Identification of the spatial pattern of biodiversity is one of the requirements for the successful conservation planning. Delineation of boundaries (i.e. detection of areas of sharp change for the variable of interest) is potential approach to describe such patterns. Boundary detection methods are widely used in ecology, and are recently receiving more attention in population genetics (for detection of boundaries to gene flow). Here we introduce an original Wombling-based method for detection and testing the significance of boundaries in which the values of the systemic function are estimated using a local polynomial regression. Wombsoft is implemented as R package and it can be applied to codominant (i.e. microsatellites) or dominant markers (i.e. AFLP), as well as contingency (i.e. species frequency) data on geo-referenced sampling points (or individuals). Detected boundaries can be projected on a map using GIS software and correlated to environmental variables in a landscape ecology perspective. The typical application is illustrated by the analysis of a dataset from INTRABIODIV project.

Biology,Agriculture