Exploring honey bee toxicological data as a proxy for assessing dimethoate sensitivity in stingless bees

31 The high diversity and distinctive characteristics of stingless bees pose challenges in 32 utilizing toxicity test results for agrochemical registrations. Toxicity assessments were 33 performed on 15 stingless bee species, along with the honey bee, using the insecticide 34 dimethoate, following adapted OECD protocols. Median lethal doses over 24 h (24h-35 LD 50 ) were determined for exposure routes (acute oral or contact) and species. 36 Species sensitivity distribution (SSD) curves were constructed and the 5% hazard 37 doses (HD 5 ) were estimated based on 24h-LD 50 values. The SSD curve was adjusted 38 as the body weight and dimethoate response were correlated. Lighter bees (<10 mg) 39 had lower 24h-LD 50 values. Contact exposure for adjusted HD 5 suggested insufficient 40 protection for Melipona mondury , whereas the oral exposure HD 5 indicated no risks for 41 the other 14 species. Comprehensive risk assessments are crucial for understanding 42 the agrochemical impact on stingless bees, emphasizing the need for a broader 43 species range in formulating conservation strategies. 44


Introduction
The decline in bees is associated with biotic and abiotic factors, among which the toxic effects of agrochemicals emerge as a contributor to the high mortality rate 1,2 .These chemicals also trigger sublethal effects, affecting behavior, physiology, morphology, and other parameters that can hinder colony success [2][3][4] .Bees are primarily exposed to agrochemicals during foraging and the collection of contaminated resources from crops 5 .
Standardized protocols for the regulation, release, and use of agrochemicals have been proposed in different countries to minimize the impact of agrochemicals on bees.The assessment of toxicological risks in bees unravel the potential adverse effects resulting from exposure to agrochemicals [6][7][8] .In the past decades, regulatory agencies have expanded their efforts to encompass diverse bee groups from the Bombini and Meliponini tribes 1,8,9 .Despite these efforts, the extent of research remains notably limited owing to the lack of standardized guidelines for testing stingless bees, prompting regulatory agencies to extrapolate toxicological data from Apis mellifera, a globally distributed species, to protect other bees 8,10,11 .The Brazilian Institute of the Environment and Renewable Natural Resources (Instituto Brasileiro do Meio Ambiente e Recursos Naturais, Ministério do Meio Ambiente, IBAMA) takes a conservative stance by incorporating a 10× safety factor into the 24 h median lethal dose (24h-LD50) of agrochemicals observed in laboratorial assessments involving A. mellifera for non-Apis species.This precautionary factor is presumed to provide protection for 95% of non-Apis species, encompassing stingless bees 10,12 .
Improving toxicological assessments in native bees is imperative as the current representation of species diversity is deficient.Thus, the applicability of extrapolating data solely from honey bees is debatable 1,4,13 .Notably, the available limited comparative toxicological studies reveal survival disparities between A. mellifera and native bee species in their responses to various agrochemicals 12,14,15 .The emergence of compelling evidence highlighting the decline in native bee colonies in Brazil since 2009 emphasizes the urgency of addressing this issue 16 .To address these knowledge gaps, more comprehensive research should explore the toxicological response of various species to agrochemicals.Such an in-depth investigation will offer a more realistic comprehension of the susceptibility of native bees, thereby significantly enhancing efforts towards their conservation.
Stingless bees (Meliponini) represent a megadiverse clade with diverse characteristics in morphology, nesting habits, behavior, and ecology, particularly in tropical and subtropical ecosystems 17,18 .These bees play a pivotal role as essential pollinators for both native flora and crops 19,20 and have been associated with enhanced productivity in a variety of crops 21 .The recognition of their unique ecological contributions underscores the need for research that addresses the toxicological dynamics of a broad spectrum of bee species.Despite the diversity of stingless bees (>600 species, 244 in Brazil) and their unique biological characteristics 22,23 , toxicological assessments are limited by the lack of methods for maintaining colonies under laboratory conditions 4,24 .
