PP 3/10 (3): Chapter 10: honeybees

  • It is acknowledged that exposure of honeybees may also occur through routes such as the consumption of guttation droplets. However, at the time when this revision was prepared, data was too scarce to allow a description of the conditions of exposure needed for a proper risk assessment, in particular with regard to the conditions of occurrence of droplets, from which a standardized estimation of the exposure could have been deduced.
    Another problem is contact with or consumption of plant material contaminated with seed dusts originating from seeds coated with plant protection products. The issue over dusts from seed treatments has been addressed by amendment Commission Directive 2010/21/EU of 12 March 2010 amending Annex I to Council Directive 91/414/EEC as regards the specific provisions relating to clothianidin, thiametoxam, fipronil and imidacloprid. Further work is needed in these areas, which will be considered by the ICPBR. Further revisions will include this and related risks to honeybees, when information is available.


Specific scope1

This standard provides an assessment of the risks presented by plant protection products to honeybees. It was revised to include the risk presented by systemic plant protection products.

Specific approval and amendment

First approved in 1992-09.
Edited as an EPPO Standard in 1998.
Revised in 2002-09 and 2010-09 (updated with ICPBR recommendations).


The subscheme in this chapter is concerned with the potential risks to pollinating insects from the use of plant protection products. It specifically addresses the assessment of risks to honeybees (Apis mellifera) and their brood and colonies arising from exposure of worker bees to insecticides and other plant protection products while they are foraging away from their colonies.

There is also an increasing need to protect other important pollinators (e.g. bumble bees). In principle, this could be approached by adapting the subscheme so that it applies specifically to other species. However, there is insufficient information available about other pollinators to permit an assessment in comparable detail. Also, populations of other pollinators are considerably more difficult to handle and study than honeybee colonies. Therefore, it would be preferable to make predictions for other species by extrapolation from the large body of data on honeybees, among other species. This approach is being investigated under the auspices of the International Commission for Plant–Bee Relationships (ICPBR), formerly the International Commission for Bee Botany (ICBB), by examining correlations between species for susceptibility and exposure to existing products.

In its content and technical approach, the subscheme uses EPPO test methods for evaluating the side-effects of plant protection products on honeybees. OEPP/EPPO (2010) provides details of the main test protocols referred to in the subscheme. These are based on recommendations of the ICPBR (Felton et al., 1986; Lewis et al., 1998, 2009), and are fully in line with other international guidelines (e.g. FAO, 1989; Council of Europe, 1992).

The subscheme adopts the assumption that the most reliable risk assessment is based on data collected under conditions which most resemble normal practice, i.e. by field tests or by monitoring the product in use. Such studies are relatively expensive and difficult to conduct, but the results should be considered as decisive if there is any conflict with results from lower-tier testing (laboratory and semi-field testing). Field studies may also be difficult to interpret. For example, the extrapolation of results from the tested crop to others relevant to the use of a product may need to be considered. This applies particularly where exposure is expected through residues in pollen or nectar following use of soil or seed treatments with systemic products. In these cases, exposure of pollinators is not only related to the application rate, but will also be influenced by the systemic properties of the active compound and attractiveness of the crop.

The proposed tiered or stepwise approach (laboratory to semi-field to field) is thus aimed at triggering higher-tier (cage and field) testing in those cases where an unacceptable risk cannot be excluded. As for other subschemes, it is always possible to go straight to higher-tier tests after the initial risk assessment (i.e. miss out intermediate levels of testing) if it is judged that they may be triggered, and to remove any uncertainty. It should be kept in mind that field tests are demanding, and consideration should be given to how far the results from each one can be extrapolated between different crops, regional conditions, etc. A simplified diagram of the scheme is provided in Fig. 1.

Figure 1.

 Simplified diagram of the subscheme for evaluation of the risk of a plant protection product to honeybees. TER, toxicity exposure ratio; HQ, hazard quotient.

History of the scheme

This scheme is a revised version of the original EPPO Standard PP 3/10. A validation of the scheme in the UK (Aldridge & Hart, 1993), a validation of the scheme by the EPPO Panel on Environmental Risk Assessment among European Designated National Authorities, and recommendations of the ICPBR Symposia on Honey Bee Protection (Lewis, 1996; Belzunces et al., 1999) resulted in a first revision of the document (2002). The basis for this revision includes developments and updates with regard to larval testing (Aupinel et al., 2005, 2009; Becker et al., 2009), semi-field and field testing (Lewis et al., 2009) and risks posed by seed coating and soil applications (Alix et al., 2009a, 2009b).

