Landscape genetics is a ten years old discipline combining population genetics and landscape ecology. This approach is of a major interest for understanding population functioning, especially in the context of an increasing demand of environmentally-friendly agriculture and forestry. Both conservation biology and agro-ecology, including plant protection against pests and diseases, call for a better understanding of ecological continuities. In comparison with classical population genetics approaches, landscape genetics presents the advantage to explicitly take into account the role of spatial heterogeneities in driving gene flows as well as new insights in dispersal. Here, we review the approach and provide an illustration for a forest pest, the pine processionary moth (PPM). We designed a sampling grid with a mesh size of 16 km covering about 60000 km2 in the southern part of the Paris basin where the moth has colonized wide territories during the last twenty years. This region consists of different land uses, including forests with native broad-leaved trees and exotic conifers, wetland and meadow areas, wide open-fields and large urbanized territories. If we only consider the distribution of the host tree resources, the different habitats of the PPM could be grouped into two main categories: forest lands with large artificial pine stands and non-forest lands with small stands and scattered ornamental pines. Previous field observations suggested that moth expansion could have been accelerated in areas where host trees are scarce whereas large forest stands could have retained philopatric females. We modelled the distribution of the potential host trees using our own inventories of TOFs and forest inventory data from the French institute of geographic and forestry information (IGN). Assuming a philopatric behavior of the moth when large resources are available, this allowed to calculate not only the Euclidian geographic distances between all the cells of our sampling grid, but also the least cost-path distances taking into account the main landscape features for the moth. Moreover, five individuals were collected in each cell and genotyped for eleven microsatellite markers in order to assess the spatial structure of genetic data at a fine scale. The comparison between the matrices of Euclidian and landscape distances and the matrices of genetic distances revealed that gene flow pattern was influenced by the landscape features assumed to play a role. Our results highlight the role of ornamental trees in the expansion of the PPM and call to take into consideration non-forest lands and ‘invisible’ trees in forest health management.