Habitat use of the white‐headed langurs in limestone forest of Southwest Guangxi, China: Seasonality and group size effects

Abstract Understanding how animals cope with habitat‐specific environmental factors can assist in species conservation management. We studied the habitat use of four groups (two large and two small groups) of white‐headed langurs (Trachypithecus leucocephalus) living in the forest of southwest Guangxi, China between September 2016 and February 2017 via instantaneous scan sampling. Our results showed that the langurs primarily used hillsides (55.91% ± 6.47%), followed by cliffs (29.70% ± 5.48%), hilltops (7.26% ± 3.55%), flat zones (6.99% ± 6.58%), and farmlands (0.14% ± 0.28%). The langurs moved most frequently on hillsides (49.35% ± 6.97%) and cliffs (35.60% ± 9.17%). The hillsides were more frequently used (66.94% ± 7.86%) during feeding, and the langurs increased the use of hilltops during the rainy season, and the use of cliffs in the dry season. The langurs frequently rested on hillsides (49.75% ± 8.16%) and cliffs (38.93% ± 8.02%). The larger langur group used cliffs more frequently when moving and resting, whereas the small langur group used hillsides more frequently while resting. Langurs in all groups avoided the flat zones for feeding. Their use of habitat reflected the balancing of foraging needs, thermoregulation, and predator avoidance. We conclude that the ecological factors are determinants of habitat use for white‐headed langurs. Our findings suggest that conservation efforts should focus on protecting the vegetation on the hillsides and restoring the vegetation on the flat zones.

