Project R-7246

A statistical framework to correct for preferential sampling in biodiversity studies (Research)

Recently, the evaluation of biodiversity has become increasingly important, since species are disappearing at alarming rates. To better understand spatio-temporal trends in species abundances, the potential of citizen science has gained a lot of attention. However, crowd-sourced data introduce statistical challenges, since sampling locations are often not chosen randomly, but preferentially, especially when non-professionals collect data. For example, locations that have a large abundance of a species of interest might be selected more often. This causes standard statistical procedures to be inappropriate for data analysis, since they are based on the assumption that data are sampled at randomly chosen locations. In this project, a statistical framework to analyse preferentially sampled spatio-temporal data is investigated to evaluate biodiversity in the province of Limburg (Belgium), assess interactions between species and their habitats and investigate changes in the presence of endangered fauna and flora and rare visitors. LIKONA's (Limburgse Koepel voor Natuurstudie) presence/absence and presence only data on faunal and floral groups in Limburg, mainly collected by volunteers over a period of thirty years, will be used to develop the proposed methods. My goal is to investigate spatio-temporal trends in the distribution of key species, using models that correct for preferential sampling, in order to set up guidelines for nature conservation in Limburg and beyond.

01 October 2016 - 30 September 2019