Several species of stingless bees exhibit a broad distribution across the American continent.Notably, Cephalotrigona capitata, Plebeia droryana, Scaptotrigona bipunctata, Tetragona clavipes, Tetragonisca angustula, and Trigona spinipes are widely distributed throughout South America.Tetragonisca angustula is also distributed from Mexico to Central America.In contrast, the geographical distribution of some stingless bee species is restricted to specific regions in Brazil, such as Melipona scutellaris (Northeast), Melipona capixaba (mountains of Espírito Santo State), and Plebeia lucii (Southeast), imposing constraints on the availability of bees for assessments 25,26 .The toxicological responses in a few species of stingless bees (Partamona helleri, M. scutellaris, Melipona quadrifasciata, T. angustula, Scaptotrigona postica, Trigona fulviventris, and Scaptotrigona mexicana) have been compared to that in honey bees 14,[27][28][29][30][31] .This distinctive distribution and endemism of these species underscore an absence of standardization for assays considering regulatory purposes, comparing sensitivity between stingless bee species and honey bees.Such assays are pertinent in environmental risk assessments of agrochemicals, aiming to elaborate conservation strategies and legislation for stingless bees 1,13 .
The species sensitivity distribution analysis (SSD) is a crucial approach in ecological risk assessments, playing a valuable role in determining safe reference parameters for conservation 32 .The hazard dose at 5% (HD5) is a key parameter in establishing safe and sustainable limits in the use of agrochemicals.This specific indicator highlights the agrochemical quantity at which 5% of the evaluated species may be potentially affected 32 .Based on this data, conservation strategies can be devised to protect the bees.However, a significant challenge in applying this analysis lies in the limitation of the number of stingless bee species considered.This reduced representativeness compared to the vast diversity of bee species poses challenges in extrapolating data and conclusions 15 .
The present study compared the sensitivity of stingless bees and honey bees to dimethoate.Dimethoate is an neurotoxic insecticide with systemic action, widely used for pest control in various crops, including fruits, cotton, and ornamentals 6,7,33 .
Despite being banned in certain regions, such as Europe 34 , dimethoate remains approved for use in Brazil, with approximately 703 tons of this organophosphate being commercialized in the country in 2017 35 .This situation implies that bees are potentially exposed to this agrochemical in Brazil.The 24h-LD50 of dimethoate was calculated for both oral and contact exposures, covering 13 and 14 species of stingless bees (10 genera), respectively.Five of these species are prioritized on the IBAMA list for pesticide risk analyses in Brazil 13 .The present study also estimated SSD curves based on 24h-LD50 data adjusted for the weight of bees.Finally, this study contributes valuable data for evaluating the efficacy of the 10× safety factor on 24h-LD50 for honey bees, a parameter established by Brazilian authorities for stingless bees 10 .

Dose-mortality
The choice of the probit model for analyzing the dose-mortality assay results across all bee species was based on the low χ2 value and a significance level of P > 0.05.The 24h-LD50 estimates were calculated for each species (Table 1).

Oral exposure
Variations in dimethoate toxicity were evident among the different species, with P. droryana displaying the highest sensitivity (0.00165 μg a.i./bee) and M. scutellaris exhibiting the lowest sensitivity to the ingestion of the insecticide (0.05363 μg a.i./bee).
Employing a safety factor of 10×, based on the 24h-LD50 of the Africanized honey bee (0.0045 μg a.i./bee), revealed that Nannotrigona testaceicornis (0.00388 μg a.i./bee), T. angustula (0.00356 μg a.i./bee), and P. droryana were not covered by the protection provided by the safety factor adopted by IBAMA, as their 24h-LD50 values fell below the A. mellifera 24h-LD50 value divided by a safety factor of 10.However, adjusting the 24h-LD50 values based on the body weight of the bees ensured the protection of all the tested species because the adjusted LD50 values based on weight were all lower than the 0.00006 μg a.i./mg of bee (LD50 adjusted by weight and a safety factor of 10) for the Africanized honey bee (Supplementary Table 1).

