Quality of citizen science data and its consequences for the conservation of skipper butterflies (Hesperiidae) in Flanders (northern Belgium)

Citizen science projects have become important data sources for ecologists. However, opportunistic data are not only characterized by spatial and temporal biases, but probably also contain species identification errors, especially concerning morphologically similar species. Such misidentifications may result in wrongly estimated distribution ranges and trends, and thus in inadequate conservation measures. We illustrate this issue with three skipper butterflies (Hesperiidae) in Flanders (northern Belgium) using photographs uploaded with observations in data portals. Ochlodes sylvanus and Thymelicus lineola records had relatively low identification error rates (1% and 11%, respectively), but the majority (59%) of T. sylvestris records turned out to be misidentified. Using verified records only allowed us to model their distribution more accurately, especially for T. sylvestris whose actual distribution area had hitherto been strongly overestimated. An additional field study on T. sylvestris confirmed the species distribution model output as the species was almost completely restricted to sites with verified records and was largely absent from sites with unverified records. The preference of T. sylvestris for unimproved grasslands was confirmed by the negative correlation between its model-predicted presence and elevated nitrogen and ammonia levels. Thus, quality control of citizen science data is of major importance to improve the knowledge of species distribution ranges, biotope preferences and other limiting factors. This, in turn, will help to better assess species conservation statuses and to suggest more appropriate management and mitigation measures.