Observer-oriented approach improves species distribution models from citizen science data

Citizen science platforms are increasingly growing, and, storing a huge amount ofdata on species locations, they provide researchers with essential information to developsound strategies for species conservation. However, the lack of information onsurveyed sites (i.e., where the observers did not record the target species) and samplingeffort (e.g., the number of surveys at a given site, by how many observers, andfor how much time) strongly limit the use of citizen science data. Thus, we examinedthe advantage of using an observer-oriented approach (i.e., considering occurrencesof species other than the target species collected by the observers of the targetspecies as pseudo-absences and additional predictors relative to the total numberof observations, observers, and days in which locations were collected in a givensampling unit, as proxies of sampling effort) to develop species distribution models.Specifically, we considered 15 mammal species occurring in Italy and compared thepredictive accuracy of the ensemble predictions of nine species distribution modelscarried out considering random pseudo-absences versus observer-oriented approach.Through cross-validations, we found that the observer-oriented approachimproved species distribution models, providing a higher predictive accuracy thanrandom pseudo-absences. Our results showed that species distribution modelingdeveloped using pseudo-absences derived citizen science data outperform thosecarried out using random pseudo-absences and thus improve the capacity of speciesdistribution models to accurately predict the geographic range of species whenderiving robust surrogate of sampling effort.