OUP user menu

Sleeping-site preferences of wild Japanese macaques (Macaca fuscata): the importance of nonpredatory factors

Yamato Tsuji
DOI: http://dx.doi.org/10.1644/11-MAMM-A-095.1 1261-1269 First published online: 14 December 2011


I investigated sleeping-site preference in habituated wild Japanese macaques (Macaca fuscata) for each season (254 days total) on predator-free Kinkazan Island, northern Japan, during 2000–2007. I focused on the effects of nonpredatory, environmental factors (vegetation type, altitude, and topography), to which little attention has been paid. Macaques used 24–79 sleeping sites in each season (227 sites in total, all on the ground). The frequencies of sleeping sites in each season followed a Poisson distribution, except for spring when several sites were used repeatedly. In spring macaques preferred sleeping in Zoysia japonica grassland, where several staple food species (Berberis thunbergii and Zelkova serrata) are abundant in this season. In summer and fall macaques avoided sleeping in high-altitude forest dominated by Fagus spp., and in the latter season they also preferred Zoysia grassland; these preferences likely reflect an avoidance of strong winds rather than the lower food availability at higher altitudes. In winter macaques avoided sleeping in Zoysia grassland, mainly due to the poor food supply. Macaques preferred valleys to ridges in spring and winter, possibly due to greater densities of shelters such as rocks and fallen trees that facilitate energy conservation in the face of strong/cold winds at night. Additional quantitative data for other mammalian species are needed for generalizations to be made about the importance of nonpredatory factors on sleeping-site preferences.

Key words
  • altitude
  • Kinkazan Island
  • sleeping site
  • topography
  • vegetation

Animals spend approximately half of their lives sleeping, and this important behavior therefore requires active management (Anderson 1998,2000). Because animals are unprotected from external stimuli such as physical stresses and predators during sleep, their selection of appropriate sleeping sites is likely subject to considerable selective pressures (Chapman 1989; Heymann 1995). The extent of concealment (Garcia and Braza 1993; Heymann 1995), height above the ground (Di Bitetti et al. 2000), and cover from above (Caine et al. 1992; Di Bitetti et al. 2000) are known to influence sleeping-site selection. Similarly, animals select sleeping sites in trees overhanging water (Matsuda et al. 2008; Ramakrishnan and Coss 2001), on steep cliffs (Hamilton 1982), in shallow caves (Hayashi 1969; Huang et al. 2003), and on top of buildings (Ramakrishnan and Coss 2001), all of which contribute to safety from predators. However, predation is not the only factor that can affect sleeping-site preferences.

Several studies have shown that animals prefer to sleep where energy costs during sleep are minimized (Aquino and Encarnación 1986). For example, mouse lemurs, Microcebus murinus, increase the amount of energy they conserve through hypometabolism during their daily torpor by an additional 5% by occupying tree holes (Schmid 1998). Other studies have shown that animals use sleeping sites at which staple foods are abundant and energy is readily available to compensate for energy losses incurred during sleep; e.g., spider monkeys (Ateles geoffroyiChapman 1989), common marmosets (Callithrix jacchus—Mendes Pontes and Soares 2005), gray-brown mouse lemurs (Microcebus griseorufusGenin 2010), and chimpanzees (Pan troglodytesFuruichi and Hashimoto 2004). These studies indicate that nonpredatory, environmental factors that contribute to energy conservation can be important determinants of sleeping-site preferences.

The actual use of sleeping sites most likely is determined by trade-offs between the costs and benefits of several competing factors, which ideally should be considered simultaneously. Nevertheless, few studies have focused on factors unrelated to predation (Sugardjito 1983). Several researchers have provided post hoc considerations related to nonpredatory factors after evaluating the predatory factors (Caine et al. 1992; von Hippel 1998; Zhang 1995). Thus, our current understanding of the effects of nonpredatory factors on sleeping-site preference is insufficient.

