Honeybee dietary neonicotinoid exposure is associated with pollen collection from agricultural weeds

Abstract

Neonicotinoid insecticides have been linked to bee declines. However, tracking the primary exposure route for bees in the field has proven to be a major logistical challenge, impeding efforts to restore pollinator health in agricultural landscapes. We quantified neonicotinoid concentrations and botanical species composition in 357 pollen samples collected from 114 commercial honeybee colonies placed along a gradient of agricultural intensity between June and September. Neonicotinoid concentrations increased through the season, peaking at the end of August. As a result, concentrations in pollen were negatively associated with collection from woody and crop plants that flower early-mid season, and positively associated with collection from herbaceous plants that flower mid-late season. Higher clothianidin and thiamethoxam residues were correlated with samples containing a greater proportion of pollen collected from agricultural weeds. The percentage of agricultural land within 1500 m was positively correlated with thiamethoxam concentration; however, this spatial relationship was far weaker than the relationship with the proportion of pollen collected from herbaceous plants. These results indicate that both plant species identity and agricultural dominance are important in determining honeybee neonicotinoid exposure through the pollen diet, but that uncultivated plants associated with agriculture are the source of the greatest acute exposure.

Keywords: bee declines; clothianidin; environmental fate; non-crop plants; pollen analysis; thiamethoxam.

Figure 1.

(a) Map of the study area in eastern Michigan showing 2017 field sites in red and 2018 field sites in black; insets: (b) low intensity site, apiary surrounded by 10.0% agricultural land; (c) intermediate intensity site, apiary surrounded by 66.7% agricultural land; (d) high intensity site, apiary surrounded by 92.7% agricultural land. (Online version in colour.)

Figure 2.

The relationship between the concentration (ppb) of thiamethoxam (a–c) and clothianidin (d–f) in honeybee-collected pollen samples and the percentage of these pollen loads comprised of herbaceous (a,d), crop (b,e) and woody (c,f) pollen types. Samples from 2017 are marked with filled circles and samples from 2018 are marked with open circles. Lines represent significant relationships with 95% confidence intervals, indicated by grey shaded areas.

Figure 3.

Non-metric multidimensional scaling (NMDS) plot using Bray Curtis dissimilarity distances for the species composition of honeybee-collected pollen samples (open circles). Arrows represent significant correlations between pollen load composition and neonicotinoids (red arrows) and pollen types (black arrows). Only correlations for neonicotinoids (α = 0.017) and individual pollen types (α = 0.001) that are significant after Bonferroni correction are included. The longer the arrow, the stronger the relationship. (Online version in colour.)

Figure 4.

Relationship between the percentage of agricultural land within 1500 m of sampled apiaries and the concentration (ppb) of (a) thiamethoxam, (b) clothianidin and (c) imidacloprid in honeybee-collected pollen samples. Samples from 2017 are marked with filled circles and samples from 2018 are marked with open circles. Lines represent significant relationships with 95% confidence intervals, indicated by grey shaded areas.