Contact exposure
Similar to the case of oral exposure, a spectrum of dimethoate toxicity was noted among the tested species, with T. angustula and M. scutellaris demonstrating the highest (0.0020 μg a.i./bee) and lowest (0.0682 μg a.i./bee) sensitivity, respectively.
Using a safety factor of 10× based on the 24h-LD50 of dimethoate obtained for A. mellifera (0.0405 μg a.i./bee), the 24h-LD50 values for T. angustula, Friesella schrottkyi (0.0024 μg a.i./bee), and N. testaceicornis (0.0025 μg a.i./bee) were lower than the 24h-LD50 value of A. mellifera divided by the established safety factor of 10 10 .However, adjusting the 24h-LD50 based on the body weight of the bees ensured the protection of all the tested species as the adjusted LD50 values based on weight were all lower than 0.000055 μg a.i./mg of bee (LD50 adjusted for weight and safety factor of 10) for A.

Oral exposure
The SSD curve resulting from oral exposure to dimethoate showed sensitivity variations among species.Light bees, such as Scaptotrigona xanthotricha, N. testaceicornis, T. angustula, and P. droryana, exhibited high sensitivity to dimethoate (Figure 2A).However, a shift occurred when assessing 24h-LD50 values based on body weight (Figure 2B).This suggests that the variability in toxicity, expressed in μg a.i./mg of bee, is more pronounced than when expressed in μg a.i./bee, disregarding the weight of the bees.The HD5 value was 0.0028 μg a.i./bee (Figure 2A), whereas 0.00024 μg a.i./mg of bee when adjusted to weight (Figure 2B).Plebeia droryana had a lower 24h-LD50 than HD5, indicating that over 5% of the species are potentially at risk.However, when calculating the HD5 considering body weight, the protection estimate for all evaluated species exceeded 95%

Contact exposure
The HD5 value derived from the SSD analysis of contact exposure to dimethoate was 0.0022 μg a.i./bee.Tetragonisca angustula exhibited the highest sensitivity after exposure to the insecticide and was not covered by the HD5, as the 24h-LD50 for dimethoate is lower for this species (0.0020 μg a.i./bee) (Figure 3A).However, this changed when evaluating 24h-LD50 values adjusted to body weight.This modification indicated that M. mondury is the most susceptible to dimethoate, as the 24h-LD50 was found to be 0.0001 μg a.i./bee for this species.However, when calculating the HD5 considering body weight, the protection estimates for all evaluated species was at least 95% (Figure 3B).