Food resource availability significantly impacts habitat use in primates (Camaratta et al., 2017;Nagy-Reis & Setz, 2017;Pennec et al., 2020). In general, food resources are spatiotemporally distributed, and primates prefer to forage in food-abundant areas (Hongo et al., 2017;Ni et al., 2018). However, primates may choose to forage in more extensive home ranges to obtain adequate food intake when food resource availability declines (Nagy-Reis & Setz, 2017).
Normally, primates stay vigilant as much as possible, and several activities (i.e., resting and grooming) could decrease activity time spent on vigilance, forcing them to perform these activities in safe locations (Cords, 1995;Cowlishaw, 1997a;Maestripieri, 1993).
Accordingly, primates prefer resting in tall trees or cliffs, which is easier for them to detect and/or avoid predator (Enstam & Isbell, 2004;Li et al., 2021). For example, resting Angolan colobus monkeys (Colobus angolensis ruwenzorii) stay in larger trees and at higher canopy heights rather than at lower canopy heights to decrease predation risk (Adams & Teichroeb, 2020).
Food availability influences primate group sizes (Chapman et al., 1995;Wrangham et al., 1993), and group sizes significantly affects travel distance/time as a reflection of intragroup feeding competition (Chapman, 1990;Liu et al., 2013;Zhang et al., 2020). Specifically, larger groups have greater energy requirements than smaller groups, making the former spend more time traveling and feeding to harvest adequate foods (Wang et al., 2021). Predation is another factor affecting primates' group sizes, and larger groups could decrease predation risk as they have an increase in predator detection and defense than smaller groups (Hill & Lee, 1998;Matsumoto-Oda et al., 2018;Suscke et al., 2017). Consequently, for those primates inhabiting in smaller forest fragments than those inhabiting continuous forests (Ni et al., 2018), the larger primate groups are more likely to forage in food-rich and high-risk patches than smaller ones to obtain adequate foods (Adams & Teichroeb, 2020;Albani et al., 2020). In the South Sulawesi karst forest, larger groups of moor macaques (Macaca Maura) use food-rich and high-predation risk patches more frequently than small groups, which prefer areas with lower food availability but lower predation risk (Albani et al., 2020).
Fragmentation affects primate behaviors, habitat use, and foraging options Mekonnen et al., 2018). The forest structure and composition in fragments may be altered, which possibly lead to the plant species diversity and abundance reduced (Arroyo-Rodríguez & Mandujano, 2006;Mekonnen et al., 2017). As a result, primates living in small and isolated forest fragments may face challenges caused by the reduced food availability and smaller homerange size Mekonnen et al., 2018). Furthermore, primates living in fragments are likely to accept more anthropogenic disturbance and may avoid to utilize areas where human occur frequently (Waterman et al., 2019).
White-headed langurs (Trachypithecus leucocephalus) from southwest Guangxi, China, are distributed in a 200-km 2 area composed of degraded and fragmented forests (Huang, 2002;. The small sizes of the fragments provide limited food species for these individuals . White-headed langurs are extremely folivorous (Huang, 2002;Li et al., 2003) and prefer young leaves and fruits (Lu et al., 2021). However, the distribution of vegetation in limestone forests significantly differ in hill zones Li & Rogers, 2006). According to the topography and the vegetation of the region, limestone hills can be roughly divided into five zones: hilltop, cliff, hillside, flat zone, and farmland (more details can be seen in the methods section) (Li & Rogers, 2006). Briefly, hilltops and cliffs have sparsely distributed vegetation and many bare rocks. The hillsides are covered by thick vegetation and lay below the cliffs. The flat zones have small amounts of plants, lying at the bottoms of the hills. The farmland is the part of flat zones and is used as agricultural areas. Moreover, since the variaitons in rainfall (with a distinct dry season and a rainy season), young leaves and fruits are seasonally abundant and unevenly distributed in the limestone forest Li et al., 2016;Zhou et al., 2006). Furthermore, group size makes different nutritional requirements for langur groups which do not have same group size (Zhang et al., 2020) Previous studies have showed that the habitat use of limestone-living primates are usually associated with the distribution of food resource and predation risk (Huang et al., 2000;Li et al., 2021;Zhou et al., 2013).
Terrestrial carnivores (e.g., leopard cat Prionailurus bengalensis) and raptors (e.g., crested goshawk Accipiter trivirgatus) in the limestone forests (Duan et al., 2020;Wu, 1983) are reported as significant death threats to infant primates (Huang, 2002). However, more information on how these factors (including food availability, group size and predation risk) affect the hill zones use by white-headed langurs is unavailable.
In this study, our goals are to explore how the spatial distribution of food and seasonal variations, predation risk, and group size affect the habitat use pattern of white-headed langurs. This may also provide critical information for this primate species conservation.
We collected data on the individual habitat use patterns of whiteheaded langurs and tested the following predictions: 1. We predict white-headed langurs would mainly forage on the hillsides, and seasonally use other hill zones for foraging. Whiteheaded langurs are folivorous and prefer young leaves and fruits Lu et al., 2021). White-headed langurs forage on the hillsides can harvest abundant foods, due to the fact that vegetation is densely distributed on the hillsides Li & Rogers, 2006). Moreover, folivorous primates tend to look for preferred foods (young leaves and fruits) when their availability is higher during the rainy season (Chen et al., 2019).
2. We predict white-headed langurs would use cliffs more frequently when resting. Primates resting on cliffs not only can more easily detect predators with a broader vision, but also can effectively prevent terrestrial carnivores approaching when resting Zhou et al., 2013), because cliffs in limestone forests are located in the upper part of the hills and are covered by sparse vegetation and consist of vertical bare rocks (Li & Rogers, 2006).
3. We predict white-headed langurs both in large and in small groups would mainly forage on hillsides, and the langurs in large groups would expand foraging areas to flat zones for satisfying larger energy needs. White-headed langurs in large groups have more intensive intragroup feeding competition than small groups (Zhang et al., 2020). Moreover, the vegetation on flat zones is relatively better compared with the cliffs and hilltops (Li & Rogers, 2006).

