Distance metrics for ecological niche modeling: beyond a chi-squared correction

Ecological Niche Modeling benefits from presence-only algorithms mostly due to the lack of absence data from species. In this sense, the Mahalanobis distance stands out, since a clear relationship between distance values and the chi-squared distribution supplies an understandable translation from distances to suitability. Other distances could benefit from this same relationship, but no clear method has been proposed so far. Here, we propose the use of the Empirical Cumulative Distribution Function (ECDF) as an alternative to obtain suitability from distances. We used a virtual ecologist approach to test both distributions against a simulated niche. ECDF suitability had a correlation higher than 0.94 with simulated data, highlighting the potential of this technique to be used with different distance metrics, specially when involving non-normal data. These results can have a deep impact on how we obtain Ecological Niche Models by allowing the use of new presence-only algorithms, which can be used to improve theoretical assessments and pseudoabsence generation.