Turnover and trends in butterfly communities on two British tidal islands: stochastic influences and deterministic factors

Aim Community trends were investigated for two small islands and two local mainland butterfly communities within the UK over a period of 20–30 years. Location Hilbre Island off the Wirral Peninsula at 53.33° N, 3.10° W; Lindisfarne, an island off the Northumberland coast at 56.41° N, 1.48° W; Leighton Moss at 54.08° N, 2.26° W; Wyre Forest at 52.23° N, 2.14° W, UK. Methods Butterfly species data were collected on Hilbre and two mainland sites (Leighton Moss and Wyre Forest) from 1983 to 2006, and on Lindisfarne from 1977 to 2006, as part of the National Habitat Survey, the UK Butterfly Monitoring Scheme and ‘Butterflies for the New Millennium Atlas’ recording. Matrices of associations (Sokal and Michener’s matching coefficient S_SM; resemblance coefficient) were computed between years and subject to non‐metric multidimensional scaling (NMDS) and Mantel tests. The pattern of extinctions and colonizations at sites were examined, their heterogeneity tested by applying a Friedman test to fractional incidences for the same years. Regression analysis (multiple regression and logit regression) was used to relate butterfly numbers and incidences to climate variables, time and previous records. Results Significant community trends based on population counts and species’ incidences were found for all four sites. There was a significant climatic signal for Hilbre; although this was not apparent for the remaining sites, significant associations occurred between records for a number of species and climatic variables at all sites. Substantial turnover of species on the islands was inversely related to numbers of records for species but not to their conspicuousness to recorders. Main conclusions We argue that time trends are widespread in butterfly communities, even for relatively short periods; they are largely generated by stochastic influences rather than by more substantive factors such as climate change. Potential biases in surveying and recording history are shown to be unlikely. A clear climate signal was found only for the small Hilbre Island, for which there was also evidence for the significant influence of colonization capability of individual source species. We conclude that for many species, small islands will be sinks or pseudosinks and their ‘populations’ vulnerable to small changes in source–sink dynamics.