Lithologic data improve plant species distribution models based on coarse-grained ocurrence data

Abstract

The aim of this study was to assess the improvement of plant species distribution models based on coarse-grained occurrence data when adding lithologic data to climatic models. The distributions of 40 woody plant species from continental Spain were modelled. A logistic regression model with climatic predictors was fitted for each species and compared to a second model with climatic and lithologic predictors. Improvements on model likelihood and prediction accuracy on validation subsamples were assessed, as well as the effect of calcicole–calcifuge habit on model improvemenClimatic models had reasonable mean prediction accuracy, but adding lithologic data improved model likelihood in most cases and increased mean prediction accuracy. Therefore, we recommend utilizing lithologic data for species distribution models based on coarse-grained occurrence data. Our data did not support the hypothesis that calcicole–calcifuge habit may explain model improvement when adding lithologic data to climatic models, but further research is needed.