscus chamek). BEHAVIOUR 105:117-134,1988. Takatsuki, S. Food habits of Sika deer on Kinkazan Island. THE SCIENCE REPORTS OF THE TOHOKU UNIVERSITY, FOURTH SERIES, BIOLOGY 38:7-31, 1980. Waser, P.M. Feeding, ranging and group size in the mangabey (Cercocebus albigena). Pp. 183-222 in PRIMATE ECOLOGY. T.H. Clutton-Brock ed. New York, Academic Press, 1977. Whitten, P.L. Diet and dominance among female vervet monkeys (Cercopithecus aethiOPS). AMERICAN JOURNAL OF PRIMATOLOGY 5139-159, 1983. Whitten, P.L. Effects of patch quality and feeding subgroup size on feeding success in vervet monkeys (Cercopithecus aethiops), BEHAVIOUR 105:35-52,1988. Yshii, Y.; Yoshioka, K. Plant communities of Kinkazan Island. ECOLOGICAL REVIEW (SENDAI) 12:84-105, 1949 (in Japanese).