RESUMEN
Wildflower mixes are often planted around field margins to provide forage for pollinators. Although seed for these mixtures is often wild-sourced, for species where agricultural cultivars are available, for example red clover (Trifolium pratense), cultivars can also be included. Previous evidence suggests that plant genetic background can have a strong influence on plant-arthropod interactions and therefore the provenance and genetic background of the plants included in wildflower mixes could impact plant-pollinator interactions. We tested the performance of five individual T. pratense cultivars against two commercially available wild-sourced T. pratense populations in terms of their ability to attract potential pollinator species (focusing on bumblebees) and their floral traits using greenhouse and garden experiments. The main bumblebee observed interacting with T. pratense was Bombus pascuorum and we found no difference in the absolute number of B. pascuorum visiting the cultivars or wild populations. However, we found variation among cultivars and between wild populations in their ability to attract bumblebees, which seems to be related to their relative investment in different floral traits. There was a positive relationship between biomass and number of inflorescences produced by the wild populations of T. pratense, which was not apparent for the cultivars. This suggests that artificial selection on the cultivars has changed the G-matrix of correlated traits. We show that agricultural cultivars of T. pratense can be as effective as wild populations at attracting pollinators such as bumblebees, but that the genetic background of both cultivars and wild populations can have a significant impact on the attractiveness of the plant to pollinators. We also show divergence in the correlated traits of T. pratense cultivars and wild populations that could lead to outbreeding depression if the plants interbreed.