Great Expectations: Correlations between Pollinator Assemblages and Floral Characters in Lamiaceae

Premise of research. Pollination syndromes are used to infer pollinators from floral attributes when field observations of floral visitors are lacking. Such inferences are often found to be incorrect or too generalized. To investigate the relationship between pollinators and observable differences in a genus of Lamiaceae with flowers that appear ornithophilous and entomophilous, we experimentally quantified floral morphology and pollinator assemblages in a phylogenetic perspective.Methodology. Floral morphology was examined using multivariate analysis of geometric morphometrics (corolla shape described by landmarks), linear dimensions, and nonmetric attributes for 17 species of Prostanthera. Pollination effectiveness was determined through censuses of floral visitors and interactions by pollinators with flowers, and multivariate statistics were used to examine correspondence between these data and nonlandmark characters. Ancestral state reconstruction was performed using geometric morphometric data and a phylogenetic tree.Pivotal results. Three floral morphologies were identified in Prostanthera, each of which was associated with a unique pollinator assemblage and corroborated by our detailed field observations. A higher abundance of bees visited flowers with the ancestral (bee) floral type, which had a long corolla, porrect adaxial lobes, and long anther appendages. Visits by birds were more frequent to the derived bird floral type, characterized by red or green flowers with larger nectar volumes, longer and narrower corollas, and a reflexed abaxial lobe. Another derived lineage attracted a higher abundance of generalist flies and beetles (beetle+fly floral type). This lineage had flowers with a shorter corolla tube, reflexed adaxial lobes, and shorter anther appendages.Conclusions. When pollinator frequency was used as a proxy for pollinator effectiveness, the correspondence between the floral types and pollinator assemblages supported traditional concepts of bird and insect pollination syndromes. Intermediates between these floral types also corresponded with phylogenetic position and pollinator assemblage. Linear measurements of floral attributes were just as reliable as landmark data in predicting pollinator assemblages. By combining a molecular phylogeny with the above data, evidence indicated that the bird and beetle+fly floral types diversified from an ancestral bee floral type, demonstrating at least one evolutionary specialization for birds, as well as another shift toward a more generalized pollination syndrome. This combination of techniques presents a potential application with which to investigate the evolution of pollination.