Traces of a neonicotinoid pesticide stimulates activity in honey bee colonies, but not productivity

How neonicotinoid contamination effects honey bees remains controversial. Studies have yielded contradictory results, and few have examined effects on colony development. Here we report the results of a comprehensive ve-year study of the effects of the neonicotinoid imidacloprid on honey bee colonies. Colonies fed 5 ng/g (ppb) imidacloprid showed increased brood production, lower temperature variability, higher CO2 production and had more foragers compared to control colonies. Even so, treatment did not affect adult bee numbers or average hive temperatures, and did not increase food stores, daily food acquisition or colony survivorship. Imidacloprid contamination increased colony metabolism without improving colony productivity, which is consistent with the pharmacological action of imidacloprid as a stimulant. Our ndings explain why some studies have reported no, or even positive, effects of neonicotinoids. We provide an explanation for the diverse effects of pesticide contamination on honey bees, and a thorough understanding of how colonies are impacted.


Introduction
The extent to which insecticides harm non-target bene cial insects continues to be highly controversial [1][2][3][4][5] . Some countries have banned neonicotinoid insecticides due to concerns about the negative impacts of neonicotinoid residues on pollinators, particularly honey bees 6 . These bans have in turn caused some workers to raise concerns about the damaging effects of insecticides that have been substituted for the neonicotinoids 3,7 . Controversies remain because it is still unclear how honey bees, and bee colonies, are affected by the low concentrations of neonicotinoids to which they can be exposed when foraging on crops 1,8 .
Imidacloprid is the most commonly used insecticide in the world, and from a toxicological perspective it is one that poses an important risk for honey bees 9,10 . Honey bees are exposed to imidacloprid through ingestion of contaminated nectar, pollen, stored honey and water, and possibly aerial dusts. Harmful but sublethal effects of imidacloprid have been described from concentrations as low as 0.1 ppb 11 . Imidacloprid is frequently found within honey bee hives 12 , and chronic imidacloprid exposure levels have been reported to be as high as 10 ppb, although eld concentrations are highly variable 12 .
Studies of the effects of eld-relevant concentrations of neonicotinoids on honey bee colony performance and survival have reported inconsistent or contradictory results. Colonies exposed to clothianidin had more eggs in Germany, fewer eggs in Hungary, and fewer workers and storage cells in the UK 1 . Honey bee colony mortality was positively correlated with imidacloprid usage in England and Wales 13 . Colony weight gain was reduced in colonies exposed to neonicotinoid treated crops in the UK 14 .
In North America imidacloprid has been shown to negatively affect brood production, queen replacement, foraging activity, winter survivorship, social immunity, and increase Varroa densities, when applied at sublethal concentrations 8, 15 , and to positively affect ight activity and colony thermoregulation 16 . In Australia honey bees exposed to 5 ppb imidacloprid were found to forage at a younger age but to accumulate less food on the long term 17,18 . Additional effects that have been reported include: increased water consumption 11 , decreased foraging performance 19 , higher bee cluster temperatures 20 and reduced learning performance when mixed with another pesticide 21 . Even so, other studies have reported no individual effects at these low levels 19,22,23 . These contrasting results point towards either increased false-discovery rates due to the low sample sizes of these studies 3 , or towards complex and dosedependent interactions between the environment, honey bees, and neonicotinoid insecticides.
As colonial insects, honey bees do not typically experience stressors as individuals 24 . Their social environment can be a powerful buffer of stress 25 . Documenting behavioral impacts on individuals may not translate to consequences that matter for agriculture and apiculture, such as colony behavior and survivorship. For growers the issues of concern are whether neonicotinoids affect crop production by affecting pollinators and pollination and whether neonicotinoids are effective in controlling pest outbreaks. For beekeepers the issues of concern are whether neonicotinoids deplete their colonies of bees or reduce colony productivity.
To better understand how imidacloprid affects honey bee colonies we conducted ve eld experiments, involving a total of 54 bee colonies, measuring the effects of chronically feeding syrup adulterated with 5ppb imidacloprid on bee colony development and behavior. In each experiment, one group of colonies were fed a six-week regimen of sugar syrup adulterated with imidacloprid at 5 ppb, and another group were fed the same regimen with blank sugar syrup. Treatments were intended to simulate nectar collection by the colonies. No commercial agriculture existed within 10 km of the sites. We collected data from hive assessments, including numbers of adult bees and capped brood cells and food reserves; and from sensors, including continuous hive weight, internal temperature and internal CO 2 concentration. We show that at this low level imidacloprid stimulated bee activity, which may explain some of the positive effects of the pesticide that have been reported. Even so, the increased activity of the colonies did not translate to increased colony productivity.

