Résumé
Ecological restoration programs are established to reverse land degradation, mitigate biodiversity loss, and reinstate ecosystem services. Following recent agricultural intensification that led to a decrease in flower diversity and density in rural areas and subsequently to the decline of many insects, conservation measures targeted at pollinators have been established, including sown wildflower strips (WFS) along field margins. Historically successful in establishing a high density of generalist bees and increasing pollinator diversity, the impact of enhanced flower provision on wider ecological interactions and the structure of pollinator networks has been rarely investigated. Here, we tested the effects of increasing flower species richness and flower density in agricultural landscapes on bee-plant interaction networks. We measured plant species richness and flower density and surveyed honeybee and bumblebee visits on flowers across a range of field margins on 10 UK farms that applied different pollinator conservation measures. We found that both flower species richness and flower density significantly increased bee abundance, in early and late summer, respectively. At the network level, we found that higher flower species richness did not significantly alter bee species' generality indices, but significantly reduced network connectance and marginally reduced niche overlap across honeybees and bumblebee species, a proxy for insect competition. While higher connectance and niche overlap is believed to strengthen network robustness and often is the aim for the restoration of pollinator networks, we argue that carefully designed WFS may benefit bees by partitioning their foraging niche, limiting competition for resources and the potential for disease transmission via shared floral use. We also discuss the need to extend WFS and their positive effects into spring when wild bee populations are established.