Country-specific effects of neonicotinoid pesticides on honey bees and wild bees

See allHide authors and affiliations

Science  30 Jun 2017:
Vol. 356, Issue 6345, pp. 1393-1395
DOI: 10.1126/science.aaa1190

Damage confirmed

Early studies of the impacts of neonicotinoid insecticides on insect pollinators indicated considerable harm. However, lingering criticism was that the studies did not represent field-realistic levels of the chemicals or prevailing environmental conditions. Two studies, conducted on different crops and on two continents, now substantiate that neonicotinoids diminish bee health (see the Perspective by Kerr). Tsvetkov et al. find that bees near corn crops are exposed to neonicotinoids for 3 to 4 months via nontarget pollen, resulting in decreased survival and immune responses, especially when coexposed to a commonly used agrochemical fungicide. Woodcock et al., in a multicounty experiment on rapeseed in Europe, find that neonicotinoid exposure from several nontarget sources reduces overwintering success and colony reproduction in both honeybees and wild bees. These field results confirm that neonicotinoids negatively affect pollinator health under realistic agricultural conditions.

Science, this issue p. 1395, p. 1393; see also p. 1331


Neonicotinoid seed dressings have caused concern world-wide. We use large field experiments to assess the effects of neonicotinoid-treated crops on three bee species across three countries (Hungary, Germany, and the United Kingdom). Winter-sown oilseed rape was grown commercially with either seed coatings containing neonicotinoids (clothianidin or thiamethoxam) or no seed treatment (control). For honey bees, we found both negative (Hungary and United Kingdom) and positive (Germany) effects during crop flowering. In Hungary, negative effects on honey bees (associated with clothianidin) persisted over winter and resulted in smaller colonies in the following spring (24% declines). In wild bees (Bombus terrestris and Osmia bicornis), reproduction was negatively correlated with neonicotinoid residues. These findings point to neonicotinoids causing a reduced capacity of bee species to establish new populations in the year following exposure.

Global declines in honey bees and wild bees have been linked to pathogens, climate change, habitat fragmentation, and pesticide use (13). The potential threat from neonicotinoid seed coatings applied to flowering crops has been the subject of considerable debate (49). Neonicotinoids have been shown to increase mortality in honey bees by impairing their homing ability (4) and to reduce the reproductive success of bumble bees (5, 8, 10) and solitary bees (8, 11); other studies have identified no effects (8, 12, 13). There is limited information from replicated studies on longer-term survival of honey bee colonies following exposure [see (12)]. Landscape-scale experiments under real-world agricultural conditions are needed to integrate spatial, temporal, and species-specific variation in order to understand the impacts of neonicotinoids on bees (8, 12, 1416). Such studies should explore the impacts of different neonicotinoid formulations, land use, and regional climate. In a large-scale experiment spanning three European countries, we tested the hypotheses that (i) exposure to seed treatments containing neonicotinoids affected the reproductive potential of managed and wild bee species and (ii) whether such effects differ between countries.

At each of 33 sites (Germany, 9; Hungary, 12; and United Kingdom, 12) an average of 63.1 ha (SE of ±2.8 ha) of winter-sown oilseed rape (OSR) was established in 2014 (Fig. 1, fig. S1, and table S1). We clustered sites into triplets (>3.2 km between sites) and randomly allocated sites to one of three treatments: (i) clothianidin applied at 11.86 to 18.05 grams of active ingredient per hectare (g a.i. ha−1) with a fungicide (thriam and prochloraz) and nonsystemic pyrethroid (beta-cyfluthrin) (trade name Modesto); (ii) thiamethoxam applied at 10.07 to 11.14 g a.i. ha−1 and combined with the fungicides fludioxonil and metalaxyl-M (trade name Cruiser); and (iii) control OSR receiving a commercial fungicide (thriam and dimethomorph in Germany and Hungary and thriam and prochloraz in the United Kingdom) but no neonicotinoid seed treatment. All treatments received typical commercial inputs of pesticide (e.g., lambda-cyhalothrin) and fertilizer, with these standardized across a triplet. Standardized colonies of honey bees (Apis mellifera) and wild bees (bumble bee Bombus terrestris and solitary bee Osmia bicornis) were introduced to each site. For honey bees, we quantified the impacts of the treatments on colony viability during the crop flowering period and in the year following exposure (hive survival and overwintering worker, brood, and storage cell numbers). Overwintering fitness defines the multiyear persistence of honey bees. For B. terrestris, we measured impacts on within-year reproductive output (colony weight gain and worker, queen, and drone production) and for O. bicornis the number of reproductive cells produced (table S2). Neonicotinoids can be persistent and widespread in agroecosystems (17, 18), so we quantified residues both in the nests of bee species and those expressed in the crop.