Risk assessment scheme

Details of the product and its pattern of use

1. From Chapter 2 take the basic information on the product and its pattern of use.

  • If the product is to be applied as a spray, enter the following information:

  •  time of treatment in relation to crop flowering
  •  any special directions for use including measures prescribed for risk mitigation.
  • go to 2a

  • If the product is an insecticide for soil treatment (e.g. granules) or a seed treatment, enter the following information:

  •  is the crop or a rotational crop attractive to bees?
  •  is the active substance or its residues systemic in plants?
  • go to 2b

Possibility of exposure

2. Is exposure of bees possible?

  • 2a. For a product to be applied by spray (see Note 1)
  • if yes go to 3a and 4

  • if no (e.g. food storage in enclosed spaces, pre-emergence herbicides etc.) go to 10

  • 2b. For a product that is an insecticide for soil treatment (e.g. granules) or a seed treatment (see Notes 2, 3 and 4)
  • if yes go to 3b and 4

  • if no go to 10

Preliminary screening based on toxicity

3. Assess the toxicity of the product to worker honeybees by conducting LD50 (contact) and LD50 (oral) laboratory tests.

  • 3a. Most products that are applied as sprays can be evaluated initially by considering the likely exposure of bees and the toxicity of the product.
  • Calculate the ratio between the application rate and toxicity (g ha−1/LD50 in μg per bee) (see Note 5).

  • if ratio < 50 go to 11

  • if ratio ≥ 50 go to 82

  • 3b. Calculate the toxicity exposure ratio (TER) between the LD50 (oral) and exposure. Exposure is assessed by estimating the amount of residues that may be ingested by a bee in 1 day (see Note 6).
  • if ratio ≥ 10 go to 11

  • if ratio < 10 go to 7

Identification of potential risks to larvae

Questions 4–6 identify cases in which honeybee larvae may be at risk, for which special tests may be appropriate, and allow indirect effects to be considered (e.g. systemic intoxication through feeding on nectar or pollen, delayed action, and alteration of behaviour).

4. Can effects on growth or development of bees be excluded (risk assessment for bee brood triggered)? (see Note 7).

  • if no go to 5

  • if yes go to 11

5. Conduct a bee brood-feeding test (see Note 8). Effects on brood may be assessed qualitatively or quantitatively depending on the test that is performed.

  • In the case where effects are assessed qualitatively, are effects observed on bee brood significant?

  • if yes go to 6

  • if no go to 11

  • In the case where effects are assessed quantitatively, calculate the ratio (TER) between the no observed effect level (NOEL) and exposure. Exposure is assessed by estimating the amount of residues that may be ingested by a bee in 1 day (see Note 9).

  • If ratio < 1 go to 6

  • If ratio ≥ 1 go to 11

6. Second-tier risk assessment for larvae.

This question applies in the cases where a quantitative assessment could be performed in question 5. At this step, a refinement of exposure may be performed in the cases where data on residue concentrations in pollen are available (see Note 9).

  • If ratio < 1 go to 8

  • If ratio ≥ 1 go to 11

7. Second-tier risk assessment for adults.

Refine the risk assessment on effects and/or exposure side.

Lethal effects can, for example, be assessed over a prolonged period that represents the duration of exposure of foragers during flowering (determination of a 10-day NOEL).

Exposure assessment may also be refined by measuring residues in pollen and nectar of the treated crop.

Calculate the new ratio between the NOEL (oral) and exposure. Exposure is assessed through the amount of residues that may be ingested by a bee in 1 day (see Note 10).

  • If ratio < 1 go to 8

  • If ratio ≥ 1 go to 11

Semi-field and field trials

The results of field tests are more directly relevant to practical conditions than those of semi-field. Therefore semi-field tests are not generally necessary if a field trial has been carried out. This stage of testing also provides an opportunity to develop means of minimizing effects, by extending the cage or field tests to examine patterns of use which might reduce risk by causing less exposure. Such extra testing is an optional supplement to the risk assessment procedure, which may aid risk management.

