A survey of bee diversity patterns and bee-plant interactions in agriculture-dominated landscapes of the South Banat region, Serbia

Abstract

Within the framework of EcoStack project, we explored the possibilities to improve the provisioning and stability of pollination services to oilseed rape in southeastern Pannonian part of Serbia, focussing on wild bees. The study region (South Banat) is largely dominated by intensive agriculture, with variable and unevenly distributed fragments of natural or semi-natural habitats. Since oilseed rape is the mid-spring mass-flowering crop managed under annual rotation, maintenance of pollinators is possibly affected by the availability of landscape scale non-crop floral resources which bloom in late-spring/early-summer. Since diversity of wild bees and their trophic interactions with local plant communities were poorly studied in this region, we initiated the baseline study of bee diversity patterns in relation to landscape scale agricultural intensity. We selected 10 sampling sites within landscape sectors (1 km radius) differing in share of annual crops and we grouped them in two broad landscape classes: High share of Agriculture (>90%; HA) and Moderate-to-Lower share of Agriculture (<85%; MLA). Bees were sampled twice per site in the period June – early July 2020, by timed hand-netting on each plant taxon found at the surveyed site visited by at least some bees. We recorded 2,084 bees from 25 bee genera, representing about 46% of the potential regional diversity. Assessment was conducted at 23 plant genera, which were unevenly recorded among landscape classes (9 in HA, 17 in MLA, only 3 found in both). Nevertheless, the bee generic richness was similar among classes (19 in HA, 21 in MLA, 15 found in both; 9.6–10.7 per average site), and Shannon diversity (H’-value) was almost identical. Ten most abundant bee genera comprised >97% of all recorded bees (range: 1.2–34.3%), while remaining 15 were represented with less than 0.5% each. The average bee abundance (per site) was somewhat higher in MLA than in HA, but more variable, with 7 out of 10 most abundant genera being markedly more abundant in MLA; this was likely affected by the unbalanced sampling intensity due to local differences in plant availability. Further indepth and species-level analysis (ongoing) should enable a more insightful, interactionbased pollinator management.