Normes OEPP EPPO Standards

Transcription

1 2003 OEPP/EPPO, Bulletin OEPP/EPPO Bulletin 33, Blackwell Oxford, Bulletin OEPP/EPPO, 33 Original EPPO Environmental Standards UK OEPP/EPPO Article Publishing 2003 risk assessment Ltd. Bulletin scheme for plant protection products Organisation Européenne et Méditerranéenne pour la Protection des Plantes European and Mediterranean Plant Protection Organization Normes OEPP EPPO Standards Environmental risk assessment scheme for plant protection products Système pour l évaluation du risque des produits phytosanitaires pour l environnement PP 3/9 (revised) Organisation Européenne et Méditerranéenne pour la Protection des Plantes 1, rue Le Nôtre, Paris, France 99

2 100 Environmental risk assessment scheme for plant protection products Approval EPPO Standards are approved by EPPO Council. The date of approval appears in each individual standard. Review EPPO Standards are subject to periodic review and amendment. The next review date for this set of EPPO Standards is decided by the EPPO Working Party on Plant Protection Products. Amendment record Amendments will be issued as necessary, numbered and dated. The dates of amendment appear in each individual standard (as appropriate). Distribution The EPPO Standards making up the EPPO/Council of Europe decision-making scheme for the environmental risk assessment of plant protection products are distributed by the EPPO Secretariat to all EPPO member governments. Copies are available to any interested person under particular conditions upon request to the EPPO Secretariat. Scope The EPPO Standards making up the EPPO/Council of Europe decision-making scheme for the environmental risk assessment of plant protection products are intended to be used by National Plant Protection Organizations in their capacity as bodies responsible for the registration of plant protection products, including an evaluation of the environmental risks arising from their use. They are also intended for use by the agrochemical companies who apply for registration of their products. Outline of requirements The decision-making scheme for the environmental risk assessment of plant protection products was developed by a joint Panel of EPPO and the Council of Europe and provides guidelines on how to assess the potential impact of a particular plant protection product on various elements of the environment, each element being presented as a separate standard. The assessment scheme is for use by agrochemical companies and by regulatory authorities, and aims to: (1) guide assessors on the questions that should be addressed, and the data that may need to be requested from registrants; (2) provide information on the test methods and approaches that are suitable in each case; (3) indicate how the data should be interpreted in a consistent manner, involving expert judgement where appropriate; (4) produce a reliable assessment of environmental risk, as a suitable aid to risk management (though the assessment scheme will not provide all the information needed for decisions on the acceptability of plant protection products). The Standards in the scheme provide a set of flexible procedures that can be adapted for use in various ways according to the priorities in different countries, yet retain the consistency of a common framework. They are not based on a series of fixed, automatic triggers for testing requirements, but are able to take full account of the particular features of each plant protection product, and to make use of expert judgement when necessary. Approbation Les Normes OEPP sont approuvées par le Conseil de l OEPP. La date d approbation figure dans chaque norme individuelle. Révision Les Normes OEPP sont sujettes à des révisions et des amendements périodiques. La prochaine date de révision de cette série de Normes OEPP est décidée par le Groupe de travail sur les produits phytosanitaires. Enregistrement des amendements Des amendements sont préparés si nécessaires, numérotés et datés. Les dates de révision figurent (si nécessaire) dans chaque norme individuelle. Distribution Les Normes OEPP composant le système de décision OEPP/Conseil de l Europe pour l évaluation du risque des produits phytosanitaires pour l environnement sont distribuées par le Secrétariat de l OEPP à tous les Etats membres de l OEPP. Des copies sont disponibles, sous certaines conditions, auprès du Secrétariat de l OEPP pour toute personne intéressée. Champ d application Les Normes de l OEPP composant le système de décision OEPP/Conseil de l Europe pour l évaluation du risque des produits phytosanitaires pour l environnement sont destinées aux Organisations Nationales de Protection des Végétaux, en leur qualité d autorités responsables de l homologation des produits phytosanitaires, et ainsi de l évaluation des risques pour l environnement liés à l utilisation de ces produits. Elles sont également destinées aux firmes agrochimiques demandant l homologation de leurs produits. Vue d ensemble Le système de décision pour l évaluation du risque des produits phytosanitaires pour l environnement a été mis au point par un Groupe d experts conjoint OEPP et Conseil de l Europe. Il donne des recommandations sur la manière d évaluer l impact potentiel d un produit phytosanitaire donné sur diverses composantes de l environnement, chacune de ces composantes étant traitée dans une norme distincte. Le système d évaluation a pour objectif de: (1) guider les responsables des évaluations sur les questions à aborder et sur les données que les demandeurs de l homologation peuvent avoir à fournir; (2) donner des informations sur les méthodes d essai et les approches appropriées dans chaque cas; (3) montrer comment interpréter les données d une manière cohérente, en ayant recours au jugement d experts lorsque cela s impose; (4) permettre une évaluation fiable du risque pour l environnement, facilitant la gestion du risque (même si le système d évaluation ne fournira cependant pas toutes les informations nécessaires pour décider de l acceptabilité des produits phytosanitaires). Les Normes qui composent le système présentent une série de procédures souples qui peuvent être adaptées et utilisées de différentes manières dans différents pays, tout en conservant la cohérence d un cadre commun. Elles ne reposent pas sur une série de seuils fixes définissant automatiquement la nécessité de conduire des essais, mais permettent de tenir compte de l ensemble des caractéristiques particulières de chaque produit phytosanitaire et de recourir au jugement d experts lorsque cela s impose OEPP/EPPO, Bulletin OEPP/EPPO Bulletin 33,

