Introduction

Group cohesion is expected to vary both within and between species, depending on factors such as predation risk, the density and distribution of food resources, and patterns of feeding competition and activity (Wrangham 1980; Isbell 1991; Heesen et al. 2015). Groups tend to become more cohesive with increasing predation risk, as the presence of many close neighbors reduces the chance of being attacked (Janson and Goldsmith 1995). At the same time, smaller inter-individual distances are likely to increase feeding interference, and foraging animals are thus likely to spread out to reduce feeding competition (van Schaik and van Noordwijk 1988; Nishikawa et al. 2014; Sugiura et al. 2014).

Due to the varying costs and benefits of different numbers of neighbors, animals are predicted to engage in risk-sensitive foraging, meaning that they have to balance a trade-off between feeding and avoiding danger (Lima and Dill 1990; Miller 2002). The outcome of this trade-off is affected by factors such as resource availability and predation risk. For example, individuals that are more vulnerable to predation, such as juveniles or females with infants, are often found closer to other group members, where they may be safer from predation (Rhine et al. 1980; Busse 1984; Collins 1984).

Dominance rank has been reported to affect inter-individual distances among females in many species (Janson 1995; Isbell and Young 2002). Higher-ranking females may be attractive or unattractive in relation to feeding competition and/or aggressive interactions (Isbell and Young 2002).

Here I report on observed variation in inter-individual distances among adult females in a group of wild patas monkeys (Erythrocebus patas) in north Cameroon. I hypothesize that higher group cohesion may reduce the predation risk among infants. Accordingly, group cohesion is expected to be higher during the birth season than the non-birth season. First, I describe the variation in inter-individual distances in the study group during the non-birth and birth seasons, and then I examine how the presence of newborns affected the cohesiveness of the group. Second, I investigate how female dominance rank may affect inter-individual distances by examining the correlation between the female rank and the number of adult female neighbors.

Methods

Study site and subjects

The present study involved a troop of patas monkeys (KK troop) that inhabit the Kala Maloué National Park, Cameroon, West Africa. The Kala Maloué National Park is situated in the Extreme North Province of Cameroon beside the Chari River (12°09′N, 15°53′E). It is roughly divided into three vegetation types: the riverine area along the Chari and its branch supports woodlands dominated by Morelia senegalensis, Acacia sieberiana, Mitragyna inermis, and Crateva religiosa, and the area away from the rivers supports grassland and open woodland dominated by Balanites aegyptiaca, Acacia seyal, and A. nilotica. The climate is characterized by two sharply defined seasons: a wet season from the latter half of May to September, and a dry season encompassing the rest of the year. The mean annual rainfall and temperature during 1985–1988 were 497 mm and 28.9 °C, respectively. Predation risk at this site has been reported to be lower than that in other populations, as there are few carnivore species (Nakagawa et al. 2003). Nevertheless, young infants are vulnerable to medium-sized carnivores such as the golden jackal (Canis aureus) and spotted hyenas (Crocuta crocuta).

The KK troop had been well habituated and identified since 1984 (Ohsawa et al. 1993), and thus I knew kin and linear dominance relationships among the members of the troop at the beginning of the study by observation. Dominance relationships were determined by approach–retreat interactions at the beginning of the study. The most dominant female was Ft, followed by Fd, Kr, Tf, and Tt, and Sk was the most subordinate. When the study began, the troop (N = 9) consisted of one adult male, six adult females (>5 years old), one juvenile female (1.5 years old), and one infant (0.5 years old). One infant disappeared on 27 November 1988. Three higher-ranking (Ft, Fd, Kr) adult females gave birth during the birth season. Two pairs of adult females, Ft–Fd and Tf–Tt, were a mother and her adult daughter, respectively. I selected all six adult females as focal subjects. Mating season is July–September (Ohsawa et al. 1993) and birth season is January–February (Nakagawa et al. 2003) for patas monkey at this study site. I collected data both outside the birth season (October and November 1988) and during it (January and February 1989).

Data collection

I collected data on inter-individual distances by following one target animal for a day and conducting group scans (‘scan sampling’, Martin and Bateson 2007). I scanned and identified adult females within a radius of 50 m from the target animal at 10-min intervals. During the non-birth season, I followed each target animal a total of about 20 h (N = 6, mean 18 h 31 min, range 15 h 50 min to 21 h 13 min). The mean number ± SE of scans for six females was 110.5 ± 11.7. During the birth season, I followed each target animal a total of about 30 h (N = 3, mean 26 h 42 min, range 22 h 32 min to 29 h 21 min) for each mother of a newborn infant (mean number of scans ± SE for three females was 151.3 ± 28.1), and 20 h (N = 3, mean 20 h 11 min, range 17 h 58 min to 21 h 45 min) for each non-mother (mean number of scans ± SE for three females was 119.7 ± 8.4). During the birth season, I started observations of the three target females (Ft, Fd, Kr) after they had given birth. Visibility was stable during the study period.

Focal animal follows lasted all day, when possible. Whenever a session was interrupted, I resumed at almost the same time on another day. Thus I scheduled observation hours such that they were almost evenly distributed between 0630 and 1730. During scans, I recorded the distance between the focal animal and other females, divided into six distance categories (A: 0–1 m, B: 1–3 m, C: 3–5 m, D: 5–10 m, E: 10–20 m, F: 20–50 m).

