A Unifying Model for the Analysis of Phenotypic, Genetic and Geographic Data

Recognition of evolutionary units (species, populations) requires integrating several kinds of data such as genetic or phenotypic markers or spatial information, in order to get a comprehensive view concerning the dierentiation of the units. We propose a statistical model with a double original advantage: (i) it incorporates information about the spatial distribution of the samples, with the aim to increase inference power and to relate more explicitly observed patterns to geography; and (ii) it allows one to analyze genetic and phenotypic data within a unied model and inference framework, thus opening the way to robust comparisons between markers and possibly combined analyzes.

We show from simulated data as well are real data from the literature that our method estimates parameters accurately and improves alternative approaches in many situations. The interest of this method is exemplied using an intricate case of inter- and intra-species dierentiation based on an original data-set of georeferenced genetic and morphometric markers obtained on Myodes voles from Sweden.

A computer program is made available as an extension of the R package Geneland.