| Study site and study subjects
This study was conducted between September 2016 and August 2017 in the Chongzuo White-Headed Langur National Nature Reserve, southwest Guangxi, China (107°16′53″-107°59′46″E, 22°10′43″-22°36′55″N; Figure 1). The altitude of limestone hills ranges from 400 to 600 m, and vegetation is typical limestone seasonal rainforest (Guangxi Forestry Department, 1993). The habitat could be roughly divided into five zones considering the topography and the vegetation of the region: hilltop, cliff, hillside, flat zone, and farmland (Li & Rogers, 2006). The hilltops have bare rocks and sparse vegetation. The cliffs contain mainly bare vertical rocks and little vegetation, given the lack of soil and water. The hillsides are covered by thick vegetation with large trees, shrubs, and vines and lay below the cliffs. The flat zones are covered by small amounts of trees and shrubs due to historical cultivation, lying at the bottoms of the hills. Farmland is part of flat zones and is used for agriculture by local residents. We did not assess the availability of each hill zone in the habitat, and made records based on observed use pattern by langurs.
Climatic factors were collected during the study period using an automatic electronic thermometer and a rain gauge. We set one thermometer in the middle forest layer, and one on bare rock to record temperatures. The mean annual temperature of bare rock was 26.3°C, and the mean temperature of forest was 22.5°C. Zhang et al. (2021) shows more detailed temperature data (e.g., mean highest/lowest temperature of forest and bare rock). The annual precipitation was 4382.9 mm ( Table 1). Based on the precipitation of the region, the study period was roughly divided into dry (September 2016 to February 2017) and rainy (March to August 2017) seasons (Zhang et al., 2020). The G-DS and G-ZWY groups were classified as large, and the G-LZ and G-NN groups were classified as small. Detailed descriptions of the groups are shown in Table 1.

| Vegetation composition
We randomly sampled 37 plots (30 plots: 20 m × 20 m; 7 plots: 10 m × 10 m) across the main study site (including flat zones, hillsides, and hilltops) to investigate the vegetation composition. Specifically, 9 plots were on the flat zones, 16 plots were on the hillsides, and 12 plots were on the hilltops. We did not set up plots on the cliffs and farmlands because cliffs were inaccessible and farmlands were agricultural areas. We recorded and identified trees, shrubs, and woody lianas with diameter at breast height (DBH) or basal diameter >2 cm within plots (the diameter at approximately 1.2 m from the ground as DBH for woody lianas). We recorded species canopy height and width to calculate canopy volume. Following (Hu, 2011), we used basal area as an index of the leaf biomass of each plant and summed the cumulative leave biomass for per species in specify hill zones.

| Behavioral data collection
The full-day observations of the langurs began after their sleeping sites were located at dawn and ended at sunset. The data sampling in the partial-day observations of the langurs began whenever they were observed and ended when the contact with the group was lost for more than 30 min or when they entered a sleeping site . During the study period, 216 full-day observations were obtained (G-DS: 52 days; G-ZWY: 52 days; G-LZ: 55 days; G-NN: 57 days) ( Table 1).
The behavioral data on the white-headed langurs were collected via instantaneous scan sampling (Altmann, 1974). We set 15 min for one scan unit. More specifically, the first 5 min of each scan were used for behavioral sampling, followed by a 10min interval until the following scan began. The group numbers were scanned from left to right or clockwise to avoid sampling bias toward specific individuals. We collected behavioral data of as many individuals as possible during each scan. No individuals were recorded twice. In each scan mission, the location (i.e., hilltop, cliff, hillside, flat zone, and farmland) of the scanned individuals and their predominant behavior (resting, moving, and feeding; Huang et al., 2017) were recorded after observing them for 5 sec.
We defined the location and predominant behavior of focal group during each scan according to the location and predominant behavior that occurs in the majority of individuals (Chen et al., 2019).
Infants were excluded from scanning, because their behaviors were not independent. When feeding, the plant species and part eaten by the langurs were recorded. A total of 9673 scans were obtained, composed of 2201 (G-DS); 2223 (G-ZWY); 2534 (G-LZ); and 2715 (G-NN) scans (Table 1).

| Data analysis
We set each scan of group as one independent record and determined the monthly utilization percentage for different hill zones by calculating the frequency of specific hill zones in the monthly total records. The annual and seasonal utilization frequencies were expressed as the average values of relevant months (Chen et al., 2019).  (m 2 /ha), and canopy volume of food species in different hill zones were calculated. Spearman's rank correlation was performed to test the correlation between species canopy volume and utilized frequency of specific hill zone, as well as was used to detect the correlation between distribution density, basal area, and canopy volume of food species and utilized frequency of each hill zone.
The density, basal area, and canopy volume of food species on hilltops were positively correlated with the langurs use of hilltops across behaviors; density of food species is correlated with moving and feeding on hilltops (Table 2). Moreover, positive correlations were detected between the density of food species on hillsides and the utilized frequency of hillsides across behaviors. The basal area of food species on hillsides had a positive correlation with the utilized frequency of hillsides when feeding ( Table 2). The basal area of food species on flat zones was positively correlated with the utilized frequency of flat zones when feeding. Similarly, significant correlations were found between the canopy volume of food species on flat zones and the utilized frequency of flat zones (Table 2).

