2nd DRAFT Report for FKZ Umweltbundesamt II 2.1 A.Z /308

Transcription

1 2nd DRAFT Report for FKZ Umweltbundesamt II 2.1 A.Z /308 Sensitivity analysis of existing concepts for application of biotic ligand models (BLM) for the derivation and application of environmental quality standards for metals and evaluation of the approaches with appropriate monitoring data sets from German waters Sensitivitätsanalyse der vorliegenden Konzepte zur Anwendung des Bioligandenmodells (BLM) für die Ableitung von Umweltqualitätszielen von Nickel, Zink, Blei und Cadmium sowie Evaluierung der Ansätze mit geeigneten Monitoringdaten für D Fraunhofer Institute for Molecular Biology and Applied Ecology (IME) Division Applied Ecology Schmallenberg, Germany Head of Institute: Prof. Dr. Rainer Fischer Head of Division: Dr. Christoph Schäfers Authors: Dr. Udo Hommen Phone: udo.hommen@ime.fraunhofer.de Dr. Heinz Rüdel Phone: heinz.ruedel@ime.fraunhofer.de Schmallenberg,

2 Abstract The aim of this paper is to support the discussion on the potential implementation of the userfriendly BLM tool (biomet) in the routine monitoring programs of the water authorities assessing the chemical quality of water bodies in Germany under the European Water Framework Directive (WFD) with respect to Cu, Ni and Zn. In particular the plausibility of the BLM-derived quality standards (QS), the practicability of the BLM approach, and the suitability for implementation should be discussed. The BLM approach in general is considered scientifically valid and accepted to consider dependence of metal bioavailability on water conditions and to achieve a more realistic assessment of the potential risks of metals toward the pelagic community. Although the plausibility of the QS calculated by the BLM tool has been checked by comparing the outputs with results of the full BLM, the analysis here has shown that this should be done in more detail. Sensitivity analysis revealed that especially the DOC has a considerable effect on the bioavailability. In example data sets from Germany up to 40 % of the samples showed ph or Ca concentrations above the upper validity boundary. Thus, handling samples with ph and Ca concentrations outside the validity range of the BLM have to be more clearly specified. The costs for implementation due to the need of additional measurements, e.g. dissolved metal concentrations and DOC, seem to be acceptable considering the benefits of a more realistic assessment of the potential hazards and the avoidance of unnecessary management and regulation. For implementation, the tool has to be refined and fully tested and documented. Guidance or harmonized approaches to deal with water conditions outside the validity range of the BLM has to be developed. Accumulation of metals in sediment or biota as well as potential effects of high concentrations measured in single samples have to be taken into account additionally. For emission and immission balances the measurements of total metal concentrations in water are considered still necessary. Zusammenfassung Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 2

3 Executive summary Background The toxicity of metals to aquatic organisms is strongly depending on water parameters, i.e. ph, hardness and DOC (dissolved organic carbon). Until now national quality standards (QS) in Germany for metals have been set for total metal concentrations and (at least for some metals) were related to suspended particular matter (SPM). Local water conditions affecting bioavailability and thus, toxicity are not taken into account yet. The current scientific state of the art approach to consider bioavailability in deriving ecotoxicological threshold concentrations such as Predicted No Effect Concentrations (PNEC) or QS is the use of Biotic Ligand Models (BLM). Derived from toxicity tests conducted for a variety of water conditions and incorporating information on metal speciation, BLM provide site and species-specific estimations of toxicity. BLM for chronic toxicity to at least one algae, crustacean and fish species, respectively, are available for Ni, Cu and Zn and have been applied in risk assessment reports (RAR) for these metals. For other metals, BLM are under development. However, derivation of site specific quality standards using the full BLM approach is complex and therefore not practicable in operative monitoring. Therefore, a user-friendly BLM tool based on look-up tables from results of the full BLM for Cu, Ni and Zn was presented during a recent EU workshop (21. June 2011) 1. This tool was the basis to discuss the advantages, obstacles and concerns related to the possible implementation of the BLM approach for these three metals in Germany. Objectives The aim of this opinion paper is to support the discussion on the potential benefits and drawbacks of an implementation of the user-friendly BLM tool in the routine monitoring programs with respect to metals. In particular the plausibility of the BLM-derived QS, the practicaility of the BLM approach, and the suitability for implementation should be discussed. Methods The project was structured into four work packages: 1. Description of status quo and potential benefits of the use of BLM. 2. Quality control and evaluation of monitoring results with respect to QS compliance. 3. Discussion of BLM application regarding emission / immission balances. 4. Cost benefit analysis of the necessary additionally measurements and calculations. 1 End of November 2011 a new version was launched at the website (user guide: Biomet 2011). However, although it was not possible to repeat all calculations with the new version at least a comparison between both versions of the sensitivity regarding ph, DOC, and Ca concentrations was conducted. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 3

4 This opinion is based on a review of the relevant literature (i.e. the report on the EU-workshop, EU risk assessment reports on metals where the BLM approach has been used yet, etc.), calculations derived from the application of the user-friendly BLM tool (sensitivity analysis, quality standard calculations for German example datasets), and interviews with representatives of relevant authorities in the German federal states. Results 1. Status quo and validity of the BLM tool: The BLM approach in general is scientifically accepted and recommended in recent European guidance documents related to chemical risk assessment and QS setting. BLM are metal- and species-specific. The full BLM for copper (Cu), nickel (Ni), and zinc (Zn) and their application to calculate site-specific species sensitivity distributions (SSD) by species to species extrapolation have been reviewed and finally accepted in the EU risk assessment reports. Thus, their validity was not further analysed here. The user friendly BLM tool calculates site-specific quality standards for Cu, Ni and Zn (expressed as dissolved metal concentrations) for the combination of site-specific ph as well as Ca and DOC concentrations. In addition, the bioavailable fraction is calculated from measured dissolved metal concentrations. Thus, the user friendly BLM tool focuses on quality standards for water and does not consider concentrations in sediment, suspended particulate matter or biota. Therefore the calculated QS correspond to the QS freshwater, eco according to the terminology of the WFD guidance document, while the so-called EQS (environmental QS) should be derived as the overall QS considering also benthic organisms, top-predators or human health. Thus it has to be decided case by case (metal) if other compartments or other protection goals are more relevant (i.e. sensitive) for EQS setting. The tool is based on look-up tables built from results of the full BLM (hazardous concentrations for 5 % of the species, HC5, from SSD with NOECs or EC10 from chronic tests).the development of the user-friendly BLM tool is described on the website To validate the tool, results of the tool were compared to the results of the full BLM for hundreds of sites in the UK and The Netherlands. For the UK sites, 84, 99 and 80 % of the estimations by the tool for Cu, Ni, and Zn, respectively, where conservative within a factor of 2. For the sites in the Netherlands 69, 59 and 87 %, respectively, of the estimation were conservative within a factor of 2. The providers of the tool conclude that the predictability of the full BLM by the user-friendly BLM is as or even more accurate (i.e. factor of < 2) and sufficiently conservative compared to the predictability of observed field toxicity by the full BLM. Therefore, the user-friendly BLM seems sufficiently accurate. However, the documentation of these tests on the bio-met website seems not sufficiently detailed. For example, no information about the maximum overestimation of the QS by the tool is given. For this report, the site specific HC5 calculated for seven EU scenarios in the EU RAR for Cu and Ni were compared to the QS calculated by the tool. For Cu the mean ratio of the HC5 of the RAR Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 4