Japanese macaques, Macaca fuscata, inhabit the northernmost regions among extant nonhuman primates. In contrast to some nocturnal strepsirrhines and the great apes, but like all other Old World monkeys, macaque species do not build nests, relying instead on structures in their habitats for sleeping sites (Anderson 2000). The habitat of Japanese macaques shows great seasonality, and many facets of their ecology, such as feeding and ranging patterns, correspond to these marked changes in the food environment (Hanya 2004; Tsuji and Takatsuki 2004; Tsuji 2010). Thus, it is plausible that they prefer to sleep at locations with the highest concentration of their staple food items. Another possibility related to environmental seasonality is that they avoid sleeping at sites in which the wind chill is high (Hill et al. 2004). Generally, winds blow more strongly at higher altitudes, on ridges, and in forests with greater gap fractions, like grasslands (Kondo et al. 2002; Saito 1996). In terms of energy conservation, such places seem to be disadvantageous. The energy intake and energy balance of macaques worsen in winter (Nakagawa 1997; Tsuji 2010; Tsuji et al. 2008), and low temperatures or strong winds at night lead to increased energy costs related to thermoregulation (Nakayama et al. 1971). The appropriate selection of sleeping sites therefore can contribute to energy conservation, especially during cold seasons. Because the macaques inhabiting Kinkazan Island (hereafter Kinkazan) have not been subjected to predation pressures since at least the 1960s, when studies on the island began, examining the effects of nonpredatory factors should be easier in this population.

The aim of this study was to investigate the importance of nonpredatory factors to sleeping-site preferences exhibited by Japanese macaques on Kinkazan. I conducted the following two-tiered analysis: at the habitat level I asked whether macaques preferred locations in specific vegetation type(s) where staple foods are abundant (i.e., forest versus grassland); and at the microhabitat level I asked whether macaques preferred valleys to ridges. In addition, I tested whether these preferences varied seasonally.

Materials and Methods

Study site.— Kinkazan (38°16′N, 141°35′E) is located 700 m off the Oshika Peninsula, northern Japan. The island is 5.1 km long and 3.7 km wide, undulating over a total area of approximately 9.6 km2 with a highest peak at 445 m. The mean annual rainfall is about 1,500 mm. The mean annual temperature is 11.4°C, being highest in summer (23.5°C in August) and lowest in winter (0.5°C in January; Fig. 1). Because of its oceanic climate the island is seldom covered with snow (Nakagawa 1989). Wind velocity is relatively lower (<4 m·s−1) from June to October and higher (>4 m·s−1) from November to May (Fig. 1). The wind typically blows from the southeast from May to August and from the northwest in other months (Ishinomaki Weather Station; http://www.data.kishou.go.jp). On the basis of mean temperatures and wind velocity, I separated the year into 4 distinct seasons: spring (March– May), summer (June–August), fall (September–November), and winter (December–February).

Fig. 1

Monthly mean temperature and wind velocity on Enoshima Island, 10 km north of Kinkazan Island (2000–2005). Data were obtained from Ishinomaki Weather Station (http://www.data.kishou.go.jp). The year was separated into 4 seasons shown by gray lines.

Study animals.—Six troops (troops A, B1, B2, C1, C2, and D) of wild Japanese macaques inhabiting Kinkazan have had no predators since at least the 1960s when the study started (Izawa 2009). Perhaps due to their lack of predators macaques of all groups usually sleep on the ground rather than in trees (Izawa 1999; Takahashi 1997). Because other populations subject to predation sleep in trees (Izawa 1972, 1982; Wada et al. 2007), the unique ground-sleeping habit of the macaques on Kinkazan implies a possible release from predation pressure. I focused on troop A in this study, a habituated troop that lives in the northwestern part of the island (Tsuji and Takatsuki 2009). Troop A has been habituated since 1982 without provisioning and could be observed from close distances. The troop size varied from 29 to 39 individuals during the study period, including 2–5 adult males (>5 years old), 14–17 adult females (>5 years old), 8–9 juveniles (1– 5 years old), and 1–11 infants (<1 year old). Almost the entire range of troop A overlaps with the ranges of troops B1, B2, and C2 (Izawa 2009). Intertroop encounters sometimes occur, although most are less agonistic than those of macaques in other habitats (Saito et al. 1998; Sugiura et al. 2000).

