Predicting the suitability area of heath alliances over France using open-source data

Abstract While fine-scale maps of vegetation alliances are required for conservation, their distribution is usually known at broad scales. Open-access field and spatial data are increasingly available worldwide, but their contribution to modelling vegetation alliances remains to be assessed. This study aimed to map the suitability area of six heath alliances distributed along ecological gradients throughout France. We used 20 broad and local variables derived from WorldClim, the European Union DEM, SoilGrids, and the Global Wind and Solar Atlases that describe bioclimatic and environmental conditions. For each alliance, two nested MaxEnt models were calibrated and validated using archive field data: one at a broad scale (1000 × 1000 m) to define its bioclimatic area, and the second at a local scale (25 × 25 m) to predict its environmental area. The results showed complementarity of the variables used to discriminate the alliances, and the accuracy of spatial modelling at broad (AUC = 0.83–0.99) and local scales (AUC = 0.91–0.99). They also highlighted the gradients of continentality and temperature that differentiate the alliances. These new maps provide additional knowledge about alliance distribution areas and could support natural vegetation conservation at a national scale.