Scientific Opinion on the science behind the development of a risk assessment of Plant Protection Products on bees (Apis mellifera, Bombus spp. and solitary bees)


Panel on Plant Protection Products and their Residues
EFSA Journal
EFSA Journal 2012;10(5):2668 [275 pp.].
Panel members at the time of adoption
Jos Boesten, Claudia Bolognesi, Theo Brock, Ettore Capri, Anthony Hardy, Andrew Hart, Karen Hirsch-Ernst, Susanne Hougaard Bennekou, Robert Luttik, Michael Klein, Kyriaki Machera, Bernadette Ossendorp, Annette Petersen, Yolanda Pico, Andreas Schäffer, Paulo Sousa, Walter Steurbaut, Anita Stromberg, Maria Tasheva, Ton van der Linden, Christiane Vleminckx

The Panel wishes to thank the members of the Working Group on Bee Risk Assessment (Robert Luttik, Gérard Arnold, Jos Boesten, James Cresswell, Andrew Hart, Jens Pistorius, Fabio Sgolastra, Noa Simon Delso, Walter Steurbaut, Helen Thompson) for the preparatory work on this scientific opinion,the hearing expert (Anne Alix) and EFSA staff (Franz Streissl, Domenica Auteri, Jean-Lou Dorne, Agnès Rortais, Klaus Swarowsky, Csaba Szentes) for the support provided to this scientific opinion.

Opinion of the Scientific Committee/Scientific Panel
On request from
European Commission
Question Number
18 April 2012
23 May 2012
European Food Safety Authority (EFSA), Parma, Italy
Download Article (3.31 MB)

The PPR Panel was asked to deliver a scientific opinion on the science behind the development of a risk assessment of plant protection products on bees (Apis mellifera, Bombus spp. and solitary bees). Specific protection goals options were suggested based on the ecosystem services approach. The different routes of exposure were analysed in detail for different categories of bees. The existing test guidelines were evaluated and suggestions for improvement and further research needs were listed. A simple prioritisation tool to assess cumulative effects of single pesticides using mortality data is suggested. Effects from repeated and simultanous exposure and synergism are discussed. Proposals for separate risk assessment schemes, one for honey bees and one for bumble bees and solitary bees, were developed.


Following a request from the European Commission, the Scientific Panel on Plant Protection Products and their Residues (PPR Panel) of EFSA was asked to deliver a scientific opinion on the science behind the development of a risk assessment of Plant Protection Products on bees (Apis mellifera, Bombus spp. and solitary bees). The opinion will be the scientific basis for the development of a Guidance Document which should provide guidance for notifiers and authorities in the context of the review of Plant Protection Products (PPPs) and their active substances under Regulation (EC) 1107/2009.

For the development of robust and efficient environmental risk assessment procedures, it is crucial to know what to protect, where to protect it and over what time period. Specific protection goals based on ecosystem services were suggested according to the methodology outlined in the Scientific Opinion of EFSA (2010). Pollination, hive products (for honey-bees only) and biodiversity (specifically addressed under genetic resources and cultural services) were identified as relevant ecosystem services. It was suggested to define the attributes to protect as survival and development of colonies and effects on larvae and honey bee behaviour as listed in regulation (EC) No 1107/2009. In addition, abundance/biomass and reproduction were also suggested because of their importance for the development and long-term survival of colonies. The magnitude of effects was defined as negligible if the natural background mortality compared to controls was not exceeded. Further work is needed to give recommendations on the deviation from the controls up to which an effect is still considered negligible. The current methods of field testing would need major improvements in order to detect for example an increase in daily mortality of foragers by 10% with high statistical power. Based on expert judgement it was considered that a small effect could be tolerated for few days without putting the survival of a hive at risk. Further research (modelling) is proposed to clarify this question and to revise the proposal for the magnitude of effects and the temporal scale of effects. The current risk assessment for honey bees relies on an Hazard Quotient (HQ) approach (application rate/LD50) in lower tiers and on semi-field and field tests in higher tiers. It is particularly difficult to ascertain whether a specific exposure percentile is achieved in field studies. Decisions need to be taken on how conservative the exposure estimate should be and what percentage of exposure situations should be covered in the risk assessment. It is recommended to design a flow chart for checking whether exposure in the semi-field or field study was indeed higher than that corresponding to the desired percentile. Factors that may be included are: the crop and its developmental stage, the dosage, measures ensuring that bees are coming into contact with the compound/formulation, weather conditions, and for instance the generation of guttation droplets by the crop. The final decision on protection goals needs to be taken by risk managers. There is a trade-off between plant protection and the protection of bees. The effects on pollinators need to be weighted against increase in crop yields due to better protection of crops against pests.

