Using landscape metrics on satellite imagery to assess conservation status of Natura 2000 habitats

International audienceHow to assess conservation status of habitats, and particularly fl oristic diversity at landscape and community scales is still an outstanding issue. Here we propose a framework including remote sensing, calculation of landscape metrics and the use of diversity indicators to achieve this goal. The framework was tested for two consecutive years in meadows of the Loire valley (France). We demonstrated that meadow vegetation can be mapped at the community scale using per pixel supervised classifi cation based on ground truth phytosociological relevés applied onto multispectral imagery. Following the spectral variation hypothesis, we assessed fl oristic diversity at the landscape scale using a Shannon indexcalculated from the vegetation map produced (p Index). Consequently, mapping inter-annual variations of p Index allowed us to detect fl oristic diversity changes. We also demonstrated that some plant communities indicating a decrease of species diversity can be mapped using remotesensing. Thus, we were able to assess the inter- annual variations of species diversity at the community scale and we improved comprehensionof observed variations of species diversity at the landscape scale (p Index). Our framework represents a useful tool for managers, allowing them to assess and monitor plant species diversity at landscape and community scales over large areas