1. The EU should adopt a coordinated, strategic approach to implementing probabilistic methods, in collaboration with other countries and other stakeholders.


  2. Probabilistic methods should be introduced gradually to assist decision-making on the ecological risks of pesticides, beginning with those cases where they are most needed. As experience accumulates, the role for probabilistic methods may expand.


  3. Probabilistic methods are not the only tool for risk assessment and should be used together with other lines of evidence. Deterministic methods are likely to remain the primary tool for lower tiers of risk assessment. Probabilistic methods constitute one of several approaches that may be used for refined ("higher tier") assessments.


  4. As probabilistic methods for specific parts of risk assessment gain acceptance they should be introduced to regulatory practice and added to existing EU guidance documents.


  5. Measures should be taken to facilitate the adoption of probabilistic methods in all parts of the expanding EU, including provision of training and expert advice, and adoption of standard computer software and databases.


  6. Appropriate procedures should be established for the peer review of probabilistic assessments.


  7. A framework should be established for the appropriate use of probabilistic methods in the regulatory process. The framework should include:
    • guidance on when to use probabilistic methods in the regulatory process, what the outputs should be, and how to use them in decision-making,
    • guidance on the selection and use of methods of uncertainty analysis, including key principles of best practice
    • guidance on ways of reporting probabilistic assessments that facilitate peer review,
    • guidance on communicating the results of risk assessments to decision-makers and the public,
    • a set of case studies to illustrate the use of the framework.
  8. Research should be conducted to develop and evaluate probabilistic approaches that can be used when data are limited.
  9. The major components for probabilistic assessment should be validated. However, the workshop recognised that comprehensive validation is probably unachievable.
  10. Specific research is needed to develop probabilistic approaches for many sources of uncertainty affecting ecological risks of pesticides. Those identified by the workshop are listed below (for details see the full report). Many of these actions are required to improve deterministic assessments, as well as probabilistic assessments.

    Aquatic organisms

    1. Extrapolation from individuals in single species tests to populations
    2. Extrapolation from individuals of single species to communities
    3. Input parameters for modelling pesticide fate
    4. Discrepancy between exposure in laboratory studies and in the field
    5. Uncertainties in the exposure scenario and variability in the landscape
    6. Variation in sensitivity between species
    7. Extrapolation of sensitivity from laboratory studies to the field
    8. Representativeness of species used in risk assessment
    9. What level of effect is acceptable

      10. Terrestrial vertebrates

    10. Intra- and inter-species variation in sensitivity
    11. Behaviour and natural history
    12. Spatial distribution of residues
    13. Residues dynamics (dissipation, bioaccumulation, etc.)
    14. Avoidance/attraction of contaminated food
    15. Effects on populations and communities
    16. Non-dietary routes of exposure (e.g. dermal exposure and inhalation)

      17. Terrestrial invertebrates and plants

    17. Factors affecting exposure
    18. Interspecies variation in sensitivity
    19. Extrapolation from effects on individuals (laboratory) to populations (field)
    20. Extrapolating acute to chronic effects
    21. Regional variation in sensitivity/concern of non-target organisms
    22. Presence of sensitive/robust life stages

  11. Where similar issues need to be addressed for different groups of organisms, it may be efficient to coordinate the necessary research in a single project or a set of linked projects. Examples of issues where this type of coordinated approach may be useful include:
    1. Methods of uncertainty analysis,
    2. Methods for dealing with variation in toxicity between species,
    3. Methods for dealing with spatial and temporal variation,
    4. Methods for defining and using regional scenarios,
    5. Methods for predicting impacts at population and community level.
  12. Priority should be given to research that is important for current deterministic assessments as well as probabilistic assessments. Examples include:
    1. Refining methods for dealing with variation in toxicity between species (species sensitivity distributions),
    2. Quantifying the spatial distribution of different types of water bodies, and their proximity to crops,
    3. Collecting improved information on aspects of ecology and behaviour affecting the exposure of terrestrial vertebrates,
    4. Developing methods for assessing the exposure of terrestrial invertebrates.
  13. There is a large amount of existing data that could be valuable in developing and using probabilistic approaches. Early action is needed to make these data accessible. Consideration should be given to whether this could be undertaken by the EU Environment Agency.
  14. When official guidelines for toxicity test guidelines are reviewed, consideration should be given to whether revisions are required to make them better suited to probabilistic assessments.
This project is supported by the European Commission under the 5th Framework Programme (http://www.cordis.lu). The contents of these pages are the sole responsibility of its publishers. These pages in no way represent the views of the Commission or its services.