Residues in different environmental matrices and bee products were combined with estimates of exposure of different categories of bees. Highest concentrations of residues were found after spray treatments in pollen and nectar. Residues in guttation droplets showed a wide variability due to the number of parameters known to influence guttation production (environmental conditions, crop type, growth stage, etc.). A potentially high exposure was highlighted for bees in some crops (e.g. maize). Exposure to dust drift from sowing treated seeds was identified as a relevant exposure route. The exposure of different categories of bees from different sources and for different application techniques suggests that the potential risk from oral uptake was highest for forager bees, winter bees and larvae. The exposure of nurse bees occurs via a combination of pollen and nectar, of larvae by contact to wax and foragers, drones, queens and swarms intercepting droplets and vapour by contact and inhalation.

Worker bees, queens and larvae of bumble bees and adult females and larvae of solitary bees were considered to be the categories that are most exposed via oral uptake. Larvae of solitary bees consume large mass provisions with unprocessed pollen thus, compared with honey bee larvae, they are more exposed to residues in pollen. Moreover, bumble bees and solitary bees may be exposed to a larger extent via contact with nesting material (soil or plants) compared to honey bees, suggesting the need for a separate risk assessment for bumble bees and solitary bees.

For the ranking of bees, the inclusion of multiple exposures with appropriate weights would need to be done with a modelling or scenario-based approach that was not available in the current assessment. It was therefore recommended that the categories of bees which represent the worst-case exposure scenarios through multiple exposures are further assessed (e.g. honey bee nurses) and that those categories which highlighted potential but unknown exposures through consumption of water and inhalation of vapour in/out field are further analysed with more studies. Further research is recommended on the testing of the presence and fate of residues (e.g. in bee relevant matrices and in-hive following spray and dust applications) and on the development of reliable exposure models.

The overview of the available studies on sub-lethal doses and long-term effects of pesticides on bees highlighted gaps in knowledge and research needs in the following areas: more toxicological studies to be performed in bees for a wider range of pesticides on both adults and larvae including sub-lethal endpoints, also including contact and inhalation routes of exposure. Few studies were conducted with non-Apis bees, considering endpoints such as fecundity (e.g. drones production in Bombus and cell production rate in solitary bees), larvae mortality rate, adult longevity and foraging behaviour. The use of micro-colonies in bumble bees appears to be well-suited to measure lethal and sub-lethal effects of pesticides with low doses and long-term effects.

Because of the specific toxicokinetic profile of bees compared with other insects, it is recognised that toxicokinetic data can provide useful information on the potential biological persistence of a pesticide which, in some cases, could have effects after continuous exposure that maybe more marked compared with their short-term effects. The integration of toxicokinetic knowledge and low (sub-lethal) dose effects generated from laboratory and field studies in the hazard identification and hazard characterisation of pesticides in Apis and non-Apis bees can provide a better understanding of short-term and long-term effects. It is therefore concluded that the conventional regulatory tests based on acute toxicity (48 to 96 h) are likely to be unsuited to assess the risks of long-term exposures to pesticides.

A testing protocol and mathematic model, based on Haber’s law, have been developed as a simple prioritisation tool to investigate the potential effects after repeated exposure to single pesticides using mortality data. However, a number of assumptions inherent to the model raise uncertainties. The protocol and model needs further validation in the laboratory and to be tested for sub-lethal endpoints in adult and bee larvae. Finally, combining basic toxicokinetic data for an active substance and its metabolites, such as the half life, will also provide more precise estimates on the potential of bioaccumulation. In the case of potential persistence of the active ingredient, half life of the parent compound and its metabolites should be determined in larvae, newly emerged bees and foragers.

The working group identified the need for improvement of existing laboratory, semi-field and field testing and areas for further research. Several exposure routes of pesticides are not evaluated in laboratory conditions, such as the intermittent and prolonged exposures of adult bees, exposure through inhalation and the exposure of larvae. Likewise, the effects of sub-lethal doses of pesticides are not fully covered in the conventional standard tests.