Results
A summary of statistical results is provided (Table 1). Colonies in both groups consumed on average at least 92% of the treatment syrup in all years except 2017, when control colonies consumed 80±9% and imidacloprid colonies consumed 86±8% of the syrup. The number of capped brood cells (a measure of reproduction) was signi cantly affected by both treatment and year ( Figure 1). The imidacloprid group produced more brood. Colonies exposed to imidacloprid had on average 6326±504 capped cells for the rst two post-treatment sampling occasions, while control colonies had on average 5400±504 capped cells. Despite the higher brood production in the imidacloprid group, there was no difference in adult bee mass between groups. The change in the amount of food stores and Varroa mite fall were likewise not affected by treatment but signi cantly differed among experiments (Table 1). This was likely due to weather and forage availability. NEB dry weights were not affected by treatment or year. Raw data on colony survivorship did not show large differences between groups (SI Figure 1). Table 1. Summary of statistical results. P-values for treatment, year and their interactions in analyses of variance. Data transformed as log values where necessary. In the Treatment and Year columns, "+" means colonies given imidacloprid had higher values and "-" lower values. "Wt." means weight; "temp." means temperature; "avg." means average; "ampl." means within-day amplitudes. Column "Covariate" indicates the pre-treatment measurements used as covariates. Signi  Temperature amplitudes were lower in the imidacloprid group, suggesting more controlled thermoregulation in that group. Average daily temperatures were not different ( Figure 2, Table 1). We performed two analyses of daily CO 2 concentrations. In the rst, average daily values were analyzed from Sept. to Nov., with pre-treatment values as covariates. In the second only concentrations at the time of hive evaluations were considered, with associated adult bee masses as covariates. In both analyses overall log CO 2 concentrations were higher in the imidacloprid group than the control group ( Figure 3, Table 1): in the rst analysis 5238±187 ppm compared to 4399±170 ppm; and in the second analysis 4382±423 ppm compared to 3461±344 ppm. Hive CO 2 concentrations were also inversely proportional to adult bee mass (imidacloprid: slope=-0.151±0.027, adjusted r 2 =0.45; control: slope=-0.104±0.026, adjusted r 2 =0.27) (Figure 4).
Within-day measures of hive weight changes were analyzed using piecewise regression (Table 1). We have shown previously that the transient rapid decline in mass at dawn is due to the departure of forager bees from the hive 26 and this parameter was affected by treatment. Imidacloprid treated hives lost more weight during forager departure (118±12 g) than control hives (100±9 g) ( Figure 3, Table 1). These values were calculated from 1 Sept. to 31 Oct. and so include many days when foraging was poor (nectar ows in general ceased by early Oct.). Some days each year were excluded from the analysis because of hive manipulations, rain, or poor curve ts 27 . These results indicate that imidacloprid colonies overall had a larger population of active foragers despite not having larger bee populations.
Imidacloprid was stable in honey stores for months after the treatment (SI Table 1

