Abstract:
|
Network modeling techniques provide a means for quantifying social characteristics of populations of animals. However, data used to estimate the social associativity within a group are typically in the form of counts of interactions between individuals based on ad hoc thresholds of physical proximity. In most applications, collecting these data is expensive, time consuming, and potentially invasive. Telemetry data offer an alternative way of estimating the pairwise associativity among individuals in a group. We examine some possible measures of associativity based on comparing animal paths. Once these measures of associativity are established, we develop latent space models to visualize and interpret social structure as a network. We also investigate the implied spatial dependence that arises based on the type of latent process, and compare it with other spatial dependence structures. Our method is flexible, allows for the incorporation of expert knowledge, and makes it possible to infer dynamic social structure for groups of animals that are not easily observed visually. We demonstrate our method using telemetry data gathered on northern fur seals.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.