Seasonal changes in the structure of rhesus macaque social networks

Abstract

Social structure emerges from the patterning of interactions between individuals and plays a critical role in shaping some of the main characteristics of animal populations. The topological features of social structure, such as the extent to which individuals interact in clusters, can influence many biologically important factors, including the persistence of cooperation, and the rate of spread of disease. Yet the extent to which social structure topology fluctuates over relatively short periods of time in relation to social, demographic or environmental events remains unclear. Here, we use social network analysis to examine seasonal changes in the topology of social structures that emerge from socio-positive associations in adult female rhesus macaques (Macaca mulatta). Behavioral data for two different association types (grooming, spatial proximity) were collected for females in two free-ranging groups during two seasons: the mating and birth seasons. Stronger dyadic bonds resulted in social structures that were more tightly connected (i.e. of greater density) in the mating season compared to the birth season. Social structures were also more centralized around a subset of individuals, and were more clustered in the mating season than the birth season, although the latter differences were mostly driven by differences in density alone. Our results suggest a degree of temporal variation in the topological features of social structure in this population. Such variation may feed back on interactions, hence affecting the behaviors of individuals, and may therefore be important to take into account in studies of animal behavior.

Keywords: network topology; rhesus macaques; social network analysis; social structure.

Fig. 1

Social structures for two groups of female rhesus macaques based on spatial proximity during the mating and birth seasons (spring-embedded sociograms generated in UCINET: Borgatti et al. 2002). Symbols represent individual females. Line thickness represents the number of times pairs of females were observed in proximity (<2m) to one other relative to the amount of time they were observed, with thicker lines representing more frequent proximity

Fig. 2

Social structure for two groups of female rhesus macaques based on grooming interactions during the mating and birth seasons (spring-embedded sociograms generated in UCINET: Borgatti et al. 2002). Symbols represent individual females. Line thickness represents the amount of time pairs of females spent grooming one another (seconds per hour a pair were observed), with thicker lines representing more frequent grooming

Fig. 3

Violin plots of the bootstrapped (1,000 iterations) densities of female rhesus macaque proximity (a–b) and grooming (c–d) social structures. Network densities were determined for two social groups, group F (a and c) and group V (b and d). Violin plots are composed of frequency distributions mirrored on both sides of box plots. Plots were generated using R (http://www.r-project.org). Overlap of 95% confidence intervals was measured for pairwise comparisons. **p < 0.05

Fig. 4

Violin plots of the bootstrapped (1,000 iterations) network centralization scores of female rhesus macaque proximity (a–b) and grooming (c–d) social structures. Network centralization was determined for two social groups, group F (a and c) and group V (b and d). **p < 0.05. Violins and whiskers do not appear on plots due to narrow ranges of variation

Fig. 5

Violin plots of the bootstrapped (1,000 iterations) mean clustering coefficients of female rhesus macaque proximity (a–b) and grooming (c–d) social structures. Mean clustering coefficients were determined for two social groups, group F (a and c) and group V (b and d). **p < 0.05. Distributions of clustering coefficient values for individual females in the mating (m) and birth (b) seasons are presented in the inset graphs