Results
We found, on average, no systematic shift in any direction across the species responses (Fig. 3). Among the 289 species studied, 45.3% of species expanded their NRB towards the north and 27% of species advanced their phenology (Fig. 3). By contrast, 40.8% of species contracted their NRB towards the south and 35.6% delayed their phenology. This lack of a systematic directional shift was evidenced by the fact that the estimates of the intercept-only PGLS -models were not significantly different from zero. We also found no difference in the estimates between the two main taxonomic groups, butterflies and moths (Table S2).
Almost half of the species (46.7%) responded according to hypothesis 1, i.e. they either shifted their NRB northwards or their phenology earlier, but not both. A minority of the studied species (12.8%) responded according to hypothesis 2, i.e. they shifted both their NRB northwards and phenology earlier. Finally, 40.5% of the species showed no adaptive response, i.e. they neither shifted their NRB northwards nor phenology earlier (hypothesis 0). Instead, the NRBs remained stable or shifted southwards and their phenology remained stable or delayed.
More than half of the studied species (61.5%) showed positive or stable population trends, but on average there was no systematic trend for neither positive nor negative population trends across the studied species (Fig. 3). Nevertheless, population trends differed between species that responded differently in NRB and phenology shifts. Species that advanced their phenology showed more positive (although insignificant) population trends over the study period than those that delayed or did not change their phenology (Table 2 - model a). Species that shifted their NRBs further north (>20 km) showed significantly stronger positive population trends compared to other species (Table 2 – model b; Fig. 4a). In addition, species that advanced their phenology tended to move their NRB more towards the north, but this effect was not significant (Table 2 – model c). The positive effect of a northwards shift in NRB on population trends in model b) was also mirrored in model d) which indicates that both combined responses including a northwards shift, no matter how the species reacted phenology-wise, showed stronger positive population trends (Table 2 – model d). This effect was stronger for species that also advanced their phenology. Thus, species able to utilize a combined response as postulated by Hypothesis 2 (both northwards shift of NRB and advance in phenology) showed significantly stronger population trends (Table 2 – model e; Fig. 4b). An ability to utilize either of the responses (as postulated by Hypothesis 1) showed, on average, lower but also positive population trend, but this effect was not significant. The species that were not able to utilize either of the presumed adaptive responses (Hypothesis 0) showed the lowest, and on average negative population trends (Table 2 – model e; Fig. 4b).
None of the four life-history traits tested showed a significant connection with population trends. Overwintering stage was the only trait that had an effect on shift in phenology and on NRB (Fig. S6; Table S3). Species overwintering as adults were more likely to advance their phenology while species that overwinter as pupae tended to retreat their NRB towards the south. However, due to imbalance in the number of species representing different host plant-use categories, we had combined species feeding on lichen and fungi into the specialist group (8 species; 2.7% of studied species: see Methods ). In an additional PGLS analysis treating lichen and fungi feeders as a separate group, the eight species that feed on lichen and fungi showed a significant shift in their NRB further north (t= 1.97; p<0.05) and also had more positive population trends (t= 3.93; p<0.001; Fig. S7).