Predicting soil micro-variables and the distribution of an endogeic earthworm species through a model based on large-scale variables

Abstract

Studies on spatial patterns of distributions of soil dwelling animals have usually relied on soil micro-variables or statistical analyses based on presence/absence data. Geographic Information Systems (GIS) allow easy access to large-scale variables to build species distribution models. In this study, we used MaxEnt to model the distribution of the endogeic earthworm Hormogaster elisae. Significant differences were found between the predicted suitability values of localities where the species was present and those where it was absent, validating the predictive model. Most of the large-scale training variables showed significant correlation with soil micro-variables known to influence the biology of the species, proving the ability of the model to predict (to an extent) soil variables from environmental ones. The methodology could be extended to other soil fauna.