| Seasonal hill zones use variations
Seasonal variations were not detected for most hill zones use by white-headed langurs, except for the flat zones (Table 4 and Figure 2). These langurs exploited the flat zones more frequently during the dry season than during the rainy season (9.37% ± 8.26% vs. 4.61% ± 4.97%, χ 2 = 6.121, df = 1, p = .013).

| Group size effect on hill zones use
There were significant variations in use frequencies for cliffs and hillsides between the large and small groups (Table 5 and Figure 3).

TA B L E 4
Comparisons of utilization frequency on hill zones for white-headed langurs: Effect of season Furthermore, the canopy volume of all plant species on the hillsides is highest, and dense and high canopy offers primates more predation protection with more escape routes and concealment sites (Adams & Teichroeb, 2020;Madden et al., 2010). Therefore, hillsides are food-rich low-risk areas for white-headed langurs, and feeding on the hillsides assists them in safely harvesting adequate foods. The flat zones were used less frequently, likely because some are cultivated zones Li & Rogers, 2005). Common human activities (i.e., sugarcane cultivation) prevented the langurs from using those zones for foraging. Given the illegal hunting on langurs commonly occurs in the past Wang et al., 2005), the langurs generally avoid to utilize areas with high human activities and/or utilize these areas only when human disappear (Waterman et al., 2019).
The langurs spent more time foraging on the hilltops during the rainy season, probably because their preferred food items (i.e., young leaves and fruits) are available in these areas. Food resource distribution varies spatiotemporally (Chen et al., 2019;Hendershott et al., 2018). The langurs spent more time foraging on the hilltops during the rainy season, probably because their preferred food items (young leaves and fruits; Li & Rogers, 2006;Lu et al., 2021) are more available at this time (Zheng et al., 2021). Sympatric Assamese macaques (M. assamensis) show similar trends linked to the temporally increase in availability of hilltop fruits (Li et al., 2017). Contrarily, white-headed langurs used cliffs more frequently during the dry season than during the rainy season when feeding, a trend that was probably related to the reduced availability of young leaves and fruits during the dry season. White-headed langurs consume more mature leaves during the dry season (Li et al., 2016), and the consumption of mature leaves is negatively correlated with young leaves and fruits availability (Lu et al., 2021). In limestone forests, mature leaves are available year round and are distributed throughout the vegetation (Li et al., 2016;Li & Rogers, 2006), likely allowing the langurs to harvest less-preferred food items in a wider-ranger of hill zones, including cliffs. Moreover, white-headed langurs exclusively use natural caves on the cliffs as their night-sleeping sites and sunbathe on bare rocks in winter to reduce the energy in thermoregulation (Huang et al., 2003;Huang & Li, 2005;Li et al., 2011). Foraging on the cliffs is helpful for them to quickly sunbathe after feeding. Limestone-living primates generally sunbathe on bare rocks to save energy for thermoregulation during the winter , and they likely do not have to travel from bare rocks/cliffs when young leaves and fruits are not available elsewhere, and adopt an energy-conservation strategy by reducing travel time when feeding on mature leaves (Li et al., 2019). Additionally, white-headed langurs inhabiting karst forests maintain a small home-range size with shorter daily path lengths during the dry months than during the rainy months, constituting an effective energy-conserving strategy Zhou et al., 2011).
White-headed langurs used hillsides and cliffs more frequently while resting, supporting prediction 2, although they also rested frequently on the hillsides. Primates reduce vigilance behaviors F I G U R E 2 Seasonal variations in hill zones use (overall, moving, feeding and resting) of white-headed langurs in limestone forest of Southwest Guangxi, China. Asterisks represent statistically significant differences between the dry and rainy season when resting and grooming (Cords, 1995). Consequently, primates prefer using refuges (e.g., tall trees and cliffs) that increase their visibility of predators and reduce predation risk (Cowlishaw, 1997a;Zhou et al., 2013). In limestone forests, cliffs located in upper hills are often vertical rock covered by few vegetation, making the terrestrial predators (such as leopard cats) inaccessible and offering good visibility to the langurs while resting . A similar strategy that relies on using the cliffs to reduce predation risk when resting also observed in other primates inhabiting limestone forests, including François's langurs (Trachypithecus francoisi) (Chen et al., 2019) and Assamese macaques . Here, these langurs also spent more time resting on hillsides, even more frequently in the rainy season, probably being linked to food distribution, predation risk, and temperature. As mentioned above, these langurs mainly fed on hillsides and had a folivorous diet. Resting in foraging patches might assist them in digesting fibrous foods . Moreover, hillsides have a dense and high canopy, offering protection against predators (Adams & Teichroeb, 2020). The langurs utilized the flat zones in the dry season more frequently than in the rainy season when resting, possibly because the langurs temporally increase the use frequency of this zone when feeding.