3 EPPO Standards 101 Existing EPPO Standards in this series The first 11 Standards in the scheme were originally published in Bulletin OEPP/EPPO Bulletin in 1993/1994: OEPP/EPPO (1993) Decision-making scheme for the environmental risk assessment of plant protection products. Chapters 1 6, 8 & 10. Bulletin OEPP/EPPO Bulletin 23, OEPP/EPPO (1994) Decision-making scheme for the environmental risk assessment of plant protection products. Chapters 7, 9 & 11. Bulletin OEPP/EPPO Bulletin 24, Over the period from 1994 to the present day, these texts were first reformatted as EPPO Standards, to constitute Series PP 3. All the Standards were then updated to take account of new developments, and two additional Standards were approved. Chapters 7 and 8 were combined. The present set is the first of several, published in Bulletin OEPP/EPPO Bulletin, which will together include all the new and revised Standards of this series. Although the original set of Standards was published bilingually in English and French, the present set is now published only in English. The new French versions of these Standards will be published at a later date. The corresponding old English and French versions of the Standards in this set are withdrawn. Normes OEPP déjà existantes dans cette série Les 11 premières Normes de ce système ont initialement été publiées dans le Bulletin OEPP/EPPO Bulletin en 1993/1994: OEPP/EPPO (1993) Système pour l évaluation des effets non intentionnels des produits phytosanitaires sur l environnement. Chapitres 1 6, 8 & 10. Bulletin OEPP/EPPO Bulletin 23, OEPP/EPPO (1994) Système pour l évaluation des effets non intentionnels des produits phytosanitaires sur l environnement. Chapitres 7, 9 & 11. Bulletin OEPP/EPPO Bulletin 24, Entre 1994 et 2002, ces textes ont d abord été reformatés sous forme de Normes OEPP, pour constituer la série PP 3. Ces Normes ont ensuite été mises à jour pour tenir compte des évolutions dans ce domaine, et deux Normes supplémentaires ont été approuvées. Les chapitres 7 et 8 ont été combinés. L ensemble des normes nouvelles et révisées de cette série sera publié dans le Bulletin OEPP/EPPO Bulletin, qui en présente ici les premières. Bien que la série originale de Normes ait été publiée sous forme bilingue, en anglais et en français, la présente série n est publiée qu en anglais. Les nouvelles versions françaises de ces Normes seront publiées ultérieurement. Les anciennes versions, anglaises et françaises, des Normes présentées ci-dessous sont retirées OEPP/EPPO, Bulletin OEPP/EPPO Bulletin 33,

4 Blackwell Publishing Ltd. European and Mediterranean Plant Protection Organization PP 3/9(2) Organisation Européenne et Méditerranéenne pour la Protection des Plantes Environmental risk assessment scheme for plant protection products Chapter 9: Non-target terrestrial arthropods Specific scope This standard provides a scheme for assessment of the potential environmental risks presented by plant protection products for non-target terrestrial arthropods. Specific approval and amendment First approved in Edited as an EPPO Standard in Revised in Introduction The subscheme in this chapter is concerned with assessing the potential risks of spray formulations of plant protection products to non-target terrestrial arthropods that are living mostly above the ground, in particular to species dwelling on plants. It is not concerned with effects of formulations for treating the soil, e.g. granules and seed treatments; these are considered in Chapter 8 (Soil Organisms and Functions). Nontarget arthropods play a vital role in the ecosystem and therefore plant protection products that are used as prescribed should not cause unacceptable effects on populations of nontarget arthropods both in- and off-field. The objective of the subscheme is therefore to ensure that any risk to non-target arthropod populations, both within and off the cropping field, are highlighted and to assess the nature of the risk so that adequate risk reduction measures can be taken. It is not the objective of this subscheme to contribute to protection of non-target arthropods for agricultural reasons alone, as in integrated pest management (IPM). Many species of non-target arthropods are natural enemies of pests. Their presence and healthy activity keeps primary and secondary pests in check. Protection of natural enemy populations, e.g. in IPM systems, has received much attention in the past mainly for these agricultural reasons, and many studies for test development and effect assessment are available. Such studies have provided an invaluable base for development of the present risk assessment scheme. Protection of non-target arthropods for environmental reasons (as in the present scheme) is a separate and broader issue than their protection for agricultural reasons, while the latter requires additional and more specific data than is generated by the present scheme. However, the possibility is not excluded that data generated under this scheme could be used for risk assessment in IPM as well. About the subscheme Applications of plant protection products are classified by this subscheme for risk to non-target arthropods on the basis of toxicity data and of estimates of the environmental exposure in off-field and within-field areas. The effects of applications of a plant protection product on non-target arthropods are assessed at each stage of a tiered testing scheme. After each assessment, applications of a product can be categorized as of low or high risk. Risks of multiple applications are evaluated by increasing the single-use application rates according to specific prescriptions or repeating applications in higher-tier tests. No further testing is needed for the negligible category (where exposure is unlikely) or for the low-risk category. Products of high risk to arthropods may be tested in higher tiers to show harmlessness in more realistic conditions or, alternatively, appropriate risk reduction measures may be used to ensure acceptability. The final assessment is made on a crop-by-crop basis but, as many non-target arthropods are common to more than one crop and exposure is similar for various crop groupings, this subscheme takes into account all intended uses during the testing stages. The classification should provide support for decisions to be