THE SECRET LIVES OF MARTENS: LEVERAGING COLLAR DATA TO REVEAL ACTIVITY PATTERNS AND REST SITE COMPOSITION IN PACIFIC MARTENS

Jessica Buskirk; NCASI; jbuskirk@ncasi.org; Mathew Titus, Holly Munro, Dalton Hance, Katie Moriarty

Understanding behavior and how to identify crucial landscape features is essential for conserving rare species. One such species are Pacific martens (Martes caurina), including the threatened Humboldt subspecies (M.c. humboldtensis). Understanding where these martens rest and how they partition time between behaviors is critical for informing conservation, yet behavior and natural history information is largely unavailable for the species. Here, we describe a multi-step process of leveraging collar data and vegetation surveys to understand activity patterns and to identify the locations, composition, and thresholds of resting areas. We collared martens with devices that recorded acceleration and GPS locations. We analyzed these data using a two-stage approach: first, a Hidden Markov Model classified behavioral states from accelerometer signatures; second, a State Space Model estimated true spatial positions while accounting for GPS error. From these, we identified resting locations across five study areas. We then collected plot-level conditions at a subset of used and random locations (n = 270, 221, respectively) and used a boosted C5.0 algorithm to build decision trees describing thresholds of marten resting conditions. These methods offer a transferable approach for quantifying activity patterns, more efficiently identifying resting habitat, and may be especially helpful for understanding cryptic species’ behavior.

Ecology and Conservation of Small Mammals II 
Wednesday 3:30 PM
 

Speaker Bio: