Evidence for reward costs in two bee-pollinated Penstemon

Floral rewards are generally neglected in studies of natural selection on flowers, despite their importance in attracting pollinators. While evolution of rewards is evident from a phylogenetic and pollination system perspective, estimating selection on rewards is critical to understand how they are likely to evolve in the future. Therefore, we examined natural selection in two bee-pollinated species to compare selection on rewards between the species and among floral traits. Using Penstemon mensarum and P. strictus , we estimated phenotypic selection on several floral traits, including nectar and pollen rewards. We measured selection on plant height, flower number, petal width, corolla tube gape width, total corolla tube length, and proximal tube length, as well as rewards, to compare selection on these floral traits to nectar volume and pollen number. We also determined if any floral traits were honest signals of reward. Selection on floral traits generally differed between the two bee-pollinated species. Notably, there was selection to reduce nectar in P. strictus and stabilizing selection on pollen number for both species. We found little evidence for honest signalling of rewards or correlational selection on signals and traits, although our sample size limits these tests. Despite strong positive correlations in corolla tube dimensions, there was selection for longer corolla tubes but shorter proximal tube in P. strictus . Higher floral rewards were related to reduced female fitness in these species, suggesting costs for reward production. For nectar this pattern was linear but for pollen it was only the highest pollen grain numbers that showed a cost.