Can we model the probability of presence of species without absence data

摘要: In ecological studies, it is useful to estimate the probability that a species occurs at given locations. The of presence can be modeled by traditional statistical methods, if both and absence data are available. However, challenge most records contain only data, without reliable data. Previous presence-only methods relative index habitat suitability, but cannot actual presence. In this study, we develop background learning algorithm (PBL) successful in modeling conditional simulated species. model trained two completely separate sets: observed Assuming one for ‘prototypical presence’ locations where habitats maximally suitable species, constant calibrate into presence. Experimental results show PBL method performs similarly presence-absence method, significantly better than widely used maximum entropy method. new enables us on environmental covariates Hence, has potential improve geographical distributions