A design-based view of species richness estimation in environmental surveys

In this work, the estimation of species richness is approached from a design-based perspective, considering the probabilistic sampling of species and checking the performance of the estimators automated in the SPADE software. As shown theoretically and by a simulation study, these estimators are affected by a massive negative bias. To reduce the underestimation of species richness, data integration is attempted, by exploiting the list of rare species compiled by purposive surveys. Richness estimation is then performed on the residual community of species not in the list, and a bootstrap mean squared error estimator is applied. A simulation study and the application to four case studies produce encouraging results.