Extensive population-level sampling reveals clinal variation in (R)-(-)-linalool produced by the flowers of an endemic evening primrose, Oenothera harringtonii.

The study of floral trait diversity has a long history due to its role in angiosperm diversification. While many studies have focused on visual traits including morphology and color, few have included floral scent despite its importance in pollination. Of the studies that have included floral scent, sampling has been limited and rarely explores variation at the population level. We studied geographic variation in the flowers of Oenothera harringtonii, a rare plant endemic to a vulnerable shortgrass prairie habitat, whose population structure and conservation status are well studied. The self-incompatible flowers of O. harringtonii open at dusk, produce nectar and a strong fragrance, and are pollinated by hawkmoths. We collected floral trait (morphology, scent chemistry and emission rates) data from 650 individuals from 19 wild populations to survey floral variation across the entire range of this species. Similarly, we collected floral data from 49 individuals grown in a greenhouse common garden, to assess whether variation observed in the field is consistent when environment factors (temperature, watering regime, soil) are standardized. We identified 35 floral volatiles representing 5 biosynthetic classes. Population differentiation was stronger for floral scent chemistry than floral morphology. (R)-(-)-linalool was the most important floral trait differentiating populations, exhibiting clinal variation across the distribution of O. harringtonii without any correlated shifts in floral morphology. Populations in the north and west produced (R)-(-)-linalool consistently, those in the east and south largely lacked it, and populations at the center of the distribution were polymorphic. Floral scent emissions in wild populations varied across four years but chemical composition was largely consistent over time. Similarly, volatile emission rates and chemical composition in greenhouse-grown plants were consistent with those of wild populations of origin. Our data set, which represents the most extensive population-level survey of floral scent to date, indicates that such sampling may be needed to capture potentially adaptive geographic variation in wild populations.