This study found adverse effects of imidacloprid on honey bee queen behavior, worker bee activity, brood production, and pollen stores. Not all responses presented in a dose-dependent or monotonic manner, which is similar to other toxicological studies on neonicotinoid effects40,41. In general, effects were less evident at lower doses in larger colonies likely due to the colony’s ability to regulate resources fed to the queen and the greater number of foragers collecting outside (untreated) resources that may dilute imidacloprid levels, thus lessening potential effects. These findings elucidate the complexity of quantifying exposure effects on highly social honey bees and are in line with previous work42 suggesting that honey bees are less susceptible or better at detoxifying neonicotinoids compared to other bee species. Nonetheless our results indicate that small colonies may not be capable of buffering agrochemical exposure and are therefore at increased risk.

Environmentally relevant concentrations of imidacloprid, based on plant residue studies, were fed to colonies. Feeding occurred every other day (pulse exposure) over three weeks and the quantity of syrup fed was proportional to the colony population, but was insufficient to sustain colony development. Bees fully consumed or stored each treatment within 24 hours and all colonies were observed foraging for floral resources. Eighteen of the 216 syrup treatments, mixed over three years, were randomly tested for residues to provide an estimate of imidacloprid concentration levels fed to bees. The average residue levels in treated syrup were close to the intended dosage with the exception of 20 ppb, which was higher than intended (32.9 ± 2.1 ppb), due either to mixing error during treatment dilutions or sensitivity of analytical equipment and residue recovery rate (112.6–119.8%) by testing facilities. The imidacloprid treatments reflect residues found in the nectar and pollen of some treated agricultural crops (10–30 ppb) and ornamental plants (50, 100 ppb). Notably, most studies on seed-treated crops, such as soy, maize, and canola, reported residues at <10 ppb in nectar and pollen43,44,45 thus the results of this study are more relevant to colonies that are exposed to higher levels in these or other crops. Regardless, there are examples in which seed-treated maize, sunflower, and canola have yielded clothianidin residues >10 ppb in pollen and or nectar45,46. In addition to collecting potentially contaminated nectar and pollen sources, bees must collect water for thermoregulation of the colony47. Neonicotinoid contamination in water puddles near seed-treated maize fields during planting has been as high as 63.4 μg/L (thiamethoxam)21. Other studies have found guttation, or plant exudates derived from xylem collected by bees as a water source48, of seed-treated maize can also exhibit high levels of neonicotinoids from >10 mg/L to 346 mg/L19,20. Another consideration in estimating relevant exposure to bees is the uptake of neonicotinoids in non-target wildflowers, such as dandelions (Taraxacum officinale) and clovers (Trifolium repens, Melilotus spp.) that may exhibit residues from <10 ppb (dandelions near seed-treated maize) to 89–319 ppb in clover nectar when turfs were treated with clothianidin spray application32,49. These examples illustrate the immense need for more residue data to better assess environmental exposure and risk of systemic insecticides to bees.

We exposed colonies to imidacloprid and examined potential effects of indirect imidacloprid exposure on queen behavior and colony development. Neonicotinoids bind to nicotinic acetylcholine receptors (nAChRs) in the central nervous system activating constant transmission of nerve signals, an excitatory action that at low doses may cause hyperactivity but with increasing concentration and exposure time can cause severe tremors or paralysis in exposed bees as more nAChRs are bound50. Queens in treated colonies exhibited reduced fecundity likely due to imidacloprid acting directly on sensory and motor functions of the central nervous system that impacted egg-laying behavior and activity. A recent study, also suggests that neonicotinoids can compromise the viability and quantity of stored sperm in mated queens thereby further reducing queen success31. In nature, queen bees are indirectly exposed to environmental toxicants via trophallaxis when fed by worker nurse bees. Trophallaxis in social insects, such as ants, can attenuate toxicity of lethal toxicants particularly those that elicit delayed-action toxicity, such as neonicotinoids, by evenly distributing toxicants among nestmates and rendering them benign likely through dilution by other uncontaminated food or bodily fluids already in the gut51,52. In honey bees, the same mechanistic explanation may apply for the influence of population size on a colony’s ability to buffer pesticide exposure and toxicity. Through trophallaxis, queen bees and brood are fed royal jelly and brood food, proteinaceous glandular secretions derived from fresh and stored pollen. Stored pollen, or beebread, is eaten directly by nurse bees to stimulate production of secretions from the mandibular and hypopharyngeal glands47. The pathway by which contaminated food reaches queens and brood through trophallaxis might originate from the transfer of toxicants through the mandibular and hypopharyngeal glands located in the heads of nurse bees. Although little has been reported about neonicotinoid contamination in glandular secretions, imidacloprid has been detected in products containing glandular secretions such as brood food (>170 ppb acetamiprid and thiacloprid)53 and royal jelly (0.3–1 μg/kg) when bees were fed imidacloprid (100 ug/kg) in supplemental pollen but not syrup35. Imidacloprid and highly toxic metabolites (olefin and 5-hydroxy imidacloprid) have been detected in the heads of worker bees where the glands are located after they were fed 14C-labeled imidacloprid in syrup54.

