LAUSR

Pollinator-mediated selection is expected to constrain floral color variation within plant populations, yet populations with high color variation (at least in human visual space) are common in nature. To explore this paradox, we collected floral reflectance spectra for 34 populations of 14 putatively bee-pollinated plant species of north-central New Mexico, USA, and translated them into three different animal visual spaces. We asked, (1) is intrapopulation variation in flower color constrained to be lower than the discrimination threshold of the putative dominant pollinators? And, (2) is perceived intrapopulation variation in flowers higher for non-pollinating animals than it is for the presumed dominant pollinator group? We found evidence consistent with some pollinator-imposed constraints on floral color in our species, with the majority (70.6%) of populations having > 95% of flower–flower or flower-centroid comparisons (where the centroid represents the mean phenotype) estimated to be visually indiscriminable to bees. We also found that perceived floral color variation was significantly greater for two non-pollinating groups—birds and humans—than for bees. Our results suggest that a large portion of human-perceived floral color variation within populations persists because it is effectively invisible to pollinators. In this scenario, human-perceived color may evolve neutrally (via drift) or via indirect selection on correlated characters such as herbivore- or drought-resistance, consistent with previous studies identifying non-pollinator agents of selection on flower color.