Patterns in swimming by a scyphomedusa: a novel approach to quantifying behavior in individuals

Behavior is commonly studied at the group level using several individuals, but there is increasing evidence that the behavior of a few individuals often has a disproportionate effect on the response of a population to its environment. The present study used a suite of statistical techniques, random series analysis, analysis of variance, spectral analysis, and goodness-of-fit tests of frequency histograms, to quantitatively describe the time-dependent changes in individual behavior. Each technique reveals a different facet of the behavior and, when simultaneously applied to the data, distinguishes significant differences among the behaviors of several individuals. The approach was developed and tested on the swimming behavior of four specimens of the scyphomedusa Aurelia aurita (Linnaeus, 1758), which were observed for 19 days, beginning 16 January 1998, and videotaped under identical environmental conditions during that period. The analyses showed that each medusa swam in a unique pattern, varying swimming at characteristic frequencies. Application of the approach to individual-based numerical modeling, to the role of endogenous stimuli in the behavioral repertoire, and to in situ studies of animal behavior is discussed.