Title: Markov-switching conditional logistic regression with application to animal movement data of interacting individuals
Authors: Jennifer Pohle - University of Potsdam (Germany) [presenting]
Johannes Signer - University of Goettingen (Germany)
Ulrike Schlaegel - University of Potsdam (Germany)
Abstract: Integrated step selection analysis is a popular statistical tool in ecology to study animal's movement and habitat selection based on conditional logistic regression. It has also successfully been applied to detect interactions (such as avoidance or attraction) between simultaneously tracked individuals. However, animals usually switch between different behavioural modes (such as resting and foraging), which might influence their preferences, habitat selection, and movement patterns. Ignoring such behavioural states in the analyses might lead to biased results and possible erroneous conclusions. To account for the usually unobserved behaviour, we present an approach where an underlying latent Markov chain is introduced into the framework, which allows preferences and movement patterns to vary over time. A simulation study is used to investigate the performance of the resulting Markov-switching integrated step selection analysis and to compare it to alternative candidate models. Furthermore, the approach is illustrated in a case study on location data from simultaneously tracked bank voles. Besides animal movement data, the inherent Markov-switching conditional logistic regression is also applicable to longitudinal discrete choice and case-control studies.