The Effects of False Positive Errors in Species Occurrence Data on the Performance of Species Distribution Models: Case Study on the Wild Duck (Anas platyrhynchos)

Abstract

In ecological research, particularly in species distribution modelling (SDM), the uncertainty resulting from data deficiencies is of recent interest. Beside other things, it includes positional accuracy of species occurrence data. In our study, we investigated the influence of the positional error, caused by false positive detection of species, in species occurrence data on the performance of the models. We used occurrences of Wild Duck (Anas platyrhynchos) mapped in fine scale resolution (300 x 300 meters) and habitat variables derived from the Base map of the Czech Republic. Generalized additive models (GAM) with a stepwise selection procedure were used to select relevant habitat variables. Model performance was evaluated using area under the receiver operating characteristics curve (AUC), sensitivity and specificity. Incorporated positional error led to a reduction in model prediction accuracy, although not enough to reject the models. Vliv nesprávného určení výskytu druhu na modelování distribuce druhů: Případová studie s kachnou divokou (Anas Platyrhynchos)