Modelling spatial deficits in citizen science recording: a case study from UK butterflies

Abstract

Citizen science data are increasingly used to infer species distributions and monitor biodiversity at large spatial scales. However, recording effort is often uneven across space, potentially biasing inferences about species’ distributions. We analysed 51,045 butterfly records submitted via the iRecord Butterflies app in Devon and Cornwall, in the southwest of the United Kingdom. We used negative binomial models to examine how land cover and three public access variables influenced the number of butterfly records in two groups of species: wider countryside species (WS) and habitat specialists (HS). Results show that the area of public access land was a consistently important explanatory variable of recording effort across both WS and HS. Footpath density showed a significantly stronger effect on number of records for WS than for HS. Moreover, predicted recording deficits, defined as the difference between model predictions of record density under current and maximum access scenarios, were widespread. Deficits tended to occur in areas predicted to provide suitable butterfly habitat but with relatively low public access, whereas highly accessible areas were more likely to be recorded. Recording deficits for HS are predicted in areas with a high density of semi-natural habitats but low public access; by contrast for WS recording deficits are predicted in built-up areas and gardens, where these common species may be widespread but recorder effort is highly localised. Implications for insect conservation. Improving the spatial representativeness of citizen science data may therefore require complementary strategies: encouraging recording close to residences for WS and securing access to restricted semi-natural habitats for HS. By explicitly quantifying effects of public access, our f3amework moves beyond intuitive explanations and offers empirical, operational guidance for identifying and addressing spatial data gaps, thereby helping to improve the spatial representativeness and reliability of citizen science data. Improving the spatial representativeness of citizen science data may therefore require complementary strategies: encouraging recording close to residences for WS and securing access to restricted semi-natural habitats for HS. By explicitly quantifying effects of public access, our framework moves beyond intuitive explanations and offers empirical, operational guidance for identifying and addressing spatial data gaps, thereby helping to improve the spatial representativeness and reliability of citizen science data.