Article révisé par les pairs
Résumé : In group-living species, individuals gain significant advantages from establishing an extensive network of social relationships. This results in complex organizations that are difficult to quantify in a comprehensive manner. In this respect, network analyses are an ideal means to pinpoint the overall properties of social structures, and the place of each individual within these structures. We used network measurements to investigate cross-species variations in the social style of macaques, and studied 12 groups from four species. Two species (Macaca mulatta, Macaca fuscata) were characterized by a relatively weak social tolerance, a steep gradient of dominance and a strong preference for kin. The other two species (Macaca nigra, Macaca tonkeana) were known to display higher levels of tolerance, relaxed dominance and low kinship bias. We used a centrality index based on eigenvector centrality to show that in a comparison of intolerant and tolerant species, top-ranking individuals were more central than other group members in the former species than the latter. We also found that networks had higher modularity in intolerant species, indicating that kin-related partners interacted more frequently in subgroups of these species than in those of tolerant species. Consistently, the matrix of body contacts was more strongly correlated with the kinship matrix in intolerant species. This study demonstrates the efficiency of network methodology in detecting fine and overall contrasts in social structures, and also reveals novel dimensions in the social style of macaques. © 2011.