5 to the QS derived with the tool is very close to one (1.09). For Ni, the ratio is 2.9 which is also close to the assessment factor of 2 proposed in the RAR to derive a PNEC from the HC5. The BLM tool is only valid within parameter and metal specific boundaries of the water parameters. The common validity range for all three metals is 6.5 to 8 for the ph and 5 to 88 mg/l for Ca. For DOC boundaries are not relevant. 2. Monitoring of metal concentrations and BLM-based evaluation of compliance with QS The BLM tool requires other monitoring parameters than applied to until now in Germany (dissolved metal concentration instead of total metal concentration or concentration in sediment or SPM) and additional measurement parameter(i.e. Ca and DOC concentrations are not regularly measured yet). The dissolved metal concentration is the recommended type of concentration for metal QS according to the WFD and the site-specific QS calculated by the BLM tool are given as dissolved metal concentration. Thus, the measured concentration can directly be compared to the QS. Alternatively the calculated bioavailable concentration can be compared to a quality standard expressed as bioavailable metal concentration. A QS can be calculated from the ph, DOC and Ca measurements over the year and averaged to calculate a site specific Annual Average (AA) QS which can be directly compared to the annual average metal concentration. Maximum allowable concentration (MAC) QS based on acute data and considering bioavailability are not included in the BLM tool yet. Therefore, generic MAC-QS have to be applied to check for QS exceedances of single samples. To analyse the effects of the input parameters on the local QS 500 parameter sets were created by selecting ph, Ca and DOC randomly from their validity ranges. This analysis demonstrated that the DOC has the strongest impact on the QS showing a clear positive correlation. In addition, the sensitivity to input parameter changes of the QS derived by the BLM tool was analysed by systematic variation of one parameter while keeping the two other fixed (standard scenario ph 7.6, 2.5 mg/l DOC, 40 mg/l Ca). The analysis demonstrated the effect of the lookup table approach by sometimes pronounced stepwise changes of the calculated QS. However, also some other unexpected patterns were found: o ph dependence was totally different for the three metals: The QS for Zn show a minimum around ph 7, while the QS for Cu showed a maximum around ph 7.5 and the QS for Ni decreased over the full range of valid ph values. QS values outside the validity range of ph kept constant at the value of the boundary. o For all metals the QS increased with increasing DOC with a clear step pattern. The Ni QS was only slightly affected by the DOC. o The Ni QS was not affected by the Ca concentration at the tested ph and DOC concentration. The QS for Zn and Cu showed a complex behaviour over the tested Ca concentration range. Also, the QS were not kept constant to the boundary values outside the validity range of the Ca concentrations. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 5

6 o Comparison with a BLM tool based on functions fitted to the results of the BLM and used by UK authorities revealed smooth curves of the QS dependence and a total different pattern for the dependence of ph dependence of the Zn QS. Thus, the example analysis of the sensitivity of the QS against changes in the input parameters conducted here revealed that before a regulatory use of the tool will be implemented a more detailed analysis of the sensitivity of the QS to the input parameters should be conducted by the developers of the tool. Especially the comparison with the full BLM could not be done within this project. The look-up table approach causes sometimes considerable changes of the QS due to a small change in one input parameter. To avoid this artificial effect alternatives like are an approach based on multiple regression models which results in steady functions should be discussed. However, also the deviation between the predictions of both tools to the predictions of the full BLM should be considered. The BLM-tool is only valid within specific boundaries of the water parameters. If parameters are outside these boundaries, this is indicated by the tool. However, based on the sensitivity analyses conducted here, the prediction of QS outside the boundaries should be analysed in more detail and a systematic approach for handling these conditions should be discussed with the potential users. Until now there seems to be no clear guidance how to handle sites with water parameters outside the validity range of the tool. For Ni und Zn it should be discussed if the QS calculated at the boundaries could be extrapolated to higher ph values or if the generic QS should be used as the most conservative approach. However, the last option could result in a large change of the QS if the ph is just above the boundary. Based on example monitoring data sets from federal states in Germany the relevance of the validity range of ph and Ca concentrations for water bodies in Germany was analysed. For simplicity, only the ranges where the BLM for all three metals are valid were considered. From the more of analytical data sets available for one Federal state (Nordrhein-Westfalen), 1.5 % had a ph below 6.5 but 21.7 % a ph above 8. For Ca concentrations, 2.9 % of the samples were below 5 mg/l while 15.9 % above 88 mg/l. In a second state, Baden-Wuerttemberg, the percentage of sites with values below the lower ranges was similar but more than 40 % showed ph or Ca concentrations above the validity range (however, only one river was covered). For Sachsen-Anhalt, a third state, the data set was considerable smaller but the water bodies sampled had often (40.5 %) ph above the validity range and all were above the validity range for the Ca concentrations. However, the Sachsen-Anhalt rivers were chosen for this evaluation because of their high salt concentrations and thus might be not representative. Especially the frequency of samples with high ph values (> ph 8) should be considered because the ph dependence of the QS seems to be very variable between the metals and it cannot be assumed that higher ph values correspond to reduced bioavailability. For Ca concentrations the situation seems less critical because it can be assumed that the bioavailability will not increase at higher water hardness. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 6

7 3. Potential implications of the use of BLM for emission and immission balances Plant permissions refer to certain loads of substances emitted to the river. For the assessment of metal loads of waters the BLM are not applicable since they only consider concentrations in the water phase (at least not for those metals of which a large fraction is bound to suspended particulate matter). Thus total water concentrations of metals need to be determined in addition to dissolved metal concentration. The potential influence at downstream sites is discussed for the case of data for the river Wupper. It was observed that discharge concentrations of metals were partly above the QS local of the respective upstream sampling site (especially for Ni and Zn). However, comparisons were only qualitative since input data were total metal data measured in effluents and dilutions could not be considered. 4. Cost benefit analysis Currently, dissolved metal concentrations are not determined routinely in Germany. Also DOC and Ca measurements are not performed regularly. ph measurements, on the other hand, are more often available. Costs for the additional required measurements were estimated by experts from relevant authorities of federal states to be about per sample. It is expected that the advantage of the BLM implementation would be to avoid risk mitigation at sites where the water conditions significantly reduce the bioavailability and thus the potential risk to the aquatic community. Conclusions (related to the main questions defined by the sponsor) 1. The BLM approach in general is considered suitable for regulatory risk assessment of metals. The approach has been developed over several years, is described in many peer-reviewed publications, has been used in EU risk assessment reports and is recommended under the WFD. 2. In principal, background concentrations can be considered independently of the use of BLM by the added risk approach (ARA). However, the WFD guidance document recommends a tiered approach for the QS derivation for metals, starting with generic QS on tier 1, consideration of bioavailability, e.g., by BLM on tier 2, and - if potential hazard is indicated - consideration of background levels at tier 3. The major problem for the ARA is seen in the determination of reliable natural background concentrations. For Zn the Added Risk Approach was used in the RAR to derive PNEC and also the QS derived by the BLM tool is a QSadd (thus, if information on background levels is available, the background concentration can be added to the QS). 3. The benefits of considering bioavailability for the assessment of the protection goals compared to the current approach would be that it would use the current scientific state of the art to achieve a more realistic assessment of the potential effects of metals on the pelagic community. 4. There are some methodological constraints to achieve harmonized results of monitoring programmes in Europe. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 7