taken on product registration, and on use restrictions which appropriately reduce risk. Risk management and risk reduction are the responsibility of national policy makers and are not as such covered by this scheme. The endpoint of the scheme is risk classification but, as the scheme needs to support risk management decisions, the risk assessment methodology has been developed with recommended management options for guidance. A simplified diagram of the subscheme is provided in Fig. 1. Risk assessment and extrapolation The group of non-target arthropods in general comprises species from a wide taxonomic and ecological range, whose composition will vary temporally and spatially between different crops. It is, of course, not possible to have all arthropod species tested, and results will therefore have to be extrapolated from the few species tested to broader groups. As indicator species Aphidius spp. (Hymenoptera: Braconidae) and Typhlodromus pyri (Acarina: Phytoseiidae) are used, since these have been shown to rank among the most sensitive species in a data set of nine arthropod families and 95 plant protection products by a SETAC/ESCORT2 workshop (Candolfi et al., 1999, 2001). 131

5 132 Arthropods The subscheme has a sequential nature and starts with an assessment of exposure. If there is no exposure, then risk is negligible. If exposure is possible, the subscheme proceeds from simple laboratory tests under controlled conditions to complex field trials. Laboratory tests are suitable for assessing the toxicity of a product, and for demonstrating harmlessness if toxicity is very low in comparison with the application rate. They give definitive, reproducible results on single indicator species. However, harmful results reported from laboratory tests conducted under worst-case exposure conditions should not be directly extrapolated to the field where exposure conditions are unlikely to be so severe. Field trials incorporate realistic exposure, and their results may cover a wider range of species. However, it is usually more difficult to interpret them and to determine the mechanisms of effects. Semi-field tests are intermediate between these two levels in terms of both realism and ease of interpreting the results. Laboratory and semi-field tests involve the study of one or more individual arthropod species, whereas field trials mostly involve a range of arthropod groups, potentially of many different species. History of this scheme This risk assessment scheme is a revised version of the original EPPO Standard PP3/9(1) and is intended to be an important simplification and improvement. The need for updating has been outlined by Oomen et al. (1998). The scheme also takes into account new developments in this area, in particular Annexes II, III and VI of EU Council Directive 91/414/EEC (EU, 1991), the first and in particular the second SETAC/ESCORT Guidance Document (Barrett et al., 1994; Candolfi et al., 2001), the British position paper on risk management (Campbell, 1998), the German position paper on risk management (Forster & Rothert, 1998), recommendations of the BART Group (Beneficial Arthropod Regulatory Testing Group), the recommendations of the EPPO/Council of Europe Workshop on Environmental Risk Assessment in Bilthoven (NL /13), and the testing guidelines developed by the Joint Initiative of IOBC, BART and EPPO (Candolfi et al., 2000a). Particularly important contributions came from the SETAC/ESCORT2 workshop, held in in Wageningen (NL), and the many studies supporting the recommendations of this workshop (Candolfi et al., 2001). Risk assessment scheme Possibility of exposure 1. Assess the possibility for non-target terrestrial arthropods to be exposed (see Note 1). Is exposure of non-target terrestrial arthropods possible? if yes go to 2 if no go to 9 Laboratory toxicity test on two standard species 2. Conduct dose response tests on inert substrates using two standard species (Aphidius rhopalosiphi and Typhlodromus pyri) and assess the LR 50 (see Note 2) go to 3 Assessment of the predicted environmental concentration (PEC) 3. Take the highest recommended application rate. Assess the rate in-field (PEC in-field) and calculate the drift rate off-field (PEC off-field) (see Note 3) go to 4 Calculate hazard quotients for the two standard species 4. The hazard quotient is the PEC/LR 50 (see Note 4). Both the PEC and the LR 50 should be expressed in g ml 1 formulated product per ha or in g ml 1 active substance per ha. Calculate hazard quotients for both standard species, separately in-field and off-field. Continue in-field and off-field risk assessments separately go to 5 Risks in-field and off-field 5. Consider the in-field hazard quotients of both standard species (see Note 5). Are both in-field hazard quotients < q? if yes for in-field risks go to 10 if no (i.e. if at least one HQ q) for in-field risks go to 6 Consider the off-field hazard quotients of both standard species (see Note 5). Are both off-field hazard quotients < q? if yes for off-field risks go to 10 if no (i.e. if at least one HQ q) for off-field risks go to 6 Higher-tier tests with the formulated product 6. Accept (separately for in-field and off-field) a preliminary classification of high risk as definitive go to 11 Or, conduct higher tier tests with the formulated product (see note 6) go to 7 Assess level of effects in second-tier studies (see note 7) 7. Consider effects in-field and off-field separately. In second-tier tests with extended laboratory, aged residue and/or semi-field studies, are effects (on survival and reproduction; see Note 7) less than 50%? if yes go to 10 if no, then accept classification as high risk go to 11 Or, conduct third-tier test go to 8 Assess level of effects in field studies (see note 8) 8. In third-tier tests with field studies, test results need interpretation by expert judgement (see Note 8). Do results after a toxic effect in-field show potential for recolonization for NTA populations within 1 year? Do results after a toxic effect at off-field PEC values show ecologically insignificant effects on the NTA populations? if yes go to 10 if no go to 11 Preliminary risk classification (see notes 9, 10 and 11) 9. Classify as negligible go to Classify as low risk (L) go to Classify as high risk (H) go to 12 Analysis of uncertainty 12. At the end of each assessment and evaluation procedure, consider the data in more detail (see Note 12). Does this alter the risk classification?

