Following Regulation, Imidacloprid Persists and Flupyradifurone Increases in Nontarget Wildlife

After regulation of pesticides, determination of their persistence in the environment is an important indicator of effectiveness of these measures. We quantified concentrations of two types of systemic insecticides, neonicotinoids (imidacloprid, acetamiprid, clothianidin, thiacloprid, and thiamethoxam) and butenolides (flupyradifurone), in off‐crop nontarget media of hummingbird cloacal fluid, honey bee (Apis mellifera) nectar and honey, and wildflowers before and after regulation of imidacloprid on highbush blueberries in Canada in April 2021. We found that mean total pesticide load increased in hummingbird cloacal fluid, nectar, and flower samples following imidacloprid regulation. On average, we did not find evidence of a decrease in imidacloprid concentrations after regulation. However, there were some decreases, some increases, and other cases with no changes in imidacloprid levels depending on the specific media, time point of sampling, and site type. At the same time, we found an overall increase in flupyradifurone, acetamiprid, thiamethoxam, and thiacloprid but no change in clothianidin concentrations. In particular, flupyradifurone concentrations observed in biota sampled near agricultural areas increased twofold in honey bee nectar, sevenfold in hummingbird cloacal fluid, and eightfold in flowers after the 2021 imidacloprid regulation. The highest residue detected was flupyradifurone at 665 ng/mL (parts per billion [ppb]) in honey bee nectar. Mean total pesticide loads were highest in honey samples (84 ± 10 ppb), followed by nectar (56 ± 7 ppb), then hummingbird cloacal fluid (1.8 ± 0.5 ppb), and least, flowers (0.51 ± 0.06 ppb). Our results highlight that limited regulation of imidacloprid does not immediately reduce residue concentrations, while other systemic insecticides, possibly replacement compounds, concurrently increase in wildlife. Environ Toxicol Chem 2024;43:1497–1508. © 2024 The Authors. Environmental Toxicology and Chemistry published by Wiley Periodicals LLC on behalf of SETAC.


INTRODUCTION
Agricultural intensification has had significant impacts on ecosystems globally (McLaughlin & Mineau, 1995;Tilman et al., 2002Tilman et al., , 2011;;Venter et al., 2016), and the use of agrochemicals in intensive agricultural systems poses a variety of threats to biodiversity (Potts et al., 2010).For instance, agrochemical use has contributed to a global decline in invertebrate abundance, with knock-on effects including decreased agricultural yields in crops that rely on ecosystem services (Bartomeus et al., 2014;Potts et al., 2010;Seufert et al., 2012) and decreased abundance of aerial insectivores by limiting prey availability (Spiller & Dettmers, 2019;Wagner et al., 2021).In response, regulatory bodies establish guidelines for use of agrochemicals within food systems and the environment (European Food Safety Authority, 2023; Government of Canada, 2023;US Environmental Protection Agency, 2021).Despite the role that those regulations play in safeguarding food systems and ecosystems, knowledge of how pesticide loads in biological media change in response to regulations is often lacking.
Regulation of agrochemicals presents a number of challenges for diverse stakeholders.For instance, regulators must establish safe concentration thresholds for chemicals that most often occur as pesticide mixtures in media where synergistic effects with other stressors may also occur (Bishop et al., 2022;Botías et al., 2015;Straub et al., 2019;Zhao et al., 2020).Furthermore, toxicological assays are not always sufficiently representative of the context or breadth of wildlife exposure scenarios or species-specific toxicity thresholds (Morrissey et al., 2015).That can result in revisions to the recommended uses of the pesticide as new data become available (European Food Safety Authority, 2016;Rundlöf et al., 2015;Hernandez Jerez et al., 2022).Pesticides can persist, accumulate, and spread throughout the environment and cause adverse effects in off-crop systems and nontarget organisms (Bonmatin et al., 2015;Tang et al., 2021), thus creating additional challenges for regulators aiming to mitigate the effects of these compounds.