8 Although the plausibility of the QS calculated by the BLM tool has been checked by the developers by comparing the outputs with results of the full BLM, the here performed sensitivity analyses revealed some unexpected finding. Some of these problems have been solved in a revised version of the tool which became available in October 2011 An analysis of the influence of the measurement uncertainty on the results revealed that especially ph value variations and DOC concentration variations (at low DOC levels) result in significant changes of QS local. Thus, variability and measurement uncertainty of ph and DOC values can result in large uncertainty of the derived QS. On the other hand, the uncertainty on QS without consideration of bioavailability is larger but probably only related to an underestimation of the true QS. For the conditions tested Ca measurement uncertainty had only a minor influence (however, it has to be considered that Ca measurement uncertainty would be relevant especially in the vicinity to step changes of Ca levels). The original full BLM and also the BLM tools are only applicable within the boundaries of the water parameters where they have been developed and validated. The tool indicates when water parameters are out of these boundaries. Using example monitoring data sets from three German federal states it was checked if the water conditions (ph and hardness) in German water bodies can expected to be covered by the tool. Despite that the BLM were developed to cover at least 90 % of the water bodies in Europe, it was found that there were considerable amounts of samples (about 20 % of the samples analysed here for one federal state from which a larger data set was available) with values outside the boundaries for ph and Ca-concentrations. The ph was often above the upper validity limit. Thus, H + concentrations were lower and thus, e.g., the QS for Ni can be expected to decrease. On the other side, if Ca was outside the boundaries it was in most cases above the higher limit where the QS potentially increases. In a recent paper Verschoor et al. (2011) report on an alternative approach. They assessed spatial and temporal variations of metal levels considering water-type specific sensitivities for a range of aquatic species in The Netherlands. Cu, Zn, and Ni BLM were used to normalize chronic NOEC data determined in test media into site-specific NOEC for about 370 sites. Then, site-specific SSD were constructed. Sensitivity of species (as NOEC) and of the ecosystem (as HC5) for Cu, Ni, and Zn revealed variations of up to 2 orders of magnitude. The authors report that the application of space-time specific HC5 for Cu and Zn resulted in a reduction of sites at risk, but in an increase of sites at risks for Ni. It should be considered whether this site-specific approach via locally adapted SSD should be tested for German waters, too. By this means a validation of the user-friendly BLM tool could be performed. However, a decision may be taken after further evaluations with a refined (final) version of the EU BLM tool are available. 5. Finally the validity of BLM with respect to the protection of ecosystems was discussed. The BLM tool focuses only on the potential effects on the organisms exposed mainly via the water. Accumulation of metals in the sediment or biomagnification in the food web are not taken into Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 8

9 account. However, ecotoxicological data for setting quality standards for organisms living in the sediment are rarely available, the exposure via the water seems to be the most relevant pathway, and thus standards for dissolved concentrations in the water are usually considered to be protective. The BLM tool is also restricted to long-term exposure respectively chronic effects. Short-term effects have to be assessed separately (e.g. by comparing measured metal concentrations with MAC-EQS). In total, the BLM is considered as the state of the art scientific approach in generic risk assessment. The transfer of the BLM approach into compliance monitoring has shown some open questions concerning new sources of uncertainty that need to be discussed. The applicability of the BLM tool for German rivers seems to be restricted, due to the limited monitoring data available for this report. The assumed applicability for 90% of European rivers has to be proved also. The costs for implementation seem to be proportionate considering the benefits of a more realistic assessment of the potential risks.. For the implementation in compliance monitoring, the tool has to be refined, boundaries need to be readjusted, fully tested with monitoring data from different European regions and documented. Guidance or harmonized approaches to deal with water conditions outside the validity range of the BLM has to be developed. Accumulation of metals in sediment or biota as well as potential effects of high concentrations measured in single samples have to be taken into account additionally. For emission and immission balances the measurements of total metal concentrations in water are considered still necessary. For substances that are highly adsorbed to suspended matter the EQS alternatively might be based on SPM concentrations. These data are more appropriate for calculating substance fluxes in streams. This is even preferable to expressing the EQS as a total water concentration because the latter is dependent on the highly variable SPM fraction (e.g., depending on season and turbidity) and so may be highly uncertain. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 9

10 Contents Abstract 2 Zusammenfassung 2 Executive summary 3 Contents 10 Figures 11 Tables 13 Abbreviations Motivation and aims General considerations on the BLM approach and its potential benefits Monitoring of environmental metal concentrations and evaluation of compliance with QS Potential implications of the use of BLM for emission and immission balances in the future Cost-benefit analysis for application of BLM based QS for compliance monitoring Final discussion Acknowledgements References 78 Appendix 82 Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 10

