Estimating stop over duration in the presence of trap-effects

Detection probability of individuals is increasingly taken into account during field monitoring schemes and in demographic models. Conversely, it is often taken for granted that trappability of animals will remain fairly constant and broadly similar between individuals present in a given area. However, animals may change their behaviour after being trapped. In this paper, we introduce a new hidden Markovian model to estimate stop over duration in the presence of trap-effects. This model combines nonhomogeneous Markovian states with semi-Markovian states in the non-observable state process, and simple distributions with first-order Markov chains as observation models. This model generalizes previously proposed models and enables the joint modeling of the time of residence and the trap effect. Two cases are considered, depending on whether or not emigration is time-dependent since arrival. We illustrate the latter with teal Anas crecca wintering in Camargue, Southern France and we demonstrate the importance of handling trap-effects.

► A similar detection probability of individuals is increasingly taken into account models.
► However, animals may change their behaviour after being trapped.
► We introduce a new hidden Markovian model to estimate stop over duration in the presence of trapeffects.
► Two cases are considered, depending on whether or not emigration is time-dependent since arrival.
► We illustrate the latter with teal Anas crecca wintering in Camargue.