Social networks in group-living animals

Jens Krause

The School of Biology, University of Leeds, Clarendon Way, Leeds LS2 9JT, UK

Network theory, used to describe local and global properties of many interconnected agents, is highly interdisciplinary, attracting the attention of, among others, mathematicians, sociologists and physicists. It has been successfully applied to a wide range of systems including human association networks, neural networks, and the internet. However, the current approach is characterised by an absence of replication (i.e. only single networks are considered), and little consideration is given to the network development (with the emphasis being on pattern description). We present data on the social network structure of shoaling fish in streams and lakes that will go some way towards addressing these problems. Social network structure was quantified using mark, release and recapture data, whereby individual fish are given unique identification tags and their social companions documented over a 1-month period. We found all populations to have highly structured social networks that exhibited characteristics consistent with the 'small world' phenomenon (i.e. short path length between individuals but social cliquishness). Furthermore, we observed a significant co-occurrence of pairs of individuals, a finding that fulfils the basic prerequisite for the evolution of reciprocal altruism. We discuss our results in the context of agent-based modelling as a technique that might give an insight into network development.