Effects of plant protection products on bees
The effects of plant protection products on bees, as well as on other organisms, are evaluated using a methodology laid down at European level. The Guidance document on terrestrial ecotoxicology (SANCO/10329/2002 rev. 2) from 2002 is currently applicable. As the science is constantly evolving, this evaluation methodology is also regularly adapted and updated. In 2013, at the request of the European Commission, the European Food Safety Agency (EFSA) developed a new methodology. This "EFSA Bee guidance document" was used by the EFSA for the evaluation of neonicotinoids, among other things. However, to date, this methodology has not been widely applied to other active substances or for the evaluation of plant protection products by Member States. We will examine below why this is the case.
The guidance document drawn up in 2013 is currently being revised by the EFSA. However, certain steps that are necessary for this revision have raised a number of questions. These are also explained below (updated in August 2021).
- There has been an EFSA bee guidance document since 2013. Why has this never been used?
- What effects on bees are considered acceptable in the context of a risk assessment?
- The EFSA guidance document from 2013 defines a 'negligible effect' as an effect of 7%. Why is this value not simply retained?
- What should be understood by "negligible effects"?
- So within that option 2, what effect of pesticides on colonies can be considered acceptable?
- Using the full operating range, what percentage decrease in colony strength can be considered acceptable?
- Does accepting a value higher than 7% not then mean that the level of protection for bees will be lower than the protection that implementing the previous GD would have provided?
- Did the EFSA have enough scientific knowledge to initiate such a project?
- Is it true that the EFSA used the BEEHAVE model, which was developed by industry, for this scientific evidence?
- Besides protection for honey bees, will the revised guidance document also provide protection for other bee species such as bumblebees and solitary bees, or even other pollinators such as hoverflies and butterflies?
- Where can I find more information about the progress of this project?
1.There has been an EFSA bee guidance document since 2013. Why has this never been used?
The EFSA bee guidance document was applied for the interim evaluation of neonicotinoids on an explicit mandate from the European Commission. However, a guidance document is normally only adopted and implemented if there is sufficient support from Member States. Moreover, this guidance document from 2013 has been the subject of a lot of discussion between Member States and the European Commission. Some Member States (including Belgium) supported its full implementation. Other Member States were reluctant to accept a guidance document which they considered unrealistic, arbitrary and therefore too stringent. It would result in a ban on plant protection products for which no negative effects on bees have been observed in practice, which is hard to justify.
After years of lengthy discussions on this subject, the European Commission proposed working in 2 phases: first, only the part relating to the acute risk for honey bees would be implemented. The part relating to the risk to honey bees after chronic exposure would be reviewed by the EFSA and adopted at a later stage, together with the part relating to the risk to bumblebees and solitary bees. This approach was approved by Member States but then rejected by the European Parliament. As a result, there is still no progress and therefore no new methodology.
As this document from 2013 has been the subject of discussion for such a long time and it was clear that no majority could be found to put it into use in its current form, the European Commission requested the EFSA to comprehensively revise this Guidance Document. This exercise is currently in progress.
Precisely because this situation has dragged on for so long, Belgium has taken the initiative to implement the EFSA Bee guidance document to the extent possible. To this end, an approach based on this guidance document from 2013 has been developed at the national level.
2.What effects on bees are considered acceptable in the context of a risk assessment?
The general consensus is that the use of plant protection products or other products should not have any negative impact on bees or other beneficial organisms, or even on the environment in general. However, from a scientific perspective, it is not possible to rule out all effects. There will always be some effect with any product used or practice applied. There is only so-called "zero risk" if there is no use at all. If the legislator accepts that it must be possible to use a product, then it must also accept that there will be effects.
For this reason, European legislation provides that the use of plant protection products can be allowed, but on condition that the effects are "acceptable". What is considered acceptable is enshrined in legislation through the available democratic processes. Before an authorisation is granted, a risk assessment will always be carried out, in which it is examined whether the use of a plant protection product is expected to result in an effect that is larger than 'acceptable'. In order to produce an effective risk assessment, it is therefore necessary to determine in advance what effects we still consider to be acceptable and what effects are no longer acceptable. In this regard, not only the extent of the effect is important, but also the level at which this effect occurs (e.g. on the individual organism or on the entire population).
