International Office for Water Review of the list of priority substances (Decision 2455/2001/EC) Testing robustness and limits of the prioritisation methodology applied for the 1st stage ranking exercise of substances under the WFD & Possible answers to comments WG E (5), Brussels, 18 March 2009
2 Presentation based on new documents... “Prioritisation process: compilation of post WG-E (4) comments” (James, 2009) “Testing robustness and limits of the prioritisation methodology applied for 1st stage ranking exercise” (James et al., 2009)...and comments made for WG E 4
Contents I.PEC issues : 1.Quality of the data : identification and treatment of outliers 2.Reminder on the calculation of PECs 3.Representativeness of monitoring data 4.Calculation of PEC values with 90 th percentile versus 50 th percentile 5.Metals issue : calculation of risk ratios based on PEC total expressed as dissolved using the equilibrium partitioning approach 6.Issue of PEC dissolved versus PEC total 7.PECs substances specific issues II.PNEC issues 1.PNECs general issues 2.PNECs substance specific issues 3
4 Quality of the data : identification and treatment of outliers Aim : Discard poor analytical results i.e. measurements with associated DL values that are > a threshold value Threshold value = 99 th percentile of all DL values for a given substance and a given analytical fraction (provided that min(DL) max(DL)) Distribution of DL values for a given substance and a given analytical fraction See Fribourg-Blanc (2008)
Member states 90th centile 95th centile 99th centile AT600 BE651 CH111 CY11103 CZ731 DE1073 DK211 EE100 ES000 FI300 FR20157 HU15138 IE330 IT13115 LT15108 LU LV000 NL500 PL PT22166 RO18136 SE210 SI621 SK1742 UK321 Mean % of discarded values Quality of the data : identification and treatment of outliers Application of various threshold values (99 th,95 th and 90 th percentile) to estimate impact of such a choice Almost 10% of the dataset is discarded when using the 90 th percentile as the threshold value Keep 99th percentile of all DL values threshold which leads to the discard a reasonable percentage of data Proposal :
6 Arithmetic mean 1 90 th percentile of all measures of all stations for : - A given substance -A given analytical fraction Raw data = all measures for : - A given substance - A given analytical fraction - A given station - At a given time station 2 station 1 station 3 Arithmetic mean 3 Arithmetic mean 2 Arithmetic means of all measures for : - A given substance - A given analytical fraction - A given station EUROPEAN PECs PECwater PECsed, 2mm PECsed, 20µm PECsed, 63µm PECbiota, fish PECbiota, invertebrates Reminder : calculation of PECs See Fribourg-Blanc (2008) Selection of manageable list = substances monitored by more than 3 countries
7 Representativeness of monitoring data Representativeness of calculated PECs by year INERIS mentions : >2000 = after COMMPS 2006 and 2007 data recognised to be more relevant in the context of WFD Heterogeneity of monitoring data between years hampers the use of 2006 and 2007 data only Comments from : NL (WG E 5) : suggestion to check if differences between PEC values and if yes, use 2006 and 2007 data only Figure 14 (page 29) of report « The central database … » (Fribourg-Blanc, 2008)
8 Representativeness of monitoring data Representativeness of calculated PECs by country Heterogeneity of monitoring data : In time ( ) In space (country, river basin, stations) Between matrix (water, sediment, biota) Between fractions (whole water vs dissolved phase, sediment 2mm vs 20µm vs 63 µm, fish vs invertebrates) Comments from : WG E 4 : DE, CEFIC, metals IND WG E 5 : NL, metals IND
9 Representativeness of monitoring data Representativeness of calculated PECs by country Nb of countries which supplied monitoring data Large variations between countries in the number of submitted analyses Ex. above : substances ranked as « Very high » and « High » priority (1 st stage ranking exercise, 2008)
10 Representativeness of monitoring data Representativeness of calculated PECs by station and by country % of stations per country contributing to PEC calculation PEC value can be highly influenced by the monitoring data supplied by a limited number of countries Ex. above : substances ranked as « Very high » and « High » priority (1 st stage ranking exercise, 2008)
