Case Study Part 1: How to Calculate Appropriate - ResearchGateкод для вставки
Ecotoxicology, 14, 877вЂ“893, 2005 Г“ 2005 Springer Science+Business Media, Inc. Printed in The U.S.A. DOI: 10.1007/s10646-005-0034-4 Case Study Part 1: How to Calculate Appropriate Deterministic Long-Term Toxicity to Exposure Ratios (TERs) for Birds and Mammals R.F. SHORE,1,* D.R. CROCKER,2 H.R. AKCAKAYA,3 R.S. BENNETT,4 P.F. CHAPMAN,5 M. CLOOK,6 M. CRANE,7 I.C. DEWHURST,7 P.J. EDWARDS,5 A. FAIRBROTHER,8 S. FERSON,3 D. FISCHER,9 A.D.M. HART,2 M. HOLMES,2 M.J. HOOPER,10 M. LAVINE,11 A. LEOPOLD,12 R. LUTTIK,13 P. MINEAU,14 D.R.J. MOORE,15 S.R. MORTENSON,16 D.G. NOBLE,17 R.J. OвЂ™CONNOR,18 W. ROELOFS,2 R.M. SIBLY,19 G. C. SMITH,2 M. SPENDIFF,20 T. A. SPRINGER,21 H.M. THOMPSON2 AND C. TOPPING22 1 Centre for Ecology and Hydrology, Monks Wood, Cambridgeshire, UK 2 Central Science Laboratory, Sand Hutton, York, UK 3 Applied Biomathematics, Setauket, NY, USA 4 USEPA/NHEERL/MED, Duluth, MN, USA 5 Jealotts Hill International Research Station, Bracknell, UK 6 Pesticides Safety Directorate, York, UK 7 Crane Consultants, Faringdon, Oxfordshire, UK 8 USEPA/NHEERL/WED, Corvallis, OR, USA 9 Bayer Corp, Research & Development, Stilwell, KS, USA 10 Texas Tech University, Lubbock, TX, USA 11 Duke University, ISDS, Durham, NC, USA 12 Wildlife International, Berkampweg 1, 7231, Warnveld, CL, The Netherlands 13 RIVM, CSR, Utrecht, The Netherlands 14 Canadian Wildlife Service, National Wildlife Research Centre, Ottawa, Canada 15 Ecological Risk Assessment Group, Cadmus Group, Ottawa, Canada 16 Syngenta Crop Protection, Inc., Greensboro, NC, USA 17 British Trust for Ornithology, Norfolk, Thetford, UK 18 Wildlife Ecology, University of Maine, Orono, ME, USA 19 University of Reading, Whiteknights, Reading, UK 20 Health and Safety Laboratory, Broad Lane, Sheп¬ѓeld, UK 21 Wildlife International, Easton, MD, USA 22 EcoSol, FaЛљrupvej 54, DK-8410, RГёnde, Denmark Accepted 25 March 2005/Published online 23 November 2005 Abstract. In the European Union, п¬Ѓrst-tier assessment of the long-term risk to birds and mammals from pesticides is based on calculation of a deterministic long-term toxicity/exposure ratio (TERlt). The ratio is developed from generic herbivores and insectivores and applied to all species. This paper describes two case studies that implement proposed improvements to the way long-term risk is assessed. These reп¬Ѓned methods require calculation of a TER for each of п¬Ѓve identiп¬Ѓed phases of reproduction (phase-speciп¬Ѓc *To whom correspondence should be addressed: Tel.: +44-1487-772517; Fax: +44-1487-773467; E-mail: firstname.lastname@example.org 878 Shore et al. TERs) and use of adjusted No Observed Eп¬Ђect Levels (NOELs) to incorporate variation in species sensitivity to pesticides. They also involve progressive reп¬Ѓnement of the exposure estimate so that it applies to particular species, rather than generic indicators, and relates spraying date to onset of reproduction. The eп¬Ђect of using these new methods on the assessment of risk is described. Each reп¬Ѓnement did not necessarily alter the calculated TER value in a way that was either predictable or consistent across both case studies. However, use of adjusted NOELs always reduced TERs, and relating spraying date to onset of reproduction increased most phase-speciп¬Ѓc TERs. The case studies suggested that the current п¬Ѓrst-tier TERlt assessment may underestimate risk in some circumstances and that phase-speciп¬Ѓc assessments can help identify appropriate risk-reduction measures. The way in which deterministic phase-speciп¬Ѓc assessments can currently be implemented to enhance п¬Ѓrst-tier assessment is outlined. Keywords: risk assessment; pesticide exposure; no observed eп¬Ђect level; skylark; wood mouse Introduction The European Union (EU) п¬Ѓrst-tier assessment of long-term risk to birds and mammals from pesticides (European Commission, 2002) is based on calculation of a long-term toxicity/exposure ratio (TERlt); the lower the ratio, the greater the risk to wildlife. If the TERlt is less than a trigger value of п¬Ѓve, this indicates that there should be no authorisation of the pesticide unless an appropriate risk assessment is carried out. The toxicity component of this ratio is derived from the lowest No Observed Eп¬Ђect Level (NOEL) determined for birds using the avian reproduction test (OECD, 1998) and for mammals using the suite of endpoints measured in medium and long-term toxicity tests (European Commission, 2002). The exposure term is an вЂ�вЂ�estimated theoretical exposureвЂ™вЂ™ (ETE) based on standard crop and wildlife scenarios using default values for generic indicator species. It is meant to represent a realistic worst-case assessment. TERs are a relatively crude measure of risk and their use has been criticised (Tiebout and Brugger, 1995; Calow, 1998; Bennett et al., 2005; Mineau, 2005). For example, there is disagreement whether the exposure estimates and the toxicity endpoints are over or under protective, there is no indication of the type or scale of eп¬Ђects, there is uncertainty about the validity and accuracy of extrapolating laboratory toxicity data to wild species in the natural environment, and there is no objective scientiп¬Ѓc justiп¬Ѓcation for a TERlt trigger value of п¬Ѓve. Furthermore, the TERlt, in providing a single deterministic value, tends to mask the variability and uncertainty that exists in both the exposure and toxicity data (Solomon et al., 2000). A workshop was convened in January 2004 with the aim of improving long-term risk assessment for birds and mammals (Hart and Thompson, 2005). Better methods for estimating exposure were outlined (Crocker, 2005) and a novel approach for assessing toxicity was proposed; appropriate NOELs were determined for key stages in the breeding cycle and the possible consequence of toxic impact at each stage were estimated (Bennett et al., 2005). Ways of incorporating inter-species diп¬Ђerences in toxicological response into the assessment were considered (Luttik et al. 2005). Methods were also developed to link the timing of reproduction to the time-course of exposure (Fischer, 2005), and to translate the assessment of risk to individuals into likely eп¬Ђects on populations (Sibley et al., 2005). The current paper is п¬Ѓrst of two that describe case studies which implement the approaches developed at the workshop. There were two case studies, one for birds and another for mammals. Both used speciп¬Ѓc, but diп¬Ђerent, exposure scenarios. The aim of this п¬Ѓrst paper is to illustrate the way in which deterministic TERs can be generated for diп¬Ђerent reproductive phases (phase-speciп¬Ѓc TERs), as advocated by Bennett et al. (2005), and then be progressively reп¬Ѓned using more realistic exposure estimates. The way in which these approaches can be incorporated into current risk assessment procedures is also described. The second case study paper (Roelofs et al., 2005) outlines a more radical departure from current deterministic TER assessments. It shows how probabilistic approaches can be applied to the Long-Term Toxicity to Exposure Ratios (TERs) for Birds and Mammals 879 production of phase-speciп¬Ѓc TERs, and how these can then be developed to model overall impacts on individual breeding success and on subsequent population eп¬Ђects. Description of case study scenarios Crop type Six standard scenarios (grassland, early cereals, late cereals, early/late leafy crops, orchard/vine/ hops, seed treatment) are used for EU risk assessments (European Commission, 2002). Winter wheat, a cereal crop widely grown in Britain and that currently occupies 42% of arable land (Garthwaite et al., 2003), was the scenario that was taken for both the bird and mammal case studies. Indicator species When using the EU