During and after applications, agrochemicals can enter the environment by spray drift (Krupke et al., 2012), through groundwater, or by accumulation and translocation in nontarget organisms (Bonmatin et al., 2015).Concentrations in offcrop and nontarget plants, including wildflowers, may vary both spatially and seasonally in relation to agricultural activities (Bishop et al., 2022;Botías et al., 2015).Therefore, pesticide exposure among pollinators may be underestimated because of highly contaminated wildflowers (Botías et al., 2015;Rundlöf et al., 2015).Impacts of pesticide exposure not only have affected global bee populations (Potts et al., 2010;Rundlöf et al., 2015) but can also affect vertebrate pollinators including hummingbirds (English, Sandoval-Herrera, et al., 2021) and bats (Williams-Guillén et al. 2016).
In April 2021, regulations were introduced in Canada to control the use of a common neonicotinoid, imidacloprid, on highbush blueberry.Highbush blueberry farms in Canada are susceptible to loss from pests including aphids (Ericaphis fimbriata), which damage crops by spreading disease (Government of British Columbia Ministry of Agriculture, 2023).Application of 240 g/L imidacloprid at 175 mL/ha on highbush blueberry is a recommended chemical treatment for aphid population control in British Columbia.The label-change in 2021 restricted timing of imidacloprid application for blueberry aphids and other pests to occur only after the bloom drop, rather than two spray periods which were previously pre-bloom (typically April) and post-bloom (typically mid-to late May).
In the Fraser River valley in British Columbia, Canada, where blueberry fields are common, we address three main questions regarding the effects of the 2021 imidacloprid regulation on pesticide loads in hummingbird cloacal fluid, wildflowers, and honey bee (Apis mellifera) nectar and honey.First, we tested whether overall pesticide loads (calculated as the sum concentration of imidacloprid, flupyradifurone, acetamiprid, clothianidin, thiacloprid, and thiamethoxam) changed in nectar, flower, or hummingbird cloacal fluid after imidacloprid regulations were in effect.We then asked whether concentrations of imidacloprid changed in those nontarget media at varying spatiotemporal distances to agriculture.Finally, we asked whether flupyradifurone concentrations, a leading replacement for imidacloprid, changed in response to regulation of imidacloprid.All but thiacloprid and clothianidin are registered for application to highbush blueberry in British Columbia (Government of British Columbia Ministry of Agriculture, 2023); however, clothianidin is a common breakdown product of thiamethoxam (Fan & Shi, 2017).To address these questions, we collected hummingbird cloacal fluid, honey bee nectar, and wildflowers annually in April through July in 2019 to 2022 at three levels of spatiotemporal proximity to agriculture before and after regulation of imidacloprid spraying on blueberry crops.We also collected honey samples in 2021 and 2022.We predicted that total agrochemical load would not change following the regulation of imidacloprid in April 2021; instead, we predicted that imidacloprid concentrations would decrease, while the recently introduced butenolide insecticide flupyradifurone would increase in concentration to compensate for the difference.

Sample collection
We collected hummingbird cloacal fluid (n = 115), honey bee nectar (n = 240), honey (n = 78), and flower (n = 390) samples from 17 sites located at distances between 10 m and 22.8 km of agricultural areas (Supporting Information, Figure S1 and Table S1).Our study area was located in the primary berrygrowing region in Canada, with over 10,000 ha of land in British Columbia producing nearly all of the highbush blueberries cultivated in Canada, in addition to approximately 1000 ha producing >80% of Canada's raspberries (Raspberry Industry Development Council, 2023;Statistics Canada, 2022).We collected 823 samples: 296 from our reference sites, 184 from sites within 500 m of a conventionally sprayed highbush blueberry field before the bloom drop, and 343 samples from sites within 500 m of a conventionally sprayed highbush blueberry field after the bloom drop.In addition, three of our reference sites were located at a distance >1 km from agriculture but were within or adjacent to residential properties.Hummingbird cloacal fluid, honey bee nectar, and flower samples were collected for 4 consecutive years in April to July (2019-2022), representing 2 sampling years prior and 2 sampling years following governmental regulations on imidacloprid use in April 2021.Honey sampling began in 2021; therefore, we were not able to assess changes in contamination of honey resulting from the regulations.Highbush blueberry was the only crop impacted by the regulations implemented in 2021, though the compounds quantified in our study are also registered for other crops grown in this valley and for pest management in residential areas.
