Detecting changes in dynamic social networks using multiply-labeled movement data

TL;DR

This paper introduces a Bayesian hierarchical model that effectively analyzes animal social networks from movement data with labeling issues, enabling insights into social structure changes under external influences like sonar exposure.

Contribution

It extends existing social movement models to handle multiply-labeled data, allowing for more accurate analysis of social networks in animal populations.

Findings

Model successfully applied to dolphin movement data

Estimated impact of sonar on dolphin social structure

Framework adaptable to various social movement studies

Abstract

The social structure of an animal population can often influence movement and inform researchers on a species' behavioral tendencies. Animal social networks can be studied through movement data; however, modern sources of data can have identification issues that result in multiply-labeled individuals. Since all available social movement models rely on unique labels, we extend an existing Bayesian hierarchical movement model in a way that makes use of a latent social network and accommodates multiply-labeled movement data (MLMD). We apply our model to drone-measured movement data from Risso's dolphins (Grampus griseus) and estimate the effects of sonar exposure on the dolphins' social structure. Our proposed framework can be applied to MLMD for various social movement applications.