Pollination syndromes are the convergent expression of floral traits in unrelated species reflecting specialized interactions between plants and pollinators exerting similar selection pressures. I addressed the controversial claim that pollinator-mediated selection is unlikely to be a major factor underlying floral evolution because plants often have many functionally different floral visitors. Detailed pollination data and pollinator-mediated selection studies are needed to address the notion that specialized plant-pollinator interactions are a major mechanism of floral evolution. I developed statistical methods to measure the importance of pollinators (Chapter 1). I addressed whether floral morphological differences of the related Silene species, S. caroliniana, S. virginica, and S. stellata, correspond to predicted specialized pollination systems (Chapter 2). I asked whether contemporary selection pressures on floral traits were detectable in a population of S. virginica (Chapter 3). I investigated the non-obligate interaction of S. stellata and the moth Hadena ectypa, that pollinates it and uses its immature seed for the development of larval offspring (Chapter 4).

Using my novel methodology (Chapter 1), I demonstrated that S. virginica and S. stellata were specialized on hummingbirds and nocturnal moths, respectively (Chapter 2). S. caroliniana was least specialized with long-tongued diurnal hawkmoth (Hemaris sp) and large bee pollinators (Bombus spp. and Xylocopa virginiana). These results matched predictions based on interspecific differences in Silene floral trait expression and were consistent with the notion that the important pollinators are the major selective agents on floral design. Positive directional but mainly nonlinear hummingbird-mediated phenotypic selection (Chapter 3) on S. virginica floral traits was detected through lifetime fitness components, supporting predictions from the syndrome concept. Flowering date predicted the relative density of H. ectypa and other moth pollinators of S. stellata, and H. ectypa density varied by population and year, which may determine the sign of the H. ectypa-S. stellata interaction. Both curvature and directional selection in S. stellata's floral trait selection surface were context dependent on the intensity of H. ectypa larval fruit predation. Overall pollinators are important sources of selection underlying floral evolution in these Silene, and S. stellata floral evolution is subject to additional selection pressures from H. ectypa larvae.