11 Figures Figure 1: Schematic description of the processes which are the base of biotic ligand models (here for a divalent cation; 19 Figure 2: Validation of the Ni BLM for the alga Pseudokirchneriella subspicata (taken from the Ni EU RAR 2008) Figure 3: Performance of the user-friendly BLM tool against calculations of the full BLM (copied from the Microsoft PowerPoint file BLM workshop Technical session (Final) provided with the EU workshop report, David et al. 2011) Figure 4: Overview of assessments needed and selection of an overall EQS (copied from EQS TGD 2011, Figure 2-3) Figure 5: Input and output data from the EU BLM EXCEL sheet (Biomet Tool_Spreadsheet_ Workshop_ xlsm version; June 2011). Output data are only presented for copper. The generic QS for copper is 1 µg/l. The grey shaded area shows the input data for dissolved metal concentrations and those for ph values, DOC- and Ca concentrations. Further explanations are given in the text. The data designated as EQS are here considered as QS (see comment in footnote 2, page 9) Figure 6: Input and output data from the EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm; available since end of November 2011). Output data are only presented for copper. The generic QS for copper is 1 µg/l. The grey shaded area shows the input data for dissolved metal concentrations and those for ph values, DOC- and Ca concentrations. Further explanations are given in the text. The data designated as EQS are here considered as QS (see comment in footnote 2, page 9) Figure 7: Correlation of QS to parameters randomly selectedplots for Figure 8: top: ph dependence of QS local as calculated with the EU BLM EXCEL sheet (Biomet Tool_Spreadsheet_Workshop_ xlsm) provided at the EU BLM workshop in June 2011; bottom: ph dependence of QS local as calculated with the revised EU BLM EXCEL sheet (Biomet_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November Boundaries are similar in both diagrams (see also Table 843) Figure 8: top: DOC dependence of QS local as calculated with the EU BLM EXCEL sheet (Biomet Tool_Spreadsheet_Workshop_ xlsm) provided at the workshop in June broad DOC range. bottom: DOC dependence of QS local as calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November broad DOC range Figure 9: top: DOC dependence of QS local as calculated with the EU BLM EXCEL sheet (Biomet Tool_Spreadsheet_Workshop_ xlsm) provided at the workshop in June performance for low DOC concentrations. bottom: DOC dependence of QS local as calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November performance for low DOC concentrations Figure 10: top: Ca concentration dependence of QS local as calculated with the EU BLM EXCEL sheet (Biomet Tool_Spreadsheet_Workshop_ xlsm) provided at the workshop in June broad Ca range. bottom: Ca concentration dependence of QS local as calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November broad Ca range. Boundaries are similar in both diagrams (see also Table 843) Figure 11: top: Ca concentration dependence of QS local as calculated with the EU BLM EXCEL sheet (Biomet Tool_Spreadsheet_Workshop_ xlsm) provided at the workshop in June performance for low Ca concentrations. bottom: Ca concentration dependence of QS local as calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November performance for low Ca concentrations. Boundaries are similar in both diagrams (see also Table 843) Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 11

12 Figure 12: Comparison of Zn results from the BLM sheet provided at the EU workshop (versin of June 2011; solid line) and the UK BLM EXCEL sheet provided by G. Merrington (dotted line). QS values are µg/l, and Ca and DOC concentrations mg/l. For Ca concentration < 1 mg/l a default QS of 10.9 µg/l Zn is applied (this corresponds to the applied generic QS; shown as red bar in the bottom diagram) Figure 12: Map of the rivers Rhine, Lipper, Emscher and Wupper in Nordrhein-Westfalen considered in this evaluation (courtesy of LANUV Nordrhein-Westfalen) Figure 14: Map of the Neckar river in Baden-Wuerttemberg considered in this evaluation (this map was Figure 15: retrieved from the website 51 Map of the rivers in Sachsen-Anhalt considered in this evaluation (this map was retrieved from the website 52 Figure 16: ph and Ca concentrations measured in the river Rhine at Bad Honnef (km 640). The shaded areas indicate the range where all three BLM (Cu, Ni, Zn) are valid Figure 17: ph and Ca concentrations measured in the river Rhine at Kleve-Bimmen (km 865). The shaded areas indicate the range where all three BLM (Cu, Ni, Zn) are valid Figure 18: ph and Ca concentrations measured in the river Emscher at the mouth into the river Rhine. The shaded areas indicate the range where all three BLM (Cu, Ni, Zn) are valid Figure 19: ph and Ca concentrations measured in the river Lippe at Wesel. The shaded areas indicate the range where all three BLM (Cu, Ni, Zn) are valid Figure 20: ph and Ca concentrations measured in the river Neckar, sampling site Rottweil. The shaded areas indicate the range where all three BLM (Cu, Ni, Zn) are valid Figure 21: ph and Ca concentrations measured in the river Saale, sampling site The shaded areas indicate the range where all three BLM (Cu, Ni, Zn) are valid Figure 21: Influence of the variation of the ph values by + 5 % on the calculated EQS (as error bar). Calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November Figure 22: Influence of the variation of the DOC concentration by 30 % on the calculated EQS (as error bar). Calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November Figure 23: Influence of the variation of the Ca concentration by 10 % on the calculated EQS (as error bar). Calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_ tool_-v1.4_ xlsm) downloaded from the Biomet website in November Figure 24: Assessment of the influence of the measurements uncertainty - variation of input parameters. Data sets for five month were varied (only one: either dissolved metal concentrations or one of the parameters ph, DOC, Ca per dataset). Calculated with the revised EU BLM EXCEL sheet (Biomet_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November Figure 25: Figure 26: Dissolved metal concentrations at sampling stations, total metal concentrations of effluents, and QS local calculated with the EU BLM tool (version of June 2011) for the river Wupper (Nordrhein- Westfalen, Germany) Annual average dissolved metal concentrations for Cu, Ni and Zinc at sampling stations and QS local calculated with the EU BLM tool (version of June 2011) for the river Neckar (Baden- Wuerttemberg, Germany). Data for The sampling sites are Rottweil, Börstingen, Kirchentellinsfurt, Deizisau, Poppenweiler, Besigheim, Kochendorf, Mannheim (first three letters are used as abbreviations) Figure 27: Schematic description of the derivation of QS and the consideration of natural background values of metals. Scheme from EQS TGD (2011) Figure 28: Schematic description of the derivation of PNEC and the consideration of natural background values of metals. Scheme from MERAG (2007). (Note that this scheme is for prospective risk assessment (PNEC) but the general principle is also valid for QS estimation Figure 29: Tiered assessment scheme for implementing an EQS for zinc (UK Environment Agency 2010). 74 Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 12

13 Tables Table 1: Metal EQS according to the Directive 2008/105/EC (EC 2008). AA: annual average; MAC: maximum allowable concentration Table 2: Metal EQS according to OGewV (2011). AA: annual average Table 3: Comparison of the physical-chemical conditions of the different scenarios versus EU surface waters (Swad database). (Table 3 14 in the Cu RAR 2008) Table 4: Summary of the physical-chemical characteristics of the different selected scenarios. (extract from Table 3 13 of Cu RAR 2008) Table 5: HC5-50 (i.e. at 50th % confidence limit together with 5th and 95th confidence limits) derived from the best fitting distribution and log normal distribution. All values in µg/l. (Table 3 20 of the Cu RAR 2008) Table 6: HC5-50 (i.e. at 50th % confidence limit together with 5th and 95th confidence limits) derived from the best fitting distribution and log normal distribution for Ni. All values are in µg/l. (Table of the Ni EU RAR 2008) Table 7: Comparison of median HC5 for log-normal SSD given in the risk assessment reports and the QS for the specific water parameters by means of the BLM tool V1.4 (version November 2011). For Ni the BLM tool states that the Ca concentration is above the upper validity bound Table 8: Validity range of the BLM tool for Cu, Ni and Zn and the generic QS used in the tool. The total validity range is the range where all three BLM are valid Table 9: Results of the comparison of monitoring data for ph, Ca and DOC and the validity ranges of the EU BLM tool (Version of June 2011) for Cu, Ni and Zn. Percentage values given are for the fraction of data which are outside the respective boundary Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 13