standard scenarios for risk assessment, the calculated ETE for the п¬Ѓrst-tier risk assessment is calculated for generic indicators. In cereals, these are a large herbivorous bird (goose, unspeciп¬Ѓed species), a small insectivorous bird (wren, Troglodytes spp or tit, Parus spp), a small herbivorous mammal (vole, unspeciп¬Ѓed species) and an insectivorous mammal (shrew Sorex spp) (European Commission, 2002). In Britain, the skylark (Alauda arvensis), an omnivorous bird that feeds its chicks on insects (Green, 1978; Poulsen et al., 1998; Donald et al., 2001) and the wood mouse (Apodemus sylvaticus), a ubiquitous rodent that eats mostly seeds and insects in arable systems (Tattersall and Macdonald, 2003), occur widely in arable п¬Ѓelds where they forage extensively and also breed. Both species are relatively small (and so have high food intake to body mass ratios), and are likely to ingest pesticide-contaminated forage. They are therefore likely to be good speciп¬Ѓc (rather than generic) indicators for pesticide-related eп¬Ђects on birds and mammals in cereals systems, and so were used in the case studies. An added advantage was that the biology of these species is well studied and the ecological data required for the case studies was readily available. Pesticides in case studies Although the pesticides in the case study scenarios were п¬Ѓctitious and deliberately given qualities that would create long-term risk, all of their toxicological and agronomic attributes were realistic. An insecticide, compound I, was used in the skylark case study and was borrowed from a previous case study (Mineau et al., 2001). Compound I was assumed to be applied once only between April and July at a rate of 0.25 kg/ha. A fungicide, compound F, was used for the wood mouse case study. It was assumed that this was applied between March and July at a rate of 0.75 kg/ha and that there were two applications separated by seven days. The timing of spraying for these compounds reп¬‚ected typical usage patterns for insecticides and fungicides in Britain (Garthwaite et al., 2003). The relevant toxicity data needed for calculation of the TERs for compounds I and F, are given in Tables 1 and 2 respectively, together with values adjusted to take into account inter-species variability in sensitivity (Luttik et al., 2005). Calculation of toxicity/exposure ratios Reп¬Ѓnement stages of the risk assessment The case studies begin by using the current EU guidance to assess the risk of insecticide I to birds and fungicide F to mammals. They then gradually increase the realism of the toxicity and exposure scenarios. The progression is as follows: 1. Standard п¬Ѓrst-tier assessment using generic wildlife species and default values for pesticide residues and half-life as described in current EU guidance (European Commission, 2002). 2. Phase-speciп¬Ѓc long-term TERs for default indicator species. The toxicity endpoints identiп¬Ѓed by Bennett et al. (2005) for diп¬Ђerent reproductive phases are used to calculate phaseвЂ“speciп¬Ѓc TERs. The exposure estimate is reп¬Ѓned so that it is weighted over a time period appropriate to the endpoint. 3. Phase-speciп¬Ѓc long-term TERs for skylark and wood mouse. Generic species are replaced with wildlife species that are vulnerable to 880 Shore et al. Table 1. Avian toxicity data for compound I. Data are from Mineau et al. (2001) a Toxicity end-point Acute NOEL for adult body weight prelaying NOEL for eggs laid per hen per day NOEL for mean eggshell thickness per hena NOEL for eggshell cracking NOEL for % fertile eggs per egg set per hen NOEL for proportion of hatching per egg set per hen NOEL for proportion of 14-day chicks per no/hatchlings per hen NOEL for 14-day chick body weight d Five-day dietary LC50 Estimated п¬Ѓve-day LC05 d,e Test speciesb Toxicity value Adjusted toxicity valuec Unit for toxicity value BQ JQ M JQ M JQ M JQ M JQ M JQ M JQ M M M 20 2.10 0.157 6.66 0.261 6.66 0.261 6.66 0.261 6.66 0.261 2.10 0.157 6.66 0.261 22.1 9.53 0.608 0.029 mg/kg mg/kg BW/d 0.067 mg/kg BW/d 0.067 mg/kg BW/d 0.067 mg/kg BW/d 0.067 mg/kg BW/d 0.029 mg/kg BW/d 0.067 mg/kg BW/d 0.672 0.289 mg/kg BW/d mg/kg BW/d a No observed eп¬Ђect level (NOELs) for dietary intake (mg/kg BW/d) calculated from reported no observed eп¬Ђect concentrations (NOECs) using data on the feeding rates and body weights of test animals. b Species are BQ: bobwhite quail (Colinus virginianus); JQ: Japanese quail (Coturnix coturnix japonica), M: mallard (Anas platyrynchos). c Toxicity factors adjusted to incorporate inter-species extrapolation using factors calculated following Luttik et al. (2005). These values are used in Tables 4вЂ“7. d Dietary lethal concentration (LC) data converted to daily intake per unit body weight using data on the feeding rates and body weights of test animals. e Assuming a probit slope of approximately 4.5 (Mastrota, pers. comm.), LC50 may converted to LC05 by multiplying the LC50 value by 0.431. Table 2. Toxicity data for compound F in rats NOEL value for various toxicity end-points (mg/kg BW/24 h)a Toxicity value Adjusted toxicity valueb 90-day toxicity test Behaviour in 2-generation testc no/pups per mated female teratogenicity No/weanlings per mated female Juvenile survival to 4 weeks Body weight at 2 weeks old No/weanlings per F2 female 22.4 4.5c 90 30 90 90 27 4.5 1.50 0.302 6.04 2.01 6.04 6.04 1.81 0.302 a No observed eп¬Ђect level (NOELs) for dietary intake (mg/kg BW/d) either as reported or calculated from no observed eп¬Ђect concentrations (NOECs) using data on the feeding rates and body weights of test animals. b Toxicity factors adjusted to incorporate inter-species extrapolation using factors calculated following Luttik et al. (2005). These values are used in Tables 4вЂ“7. c Lowest value taken from measurements made on eп¬Ђects on food or water intake or body weight. exposure due to the proposed use of the pesticide. Exposure calculations make use of speciп¬Ѓc body weight, diet composition and habitat use. 4. Chronological TERs. The species breeding calendar and spraying dates are applied to the phase-speciп¬Ѓc long-term TERs for the skylark and wood mouse. Long-Term Toxicity to Exposure Ratios (TERs) for Birds and Mammals 881 Standard п¬Ѓrst-tier assessment of long term toxicity/ exposure ratio (TERlt) Phase-speciп¬Ѓc long-term TERs for default indicators The п¬Ѓrst-tier TERlt, is calculated as the NOEL/ ETE ratio. The toxicity term used in the calculations was the most sensitive of the range of possible endpoints determined for compound I in birds (avian case study, Table 1) and compound F in laboratory mammals (mammalian case study, Table 2). The NOEL is often calculated from a dietary No Observed Eп¬Ђect Concentration (NOEC) using the reported feeding rate and body weight of the test animals. The ETE is for generic insectivorous and herbivorous birds and mammals that have п¬Ѓxed default body weights and food intake rates (Table 3). Calculation of the ETE is described in detail elsewhere (European Commission, 2002; Crocker, 2005). Default values (Table 3) are used for all exposure inputs other than application rate and the number of repeat applications of the pesticide. It is assumed that contaminated diet is not avoided (avoidance factor (AV) = 1), animals feed wholly on a single food type (proportion of food type in the diet (PD) = 1) and all diet is from the treated area (proportion of the diet obtained in the treated area (PT) = 1). In the case study scenarios, the п¬Ѓrst-tier TERlt values for birds and mammals (Table 3) were less than п¬Ѓve, the trigger value. Although the п¬Ѓrst-tier TERlt values in the case studies were all less then п¬Ѓve, they give no indication of the exact nature, likely scale or frequency of toxic eп¬Ђects. They do not even indicate that one or more reproductive