Main content area

Advances in methods for estimating stopover duration for migratory species using capture–recapture data

Guérin, S., Picard, D., Choquet, R., Besnard, A.
Ecological applications 2017 v.27 no.5 pp. 1594-1604
breeding, computer software, data collection, females, habitats, males, migratory behavior, migratory species, models, probability, salamanders and newts, wildlife management
Many species are migratory, resulting in a life cycle divided into periodic stages occurring in different habitats occupied for a limited amount of time. Estimating the time spent in each habitat is crucial to understanding how individuals modulate their activities and thus to evolutionary ecology and conservation biology. Several methods, including some recent promising advances, can be used to estimate stopover duration as well as arrival and departure probabilities at sites where individuals are monitored using capture–recapture sampling. Our objectives in this study were to (1) describe the available models to estimate stopover duration, (2) illustrate with an original data set what kinds of questions can be addressed using the most recent methods, and (3) to provide in a detailed appendix a practical guide for implementing these methods in E‐SURGE software. To illustrate the potential of these models for testing biological hypotheses, we used a capture–recapture data set on marbled newts (Triturus marmoratus). We used time‐dependent and time‐elapsed‐since‐arrival effects (using both Markovian and semi‐Markov processes for the latter) to model stopover duration and the probability of arriving in and departing from a breeding pond for this species and compared the relative performance of the resulting models. Our findings showed a strong sex effect on stopover duration: females stayed on average 5.63 weeks in a breeding pond whereas males stayed only 3.03 weeks. In both sexes, the retention probability was mainly influenced by the time already spent there. Consequently, individuals of the same sex stayed a similar amount of time in a pond, although they did not arrive simultaneously but successively. The selected data set demonstrated the flexibility of these methods and their potential relevance for applications in evolutionary ecology and conservation.