For most organisms, including bees, a limited effect is currently accepted, as laid down in European legislation. When the new guideline for bees was developed in 2013, the latest scientific insights were taken into account to keep acceptable effects at the lowest possible level. Using the correct scientific terminology, we then refer to "negligible" effects.
For honey bees, it was also agreed to focus on effects at the level of the colony. This choice was made because a colony of honey bees is the 'entity' that provides various ecosystem services such as pollination. Furthermore, there is not much point in taking individual honey bees into account, since many bees die every day in each hive for all sorts of natural reasons, while new bees are born. Moreover, the impact of an individual bee that dies on the ecosystem services provided by the colony (e.g. pollination) will be very limited. Only if the colony as a whole experiences an effect, e.g. due to a large number of bees dying, or due to a serious disruption to the development of new bees, will there be an impact on the ecosystem services. From a scientific perspective, it is therefore more beneficial to evaluate the final result for the colony as a whole.
Even the current revision of the EFSA Bee guidance document still assumes that only a "negligible effect" may occur on bees when plant protection products are used. However, the question of what should be understood by 'negligible effect' is still the subject of ongoing discussions (see question 4).
3.The EFSA guidance document from 2013 defines a 'negligible effect' as an effect of 7%. Why is this value not simply retained?
To fully answer this question, it is important to understand what this value of 7% refers to. As already mentioned under question 2, it was agreed to look at effects on the colony. As a measure of how well a colony of honey bees is doing, the 2013 EFSA guidance document looked at colony strength (the number of adult bees in a colony).
To examine any possible effect when a plant protection product is applied, you can carry out a study in which you place two colonies in a similar environment, and in which you expose one colony to the plant protection product but not the other one (the control). If exposure to the plant protection product causes the number of bees in the first colony to be more than 7% lower than the number of bees in the control colony, this would mean that the plant protection product has an effect larger than negligible, and therefore cannot be authorised for use.
In 2013, however, the necessary scientific data were not available to start defining in a clear and substantiated manner where the limit is of an effect that is still just negligible. Using expert judgement, including consultation with beekeepers, it was estimated that this limit was probably 7%.
As this value of 7% is chosen rather arbitrarily, it has been the subject of extensive discussion in recent years. Consequently, in the context of the revision of the guidance document, different possible methods have been sought that could ensure that the chosen limit value for a negligible effect is more scientifically substantiated (see also question 4).
4.What should be understood by "negligible effects"?
Following the decision that only negligible effects are acceptable, in 2020 the EFSA proposed four different approaches for defining what constitutes a negligible effect:
- Survival of bee colony after pesticide application.
- An effect on the size of a bee colony no greater than the normal variability in an "average" colony without pesticide application. For this approach, it is necessary to determine how large the normal variability is. To do this, a mathematical model is used.
- An effect on the size of a bee colony no greater than the normal variability in an "average" colony without pesticide application. As is the case for approach 2, it is necessary here to determine how large the normal variability is. However, in this approach no specific data is made, but expert judgement is used.
- The effect that is permissible on the size of a bee colony is determined by the magnitude of the effect on the ecosystem services provided by the bee colonies that is considered acceptable. Ecosystem services are the services that bees provide to the ecosystem, i.e. nature as a whole and in particular the services that are considered useful to humans, for example crop pollination, genetic diversity, aesthetic value (e.g. of wild plants that exist because of pollination by bees), etc.
After an extensive discussion between EFSA, the European Commission and Member States, option 2 was selected. Belgium also supported this option. Our reasoning for this is as follows:
Option 1 was considered to be insufficiently stringent. Indeed, just because a colony survives does not mean it had "normal" activity during the year. There could therefore be non-negligible effects (e.g. reduced pollination) without the entire colony dying out.
The idea behind options 2 and 3 is comparable. Actually, these only differ in the substantiation of the limit value of what can be considered normal variability. With option 3, a value is selected that is perceived to be acceptable. The 7% limit was also set in this way in the 2013 EFSA methodology, but precisely because this is scientifically debatable, no agreement could ever be found. Using the same approach for this new guidance document would therefore probably lead to the same impasse. In option 2, on the other hand, a fixed value is selected that is more scientifically substantiated.
In theory, option 4 is the ideal option, but the necessary knowledge to make the link between effects on a colony and effects on ecosystem services provided does not currently exist. It would take many years of research before a scientifically sound evaluation methodology could be developed on that basis. If we therefore wish to use a guidance document in the short term, this approach is out of the question. It should be expected that once the research has progressed far enough, this approach will be retained in the future.