stage is likely to be aп¬Ђected, nor indicate a likely adverse eп¬Ђect on the speciп¬Ѓc toxicity endpoint used in calculating the TERlt values. This is, in part, because the default time period of three weeks (post-application), over which a time weighted average (TWA) exposure is calculated, may bear little relation to the exposure period that is actually critical to the speciп¬Ѓc toxicity endpoint. It could be argued that any potential underestimate of exposure due to the use of a TWA exposure factor is more than compensated for by use of the lowest appropriate NOEL in the TERlt. The result is likely to be a conservative, and therefore precautionary, п¬Ѓrst-tier assessment. Whether this is the case is unproven. If true, it is probable that acceptable products will be inappropriately rejected. Bennett et al. (2005) argued that a more objective and informative assessment of likely eп¬Ђects on reproduction can be gained by calculating speciп¬Ѓc TERs for each phase of reproduction. This also has the advantage of explicitly aligning the exposure term to the speciп¬Ѓc toxicity endpoint. Phase-speciп¬Ѓc Table 3. Default values (European Commission, 2002) for the various parameters used to calculate the п¬Ѓrst-tier estimated theoretical exposure (ETE), application rate for pesticides, and the calculated values for the п¬Ѓrst-tier ETE and TERlt for insecticide I and birds and for fungicide F and mammals in a cereals scenario Indicator species Compound I Insectivorous bird (10 g) Large herbivorous bird (3000 g) Compound F Insectivorous mammal (10 g) Small herbivorous mammal (25 g) a Application rate (kg/ha) FIR/bw a Category 0.25 0.25 1.04 0.44 Small insects Short grass 0.75 0.75 0.63 1.39 Large insects Short grass Mean (RUD)b c ftwa (MAF)d ETE e TERflt 29 76 n/a 0.53 n/a 1.00 7.54 4.43 0.021 0.035 5.1 76 n/a 0.53 n/a 1.58 2.41 67.2 1.87 0.067 Food intake rate/body weight. Residue per Unit Dose (residue after spraying normalised to application rate of 1 kg ai/ha). c Time-weighted average factor. d Multiple Application Factor. e ETE (mg/kg BW/d)=(FIR/bw)*RUD*MAF*ftwa. Other factors used to calculate the ETE (Crocker, 2005) are all assumed to be equal to one (see text for details). f First-tier toxicity exposure ratio. Toxicity data used to calculate the TER were (i) the lowest NOEL in the avian reproduction test (0.157 mg/kg BW/24 h) for insecticide I (Table 1); (ii) the two-generation rat toxicity NOEL (4.5 mg/kg BW/24 h) for fungicide F (Table 2). b 882 Shore et al. TERs were calculated as the next step in the case studies. It was assumed that onset of each reproductive phase occurred on the day after the compound was applied (п¬Ѓrst of the two applications for compound F) and so each TER represented a worst case scenario. The TERs were calculated for the default indicator species used in п¬Ѓrst-tier assessments (herbivorous and insectivorous bird and mammal) and mostly used their default values (Table 3) when calculating the ETE term. However, the default (21-day) time period used to calculate the TWA factor (ftwa) (Table 3) was replaced by a time period appropriate to the speciп¬Ѓc reproductive phase (Bennett et al., 2005). Furthermore, the EU guidance does not provide default ftwa or multiple application factor (MAF) values (Table 3) when calculating residues on insects because of a lack of empirical data and diп¬ѓculties in making theoretical predictions. The default is to assume that there is no decay of residues on insects (European Commission, 2002). In the case study with compound I, however, data were available for rate of decay of residues on insects (DT50 = 5.66) and this was used to calculate TWAs for the insectivore phasespeciп¬Ѓc TERs. An example ETE calculation is given in Box 1. The toxicity values were also adjusted when calculating the phase-speciп¬Ѓc TERs. This involved using an extrapolation factor to incorporate species diп¬Ђerences in sensitivity to the pesticide (Luttik et al., 2005). Unadjusted and adjusted TERs, calculated using unadjusted and adjusted NOELs, respectively, are both presented for the avian (Table 4) and mammalian (Table 5) case studies. In the avian case study, all phase-speciп¬Ѓc, unadjusted TERs were less than п¬Ѓve (Table 4). This indicated that any reproductive phase might be adversely aп¬Ђected if application occurred on the day before onset of that phase. This assessment provides clearer evidence than the п¬Ѓrst-tier TERlt evaluation (Table 3) of the risk associated with the proposed use of the compound. Furthermore, the application of an appropriate exposure duration to each toxicity endpoint means the TERs can eп¬Ђectively be used as a crude ranking system to identify which reproductive phase may be most sensitive to the pesticide; in this case study, the lowest unadjusted TERs were associated with copulation and egg-laying (Table 4). One other noticeable diп¬Ђerence between the п¬Ѓrst-tier and phase-speciп¬Ѓc assessments was that the п¬Ѓrst-tier ETE was almost twice as high for the insectivore as the herbivore (Table 3) whereas the phase-speciп¬Ѓc ETEs were similar for both indicators, but always a little higher for the herbivore (Table 4). This diп¬Ђerence was largely because of the ways in which residues on forage were estimated. First-tier assessments tend to overemphasise the exposure of insectivores relative to herbivores because of their assumption of no residue decay on insects and use of a TWA Box 1. Example calculation of ETE for a generic insectivorous bird in the course of egg formation at the time of spraying (see Table 4). Default values are from Table 3. Equation ETE = (FIR/BW) Г‚ C Г‚ AV Г‚ PT Г‚ PD (mg/kg BW/d) where C = C0 Г‚ MAF Г‚ ftwa C0 = Application Rate (kg) Г‚ RUD (residue (mg/kg) per kg dose) ftw = (I-e)kt)/kt k = ln(2)/DT50 Speciп¬Ѓc values t = 3 days DT50 = 5.66 FIR/BW = 1.04 Calculations 1. k = ln(2)/5.66 = 0.122 2. kt = 0.122 Г‚ 3 = 0.367 3. C0 = 0.25 Г‚ 29 = 7.25 4. ftwa = (1)e)0.367)/0.367 = 0.837 5. C = 7.25 Г‚ 0.837 = 6.068 ETE = 1.04 Г‚ 6.068 = 6.311 Default values Application rate = 0.25 RUD = 29 Assumptions AV = 1 (no avoidance) PT = 1 (all of diet contaminated) PD = 1 (only insects eaten) MAF = 1 (only 1 application) Phase-speciп¬Ѓc eп¬Ђect of concern Reduced п¬‚edgling survival from direct exposure Reduced juvenile survival and growth from in ovo exposure Adult behavioural eп¬Ђects leading to abandonment of nesting attempt Embryotoxicity leading to reduced hatchability Adult behavioural eп¬Ђects leading to brood abandonment/abnormal parental care Reduced juvenile survival from direct exposure Reduced juvenile survival and growth from in ovo exposure 5-day dietary toxicity test with juvenilesf Proportion of 14-day juveniles per no/hatchlings per hen 5-day dietary toxicity test with juveniles 14-day-old juvenile weights/hen 0.261 Mean eggshell thickness/hen Proportion of fertile eggs/eggs set/hen Change in adult BW before egg laying Proportion hatching/eggs set/hen Change in adult BW before egg laying 0.289 0.067 0.261 0.029 0.157 9.53 0.289 9.53 0.608 0.067 0.261 20 0.608 0.067 0.067 0.029 0.608 ovum dvt TWA 5-day TWA ovum dvt TWA 5-day TWA 2-day TWA ovum dvt TWAd 1-day ETE 1-day ETE 1-day ETE 1-day ETE 1-day ETE Adjusted Duration of NOELb exposure 20 0.261 0.157 20 a NOEL No/eggs laid/hen Change in adult BW prior to egg laying1 NOEL test endpoint used as surrogate ETE Insectivore Herbivore Insectivore Herbivore Insectivore Herbivore Insectivore Herbivore Insectivore Herbivore Insectivore Herbivore Insectivore Herbivore Insectivore Herbivore Insectivore Herbivore c 5.46 6.37 6.31 6.81 5.46 6.37 6.31 6.81 7.54 8.36 7.54 8.36 7.54 8.36 6.31e 6.81 6.60 7.04 Insectivore 7.54 Herbivore 8.36 Insectivore 7.54 Herbivore 8.36 Indicator species 0.081 0.073 Adjusted TER 1.75 1.50 0.041 0.038 1.75 1.50 0.025 0.023 