Efficient spatial multi-state capture-recapture model to study natal dispersal: An application to the Alpine marmot

Studying natal dispersal in natural populations using capture-recapture data is challenging as an unknown proportion of individuals leaves the study area when dispersing and are never recaptured. Most dispersal (and survival) estimates from capture-recapture studies are thus biased and only reflect what happens within the study area, not the population. Here, we elaborate on recent methodological advances to build a spatially explicit multi-state capture-recapture model to study natal dispersal in a territorial mammal while accounting for imperfect detection and movement in and out of the study area. We validate our model using a simulation study where we compare it to a non-spatial multi-state capture-recapture model. We then apply it to a long-term individual-based dataset on Alpine marmot Marmota marmota. Our model was able to accurately estimate natal dispersal and survival probabilities, as well as mean dispersal distance for a large range of dispersal patterns. By contrast, the non-spatial multi-state estimates underestimated both survival and natal dispersal even for short dispersal distances relative to the study area size. We discuss the application of our approach to other species and monitoring setups. We estimated higher inheritance probabilities of female Alpine marmots, which suggests higher levels of philopatry, although the probability to become dominant after dispersal did not differ between sexes. Nonetheless, the lower survival of young adult males suggests higher costs of dispersal for males. We further discuss the implications of our findings in light of the life history of the species.