A massive community-science dataset reveals convergent evolution of delayed flowering phenology in North American red-flowering plants

Abstract

The radiation of angiosperms is marked by a phenomenal diversity of floral size, shape, color, scent, and reward. Through hundreds of years of documentation and quantification scientists have sought to make sense of this variation by defining pollination syndromes as the convergent evolution of common suits of floral traits across distantly related species that have evolved by selection to optimize pollination strategies. Now, with the rise in popularity of community science platforms like iNaturalist, anyone - not just scientists - can collect data. Thanks to the availability of high-quality community-science datasets, we have an unprecedented number of observations of natural flowering plant diversity. These datasets provide an opportunity to develop new tools and to examine new traits, such as flowering time, that may help further characterize pollination syndromes. Here we test the hypothesis that flowering phenology can also be an important trait of a pollination syndrome; particularly of the "hummingbird flower" syndrome, which is usually characterized by red color, long corolla tubes, and exserted stamens. We produced a novel flower color dataset by using GPT-4 with Vision (GPT-4V) to assign flower color to 11,729 North American flowering plant species from community science photographs. We then mapped those species-specific colors to 1,674,908 citizen scientist observations of flowering plants. We demonstrate constrained flowering time in the eastern U.S. for species with red or orange flowers relative to species with flowers of other colors. Importantly, the onset of flowering corresponds closely to the arrival of hummingbirds. Our findings reveal an opportunity to expand the suite of traits included in pollination syndrome and suggest that the hummingbird pollination syndrome can include flowering phenology. Our methods demonstrate an effective pipeline for leveraging the enormous amount of community science data by extracting valuable information about patterns of trait variation.