11 Representativeness of monitoring data Representativeness of calculated PECs by station and by country % of stations per country contributing to PEC calculation PEC value can be highly influenced by the monitoring data supplied by a limited number of countries Ex. above : substances ranked as « Very high » and « High » priority (1 st stage ranking exercise, 2008)
12 Representativeness of monitoring data Representativeness of calculated PECs by station and by country % of stations per country contributing to PEC calculation PEC value can be highly influenced by the monitoring data supplied by a limited number of countries Ex. above : substances ranked as « Very high » and « High » priority (1 st stage ranking exercise, 2008)
13 Representativeness of monitoring data Representativeness of calculated PECs by station and by country % of stations per country contributing to PEC calculation PEC value can be highly influenced by the monitoring data supplied by a limited number of countries Ex. above : substances ranked as « Very high » and « High » priority (1 st stage ranking exercise, 2008)
Nb of stations per river basin : example of Arsenic and its compounds 14 Representativeness of monitoring data Representativeness of calculated PECs by station and by river basin High level of heterogeneity in the number of stations contributing to PEC within a river basin UK AT
15 Representativeness of monitoring data Proposal for another PEC calculation level for next ranking exercise 90 th percentiles of all measures of all stations for : - A given substance -A given analytical fraction -A given river basin Arithmetic mean 1 Raw data = all measures for : - A given substance - A given analytical fraction - A given station - At a given time Arithmetic mean 3 Arithmetic mean 2 Arithmetic means of all measures for : - A given substance - A given analytical fraction - A given statioN 90 th percentiles of all measures of all stations of all river basins for : - A given substance -A given analytical fraction To be discussed EUROPEAN PECs PECwater PECsed, 2mm PECsed, 20µm PECsed, 63µm PECbiota, fish PECbiota, invertebrates Selection of manageable list = substances monitored by more than 3 countries
16 Representativeness of monitoring data Influence of the matrix and fraction analyzed on the representativeness of calculated PECs Ex. above : substances ranked as « Very high » and « High » priority (1 st stage ranking exercise, 2008) PEC value can be highly influenced by the monitoring data supplied by a limited number of countries for a limited number of matrix and/or fraction Ex. of Arsenic and its compounds
Fraction level : Takes account of analytical issue but does not address representativeness of data Notable differences in monitoring practices between countries No opportunity for an additional requirement because data are too widely distributed Approach 1 st stage ranking : taking into account most stringent risk ratio within all level, i.e. fractions and matrices and between matrices 17 Representativeness of monitoring data No opportunity for additional requirements in prioritisation ranking process at fraction level Need for harmonisation in reporting Need for more careful reporting (template)
Matrix level : Does not address representativeness of data Notable differences in monitoring practices between countries 18 Representativeness of monitoring data Proposal for additional requirement in prioritisation ranking process at matrix level Opportunity for an additional requirement for definition of the actual manageable list (nb MS>3): number of country supplying data should be at least 2 within one matrix Nb of substances retained by application of the condition Condition on the nb of countries monitoring the substance in each matrix WaterSedimentBiota At least At least At least Approach 1 st stage ranking : taking into account most stringent risk ratio within all level, i.e. fractions and matrices and between matrices To be discussed
19 Arithmetic mean 1 90 th percentile of all measures of all stations and RB for : - A given substance -A given analytical fraction Raw data = all measures for : - A given substance - A given analytical fraction - A given station & river basin - At a given time station 2 station 1 station 3 Arithmetic mean 3 Arithmetic mean 2 Arithmetic means of all measures for : - A given substance - A given analytical fraction - A given station & river basin EUROPEAN PECs PECwater PECsed, 2mm PECsed, 20µm PECsed, 63µm PECbiota, fish PECbiota, invertebrates Reminder : calculation of PECs See Fribourg-Blanc (2008)