As such, option 2 proved to be the most scientifically accurate approach achievable within a reasonable period of time. Moreover, if no effects greater than normal variability are accepted, it is also to be expected that there will be no effects on the ecosystem services provided.
5.So within that option 2, what effect of pesticides on colonies can be considered acceptable?
The EFSA has determined on a scientific basis the normal variability in the magnitude of bee colonies. The results of this work have been submitted to the European Commission and Member States, and have already been discussed for the first time at a workshop. Consequently, for the time being, no decision has been taken on this matter. This document can be consulted via this link.
This document describes the outcome of a modelling exercise that examined the normal variability of bee colonies in 19 different scenarios (for 19 sites across the EU). This shows that bee colonies that are not exposed to pesticides will show a certain variation in bee numbers in any case. The principle of option 2 is that an effect of pesticides smaller than this normal variability within and between unexposed colonies can be considered negligible.
To express the normal variation quantitatively, we ran the model 500 times for each of the 19 scenarios. For each day of the year, the average colony strength of those 500 runs was then calculated. The difference between this average value and the size of the smallest colony was then calculated. This difference is a measure of the normal variation.
To determine the normal variation, you can take into account all modelled colonies, and thus effectively determine the difference between the average and the very smallest colony. In the EFSA document this is described as the 'full operating range'. However, it is also possible to exclude a part of the smallest colonies (e.g. the 5 or 10% smallest). In the EFSA document this is described as the 'restricted operating range'. By not including the most extreme colonies in the calculation, the lower limit shifts upwards, and the variation from the average becomes smaller. In this way, some additional 'caution' can be built into the calculated variation, in order to compensate for any uncertainties that are inherent in using a model.
However, the EFSA also compared the results of the modelling with data from field studies. In addition, the impact of certain assumptions in the model on the outcome was also analysed. This showed that the calculated variability is more of an underestimate of the variability that would be found in reality. Belgium has therefore taken the position that there is no need for a further restriction of the operating range in order to calculate normal variability and to determine the specific protection objective for bees.
6.Using the full operating range, what percentage decrease in colony strength can be considered acceptable?
In the EFSA document, calculations were made for 19 scenarios. Since a value for the normal variation was calculated for each scenario, the result is a certain range of possible percentages. However, the Commission has decided to summarise this in a single value, i.e. the median of the variation over a full year, both for the full operating range and for the restricted operating ranges. If the full operating range is considered (all modelled colonies, including the smallest), this corresponds to 23.2%. If the operating range is restricted by 5% (not including the 5% smallest colonies) this is 12.8%.
However, Belgium believes that taking a decision on whether to use the full operating range or a more restricted operating range should have been the first step in the process. The choice of the specific percentage from the resulting range of possible values (e.g. the lowest value, the average, etc.) should have been the subject of a subsequent discussion. In addition, it also has an impact on whether the situation is looked at per season or over a full year. The possible percentages of normal variability corresponding to the full operating range are:
- Full year: 20.0-31.1% (median: 23.2%)
- Spring: 17.4-28.2% (median: 20.9%)
- Summer: 10.4-47.1% (median 18.6%)
- Autumn: 19.1-44.5% (median: 28.6%)
At a meeting of the European Ministers of Agriculture on June 28 and 29, it was decided at that level to use 10% as an acceptable decrease in colony size.
7.Does accepting a value higher than 7% not then mean that the level of protection for bees will be lower than the protection that implementing the previous GD would have provided?
As already explained in question 3, the figure of 7% was based on expert judgement, and not, or only to a very limited extent, on scientific data as this was simply not available. Consequently, 7% was a very conservative approximation, as there were many uncertainties. Today we have access to much more scientific data on the variability of colony strengths. It can be seen from the work of the EFSA that variations in colony strength larger than 7% can occur within a normal colony, which was not exposed to pesticides. As such, these larger variations in colony strength do not pose a problem for the normal functioning of bee colonies. Because there are now far fewer uncertainties than 8 years ago, we can more accurately state what percentage of decline in colony strength is not a problem, and so there is less need to include additional caution into the protection goal.