Discussion
Here we report a comprehensive and long-term study of colony-level effects of a chronic, low level neonicotinoid exposure on honey bee colonies. We believe we now understand why studies of colonylevel effects of low neonicotinoid doses have yielded variable and contradictory results. The best explanation for the effects we found is that imidacloprid increases adult bee activity, but in our experiments this did not translate to better colony outcomes.
While there is still debate about the level at which bee colonies are exposed to imidacloprid in the eld, we focused on exposure to 5 ppb imidacloprid, which is at the lower range of bee exposure estimates 12 .
Colonies provided with this imidacloprid treatment had signi cantly higher brood production in the two months after treatment, producing about 17% more brood on average than control colonies. Even so this did not translate to higher adult bee populations, suggesting an increased mortality of brood, adults or both. Imidacloprid treated colonies also had a larger proportion of active foragers, as measured by the dynamic change in weight at dawn as foragers leave the hive. Even though the foraging force was larger in the treated than in control colonies, it did not translate into increased food gain.
We observed lower temperature variability in the imidacloprid treated colonies. Honey bees actively heat or cool the hive to maintain an optimal temperature for immature bees 28, 29 . Tighter regulation of the brood nest temperature indicates a greater investment by adult bees in thermoregulation. The differences seen were greatest midwinter, typically when bees invest energy to heat the colony.
Daily hive CO 2 concentrations were inversely proportional to the adult bee population sizes, i.e. larger colonies had lower hive CO 2 concentrations than smaller colonies. Clearly larger colonies would produce more CO 2 , but having more workers could also ventilate better. While imidacloprid treatment did not affect adult bee numbers, hive CO 2 concentrations in small colonies in the imidacloprid group were measurably higher than those of small colonies in the control group, yet among larger colonies the two groups were indistinguishable. This suggests that the imidacloprid treatment increased CO 2 production but did not affect CO 2 tolerance.
All of these effects indicate increased activity in the imidacloprid treated colonies. This interpretation is consistent with the molecular action of imidacloprid as a competitive agonist of nicotinic acetylcholine receptors in the insect nervous system [30][31][32] . At very low levels imidacloprid excites neural activity and increases neural oxidative stress 33 , and such exposure has been associated with stimulant effects 34,35 .
The stimulant effects of imidacloprid would explain the changes in colony physiology seen here. The pharmacological action of imidacloprid is not dissimilar to nicotine 37 , which increases general arousal, motor responsiveness and sensitivity in both insects and mammals at low doses. This would explain both the increased investment in thermoregulation and increased CO 2 levels in treated colonies.
These neuroexcitatory effects would also explain behavioral changes observed here. Stimulation of muscarinic acetylcholine receptors by pilocarpine induces a precocious onset of foraging in honey bees 38 . Precocious foraging is a common stress response in bees and triggered by disease and starvation [39][40][41] . The 5 ppb imidacloprid treatment we used here has also been found to induce precocious foraging 18 . In this study we did not measure the age of foraging onset, but we did observe increased foraging activity. From this pharmacological perspective we would expect the effect of imidacloprid on bees to be highly dose dependent. Indeed, at higher levels it overexcites central and peripheral neural circuits and leads to motor dysfunction and death 36 . At the colony level, treatment with a higher dose of imidacloprid at 100 ppb in syrup has been shown to reduce adult bee and brood populations, and increase temperature variability 16 .
The effects documented here are consistent with hormesis, a biphasic physiological response to increasing concentrations of a compound [42][43][44] , which is usually excitatory at low doses and inhibitory at higher doses. Such effects may imbalance the fragile metabolic optimum that bee colonies depend on to maintain homeostasis and growth. We emphasize that the increased foraging or temperature regulation triggered by imidacloprid in this study did not signi cantly improve outcomes for colonies. It is possible that, in some environments, spending more energy foraging and hive thermoregulation could bene t a colony while in other environments it may not. Our ndings offer a simple way to explain why previous studies have found highly variable and dose dependent effects of sublethal neonicotinoid exposure. We offer a single physiological framework to understand the diverse and dose-dependent effects on honey bees of neonicotinoid insecticides at sublethal concentrations.

