The population dynamics of most animal species inhabiting agroecosystems may be determined by landscape characteristics, with agricultural intensification and reduction of natural habitats influencing dispersal patterns. Increasing landscape complexity would thus benefit endangered species by providing different ecological niches, but it could also lead to undesired effects in species that can act as crop pests and disease reservoirs. We tested the hypothesis that a highly variegated landscape influences the dispersal, and hence patterns of genetic structure, in agricultural pest voles. Ten populations of fossorial water vole, Arvicola scherman, located in a bocage landscape in Atlantic NW Spain were studied using DNA microsatellite markers and a graph-based model. The results showed a strong isolation-by-distance pattern with a strong and significant genetic correlation at smaller geographic scales, while genetic differentiation at larger geographic scales indicated a hierarchical pattern of up to eight genetic clusters. A metapopulation-type structure was observed, immersed in a landscape with a low proportion of suitable habitats. Matrix scale rather than matrix heterogeneity per se may have an important effect upon determining gene flow, acting as a demographic sink. The identification of sub-populations, considered to be independent management units, allows the establishment of feasible population control efforts in this area. These insights support the use of agro-ecological tools aimed at recreating enclosed field systems when planning integrated managements for controlling patch-dependent species such as grassland voles.

Graph-based landscape model: Habitat patches were represented by polygons of soil occupancy with a 5 m² resolution, extracted from the Land Cover and Use Information System of Spain and analysed in a vector-based geographic information system (GIS). Soil-occupancy polygons were classified into nine categories: meadows, fruit orchards, pastures or grasslands, annual crops, shrubs, eucalyptus plantations, deciduous woodlots, settlements and roads and bodies of water.

Microsatellite genotyping: PCR-amplification of 12 microsatellite loci was carried out in two multiplex panels: panel 1: AV3, AV8, AV11, AV15, AT2 and AT24; and panel 2: AV12, AV14, AV13P, AT9, AT13 and AT22 (Stewart et al. 1999; Berthier et al. 2004, 2005). AV13 reverse microsatellite primer was labelled with a "pig-tail" to increase the accuracy of genotyping. Forward primers were labelled using FAM, NED, PET and VIC fluorochromes.