0.035 0.031 0.035 0.031 2.65 2.39 0.041 0.038 3.03 2.84 0.053 0.045 0.011 0.010 0.053 0.045 0.005 0.004 0.009 0.008 0.009 0.008 0.081 0.073 0.011 0.010 0.092 0.086 0.021 0.004 0.019 0.003 2.65 2.39 TER No observed eп¬Ђect level (NOELs) for dietary intake (mg/kg BW/d) calculated from no observed eп¬Ђect concentrations (NOECs) using data on the feeding rates and body weights of test animals. Lowest value used when data for more than one species given in Table 1. b Toxicity factors adjusted to incorporate inter-species extrapolation using factors following Luttik et al. (2005). c ETE in mg/kg BW/d; case study data available for residues on foliage, so default DT50 of 10 days was used to calculate TWA factors. d Time weighted average (TWA) for an exposure period equivalent to length of time for an ovum to develop (ovum dvt) in species of interest (3 days in this calculation) вЂ“ primary in ovo exposure assumed to be from material deposited in the yolk during ovum formation. e The working for this calculation is shown in Box 1. f NOEC estimated as equivalent to LC05 and LC05 extrapolated from LC50 (Table 1). a Post-п¬‚edging survival Juvenile growth and survival until п¬‚edging Incubation and hatching Reduced fertility Adult behavioural eп¬Ђects leading to territory abandonment/delayed breeding Copulation and egg laying Adult behavioural eп¬Ђects leading (5 days pre-laying through to reduced clutch size/abandonment end of laying) of nesting attempt Reduced eggshell quality Pair formation/breeding site selection Breeding phase Table 4. Phase-speciп¬Ѓc avian toxicity data for insecticide I. Endpoints from Bennett et al. (2005) and toxicity values from Table 1 Long-Term Toxicity to Exposure Ratios (TERs) for Birds and Mammals 883 Adult behavioural eп¬Ђects leading to abnormal litter care Reduced litter survival Increased developmental abnormalities 4.5 Behavioural observations in 2-generation test No/pups per mated female in 2-generation test Teratological eп¬Ђects (2-generation or prenatal development test) 28-day or 90-day toxicity tests 1.51 1.81 0.302 2.01 90 27 4.5 30 Insectivore Herbivore Insectivore Herbivore Insectivore Herbivore Insectivore Herbivore 2.41 127 2.41 127 2.41 127 2.41 127 c ETE 9.29 0.176 1.87 0.035 37.3 0.709 12.4 0.236 TER 1-day ETE 1-day ETE Insectivore 2.41 Herbivore 127 Insectivore 2.41 Herbivore 127 1.87 0.035 12.4 0.236 Insectivore 2.41 9.29 Herbivore 127 0.176 1-day ETE Insectivore 2.41 37.3 Herbivore 127 0.709 14-day TWA Insectivore 2.41d 37.3 Herbivore 81.1 1.11 14-day TWA Insectivore 2.41d 11.2 Herbivore 81.1 0.333 1-day ETE 1-day ETE 1-day ETE 1-day ETE 1-day ETE Indicator species 0.125 0.002 0.834 0.016 0.627 0.012 2.61 0.050 0.627 0.019 0.751 0.022 0.627 0.012 0.125 0.002 2.61 0.050 0.834 0.016 Adjusted TER No observed eп¬Ђect level (NOELs) for dietary intake (mg/kg BW/d) either as reported or calculated from no observed eп¬Ђect concentrations (NOECs) using data on the feeding rates and body weights of test animals. b Toxicity factors adjusted to incorporate inter-species extrapolation using factors following Luttik et al. (2005). c ETE in mg/kg BW/d. d Data not available for residues on insects and European Commission (2002) guideline (assumes no decay over time in residue concentrations in insects) followed when calculating time-weighted ETE. a 6.30 1.51 2.01 6.30 0.302 1.51 Adjusted Duration of NOELb exposure 90 22.4 30 90 22.4 a NOEL 28- or 90-day toxicity tests NOEL test endpoint used as surrogate No/weanlings per mated female in 2-generation test Post-weaning survival Reduced juvenile survival Survival to 4 weeks in until maturity 2-generation test Reduced juvenile growth and development Body weight of 4-week-old juveniles in 2-generation test Reproduction of F1 Reduced productivity of F1 generation No/weanlings per mated F1 generation female in 2-generation test Increased developmental Teratological eп¬Ђects in abnormalities in F2 2-generation test Pup growth and survival until weaning Adult behavioural eп¬Ђects leading to territory abandonment or delayed or abnormal mating Establishing breeding site, pairing, and mating Pregnancy Reduced litter size Phase-speciп¬Ѓc eп¬Ђect of concern Breeding phase Table 5. Phase-speciп¬Ѓc mammalian toxicity data for fungicide F. Endpoints from Bennett et al. (2005) and toxicity values from Table 2 884 Shore et al. Long-Term Toxicity to Exposure Ratios (TERs) for Birds and Mammals 885 for residues on foliage. The phase-speciп¬Ѓc ETEs adopted a more equitable approach by using either a one-day exposure or relatively short TWAs for residues on both insects and foliage. In the mammalian case study, the phase-speciп¬Ѓc TERs (Table 5) generally conп¬Ѓrmed the п¬Ѓrst-tier assessment for compound F in that many of the TERs were lower than the trigger value. The level of risk indicated for the herbivore was higher in the phase-speciп¬Ѓc than in the п¬Ѓrst-tier assessment because the exposure associated with the toxicity endpoint was generally higher. As in the avian case study, the fact that all phase-speciп¬Ѓc TERs were below п¬Ѓve for the herbivore provides stronger grounds, compared with a single п¬Ѓrst-tier TERlt, for requiring additional information prior to authorisation of the product. Unlike the herbivore, the insectivore unadjusted, phase-speciп¬Ѓc TERs fell below the trigger value only on two reproductive phases, establishment of breeding site and productivity of the F1 generation (Table 5). The TERs for both phases were based on a worst case (onset of phase the day after spraying) one-day ETE. This suggests that further reп¬Ѓnement of the exposure term, or alteration of spraying date so that it does not coincide with onset of breeding, may be suп¬ѓcient to reduce risk to insectivores to an acceptable level. Incorporation of an inter-species extrapolation factor in the risk assessment (by using adjusted NOELs) reduced avian (Table 4) and mammalian (Table 5) phase-speciп¬Ѓc TERs by 4вЂ“33-fold and 14вЂ“60-fold, respectively. Bennett et al. (2005) proposed that the uncertainty factor of п¬Ѓve applied to the п¬Ѓrst-tier TERlt should be reduced to one when using adjusted TERs because the main purpose of the uncertainty factor was to allow for the possibility that some wildlife species may be more sensitive to the pesticide than the test species. The present case studies indicated that, if the extrapolation factors for toxicity values proposed by Luttik et al. (2005) are accepted, a п¬Ѓvefold uncertainty factor is often likely to be inadequate for allowing for variation in species sensitivity. Therefore, the overall eп¬Ђect of using adjusted NOELs was that the phase-speciп¬Ѓc TERs for pesticides I and F more often fell below their trigger value. Phase-speciп¬Ѓc long-term TERs for skylark and wood mouse The exposure data used to calculate the phasespeciп¬Ѓc TERs in Tables 4 and 5 applied generically to any herbivore or insectivore of a given weight and assumed that diet was monophagous and always contaminated (PD and PT were both 1). Thus, they again were relatively worst-case scenarios that did not account for factors that might moderate the risk to species actually present in cereals. The next stage of the case studies was to calculate phase-speciп¬Ѓc TERs for the skylark and wood mouse, two species potentially at high risk of pesticide exposure in cereal п¬Ѓelds in Britain. The ways in which the ETE component of the TER can be reп¬Ѓned for particular species is discussed by Crocker (2005). Relevant ecological data for the skylark and the wood mouse are summarised by Roelofs et al. (2005) and were used here to reп¬Ѓne the ETE for compounds I and F, respectively. When reп¬Ѓning the ETE, it was assumed that exposure to the pesticide was exclusively via the diet and not through skin contact or inhalation, although such routes