Home range and sleeping sites.—I observed the macaques of troop A from dawn to dusk on 254 days during 35 intermittently conducted field surveys between May 2000 and November 2007. No observations were conducted on rainy days. Over this period sunrise and sunset occurred at 0415– 0653 h and 1614–1904 h, respectively (Ishinomaki Weather Station; http//:www.data.kishou.go.jp). During observations the location of troop A, represented by the central part of the troop, was recorded every 10 min by a global positioning system (IPS-5100; Sony, Tokyo, Japan; and GPSMAP 60CSx; Garmin International, Inc., Olathe, Kansas). Locations of the macaques were superimposed on a map divided into 1-ha quadrats. This quadrat size was considered appropriate because macaques on Kinkazan usually are distributed within an area <100 m in diameter (Sugiura et al., in press), an area smaller than the estimated quadrat size used here. In total, location data were plotted in 374 quadrats during the study period and defined the annual home range of troop A (Fig. 2a). Quadrats in which the troop as a whole ceased daily movements and formed sleeping clusters in the evening were defined as sleeping sites. The total number of recorded sleeping sites during the study period was 227. Research methods complied with guidelines of the American Society of Mammalogists (Sikes et al. 2011) and adhered to Japanese legal requirements.

Fig. 2

a) Annual home range of troop A and its vegetation types. Seasonal home ranges (gray area) and sleeping sites in b) spring, c) summer, d) fall, and e) winter. Contours show 100-m intervals. Bold lines represent shoreline. The area of each quadrat is 1 ha (note that parts beyond the shoreline were not included in the analysis). Filled circles indicate sleeping sites. The percentage of sampled days each sleeping site was used by macaques is represented by its size.

Vegetation types.—In accordance with Izawa and Komuro (1993) quadrats constituting the annual home range were categorized into 7 vegetation types (next paragraph; Fig. 2a). If a quadrat contained more than 1 vegetation type, it was categorized according to the dominant vegetation type (i.e., the one that covered the most surface area).

Carpinus forest is mixed coniferous and broadleaf forest of Carpinus spp. (Betulaceae, including C. tschonoskii and C. laxiflord) and Abies firma (Pinaceae). Macaques feed on the leaves (spring) and nuts (fall) of Zelkova serrata (Ulmaceae), a tall tree species that grows at high densities in the valleys of this forest (Tsuji and Takatsuki 2004). Pirns forest is coniferous forest consisting of Pinus densiflora (Pmaceae).

Fagus forest is deciduous broadleaf forest of Fagus crenata (Fagaceae). Macaques consume the nuts of this tree mainly in fall (Tsuji and Takatsuki 2009; Tsuji 2010). Illicium forest is evergreen forest of Illicium anisatum (Illiciaceae). Zoysia grassland is grassland of Zoysia japonica (Gramineae). In the spring macaques feed on the leaves and flowers of the shrub tree species Berberis thunbergii (Berberidaceae), which grows at high densities in this grassland (Tsuji and Takatsuki 2004). Miscanthus grassland is grassland of Miscanthus sinensis (Gramineae). Cryptomeria forest is forest of planted Cryptomeria japonica (Taxodiaceae).

Altitude and topography.—Using a map, I recorded the altitude (with an accuracy of 10 m) and topography (ridge or valley) at the center of each quadrat constituting the annual home range. A ridge was defined as a quadrat in which the altitude was > 10 m higher than that of 2 adjacent cell quadrats with the same slope direction.

Survey of food availability and protected locations.—To evaluate the availability of each food type I conducted a survey of shrub plants. From November 2003 to November 2004 I made 76 100-m2 quadrats at ground level (Carpinus forest: n = 14; Pinus forest: n = 16; Fagus forest: n = 8; Illicium forest: n = 7; Zoysia grassland: n =10; Miscanthus grassland: n = 14; and Cryptomeria forest: n = 7) and counted the number of 4 important shrub tree species within these quadrats: B. thunbergii, Viburnum dilatatum (Caprifoliaceae), Rosa multiflora (Rosaceae), and Zanthoxylum piperitum (Rutaceae). Macaques spend much time feeding on the leaves and flowers of B. thunbergii in spring (18–29% of their total feeding time), on the fruits of berries of V. dilatatum and R. multiflora in fall (14–36%), and on the thorny bark of Z piperitum in winter and spring (4–15%), respectively (Tsuji and Takatsuki 2004; Tsuji et al. 2006). The densities of these plant species therefore seemed to affect the foraging locations of the macaques before sleep.

