To represent JTM’s climatic niche, we used four climatic variables:
average maximum temperature, coefficient of variation in average maximum
temperature, total precipitation, and coefficient of variation in total
precipitation. These four variables have been found to be dominant
factors in the range shift and migration of other
lepidoptera24,25. Climatic data were all calculated
per year for the flying time of JTM (July –
September)22. Climatic data were gathered from
WorldClim at a 2.5 minute spatial resolution (~21
km2)26,27. We note that the CHELSA
dataset is an improvement on WorldClim, but the differences are very
small within Europe, and within the non-mountainous habitat that JTM
largely occupies. Our goal was to understand the impact of social media
data and recorder effort on biomonitoring of range-shifts, rather than
make the most accurate range-shift prediction possible, and it is highly
unlikely the small, unsystematic, differences between CHELSA and
WorldClim in our study region would affect these results. Other climatic
layers were not included to avoid overfitting HSMs, under-predicting
potential distributions and tolerances under climatic conditions where
species may be underreported28. A fifth environmental
layer, night light, was used to capture the degree of
urbanisation29. Night light data were collected from
December of every year (data from summer months may not be an accurate
representation due to the lighter summers in the northernmost parts of
the study region). Data were collected from the National Centers for
Environmental Information30 and converted to a 2.5
minute spatial resolution (~21 km2) by
averaging. Stray light, lightning, lunar illumination, and cloud cover
are all removed from the average measure of illumination prior to
calculation of averages for each layer. Only data from 2012 onwards were
comparable between years, so for all years prior to 2012, the night
light dataset from 2012 was used.