Response of Soil Fauna to Landscape Heterogeneity: Determining Optimal Scales for Biodiversity Modeling

The concept of scale is inherent to ecological processes that occur over landscapes. To address this issue, we examined the response of soil fauna (Collembola) to landscape heterogeneity at different spatial scales. We derived spatial features from satellite images of the Massif of Arize in the French Pyrenees. We developed a multiscale approach to test the effect of scale of the landscape pattern on species abundance and biodiversity, with a Mantel test and stepwise multiple linear regression analysis, respectively. We distinguished the effect of the "regional" habitat (landscape descriptors) on soil fauna, which is a spatially explicit concept, and that of the "local" habitat (topography and forest type). We carried out a variance-partitioning analysis of species composition and richness to assess the relative contribution of regional and local habitat. The effect of the spatial structure on species abundance was especially strong at a well-defined scale of 48 ha. The surrounding environment that primarily structured the soil fauna assemblages was vegetation cover, measured as the mean of a radiometric vegetation index. Our results show that regional habitat can play an important role in structuring species composition, which is frequently hidden when the scale component is not taken into consideration. Further, we modeled endemic Collembolan richness based on a combination of landscape features at two spatial scales (2.25 ha and 292 ha). The landscape descriptors accounted for 72% of endemic richness variance, although within that variation 41 % is shared with local habitat and spatial terms. We used the linear regression model to extrapolate endemism scores over the landscape. Such spatial models could be powerful tools for the selection and delimitation of high-biodiversity areas in conservation policies.