Is variation in flower shape and length among native and non-native populations of Nicotiana glauca a product of pollinator-mediated selection?

Abstract

Due to drastic changes in pollinators between native and invaded habitats, we might expect that pollinator-mediated selection on floral traits of alien plants differ from that in their native ranges. Here, through geometric morphometric tools and phenotypic selection analyses, we examined whether adaptation in flower shape and length occurred in Nicotiana glauca as a response to pollinator selection in contrasting pollination environments. We assessed populations of this plant species in the native range (South America), where plants depend on hummingbird pollination, and in two invaded areas, one where sunbirds act as pollinators (South Africa), and another where nectar feeding birds are absent and reproduction is entirely by autonomous self-pollination (Mallorca, Spain). Corolla length and shape varied significantly among pollination environments. Non-native sites were less variable and their range of variation fell within the native range of variation. Flower length in native populations and in a South African population matched the bill length of their respective pollinators. In contrast with the straight floral tubes in the native range, both non-native areas had significantly curved tubes. Curvature may improve the fit with the curved bills of sunbirds in South Africa (versus straight beaks of hummingbirds) and may enhance self-pollination in Mallorca, but this similarity between invaded areas may equally be due to drift and a shared colonization route. We found spatial variation in selection acting on corolla length but not on corolla shape. Overall, selection patterns were not consistent with floral trait variation. Although some results are consistent with both drift and selection, our study suggests that population divergence in flower shape and length is more likely the result of long-term diversifying pollinator-driven selection, which is difficult to detect by studying a single selection event.