Geographically structured phenotypic selection can lead to adaptive divergence. However, in flowering plants, such divergent selection has rarely been shown, and selection on floral signals is generally little understood. In this study, we measured phenotypic selection on display size, floral color, and floral scent in four lowland and four mountain populations of the nectar-rewarding terrestrial orchid Gymnadenia odoratissima in two years. We also quantified population differences in these traits and pollinator community composition. Our results show positive selection on display size and positive, negative, or absence of selection on different scent compounds and floral color. Selection on the main scent compounds was consistently stronger in the lowlands than in the mountains, and lowland plants emitted higher amounts of most of these compounds. Pollinator community composition also differed between regions, suggesting different pollinators select for differences in floral volatiles. Overall, our study is the first to document consistent regional differences in selection on floral scent, suggesting this pattern of selection is one of the evolutionary forces contributing to regional divergence in floral chemical signaling.

Local pollinator adaptation through pollinator attraction was shown in the mountain populations, possibly due to adaptation to pollinating flies. The mountain plants were also observed to receive pollination from a greater diversity of pollinators than the lowland plants. The different floral phenotypes of the altitudinal regions are likely to be the consequence of adaptations to local pollinator guilds.

The crucial role of reproductive isolation in speciation has long been recognized; however, a limited number of studies quantify different isolation barriers and embed reproductive isolation in a phylogenetic context. In this study, we investigate reproductive isolation between the often sympatrically occurring orchid species, Gymnadenia conopsea and G. odoratissima. We examine the phylogenetic relationship between the two species and analyse floral isolation, fruit set and seed viability from interspecies crosses, as well as the ploidy level. Additionally, we quantify interspecies differences in floral signals and morphology. The results suggest that the two species have a sisterspecies relationship. In terms of reproductive isolation, we found complete floral isolation between the two species, but little to no post-pollination isolation; the species also mostly had the same ploidy level in the studied populations. We also show clear distinctions in floral signals, as well as in floral size and spur length. We propose that respective adaptation to shortvs. long-tongued pollinators was the driver of speciation in the here studied Gymnadenia species. Our study supports the key role of floral isolation in orchid speciation and shows that floral isolation is not restricted to highly specialized pollination systems, but can also occur between species with less specialized pollination.

No abstract