The assumption that flowers readily evolve specializations for pollination by particular animals has been central to a standard view of pollinator-mediated adaptive divergence in angiosperms. Stebbins’ Most Effective Pollinator Principle (MEPP) formalized this assumption in proposing that a plant should always evolve specializations to its most effective pollinator. I argue that the MEPP and its corollaries are unsupported by basic models of phenotypic selection which predict that a plant should evolve greater specialization to a particular pollinator when the marginal fitness gain exceeds the marginal fitness loss from becoming less adapted to all other pollinators. Differences in pollinator effectiveness are neither necessary nor sufficient to predict specialization. Differences in effectiveness certainly can foster floral specialization to the most effective pollinator in some cases, but when adaptation to a relatively ineffectivepollinator requires little loss in the fitness contribution of a more effective pollinator, plants may exhibit striking specializations for the less effective pollinator. Recognizing that the effectiveness of pollinators is not tightly coupled to their importance in selecting for phenotypic novelty may resolve the mismatch between floral features that appear to represent clear evolutionary responses to specific pollinators and patterns of flower visitation that often seem generalized. To shed light on agents of selection and the adaptive value of floral traits I argue that we must go beyond measures of pollinator effectiveness to investigate pollinator-mediated fitness trade-offs over a range of floral pheno-types.