14 Abbreviations AA ARA BAT Bioavailability BioF BLM DOC Annual Average Added Risk Approach; the added risk approach assumes that only the anthropogenic added fraction of a natural element that contributes to the risk for the environment should be regulated/controlled (MERAG 2007). Thus, the natural background concentration is subtracted from the measured concentration for comparison with the QS. See also TRA Best Available Technology combination of the physicochemical factors governing metal behaviour and the biological receptor, its specific pathophysiological characteristics (such as route of entry, and duration and frequency of exposure; David et al. 2011) Bioavailability factor - the ratio of the generic QS bioavailable divided by the local QS. The BioF is based on a comparison between the maximum bioavailability and that relating to site-specific conditions. The BioF is always 1 (= 100% bioavailable metal) or less. Biotic Ligand Model(s). A predictive metal and species specific model that accounts for variation in metal toxicity due to the chemistry of the water, i.e. ph, DOC, Ca concentration, etc. Dissolved organic carbon EQS Environmental Quality Standard(s) 2 Generic EQS HC5 Local EQS EQS representative of conditions of high bioavailability; it is expressed as bioavailable metal concentration (also reference EQS) Hazardous Concentration for 5 % of species calculated as the 5th percentile of a SSD (Species Sensitivity Distribution). For example, a HC5 of a SSD based on NOECs means that at this concentration it is expected that the NOEC of 5 % of the species is exceeded. In order to express the uncertainty on the HC5, often additionally a % value of the probability that the true HC5 is lower than the estimated one is given, e.g., HC5 (50%) for the median HC5 calculated dissolved concentration of metal that is equivalent to the EQS bioavailabe at the local water conditions at the site (also EQS local ) 2 In this opinion we use the term EQS in the sense of the overall EQS, thus considering the QS for water, sediment, secondary poisoning and drinking water. BLM covered here are only related to the QS fw,eco, the standard to protect the aquatic community exposed via the water. The overall EQS which considers all protection goals can be lower than the QS fw, eco. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 14

15 LOEC LOQ MAC NOEC PEC PNEC Lowest Observed Effect Concentration = lowest concentration in a test which gives a (statistically) significantly different response compared to the control Limit Of Quantification Maximum Allowable Concentration No Observed Effect Concentration; lowest concentration in a test which gives a (statistically) not significantly different response compared to the control Predicted Environmental Concentration. Used in prospective risk assessments, e.g. under REACH. Under the WFD, measured concentrations instead of PECs are used. Predicted No Effect Concentration (derived from a set of ecotoxicological tests and used, e.g., under REACH) QS Quality Standard(s) (see footnote 2) RAR REACH RCR SPM STP SSD Risk Assessment Report(s) Regulation on Registration, Evaluation, Authorisation and Restriction of Chemicals (EU REACH Directive 2006) Risk Characterization Ratio (or risk quotient, also, and more precisely, called hazard quotient); calculated by dividing the PEC by the PNEC; values >1 present a potential risk. In the context of the WFD, the hazard quotient is the ratio of the measured concentration and the EQS Suspended Particulate Matter Sewage Treatment Plant Species Sensitivity Distribution(s) TGD Technical Guidance Document (EQS TGD 2011) TRA Total Risk Approach. The TRA assumes that exposure and effects should be compared on both, the fraction compiling the natural and the added anthropogenic background; the risk characterization can be done at different levels, for example, on total, dissolved, or bioavailable fractions (MERAG 2007). See also ARA WFD Water Framework Directive (EU WFD 2000) WHAM Windermere Humic Aqueous Model; metal chemical speciation model (Tipping et al. 1994) Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 15

16 1. Motivation and aims The load of waters with chemical substances is controlled via emission and immission regulations. Numerous regulations are covering emissions mostly referring to technical standards (best available technology, BAT). They are national or regional, may be only valid for a certain industrial sector or even for specific products or compounds, or for certain applications. However, for immission assessment a uniform European-wide method was adopted for chemical risk characterisation and the derivation of Environmental Quality standards (EQS) under the EU Water Framework Directive (EU WFD 2000). On base of the daughter directive 2008/105/EC environmental quality standards (EQS) for a number of priority substances were derived and implemented (EC 2008).Compliance with the provisions is proved on the basis of chemical monitoring and comparison of the measured concentrations with the EQS the chemical quality of waters. Metals are a specific group of substances in this context because they can be present in water bodies naturally due to geochemical processes. Some of them are essential elements for organisms, and their toxicity can strongly depend on the water conditions. Part B of the Directive 2008/105/EC, a daughter directive of the WFD (EU WFD 2000), states that in case of metals the EQS refer to the dissolved concentration, i.e. the concentration measured in water sample obtained by filtration through a 0.45 μm filter or any equivalent pre-treatment. Member States may, when assessing the monitoring results against the EQS, take into account: (a) natural background concentrations for metals and their compounds, if they prevent compliance with the EQS value; and (b) hardness, ph or other water quality parameters that affect the bioavailability of metals. Guidance Document No 19 on surface chemical water monitoring under the WFD (EC 2009) also states that bioavailable metal concentrations depend on various parameters including ph, Ca and Mg concentrations, as well as dissolved organic carbon (DOC) concentration. Hence, measuring these parameters in parallel with the metals can assist in the interpretation of results, where appropriate. In case of cadmium, the measurement of hardness is mandatory (see Table 1). Currently only a part of the metals are regulated via the WFD. The Directive 2008/105/EC states EQS for the following metals: lead, nickel, mercury, cadmium (and for one group of organo-metal compounds, tributyltin compounds, which is not covered here). The current EQS for the priority substances Cd, Pb, Ni and Hg are listed in Table 1. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 16

17 Table 1: Metal EQS according to the Commission draft ( ) amending Directive 2008/105/EC (EC 2008). AA: annual average; MAC: maximum allowable concentration. No Name of substance (6) Cadmium and its compounds (depending on water hardness classes)# (20) Lead and its compounds (21) Mercury and its compounds (23) Nickel and its compounds CAS number AA-EQS Inland surface waters [μg/l] AA-EQS Other surface waters [μg/l] MAC-EQS Inland surface waters [μg/l] MAC-EQS Other surface waters [μg/l] (Class 1) 0.08 (Class 2) 0.09 (Class 3) 0.15 (Class 4) 0.25 (Class 5) (Class 1) 0.45 (Class 2) 0.6 (Class 3) 0.9 (Class 4) 1.5 (Class 5) 0.45 (Class 1) 0.45 (Class 2) 0.6 (Class 3) 0.9 (Class 4) 1.5 (Class 5) ,2 1,2 13 7,2 1,3 14 not applicable , EQS Biota # For cadmium and its compounds the EQS values vary depending on the hardness of the water as specified in five class categories (Class 1: < 40 mg CaCO 3 /L, Class 2: 40 to < 50 mg CaCO 3 /L, Class 3: 50 to < 100 mg CaCO 3 /L, Class 4: 100 to < 200 mg CaCO 3 /L and Class 5: 200 mg CaCO 3 /L). The German OGewV (2011) specifies: the hardness class is derived from CaCO 3 -concentrations which result from the 50th percentile of CaCO 3 -concentrations determined parallel to the cadmium concentrations (not defined in the original EQS Directive) In Germany further metals are covered by a national regulation. Since 2011 the respective EQS are regulated via the Surface Water Ordinance (Oberflächengewaesserverordnung, OGewV 2011, EQS see Table 2). Before this regulation was enacted the EQS were regulated on the level of the German federal states (Bundeslaender). However, even then the values were harmonized by the State- Laender-working-group water (LAWA) and the same as now covered by the OGewV Note that no quality standards for short-term exposure (MAC-EQS) are given and that for As, Cr, Cu and Zn the EQS is expressed as the concentration in sediment and suspended particulate matter (SPM) instead of concentrations in the water phase. Table 2: Metal EQS according to OGewV (2011). AA: annual average. No CAS number Name of substance AA-EQS inland surface waters including transitional and coastal waters Water [μg/l] Sediment or suspended particulate matter [mg/kg] Arsenic Chromium Copper Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 17