8. Conduct a semi-field field trial in conditions representative of use (see Note 11). Are effects on colony survival and development significant (see Note 13)?

  • If yes go to 9

  • If no go to 11

Field trials

9. Conduct a field trial in conditions representative of use (see Note 12). Are effects on colony survival and development significant (see Note 13)?

  • If no go to 11

  • If yes go to 12 or 13– depending on the level of effect.

Categories of risk

The preceding stages of assessment allow uses of plant protection products to be allocated to three categories of potential risk to honeybees.

10. Categorize as negligible risk to bees.

11. Categorize as low risk to bees.

12. Categorize as medium risk to bees go to 14.

13. Categorize as high risk to bees go to 14.

Analysis of uncertainty

After completing the risk assessment based on data reflecting normal use of the product, the assessor should consider whether errors in measurements, or variations in conditions of use, might alter the conclusions.

14. Review the data which led to the medium- or high-risk category and check whether the conclusions are correct (see Note 14).

  • If yes, confirm assessment go to 15

  • If no, obtain more information as needed go to 8

Risk management

15. The following points give guidance on the steps that might be appropriate in order to mitigate effects on honeybees, for products in each of the categories of risk.

  • If risk is low (i.e. level of exposure leads to acceptable risks) or negligible (i.e. no exposure): set no restrictions on use.

  • If there is a medium to high risk: consider conditions that would limit or exclude exposure of bees. For example, allow use only in crops which are not visited by bees. Consider the persistence of residues in soil and possible exposure through rotational crop, and consider related recommendation with regard to rotational crops in contaminated soils. Mitigation measures should be proposed, e.g. moving beehives away from the treated crops or application outside the flowering period.

According to the Directive 2003/82/EC (standard phrases for special risks and safety precautions for plant protection products), these indications or restrictions should be mentioned in standard phrases for safety precautions for the environment, in the case of honey bees SPe8: ‘Dangerous to bees/To protect bees and pollinating insects do not apply to crop plants when in flower/Do not use where bees are actively foraging/Remove or cover beehives during application and for (state time) after treatment/Do not apply when flowering weeds are present/Remove weeds before flowering/Do not apply before (state time).…’. Specific phrases may be proposed based on the conditions that would lead to a limited or excluded exposure.

Explanatory notes

Note 1. Possibility of exposure to sprayed products

The most important route of exposure of honeybees to plant protection products is by direct exposure to field sprays. In some cases, exposure of bees is not possible and there is no need for a detailed assessment of risks. Examples are: use during winter when bees are not flying; pre-emergence use of herbicides, indoor use and use in glasshouses where bees are not used for pollination; seed treatments and granules except when there is systemic activity; products for dipping bulbs, etc. However, any crops in which there are flowering weeds, or which might be overflown by bees visiting other crops, may present a risk of exposure, even if the crops themselves are not attractive to bees. In such cases, it is prudent to regard exposure as possible and to continue with the assessment.

Note 2. Establish if the crop is attractive to bees

The attractiveness of the crop plant to honeybees may be considered as an entry point for this risk assessment. Useful guidance in this respect may be found in the document of the MRL Working Group (EC, 2009). There are also recommendations on additional criteria to consider, such as the presence in the foraging area of other sources of nectar/honeydew of higher/lower level of attractiveness, i.e. weeds, which may influence the behaviour of bees towards the crop of interest. In general, a crop can be considered as unattractive to bees when it is harvested before flowering. Some plants that are intrinsically not attractive to bees may be visited due to extra floral nectaries, e.g. in field beans, or due to honeydew produced by aphids on crops otherwise not attractive to bees. Similarly, the presence of bee-attractive flowering weeds or of ‘secondary’ crops (e.g. associated flowering crop, mixed cropping, intercropping) in an unattractive crop may result in bee activity that leads to some exposure. A description of agricultural practices associated to the crop of concern may help in deciding whether or not visits and exposure are to be expected.

Note 3. Establish if rotational crops have to be considered in the risk assessment

The persistence of the product in soil may result in exposure of bees, in the case of the growth of an attractive plant in the rotation. Criteria to identify persistent substances have been defined in Regulation 1107/2009; in general these require additional residue studies involving crop rotation. In the case of residue transfer into rotational crops, investigations specifically to address the risks to bees from attractive plants grown during the rotation with the treated crop become necessary.