6 Environmental risk assessment for plant protection products 133 if yes go back to 9, 10 or 11 if no go to 13 Integration of results of different species and tests into a final classification 13. Consider first-tier and higher-tier test results and all risk classifications in the context of the claimed uses. Keep classifications of in-field and off-field risks separate (see Note 13). Risk classifications for the tiers are integrated into a final classification: Tier 1. Both standard species low risk = final classification low risk. All other combinations lead to the final classification of high risk, or to considering next-tier study results. Tier 2. All species low risk = final classification low risk. All other outcomes lead to group-specific risk classifications or to considering next-tier study results. Tier 3. A case-by-case evaluation by expert judgement is required to combine results of different field trials to a groupspecific final risk classification go to 14 Risk Management 14. Expert judgement is required to specify the separate risk management strategies (see Note 14) to reduce unwanted risks for non-target arthropods inhabiting in-field habitats and off-field habitats go to 15 Last verification and definitive risk classification 15. Do the proposed risk management and mitigation measures influence the risk classification? if yes go back to 12 if risk classification appears to be stable end of risk assessment Explanatory notes Risk assessment in general This subscheme is developed to assess the risks of all formulations intended for spray treatments above the ground, in particular those directed onto the crop. It does not cover formulations for treating the soil such as seed treatments, granules and pellets. Risks to non-target arthropods from soil-incorporated products are covered in Chapter 8 Soil Organisms and Functions. The first tier (step 2 5) is a standard procedure, required for compounds based on all active substances 1. However, the procedure can also be used for risk assessment of formulated products. In principle, the scheme assesses the risk of a product by comparing, in the first tier, the toxicity to non-target arthropods with the expected exposure (PEC), which depends on the recommended application rate. All testing in this tier is with a formulated product, either the product under test or a typical formulation representative of the characteristics of the active substance. In higher tiers 2 effects of formulated products are studied under more realistic conditions. The risks are examined separately for the in-field and off-field environments as these require separate risk assessment and risk management approaches. The off-field risk assessment is concerned with non-target arthropods inhabiting natural and semi-natural off-field habitats, such as hedgerows and woodland. Note 1. Possibility of exposure Non-target arthropods occur in and near most crop types in the open field. For most uses of a plant protection product, at least one arthropod species or group would be exposed. For example, leaf-applied treatments may expose non-target arthropods living on the plants or on the soil by direct exposure, drift or residues. However, preparations such as pastes or wound sealants will not expose non-target arthropods and are thus unlikely to affect them adversely. Non-target arthropods in artificial environments such as glasshouses will be exposed but, as these have been intentionally introduced or maintained, these populations are not subject to risk assessment as elaborated in this scheme and are not considered further. Note 2. Laboratory toxicity tests with the active substance Standard species are Aphidius rhopalosiphi (a parasitoid hymenopteran insect) and Typhlodromus pyri (a predatory phytoseiid mite). These standard indicator species have been shown to be relatively sensitive and representative of non-target arthropods in a comparative study of arthropods of 9 families exposed to 95 plant protection products (Candolfi et al., 1999) and are therefore suitable as indicator species for this group of terrestrial non-target arthropods (SETAC/ESCORT2: Candolfi et al., 2001). The Tier 1 laboratory test is a dose response test assessing lethal effects of the active substance on the test species. This tier has a screening character and is required for all active substances intended for use in plant protection where exposure of non-target arthropods can occur. At this tier, detailed information is not considered, e.g. about sublethal effects or about the effects of specific formulations. A representative formulation should be used for this test instead of the active substance as relevant experimental exposure of non-target arthropods to the pure active substance is impractical. There may be reasons to use a specific formulated product under assessment as well, where extrapolation from the typical formulation is not possible, in particular for assessing the effects of combination products containing more than one active substance. Testing on inert substrate according to validated methods is required as this substrate allows maximal standardization of toxicity-assessing methodology. Testing guidelines validated by the Joint Initiative of EPPO, IOBC and BART (see the guidelines indicated in Candolfi et al., 2000a) should be used for dose response tests. The study should include different application rates of the test compound 1It is suggested that this first-tier procedure (step 2 5) is fitted into the Annex II requirements for active substances of the Uniform Principles (EU, 1994) of EU Directive 91/414/EEC. 2It is suggested that higher tier procedures (step 6 8) are fitted into the Annex III requirements for formulated products of the Uniform Principles (EU, 1994) of EU Directive 91/414/EEC.

