Estimating spatial location of attraction through time with animal movement data
Loading...
Authors
Keating, Meghan Patricia
Issue Date
2021
Type
Thesis
Language
Keywords
Attraction Point , Bayesian , Change Point , Markov chain Monte Carlo , Movement , RJMCMC
Alternative Title
Abstract
As statistical methodology catches up with GPS technology, researchers are gaining new insight on movement behavior and wildlife decision-making that were previously unobtainable. New areas of research target underlying drivers of animal movement including understanding how animal movement is driven by points of attraction on the landscape. Estimating attraction points using movement data can be challenging when an animal changes behavior, and the change is not obvious. I have developed a parametric statistical modeling framework that mechanistically models animal movement data to identify if animal movement is driven by points of attraction on the landscape. The framework is flexible enough to identify an arbitrary number of attraction points and change points. We optimally estimate the number of attraction points in an animal movement path by framing the question in a model selection context and using reversible-jump Markov chain Monte Carlo methods. I validated the framework using extensive simulation experiments which all suggested appropriate model fit and selection. I applied the modeling framework to a case study of four mule deer in the Mojave National Preserve. Thirty-seven percent of attraction points were associated with known water sources, a disproportionately large amount, given the composition of water in the Mojave National Preserve. These preliminary results support the importance of water sources on mule deer movement behavior.