How flower colour signals allure bees and hummingbirds: a community-level test of the bee avoidance hypothesis.

Colour signals are the main floral trait for plant-pollinator communication. Owing to visual specificities, flower visitors exert different selective pressures on flower colour signals of plant communities. Although they evolved to attract pollinators, matching their visual sensitivity and colour preferences, floral signals may also evolve to avoid less efficient pollinators and antagonistic flower visitors. We evaluated evidence for the bee avoidance hypothesis in a Neotropical community pollinated mainly by bees and hummingbirds, the campo rupestre. We analysed flower reflectance spectra, compared colour variables of bee-pollinated flowers (bee-flowers; 244 species) and hummingbird-pollinated flowers (hummingbird-flowers; 39 species), and looked for evidence of bee sensorial exclusion in hummingbird-flowers. Flowers were equally contrasting for hummingbirds. Hummingbird-flowers were less conspicuous to bees, reflecting mainly long wavelengths and avoiding red-blind visitors. Bee-flowers reflected more short wavelengths, were more conspicuous to bees (higher contrasts and spectral purity) than hummingbird-flowers and displayed floral guides more frequently, favouring flower attractiveness, discrimination and handling by bees. Along with no phylogenetic signal, the differences in colour signal strategies between bee- and hummingbird-flowers are the first evidence of the bee avoidance hypothesis at a community level and reinforce the role of pollinators as a selective pressure driving flower colour diversity.

Flower colour within communities shifts from overdispersed to clustered along an alpine altitudinal gradient.

Altitudinal gradients are interesting models to test the effect of biotic and abiotic drivers of floral colour diversity, since an increase in UV irradiance, decrease of pollinator availability and shifts from bee- to fly-pollination in high relative to low altitudes are expected. We tested the effect of altitude and phylogeny, using several chromatic and achromatic colour properties, UV reflectance and pollinators' discrimination capacity (Apis mellifera, Bombus terrestris, Musca domestica and Eristalis tenax), to understand the floral colour diversity in an alpine altitudinal gradient. All colour properties were weakly related to phylogeny. We found a shift from overdispersed floral colours and high chromatic contrast with the background (for bees) in the low altitude, to clustered floral colours (UV and green range for bees and flies) and clustered chromatic and achromatic properties in the high altitude. Different from flies, bees could discriminate floral colours in all altitudinal ranges. Low altitudes are likely to exhibit suitable conditions for more plant species, increasing competition for pollinators and floral colour divergence. Conversely, the increase in UV irradiance in high altitudes may filter plants with specific floral UV-reflectance patterns. Overall, floral colour diversity suggests that both biotic (pollinator fauna) and abiotic (UV irradiance) drivers shape floral communities, but their importance changes with altitude.