7 134 Arthropods and should allow calculation of a dose response regression curve. The report should include regression analysis and calculation of the required LR 50 3, to be expressed in g of active substance or formulated product per ha. In certain conditions, e.g. when the toxicity is very low compared with maximum field rates or in other cases where experimental assessment of a reliable LR 50 value is not reasonably possible, a limit test 4 may be sufficient. This is the case if it can be shown with the limit test results that LR 50 > MAF* application rate, where MAF is the multiple application factor (see Note 3). Note 3. Assessment of the PEC Assessment of PEC in-field The estimate of the in-field exposure (or in-field PEC, i.e. predicted environmental concentration 5 ) of non-target arthropods is the highest recommended use rate of the product, times a multi-application factor (MAF). It is expressed in g of active substance or product per ha: In-field PEC = application rate MAF In Tier 1, only single applications are tested. Multiple applications are simulated by applying a MAF to the single application rate. As the main route of exposure of non-target arthropods to plant protection products is through the (accumulated) residues on (leaf) surfaces, the MAF quantifies the expected residue accumulation. The MAF depends on the ratio between the half-life of the product, the spray interval and the number of applications (see references cited below). For multiple application products where no information on residue accumulation or on the half-life of the product and the spray interval is available, a default MAF value 6 can be used, based on the 90th percentile value of the distribution of the DT 50 : spray interval ratios for 32 representative multiple application 3LR 50 = Application rate causing 50% mortality of the organisms under worst-case laboratory conditions. LR 50 is expressed in g or ml formulated product per ha (or in g or ml active substance per ha). Application rate and LR 50 must not differ in their units, i.e. both must be related to either formulated product or active substance rates. In this scheme the LR 50 is used throughout as the toxicity parameter of a plant protection product for a nontarget arthropod species. The value of 50% mortality is chosen for statistical reasons of reliability. The mortality criterion (lethal effects) is chosen because it is often not feasible to generate reliable dose response data on sublethal parameters due to the high variability of reproduction in arthropods. 4A limit test is defined as a single-dose laboratory test that reliably shows that the LR 50 of the compound is higher than the single dose tested. If the LR 50 is shown to be > MAF*PEC, then the evaluation according to this scheme can be completed successfully without having to know precisely the value of the LR 50. However, LR 50 values are always preferable as only these allow an efficient analysis of uncertainty. 5The PEC (predicted environmental concentration) in the context of this scheme is the predicted deposit per surface area (in g ha 1 product or active substance) in the field or at a given distance r from the field, i.e. the maximum recommended field rate in-field or the drift-related product deposit off-field respectively. 6SETAC/ESCORT3 recommends to apply a default MAF factor of 3 in cases where neither dissipation rate nor number of treatments is specified. plant protection products. For guidance, calculations and a reference table of MAFs, see SETAC/ ESCORT2 (Candolfi et al., 2001, appendix 5; Gonzalez-Valero et al., 2000). Assessment of PEC off-field The estimate of the off-field exposure is the drift deposit (or off-field PEC), i.e. the predicted deposit per surface area in g ha 1 product at a given distance from the field border, again expressed in g of active substance per ha. The given distance is defined in the acceptability criterion for non-target arthropods. If effects are greater than a certain trigger value at this distance r from the field border, then risk is classified as high unless acceptability of these effects is justified with higher-tier evaluations. The SETAC/ ESCORT2 workshop recommends to set this distance at default values of r = 1 m for arable crops and 3 m for orchard and vineyard crops (Candolfi et al., 2001). For calculation of the PEC values for the two indicator species and for in- and off-field exposure scenarios, the following equations should be used: In-field PEC = Application rate MAF Off-field PEC = Application rate MAF drift factor/vegetation distribution factor In these formulae, application rate refers to a single application and is expressed in g or ml formulated product per ha (or in g or ml active substance per ha); MAF (multiple application factor) is the ratio between the initial concentration after the last of several applications and the initial concentration after a single application; drift factor = %drift/100. Spray drift is the most relevant exposure route for non-target arthropods in off-field areas. Usually, the overall 90th percentile drift data according to Ganzelmeier et al. (1995), recently recalculated by the German BBA and UBA and published by the BBA (2000), should be used to estimate off-field drift deposition values (%drift). This proposal is based on the recommendations of the FOCUS surface water group which has defined that a reasonable worstcase scenario for drift estimation is represented by an overall 90th percentile probability. Default drift values for a distance from the field border of 1 m for arable crops and 3 m for orchards/vineyards should be used to estimate the off-field PEC. Overall 90th percentile drift values are presented in the Annex of Chapter 3 (Air), Table 3.3 and in SETAC/ESCORT2 report (Candolfi et al., 2001). Finally, the SETAC/ESCORT2 report recommends to use a vegetation distribution factor of 10 as the 90th percentile drift values overestimate drift in vegetated areas (Candolfi et al., 2001). Note 4. Calculation of hazard quotients The risk assessment evaluates the appropriate realistic worstcase exposure, both within-field and off-field, and compares this with the toxicity of the plant protection product to each of the non-target arthropod species. The risk assessment is done by dividing the predicted exposure (PEC) of the organisms by the toxicity (LR 50 ) in the form of the hazard quotient