The dense vegetation on hillsides provides shade options for the langurs, particularly during hot and rainy months .
Primates frequently stay within the forest shade, avoiding daytime high temperatures and reducing their energetic thermoregulation demands (Duncan & Pillay, 2013;McFarland et al., 2020;Thompson et al., 2016). The temperature of exposed bare rock surfaces rises rapidly during hot summers, while the temperature within the forest is significantly lower (Huang, 2002), supported by our study . Consequently, time resting in shaded hillsides is helpful for the langurs to reduce body temperature. Similar behaviors are reported for other sympatric primates [e.g., François' langurs  and Assamese macaques ].
The langurs in both large and small groups primarily foraged on hillsides; however, all groups less frequently foraged on flat zones, partially supporting prediction 3 that the langurs both in large and small groups would mainly forage on hillsides and large groups would expand foraging areas to flat zones for satisfying larger energy needs. As we mentioned previously, hillsides are covered by the most abundant food resources and these langurs could safely harvest foods in these zones. However, the langurs in large groups did not forage more frequently on flat zones than the small groups, despite of having heavier food competition (Zhang et al., 2020).
Large groups in current study were closed to a roadside and likely perceived more anthropogenic disturbance from agricultural cultivation and automobile traffic than small groups (Yuan, 2013). As a result, the langurs may infrequently use the flat zones exposed to more human disturbance. The sympatric François' langurs show similar trends in reducing use of bottom zones with heavy human interference . In contrast, the langurs in large groups are more likely to increase their daily path length or decrease their resting time to obtain more foods (Zhang et al., 2020).
The utilization frequency of cliffs by large groups was significantly higher than that by small groups while moving or resting; however, the utilization frequency for hillsides by large groups was smaller than that of small groups when resting. This probably due to predators are mostly a threat to infants (Huang, 2002), and large groups with infants might be more wary of predators. In contrast, groups with more adults/subadults likely suffer less predation presure than groups with more immatures. One limitation of our study is that we do not have data on predator abundance, distribution, and events, due to the topographic conditions.

F I G U R E 3
Difference in hill zones use (overall, moving, feeding and resting) for the large and small groups of white-headed langurs in limestone forest of Southwest Guangxi, China. Asterisks represent statistically significant differences between the large and small groups Our study provides a general pattern of habitat use in the whiteheaded langurs. We conclude that the langurs use their habitat differently depending on behaviors, group sizes, spationl distribution of food resources and seasonal variation. We propose this reflects the langurs balancing foraging and thermoregulatory needs with predation risk and human disturbance. Moreover, our results showed that these langurs mostly fed on hillsides and seldom used the flat zones, which may provide conservation management strategies for these langurs. This study suggested that potecting the vegetation on hillsides is important, and vegetation restoration on flat zones is needed.
Furthermore, human activities should be reduced or prohibited on flat zones, which would make the langurs increse the use of this zone.

ACK N OWLED G M ENTS
We are very grateful to the Department of Forestry of Guangxi Zhuang Autonomous Region, China for permitting us to conduct research. We also thank the editors and three anonymous reviewers for their comments and suggestions on the earlier manuscript versions. This study was supported by the National Nature Science Foundation of China (No. 32170488, 31960106, 31960104, 31301893).

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data are available in the figshare repository at https://doi.