Note 4. Establish if the substance or its residues present systemic properties

The exposure of honeybees to plant protection products used for soil or seed treatments may occur in the case of transfer of the active substance itself, or its degradation products, to the parts of the plant that may be consumed by bees, i.e. nectar, pollen or honeydew. Exposure to contaminated honeydew is not considered a relevant route in the case of soil and seed treatments. This is because the concentration of a systemic compound that could circulate in the phloem and reach honeydew without harming aphids should, in principle, not be capable of harming bees foraging on the honeydew, unless the compound is highly selective towards non-aphid insects. Selectivity information (available in the registration dossier) should, in principle, allow the highlighting of such a selectivity, which would then trigger a dedicated risk assessment according to the present subscheme.

Information derived from residue studies and plant metabolism studies (residue section of Annex II and Annex III dossiers according to Regulation 1107/2009) is, in general, sufficient to identify if the substance is transferred into the plant during its growth, and if it is further degraded into major degradation products. Similarly, possible uptake in plants of major soil degradation products will be identified in these residue studies. The plant protection product is considered systemic in case of uptake and transfer into the plant.

The sensitivity (i.e. limit of quantification and detection) of the analytical methods that are used in the residue studies should be checked in order to ensure that they are low enough to detect residue levels that exert toxic effects to honeybees. If it is uncertain whether the detection methods were sufficiently sensitive, additional investigations have to be considered to demonstrate the absence of residue translocation at toxic levels. Beside this verification, studies that specifically investigate the presence of residues in flowers, nectar or pollen are not necessary at this stage.

Note 5. Toxicity tests

Suitable methods for toxicity tests are described by OEPP/EPPO (2010), OECD (1998a, 1998b).3 Contact and oral toxicities (LD50) tend to be of the same order of magnitude. If these deviate considerably, and this cannot be explained by the mode of action of the compound, this may indicate unreliability of the data. As the main route of hazardous exposure to acutely toxic compounds is through contact action, the contact LD50 is most important for insecticides, while the oral LD50 is more relevant for the assessment of compounds that are not acutely toxic, such as herbicides. To achieve a good margin of safety, the risk assessment should be carried out selecting the lowest of the oral and contact LD50 values.

The ratio between application rate and toxicity (also referred to as a hazard ratio) gives an approximation of how close the likely exposure of bees is to a toxicologically significant level.

In calculating the ratio (dose ha−1/LD50), dose ha−1 is the highest application rate in g active substance per ha, and LD50 is measured in μg active substance per bee. The threshold is determined on the basis of bee toxicity, dosage rate and an independent classification of risk verified by extensive practical experience of plant protection products.

For more information about the derivation of the hazard ratio and the suggested thresholds, see Oomen (1998), Smart & Stevenson (1982), and Mineau et al. (2008).

For decision making in this first tier, both the toxicity value (LD50) and the application rate in the hazard ratio should always be expressed either in active substance or in the formulated product, but these should never be used together within one evaluation. Products containing mixtures of active substances should be evaluated by entering the toxicity and application rate of the formulated product only.

Note 6. First-tier risk assessment for soil and seed treatments

Suitable methods for acute oral and contact toxicity tests are described by OEPP/EPPO (2010), OECD (1998a, 1998b)– see Note 5 Toxicity tests.

The main route of exposure of honeybees to soil/seed treatment is oral through the consumption of contaminated pollen and nectar, although a contact exposure cannot be excluded for bees carrying pollen that contains residues. It has to be noted, however, that topical exposure through contaminated nectar may also occur for sprayed, non-systemic compounds.

