Analysing animal social structure

This paper presents a framework for analysing the social structure of populations in which interactions between some identified individuals can be observed. Statistics describing the nature, quality and temporal patterning of one or more interaction measures are used to define relationships between pairs of individuals or classes of individual. Multivariate techniques can then be used to display the social structure of the population. These displays indicate the social complexity of the population and can be used to classify relationships and examine patterns of relationship between classes of animal. They can also be used to define and delineate groups. This framework and these techniques should be particularly useful when analysing complex fission–fusion societies, as are found among the primates and cetaceans. ? 1997 The Association for the Study of Animal Behaviour A knowledge of the social structure (here synonymous with social organization) of a population is important for a range of fundamental and applied purposes. Social structure defines an important class of ecological relationships, those between nearby conspecifics. It may include competition, cooperation and dominance in the acquisition of mates or resources, as well as competitive or cooperative care of offspring, and even cannibalism. Thus social structure is often an important element of the population biology of a species, influencing gene flows, spatial pattern and scale, and the effects of predation or exploitation by humans (Wilson 1975). Social structure also sets the scene within which communication and cognition take place, and appears to be an important correlate, and perhaps evolutionary determinant, of communicative and cognitive behaviour (Byrne & Whitten 1988). For some phylogenetic groups of animals, the range of social structures is sufficiently constrained (by anatomy, environment, physiology or life history) that a fairly simple set of principles can be used to describe and classify them (e.g. Michener 1969 for the social insects). In contrast, the more cognitively advanced mammals often have complex social structures, which vary considerably and nearly continuously between and Correspondence: H. Whitehead, Department of Biology, Dalhousie University, Halifax, Nova Scotia, Canada B3H 4J1 (email: [email protected]). 0003–3472/97/051053+15 $25.00/0/ar960358 ? 1 105 within species (e.g. Dunbar 1988). Describing and classifying these societies is complex and challenging (Costa & Fitzgerald 1996). Hinde (1976) proposed a conceptual framework for examining social structure involving three principal levels: interactions between individuals, relationships among individuals described by the content, quality and temporal patterns of interactions, and social structure described by the content, quality and patterning of relationships. Hinde (1976) developed this framework to structure the analyses of primatologists, anthropologists, sociologists and social psychologists, and it has been implicitly or explicitly referred to in a number of subsequent analyses of primate social organization (e.g. Cheney et al. 1987; Dunbar 1988). To analyse the social structure of a population in the manner proposed by Hinde (1976) requires detailed information on the interactions between individual members of the population collected over a considerable period of time (so that the temporal patterning of interactions can be described). This has been possible for some primate populations (e.g. Goodall 1986). Many animals whose social structures appear complex and interesting, however, live in situations that make it impossible to collect detailed data on interactions between individuals. These include nocturnal animals, migratory animals, animals that live in large groups or spend much of their time below 997 The Association for the Study of Animal Behaviour