Sampling protocols for hummingbirds, bee products, and wildflowers adhered to those described previously (Bishop et al., 2022).Anna's (Calypte anna) and rufous (Selasphorus rufus) hummingbird cloacal fluid samples were collected from 13 sites, of which seven were putative reference sites.Individual samples were pooled by site and time point to attain a minimum sample volume of 300 μl for chemical analysis.Honey bee nectar was collected from within bee hives at seven sites, three of which were reference sites.Capped honey was collected from within bee hives at five sites, three of which were reference sites.
Flower heads were collected at 13 sites, seven of which were reference sites (Supporting Information, Table S1).The flower species collected depended on which species were in bloom at the time of sampling.Sampling took place at weekly to monthly intervals between April and August, depending on the year (Supporting Information, Table S1).Sampled flower species included primarily dandelions of the Taraxacum genus and Himalayan blackberry (Rubus armeniacus), though creeping buttercup (Ranunculus repens) and foxglove (Digitalis purpurea) were also commonly sampled (Supporting Information, Table S1).Flower heads were collected by hand from the top of the stems.Each sample for chemical analysis comprised samples collected from a single species and was analyzed as a single homogenate.The taxonomy of collected samples was confirmed using the iNaturalist software (2018) and field guides (MacKinnon & Pojar, 1994).Samples were stored on ice (-5 °C) within 20 min of collection and then transferred to −20 °C within 72 h of collection.
Hummingbird cloacal fluid was collected following noninvasive sampling methods (Bishop et al., 2018).The handling of hummingbirds was conducted under the following Canadian Wildlife Service permit numbers: 10899, 10883L, 10720E, 10761V, 10761H, and 10805.Sampling of biological specimens was approved under AM-MIN2019 and Animal Care Certificate 17CB03.All data, including sample proximity to agriculture, sample species identification, and analytical chemistry results, are provided as online supplements (Supporting Information, Table S1).

Chemical analysis
Chemical analyses were conducted at the National Wildlife Research Centre, Ottawa, Ontario, Canada.Hummingbird cloacal fluid, nectar, honey, and flowers were analyzed following methods described previously (Bishop et al., 2018(Bishop et al., , 2020(Bishop et al., , 2022)).Detection limits varied by compound, year, and media.Consequently, there were 559 unique detection limits for 4938 chemical quantitations.The lowest detection limits were for flupyradifurone and thiacloprid at 0.014 ng ml −1 for honey and nectar; the highest detection limit was for clothianidin at 1.63 ng g −1 for flower tissue.Samples were blind-coded to avoid revealing the sampling information to the analytical laboratory.
Cloacal fluid samples were prepared as aliquots of 200 μL of cloacal fluid pools spiked with 50 μL of acetonitrile containing internal standard (IS).For flowers, approximately 1 g of chopped flower sample was spiked with 25 µL of IS solution (1000 parts per billion [ppb] in 80:20 water:acetonitrile).For both honey and nectar samples, a 4-mL aliquot of deionized water was added to approximately 2 g of thawed nectar or honey, and the sample was spiked with 12.5 µL of IS solution (1000 ppb in 80:20 water:acetonitrile).The IS consisted of six stable isotope-labeled standards: acetamiprid-d3, clothianidin-d3, imidacloprid-d4, and thiamethoxam-d3 from Sigma Aldrich; flupyradifurone-d5 from Toronto Research Chemicals; and thiacloprid-d4 from CDN Isotopes.
Solvent blanks (water:acetonitrile 80:20) were injected at the beginning and the end of each set of samples to monitor injection cross-contamination.One random sample per set was run in duplicate to evaluate method precision; all duplicate results above the minimum reporting level (3 times the minimum detectable limit) were within 15% relative percent difference of one another.Method accuracy was quantified by spiking a clean cloacal fluid pool with a mixed neonicotinoid standard and calculating the recoveries; all values were between 85% and 115%.To monitor quantification accuracy, we analyzed a second source standard (commercially prepared solution; ChemService) against a calibration on a daily basis and calculated that concentrations were within 85% to 115% of the expected concentrations.