8 Environmental risk assessment for plant protection products 135 (HQ) 7. PEC and LR 50 are both expressed in the same units of g ha 1 and are both expressed in either active substance or formulated lead product: Hazard Quotient HQ = (PEC/LR 50 ) correction factor 8 The hazard quotients are calculated for each of the standard species (A. rhopalosiphi and T. pyri), at PEC in-field (distance 0 m) and at PEC off-field (distance r m). These calculations thus yield four different values. Note 5. Assess the risks in-field and off-field The hazard quotients are compared with the trigger value q which should be used as a decision threshold. Specific values for the trigger value q have been recommended by SETAC/ ESCORT2, based on an extensive empirical comparison of the hazard quotients and known acceptable effects from field and semi-field studies for the two indicator species. Significant ecological effects of plant protection products on non-target arthropod populations are not expected at a hazard quotient of q or less (Campbell et al., 2000; Candolfi et al., 2001). Trigger values recommended by SETAC/ ESCORT2 are q = 2 for both species (A. rhopalosiphi and T. pyri). The result of this evaluation is a preliminary risk assessment. The assessment need not be continued if the assessment result for both indicator species at this stage is low risk, unless the evaluated compound is an unusual product such as an insect growth regulator (IGR). In these cases, higher-tier studies where other organisms may be tested (step 6) should be carried out with exposure conditions representative of those in the field. The infield and off-field risks are assessed separately as they require separate exposure estimates and risk management measures. Note 6. Higher-tier tests with the formulated product The classifications for the in- and off-field areas, respectively, resulting from the previous hazard quotient assessment are either accepted as definitive, or higher-tier testing may be used to show that effects under more realistic conditions are nevertheless acceptable as prescribed by Annex VI of the EU Directive. Higher-tier tests may comprise extended laboratory tests (e.g. on natural substrates), semi-field bioassays and field 7Hazard quotient = PEC/LR 50 (both in g product or active substance ha 1 ). The hazard quotient is also called ETR or Effect Toxicity Ratio. In this scheme, dosages and concentrations are expressed in quantity of test material. This test material represents either active substance or formulated product, provided that in the whole scheme only one is used. For products with more than one active substance, formulated product should always be used. 8SETAC/ESCORT2 advises the inclusion of a correction factor in the offfield hazard quotient for off-field species diversity. Since the species sensitivity analysis as reported by SETAC/ESCORT2 was mainly based on a comparison of in-field species which represent a lower species diversity than expected within off-field habitats, it recommended to integrate an uncertainty (safety) factor of 5 into the off-field HQ calculation to account for uncertainty with the applicability of T. pyri and A. rhopalosiphi as indicator species for off-field non-target arthropods. Evidently, the in-field correction factor is 1. experiments. They are not necessarily executed in this order but depending on the question that needs to be answered one may choose, for instance, to skip extended laboratory tests and move directly to semi-field or field experiments. If one or both indicator species are affected in tier 1 testing, the affected species should be tested further in higher-tier studies. Moreover, one (if only the in-field HQ value is exceeded) or two (if the off-field HQ value is also exceeded) additional species should be included in higher-tier tests. For these extended laboratory, aged residue and/or semi-field tests, the following species are suitable: Orius laevigatus, Chrysoperla carnea, Coccinella septempunctata and Aleochara bilineata. The choice of additional species including criteria for the choice, testing methodology, how to deal with compounds for multiple application, etc., is discussed in SETAC/ESCORT2 (Candolfi et al., 2001). IGRs should be tested in these higher tiers on susceptible, immature life stages. Further research will be needed on the ecology of off-field non-target arthropods especially with regards to species distribution between in-field and off-field habitats, and their sensitivity to plant protection products. When such data becomes available and appropriate test systems have been validated, the list of species should be reconsidered. In the meantime, it is proposed to apply an uncertainty factor for evaluating the offfield hazard quotients to cover potential sensitivity differences between the tested indicator species and other arthropods. Since the species sensitivity analysis as reported by SETAC/ ESCORT2 was mainly based on a comparison of in-field species which represent a lower species diversity than expected within off-field habitats, an uncertainty (safety) factor of 5 is considered appropriate for the off-field HQ calculation to account for uncertainty with the applicability of T. pyri and A. rhopalosiphi as indicator species for off-field non-target arthropods (see Note 4). Semi-field studies Semi-field studies, and indeed field studies, are more representative of the type of expected exposure than laboratory studies and therefore bring more weight to the final classification. Semi-field trials usually involve the release of a single species of arthropod, preferably reared in the laboratory, into a cage which encloses an area of the crop (arable crops) or part of the plant (fruit tree) grown under conditions as close to commercial practice as possible. The cage should be such as to retain a high proportion of the arthropod population and allow a high rate of collection of surviving and dead individuals at the end of the trial, while affecting the microclimate and environment as little as possible. In some circumstances, it may be more practical to proceed from the laboratory tests to field trials directly, omitting this stage. The product should be applied according to good standard practice at the highest recommended rate. In some test methods, survivors may be returned to the laboratory for further studies, such as measuring reproductive rates. Because semi-field trials involve the release of single species, analysis and interpretation of the results is usually straightforward. In most cases, one test may be sufficient, if all test conditions are truly relevant and study design and replication is statistically robust (Candolfi et al., 2000b). A toxic

