Establishing wildflower strips has been suggested as an effective measure to promote pollination services, pest control or general insect biodiversity, but little is known about the integration of these different objectives when selecting flower seed mixtures. In ten agricultural landscapes in the Netherlands, we established a wildflower strip (0.4 – 4.9 ha) with half of each strip sown with a mixture targeting longer-tongued pollinators and the other half sown with a mixture targeting shorter-tongued pollinators and natural enemies. We determined establishment success of sown wildflowers and evaluated the attractiveness of the established flower communities to multiple functional groups of flower visitors: bumblebees (long-tongued pollinators), hoverflies (short-tongued pollinators and natural enemies), and butterflies and total flower-visitor richness (indicators of wider biodiversity values). Bumblebees clearly preferred the pollinatortargeted seed mixture and were positively associated with cover of Fabaceae and negatively with Apiaceae. Hoverflies consistently preferred the natural enemy mixture and were positively associated with Apiaceae. The other target groups displayed no clear responses to seed mixture type but instead were associated with local flower richness within strips. Across sites, responses of flower-visitors to sown mixture types did not depend on wildflower strip size, proportion of surrounding semi-natural habitat, or flower variables. However, all flower-visitors except butterflies increased with increasing cover or richness of (sown) flower species across sites. Our results suggest that, although species-rich wildflower strips may benefit several species groups, maximising different objectives involves trade-offs between functional groups that prefer short- or long-corolla flowers. Furthermore, our study suggests that sowing a wildflower mixture does not necessarily result in a vegetation with the same composition as the seed mixture as species may establish poorly or not at all. Selection of flower species for seed mixtures should therefore, in addition to insect target group, take the establishment characteristics of plant species into account.

Abundance and richness of bumblebees, hoverflies, and flower cover and richness were surveyed in ten wildflower strips in the Netherlands in 2013 and 2014. Half of each strip was sown with a flower seed mixture targeting pollinators, the other half was sown with a seed mixture targeting natural enemies. Flower-visitors and flowers were surveyed in five 10 m  x 5 m transects per mixture-type, three times in 2013 (June, July and August) and two times in 2014 (June and August).