Statistical mechanics techniques applied to the analysis of humpback whale inter-call intervals

Abstract

Techniques developed in statistical mechanics recently have been applied to the analysis of the topology of complex human communication networks. These methods examine the network's macroscopic statistical properties rather than the details of individual interactions. Here, these methods are applied to the analysis of the time intervals between humpback whale calls detected in passive acoustic monitoring data collected by the bottom-mounted hydrophones on the Pacific Missile Range Facility (PMRF) west of Kauai, Hawaii. Recently developed localization and tracking algorithms for use with PMRF data have been applied to separate the calls of an individual animal from those of a collection of animals. As with the distributions of time intervals between human communications, the distributions of time intervals between humpback whale call detections are distinctly different than those expected for a purely independent, random (Poisson) process. This conclusion holds both for time intervals between calls from individual animals and from the collection of animals vocalizing simultaneously. although significant differences in these probability distributions occur. A model based on the migration of clusters of animals is developed to fit the distributions. Possible mechanisms giving rise to aspects of the distributions are discussed. [Work supported by the Office of Naval Research, Code 322-MMB.]