Spatial modelling of spider biodiversity: matters of scale

In order to choose adequate conservation strategies to face the deterioration of natural ecosystems and the decline of species, it is essential to know the spatial distribution of diversity. Here, we use predictive modelling in spiders, which is a group of highly diverse generalist predators that show a great potential as diversity indicators. We built a predictive model of spider species richness within a protected area assessing those environmental factors that have the strongest effect in the distribution of spider species richness. Our results show a strong relationship between spider species richness and landscape descriptors of land cover. We also assessed the importance of the spatial scale to identify patterns of spider diversity and we selected the optimal spatial scale for species richness and composition by a multiscale approach. We found that this relationship in spiders occurs at relatively fine scales, i.e., 220 x 220 m. The multiple linear regression model at the optimal scale explained 82% of the total variance in species richness. We used the Jackknife procedure to validate the model and we obtained a predictive map of spider richness by extrapolating the model to the entire range of the protected area. Our results show that predictive modelling is a useful tool to estimate the spatial patterns of diversity in a widespread group of arthropod generalist predators.