To evaluate the availability of herbaceous plants during winter, the season in which the macaques feed intensively upon such items (Tsuji and Takatsuki 2004; Tsuji et al. 2006), I conducted a biomass survey from December 2004 to February 2005. I set 14 0.25-m2 quadrats at ground level (2 quadrats for each vegetation type) and harvested the aboveground parts of herbaceous plants on which macaques feed within these quadrats. Collected plants were dried at 70°C for 24 h and weighed.

To evaluate the density of protected locations I prepared 63 20- × 50-m line transects (1,000 m2; Carpinus forest: n = 23, Pinus forest: n = 5, Fagus forest: n = 20, Illicium forest: n = 3, Zoysia grassland: n = 4, Miscanthus grassland: n = 6, and Cryptomeria forest: n = 2) and counted all rocks and fallen trees >1 m in height and >2 m in width. These habitat features provide cover for macaques, and multiple individuals can sleep together when sheltered by rocks of such size (Takahashi 1997). For Carpinus and Fagus forest, both of which had larger areas and included both ridges and valleys (Fig. 2a), I selected transects incorporating both topographies {Carpinus forest ridge: n = 8, Carpinus forest valley: n = 15; Fagus forest ridge: n = 11, Fagus forest valley: n = 9) so that the effect of topography on densities of protected locations could be addressed.

Statistical analyses.—Using the variance-to-mean ratio (FSokal and Rohlf 1994), I tested whether the pattern of sleeping-site use in a given season followed a Poisson distribution. If the observed F-value, which represents the variance-to-mean ratio for sleeping sites used in each quadrat, was significantly >1, the observed use of sleeping sites is concentrated at specific locations. Next, I investigated the nonpredatory factors influencing habitat-level sleeping-site preference. As both biotic and abiotic factors might be involved in the choice of sleeping sites, and all variables potentially can interact (e.g., Fagus forest is located at higher altitudes; Fig. 2a), I applied principal component analysis (PCA) to account for the variables for vegetation type and altitude. I then examined the effects of principal components for vegetation types and altitude, and topography (ridge or valley) on the sleeping-site preference using generalized linear models (GLM) with a Poisson error distribution (Crawly 2005). Models of best fit were determined by removing independent variables that did not improve the Akaike information criterion compared with that of the full model (Crawly 2005). All data analyses in this study were carried out using the statistical software R version 2.9.1 (R Developmental Core Team 2009). Effects of neighboring troops were not included in the analyses because frequency of intertroop encounters on Kinkazan are quite low and less aggressive than those in other habitats (Saito et al. 1998; Sugiura et al. 2000) due to their larger home range (Maruhashi et al. 1998), and thus the effects of adjacent groups on sleeping-site preference seemed minimal. Finally, I investigated the nonpredatory factors influencing sleeping-site preference at the microhabitat level. I used a Kruskal–Wallis nonparametric analysis of variance (H) to compare the densities of food availability and protected locations, using Scheffe's method for post hoc analysis (Sokal and Rohlf 1994). Significance levels (a) were set at 0.05 for each test.


Seasonal sleeping sites.—In spring macaques used 25 sleeping sites over 35 nights. Four quadrats were selected as sleeping sites twice, 1 quadrat was used 3 times, and another quadrat was used 5 times. Many sleeping sites were located at lower altitudes, <200 m (22 of 25 sites; Fig. 2b). Sleeping sites did not follow a Poisson distribution (F34,∞ = 1.77, P < 0.001), meaning that the macaques used several sites more intensively than others.

In summer macaques used 24 sleeping sites over 26 nights, 3 of which were used on 2 separate occasions. Similar to the spring, all sleeping sites were located at altitudes <200 m but were distributed in the western part of the home range at this season (Fig. 2c). Use of the sleeping sites followed a Poisson distribution (F25,∞ = 1.05, P > 0.05), suggesting that each quadrat was used randomly as a sleeping site.

