The colors of fleshy fruits play a critical role in plant dispersal by advertising ripe fruits to consumers. Fruit colors have long been classified into syndromes attributed to selection by animal dispersers, despite weak evidence for this hypothesis. Here, we test the relative importance of biotic (bird and mammal frugivory) and abiotic (wet season temperatures, growing season length, and UV-B radiation) factors in determining fruit color syndrome in 3,163 species of fleshy-fruited plants. We find that both dispersers and environment are important, and they interact. In warm areas, contrastive, bird-associated fruit colors increase with relative bird frugivore prevalence, whereas in cold places these colors dominate even where mammalian dispersers are prevalent. We present near-global maps of predicted fruit color syndrome based on our species-level model and our newly developed characterizations of relative importance of bird and mammal frugivores.

In brief, data were derived from publicly available sources (listed in the README file).  To process plant species data, we took data on fruit coloration from Sinnott-Armstrong et al (2018), and then combined those data with occurrence records from GBIF, removing all species with fewer than 20 occurrences.  We classified fruit colors into "contrastive" (black, blue, red, and white) and "cryptic" (green, brown, yellow, orange) syndromes traditionally associated with bird and mammal dispersal, respectively.  Finally, we produced predictor layers for growing season length, UV radiation, and wet season temperatures as well as a biotic variable, relative bird frugivore prevalence.  Relative bird frugivore prevalence values are the proportion of frugivores, per grid cell, that are birds out of frugivorous birds and mammals.  This variable is weighted by diet (e.g., a bird species that is entirely frugivorous has a weight of 1; a mammal species whose diet is 20% fruit has a weight of 0.2).  We also conducted alternative analyses wherein this variable is also weighted by the range size of the bird and mammal species in order to downweight species with very large distributions.

More details can be found in the associated manuscript.

The README file contains a description of the data sets and variable names.