Skip to main content

Wild bees and hoverflies respond differently to urbanisation, human population density and urban form

Cite this dataset

Persson, Anna; Ekroos, Johan; Olsson, Peter; Smith, Henrik (2021). Wild bees and hoverflies respond differently to urbanisation, human population density and urban form [Dataset]. Dryad.


While urbanisation contributes to global biodiversity declines, flower-rich urban habitats may provide beneficial pollinator habitats. We investigated the potential of urban residential areas to contribute to pollinator diversity by analysing wild bee and hoverfly species richness and composition of species assemblages of summer-active species, sampled in 53 gardens across urban and rural landscapes of Malmö, the regional capital of Sweden’s southernmost county. Species richness differed between urban and rural gardens, and between four urban residential types (ranging from low human density and high vegetation cover, to high human density and low vegetation cover), and taxonomic groups responded differently. Solitary bee species richness was higher in urban than rural gardens, driven by a higher richness in low-density urban gardens compared to both high-density urban gardens and rural gardens. In contrast, bumblebee species richness was higher in rural than urban gardens, whereas differences among the urban types were less clear. Hoverfly species richness was consistently higher in rural gardens than any urban garden type. Species richness of all groups was negatively related to human population density at the landscape scale (radius 500 m), but unrelated to vegetation cover. This indicates that population density affects pollinator habitat quality through associated green space management and design. Rural and urban wild bee species assemblages consisted of different species (significant species turnover), whereas urban hoverfly assemblages were a subset of rural ones (significant nestedness). Species nestedness of hoverflies, but not bees, increased with human population density. We show that urban areas can complement the regional wild bee species pool, mainly caused by large variation in tenure and management at small spatial scales, while urbanisation drives a systematic loss of hoverfly species. We suggest alternatives to improve dense residential areas for pollinators.


We surveyed pollinators in residential gardens of neighbourhoods that differed qualitatively in morphological type relevant to urban planning, as well as quantitatively in vegetation cover and human population density at spatial scales relevant for pollinators. We focused on wild bees (Apoidea) and hoverflies (Syrphidae) but also included soldier flies (Stratiomyidae), because although often not treated as pollinators they have a similar life history to hoverflies.

We surveyed pollinators using pan-traps for 30 days during summer 2017, from 2nd July until 2nd August. Three pan-traps (one yellow, one blue, one white) 15 cm in diameter and 5.5 cm deep, sprayed with UV-fluorescent paint (Sparvar Leuchtfarbe), were placed together on a black plastic tray in each garden in connection to vegetation typical of each garden, clearly visible and without shade or with only light shade for part of the day. Each bowl was half-filled with 50% propylene glycol and a drop of non-scented detergent. Traps were left in gardens, and contained liquid, for the whole period and emptied every 10 days (i.e. three times). Insects were preserved in ethanol until pinned and identified to species by entomologists at the Biological Museum, Lund University. We pooled data per pollinator taxon (bumblebees, solitary bees and hoverflies) at the individual garden level (three pan-traps) and calculated species richness, giving us three data points per garden. One urban site (multifamily open yard) was dropped from analyses because pan-traps were tampered with, leading to reduced sampling effort, resulting in 39 urban gardens. Two rural gardens were not surveyed for flowers because of time constraints, and were thus not included in models where flower data was a predictor, resulting in 12 rural gardens in those models.

We surveyed plants in each garden twice during the study period. Each time we surveyed 50 m2 each of lawn/grass and flowerbeds/ornamental shrubs, depending on the characteristics of each garden. Hence, in total 100 m2 per garden was surveyed. However, in 12 urban gardens less than 50 m2 of either habitat was available, and for these we surveyed the entire habitat whilst assessing the total area surveyed to be able to control for area in analyses. We noted all flowering entomophilous plant species and estimated the total number of “flower units”, i.e. an easily countable structure that requires pollinators to fly between units.

Usage notes

Land use gradients (proportion vegetation) and human population per sealed surface are only available for rural sites. Hoverflies include both Syrphidae and Stratiomyidae (soldierflies), as they are both flower visiors with similar life histories.


Swedish Research Council for Environment Agricultural Sciences and Spatial Planning, Award: 2014-01313