In this respect, the first-tier risk assessment focuses on acute oral risks. A first-tier toxicity exposure ratio (TER) is calculated based on the acute oral toxicity figure for adult bees and on an assessment of the exposure through, ideally, pollen and nectar. Residues in pollen and nectar are rarely quantified in residue studies that are available in the residue section of dossiers, as these studies are performed for other (risk to consumers) purposes. The transfer and fate of products and their residues in plants is not homogeneous, and transfers to the blossom depend on their ability to cross the flower barrier. Thus estimates of the concentration in the aerial parts of the plant may be considered as an overestimation of residual concentration in nectar and pollen, and provide a useful margin of safety as a first assessment step. In the case where such data on residues in plant material are considered not reliable or available, a generic worst case value of 1 mg (a.s.) kg−1 plant matrix is proposed. This value is deduced from a compilation of the data generated in various plant species treated with systemic insecticides and the consequent residue concentrations measured in all types of plant parts (leaves, fruit, green parts, inflorescence, whole plant, grain) at the period as close as possible to blossom, as well as residues measured in nectar and pollen. The results displayed a majority of samples with <1 mg active substance (a.s.) kg−1 matrix (95th percentile = 0.55 mg kg−1, n = 62), the same being observed for degradation products. Taking the matrices nectar and pollen separately, residue concentrations would not reach more than 0.1 mg a.s. kg−1.

Because it is a worst case assessment, exposure estimates should reflect the maximal expected residue levels. When based on measured residue in plant matrices, the 90th percentile of the data set of residue data for the relevant crop should be selected at this step.

The oral LD50 is measured in μg active substance per bee and residues in plant parts are expressed in mg kg−1. Therefore a conversion of residue data is necessary to express exposure as an amount of residue ingested. This conversion may be done by multiplying the 90th percentile of residue concentration (mg a.s. kg−1 plant part) by the daily food ingestion that reflects the dietary need in sugar for a bee. The maximum food ingestion may be estimated from Rortais et al., 2005 at 128 mg per bee per day for nectar foragers. The data set provided by Rortais et al. (2005) is proposed as it is considered to represent food consumption estimates of the different categories of bees satisfactorily. Other figures for food ingestion may become available and could be used if it is demonstrated that they provide a better estimate of this exposure route.

The calculation of a TER gives an approximation of how closely the likely exposure of bees is to a toxicologically significant level. The margin of safety achieved should be sufficient to cover the uncertainty related to longer exposure periods and possible related increased effects. To quantify the range of this uncertainty, the comparison of toxicity values for adults from acute tests and from chronic (10-day) tests was done for 7 substances (Defra, 2007). The results show that the LD50 expressed in μg a.s. per bee per day as derived from 10-day studies can be derived from 48 h LD50 by applying an adjustment factor of 10, for acute toxicity data ranging from 0.13 to 90 μg per bee. Despite the need for further work to confirm this correlation with a wider range of compounds, this factor is considered sufficient to cover uncertainties related to the influence of exposure duration on toxicity levels.

Note that for low-toxicity figures (e.g. LD50 of 10 μg a.s. per bee and above), TER calculations will always result in values above the trigger (= low risk) even with exposure levels estimated from concentrations in aerial parts. However, a definite cut-off value for identifying the need to enter to enter the risk assessment scheme through a Tier 1 TER is difficult to establish but as this calculation involves only the acute oral toxicity test in adults and no additional experiments, a toxicity based trigger is not deemed necessary.

Note 7. Insect growth regulators

Insect growth regulators (IGRs) and substances that display effects specifically on juvenile stages, apparent from screening and efficacy studies and from tests with other non-target arthropods (including terrestrial and aquatic), have to be assessed more precisely with a bee brood-feeding test (Note 8).

Note 8. Bee brood-feeding tests

A suitable method is described by Oomen et al. (1992). The test should be performed at the highest expected level of exposure (the maximum level of exposure is supposed to kill foragers) as measured in plant parts, or, if available, in nectar or pollen, or other environmentally relevant exposure concentration determined experimentally.

For systemic treatments, as the level of exposure may vary from one crop to another and probably also between samples of the same crop, it is not necessary to duplicate the study to take the variability of exposure levels into account. Rather, the test should allow the determination of a NOEL in order to assess the risk for bee brood with, for example, the calculation of a TER that would give an approximation of how close the likely exposure of bee brood is to a toxicologically significant level for a particular crop. The test methodology developed by Aupinel et al. (2005) may be considered in this respect, although it is not available as an OECD or an EPPO method at the time of this risk assessment revision. Note that, as exposure level may differ from one crop to another, and considering possible persistence issues in soils, TERs should be calculated for each crop separately to ensure that the trigger (see question 5) is reached in each case.

