Figure 2 | Effect of standardised values of habitat suitability (panels a, c, e, g, and i) and recorder effort (b, d, f, h, and j) on the presence of JTM for GBIF data (a and b), iNaturalist (c and d), Flickr (e and f), and Instagram (g and h). 0 on the y axis refers to a pseudoabsence, 1 refers to presence of JTM. Lines predicted from GLMMs with year as random effect, however, due to a low variance explained by year (mean standard deviation = 0.020), only the average effect across years was plotted per panel. Grey area denotes 95% confidence interval.
Even though the geographic background for the HSMs did not include the species’ entire range, this did not affect our results, since all post-2009 JTM records are found within climate conditions that is analogous to the historical range (Figure S5). GBIF-calculated suitable habitat across the study region was relatively consistent between years, with notable exceptions in the UK, Republic of Ireland, Italy, and Denmark (Figure 2 and S6). The mean (± standard deviation) AUC of HSMs was 0.743 (± 0.057) and mean Boyce Index was 0.367 (± 0.246).