Sub-lethal effects should be taken into account and observed in laboratory studies. Potential laboratory methods to investigate sub-lethal effects would be testing of Bombus microcolonies to investigate effects on reproduction, proboscis extension reflex (PER) test for neurotoxic effects and homing behaviour for effects on foraging, including orientation. Further research is needed in order to integrate the results of these studies in the risk assessment scheme.

Semi-field testing appears to be a useful option of higher tier testing. Nevertheless, weaknesses have been identified for each of the test guidelines e.g. the limited size of crop area, the impossibility to evaluate all the possible exposure routes of the systemic compounds used as seed- and soil-treatments (SSST), the limited potential to extrapolate the findings on larger colony sizes used in field studies or the relatively short timescale (one brood cycle).

The guideline for field testing (EPPO 170) (4) has several major weaknesses (e.g. the small size of the colonies, the very small distance between the hives and the treated field, the very low surface of the test field), leading to uncertainties concerning the real exposures of the honey bees. The guideline is better suited to the assessment of spray products than of seed- and soil-treatments. Points for research and improvement of methods used in field testing are highlighted (e.g. methods for detection of mortality).

The available protocols for testing of solitary bees are suitable to study the oral and contact toxicity in adults and larvae for several species of solitary bees (Megachile rotundata, Osmia spp.) but they need to be ring tested. More studies are necessary to compare the susceptibility of honey bees with other non-Apis species in order to see to which extent honey bee endpoints also cover non-Apis bees.

Future research is recommended to improve laboratory, semi-field and field tests (e.g. extrapolation of the endpoints in first tier to the colony/forager effects, extrapolation of the toxicity between dust and spray, extrapolation of laboratory based Bombus micro colonies to Apis and solitary bees).

Pesticides are often applied in tank mixes (2 to 9 active ingredients at the same time) and in addition non target organisms will be exposed to mixtures of compounds following sequential applications to crops. There is a consensus in the field of mixture toxicology that the customary chemical-by-chemical approach to risk assessment is too simplistic. At low levels of exposure concentration, addition has been observed more often than synergistic or antagonistic effects for mixtures of pesticides with a common mode of action and independent action (response addition) has been observed for compounds with a different mode of action. In some cases synergistic and antagonistic effects have also been observed.

Honey bees and hymenoptera are known to have a specific metabolic profile with the lowest number of copies of detoxification enzymes within the insect kingdom. A number of studies have shown synergistic effects of pesticides and active substances applied in hives as medical treatments against Varroa mites in honey bees, for which toxicokinetic interactions were most commonly involved. There is also a growing body of evidence of interaction between honey bee disease (fungi, bacteria and viruses) and pesticides. Currently, full dose responses for synergistic effects between potential inhibitors and different classes of pesticides are rarely available for either lethal effects or sub-lethal effects in bees so that predictions of the magnitude of these interactions at realistic exposure levels cannot be performed. However, there is evidence that where realistic exposure levels have been investigated, deviations from concentration addition, such as synergy, is rarely more than a factor of 2 to 3. Such deviations have been observed for mixtures containing small numbers of chemicals and decreases as the complexity of the mixture increases.

In the case of synergism which can be predicted based on the mode of action of the chemical classes involved (e.g. EBI fungicides and insecticides), and in the absence of existing data on toxicity of the mixture, it is recommended to design full dose-response studies in adult bees and larvae for mixtures of potential synergists. Further work is also required to identify the molecular basis of interactions between environmentally realistic exposure to pesticides and the range of honey bee diseases (fungi, bacteria and viruses) to determine whether and how these may be included in risk assessment.

Separate risk assessment schemes are proposed, one for honey bees and one for bumble bees and solitary bees. In the first tier it is suggested to include toxicity testing that covers a longer period of exposure (7 to 10 days) for adult bees as well as larval bees. Both life stages can be exposed for more than one day and this risk was not covered by the standard OECD tests (213 and 214) for oral and contact exposure. Currently there is insufficient evidence that toxicity following extended exposures can be reliably predicted from acute oral LD50 data. It is also proposed to investigate whether there are any indications of cumulative effects for each compound. A new method to detect cumulative toxicity is proposed based on Haber`s law. If there is an indication that a compound is a cumulative toxin then this needs further evaluation since the potential effects of prolonged or repeated exposure to low doses may be underestimated.

Guidance Document, PPR opinion, honey bees, bumble bees, solitary bees, pesticide, risk assessment
Print on demand
Number of Pages