Modeling large-scale dispersal distances

Abstract

We present an approach to analyze dispersal distance data. This approach allows one to take into account accuracy of the recorded dispersal distances. Three distributions were used, all assuming continuous space; a maximum likelihood approach was used for parameter estimation and model selection. Numerical simulations showed that our method is statistically consistent since it selected the correct model with increasing frequency when sample size increased. Ringing data on two species of tits ( Parus caeruleus and Parus major) in Britain and Ireland were used to illustrate the potentialities of our method. In both species, adults dispersed significantly further than juveniles. The differences between species within an age-class were not statistically significant. In all species and age-classes, the model finally selected was the one assuming a heavy-tailed half-Cauchy distribution where long-distance dispersers are predicted to be more frequent than in the exponential model. The proposed methodology can potentially be applied to any organisms, and the model selection procedure can be used with any model of the distribution of dispersal distances (DDD). Several extensions are presented in the discussion, such as generalized linear modeling of the dispersal parameters, or interfacing with capture–recapture models.