Major environmental changes due to a great variety of causes (such as anthropic pressure or climate change) are shifting the capacity of ecosystems to harbour vectors and the diseases they transmit. Bluetongue (BT), an arboviral disease of ruminants which recently emerged in the Mediterranean Basin, is a pertinent model to study the conditions enabling an exotic vector to settle in a new environment. This work aims to use remote sensing and landscape analysis to characterise favourable environments for BT and its main vector, Culicoides imicola, at a local scale, in Corsica. The environment related to the vector's biology was characterized within neighbourhoods surrounding BT-infected or BT-free farms (disease approach) and trapping sites where the vector was present or absent (vector approach), using three sizes of neighbourhoods. Environmental characteristics were extracted from a digital elevation model, a hydrographical data base and a SPOT high spatial resolution image. The image was classified to produce a land-cover map from which landscape metrics relative to the spatial structure and composition of the vegetation were extracted. The models of the disease and vector approaches were validated both internally and externally. The results confirm that favourable environments for the vector and the disease can be characterised by non-climatic environmental factors and show the importance of landscape metrics for the three neighbourhood scales. They show that favourable landscapes present a high diversity and an important fragmentation. Latitude, which could reflect proximity to Sardinia, is also related to the presence of both the vector and the disease. Most models have a good discriminating capacity, allowing their use as predictive tools in Corsica and elsewhere, in particular to help focus surveillance activities on French mainland coasts. This works opens up many perspectives among which developing similar approaches for other species of Culicoides implicated in BT transmission, integrating climatic factors, identifying new remote sensing parameters useful to epidemiology and building dynamic vectorial capacity models.
