Six drought-sensitive British butterflies fell less after the 2003, 2018 and 2022 dry summers than after 1995, and semi-natural land cover did not reduce the later falls

Question: Oliver et al. (2015) used the 1995 drought and UK Butterfly Monitoring Scheme (UKBMS) data to identify six drought-sensitive butterfly species. They found that more semi-natural habitat within 3 km reduced the 1995 fall, and they projected local extinctions on the assumption that each future 1995-like drought has the 1995 effect. We tested these results out of sample with the 2003, 2018 and 2022 summers, which had England April-September aridity equal to or above 1995 (4.50, 4.85, 5.50 against 4.52). Method: UKBMS site indices for 37 species on Great Britain transects (1985-2024), Met Office and E-OBS aridity, and ESA WorldCover 2021 land cover within 3 km of 2,719 transects. The main measure is the change in log index from the drought summer to the next summer, compared between the six species and the other 31 species, with permutation tests and fixed-effect models. Results: the six species fell more than other species after 2003 (-0.29 log units, p = 0.039) and 2018 (-0.46, p = 0.006), but not clearly after 2022 (-0.11, p = 0.27). These contrasts were 15-62% of the 1995 contrast (-0.75). Species drops after each later drought correlated with 1995 drops (Spearman 0.38-0.49). The relative response to aridity was stable (-0.09 log units per index unit in 1985-2013 and 2014-2023), but 1995 was far below this relation. More semi-natural land cover reduced the 1995 fall of the six species (+0.10 per SD, SE 0.035) but not the later falls (-0.009, SE 0.015). An edge effect after the later droughts also occurred in the other species, and recovery was not faster at sites with less edge. From 1995 to 2024 the six species did not decline more than the other species (-0.54 against -0.60 log units), although they declined faster at high-edge sites. Limitations: only four drought events; land cover from 2021 with improved grassland in the grassland class; counts can increase in sunny summers; transects are not a random sample; some test choices followed an exploratory table. The evidence does not support the assumption that each 1995-like drought repeats the 1995 effect.

butterflies · drought · UKBMS · habitat fragmentation · semi-natural habitat · out-of-sample test · extreme weather · aridity index · E-OBS · ESA WorldCover · population collapse · climate adaptation · Great Britain · replication