Discussion
Studies comparing the sensitivity of various stingless bees to that of A. mellifera are scarce 4,13 .Additionally, previous studies have employed a restricted range of species to assess sensitivities in comparison to that of A. mellifera 15,36,37 .Therefore, the findings from the current study, utilizing protocols adapted to the Organization for Economic Cooperation and Development (OECD) guidelines 213 and 214 for assessments, involving 15 species of stingless bees and Africanized A. mellifera (16     species in the total), stands out as a comprehensive approach in evaluating dimethoate toxicity.
As speculated, exposure to dimethoate adversely affected the survival of all tested species, given its classification as a broad-spectrum insecticide within the organophosphate group 38 .Its high toxicity through oral and contact exposure is welldocumented in both adult and larval honey bees 33,39,40 .Dimethoate is commonly used in various crops, such as cotton, tomatoes, and carrots, pollinated by stingless bees including Friseomelitta varia, M. scutellaris, S. bipunctata, P. droryana, P. helleri, T. angustula, T. clavipes, and T. spinipes 13,41 .The identification of dimethoate residues in both pollen and nectar suggests a potential route of exposure for bees associated with these crops 42 .However, owing to the lack of information regarding the consumption of nectar and pollen by stingless bees, estimating whether the dimethoate residues are in a quantity sufficient to cause toxicity becomes challenging.
Body size is a crucial characteristic in assessing the risks of pesticide exposure in bees 43,44 , serving as a key factor in predicting interspecific sensitivity [44][45][46] .A negative correlation exists between body size and sensitivity, significantly influencing insecticide absorption on the body surface 43,44 .Our findings align with similar patterns previously reported for solitary and social species 47 , indicating that stingless bees with lighter body exhibit higher sensitivity to dimethoate.However, M. mondury stands out as an exception.Despite having a weight comparable to the Africanized honey bee, it demonstrated heightened sensitivity to dimethoate.This implies that body weight should be considered in conjunction with other variables such as nutritional status, health, detoxification pathways 48 , and other variables to more accurately explain the sensitivity of certain bees, like M. mondury, to agrochemical exposure.
Depending on the exposure routes, the 10× safety factor failed to protect certain species of stingless bees when LD values are not adjusted for the weight, including F. schrottkyi (oral exposure), P. droryana (contact exposure), N. testaceicornis, and T.
angustula (oral or contact exposure).As discussed above, this outcome might be associated with the lighter weight of these bees relative to the honey bee 46,49 .However, differences in body size could not be the sole explanation in these cases.In contrast to the majority of stingless bee species, where older or foraging bees were used, F. schrottkyi and P. droryana comprised a mixture of nurses (young) and foragers.In the case of N. testaceicornis, swarming drones (aged) were used.
Regarding the age factor, a consensus is absent in the literature regarding the differential sensitivity to agrochemicals concerning honey bees.Ageing is associated with a greater capacity for detoxification in A. mellifera 50,51 .However, mortality in this species increases with aging when considering contact exposure to organophosphates 51,52 .Another important factor related to age that could influence the level of exposure is the ability of the agrochemical to penetrate the cuticle.In this sense, P. droryana and T. angustula might be highly affected by contact exposure, as the collected individuals included both nurses with less rigid cuticles and foragers with fully formed integuments 53 .Drones (haploid) may also respond differently from workers.In A. mellifera, drones have been demonstrated to be more susceptible to neonicotinoids than workers, and heterozygosity may be important in minimizing sublethal effects 54 .
By adjusting the 24h-LD50 of dimethoate relative to body weight, the safety factor of 10 successfully protected all species.Our findings align with observations in other wild bees, such as Bombus terrestris and Osmia bicornis, where this safety factor demonstrated protective efficacy 47,55,56 .This emphasizes the suitability of the honey bee as a surrogate species for assessing acute oral and contact exposure to dimethoate.However, the necessity for further research on bee response to diverse agrochemicals is crucial, as relying solely on safety factor extrapolation may not ensure comprehensive protection in all situations, as was particularly evident in the context of neonicotinoids and various Brazilian non-Apis species 37 .Hence, comprehensive risk assessments should compare sensitivity across different species, concomitantly consider the implemented safety factor for the protection of non-Apis species.These assessments should encompass various species and agrochemicals 25 .
Species distribution on the SSD curve, considering body weight in this study, revealed a relatively minor difference in sensitivity to dimethoate among stingless bees compared to that in A. mellifera.The application of the SSD approach emerges as a practical alternative to intricate studies performed in field and semi-field environments 57,58 .The HD5, after adjustment for body weight in oral exposure, is lower than the dimethoate 24h-LD50 value for all species.However, through contact, M.
mondury had a higher 24h-LD50 than HD5 value.These findings reinforce the conclusion that smaller bees do not face a higher risk when using HD5 as a parameter in the assessment of dimethoate toxicity.These results align with previous studies that adopted the SSD approach and calculated the HD5, thus consolidating the reliability of this method in toxicological assessments in bees 37,47,59 .Therefore, analyses that do not consider the weight adjustment in 24h-LD50 values for non-Apis species must be carefully evaluated.
The present study provided significant insights into the toxic effects of dimethoate on stingless bees in comparison to A. mellifera.It serves as an initial investigation assessing the impact of this insecticide, following OECD protocols, across an extensive range of species spanning 10 evolutionarily distinct genera of Meliponini.Moreover, it marks the first instance of M. mondury and C. capitata being included in ecotoxicological studies.Investigating different routes of exposure and correlating the presence of stingless bees in crops subject to the application of agrochemicals is fundamental for determining which species are genuinely exposed 13 .
The results of the present study also emphasized the influence of body weight on susceptibility, highlighting that lighter-bodied tropical eusocial bees do not face a higher risk after exposure to dimethoate.Adjusting 24h-LD50 values based on body weight indicated that all species remained within safe limits, considering the 10× safety factor applied to the A. mellifera regulatory endpoints.Notably, HD5 emerged as a risk parameter for most bee species, except for M. mondury in contact exposure.This distinct response within Melipona underscores the importance of considering multiple exposure routes for a more comprehensive assessment.Finally, these results highlight the need to deepen our comprehension of the distinct vulnerabilities exhibited by eusocial bees in tropical environments.Recognizing and understanding these unique vulnerabilities are essential for formulating targeted conservation strategies that can effectively preserve pollinators in the tropics.
discarded.Specimen identification was based on their phenotype and on the structural characteristics of the colonies, including the entrance and internal organization of the combs 26 .Details on bee collection are provided in the Supplementary Information (S1.Supplementary Methods).

