Investigating structure and temporal scale in social organizations using identified individuals

Studies of individually identified animals can produce substantial data sets containing information on the structure and temporal scale of social organizations. However, methods of analyzing such data are not well established. Important features of a social organization are revealed by plotting the rate of persistence of the associations between pairs of individuals over a range of time lags (lagged association rate). The consistency of long-term relationships can be characterized using the rate of association of pairs of individuals between their first and last observed associations (intermediate association rate). A hierarchical series of models featuring exponentially decaying lagged association rates may be fitted to these data. This technique retrieved the essential parameters of five simulated social organizations and, when used on real data, portrayed the essential features of the patterns of temporal change in relationships between animals. The method should be especially useful for analyzing fissionfusion societies containing 10-10, 000 individually identifiable animals.