The Influence of familiarity, resources, and sampling on social structure: A Simulation-based study

Abstract

The wide diversity in social structures is thought to be the result of selection for individual behavioural strategies and inter-individual interaction patterns that maximise fitness under different environments (van Schaik and van Hoff 1983, Kappeler and van Schaik 2002). Recently, several simulation studies have attempted to generate social structure through simple and general models (Ilany and Akçay 2015, Rios and Kraenkel 2017, Cantor and Farine 2018) but have not simultaneously examined the effect of resource conditions on social structure. Here, I describe a simulation study that I conducted to examine whether a social structure emerged when individuals associated with others to different extents based on familiarity, in a habitat with limited and patchy ephemeral resources that either varied in quantity or stayed constant over time. Additionally, I also examined the sampling conditions under which a social structure would be wrongly inferred even when associations were random. Thus, the first question dealt with possible social structure emerging due to simple rules, whereas the second dealt with possible apparent social structure when there was none. Results from my simulations showed that when resources were limited and ephemeral, social structure, characterised by low network density and high modularity, emerged only when all the associations were with familiar individuals and there was no temporal variation in resources. When there was temporal variation in resources, this structure broke down even when many associations were with familiar individuals, becoming similar to that obtained for random association. When associations were only with familiar individuals, social structure could also emerge if resources were not limited. Simulations to address the second question showed that low sampling intensity and a small sampling period could lead to apparent social structure (with high modularity and low density) even in a population with random associations. Moderately intense sampling conducted over long periods of time was essential to detect social structure close to the true structure.