In this study, indirect imidacloprid exposure through trophallaxis likely resulted in a diluted or “filtered” exposure to queens and brood, but we were unable to quantify the actual exposure levels. Individual queen bees were tested for residues but no imidacloprid or metabolites were detected possibly because individual queens (weighing < 1 g) provided insufficient sample weight to obtain results. Another possibility is that only metabolites were present in queen bees. Chemical analyses of metabolites had limits of detection of 10 and 25 ppb for olefin and 5-OH imidacloprid, respectively. Worker nurse bees attending to queens, or retinue bees, were observed feeding and grooming queens in all colonies throughout the experiment, indicating it is unlikely that reduced egg-laying was the result of poor queen attendance but was rather due to some physiological effect from exposure to imidacloprid and/or metabolites.

Chemical residue analysis of adult worker bees, and stored nectar or honey collected from inside comb cells after the chronic exposure period provided confirmation of imidacloprid exposure and contamination of food stores within the colony. For each colony size, imidacloprid detection in worker bees increased with treatment dose as expected. However, residue levels were lower than the intended dose, particularly in 7000-bee colonies, possibly due to greater numbers of foragers able to collect outside (untreated) resources and social nestmate interactions (trophallaxis) mediating or diluting exposure levels.

The combination of observed responses and chemical analysis indicates that colony size was a significant factor in reducing the toxicity and degree of affliction in treated colonies but only between the smaller (1500- and 3000-bees) and larger (7000-bees) colonies. Queens from 7000-bee colonies exhibited more gradual and graded (dose-dependent) responses, laid twice as many eggs and travelled greater distances per observation compared to the smaller colonies of the same treatment further supporting the hypothesis that imidacloprid concentrations were diluted within the larger colonies (Fig. 1). Few differences were observed in egg-laying rates and inactivity of queens between 1500- and 3000-bee colonies indicating that the smaller two sizes were not very different from each other. In epidemiological terms, social network interactions that comprise organizational immunity55,56 against pathogen transmission may be extended to pesticide exposure and attenuation of toxicity. Organizational immunity has the effect of isolating infected or intoxicated individuals through reduced social interactions and spatial segregation of diseased bees55,56. Though little is understood about the triggers and mechanisms of organizational immunity, the induction of detoxification through increased activity of enzymes in individual honey bees is negatively correlated with population size57. Thus small colonies may rely more on social isolation and metabolic detoxification of older foraging bees to avoid transmission of toxicants to younger hive bees that are more sensitive to pesticides and have lower detoxification capacities58. In contrast, larger colonies have more workers that may bring back uncontaminated forage to directly dilute collected toxicants via trophallaxis with nestmates. It may thus be more advantageous for larger colonies to increase social interactions to attenuate toxicants rather than rely on metabolic detoxification that can be energetically costly, further reinforcing the “buffering” capacity of population size to environmental toxicants, although this hypothesis would require validation59.