Dose-response assays
Thirteen and 14 species of stingless bees, along with A. mellifera, underwent oral and contact exposures, respectively (Supplementary Table 2).Diets (with or without dimethoate) were administered through perforated 2 mL centrifuge microtubes fitted into a hole (~13 mm in diameter) at the bottom of plastic pots 24 .In the case of oral exposure, bees underwent a 1-h fasting period to stimulate consumption of the contaminated diet.These bees had access to a diet containing dimethoate for 6 h.
After this period, the feeder with dimethoate was replaced with a new one containing only sucrose solution, available until the end of the experiment.For both oral and contact exposure assays, mortality assessments were conducted 24 h after the initiation of exposure and monitored daily for up to 96 h from the start of the experiment, with insects considered completely immobile classified as dead.
The bees were grouped in sets of 10 individuals in plastic pots.Each dose of the insecticide was assessed in bees selected from three different pots, resulting in three replicates and a total of 30 individuals per dose.The total number of individuals used for each species is described in the Supplementary Table 2.In the case of bees originating from three or more colonies, each of the three replicates was represented by bees from distinct colonies.For species originating from two colonies, two replicates comprised individuals from the same colony.For species in which only one colony was used, the three replicates consisted of individuals from the same colony.Additionally, a control group, also consisting of three replicates, was established for comparative purposes.
To assess the relationship between body weight and sensitivity to dimethoate, a separate set of 20 bees from different colonies was weighed while still anesthetized to determine the mean body weight using a precision scale (AG 200, Gehaka, São Paulo, Brazil) (Supplementary Table 1).

Oral exposure
For oral exposure, the insecticide was diluted in a 50% sucrose solution.
According to the OECD protocol 6 , the reference 24h-LD50 for dimethoate falls within the range of 0.10 to 0.35 µg of a.i.per forager of A. mellifera.Therefore, preliminary range-finder tests were performed to establish five dimethoate doses within the OECDrecommended range, covering the range of 1 to 100% mortality.These doses are necessary to estimate the 24h-LD50 for each species (Supplementary Table 3).Each insecticide dilution was administered as described above.
The feeder with the contaminated diet was weighed before and after exposure using the precision scale to determine diet consumption.This allowed for the determination of the average amount of food consumed per bee.Three plastic pots, each with its respective feeder containing sucrose solutions but without bees, were placed in the incubator during the experiment to estimate the evaporation rate.This rate was then subtracted from the bees' diet consumption to determine the average food consumption per bee 24 .The 24h-LD50 values for oral exposure were calculated based on the average insecticide consumption per bee.