9 136 Arthropods reference product should be included so that the ability of the trial design to detect effects can be confirmed. Other reference products may also be included, e.g. a soft or non-hazardous reference product or product known to be harmless in comparable conditions. SETAC/ESCORT2 and Candolfi et al. (2000b) provide guidance on these semi-field test methods. Field studies The choice of crop, or crops, to be tested depends on the integration of results from the earlier laboratory and semi-field tests and on expert assessment of where harmful effects are most likely to be observed, with particular emphasis on major crops. Sampling should be done before as well as after treatment, to establish baseline measurements. If an indication of test sensitivity under different conditions is required, at least two tests should be done, but one test may be sufficient if test conditions are demonstrated to be relevant and study design and replication is statistically robust. The product should be applied according to good standard practice at the highest recommended rate. Sufficient replication should be included in the trial to enable effects, if they occur, to be detected. A toxic reference product should be included so that the ability of the trial design to detect effects can be confirmed. Other reference products may also be included, e.g. a soft or non-hazardous reference product. Useful further guidance is given in the SETAC/ ESCORT2 guidance document (Candolfi et al., 2000b, 2001). Note 7. Assessment of the level of effect in second tier For extended laboratory, aged residue and semi-field studies, lethal (mortality) as well as sublethal (reproduction, parasitization) parameters should be used to evaluate the data. Assessments for in- and off-field exposure scenarios should be performed separately. A trigger value for lethal or sublethal effects of 50% 9 after exposure of the test organisms to fresh or aged residues of plant protection products should be used, according to SETAC/ ESCORT2. The same threshold values should be used for both in- and off-field habitats. Data on lethal and sublethal effects in aged residue studies should be used to demonstrate the potential for recolonization of an affected non-target arthropod population. A potential for recolonization should be demonstrated within 1 year for in-field habitats and within an ecologically relevant time for off-field habitats. For a detailed description of evaluation and interpretation of semi-field data, see the Joint Initiative guidance document (Candolfi et al., 2000b). 9The effects observed in the test treatment(s) should be corrected for control performance (e.g. Abbott corrected mortality or effect according to Henderson & Tilton). Note 8. Assess level of effect in third tier In field trials, effects on populations rather than on individuals should be the testing endpoint. The effect on the population of a species including time to recolonization/recovery should be analysed by comparison with control plots. There is no fixed threshold (trigger) value for acceptability of effects, because the consequence/impact of treatments and potential for recovery can be markedly different for different organisms and circumstances. Factors such as mobility of a species, reproduction time and developmental stage at risk can influence the severity of product effects on a population. Therefore, acceptability has to be assessed on a case-by-case basis for each arthropod taxon or group under investigation. Expert judgement is required to interpret field study results. Assessments for in- and off-field exposure scenarios should be performed separately because the criteria for acceptability of effects are different in-field and offfield, as are the subsequent risk management options. In some cases, in particular with insecticides, a certain level of adverse impact on in-field arthropods is likely and almost unavoidable. The final criterion for high risk in-field is that non-target arthropods affected by plant protection products should be able to recover to levels of corresponding control fields within, at most, 1 year after treatment(s). For high-risk off-field at a given distance of r m from the field border, no ecologically significant effects on populations of non-target arthropods should occur, as recommended by SETAC/ ESCORT2 (Candolfi et al., 2001). Significant effects here imply that population recovery is not expected to occur within a reasonable delay. Where significant off-field effects are detected, the duration of effect and the range of taxa affected should be taken into consideration. For a detailed description of evaluation and interpretation of field data and acceptability of effects, see SETAC/ ESCORT2 (Candolfi et al., 2001) and the Joint Initiative guidance document (Candolfi et al., 2000b). These expert judgements are, however, difficult to implement. Instead, as a more practical decision criterion, the effects can be compared with the water-treated or non-hazardous control (i.e. with the soft reference product), provided that the environmental conditions are suitable for the detection of hazards to the test organisms and that the test has been statistically reliable. However, this evaluation also requires a certain amount of expert judgement. Treatment effects equal to or less than those caused by the non-hazardous control (or if not available, by the water-treated control) are classified as low risk, others as high risk. Of course, this conclusion is only justified when the test has sufficient statistical power. At each higher-tier risk assessment step (extended laboratory, semi-field or field testing level), a refined risk assessment (e.g. taking into account new ecotoxicological data or refined exposure assessments) can be conducted or risk reduction measures can be imposed to demonstrate that the plant protection product has no unacceptable impact on habitats of concern. Notes 9, 10 and 11. Risk classifications Categories of risk can be interpreted as follows. Negligible risk means that there is no reasonable chance of exposure of non-target arthropods. Low risk means that significant ecological and between-season effects are not expected for non-target arthropods. Medium risk is not a possible outcome of this subscheme, as there are no endpoints leading to a classification of medium risk. High risk off-field means that significant ecological effects are expected for non-target arthropods. Such effects are statistically significant adverse effects on populations of several taxa of non-target arthropods