Calculation of PECs : 90 th vs 50 th percentile 20 Comments from CEFIC and the metals IND : The 50 th percentile of the distribution of the mean values should be calculated for a more accurate and reliable reflection of general EU water quality INERIS mentions that the approach of using the 90 th percentile : is consistent with COMMPS procedure (1999) is a conservative approach although it may tend to estimate risk based on a few stations Approach using the 50 th percentile of arithmetic means tested
Calculation of PECs : 90 th vs 50 th percentile 21 Results of the test : variations in the priority ranking of substances when using 50th percentile instead of 90th percentile OrganicsMetals Nb of substances % Nb of substances % Decreasing priority ranking No change in priority ranking Increasing priority ranking **11500 OrganicsMetals Number of substances classified asPEC 90th p. PEC 50th p. * PEC 90th p. *PEC 50th p. “Very high” “High”16440 “Medium” “Low” “Further investigation needed”16700 * Results are different from 1 st stage ranking because new IT data are included here Approach using the 90 th percentile of arithmetic means is more conservative Proposal * « Increasing » from « Further investigation needed » to « other ranking »
22 Metals issue : risk ratios based on PEC dissolved via EqP Comments from :CEFIC and the metals IND : WG E 4 : SK : EqP approach is a general and simplified approach UK, DK and ECPA : accepts the use of EqP approach but believes that further consideration for sediment approach is needed (e.g. availability of experimental data) CEFIC and metals IND : prioritisation should focus on water were more monitoring data is available. Sediment information should only be used as complementary to support water based conclusions EqP approach was used to derive PEC dissolved from PEC total in order to compare PEC and PNEC on the same basis WG E 5 : Metals IND : EqP approach should be used to convert PEC total into PEC dissolved in order to proceed to comparison of PEC and PNEC on the same basis within a tiered approach (as recommended in TGD (E.C., 2003) and REACH guidances)
23 Metals issue : risk ratios based on PEC dissolved via EqP Subst.CAS PNEC Water (µg/L) 90th p. µg/L) Kp* log Kp* PEC1 (µg.l -1 )PEC2 (µg.l -1 )PEC1/PNECPEC2/PNEC Nb Anal>DL / Nb all anal PEC2 / 90th p. BkgPEC1 / 90th p. Bkg Final Priority for WATER wWdW EqP wWdW EqP wWdW EqP wWdW EqP wWdW EqP wWdW EqP wWdW EqP wWdW EqP Ag ,0510,000,202,500,25200,004,0050,005,000,080,00FIV Low Al ,44112,50303,0051,2773,0922,5060,6010,250,850,391,450,251,750,54High As ,22,453,915,902,77 0,931,400,66 0,490,401,13 1,602,41Medium B ,5632,50169,12405,71161,242,900,781,860,740,630,814,291,716,691,79Medium Ba ,868,1245,8363,9848,961,170,791,100,84 0,980,830,640,890,60Medium Low Be ,0380,0560,372,500,109,820,0065,792,630,210,0044,641,796,660,00FIV Low Cd ,080, ,110,681,030,230,380,500,138,5512,942,904,696,251,590,250,130,257,089,432,4012,9119,534,38Medium Co ,280, ,598,421,795,323,151,101,9930,096,4119,0011,253,927,100,310,880,315,411,893,4214,473,089,14High Medium Cr ,41,44,889,002,353,811,442,650,691,120,290,241,682,723,496,43Medium Cu ,42, ,485,6418,603,885,0810,943,504,0313,292,773,637,812,500,910,350,912,084,461,432,307,591,58Medium Fe ,44149,29493,73141,001,800,501,650,470,800,550,660,190,730,20Medium Low Hg ,050,451,500,100,509,0030,002,0010,000,080,02Low Mo ,41, ,456,6024,806,335,0024,804,800,491,850,470,371,850,360,281,000,284,6723,184,486,1723,185,92Medium HighMedium Ni , ,428,617,746,174,966,913,560,430,390,310,250,350,180,480,860,481,051,460,751,821,641,31Medium Low Pb ,20, ,473,504,190,642,712,480,500,490,580,090,380,340,070,590,170,596,305,761,168,149,741,50Medium Sb , ,655,780,895,422,500,642,340,050,010,050,020,010,020,210,260,2111,903,0311,1627,514,2625,79High MediumHigh Se ,951,16,000,980,942,506,321,030,992,630,020,280,852,275,450,89Low Sn ,516,905,0011,270,003,330,000,09Low V ,51,6610,00 4,00 0,106,02 Low Zn ,110, ,0441,8070,0015,7729,4950,0011,1313,4822,585,099,5116,133,590,650,240,652,894,901,094,106,861,55Medium HighMedium * proposed by EUROMETAUX Table 5 of Discussion paper (James et al., 2009)