18 No CAS number Name of substance AA-EQS inland surface waters including transitional and coastal waters Water [μg/l] Sediment or suspended particulate matter [mg/kg] Zinc Selenium Silver Thallium For a number of metals Risk Assessment Reports (RAR) were prepared under the European existing chemicals regulation (i.e., cadmium, zinc, nickel 3 ; for copper and lead voluntary RAR 4 prepared by industry stakeholders). The research data which are compiled in these reports revealed that the measured total metal contents in water do not correlate with the toxic effect on aquatic organisms. The reference to dissolved concentrations (operational defined as filtered through a 0.45 µm membrane filter) eliminates extreme outliers, but still yields no satisfying correlation of concentrations and effects. Only with consideration of site-specific bioavailability of metal ions the connection of concentration and effect becomes clearly. The free metal ions bind in particular to gills or gill-similar structures of water organisms, since these exhibit negatively charged binding sites. After uptake the metal ions may be distributed in the organism via different cellular transport systems to specific sites of action. Other metals (especially calcium and magnesium) may compete with the binding sites. Since protons also compete, the ph value is an important factor, too. On the other hand dissolved organic carbon (DOC) can also provide binding sites, thus reducing the fraction of bioavailable metal ions. In a simple way bioavailability can be considered in metal risk assessment by deriving PNECS or EQS for different Ca/Mg concentrations (hardness banding; e.g. as implemented for Cd; see Table 1). However, a more sophisticated way to consider bioavailability is by so called Biotic Ligand Models (BLM; for reviews see, e.g., Paquin et al., 2000, 2002). These BLM are based on the one hand on speciation models which provide information on the relevant metal species under the prevalent physic-chemical water quality parameters (e.g., Windermere Humic Aqueous Model WHAM; Tipping et al. 1994). On the other hand they model the binding of metals by gills or gill-like structures of organisms (see scheme in Figure 1). BLM are metal- and species-specific. By the use of BLM and the reference to bioavailable concentrations the variability of toxicity of metals for 3 RAR for Cd, Ni, Zn are published on the website of the Joint Research Center (former European Chemical Bureau; Cd: Ni: (Ni EU RAR 2008); Zn: (Zn EU RAR 2010). 4 Voluntary risk assessments are to be found at the ECHA website: Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 18

19 different taxa (and also the variability of intra-species toxicity) can clearly be decreased (see e.g. the Ni EU RAR 2008). Figure 1: Schematic description of the processes which are the base of biotic ligand models (here for a divalent cation; The PNEC derivations in the EU RAR for Cu, Ni, and Zn were already based on BLM. The use of BLM for risk characterisation purposes is also approved by the European Union Scientific Committee on Health and Environmental Risks (e.g., SCHER 2010) and the BLM application is also recommended in the new guidance document for the derivation of EQS for the WFD (EQS TGD 2011). In June 2011 The European Commission organized a workshop where a ready to use tool to calculate site specific EQS based on BLM was presented. The provided EXCEL-file (Biomet Tool_Spreadsheet_Workshop_ xlsm, allows bioavailability assessments for three metals (Cu, Ni, Zn). Beside dissolved metal concentrations ph values, DOC- and Caconcentrations have to be available as input parameters for the calculations. Basis for the provided BLM EXCEL versions are look-up tables with results of the full BLM which were derived for the EU risk assessments of these metals. In November 2011 a new version of the BLM tool was launched at the website (Bio-met_bioavailability_tool_-v1.4_ xlsm; user guide Bio-met 2011). It was not possible to repeat all calculations with the new version. However, at least a comparison between both versions of the sensitivity regarding ph, DOC and Ca concentrations was conducted. The aim of this study is to evaluate the pros and cons of implementing the BLM approach, i.e. by using the proposed BLM tool, in the monitoring programmes under WFD in Germany. Especially the plausibility of the derived values, the practicability in routine monitoring and the potential consequences for decisions on compliance should be discussed. In the description of work the UBA has listed the following six main questions to be addressed in this report: Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 19

20 I. Are BLM generally applicable for the regulatory risk assessment of metals? II. Should natural background concentrations be explicitly considered? III. What are the benefits of using the BLM approach for reaching the protection targets compared to the current approach? IV. What are the methodological constraints for the implementation of the BLM approach referring to consistent results throughout Europe? V. How can the validity of the BLM approach for the protection of the ecosystems be assessed? VI. Which definition of substance is applied for BLM since they are metal-specific? In addition a couple of more detailed questions are bundled in four work packages: 1. Status quo and general consideration of the BLM approach; 2. Potential problems of the implementtation of the BLM approach in routine monitoring in Germany; 3. Consequences for emission and immission balances in the future; and 4. Cost benefit consideration for BLM implementation in Germany. The work packages will build the next four chapters of this report before the six main questions are answered in the final chapter based on the results of the work packages as well as additional considerations. 2. General considerations on the BLM approach and its potential benefits What are general validity criteria for a BLM allowing its use in the routine water quality monitoring? The validity of the use of BLM for implementing the WFD can be assessed on different levels: General validity of the BLM approach The BLM is a mechanistic description how water chemistry affects the speciation and the binding of metal ions to the biotic receptor (the biotic ligands). BLM can be considered as a state of the art scientific approach. The BLM approach is based on long history of development and refinements published in the peer-reviewed literature (see, e.g., reviews of Paquin et al. 2002, Niyogi & Wood 2004, or the book by Meyer et al. 2007). BLM have been used in the US for refining the water quality criteria for copper and in EU risk assessment reports (e.g. for Ni and Zn). The use of BLM is recommended in recent guidance documents related to chemical risk assessment and quality standard setting (ECHA 2008a, b; EQS TGD 2011) and support by SCHER (2010). Validity of specific BLM Basically, BLM are species and metal specific. The validity of the BLM used in EU RAR has been tested by comparison of the BLM predictions with metal toxicities measured in ecotoxicological tests using water samples fromt different sites across Europe (see e.g. the Zn and Ni risk assessment reports; Zn EU RAR 2010, Ni EU RAR 2008). As an example, the validation of the algae BLM for Ni is Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 20