Statistical analyses
All statistical analyses were conducted in R (Ver.4.2.3;2023).Nonparametric descriptive statistics of left-censored chemistry results were computed using the Kaplan-Meier method (Barker, 2012;Kaplan & Meier, 1958) as implemented in the package EnvStats (Ver.2.7.0;Millard, 2013).To calculate a mean pesticide load in the four media, we calculated mean concentrations of each compound detected in each medium and summed concentrations across compounds (Helsel, 2011).When all quantitations for a given compound and medium were below the method detection limit, we assumed a concentration of 0 ppb.Statistical modeling of contaminant concentrations in media was performed using Stan (Carpenter et al., 2017;Stan Development Team, 2022) interfaced via the package brms (Ver.2.19.0;Burkner, 2017).Our Bayesian models assumed a log-normal response distribution on censored data with an identity link function.
We built three separate models to address each of our questions regarding the effect of regulations of imidacloprid on pesticide contamination in nontarget media.In our model of overall pesticide loads, we evaluated concentrations of each pesticide in nectar, flower, or hummingbird cloacal fluid before and after imidacloprid regulations were in effect.To accommodate nonindependence of analyte quantitations from the same sample, we included a normally distributed group effect for sample with a mean of 0. Next, we modeled changes to imidacloprid concentrations in response to regulation among nontarget media and at different levels of spatiotemporal proximity to agriculture (i.e., before or after the bloom drop or located in reference areas).Finally, to address our hypothesis that flupyradifurone concentrations would increase following imidacloprid regulation, we modeled changes to flupyradifurone concentrations using the same model structure as for imidacloprid, where concentration was estimated as a function of media type, spatiotemporal proximity to agriculture, and regulation.We included a normally distributed group effect with a mean of 0 for the sampling location in each of our three models.
For each model, we ran four independent chains and assessed model convergence when all monitored parameters yielded r-hat statistics <1.01.We evaluated contrasts and performed Bayesian nonlinear hypothesis testing using the hypothesis function in the brms package to calculate joint posterior distributions of model parameters.We considered evidence of one-sided contrasts to have strong, moderate, or weak support when the posterior probability of a given hypothesis (p-tail) was 0.05, 0.10, or 0.20, respectively.We therefore also considered one-sided contrasts to have strong, moderate, or weak evidence to the contrary of a given hypothesis when the posterior probability (p-tail) was 0.95, 0.90, or 0.80, respectively.Parameter point estimates are presented on the log scale with 95% confidence intervals (CIs) unless otherwise indicated.Descriptive statistics of left-censored chemistry results are presented as means ± standard error (percentage nondetects [ND]).

Pesticide loads in media before and after regulation
We detected flupyradifurone in 54% of all samples, imidacloprid in 38% of samples, clothianidin in 8% of samples, acetamiprid in 24% of samples, thiamethoxam in 15% of samples, and thiacloprid in <1% of samples.All thiacloprid detections occurred in flowers.

DISCUSSION
Supporting our predictions, we did not find evidence of an overall decline in pesticide concentrations in nontarget media after imidacloprid regulations came into effect in 2021 in Canada.Further supporting our predictions, we observed substantial increases in flupyradifurone concentrations in flower, nectar, and hummingbird cloacal fluid samples; however, we did not find evidence for an overall decrease in imidacloprid concentrations.Our study sites were embedded primarily in agricultural areas cultivating highbush blueberry (Bishop et al., 2018(Bishop et al., , 2020(Bishop et al., , 2022)); however, uses of imidacloprid in the urban context or on other common crops in the Fraser Valley such as raspberry and potato were not affected by the 2021 regulation.Therefore, while we expected that any changes in off-target contamination would be the result of regulatory changes on blueberry farms, neonicotinoid and butenolide residues could also originate from other crops or urban garden uses (Bishop et al., 2018).Our results highlight that regulating pesticides for all land uses concurrently and enacting more stringent regulations on novel pesticides may be required to protect wildlife from accumulating high concentrations of these chemicals.
