Abstract 328 – Effectiveness of individual-based activity analysis and hidden markov models at predicating animal activity of a wild Eurasian brown bear

David Blount, Ryan Fregmen, Mark Chynoweth, Josip Kusak, Cagan Sekercioglu

As GPS collars grow in popularity and decrease in price, many research programs have
incorporated these tools to understand temporal trends of their study animal. Specifically,
activity patterns have been used to understand how animals persist on landscapes, what
threats they face, and how they may avoid these threats by changing their temporal patterns.
Changes in activity patterns can have drastic consequences, with animals navigating a matrix of
risk and reward through time, affecting fitness, mortality risk, and risk of starvation. However,
many of the analysis used to understand activity have only been tested in simulations or in
captive populations. In this study we use a GPS collar with a camera to compare binomial
activity estimates and activity states estimated with hidden Markov models to observational
camera data from a wild Eurasian brown bear. We hope to show how well each of these activity
analysis predict actual activity in the wild.