Modeling animal movement with directional persistence and attractive points

TL;DR

This paper introduces a novel hidden Markov model with a new emission distribution to effectively capture animal movement behaviors, including directional persistence and attraction points, demonstrated on GPS data of a sheepdog.

Contribution

A new Bayesian hidden Markov model with a specialized emission distribution that models both directional persistence and attraction points in animal movement data.

Findings

Model outperforms existing competitive models.

Results are easily interpretable.

Effective in analyzing GPS data of animal movement.

Abstract

GPS technology is currently easily accessible to researchers, and many animal movement datasets are available. Two of the main features that a model which describes an animal's path can possess are directional persistence and attraction to a point in space. In this work, we propose a new approach that can have both characteristics. Our proposal is a hidden Markov model with a new emission distribution. The emission distribution models the two aforementioned characteristics, while the latent state of the hidden Markov model is needed to account for the behavioral modes. We show that the model is easy to implement in a Bayesian framework. We estimate our proposal on the motivating data that represent GPS locations of a Maremma Sheepdog recorded in Australia. The obtained results are easily interpretable and we show that our proposal outperforms the main competitive model.