24 Metals issue : risk ratios based on PEC dissolved via EqP PEC1/PNECPEC2/PNEC PEC2 / 90th perc. Bkg PEC1 / 90th perc. Bkg Final Priority Sub st. CASKp log Kp wWdW EqP wWdW EqP wWdW EqP wWdW EqP wWdW EqP Cd ,118,5512,942,904,696,251,597,089,432,4012,9119,534,38 Medium Co ,5930,096,4119,0011,253,927,105,411,893,4214,473,089,14 HighMedium Cu ,484,0313,292,773,637,812,502,084,461,432,307,591,58 Medium Mo ,450,491,850,470,371,850,364,6723,184,486,1723,185,92 MediumHighMedium Ni ,420,430,390,310,250,350,181,051,460,751,821,641,31 Medium Low Pb ,470,490,580,090,380,340,076,305,761,168,149,741,50 Medium Sb ,650,050,010,050,020,010,0211,903,0311,1627,514,2625,79 HighMediumHigh Zn ,0413,4822,585,099,5116,133,592,894,901,094,106,861,55 MediumHighMedium WARNING !!! PNEC to be discussed / reviewed for next ranking exercise To be discussed
25 Issue of PEC dissolved versus PEC total Comments from many stakeholders about PECs for metals : PEC dissolved > PEC total WARNING !!! This is not due to a bias/error in the calculation of PECs
Issue of PEC1 dissolved versus PEC1 total Example of Zn on % monitoring data : dissolved vs total
Issue of PEC1 dissolved versus PEC1 total Example of Zn on PEC1 values : dissolved vs total Comparison between member states
Issue of PEC1 dissolved versus PEC1 total Example of Cd on % monitoring data : dissolved vs total
Issue of PEC1dissolved versus PEC1total Example of Cd, Ni and Cr on PEC1 values : dissolved vs total Comparison between member states
30 PECs Substances specific issues The data collection aims at browsing all monitoring data in Europe, in particular to derive European PECs Despite limits of the exercise, taking into consideration other PEC values is not relevant at this stage
31 PNEC General issues : cancer risk Comments from NL (WG E 5) : PNEC or ADI/TDI via oral route and risk assessment in biota : ADI used in first stage ranking corresopnd to a cancer risk of A lifetime risk of 10-6 should be used instead Which thresholds is considered a tolerable risk level ? = political issue INERIS mentions : According to REACH guidance on information requirements and chemical safety assessment – Ch. 8 : Characterisation of concentrations-response for human health : cancer risk levels of and = indicative tolerable risk levels when setting thresholds of no effect for general population To be discussed
32 PNEC General issues : metabolites Comments from NL (WG E 5) : PNECs should be derived for metabolites and parent compound Parent compounds : comparing peak measurements with MAC-EQS ? INERIS mentions : PNEC are not always available for metabolites and parent compounds no PNEC metabolites -> metabolites have been applied PNEC parent compound no PNEC parent compounds + metabolites more toxic -> parent compounds have been applied PNEC metabolites To be discussed sum PEC for parent compound and metabolites in order to compare them to a unique PNEC ?
33 PNEC Substances specific issues (1) Substances for which a Risk Assessment Report (RAR) finalised/validated at the European level is available : Pesticides : PNEC values proposed* will be included in the next prioritisation ranking exercise provided they are not derived from ecotoxicity data taking account of recovery Metals : PNEC values derived according to a methodology relevant for risk assessment in the context of WFD prioritisation ranking exercise will be included in the next prioritisation ranking exercise provided they are not derived from ecotoxicity data taking account of recovery Substances : Copper and Zinc * PNEC sediment for Pendimethalin and alpha-Cypermthrin will be calculated applying the additional assessment factor of 10
34 PNEC Substances specific issues (2) Substances for which an EU-RAR is not finalised/not validated Substances for which no RAR is available but other “thresholds of no effect” have been derived, e.g. Ambient Water Quality Criteria (Environment Canada) Revision of PNEC values adopted for first prioritisation ranking process might be reviewed for the next prioritisation ranking exercise Pesticides : Prochloraz Metals : Aluminum, Antimony, Boron and Molybdenum needs to be addressed/validated by experts consortium If not, INERIS will decide itself what PNEC value is deemed relevant