As already highlighted in question 5, there are still some uncertainties in the current modelling. Such uncertainties are inherent in the use of models, which are always a simplification of reality. However, the EFSA has analysed the impact of certain assumptions in the model (and associated uncertainties) on the outcome. This showed that the calculated variability is more of an underestimate of the variability that would be found in reality. This was also confirmed by comparing the outcome of the modelling with data from field studies. In the revised version of the guidance document, caution is therefore also incorporated to cover the remaining uncertainties. This is already included in the percentages shown by this full operating range. It is therefore not the case that we accept a lower level of protection for bees compared to the 2013 guidance document. It is the case that with advances in scientific knowledge on bees, and therefore the fact there are fewer uncertainties, we are able to determine more precise percentages that do not interfere with the normal functioning of a bee colony. Choosing (one of) the values that emerge from this EFSA document, which are slightly higher than 7%, will result in a similar level of protection as that proposed in 2013. Note that this level of protection is many times higher than what is achieved with the 2001 guidance document, which is still applicable today.
It is also important to stress that the risk assessment according to the EFSA bee guidance document (both the 2013 version and the revised version) focuses on the protection of bee colonies located on the edge of a treated field. This is therefore a highly cautious approach, since most bee colonies are further away from treated fields, and will therefore have less exposure. Consequently, these more remote colonies will also experience fewer effects. This is another way of incorporating additional caution into the risk assessment.
8.Did the EFSA have enough scientific knowledge to initiate such a project?
The EFSA set up a working group consisting of various national experts and academics from relevant fields of research, selected on the basis of their experience and expertise. More information on this working group and its composition can be found on the website of the EFSA.
9.Is it true that the EFSA used the BEEHAVE model, which was developed by industry, for this scientific evidence?
The EFSA did indeed use the BEEHAVE model. However, it would be incorrect to say that the BEEHAVE model was developed by industry. This model was developed at the University of Exeter (Becher et al., 2014). It is true that 15% of research costs were sponsored by industry.
The BEEHAVE model was evaluated in detail by the EFSA PPR Panel, which was made up of independent experts, in 2015 (see the published EFSA statement on this subject). Several criticisms were made of the model at the time, but not all of them are relevant to the context in which it will now be used by the EFSA. In the document that the EFSA has made available to the European Commission and Member States (this document can be consulted via this link), and in the technical report that will be subsequently published by the EFSA, all these criticisms, the way they have been addressed, and their relevance for the current exercise, have been discussed in detail. In this way, this can be taken into account in further discussions.
The BEEHAVE model is currently considered the best available model for modelling the dynamics in a bee colony. It should be noted that the EFSA is currently developing a new mechanistic model (ApisRAM project). Among other things, this model aims to resolve some of the limitations of BEEHAVE, and would therefore also have some advantages for estimating the normal variability of a bee colony. However, this model is still under development. Consequently, it was not possible to propose this for the current exercise.
Becher, M.A., Grimm, V., Thorbek, P., Horn, J., Kennedy, P.J. & Osborne, J.L. (2014). BEEHAVE: a systems model of honeybee colony dynamics and foraging to explore multifactorial causes of colony failure. J Appl Ecol, 51, 470-482.
10.Besides protection for honey bees, will the revised guidance document also provide protection for other bee species such as bumblebees and solitary bees, or even other pollinators such as hoverflies and butterflies?
The European Commission is aware of the need to develop an evaluation methodology for bumblebees and solitary bees. The 2013 EFSA guidance document already included a risk assessment method for bumblebees and solitary bees. The mandate of the European Commission to the EFSA for the revision of the guidance document also explicitly requests a revision of the sections related to bumblebees and solitary bees.
Also for bumblebees and solitary bees, there is a consensus that there should only be a "negligible effect" after application of a plant protection product. However, because there is less data for these species, the possible approaches for defining what constitutes a negligible effect are more limited. It is therefore not yet certain which approach will be adopted. However, it is certain that the revised guidance document will still provide protection for bumblebees and solitary bees.
Other pollinators, such as hoverflies and butterflies, do not fall within the scope of this guidance document. These are classified under the group of 'non-target arthropods', along with other insects, and are evaluated according to a different evaluation methodology. The development of a new guidance document for the risk assessment of this group is envisaged in another project, for which no time scales are currently known.
11.Where can I find more information about the progress of this project?
The European Commission has a website on which it provides updates on this guidance document. It is regularly updated to reflect the latest state of play in this evolving dossier.