How well can models predict changes in species distributions? A 13-year-old otter model revisited

摘要: Species distribution and ecological niche models are increasingly used in biodiversity management conservation. However, one thing that is important but rarely done to follow up on  the predictive performance of these over time, check if their predictions fulfilled and  maintain accuracy, or they apply only set which were produced. In 2003, a model Eurasian otter ( Lutra lutra ) Spain was published, based on results a  country-wide survey published 1998. This built with logistic regression otter  presence-absence UTM 10 km 2 cells diverse  nvironmental, human spatial variables, selected according statistical criteria. Here we evaluate this against of  most recent survey, carried out decade later after significant expansion area country. Despite time elapsed evident changes species’  distribution, maintained good capacity, considering both discrimination calibration measures. Otter did not expand randomly simply towards vicinity areas,  specifically areas predicted as favourable by data from 10  years before. corroborates utility models, at least medium  term when made robust methods relevant predictor variables.