Fig. 1 Location of the 33 experimental sites in the United Kingdom, Hungary, and Germany.

See fig. S2 for a diagrammatic representation of the experimental setup.

We found that neonicotinoid seed treatment affected the interannual viability of honey bee colonies following the winter period in a country-specific manner. In Hungary, worker numbers were 24% lower where clothianidin was compared with the control [treatment × country: χ2(6) = 1.47, P = 0.01, explained variance = 59.4%] (Fig. 2), with no significant effect of thiamethoxam. Clothianidin was more likely to be expressed in the crop where it was applied as a seed treatment, which identified a mechanism of exposure to the bees [χ2(2) = 6.46, P = 0.04], but this was not so for thiamethoxam (table S3). In the United Kingdom, high hive mortality precluded a formal statistical analysis of overwintering worker numbers. However, median worker numbers were zero for all four clothianidin-treated sites but above zero for two of the control and one of the thiamethoxam sites (table S2 and Fig. 2). Worker numbers following the winter in Germany showed no treatment effect (table S4). Overwintering honey-bee brood, stored hive products (pollen and nectar), and the likelihood of hives surviving the winter were not affected by seed treatments (table S3).

Fig. 2 Summary effect sizes for the response of honey bees and wild bees to the neonicotinoid seed treatments.

An effect size represents the difference between the mean population response for a given seed treatment and the control within a country, with this difference divided by the pooled standard deviation, where asterisk (*) indicates a significant difference between the control and seed treatment [either TMX (thiamethoxam) or CTD (clothianidin)] determined from the predicted marginal means of the model “y ~ seed treatment × country + block/country.” Dagger (†) indicates where U.K. colony survival was too low for a formal analysis. Note that effect sizes differ between countries.

Neither B. terrestris queen nor O. bicornis egg cell production was directly affected by the seed treatments or its interaction with country (table S5). However, they were negatively correlated with peak [χ2(1) = 2.09, P = 0.03, explained variance = 13.5%] (Fig. 3A) and median [χ2(1) = 4.34, P = 0.04, explained variance = 0.8%] (Fig. 3B) neonicotinoid nest residues (combined clothianidin, thiamethoxam, and imidacloprid). Imidacloprid was not applied as part of the study, and its presence is most likely a result of environmental contamination from previous widespread agronomic use (17, 18). Residues of neonicotinoids detected in stored hive products did not differ in response to seed treatments for any bee species (table S6). This may be due to the amalgamation of stored hive products at the site level for residue analysis, which may have obscured within-site heterogeneity in residues. The negative correlation for B. terrestris queen production remained significant when we excluded sites with imidacloprid residues [χ2(1) = 2.14, P = 0.02], although this was not the case for O. bicornis2(1) = 0.05, P = 0.81]. Country-specific responses to neonicotinoid seed treatment were found for B. terrestris drone production, with positive and negative effects from exposure to thiamethoxam in Germany and the United Kingdom, respectively [treatment × country: χ2(6) = 13.1, P = 0.04, explained variance = 13.6%] (Fig. 2).

Fig. 3 Wild bee reproductive success in response to neonicotinoid nest residues.

Separate graphs are shown for the response of B. terrestris queen production and O. bicornis reproductive cell production to neonicotinoid residues found in nests. The significance of these relationships is based on a likelihood ratio test comparison of H0: “y ~ country” and H1: “y ~ neonicotinoid + country.” Neonicotinoid residues are based on summed concentrations of clothianidin, thiamethoxam, and imidacloprid.