Materials And Methods
Sucrose syrup preparation. To produce syrup, sucrose was added to distilled water in a 20 L bucket and mixed until sugar was dissolved to produce a solution of 1:1 sucrose weight to water volume. To produce syrup with imidacloprid, 1 kg sugar was dissolved in 900 mL distilled water to produce 1.9 kg "short" solution. A spike solution was prepared by rst dissolving 1.0 mg imidacloprid (Pestanal, CAS # 138261-41-3) in 100 mL distilled water, then mixing 1 mL of that solution into 99 mL distilled water to achieve a 100 mL spike solution. The spike was then added to 1.9 kg short sucrose solution to achieve 2 kg of 5 ppb imidacloprid syrup. Hives were assessed before assignment to treatment groups using a published protocol 26,47 . Brie y, each frame was gently shaken to dislodge bees, photographed (Rebel SL1, Canon USA, Inc., Melville, NY), weighed (EC15, OHaus), and replaced. During this rst assessment, all hive components (i.e. lid, inner cover, boxes, bottom board, etc.) were also shaken free of bees and weighed. The total adult bee population weight was calculated by subtracting the combined weights of hive components free of bees from the total hive weight with bees recorded at midnight prior to the inspection. In addition, 3-5 g of honey were collected from each hive into 50 ml tubes and stored at -80ºC; samples collected in July before treatment were pooled and subjected residue analysis of pesticides and fungicides by the Laboratory Approval and Testing Division, USDA.
After the rst assessment, hives were assigned to groups based on adult bee mass and location (to avoid clumping) and each group randomly assigned a treatment. Just prior to treatment all broodless frames with food stores were replaced with empty drawn comb. Colonies were then fed 2-3 kg treatment syrup twice per week for six weeks, from mid-July onwards. Syrup consumption per colony was recorded. Hives were assessed every 5-6 weeks after the treatment period until Nov., and once more in Feb. At each assessment, hive lid and inner cover weights were compared to previous values to correct wooden parts for moisture content and improve adult bee estimates.
We sampled NEBs before and after the treatment period by pressing an 8 cm x 8 cm wire cage into capped brood, collecting trapped NEBs into 50 mL centrifuge tubes 24 h later, and then storing them at -80°C. To determine dry weight, ve bees per hive per date were placed in Eppendorf tubes and weighed before and after drying for 72 h at 60°C. Varroa mite fall was measured within a few weeks after treatment and colonies were treated in Oct. or Nov. using an amitraz-based product. 3. Multiplying the brood surface area by 0.77 to estimate g brood mass 47 and subtracting that value from the brood + food weight.
4. Dividing the change in the stored food mass between the Oct. and Feb. hive assessments by the number of days between those assessments.
Adult bee, brood cell and food resources were subjected to repeated-measures MANOVA with treatment, year, sampling date, and all 2-way interactions as xed effects and with pre-treatment values as covariates, where possible, to control for pre-existing differences (Proc Glimmix, SAS Inc. 2002). Post-hoc contrasts with Bonferroni correction were reported for signi cant main effects. Analyses of NEB dry weight and Varroa fall were limited to a single sampling occasion immediately after the end of treatment application. Temperature and CO 2 data were transformed into daily average and within-day detrended data, calculated as the difference between the 24 hour running average and the raw data. Sine curves were t to 3-day subsamples of detrended data 47 and MANOVA analyses conducted on daily averages and sine amplitudes. To improve the detection of treatment effects across seasons, continuous data were divided into different time periods for analysis (see Table 1). Hive CO 2 concentrations were also analyzed using a second method: CO 2 concentrations at the time of hive evaluations were estimated by averaging data over three days prior to evaluations (July concentrations at start of study obtained three days post evaluation) and subjected to the MANOVA analysis described above with adult bee mass as covariates. Imidacloprid treatment (red) had no signi cant effects on the adult bee population or food mass accumulated compared to the control (green), but increased brood amounts slightly in the rst months following treatment (Sept. Oct.). Black columns show treatment period. Thick lines show group averages, thin transparent lines individual hive data, dashed lines last measurements before colony death.

Figure 2
Average temperatures in bee colonies exposed to imidacloprid (red) were not signi cantly different from those of control colonies (green), but imidacloprid treated colonies were less variable indicating greater temperature control with variations of 2.8±0.1°C on average, compared to control colonies that varied by 4.0±0.2°C on average. External temperature (yellow) and total daily rainfall (blue) are given below in-hive measurements. Black columns show treatment period. Dots show raw data, smoothed lines default ggplot2/R GAMs.

Figure 3
Upper graph: average CO2 concentrations (measured only in the last two years) were higher in the imidacloprid group, 5238±187 ppm, compared to the control group, 4399±170 ppm., suggesting either an increased metabolic activity or a change in within-hive air management. Lower graph: Imidacloprid treatment (red) was associated with higher morning hive weight loss due to forager departure in the rst months following treatment (Sept. Oct.) suggesting imidacloprid stimulated bee workers to forage more.
Extreme values and days of rain or hive assessments were excluded (see methods), smoothed lines are default ggplot2/R GAMs. Black columns show treatment period. Data are not shown during treatment period because of confounding effects of syrup feeding on hive weight. The low values in 2017 were likely due to poor forage availability that year (see rainfall data in Figure 2).