10 Environmental risk assessment for plant protection products 137 over a prolonged period of time. High risk in-field means that severe and persistent effects are expected, and the affected populations are not expected to recover within 1 year. These risk classifications are done separately for in-field and off-field situations. Note 12. Analysis of uncertainty Before the final evaluation of a plant protection product, any decision should be checked by an analysis of uncertainty. At the end of the assessment procedure, the results from all tests should be checked to ensure that the correct risk category has been allocated. All the intended uses of the plant protection product should be reviewed so that the realistic worstcase situations can be identified, given the knowledge gained during the assessment procedure, and any further work which may be needed can be considered. It is at this stage that the LR 50 data will prove particularly useful. Those plant protection products which gave effects close to threshold or trigger values (from either side) should be evaluated most carefully as these are the ones which are most likely to need reclassification. Attention should be given to the risk that uncertainty and extrapolation factors accumulate to unrealistically high threshold levels, in particular in higher-tier, off-field risk studies. It requires expert judgement to recognize and avoid these risks. Note 13. Integration of results Test results and preliminary risk classifications for a single active substance may add up to considerable numbers of intermediate results: in-field and off-field, extended laboratory, aged residue, semi-field and field tests. These are schematically represented in Table 1. These results and preliminary classifications should be integrated into final classifications which should, however, remain separate for in-field and off-field. The integration of the results of all available test and evaluation results into final risk classifications should be done by expert judgement. In these final classifications, test results in the higher tiers are more decisive (i.e. more exposure representative) than lower-tier test results (field > semi-field > extended laboratory > inert surface laboratory). At Tier 1, a final classification as low risk requires that both standard species have low-risk classifications. All other results Table 1 Risk classifications Tier Tested on Risk classification In-field 1: lab test on Species 1 inert substrate Species 2 2: extended lab, Species 1 aged residue Species 2 or semi-field Species 3 test Species 4 (not required) 3: field test Populations of key species or arthropod groups Off-field lead to high-risk classifications that can be changed only by higher-tier test results. At Tier 2, a final classification of low risk requires that all tested species have low-risk classifications. Of course, these classifications are done separately for within-field and off-field situations. At Tier 3 (field tests), integration of results requires expert judgement. Comparisons with effects of appropriate reference products may strongly facilitate this expert judgement. Note 14. Risk management and risk reduction Individual registration authorities are likely to differ in the importance they place on risk management and the methods which can be used. The points raised here provide guidance on how to deal with the assessed risks. They include label statements and warnings, restrictions on specific uses and extension messages. The exact wording of labels will involve legal considerations and be decided on at the national level. Risk management should be done separately for the in-field and offfield situations. Useful examples of national risk management policies are given for the UK (Campbell, 1998) and Germany (Forster & Rothert, 1998). For risk mitigation and management, precise definitions should be used for the different areas 10 where the risks to nontarget arthropods are of concern. A field (or within-field area) is the land intended for agricultural production activities (including horticulture and forestry). The field includes a cropped area and field margins. A cropped area is that part of the field intended to grow the crop. It extends half a row distance beyond the centre of the last crop row. A field margin is the outer part of the field, between the cropped area and the off-field area. The field margin includes the parts of land used for agricultural purposes such as roads and fences, wind breaks, and vegetation planted for reduction of drift (catch crop) and run-off. The offfield area is all area surrounding the field. The off-field area includes natural and semi-natural habitats, in particular hedgerows and woodland. The edge of the field is the boundary line between the field margin and the off-field area. Within-field risk management If risk is classified as low, there is no need to limit registration or to prescribe use restrictions. If risk is classified as high 11, it is suggested (subject to national policy) to inform the user by advisory phrases in the text on the label. These phrases should advise the user how to reduce the risks by modifying the application rate, application frequency and intervals, or timing of application (early and late crop stages, presence of concerned populations of non-target arthropods), etc. An example of an advisory label phrase for a high-risk 10Plant protection products are not necessarily always used in crops, nor are they necessarily always used during cropping periods. Therefore it is preferred in this scheme to use fields rather than crops in the definitions of the areas to be protected. 11There are also high risks that are not relevant and do not require management. Toxic effects of herbicides to non-target arthropods which live on the weeds that are controlled with the same herbicides may be considered as non-relevant. It requires expert judgement to recognize such cases.