Baptiste Brault, Jeanne Clermont, Andreas Zedrosser, Andrea Friebe, Jonas Kindberg and Fanie Pelletier

 - Abstract: Hibernation is a key phase in the life cycle of many species. In bears, it constitutes a response to harsh winter conditions, providing protection to mothers and their young after birth, as well as to all bears that need to hibernate. Various methods have been used to identify den entry and exit dates, often based on GPS data. We propose a methodology for processing activity data using hidden Markov models to identify start and end of inactivity periods in bears. The method we propose is accurate, robust, and easily reproducible. It relies on activity data continuously collected at 5-minute intervals, whereas GPS data are generally collected at longer intervals, only provide information on movements, and are subject to detection issues. We believe that start and end dates of inactivity periods provide better approximations of denning period. This methodological approach allows for more precise identification of prolonged inactivity periods and improves the detection of behaviors such as hibernation. Thus, we will be able to observe whether the phenology of these behaviors has changed with weather conditions.