Describing dispersal under habitat constraints: A randomization approach in lesser kestrels

Animal dispersal is usually studied with capture-mark-reencounter data, which provide information on realized dispersal but rarely on underlying processes. In this context, the unreliable assumption of all habitat being available is usually made when describing and analysing dispersal patterns. However, actual settlement options may be constrained by the spatial distribution of appropriate patches, so an important task to understand movement patterns is to adequately describe dispersal when the dispersers’ options are constrained by the sites that are available to them. Using a long-term monitored population of the migratory lesser kestrel, we show how randomization procedures can be used to describe dispersal strategies in such situations. This species breeds colonially in discrete patches, most individuals (83%) disperse from their natal colony, and dispersers tend to move short distances (median 1⁄4 7.2 km). Observed patterns (natal dispersal rates and median dispersal distances of birds emigrating from their natal colony) were compared with those expected from two null models of random settlement of individuals: in any colony available in the whole population, or within the subpopulation (cluster of colonies) of origin. Our simulations indicate that philopatry to the natal colony was much higher than expected under both null expectations, and observed distances were much lower than expected in the whole population. When individuals were constrained to settle within their natal subpopulation in the simulations, dispersal distances were longer than expected in females, but were higher or lower in males depending on year. Dispersal was not only constrained by the spatial distribution of settlement options, but specific hypotheses arise that can be helpful to design and conduct further research. These results challenge previous interpretations of observed dispersal patterns, which may not reflect free decisions of individuals but environmental or social constraints. We suggest using simulation procedures as a routine to advance in the understanding of dispersal ecology and evolution.