may be important under п¬Ѓeld conditions (Mineau, 2002). It was also assumed that animals were exposed to pesticides in proportion to the amount of time they spent in habitat where pesticides are used, but did not distinguish between crops or allow п¬Ѓelds to receive diп¬Ђerent spray regimes. Other assumptions were that diet composition was independent of habitat use, individuals ate the same diet on successive days, and animals did not show any avoidance of pesticide residues, although this can occur to some extent under п¬Ѓeld conditions (McKay et al., 1999). Daily food intake was estimated from information about daily energy expenditure and energy and moisture content of wildlife food (Crocker et al., 2002a). The total daily intake of pesticide (mg/kg BW/day) was calculated as shown in Box 2. For each of the parameters shown, wherever empirical data were available, calculations were based on the mean values, i.e. we did not attempt to build in worst-case assumptions by using 95th or other percentiles. The reп¬Ѓned ETEs for the skylark (Table 6) were nearly always higher than those for a generic bird 886 Shore et al. Box 2. Example calculation of ETE for an adult skylark in the course of egg formation at the time of spraying (see Table 5). Equations FIR Г‚ AV Г‚ PT BWt DEE Г°Ci Г‚ PDiГћ Гѕ Г°Cs Г‚ PDsГћ Гѕ Г°Ch Г‚ PDhГћ ETE Вј Г‚ Г‚ AV Г‚ PT BWt Г° PDi Г‚ Г°1 ГЂ MiГћ Г‚ GEi Г‚ AEiГћ Гѕ Г°PDs Г‚ Г°1 ГЂ MhГћ Г‚ GEs Г‚ AEsГћ Гѕ Г° PDh Г‚ Г°1 ГЂ MhГћ Г‚ GEh Г‚ AEhГћ where: i=insects, s=seeds, and h=herbs FIR=food intake rate (g fresh wt/day) DEE=Daily Energy Expenditure (kJ) BWt=Body Weight (g) C=Concentration (mg ai/kg wt) of pesticide residue on food PD=Proportion (fresh wt) of Diet made up by food type (0вЂ“1) M=Proportion Moisture in fresh food (0вЂ“1) GE=Gross Energy (kJ/g dry wt) provided by food type AE=Assimilation Eп¬ѓciency of food type (0вЂ“1) PT=Proportion of food eaten that is from the treated area (0вЂ“1) AV=Avoidance of treated food (0=wholly avoided, 1=not avoided) Log(DEE) = 1.0017 + 0.7034 Г‚ log(BWt ) C=C0 x MAF Г‚ ftwa ftwa=(I)e)kt)/kt k=ln(2)/DT50 ETE Вј Assumptions AV=1 (no avoidance) PT=0.51 of diet contaminated MAF=1 (only 1 application) Calculations Speciп¬Ѓc values Food B Wt g DEE kJ Insects Seeds Herbs 37.2 37.2 37.2 128 128 128 Calculated values PD GE kJ/g dry wt M AE C0 mg/kg 0.300 0.200 0.500 22.7 20.0 17.1 0.660 0.125 0.779 0.76 0.80 0.58 7.50 25.0 60.0 (Table 4). This was despite PT being reduced from the п¬Ѓrst-tier default of one to 0.51, a value derived from radio-tracking studies on skylarks in arable п¬Ѓelds in Britain (Crocker вЂ“ unpub. data). Some 50% of the adult skylark diet in spring is comprised of leaves (Green, 1978), and the foliar RUD for compound I, estimated from empirical data (Mineau et al., 2001), was approximately three times the п¬Ѓrst-tier default value. This high foliar RUD largely explained why adult skylark ETEs were usually greater than those for a generic herbivorous or insectivorous bird. Nestling skylarks are wholly insectivorous (Donald, 2004), yet the phase-speciп¬Ѓc ETEs based solely on direct dietary exposure of juveniles were lower, not higher, than the default value for a generic insectivore. This was partly due to the fact that only half of the diet was t days 3 3 3 DT50 days FIR PT g/day k ftwa 5.66 5.06 10.0 0.51 6.82 0.51 4.54 0.51 11.4 0.12 0.14 0.07 0.84 0.82 0.90 C mg/kg ETE mg/kg 6.28 20.5 54.2 Sum 0.59 1.28 8.44 10.31 assumed to be contaminated (PT=0.51). Another factor was that their daily energy expenditure (DEE) was estimated in a diп¬Ђerent way from the default п¬Ѓrst-tier calculation. The DEE of inactive nestlings kept warm by a brooding mother is unlikely to be a good measure of food intake. A large part of nestling food intake will go towards weight gain. Therefore, food intake in nestlings was based on empirical data for passerine chick feeding rates. Although it was assumed that, once out of the nest, п¬‚edgling skylarks (estimated BW 30 g), would follow the standard allometric equation (second equation in Box 2), heavier birds eat less in proportion to body weight than smaller birds. Hence it was predicted that exposure would also be lower for a 30 g п¬‚edgling skylark than for a 10 g generic passerine. 28- or 90-day toxicity tests Behavioural observations in 2-generation test No/pups per mated female in 2-generation test Teratological eп¬Ђects in 2-generation test or prenatal development toxicity test 28-day or 90-day toxicity tests No/weanlings per mated female in 2-generation test Survival to 4 weeks in 2-generation test Body weight of 4-week-old juveniles in 2-generation test No/weanlings per mated F1 female in 2-generation test Teratological eп¬Ђects in 2-generation test Change in adult BW prior to egg laying No/eggs laid/hen Mean eggshell thickness/hen Proportion of fertile eggs/eggs set/hen Change in adult BW before egg laying Proportion hatching/eggs set/hen Change in adult BW before egg laying 5-day dietary toxicity test with juveniles Proportion of 14-day juveniles per no/hatchlings per hen 5-day dietary toxicity test with juveniles 14-day-old juvenile weights per hen NOEL used as surrogate No observed eп¬Ђect level (NOELs) and ETE values expressed as mg/kg BW/d. Reproduction of F1 generation Post-weaning survival until maturity Pup growth and survival until weaning Pregnancy Wood mouse Establishing breeding site, pairing, and mating Post-п¬‚edging survival Juvenile growth and survival until п¬‚edging Incubation and hatching Skylark Pair formation/breeding site selection Copulation and egg laying Breeding phase 22.4 90 90 27 4.5 30 22.4 4.5 90 30 9.53 0.261 20 0.157 0.261 0.261 20 0.261 20 9.53 0.157 NOEL 1.51 6.30 1.51 1.81 0.302 2.01 1.51 0.302 6.3 2.01 0.289 0.067 0.608 0.029 0.067 0.067 0.608 0.067 0.608 0.289 0.029 Adjusted NOEL 7.91 7.91 5.17 5.17 7.91 7.91 7.91 7.91 7.91 7.91 1.81 10.3 11.6 11.6 11.6 11.6 11.6 10.3 10.7 1.78 10.3 Species-speciп¬Ѓc ETE 2.83 11.4 17.4 5.22 0.569 3.79 2.83 0.569 11.4 3.79 5.26 0.025 1.72 0.014 0.022 0.022 1.72 0.025 1.87 5.35 0.015 TER 0.191 0.796 0.292 0.350 0.038 0.254 0.191 0.038 0.796 0.254 0.160 0.007 0.052 0.002 0.006 0.006 0.052 0.007 0.057 0.162 0.003 Adjusted TER Table 6. Phase-speciп¬Ѓc TERs calculated for the skylark and the wood mouse. Onset of each reproductive phase assumed to occur on the day after the application of the pesticide. Duration of exposure for each reproductive phase is indicated in Tables 4 and 5 and the species speciп¬Ѓc data used to reп¬Ѓne components of the ETE are given in Roelofs et al. (2005). Long-Term Toxicity to Exposure Ratios (TERs) for Birds and Mammals 887 888 Shore et al. For the wood mouse, PT was again reduced (this time from 1 to 0.38) on the basis of radiotracking data (Crocker et al., 2002b). Arable wood mice eat seeds, insects and plant material (Watts, 1968; Gorman and Zubaid, 1993; Rogers and Gorman, 1995) and the estimated residue of compound F for this omnivorous diet (Roelofs et al., 2005) falls between those based on default RUDs for an obligate herbivore and obligate insectivore (Table 3). This largely accounted for why the phase-speciп¬Ѓc ETEs for the wood mouse (Table 6) were always lower than the default values for a generic herbivorous mammal but similar, although somewhat higher, than those for a generic insectivore (Table 5). The eп¬Ђect of using species-speciп¬Ѓc ETEs on the TER calculation diп¬Ђered between the case studies. Because almost all the ETE values increased when they were reп¬Ѓned for the skylark, the associated phase-speciп¬Ѓc TERs (Table 6) were mostly smaller than those for a generic herbivorous or insectivorous bird (Table 4), and were below п¬Ѓve in each case. The only exceptions were those associated with the direct toxicity of compound I to the chick or п¬‚edgling. In