Statistical analysis

I performed all statistical analyses with R (version 3.3.0). I examined the effects of distance from the target animal and season on the number of individuals at each distance category using generalized linear mixed models (GLMMs, Agresti 2007). Response variables were the number of animals in each distance category in each scan (/10 min). Explanatory variables were seasons (non-birth/birth) and six distance categories. The total number of scans was 1476. For the analysis, I used log as link function, Poisson distribution as error distribution, and the focal animal identity as a random factor.

When a response count y1 has an index (total number of counts of each female in each distance category in each season) equal to t1, the sample rate is y1/t1 (Table 1). The link function is as follows:

$$\log \left( {\frac{y1}{t1}} \right) = \beta 1x1 + \beta 2x2 + \beta 0.$$
Table 1 Number of adult females per scan (/10 min) at each distance category from the target animal during (a) non-birth season and (b) birth season
Full size table

This model has equivalent representation

$$\log \left( {{\text{y}}1} \right) = \beta 1x1 + \beta 2x2 + \beta 0 + { \log }(t1)$$

where y1 = number of count, βi = parameter of xi, x1 = distance category, x2 = season, and log(t1) = the offset. I used likelihood ratio statistics for the differences between models to compare the effects of distance category and season on the number of animals in each distance category in each scan (/10 min).

I then analyzed data collected during the non-birth season and those collected during the birth season separately to examine the effect of rank on the number of individuals at each distance category. To compare the number of animals at each distance category, I used Kendall’s rank correlated tests for each distance category. All probabilities reported are two-tailed.

Results

General pattern of inter-individual distance

Table 1 shows the total number of scans and the number of females per scan (/10 min) at each distance from the target animals during non-birth season and birth season, and mean number of females per scan at each distance. In general, the number of females per scan increased more as the distance became larger, and most numbers of females per scan were recorded at 20–50 m distance category in both the non-birth and birth season. However, differences between seasons were clear at 0–1 m and 20–50 distance categories, in which the number of females were higher during birth season than non-birth season.

Likelihood-ratio statistics indicated that the number of individuals at each distance category differed between seasons (df = 1, 70, residual deviance = 1496.38, p < 0.0001). Moreover, the number of animals at each distance category differed significantly (df = 5, 66, residual deviance = 426.96, p < 0.0001).

Non-birth season

Individual females generally had few close females (Table 1), with a mean of 2.4 ± 0.3 animals (range 1.9–2.6) within a radius of 50 m during the non-birth season. The number of individuals per distance category appears to have increased between 1 m and more than 1 m, and between 5 m and more than 5 m (Table 1). The higher-ranking females were likely to have fewer females (<1 m) than did lower-ranking females though no statistical difference was found (Table 1, Kendall’s tau = 0.733, p = 0.06). No statistical differences were found in any other distance categories either, or within a radius of 50 m.

Birth season

During the birth season, females had a mean number of 3.0 ± 0.2 animals within 50 m. All females had a larger number of animals within a radius of 50 m during the birth season than the non-birth season. Inter-individual distances appear to have increased between 5 and 10 m and between 20 and 50 m during the birth season (Table 1). In particular, between 0 and 1 m and between 20 and 50 m, numbers of animals were much greater during the birth season than the non-birth season (Table 1). The higher-ranking females did not have a larger number of individuals within 50 m compared to the lower-ranking females.

Discussion

GLMMs indicated that the inter-individual distances among females differed between the non-birth and the birth seasons. Moreover, the number of animals at each distance category differed significantly. These results suggest that the presence of newborn infants affected the inter-individual distances among patas females, and possibly the spacing behavior of females. Indeed, Muroyama (1994) reported that females frequently approach mothers of newborn infants before initiating grooming in exchange for allomothering, possibly leading to increased cohesiveness among adult females within 1 m.

Individual females generally had a small number of adult females (less than 0.1) within <1 m during both the non-birth and birth seasons. Such large inter-individual distances may be characteristic of wild patas females, although no comparative data exist in other primate species (cf. Sugiura et al. 2014).

During the non-birth season, inter-individual distances between female patas monkeys were generally large, with females consequently being widely dispersed and cohesion being low. In contrast, during the birth season, group cohesion was generally higher, and females with newborn infants, which might be more vulnerable to predation, were located closer to other females.

Dominance relationships appear to have had little effect on the number of neighbors during both the non-birth and birth seasons, though the higher-ranking females were likely to have fewer females within <1 m during the non-birth season than during the birth season. These results suggest that in patas monkeys, inter-individual distances may be structured by the perception of risk to newborn infants, though the effects of seasonal differences in food resources were unknown in this study.

The influence of dominance relationships on affiliative and/or agonistic interactions in patas monkeys is conflicting (Nakagawa 1992, 2008). While linear hierarchies are commonly reported in Kala Maloué National Park (Nakagawa 1992, 2008), some studies have found that dominance relationships did not have a significant effect on affiliative behaviors such as short inter-individual distances and/or grooming (Goldman and Loy 1977; Kaplan and Zucker 1980; Rowell and Olson 1983). In this study, there were no significant effects of dominance relationships on inter-individual distances between adult females during either the non-birth or birth season.

Muroyama (1994) reported that dominance relationships among female patas monkeys, which are linear in Kala Maloué National Park, did not affect the exchange patterns of grooming and allomothering. More detailed analyses involving the distribution between affiliative and aggressive interactions among females are needed to examine the social structure of patas monkeys and the relationships between ecological and social factors.