Unequal pollinator benefits from partial mowing regimes in a field experiment

Studies suggest that partial mowing, where unmown refuges are left for insects after mowing, promotes arthropods in semi‐natural grasslands. However, studies comparing community development in refuges to unmown controls are scarce, particularly for insect pollinators. As competition for floral resources is a central driver in pollinator community composition, the potential increase in pollinator density within unmown refuges is likely to filter ecological communities. We here present a partial mowing field experiment on dike grasslands to study how pollinator communities develop in unmown strips (refuges) within mown dike locations compared to unmown controls. We collected data on flower visitation frequency (counting 28,553 unique visitation events) and community composition from transect counts (identifying 4530 pollinators to the species level) in three periods: before (four days), directly after (four days) and later (until four weeks) after mowing. Pollinator visitation frequency (+85%) and abundance (+25%) increased in partially mown grasslands directly after mowing, but three weeks later pollinators had abandoned refuges, resulting in lower numbers than in control vegetation. These declines were not equally divided over pollinator groups, with significant non‐ Bombus wild bee abundance and species richness decline (−82% and −70%, respectively), whereas honeybees maintained high abundance. Subsequently, rank‐abundance curves reveal a disproportionate decline of intermediately common and rare species at partially mown sites, with no compensatory appearance of later‐season species, which we did observe at unmown controls. Synthesis and applications: We conclude that, although unmown refuges under partial mowing provide initial relief for insects, they do not substitute unmown vegetation. Our results suggest that they primarily benefit common pollinators, such as honeybees, and might not sustain pollinator diversity as a whole.