In fall macaques used 79 sleeping sites over 97 nights, 12 of which were used twice with 1 being used 3 times. In contrast to spring and summer, the macaques chose their sleeping sites at a wide range of altitudes, from sea level to 300 m (Fig. 2d). Use of sleeping sites followed the Poisson distribution (F96,∞ = 1.14, P >0.05), and thus quadrats were used randomly as sleeping sites.

In winter macaques used 61 sleeping sites over 69 nights, with 5 being used twice and 1 being used on 3 occasions. Similar to fall, the macaques chose their sleeping sites over a wide range of altitudes from the seashore to 300 m (Fig. 2e). Use of the sleeping sites followed a Poisson distribution (F68,∞ = 1.05, P > 0.05), suggesting that quadrats were used randomly as sleeping sites.

Seasonal change in habitat-level sleeping-site preferences.— The first principal component (PC1), related to vegetation characteristics and altitude (Table 1), accounted for 26.1% of the variance, whereas PC2 accounted for 16.6%, PC3 for 14.9%, and PC4 for 14.0%. The cumulative percentage of the variance accounted for by PC1 through PC4 was >70%, and I therefore used these 4 components as independent variables in the GLM. Of these, PC1 was interactive (i.e., Fagus forest is located at higher altitudes; Fig. 2a) and was thus designated as high-altitude forest. Similarly, PC2 was designated as coastal forest because the non-Carpinus, Pinus forest is located along the seashore (Fig. 2a). In contrast, PC3 and PC4 were not strongly interactive, and according to their loading I simply designated each component as Miscanthus grassland (PC3) and Zoysia grassland (PC4), respectively (Table 1).

View this table:
Table 1.

Principal component loadings for 8 variables about vegetation and altitude within the annual home range of troop A macaques on Kinkazan Island, northern Japan. Heavily loaded factors appear in bold.

VariablesPrincipal components
Vegetation types
Pinus forest−0.520.52−0.27−0.58
Carpinus forest−0.30−0.94−0.14−0.06
Zoysia grassland−0.320.29−0.030.87
Fagus forest0.870.23−0.390.01
Illicium forest−
Miscanthus grassland0.090.120.97−0.14
Cryptomeria forest−
Percentage of variance26.116.614.914.0
Cumulative percentage of variance26.142.757.671.6
Habitat designationHigh-altitude forestCoastal forestMiscanthus grasslandZoysia grassland

The GLM suggests that topography (negative effect) and PC4 (positive effect) best explain sleeping-site preferences during the spring (Table 2). Quadrats containing Zoysia grassland or in valleys had higher probabilities of both being selected as sleeping sites and being used repeatedly. In summer the only explanatory variable in the model of best fit was PC1 (negative effect; Table 2), suggesting that quadrats located at lower altitudes were used as sleeping sites more frequently. In fall the explanatory variables in the best model were PC1 (negative effect) and PC4 (positive effect; Table 2). Quadrats located at lower altitudes and those in Zoysia grassland were used more frequently as sleeping sites. In winter the GLM of best fit showed that topography (negative effect) and PC4 (negative effect) best explained variation in the model (Table 2). Quadrats located in the valley or in non-Zoysia grassland were used most frequently as sleeping sites.

View this table:
Table 2.

Best models, based on Akaike information criterion (AIC—Crawly 2005), for explaining sleeping-site selection by Japanese macaques by season, and estimates of parameters (± SE). n = number of quadrats constituting the home range in a given season. * P < 0.05, ** P < 0.01, *** P < 0.001.

Parameters (±SE)Spring (n = 162), AIC = 180.04Summer (n = 138), AIC = 131.17Fall (n = 332), AIC = 451.25Winter (n = 329), AIC = 360.20
Intercept−1.76 ± 0.24***−2.26 ± 0.34***−1,30 ± 0.11***−1.46 ±0.13***
Topography (ridge)−1.84 ± 1.02−0.92 ± 0.43*
Principal component (PC) 1 (high-altitude forest)&#x2013;−0.81 ± 0.25**−0.24 ± 0.08**&#x2013;
PC4 (Zoysia grassland)0.41 ± 0.12***0.14 ± 0.08−0.22 ± 0.13