contrast, six of the 10 reп¬Ѓned TERs for the wood mouse had values that exceeded п¬Ѓve and another two values exceeded three (Table 6); most of the generic mammalian insectivore TERs but none of the generic mammalian herbivore TERs exceeded п¬Ѓve (Table 5). Thus, on the basis of the reп¬Ѓnement of the ETE, the predicted direct toxicity of compound F to the wood mouse would appear to less than that predicted for a generic mammalian herbivore such as a vole, but largely similar to that predicted for a generic mammalian insectivore such as a shrew. However, it is arguable that if a species-speciп¬Ѓc ETE is used to calculate the TER, the appropriate toxicity endpoint is an avian or mammalian NOEL that is adjusted to account for inter-species variation in sensitivity. None of the adjusted TERs exceeded one, the trigger value proposed by Bennett et al. (2005), for the skylark while only two of the adjusted TERs exceeded this value for the wood mouse (Table 6). Use of species-speciп¬Ѓc ETEs and adjusted NOELs would, therefore, indicate that compounds I and F could potentially have signiп¬Ѓcant impacts on multiple reproductive phases of the skylark and wood mouse, respectively. Chronological TERs The phase-speciп¬Ѓc TERs in Table 6 remain worstcase scenarios in that they assume spraying occurs the day before onset of each reproductive phase. In reality, breeding may begin before or well after pesticide application and few, if any, of the speciп¬Ѓc reproductive phases may occur at, or close to, the spraying date. Thus, many of the ETE values in Table 6 could be overestimated. To evaluate how signiп¬Ѓcant the temporal dynamic between spraying date and onset of reproduction might be, limited temporal variation was incorporated into the case studies. This was done by assuming a spraying date for both compounds of May 1st, which was within the distribution of spraying times for insecticides and fungicides in Britain (Roelofs et al., 2005). Onset of the п¬Ѓrst reproductive phase (Table 6) for skylarks and wood mice in arable п¬Ѓelds was assumed to be two days after and 10 days before spraying, respectively. This was consistent with data from empirical studies on the timing of breeding for these species (Donald et al., 2001; Tattersall and Macdonald, 2003). The relevant data on breeding biology that were used to estimate the timing of onset of each subsequent reproductive phase are given in Roelofs et al. (2005). The eп¬Ђect of incorporating a breeding calendar and a spraying date of May 1st was to reduce estimated exposure for most reproductive phases (Table 7). For the skylark, the ETEs for each phase were approximately 70% of the values calculated in Table 6, although the reduction was substantially greater in some cases. The phasespeciп¬Ѓc TERs were elevated when they were based on the revised ETEs (Table 7). Despite this, the overall assessment of risk was the same as when no calendar was applied (Table 6), in that only the unadjusted TERs for 5-day toxicity to chick and п¬‚edgling exceeded their trigger value of п¬Ѓve. However, in both these cases, the adjusted TERs also exceeded their trigger value of one (Table 7). When the calendar was applied in the wood mouse case study, the eп¬Ђects on the ETE were marked (Table 7). Exposure during the п¬Ѓrst reproductive phase was zero because this phase preceded the spraying date. The ETEs during subsequent reproductive phases were reduced to an average of 29% (range 1вЂ“66%) of their values in Table 6, Body weight of 4-week-old juveniles in 2-generation test No/weanlings per mated F1 female in 2-generation test Teratological eп¬Ђects in 2-generation test 28- or 90-day toxicity tests Behavioural observations in 2-generation test No/pups per mated female in 2-generation test Teratological eп¬Ђects in 2-generation test or prenatal development toxicity test 28-day or 90-day toxicity tests No/weanlings per mated female in 2-generation test Survival to 4 weeks in 2-generation test 5 5 5 11 TWA TWA TWA TWA No/eggs laid/hen Mean eggshell thickness/hen Proportion of fertile eggs/eggs set/hen Change in adult BW Proportion hatching/eggs set/hen Change in adult BW 5-day dietary toxicity test with juveniles Proportion of 14-day juveniles per no/hatchlings per hen 5-day dietary toxicity test with juvenilesb 14-day-old juvenile weights per hen over over over over days days days days 5вЂ“7 22вЂ“23 22вЂ“26 5вЂ“7 Day 65 after 2nd spray 14 day TWA beginning Day 24 after second spray Day 6 after 2nd spray )10 to )7 )10 to )7 Day 0 of 2nd spray Day 0 of 2nd spray TWA over days 40вЂ“44 TWA over days 5вЂ“7 2 Days after sprayinga Change in adult BW prior to egg laying NOEL used as surrogate 50.6 337 0.089 28.0 93.4 4.29 17.2 NA NA 11.4 3.79 705 0.036 0.020 0.034 0.034 4.09 0.038 9.96 79.1 0.022 2.03 TER 0.089 0.964 0.964 5.22 5.22 0 0 7.91 7.91 0.014 6.93 7.77 7.77 7.77 4.89 6.93 2.01 0.120 6.93 9.87 Species-speciп¬Ѓc ETE (mg/kg BW/d) 22.6 3.39 1.88 1.57 0.289 1.21 NA NA 0.796 0.254 21.4 0.009 0.004 0.009 0.009 0.124 0.010 0.30 2.40 0.004 0.062 Adjusted TER a For the skylark, the breeding cycle was assumed to follow this calendar: arrival and pairing, May 3вЂ“5; copulation May 6; egg development (3 days) May 6вЂ“12, Egg-laying (4 eggs) May 9вЂ“12; incubation May12вЂ“22; nestling May 23вЂ“30, post-nestling May 31вЂ“June 9; п¬‚edging 10 June. For the wood mouse the breeding calendar was: arrival and pairing April 21вЂ“24, copulation April 25, pregnancy April 26вЂ“13 May; lactation 14вЂ“31 May; post-weaning 1 JuneвЂ“11 July; mature 12 July. b Juveniles assumed to weigh 30 g, with DEE of 110 kJ (see Box 2), and wholly insectivorous. Reproduction of F1 generation Post-weaning survival until maturity Pup growth and survival until weaning Wood mouse Establishing breeding site, pairing, and mating Pregnancy Post-п¬‚edging survival Juvenile growth and survival until п¬‚edging Incubation and hatching Skylark Pair formation/breeding site selection Copulation and egg laying Breeding phase Table 7. Phase-speciп¬Ѓc TERs calculated for the skylark and the wood mouse assuming that п¬Ѓrst application of pesticide is on May 1st and timing of reproduction occurs as outlined by Roelofs et al. (2005). Duration of exposure for each reproductive phase and the adjusted and non-adjusted NOEL values used to calculate the TER are indicated in Tables 4 and 5 Long-Term Toxicity to Exposure Ratios (TERs) for Birds and Mammals 889 890 Shore et al. except during pregnancy when exposure was maximal because it coincided with the second fungicide spray. The associated eп¬Ђects on the various TERs for wood mice were that eight unadjusted values were greater than п¬Ѓve and the remaining two were higher than three. The magnitude of the reduction in estimated exposure in wood mice was such that all but three of the adjusted TERs also exceeded their trigger value. It is clear from these two case studies that the eп¬Ђects of applying a calendar to the phase-speciп¬Ѓc toxicity endpoints can be pronounced, but vary substantially for diп¬Ђerent compounds or species. Discussion This paper used case studies to demonstrate how deterministic long-term risk assessments can be enhanced by use of multiple toxicity endpoints and reп¬Ѓned estimates of exposure. The studies progressed from the current EU п¬Ѓrst-tier assessment to incorporate phase-speciп¬Ѓc end-points, adjustment of the ETE so that it applied to speciп¬Ѓc (skylark, wood mouse) rather than generic indicator species, and then further reп¬Ѓnement of the ETE so that timing of pesticide application was related to onset of breeding. The other major aspect incorporated into the case studies was the replacement of unadjusted NOELs with adjusted values that accounted for species variation in sensitivity to toxic insult. Although there were only two case studies, they provided