We also found seed treatment effects during the crop flowering period that lasted between 3 and 6 weeks (tables S4 and S5). Significant interactions between seed treatment and country were identified for peak worker [χ2(6) = 16.6, P < 0.01, explained variance = 45.3%], egg cell [χ2(6) = 4.13, P = 0.01, explained variance = 49.9%], and combined pollen and nectar storage cell [χ2(6) = 40.5, P < 0.001, explained variance = 53.6%] numbers. These responses describe within-year colony performance. Neonicotinoid exposure resulted in both negative (Hungary and United Kingdom) and positive (Germany) effects on colony size (see Fig. 2; pairwise treatment comparison given in tables S4 and S5). Bombus terrestris worker and peak colony weight showed no seed treatment response.

Our quantification of neonicotinoid effects on the interannual viability of honey bees and wild bee populations represents a fundamental advance in our understanding of the impacts of these pesticides. For solitary bees and bumble bees (queen production), neonicotinoid impacts were associated with the residues found in nests rather than the experimental seed treatments. For B. terrestris, the few treatment effects and the presence of imidacloprid in stored pollen and nectar (tables S7 to S9) suggests that negative impacts of neonicotinoids may be driven by the persistence of residues in the wider landscape rather than current management alone (18, 19). The European Union (EU) moratorium meant that no neonicotinoids were applied to oilseed in the surrounding landscapes during the experiment, so such residues may originate from previous agricultural use leading to expression in nontarget plants (1719), guttation fluids, or contaminated water (19, 20). Although the reproductive potential of O. bicornis was also negatively affected by neonicotinoid residues in nests, the explained variation of these effects was small. However, a failure to detect small population changes may be due to limited experimental replication restricting statistical power. Our results suggest that even if their use were to be restricted, as in the recent EU moratorium, continued exposure to neonicotinoid residues resulting from their previous widespread use has the potential to impact negatively wild bee persistence in agricultural landscapes (14, 18, 19).

Taken together, our results suggest that exposure to neonicotinoid seed treatments can have negative effects on the interannual reproductive potential of both wild and managed bees, but that these effects are not consistent across countries. The country-specific responses of honey bees and bumble bees strongly suggest that the effects of neonicotinoids are a product of interacting factors (2023). This study has identified between-country differences in the use of oilseed rape crop as a forage resource for bees (affecting exposure to crop residues) and incidence of disease within hives. Both factors were higher for Hungarian and U.K. honey bees (tables S10 and S11). Overall neonicotinoid residues were detected infrequently and rarely exceeded 1.5 ng g−1 (w/w). As such, direct mortality effects caused by exposure to high concentrations of neonicotinoids are likely to be rare (table S12). However, our results suggest that exposure to low levels of neonicotinoids may cause reductions in hive fitness that are influenced by a number of interacting environmental factors. Such interacting environmental factors can amplify the impact of honey bee worker losses (e.g., through sublethal toxicity effects) and reduce longer-term colony viability (4, 16). Note that our common experimental approach applied across three countries revealed varying impacts and may explain the inconsistent results of previous studies conducted in single countries or at few sites (4, 5, 8, 12, 13, 15).


Materials and Methods

Figs. S1 and S2

Tables S1 to 12

References (2432)


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 29.
  29. 30.
  30. 31.
  31. 32.
  32. ACKNOWLEDGMENTS: Data are in supplementary materials. Funded by Syngenta Ltd. and Bayer CropScience (P. Campbell, M. Miles, C. Maus, D. Holah, M. Coulson). Wild pollinator work supported by NERC CEH National Capability funding (NEC05829). Thanks to K. Jaekel, P. Fisher, M. Nowakowski, R. Hails, P. Scrimshaw, N. Mitschunas, P. Nuttall, M. McCracken, S. Ball, J. Webb, B. Sutherland, R. Freckleton, T. Tscharntke, J. Memmott, K. Norris, B. Raffa, and D. Vaskor.

Stay Connected to Science

Navigate This Article