Future Directions
These results indicate that, with some caveats, there is no fundamental barrier that prevents the prediction of population-level flowering timing and duration from specimen-based phenoclimate models. Further, our results show that little more data are needed than have been used by phenoclimate models predicting simple mean (or median) phenological dates (Park and Mazer 2018, Ramirez-Parada et al. 2022). However, we have also identified certain phenological modalities, such species that flower close to the start and end of the growing season, where inferences from collections should be examined cautiously.
Although our simulations were conducted on plant flowering phenology, the underlying results may apply to the development of phenoclimate models of any taxon whose phenology can be assessed from herbaria or other natural history collections data. While the accuracy of those models was not explicitly tested, similar methods have already been used to evaluate the activity period of bee species across the Northeastern United States (Dorian et al. 2022). Thus, widespread assessment should be possible of the effects of climate change on many other taxa and on synchrony among co-occurring plant species, and plants and their pollinators, pests, and frugivores.