Identifying the time scale of temporal association

Abstract

Previously used methods of investigating the temporal association of behavioural events have tested sequences of events for temporal clumping; such an approach does not quantitatively indicate the time scale over which the association is significant. We present a novel method of analysis that specifically identifies this time scale, thereby allowing more ready elimination of competing causal hypotheses for the association observed. Our null hypothesis is that, following any event, the number of events in a given time interval is independent of the time elapsed since the initial event. The method is simple and robust against long-term changes in event frequency, and provides a more detailed quantitative description of the temporal association. We illustrate it using data on spawning in the blueheaded wrasse (Thalassoma bifasciatum). The analysis demonstrates that, after a spawning event, there is a 70% increase in mean spawning frequency which lasts for 30 s and then abruptly disappears. The effect of spawns appears to summate multiplicatively.