There is insufficient data available, particularly on exposure of brood, to relate larval toxicity (assessed by methods described by several authors, e.g. Wittman & Engels, 1981) to field application rates and brood damage. Therefore, if any effects are detected in a bee brood-feeding test, semi-field or field testing becomes necessary.

Note 9. TER calculation for larvae

In cases where a quantitative appreciation of effects could be made from the test performed, a NOEL may be derived, which may be used in a TER calculation. Exposure levels are estimated as for adults, i.e. in the first instance based on the concentration in the aerial parts of the plant as a surrogate for residual concentration in nectar and pollen. In the case where such data on residues in plant material is not considered reliable or available, a generic worst case value of 1 mg (a.s.) kg−1 plant matrix is proposed. Data on maximum food consumption, if needed for the calculation, may be found in Rortais et al. (2005).

For further refinement of TER calculation, residue concentration as measured in pollen may be used (see Note 10). As for adults, a 50th percentile may be used in the calculations.

Note 10. Second-tier risk assessment for adults

Additional information with regard to toxic effects may be incorporated by including the duration of exposure of foragers in the assessment of effect. This should be performed by conducting a toxicity test in which worker honeybees are fed treated sucrose for 10 days to calculate a 10-day NOEL (mg a.s. per bee per day). The method of Decourtye et al. (2005) could be used, although it is not available as an OECD or an EPPO method at the time of this risk assessment revision. Usually, a lower LC50 is measured over a 10-day period than after an ingestion period of several hours (Defra, 2007). Thus uncertainty with regard to chronic exposure is considered to be addressed by this test.

A refinement of the exposure may be made by measurements of the residues in pollen and nectar (if relevant) in plants grown from coated seeds or sown in a treated soil according to the intended Good Agricultural Practices (GAPs). These residue levels will reflect the exposure levels experienced in the actual crop. Possible build-up in soil due to residue persistence, based on criteria, and use of the substance in the rotation, should be considered if expected. Since exposure has to reflect a period of several days, the mean value of the concentrations measured in samples could be used in the TER calculation.

The Tier 2 TER should be calculated with the NOEL from the 10-day chronic toxicity test in bees and/or the measured level of residues in the relevant material for honeybees (mean residue data). A further refinement of both effects and exposure is not necessary, but it is rather to be considered as a possibility, especially when there is evidence that the refinement of either effect threshold or exposure level will be sufficient to reach the trigger value. If a 10-day test-derived NOEL is used in the TER calculation, the 50th percentile for residue concentration may be used, as it is considered more relevant to reflect a chronic exposure. Note, however, that the trigger value remains unchanged in the case of a single exposure refinement as the uncertainty with regard to chronic effects remains. Again, toxicity and exposure data should be expressed in the same units.

Note 11. Semi-field and field trials

Suitable methods for semi-field and field trials are discussed in OEPP/EPPO (2010). They can be adapted to soil/seed treatments (systemic activity) and may also be modified for specific assessments with honeybees, e.g. repellency and other behavioural effects, effects of aged residues or for specific testing of brood effects. Semi-field testing (cage, tunnel or tent tests) is a suitable option before field testing. The advantage is that potential mortality is easier to assess and exposure is worst-case (colonies confined to treated crop) and can be readily demonstrated. Cage studies may also be used for evaluation of the hazard of application of plant protection products to honeybees foraging the honeydew secreted by aphids.

The design of trials should be influenced by the characteristics of the chemical and its effects on bees, revealed by the earlier tests. Exposure in a cage or tunnel is more intensive than in the field. The product tested is therefore regarded as presenting a low risk if the effects on colony survival and development are similar to those in a non-pesticide control, provided that environmental conditions are suitable for the detection of hazards to bees. Semi-field and field trials should be conducted under conditions reasonably representative of the uses to be prescribed (appropriate application rate and treatment/sowing rate for seed treatments). In the first instance, a standard crop highly attractive to bees should be used as test plants. In other cases, identification of a surrogate (worst-case) test crop may be more difficult, e.g. for systemic compounds, where the test crop should be one for intended use. Crops on which use of the product is proposed may also be appropriate if significant effects are seen with the standard attractive crops. This also ensures that testing takes place under the specific conditions of exposure (e.g. in relation to duration of flowering) expected in the field. In the case of systemic activity, if the substance or its residues are persistent and the product may be used on several crops in a rotation, the accumulation in soil should be considered in the study protocol.