Because neonicotinoids came under scrutiny for their role in global pollinator declines (Goulson, 2019;Potts et al., 2010;Powney et al., 2019;Wagner et al., 2021), several agrochemicals have appeared in global pesticide markets as replacements (Martin et al., 2023;Siviter et al., 2018), including flupyradifurone.However, the effects of such replacement agrochemicals on food systems and wildlife are not yet fully understood (Siviter & Muth, 2020), while the potential for toxic effects in off-crop nontarget organisms is considerable (Graham et al., 2022;Wood & Goulson, 2017).While regulation of neonicotinoid uses on outdoor crops should reduce the ecological impact of pesticides on wildlife, permissive usage of novel replacements like flupyradifurone is expected to continue to threaten food systems and the environment (Siviter & Muth, 2020).
Flupyradifurone possesses a similar mode of action to the neonicotinoids (Jeschke et al., 2015); and although it is claimed to be safe for nontarget plants and animals based on median lethal concentration (LC50), median lethal dose, or median effective dose assays in seven taxa (Nauen et al., 2014), effects are now documented on bees (Hesselbach & Scheiner, 2019;Kablau et al., 2023;Tosi & Nieh, 2019;Tosi et al., 2021) and some nontarget aquatic organisms (Bartlett et al., 2018;Bartlett et al., 2019;Zhong et al., 2021).In addition, susceptibility to toxic effects is species-specific (Morrissey et al., 2015).For example, LC50 values for flupyradifurone reported for mayfly larvae (Hexagenia) were one to two orders of magnitude lower than those reported for aquatic invertebrates by the manufacturer (based on LC50 assays in Daphnia; Bartlett et al., 2018;Nauen et al., 2014).Furthermore, sublethal effects (median effect concentration) occurred at approximately three orders of magnitude lower than the manufacturer's LC50 for aquatic invertebrates (Bartlett et al., 2018;Nauen et al., 2014), highlighting the need for more representative species and endpoints in toxicological studies.
Concurrent increases in flupyradifurone use and the restriction of neonicotinoids have been noted in the United States (US Geological Survey, 2020a, 2020b).We report that mean flupyradifurone concentrations across all samples increased twofold in nectar, sevenfold in hummingbird cloacal fluid, and eightfold in flowers after the 2021 regulation of imidacloprid.In our study area, the recommended application rate of flupyradifurone for aphid pests is >3.5 times higher than that of imidacloprid, with no restrictions on timing of foliar spray, which can be applied as often as once every 7 days (Government of British Columbia Ministry of Agriculture, 2023).Both flupyradifurone and imidacloprid are classified as intermediate toxicity to nontarget aquatic species, exhibiting acute LC50 values between approximately two-and 10-fold of one another (Bartlett et al., 2018(Bartlett et al., , 2019)).Thus, some wildlife species may be at risk of toxic effects, given that relative toxicities of imidacloprid and flupyradifurone may be largely or entirely offset by differences in application rates and species-specific sensitivities.The effects of flupyradifurone on nontarget wildlife in our study region at the concentrations detected in our study sites are not known.However, our study further confirms that wildlife are chronically exposed to mixtures of agrochemicals near to (<500 m) and far from (>500 m) actively sprayed fields, raising the potential for impacts not predicted by conventional pesticide risk assessments (Rondeau & Raine, 2024;Tosi & Nieh, 2019).