Food availability.—Becmse only PC1 and PC4 were selected as explanatory variables in the best-fit models for each season, I compared the abundance of food resources among the 3 relevant vegetation types: Zoysia grassland, Fagus forest, and other (Fig. 3). Among the 4 shrub tree species only B. thunbergii showed a significant difference in tree density among the 3 vegetation types (H = 19.86, P < 0.001; Fig. 3a). Furthermore, the tree density of B. thunbergii in Fagus forest was significantly lower than that in other vegetation types (P < 0.05 for each case). Conversely, I found no significant differences in tree density across vegetation types for the other 3 shrub species: Zanthoxylum piperitum: H = 2.67, P = 0.264 (Fig. 3b); V. dilatatum: H = 3.54, P = 0.170 (Fig. 3c); and R. multiflora: H = 2.35, P = 0.308 (Fig. 3d). In contrast, the dry weights of herbaceous plants showed significant difference across vegetation types (H2 = 7.81, P = 0.020; Fig. 3e). The dry weights of these herbaceous plants in Fagus forest were significantly greater than in the others (P < 0.05).

Fig. 3

Mean (± SD) tree numbers of the 4 shrub tree species, and the dry weight of herbaceous plants in winter, within quadrats representing 3 vegetation types. Number of sampled quadrats per vegetation types is shown in parentheses. * P < 0.05, *** P < 0.001.

Protected location availability.—Similar to food availability, I compared the density of both rocks and fallen trees across the 3 vegetation types: Zoysia grassland, Fagus forest, and other (represented by Carpinus forest; Fig. 4). Rock density was not significantly different among vegetation types (H = 2.88, P = 0.236; Fig. 4a). However, rock densities were significantly greater in valleys than along ridges in both Fagus forest (H = 9.45, P = 0.002) and Carpinus forest (H = 8.67, P = 0.003).

Fig. 4

Mean (± SD) numbers of rocks and fallen trees observed in line transects in 3 vegetation types. Fagus forest and Others vegetation types are further divided into ridges and valleys. The number of line transects is shown in parentheses. ** P < 0.01, * P < 0.05. NS = P > 0.05.

The density of fallen trees did not differ significantly with vegetation type (H = 1.13, P = 0.567; Fig. 4b), although like rocks, fallen tree densities were significantly greater in valleys than along ridges in Carpinus forest (H = 4.14, P = 0.042). However, I found no difference in rock densities in relation to topographical features in Fagus forest (H = 2.58, P = 0.108).


I demonstrated the impacts of various nonpredatory factors on the habitat-level sleeping-site preferences of Japanese macaques on Kinkazan Island. Site selection varied seasonally, but only during spring did macaques exhibit strong preferences for certain locations, as evidenced by the aggregated distribution of sleeping sites used during this season. My analysis suggests that macaques prefer sleeping sites located in Zoysia grassland during spring, and this might indicate a preference for sleeping in areas containing several staple foods, which are found in abundance within such grassland. Their staple foods in spring consisted of the leaves and flowers of B. thunbergii (combined 18–29% of total feeding time) and leaves of Zelkova serrata (10–37%—Tsuji and Takatsuki 2004). Although the density of B. thunbergii in Zoysia grassland did not differ from other vegetation types, Zelkova serrata trees are distributed densely within Zoysia grassland (Tsuji and Takatsuki 2004). Thus, macaques might prefer such areas in spring because of their ability to use the more abundant food resources therein. The repeated use of several sleeping sites in Zoysia grassland during this season supports this possibility. Such staple food-oriented sleeping-site selection has been reported for other primate species (spider monkeys, Ateles geoffroyiChapman 1989; chimpanzees, Pan troglodytesFuraichi and Hashimoto 2004; common marmosets, Callithrix jacchus—Mendes Pontes and Soares 2005).

Despite the preference for Zoysia grassland in spring, macaques actually avoided such areas in winter. The staple foods of macaques during this season consisted mainly of herbaceous vegetation (Tsuji 2010; Tsuji et al. 2006). Because my results indicate no differences in the dry weights of such herbaceous plants collected from different vegetation types, avoidance of Zoysia grassland cannot be attributed to a lack of staple food availability in such areas. Similarly, except for B. thunbergii, no variation was seen across vegetation types in the tree densities of the main shrub species, whose fruits, berries, and nuts are eaten by macaques during summer and fall (Tsuji 2010; Tsuji et al. 2006). As a result, unlike in spring, the observed sleeping-site preferences expressed in other seasons could not be attributed to the availability of staple food items across specific vegetation type(s). Other factors, therefore, must be considered.

