Predicting the conservation status of data-deficient species.

摘要: There is little appreciation of the level extinction risk faced by one-sixth over 65,000 species assessed International Union for Conservation Nature. Determining status these data-deficient (DD) essential to developing an accurate picture global biodiversity and identifying potentially threatened DD species. To address this knowledge gap, we used predictive models incorporating species' life history, geography, threat information predict conservation terrestrial mammals. We constructed with 7 machine learning (ML) tools trained on known status. The resultant showed very high classification accuracy (up 92%) ability correctly identify centers richness. Applying best model species, predicted 313 493 (64%) be at extinction, which increases estimated proportion mammals from 22% 27%. Regions contain large numbers are already priorities, but in areas show considerably higher levels than previously recognized. conclude that unless directly targeted monitoring, classified as likely go extinct without notice. Taking into account may therefore help alleviate data gaps indicators conserve poorly biodiversity.