Spatial genetic structure of two forest plant metapopulations in dynamic agricultural landscapes

In many rural landscapes, woodland typically consists of small fragments of different habitat quality and age, embedded in a more or less intensively managed agricultural matrix. Genetic consequences on forest plant populations remain largely unknown. Here we explore whether genetic diversity, spatial genetic structure (SGS) and underlying gene flow are influenced by current and past permeability of the matrix in fragmented landscapes. We compare SGS of the generalist Geum urbanum and the forest specialist Primula elatior across three agricultural landscapes differing by their composition and management intensity. We use non-spatial and spatially informed approaches based upon multilocus genotypes to detect SGS and evaluate the respective importance of historical and contemporary gene flow. Results suggest that a low matrix permeability tends to disrupt gene flow, reducing SGS in populations of the two species. This effect is stronger for the forest specialist than for the generalist, as the former exhibits both low fecundity and dispersal limitation, and requires a higher habitat quality to maintain metapopulations. The current and past permeability of the matrix to gene vectors (seed dispersers and pollinators) explains the apparent unimodal relationship between SGS intensity and genetic connectivity. These results show the susceptibility of forest plant specialists to fragmentation and highlight the need for conserving the most ancient forest fragments and restoring a high functional connectivity among forest patches within agricultural landscapes.