Modeling Vertebrate Dispersal Distances: Alternatives to the Geometric Distribution

Abstract

The geometric probability distribution is a common way by which to model frequency distributions of vertebrate dispersal distances. This model has two weaknesses: (1) it requires that the probability of stopping, p, is constant across the range of habitat units, and (2) it provides no mechanism for modeling the maximum dispersal distance. The weaknesses of this model are discussed and a family of alternative models, which allows the stopping probability to change and which takes into account the maximum dispersal distance (dispersal scale) is presented. The new models in their simplest form reduce to the goemetric probability distribution model. The new models are applied to five sets of published field dispersal data to determine the general properties of their fit as compared to the geometric and to assess their potential as better models of dispersal patterns.