The adverse effects on queen behavior extended to colony level effects. There was significantly less brood (eggs, larvae and pupae) and more highly disrupted brood patterns observed in colonies after chronic exposure at all doses and population sizes compared to untreated colonies. Brood production is highly correlated with the population of brood-rearing nurse bees and pollen foragers and thus is a good measure for grading colony health60,61. Brood pattern is also used to assess the health of the developing brood and the queen. “Spotty” or irregular brood patterns often indicate the presence of brood diseases, a failing queen, poor brood care and or limited pollen60,62. Brood care (nursing frequency and duration) of young larvae (<4 days) is strongly correlated with the amount of pollen in the hive62. During times of pollen deficits, older larvae (>4 days) receive preferential feedings while younger larvae are more likely to be cannibalized to compensate for the protein shortage63. In this study, imidacloprid exposure had the strongest effect on the amount of pollen stored in the hive, particularly in the larger (3000- and 7000-bee) colonies likely because the amount of brood and the demand for pollen was greater. Untreated colonies had on average (SE) 4.3% ± 0.34 of all cells containing stored pollen, 6–17 times the amount compared to all treated hives, which had <2% of all cells containing pollen (10 ppb: 0.9% ± 0.51; 20 ppb: 1.5% ± 0.32; 50 ppb: 0.6% ± 0.35; and 100 ppb: 0.5% ± 0.36). Preferential cannibalism of young larvae due to pollen deficits may explain the high variation observed in larvae compared to eggs and pupae among treatments. Another explanation for lower amounts of brood in treated colonies is the potential direct toxicity due to imidacloprid exposure via contaminated brood food, which can alter the physiology and development of larvae64. The overall effects on brood production and pattern in this study were likely caused by a combination of factors, including effects on queen behavior, direct toxicity from contaminated food, reduced brood care and lack of pollen, but it is unclear which factors had the greatest impact on brood development.

A number of studies have demonstrated adverse effects of neonicotinoid exposure on foraging behavior in bees12,30,65,66. In our study, significantly lower foraging activity was observed in 3000- and 7000-bee colonies exposed to neonicotinoids, regardless of dose. Given there were no statistical differences in the initial and final worker bee populations it is likely that low pollen stores in treated colonies (61–161% less than in untreated colonies) was due to exposed bees being too intoxicated to forage efficiently or not stimulated to forage at all. Even with similar population sizes, the treated colonies were set back severely in brood production and pollen stores compared to untreated colonies. Although the colonies in our study were smaller than typical field colonies, colonies containing 4500 and 9000 worker bees can produce more brood per adult bee than colonies containing 17,000 and 35,000 bees62. Therefore population size was not a limiting factor for brood production and rearing capacity. Our findings suggest that treated colonies may appear healthy (based on population size) but may actually be performing poorly in normal colony functions based on brood and pollen stores, which have long-term consequences for colony survival and may be better indicators of colony productivity (pollination services) and health61. In addition, in-hive activity (hygienic behavior) in 7000-bee colonies was disrupted at 50 and 100 ppb treatments. Worker bees with impaired hygienic behavior may have been unable to detect dead brood or were motor-impaired and possibly inactive, similar to foragers and queens in treated colonies. Impaired hygienic performance could affect the colonies ability to prevent within-colony and apiary transmission of pests and pathogens, potentially making colonies exposed to neonicotinoids at high levels more susceptible to robbing by other bees, disease, and parasites66,67,68.

Interpretation of the environmental relevance of our findings and colony fate may require additional studies on full-sized field colonies (≥30,000 workers) and longer observation periods to determine whether queens and colonies can recover from short-term exposures. However, this study highlights the importance of mitigating neonicotinoid exposure when honey bee colonies are at low population sizes such as in early spring when colonies are small due to normal winter losses or when surviving colonies are divided by splitting the population among daughter colonies to prevent swarming. In addition, commercially available “packages” containing small populations from 7000–10000 worker bees are purchased early in spring to replace dead colonies. Small colonies, as shown in our data, which are unable to buffer or dilute neonicotinoid exposure are most vulnerable to queen effects. Risk-mitigation options should focus on reducing exposure risks when colonies are at their lowest population size due to season or management practices, for example in the early spring when risk of exposure to seed-treatment dust is at its highest during planting44. This study provides a mechanistic explanation for how sub-lethal effects of neonicotinoids may impair short-term colony functioning, and offers insights into potential effects of imidacloprid exposure on long-term colony survival10. The results have implications for promoting bee health because they offer a potential explanation for queen failures, which have been identified as a precursor to colony mortality in commercial beekeeping operations9.