Drivers of nocturnal and diurnal pollinating insect declines in urban landscapes

Abstract

Insect pollinators are essential for terrestrial ecosystems, delivering key ecosystem functions in the face of anthropogenic disturbance. Urbanization may be a key threat to pollinator communities. However, the scale of the threat remains unknown due to an overwhelming research emphasis on bees and a lack of comparative studies on hyper-diverse pollinating taxa such as nocturnal moths. As a result, it remains unclear which pollinator groups are most vulnerable to urbanization, and which habitat features are most critical for supporting them. We conducted a large-scale assessment of the effects of increasing urbanization on the diversity of bees, hoverflies and nocturnal moths in urban horticultural sites (allotments) across three cities. We report up to a 43% reduction in species richness along urbanization gradients, suggesting that a wide range of pollinators are under threat in urban landscapes. We show that these declines are driven by taxon-specific landscape drivers such as the reduction of tree canopy and semi-natural habitat, suggesting that urban insect conservation depends on the preservation or expansion of habitat features specific to different threatened taxa. We found that relative to bees, moths and hoverflies are particularly sensitive to urbanization, and we highlight the importance of including these frequently overlooked pollinator groups when assessing the biodiversity impacts of environmental change.

Keywords: Hymenoptera; Lepidoptera; Syrphidae; pollinator communities; pollinator declines; urbanization.

Figure 1.

Landscape composition of Leeds, Sheffield and Leicester. (A) The city locations within the UK. (B) The proportion of area of garden, greenspace and impervious cover surrounding eight allotment sites in each city (Leeds, Sheffield and Leicester), at three different scales surrounding the site. Circular buffers with radii of 250, 500 and 1000 m were drawn around the sites. (C) Maps of Leeds, Sheffield and Leicester showing the site locations and the distribution of greenspace (green) and impervious cover (grey) areas. (D) An example of an allotment with a high density of surrounding impervious surfaces and (E) an example of an allotment with a low density of surrounding impervious surfaces. Circles within D and E depict the circular buffers at which landscape composition was measured: 250, 500 and 1000 m.

Figure 2.

Plots showing (A) species richness and (B) abundance of moths (yellow), hoverflies (red) and bees (blue) in allotment sites (points) along gradients of increasing impervious surface (m²) in the area surrounding the site (at a 250 m buffer radius from the site centre). Lines illustrate linear mixed models testing how species richness and abundance change along increasing urbanization gradients across pollinator groups and cities (and their interactions where present). (A) Parallel lines show no significant pollinator taxon × urbanization interactions. (B) A significant pollinator taxon × urbanization interaction. Solid lines show significant effects (p < 0.05); dashed lines show non-significant effects (p > 0.05) derived from post hoc tests.

Figure 3.

Landscape drivers of changes in pollinator communities in urban allotments across multiple scales. The effect of the area of impervious cover, area of tree canopy, area of semi-natural greenspaces and area of gardens on the species richness and abundance of bees, moths and hoverflies across the three cities. Landscape composition was measured at three scales surrounding sites of urban horticulture (250, 500 and 1000 m). Colours of cells indicate positive (blue) or negative (red) responses (or non-significant (grey), p > 0.05 ). All p-values are adjusted using false discovery rates to account for multiple testing. Statistics report a significant main effect when no interaction was significant. Icons of pollinator taxon and city indicate if there were significant interactions between the environmental variable and pollinator taxon or city. Full model outputs and p-value adjustments are found in electronic supplementary material, tables S6–S13.