Following the European Union's 2013 moratorium on imidacloprid use on mass-flowering, bee-attractive crops, imidacloprid concentrations persisted at elevated levels in untreated crops for 5 years without evidence of attenuation (Wintermantel et al., 2020) and continue to pose a threat to wildlife (Woodcock et al., 2018).Nontarget wildlife may be exposed to neonicotinoids in part due to their water solubility, which allows the pesticides to spread through the environment carried by precipitation or irrigation systems (Botías et al., 2016).Neonicotinoids like imidacloprid can then accumulate in nontarget media, persisting in water for >100 days (Kumar et al., 2023;Wintermantel et al., 2020), and expose organisms that use the contaminated media.These transport mechanisms near treated crops drive the intraannual variation in neonicotinoid concentrations in off-crop media (Woodcock et al., 2018) and are likely important factors behind the variation in concentrations observed in nontarget media before and after the bloom drop and in reference sites in our study area.We are not aware of any permitted exemptions allowing for use of imidacloprid in this region, and it is, therefore, assumed that measured concentrations are the result of applications as described in the production guide (Government of British Columbia Ministry of Agriculture, 2023), transport, or accumulation.Notwithstanding the enactment of new regulations on imidacloprid use in blueberry crops, elevated neonicotinoid concentrations in reference sites in our study suggest that contamination in nontarget media may also be driven by applications on other crops besides blueberry, plants from commercial nurseries, or other urban uses (Bishop et al., 2018).
Imidacloprid concentrations >45 ppb were detected in oilseed rape nectar more than 3 years after the European Union moratorium, raising concerns that neonicotinoids were accumulating in nontarget media (Wintermantel et al., 2020), with the potential to cause chronic or acute toxic effects to wildlife (Botías et al., 2015;Woodcock et al., 2018).We detected imidacloprid at comparable concentrations in nectar from areas within 500 m of agriculture after regulation in 2021.Therefore, accumulation of neonicotinoids in off-crop nontarget media does pose a continued threat of chronic exposure to wildlife (Botías et al., 2015).That has led some authors to advocate for the total ban on outdoor applications of neonicotinoids (Goulson, 2018;Wintermantel et al., 2020), as was implemented by the European Union in 2018 (European Food Safety Authority, 2018).A total ban on systemic insecticides may be necessary to minimize threats to wildlife because those agrochemicals persist and permeate the environment widely due to their water solubility (Botías et al., 2016).
Current pesticide risk assessments of systemic insecticides focus on insect pollinators because most of the available toxicology research is dedicated to understanding impacts to species of bees.We show that exposure can extend much further and include avian pollinators.Flupyradifurone persistence at concentrations similar to those found in nectar samples in our study area resulted in chronic toxic exposure in honey bees (Tosi et al., 2021).Vertebrate toxicological studies of flupyradifurone at concentrations measured in our study are lacking.Despite the paucity of available studies, hummingbirds are at particular risk of high exposure to systemic insecticides (Etterson et al., 2023), and exposure and/or loss of insects as a food resource may be one factor contributing to population declines of migratory species in this family (English, Bishop, et al., 2021).While some studies on effects of flupyradifurone suggest lower toxicity compared with neonicotinoid predecessors (Bartlett et al., 2018(Bartlett et al., , 2019;;Nauen et al., 2014), one major concern in our study area is the unregulated use of this novel butenolide.Relatively lower toxicity to flupyradifurone compared to imidacloprid for some species may not effectively protect wildlife because the compound is applied and accumulates at substantially higher concentrations.
The widespread contamination of natural and food systems by agrochemicals can pose substantial threats to wildlife.The reliance on high-input agricultural systems undermines the resilience of the very resource that intensification strives to optimize (Malaj & Morrissey, 2022).Consistent with findings that offcrop imidacloprid application drives exposure in managed bee colonies (Graham et al., 2022), we found that management of imidacloprid use in blueberry fields did not result in a decline in imidacloprid contamination measured in wildlife media.Neonicotinoid persistence and unmanaged butenolide applications have led to elevated contaminant concentrations in nontarget media and exposure of insect and avian pollinators in Canada, while the effects of that exposure have not been adequately addressed (Bishop et al., 2020(Bishop et al., , 2022;;English, Sandoval-Herrera, et al., 2021).Hummingbirds face multiple potential routes of exposure via nectar and insect consumption and through these mechanisms are considered to be at elevated risk of high exposure to systemic insecticides (Etterson et al., 2023).Moreover, global declines of invertebrate abundance, caused in large part by agricultural intensification including pesticide use, may impact hummingbirds and other aerial insectivores indirectly by limiting prey availability (Goebel et al., 2024;Spiller & Dettmers, 2019;Wagner et al., 2021).Indeed, the persistence, accumulation, and effects of neonicotinoids and butenolides on wildlife have spurred calls to review environmental risk-assessment strategies broadly (Bean et al., 2024;Morrissey et al., 2024;Rattner et al., 2024;Sgolastra et al., 2020;Tosi et al., 2021).Our findings are consistent with the need for more field-based environmental and toxicological data to adequately assess the risk of pesticide exposure and effects on wildlife (Morrissey et al., 2024;Siviter & Muth, 2020;Wagner et al., 2021).
