Bias Correction in Species Distribution Models: Pooling Survey and Collection Data for Multiple Species

摘要

Presence-only records may provide data on the distributions of rare species,but commonly suffer from large, unknown biases due to their typically haphazardcollection schemes. Presence-absence or count data collected in systematic,planned surveys are more reliable but typically less abundant. We proposed a probabilistic model to allow for joint analysis ofpresence-only and survey data to exploit their complementary strengths. Ourmethod pools presence-only and presence-absence data for many species andmaximizes a joint likelihood, simultaneously estimating and adjusting for thesampling bias affecting the presence-only data. By assuming that the samplingbias is the same for all species, we can borrow strength across species toefficiently estimate the bias and improve our inference from presence-onlydata. We evaluate our model's performance on data for 36 eucalypt species insoutheastern Australia. We find that presence-only records exhibit a strongsampling bias toward the coast and toward Sydney, the largest city. Ourdata-pooling technique substantially improves the out-of-sample predictiveperformance of our model when the amount of available presence-absence data fora given species is scarce. If we have only presence-only data and no presence-absence data for a givenspecies, but both types of data for several other species that suffer from thesame spatial sampling bias, then our method can obtain an unbiased estimate ofthe first species' geographic range.