What governs protein content of pollen: pollinator preferences, pollen–pistil interactions, or phylogeny?

Pollen ranges from 2.5% to 61% protein content. Most pollen proteins are likely to be enzymes that function during pollen tube growth and subsequent fertilization, but the vast range of protein quantity may not reflect only pollen–pistil interactions. Because numerous vertebrate and invertebrate floral visitors consume pollen for protein, protein content may influence floral host choice. Additionally, many floral visitors pollinate their host plants. If protein content influences pollinator visitation, then pollinators are hypothesized to select for increased protein content of host plants. We analyzed or gleaned from the literature crude pollen protein concentrations of 377 plant species from 93 plant families. Using this database, we compared pollen protein concentration with (1) pollination mode, (2) pollen collection by bees, and (3) distance from stigma to ovule, after accounting for phylogeny through paired phylogenetic comparisons and a nested ANOVA including taxonomic rank. We found that pollen protein concentrations were highly conserved within plant genera, families, and divisions. We found that bees did not collect pollen that was unusually rich in protein, whether they pollinated or merely robbed their host plant. Plant species with vibratile pollination systems, which require visitation by pollen‐collecting bees in order to transfer pollen, tended to have very protein‐rich pollen, but it was not clear whether this was due to plant enhancement of pollinator rewards or to the possession of very small pollen grains. We found that zoophilous species were not statistically richer in pollen protein than anemophilous species after accounting for phylogeny, although the three most species‐rich anemophilous clades surveyed were generally poor in protein. Plant genera hosting specialist pollen‐collecting bees did not have particularly protein‐rich pollen. Both mass of protein per pollen grain and pollen grain volume were correlated with stigma–ovule distance. We suggest that the need for growing pollen tubes probably plays a more important role in determining pollen protein content than rewarding pollinators.