Abstract

The stable, long term social structures formed by animals or human groups are the outcome of repeated interactions between individuals that occur over fast temporal scales. Understanding the inter-relations that exist between pairwise interactions, relationships and complex social network architectures is a major challenge in the study of animal societies and social networks in general. We construct and study a stochastic network model where individuals search new connections and reinforce their links with the neighbours with whom they share an attribute (homophily). The model quantitatively accounts for many features of the association network of spider monkeys (Ateles geoffroyi) in the wild, including the community structure, weight distribution and dependence on the attributes. As the interactions are actually of several types (proximity, aggression, grooming,....) we further discuss the need of considering multiplex networks as a tool for representing social structures.