21 shown in Figure 2. In the USA, BLM are used without such external validation (van Sprang & Oorts 2011). Figure 2: Validation of the Ni BLM for the alga Pseudokirchneriella subspicata (taken from the Ni EU RAR 2008). Validity of species to species extrapolation QS setting for the aquatic community is preferably based on the species sensitivity distribution (SSD) approach (EQS TGD 2011). A set of test species, preferable more than 15 (but at least 10), is considered as a sample of all the species potentially exposed in the field. From the distribution of the NOEC or EC10 values of the tested species the HC5, the hazardous concentration for 5 % of the species, is calculated as the 5th percentile of the distribution. Considering potential remaining uncertainties the EQS is derived by application of an assessment factor between 1 and 5 (for details on the SSD approach see the EQS TGD 2011). The derived EQS should be site-specific and consider bioavailability extrapolation from the limited number of species for which BLM are available to the species without a species-specific BLM. Under the assumptions of similar mechanisms of action a BLM from one species is used also for related species, i.e. a rainbow trout BLM for other fish, and a Daphnia BLM for other invertebrates. The validity of such extrapolation is done by spot checking of the BLM (van Sprang & Oorts 2011). This is discussed for one example: For Ni the procedure is described in Schlekat et al. (2010). Toxicity tests were conducted in different natural waters with four species where no BLM were available for (a midge, a snail, a rotifer and a duckweed). Then the observed toxicities were compared to predictions of the Daphnia and Ceriodaphnia BLM and the algae BLM for the duck weed. In most cases the BLM were able to predict the observed toxicities within a factor of two. However, as Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 21

22 discussed in the Ni EU RAR (2008) some residual uncertainty remains in applying BLM to non-blm species. For example, the Daphnia BLM showed a better fit to the Lemna data than the algae BLM which demonstrates that there is some uncertainty in using the BLM of the taxonomically closest species. However, a sensitivity analysis indicated that the differences between the use of the bestfitting model (i.e. Daphnia BLM for Lemna) versus the most stringent model (algae-blm for Lemna) were extremely minor (Ni EU RAR 2008, appendix G.7). For the EU BLM tool applied here further information on validation is given in the BLM tool user guide provided recently (Bio-met 2011). Validity of the user-friendly BLM tool (Bio-met) To allow the use of the BLM in routine monitoring a user-friendly tool was presented at the EU Workshop in June 2011 (David et al. 2011). This tool is based on look-up tables of results from the full BLM. In total, HC5 have been calculated for more than combinations of water parameters ph, Ca and DOC concentration. The HC5 for a given set of water parameters is then calculated as the minimum of the two HC5 with the most similar water conditions (see online documentation in the appendix). When inputs values are outside the validated range the best fitting combination of validated ph, DOC and Ca is returned. In order to validate the EU BLM tool, results of the tool were compared to the results of the full BLM for hundreds of sites in the UK and The Netherlands. For the UK sites, 84, 99 and 80 % of the estimations by the tool for Cu, Ni, and Zn, respectively, where conservative within a factor of 2. For the sites in the Netherlands 69, 59 and 87 %, respectively, of the estimation were conservative within a factor of 2. The providers of the tool conclude that the predictability of the full BLM by the user-friendly BLM is as or even more accurate (i.e. factor of < 2) and sufficiently conservative compared to the predictability of observed field toxicity by the full BLM. HC5 values derived by the user-friendly EU BLM tool were correlated to the HC5 predicted by the BLM model for sites in the UK and The Netherlands (David et al. 2011). The plot for Zn is shown as an example here (Figure 3) other examples can be found in the online documentation of the tool, given also as the appendix of this report). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 22

23 Figure 3: Performance of the user-friendly BLM tool against calculations of the full BLM (copied from the Microsoft PowerPoint file BLM workshop Technical session (Final) provided with the EU workshop report, David et al. 2011). To derive the site specific quality standards, assessment factors between 1 and 5 were applied to the HC5. However, it is not documented which assessment factors were used in the end. For the project here, the HC5 calculated in the EU risk assessment reports for Cu and Ni were compared to the QS-calculations of the BLM tools. The ratio between the two values corresponds to the assessment factor applied and should be relatively stable for each metal. For the EU reports for Cu and Ni seven scenarios were identified to provide examples of typical conditions covering a wide range of physico-chemical conditions (ph between 6.67 and 8.21; hardness between 27.8 and 260 mg/l CaCO 3, DOC between 2.5 and 27.5 mg/l) occurring in EU surface waters (Cu RAR 2008) with respect to parameters driving the bioavailability of metals (see Table 3 and Table 4). In the report for Zn (Zn EU RAR 2010) these scenarios have not been used and only a generic PNEC add,aquatic of 7.8 µg/l for dissolved Zn in freshwater was derived from a SSD. Table 3: Comparison of the physical-chemical conditions of the different scenarios versus EU surface waters (Swad database). (Table 3 14 in the Cu RAR 2008). Type Scenario 5 ph Hardness (mg/l CaCO3) DOC (mg/l C) Rivers Small (ditch) Low High Very high Medium Medium-high Medium Medium-low Large Medium High Low Mediterranean High High Low Lakes Oligotrophic systems Medium Low Low Acidic system Low Low Low 5 Low (L): when the phys.-chem. characteristics in the system are covering 10 th % of abiotic factor in EU surface waters; Medium (M): covering 50 th %; High (H): covering 90 th %. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 23

24 Table 4: Summary of the physical-chemical characteristics of the different selected scenarios. (extract from Table 3 13 of Cu RAR 2008). Rivers Type Name Country ph Hardness (mg/l CaCO3) Small (ditches / The (Ca: 88.2; with flow rate Netherlands Mg: 31.6 mg/l) of ± 1,000 m³/d) Medium (rivers with flow rate of ± 200,000 m³/d) Large (rivers with flow rate of ± 1,000,000 m³/d) River Otter / River Teme River Rhine United Kingdom The Netherlands (Ca: 46.9; Mg: 11.6 mg/l) 159 (Ca: 49.9; Mg: 8.4 mg/l) (Ca: 68.9; Mg: 10.9 mg/l) Mediterranean river River Ebro Spain (Ca: 72.9; Mg: 22.1 mg/l) Lakes Oligotrophic Lake Italy (Ca: 13.6; systems Monate Mg: 3.5 mg/l) Acidic system / Sweden (Ca: 8.7; Mg: 1.5 mg/l) Boundaries Foregs database / Swad database DOC Na Alkalinity (mg/l) (mg/l) (mg/l CaCO3) The resulting median HC5 for Cu and Ni are listed in Table 5 and Table 6. These BLM calculated HC5-50 values from the log-normal distribution have been carried forward to the risk characterisation in the RAR. Table 5: HC5-50 (i.e. at 50th % confidence limit together with 5th and 95th confidence limits) derived from the best fitting distribution and log normal distribution. All values in µg/l. (Table 3 20 of the Cu RAR 2008). Scenario HC5-50 (µg/l) using the best fitting distribution Ditch in The Netherlands 22.1 ( ) beta River Otter in the United 7.8 ( ) Kingdom log-normal River Teme in the United 17.6 ( ) Kingdom beta River Rhine in The 8.2 ( ) Netherlands log-normal River Ebro in Spain 9.3 ( ) beta Lake Monate in Italy 10.6 ( ) log-normal Acidic lake in Sweden 11.5 ( ) beta HC5-50 (µg/l) using the log normal distribution 27.2 ( ) log-normal 7.8 ( ) log-normal 21.9 ( ) log-normal 8.2 ( ) log-normal 10.6 ( ) log-normal 10.6 ( ) log-normal 11.1 ( ) log-normal Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 24