Macaques avoided sleeping in high-altitude forest (corresponding to Fagus forest) during summer and fall. In addition to macaques not sleeping in Zoysia grassland during winter, this may point to another environmental factor, wind strength, as a potential variable influencing sleeping-site selection. In general, winds are stronger at higher altitudes and in forest with greater gap fractions, such as grassland areas (Kondo et al. 2002; Saito 1996). Therefore, sleeping at high altitudes in summer and fall and in Zoysia grassland during winter, where wind strength is the greatest, should be disadvantageous in terms of thermoregulatory costs. Although still speculative, this notion is supported further by macaques' preference for valleys to ridges in both spring and winter, although no such preferences were observed during summer and fall. Winds are stronger along ridges than in valleys (Kondo et al. 2001), and this study also showed more protected locations such as rocks and fallen trees in valleys than in ridges. Therefore, the use of valleys as sleeping sites should minimize the risk of being exposed to strong winds. These results suggest that, at least in several seasons, macaques select sleeping sites to reduce thermoregulatory costs. Because both energy intake and energy balance of macaques are lowest from winter to early spring (Nakagawa 1997; Tsuji 2010; Tsuji et al. 2008), and low temperatures and strong winds in these seasons increase the energy costs of maintaining a constant body temperature (Nakayama et al. 1971), an appropriate preference for sleeping sites should contribute to energy conservation, especially during this period. Plasticity in sleeping-site preference, as with huddling behavior (Takahashi 1997; Vessey 1973), sunbathing (Huang et al. 2003), and alternations in ranging altitudes (Liu and Zhao 2004; Wada and Ichiki 1980), thus might reflect an energy-saving tactic of macaques inhabiting environments that exhibit strong seasonality.

With the exception of spring, the macaques on Kinkazan chose their sleeping sites at random across the quadrats constructed within their home range. This result differed from studies of other macaque species that have been shown to use repeatedly only a limited number of sleeping sites (bonnet macaques, Macaca radiataRamakrishnan and Coss 2001; Thibetan macaques, M. thibetanaOgawa and Takahashi 2003). Because adherence to specific sleeping sites often has been considered an antipredator tactic (Cui et al. 2006; Di Bitetti et al. 2000; Matsuda et al. 2008; Reichard 1998), release from predation pressure might have caused the more random use of sleeping sites seen in this study.

Although the significance of nonpredatory environmental factors on sleeping-site preference has been addressed previously (Anderson 1998, 2000), few studies have examined such effects with appropriate empirical data. This study indicates the necessity of further considering nonpredatory environmental factors in influencing sleeping-site selection. Some researchers have reported that such habitat-level variables as vegetation type are not as useful in predicting sleeping sites as are microlevel topographical features and specific tree characteristics (Hankerson et al. 2007). However, I found that both habitat- and microhabitat-level variables were important explanatory factors for the sleeping-site preferences of Japanese macaques. The conclusions of Hankerson et al. (2007) might have been biased by their separate analyses of variables at different levels. This study, alternatively, suggests that multivariate analyses such as PCA can account for the potential interplay between environmental variables and thus explain more adequately sleeping-site preferences under natural conditions (Liu and Zhao 2004).


I thank Drs. K. Izawa, S. Takatsuki, and K. Matsubayashi for the use of their facilities during my study and Drs. H. Sugiura, T. Furuichi, and A. J. J. Macintosh for their constructive comments on an earlier version of this manuscript. This study complied with the guidelines of the Primate Research Institute, Kyoto University, Japan. Part of this study was supported financially by the Cooperative Research Fund of the Primate Research Institute, Kyoto University, and the Environment Research and Technology Development (D-1007) of the Ministry of the Environment, Japan.


  • Associate Editor was David S. Jacobs.

Literature Cited

View Abstract