Contact exposure
As per the OECD protocol 7 , the reference 24h-LD50 for dimethoate falls within the range of 0.10-0.30µg of a.i./A.mellifera bee.Adhering to these guidelines, a stock solution was prepared, comprising 15 µg of the a.i. of dimethoate in 15 mL of deionized water.The maximum treatment dose, derived from the stock solution, was determined as the application of 0.20 µg of a.i.per bee.Subsequently, five dilutions were meticulously prepared, each supplemented with the same volume of acetone to reach a final volume of 10 mL.The control group bees received only acetone at an amount equivalent to the insecticide solution received by the bees.These doses were carefully selected to encompass the spectrum of 1 to 100% mortality, facilitating the determination of 24h-LD50 values for each species (Supplementary Table 3).Acetone, recognized for its non-toxicity to A. mellifera, was selected as a solvent, considering it appropriate for use as a control compound 7 .
The insecticide solution was delivered to the dorsal region of the thorax of anesthetized bees using a micropipette with a maximum volume of 5 µL, by touching the bees with the tip of the plastic tip.Species weighing less than 9 mg received 0.5 µL of the solution, those weighing between 9 mg and 40 mg received 1 µL, and species weighing over 40 mg received 2 µL (Supplementary Table 2) The diet provided throughout the experiment was a 50% (w/v) sucrose solution.

Statistical analyses
Mortality values were subjected to probit analysis to estimate the 24h-LD50 of each bee species, accompanied by the respective 95% confidence intervals (CI-95%) using PROC PROBIT from SAS Institute 60 .To explore the potential relationship between the log-transformed 24h-LD50 and the body weight of bees, a linear model was applied, and the effect was statistically examined using the F-test (P < 0.05).
Residuals were scrutinized through visual inspection, facilitated by quantile-quantile plots.The entire analysis was executed using the R software (version 4.3.2.) 61 .
For comparing susceptibility among species, an SSD curve was obtained using the 24h-LD50 values of all stingless bee species and A. mellifera.The LD50 values for all bee species were adjusted with the fresh body weight of each species.
Subsequently, SSD curves were estimated for both unadjusted and adjusted LD50 values, using a log-linear model.The HD5 and the hazard dose for 50% (HD50) for stingless bees and A. mellifera were estimated using the ETX software (effective toxicity exponent, ETX 2.3) 62,63 .Lognormality was tested using the Anderson-Darling test at a significance level of 5%.

Figure 2 .
Figure 2. Species sensitivity distribution (SSD) curve for stingless bees and Apis mellifera (highlighted in red) exposed orally to dimethoate for 6 h.(A) Estimated SSD curve using 24h-LD50 (µg a.i./bee).(B) Estimated SSD curve with 24h-LD50 adjusted for fresh body weight, using the dose per milligram of bee.The x-axis represents species 24h-LD50 values, whereas the y-axis corresponds to cumulative probabilities.The dashed blue line indicates the hazardous dose affecting 5% of the individuals (HD5).* Species not protected by the HD5.

Figure 3 .
Figure 3. Sensitivity species distribution (SSD) curve for stingless bees and Apis mellifera (highlighted in red) exposed to dimethoate by contact.(A) Estimated SSD curve using 24h-LD50 (µg a.i./bee).(B) Estimated SSD curve with 24h-LD50 adjusted for fresh body weight, using the dose per milligram of bee.The x-axis represents species 24h-LD50 values, whereas the y-axis corresponds to cumulative probabilities.The dashed blue line indicates the hazardous dose affecting 5% of the individuals (HD5).* Species not protected by the HD5.

Table 1 .
Estimated 24h-LD50 of dimethoate through acute oral and contact exposure in honey bee and stingless bees