Is floral specialization an evolutionary dead-end? Pollination system transitions in Ruellia (Acanthaceae).

Pollination systems frequently reflect adaptations to particular groups of pollinators. Such systems are indicative of evolutionary specialization and have been important in angiosperm diversification. We studied the evolution of pollination systems in the large genus Ruellia. Phylogenetic analyses, morphological ordinations, ancestral state reconstructions, and a character mapping simulation were conducted to reveal key patterns in the direction and lability of floral characters associated with pollination. We found significant floral morphological differences among species that were generally associated with different groups of floral visitors. Floral evolution has been highly labile and also directional. Some specialized systems such as hawkmoth or bat pollination are likely evolutionary dead-ends. In contrast, specialized pollination by hummingbirds is clearly not a dead-end. We found evidence for multiple reverse transitions from presumed ancestral hummingbird pollination to more derived bee or insect pollination. These repeated origins of insect pollination from hummingbird-pollinated ancestors have not evolved without historical baggage. Flowers of insect-pollinated species derived from hummingbird-pollinated ancestors are morphologically more similar to hummingbird flowers than they are to other more distantly related insect-pollinated flowers. Finally, some pollinator switches were concomitant with changes in floral morphology that are associated with those pollinators. These observations are consistent with the hypothesis that some transitions have been adaptive in the evolution of Ruellia.