evidence of some of the major impacts these modiп¬Ѓcations can have on assessment of risk. In theory, the use of multiple rather than single toxicity endpoints should not per se aп¬Ђect the outcome of a long-term risk assessment. This is because the current п¬Ѓrst-tier TERlt uses the lowest of the various NOELs in the phase-speciп¬Ѓc assessments. The only exception is the use of the 5day dietary toxicity test to assess impacts on growth and survival of chicks and п¬‚edglings. This is not assessed in the avian reproduction test, because chicks are reared on uncontaminated diet (Mineau, 2005). Thus, a compound with high dietary toxicity to chicks could fall below the trigger value for a phase-speciп¬Ѓc assessment but exceed the trigger value for a standard п¬Ѓrst-tier TERlt. In practice however, use of phase-speciп¬Ѓc endpoints requires concurrent reduction of the time period used in the ftwa from the default 21days to values that range from one to 14 days (Tables 4 and 5). The resultant ftwa will normally be greater than the default value unless there is concentration, or at least no degradation, of dietary residues over time. Typically therefore, the lowest of the NOELs in the phase-speciп¬Ѓc assessments will be evaluated against a higher ETE than in the current п¬Ѓrst tier assessments, although the diп¬Ђerence in the case studies was, at most, only twofold. Consequently, the phase-speciп¬Ѓc risk assessment is likely to be somewhat more stringent than the current п¬Ѓrst-tier assessment if there is no subsequent further reп¬Ѓnement of the ETE. Adjustment of the exposure term so that it applies to individual species, rather than generic indicators, might have been expected to reduce the ETE values in the case studies, not least because PT was decreased by between a half and twothirds. However, it was apparent that replacing generic default values with species-speciп¬Ѓc exposure parameters can have complex impacts and, in some cases, negate the inп¬‚uence of even a relatively large reduction in PT. The reп¬Ѓned ETE for the skylark was, in fact, higher than that for the generic herbivorous and insectivorous bird (Tables 4 versus 6), demonstrating that the п¬Ѓrst-tier assessment is not always particularly conservative. Although omnivorous, skylark diets can contain substantial amounts of leaf material, and being relatively small birds, they may be more at risk than larger obligate herbivores. In contrast, the mammal case study indicated that the estimated exposure for wood mice was similar to that predicted for the generic insectivore (shrew) but much lower than that predicted for the generic herbivorous vole (Tables 5 versus 6). Intuitively, it might be thought reasonable to use the generic vole (but not the shrew) as a surrogate for wood mice as both are rodents, have similar body weights and some dietary overlap (Hansson, 1985; Corbet and Harris, 1991). However, the case study data suggests that if this is done, exposure in wood mice is markedly overestimated. In reality, most of the ETE values calculated without the implementation of a calendar are likely to be overestimated because they are assumed to be close to, or at, a maximum during each reproductive phase. Although the reduction Long-Term Toxicity to Exposure Ratios (TERs) for Birds and Mammals 891 in ETE estimates associated with use of a breeding calendar and spraying date did not raise many skylark TERs above the trigger value, most wood mouse chronological TERs did exceed the trigger value. The only exceptions were for eп¬Ђects on pregnancy and direct toxicity to pups and were due to a high level of exposure following the second fungicide application (Table 7). The chronological TERs therefore indicated a potential to reduce the predicted risk by limiting fungicide use to a single application. The simple п¬Ѓrst-tier TERlt values (Table 3) do not provide such insight. Even when the TERlt values were recalculated assuming a single fungicide application, the resultant values (1.87 for the generic insectivore, 0.107 for the small herbivore) fell below the trigger value. Although chronological TERs may provide a more realistic assessment of risk, they are deterministic values and, as such, problematic. This is because they are based on one single speciп¬Ѓc date for pesticide application and another for uniform onset and procession of breeding. The calculated TERs are likely to vary markedly with choice of dates. For example, the coincidence of the second fungicide spraying with pregnancy in the wood mouse case study could disappear if a diп¬Ђerent calendar was used. In reality, spraying date(s) and the dates for onset of each reproductive phase all vary. Chronological phase-speciп¬Ѓc assessments therefore demand a probabilistic approach that incorporates information about the variation in these parameters and how they interact with each other. Such an approach would also allow incorporation of data on the distributions of other parameters in the model (food consumption, energy expenditure, pesticide residues, species sensitivity, etc), all of which also vary, although to diп¬Ђering degrees and not all contribute equally to uncertainty in the risk assessment. The type of model needed to produce this type of probabilistic assessment and the nature of the resultant outputs are beyond the scope of this paper but are considered by Roelofs et al. (2005). The п¬Ѓnal major aspect of the long-term risk assessment that was investigated in the case studies was to examine the eп¬Ђect of using adjusted rather than unadjusted NOELs to calculate TERs. Adjusted NOELs are usually, but not always, lower than unadjusted values (Luttik et al., 2005). This was true in the case studies, and replacement of unadjusted NOELs with adjusted values increased the proportion of phase-speciп¬Ѓc TERs that fell below the trigger value, even though this was relaxed from п¬Ѓve to one. Thus, the uncertainty factor of п¬Ѓve applied to the unadjusted TERlt would seem insuп¬ѓcient to account for species variation in sensitivity to pesticides, let alone other sources of uncertainty. This suggests again that the current п¬Ѓrst-tier TERlt assessment may not be suп¬ѓciently protective, but in this case because toxicity, rather than exposure, is underestimated. Table 8. Decision tree for assessing the long-term risk to vertebrates from novel compounds Criteria if NOELs are Stage Description Unadjusted 1 Standard п¬Ѓrst tier TERlt 2 Phase speciп¬Ѓc TERs for generic indicator speciesa 3 Phase speciп¬Ѓc TERs for particular indicator speciesb 4 Deterministic chronological TERsc TERlt TERlt TERlt TERlt TERlt TERlt TERlt TERlt 5 Probabilistic assessment of risk and potential population eп¬Ђectsd a вЂЎ5 <5 вЂЎ5 <5 вЂЎ5 <5 вЂЎ5 <5 Adjusted TERlt TERlt TERlt TERlt TERlt TERlt TERlt TERlt вЂЎ1 <1 вЂЎ1 <1 вЂЎ1 <1 вЂЎ1 <1 Next stage Stop To stage 2 Stop Stage 3 Stop Stage 4 Stop Stage 5 Calculated using TWA factor appropriate to the speciп¬Ѓc toxicity endpoint. If skylark and wood mouse are used as speciп¬Ѓc indicators, the required ecological data to reп¬Ѓne the ETE are summarised by Roelofs et al. (2005). c Calculated using a single spraying date consistent with the proposed use of the compound and a single date for onset of breeding for indicator species. d As described by Roelofs et al. (2005). b 892 Shore et al. The original brief of the risk assessment workshop was to remain close to the current EU guidance for long-term risk assessment. It is therefore pertinent to ask how the approaches outlined here can be readily implemented. A decision tree of rules was suggested (Table 8) that followed the assessment sequence outlined in the present paper. It was proposed that phase-speciп¬Ѓc assessments should be carried out only for compounds that did not have п¬Ѓrst-tier TERlt values