Possible effects on adult survival and foraging behaviour and on bee colonies should be checked. Where pollen or nectar containing residues are brought back to the hive, colonies should be monitored for a sufficient period to check long-lasting or delayed effects.

For both semi-field and field trials, it should be demonstrated that the test bees were exposed under the environmental conditions of the trial (especially weather conditions in the case of field trials). In semi-field studies, this is normally achieved by the use of a toxic standard, but in field trials this should be done by assessments of flight and/or foraging activity in the crop (parameters such as pollen collection and residue analysis may also provide useful information). A quantified assessment of the exposure is particularly important for systemic products, as reference substances for systemic products are difficult to define, being also dependant on crop properties. There should always be a comparable untreated control in order to provide a reference point against which to compare the test treatment(s).

Note 12. Field trials

Field trials serve to classify all remaining plant protection products. Suitable methods are discussed in OEPP/EPPO (2010). The design of trials should be influenced by the characteristics of the chemical and its effects on bees, revealed by lower-tier tests. It should be demonstrated that the test bees were at risk under the environmental conditions of the trial (especially weather), preferably by pollen analysis, assessing the flight intensity at the time of application in the field and by observation of the activity at the hive entrance.4 There should always be a comparable, water-treated control in order to provide a reference point against which to compare the test treatment(s). A reference product known to present a low risk may also be useful, in order to enable evaluation of the effects of the test product on colony survival and development and arrive at an appropriate category of risk.

Special effects (larval toxicity, long-term residual effect, disorienting effects on bees, etc.) identified by the field test may require further investigation using specific methods. If field trials are effectively impossible (e.g. for evaluating the hazard to bees foraging on honeydew secreted by cereal aphids), large-scale (tunnel) trials may replace field trials.

Note 13. Significance of field results

Effects as a result of the experimental treatment in semi-field or field trials may be difficult to assess and to distinguish from other sources of mortality. Statistical analysis of the results may help address this problem. When interpreting the results, it needs to be recognized that there are endpoints which are intrinsically suitable for statistical evaluation (e.g. mortality data) whereas others may be unsuitable (e.g. behavioural endpoints). In addition, the evaluation needs to consider the range of parameters assessed and their relative importance, which will depend on the specific objectives and design of each study and must be considered on a case-by-case basis.

Due to the limitations on replication in field studies and the inherent variability in most of the relevant endpoints assessed, it has to be recognized that statistical analysis may not be feasible. Whether or not statistical analysis is available, expert judgement will be needed to assess the biological significance of any effects seen in the context of each colony and the test conditions. This will also be needed to consider the relative importance of the various parameters assessed, in the context of impact on overall colony health and the specific aims of each study.

Note 14. Additional investigation

Special effects (larval toxicity, long residual effect, disorienting effects on bees, etc.) identified by the field test may in some cases require further investigation using specific methods, particularly in the case where these effects are observed under realistic exposure conditions, as this means they may also be expected under the intended conditions for use of plant protection products. Such investigations should then be focused on assessing the importance and significance of effects and to help in setting any risk mitigation measures necessary. As an aid to risk management, additional testing may be incorporated into cage or field trials in order to examine whether effects on bees under normal recommended patterns of use can be reduced by changing the conditions of use (e.g. lower application rates or changing the timing of application in relation to flowering).


  • 2

    The applicant may also choose to have a compound classified as ‘high risk’ without the need to go through cage or field trials, i.e. go to 13. If it is chosen to demonstrate harmlessness in higher-tier testing, such results will supersede earlier classification based on the hazard ratio (this means that the applicant may choose to go to 9 as well).

  • 3

    The Organisation for Economic Co-operation and Development (OECD) guidelines are derived from the EPPO standard but deviate in some aspects, e.g. the use of the control, and validation limits of the toxic standard. The latest EPPO standard (OEPP/EPPO, 2010), however, incorporates the recent ICPBR recommendations on these subjects.

  • 4

    As a toxic standard is normally not included in field trials, honeybee exposure should be otherwise demonstrated, e.g. by evidence based on assessments of foraging bees before and after application.