Supporting Information-The Supporting Information is available on the Wiley Online Library at https://doi.org/10.1002/etc.5892.

FIGURE 1 :
FIGURE 1: Averaged across media and exposure groups (hummingbird cloacal fluid, honey bee nectar, and wildflowers) among all study sites (n = 17), estimated change and 95% credible intervals for each pesticide between 2019-2020 and 2021-2022.Mean concentrations of imidacloprid, flupyradifurone, acetamiprid, clothianidin, thiamethoxam, and thiacloprid either increased or did not change after Canadian federal regulation of imidacloprid in 2021.Credible intervals overlapping with 0 (red dashed line) indicate little or no evidence of an average change in concentration, while intervals above or below 0 indicate an increase and decrease, respectively.PPB = parts per billion.

FIGURE 2 :
FIGURE 2: Mean summed concentrations of imidacloprid, flupyradifurone, acetamiprid, clothianidin, thiamethoxam, and thiacloprid in media before (dark shades) and after (light shades) Canadian federal regulation of imidacloprid in 2021.For each compound, dark shades represent mean concentration for the period 2019-2020, and light shades represent mean concentration for the period 2021-2022.Exposure groups include sites within 500 m of agricultural areas sampled either before or after the bloom drop as well as reference sites more than 500 m to the nearest agricultural area.Concentrations of all six compounds were quantified in each sample, and the percentages of quantitations below the detection limit are shown as percent nondetects.Summed concentrations are shown in nanograms per liter for nectar, honey, and hummingbird cloacal fluid and in nanograms per gram for flower homogenates.When all quantitations for a given compound in a given exposure group were below detection limits, concentrations were assumed to be 0 for summation with other compounds.Otherwise, descriptive statistics of left-censored sample-bygroup means were computed using the Kaplan-Meier method and summed.PPB = parts per billion; ND = nondetects; NS = not sampled.

FIGURE 3 :
FIGURE 3: Imidacloprid concentrations in media before (dark shades) and after (light shades) Canadian federal regulation of imidacloprid in April 2021.Exposure groups include sites within 500 m of agricultural areas sampled either before or after the bloom drop as well as reference sites more than 500 m to the nearest agricultural area.Percentages of samples below the detection limit are shown as percent nondetects.Imidacloprid concentrations are shown in nanograms per liter for nectar, honey, and hummingbird cloacal fluid and in nanograms per gram for flower homogenates.Evidence of decreased imidacloprid concentrations following regulation in each group is indicated as strong (***) or weak (*) when 0.95, 0.90, or 0.80 Bayesian credible intervals of joint posterior probabilities did not overlap 0, respectively.PPB = parts per billion; ND = nondetects; NS = not sampled.

FIGURE 4 :
FIGURE 4: Flupyradifurone concentrations in media before (dark shades) and after (light shades) Canadian federal regulation of imidacloprid in 2021.Exposure groups include sites within 500 m of agricultural areas sampled either before or after the bloom drop as well as reference sites more than 500 m to the nearest agricultural area.Percentages of samples below the detection limit are shown as percent nondetects.Flupyradifurone concentrations are shown in nanograms per liter for nectar, honey, and hummingbird cloacal fluid and in nanograms per gram for flower homogenates.Evidence of increased flupyradifurone concentrations following regulation in each group is indicated as strong (***), moderate (**), or weak (*) when 0.95, 0.90, or 0.80 Bayesian credible intervals of joint posterior probabilities did not overlap 0, respectively.PPB = parts per billion; ND = nondetects; NS = not sampled.