of п¬Ѓve or more. This conforms to current EU guidance but the case studies have shown that the п¬Ѓrst-tier assessment may under-estimate risk in some cases. One solution would be to increase the conservatism of standard п¬Ѓrst-tier assessments. Alternatively, it might be argued that all compounds should be subject to both the п¬Ѓrsttier TERlt and the subsequent assessment stages in Table 8, thereby making more use of the available information. Unadjusted or adjusted NOELs (following Luttik et al., 2005) could be used. The procession from deterministic chronological TERs to full probabilistic assessment (Stage 4 to 5 in Table 8) is a large and complex step and application of uncertainty bounds analysis to the chronological TERs may be merited п¬Ѓrst. This would assess the likely range of variation in the chronological TER values and help determine if a full probabilistic assessment was necessary. In conclusion, this paper has shown how the toxicity data required under current registration procedures can be used to produce more realistic assessments of long-term risk to birds and mammals. Implementation of staged reп¬Ѓnements to the deterministic assessment is currently possible using existing registration data for the compound and available ecological data for indicator species. The advantages of using such phase-speciп¬Ѓc assessments are that they potentially give a more accurate assessment of risk and identify the most vulnerable phase(s) of reproduction. This information can further be used to formulate riskreduction measures, as demonstrated in the wood mouse case study, or to target data-gathering needed for reп¬Ѓning the assessment. Phase-speciп¬Ѓc assessments are also a necessary prerequisite for developing methods to predict impacts at the population level. Acknowledgments We acknowledge the UK Department for Food and Rural Affairs and its Pesticide Safety Directorate for funding the workshop on long-term risk assessment. References Bennett, R.S., Dewhurst, I., Fairbrother, A., Hooper, M., Leopold, A., Mineau, P., Mortensen, S., Shore, R.F. and Springer, T. (2005). A new interpretation of avian and mammalian reproduction toxicity test data in ecological risk assessment. Ecotoxicology, this volume. Calow, P. (1998). Ecological risk assessment: risk for what? How do we decide? Ecotoxicol. Environ. Saf. 40, 15вЂ“8. Corbet, G.B. and Harris, S. (1991). The Handbook of British Mammals. Third Blackwell Scientiп¬Ѓc Publications, Oxford. Crocker, D.R., Hart, A., Gurney, J. and McCoy, C. (2002a). Project PN0908: Methods for estimating daily food intake of wild birds and mammals. Central Science Laboratory п¬Ѓnal report to Defra. Crocker, D.R., Prosser, P., Irving, P.V., Bone, P. and Hart, A. (2002b). Estimating avian exposure to pesticides on arable crops. In N.D. Boatman, N. Carter, A.D. Evans, P.V. Grice, C. Stoate and J.D. Wilson (eds). Birds and Agriculture, pp. 237вЂ“44. Wellesbourne: Association of Applied Biologists. Crocker, D.R. (2005). Estimating the exposure of birds and mammals to pesticides in long-term risk assessments. Ecotoxicology, this volume. Donald, P.F., Buckingham, D.L., Moorcroft, D., Muirhead, L.B., Evans, A.D. and Kirby, W.B. (2001). Habitat use and diet of skylarks Alauda arvensis wintering on lowland farmland in southern Britain. J. Appl. Ecol. 38, 536вЂ“47. Donald, P.F. (2004). The Skylark. Poyser, London. European Commission (2002). Guidance document on risk assessment for birds and mammals under council directive 91/414/EEC, SANCO/4145/2000. Fischer, D. (2005). Accounting for diп¬Ђering exposure patterns between laboratory test and the п¬Ѓeld in the assessment of long-term risks of pesticides to terrestrial vertebrates. Ecotoxicology, this volume. Garthwaite, D.G., Thomas, M.R., Dawson, A. and Stoddart, H. (2003). Pesticide Usage Survey Report 187: Arable Crops in Great Britain 2002. Defra Publications, London. Gorman, M.L. and Zubaid, A.M.A. (1993). A comparative study of the ecology of woodmice Apodemus sylvaticus in two contrasting habitats: deciduous woodland and maritime sand-dunes. J. Zool. (Lond.) 229, 385вЂ“96. Green, R.E. (1978). Factors aп¬Ђecting the diet of farmland skylarks Alauda arvensis. J. Anim. Ecol. 47, 913вЂ“28. Hansson, L. (1985). The food of bank voles, wood mice and yellow-necked mice. Symp. Zool. Soc. Lond. 55, 141вЂ“68. Hart, A.D.M. and Thompson, H.M. (2005). Introduction to workshop. Ecotoxicology, this volume. Long-Term Toxicity to Exposure Ratios (TERs) for Birds and Mammals 893 Luttik, R., Mineau, P. and Roelofs W. (2005). A review of interspecies extrapolation in birds and mammals and a proposal for long-term toxicity data. Ecotoxicology, this volume. McKay, H.V., Prosser, P.J., Hart, A.D.M., Langton, S.D., Jones, A., McCoy, C., Chandler-Morris, S.A. and Pascual, J.A. (1999). Do wood-pigeons avoid pesticide-treated cereal seed? J. Appl. Ecol. 36, 283вЂ“96. Mineau, P., Hooper, M., Elmegaard, N., Grau, R., Luttik, R. and Ringer, B. (2001). Case Study 5: foliar insecticide II. In A.D.M. Hart, D. Balluп¬Ђ, R. Barfknecht, P.F. Chapman, T. Hawkes, G. Joermann, A. Leopold and R. Luttik (eds). Avian Eп¬Ђects Assessment: A Framework for Contaminants Studies, pp. 113вЂ“36. Pensacola, FL: Society of Environmental Toxicology and Chemistry (SETAC). Mineau, P. (2002). Estimating the probability of bird mortality from pesticide sprays on the basis of the п¬Ѓeld study record. Environ. Toxicol. Chem. 21, 1497вЂ“506. Mineau, P. (2005). A review and analysis of study endpoints relevant to the assessment of вЂ�long termвЂ™ pesticide toxicity in avian and mammalian wildlife. Ecotoxicology, this volume. OECD (1998). Organisation for Economic Co-operation and Development (OECD) guidelines for the testing of chemicals (up to and including tenth addendum) Version consulted for guidelines 206, 401, 402, 407, 408, 409, 410, 411, 412, 413, 415. Paris, OECD. Poulsen, J.G., Sotherton, N.W. and Aebischer, N.J. (1998). Comparative nesting and feeding ecology of skylarks Alauda arvensis on arable farmland in southern England with special reference to set-aside. J. Appl. Ecol. 35, 131вЂ“47. Roelofs, W., Crocker, D.R., Shore, R.F., Topping, C., Akcakaya, H.R., Bennett, R.S., Chapman, P.F., Clook, M., Crane, M., Dewhurst, I.C., Edwards, P.J., Fairbrother, A., Ferson, S., Fischer, D., Hart, A.D.M., Holmes, M., Hooper, M.J., Lavine, M., Leopold, A., Luttik, R., Mineau, P., Moore, D.R.J., Mortenson, S.R., Noble, D.G., OвЂ™Connor, R.J., Sibly, R.M., Smith, G., Spendiп¬Ђ, M., Springer, T.A. and Thompson, H.M. (2005). Case Study Part 2: Probabilistic modelling of long-term eп¬Ђects of pesticides on individual breeding success in birds and mammals. Ecotoxicology, this volume. Rogers, L.M. and Gorman, M.L. (1995). The diet of the wood mouse Apodemus sylvaticus on set-aside land. J. Zool. 235, 77вЂ“83. Sibley, R.M., AkcВёakaya, H.R., Topping, C.J. and OвЂ™Connor, R.J. (2005). Population-level assessment of risks of pesticides to birds and mammals in the UK. Ecotoxicology, this volume. Solomon, K., Giesy, J. and Jones, P. (2000). Probabilistic risk assessment of agrochemicals in the environment. Crop Protection 19, 649вЂ“55. Tattersall, F.H. and Macdonald, D.W. (2003). Wood mice in the arable system. In F. Tattersall and W. Manley (eds). Conservation and Conп¬‚ict: Mammals and Farming in Britain, pp. 82вЂ“96. London: Occasional Publication of the Linnean Society, The Linnean Society. Tiebout, H.M. and Brugger, K.E. (1995). Ecological risk assessment of pesticides for terrestrial vertebrates: Evaluation and application of the U.S. Environmental Protection AgencyвЂ™s quotient model. Conserv. Biol. 9, 1605вЂ“18. Watts, C.H.S. (1968). The foods eaten by wood mice (Apodemus sylvaticus) and bank voles (Clethrionomys glareolus) in Wytham Woods, Berkshire. J. Anim. Ecol. 37, 25вЂ“41.