25 Table 6: HC5-50 (i.e. at 50th % confidence limit together with 5th and 95th confidence limits) derived from the best fitting distribution and log normal distribution for Ni. All values are in µg/l. (Table of the Ni EU RAR 2008). The site specific data for ph, DOC and Ca in Table 4 were used as inputs in the user-friendly EU BLM tool to calculate the QS. Results are shown in Table 7, together with the HC5 based on the lognormal SSD. The HC5/QS ratio for Cu varies between 0.74 and 2.22 but is close to 1 for the five other scenario and thus, also close to the assessment factor applied to the HC5 in the Cu RAR (2008). For Ni the ratio varies between 1.8 and 3.7 and thus also in the range of the proposed assessment factor in the Ni EU RAR (2008). Reasons for this higher variability are unclear. It is also not clear why in the EU report the BLM validity range for hardness is given as mg CaCO 3 /L (corresponding to approximately 1300 mg Ca/L) but why in the BLM tool the upper boundary for Ca is set at 88 mg/l. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 25

26 Table 7: Comparison of median HC5 for log-normal SSD given in the risk assessment reports and the QS for the specific water parameters by means of the BLM tool V1.4 (version November 2011). For Ni the BLM tool states that the Ca concentration is above the upper validity bound. ID Sample Name ph Input data Cu results Ni results DOC [mg/l] Ca [mg/l] QS [µg/l] HC5 [µg/l] HC5/QS QS [µg/l] HC5 [µg/l] 1 Small ditch, NL River Otter, UK River Teme, UK River Rhine, NL River Ebro, ES Lake Monate, IT Neutral-acidic 7 system SE HC5/ QS The EU BLM tool was developed and validated on data sets covering specific ranges of water quality conditions. Therefore, it is only valid within these boundaries. If an input value for ph or Ca is outside the validated range a QS local prediction using the best-fitting combination of validated ph, DOC and Ca values is applied. Some of the unexpected outcomes of the tool found during the sensitivity analysis conducted in chapter 3 may be explained by these facts. The relevance of situations with water parameters outside these boundaries for water quality monitoring in Germany is discussed in chapter 3. How is the extrapolation from species specific results to community level be assessed? The Species Sensitivity Distribution (SSD) approach is a well-accepted tool to extrapolate from the intrinsic sensitivity to a toxicant of a set of tested species to the whole set of species potentially exposed in the field. It is therefore used in different chemical regulatory frameworks as well as for EQS-setting under WFD. Its protectiveness against effects in (model) ecosystems has been confirmed in several reviews for different classes of chemicals (Emans et al. 1993, Okkerman et al. 1993, Posthuma et al. 2002, Maltby et al. 2005, van den Brink et al. 2006). Remaining uncertainty in the estimation of a PNEC or QS based on the SSD approach is considered by the application of an assessment factor between 1 and 5 on the median HC5 of the SSD and some criteria for selecting the appropriate assessment factor are given in the EQS TGD (2011). However, in the case of metals, additional uncertainty arises due to the need to apply BLM developed for a small set of species to the non-blm species. Per se, there is no reason to assume a biased error in this extrapolation, i.e. tendency to over- or underestimate toxicity for the non-blm species. As discussed before, the validity of this cross-species extrapolation of BLM has been tested and discussed in the risk assessment report. Spot checking for Ni revealed that the BLM could Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 26

27 predict the toxicity observed for non-blm species in almost all cases within a factor of two (see above). Which exposure pathways are considered by the use of the BLM and how relevant are other exposure pathways? The BLM tool only considers exposure via the water phase. Thus, they are related to the QS freshwater, eco. Potential risks related to contamination of sediment or biota (food chain) therefore have to be considered separately. Figure 4 gives an overview on the different QS and how an overall EQS should be selected (copied from EQS TGD 2011). Water Sediment Predators (secondary poisoning) Human health (consumption of fishery products) Human health (drinking water) Y Are derivation triggers met? (Sections ) Y Y N No further assessment required Derive QSfw, eco Derive QSsediment Derive QSbiota, secpois, fw Derive QSbiota,hh food Derive QSdw, hh Convert QSbiota, secpois, fw and QSbiota, hh food into equivalent water concentration (Section 2.5.1) Are back calculated QSfw, secpois and/or QSwater, hh foodlower (i.e. more stringent) than QSfw, eco and/or QSdw,hh? Adjust QSfw, eco to level equivalent to the back-calculated QSfw, secpois or QSwater, hh food Y N Adopt lowest QSfw (QSfw,eco or QSdw,hh *) as overall EQS Do not implement EQSbiota (it will not be sufficiently protective)** Adopt this as overall EQS Retain option to implement EQSbiota * QS dw,hh can only be adopted as the lowest QS water for waters intended for drinking water use ** unless monitoring in biota is strongly preferred. Under these circumstances, calculate QSbiota that is equivalent to lowest (i.e. most protective) QSwater and select this value as biota Figure 4: Overview of assessments needed and selection of an overall EQS (copied from EQS TGD 2011, Figure 2-3). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 27

28 Thus, if QS for other compartments are adjusted to concentrations in the water (i.e. for monitoring purposes), this might result in a lower concentration than the QS fw, eco, which then drives the overall EQS. Therefore, in the user-friendly BLM tool it would be more correct to use the term QS instead of EQS to make clear, that the BLM only covers the effects of exposure on the aquatic community via the water phase and estimates athe QS fw, eco. Concerning the WFD Ahlf & Heise (2009) state that generalized EQS would face a high uncertainty. Case studies show that ignoring sediment in the risk assessment schemes could lead to major risks. Hence, sediments should be included in EQS derivation, the metalloregion concept should be applied and metal-specific biodynamics should be considered using the Dynamic Multi-pathway Bioaccumulation Model (DYNBAM). Independent from the use of BLM to derive the QS fw, eco the need to determine QS for the benthic community, predators or human health has to be checked to derive the overall AA- and MAC-EQS. For example for Zn, these were considered not to be relevant (Zn fact sheet, UK Environment Agency, October 2010). For Ni, a draft EQS dossier from 2011 is available 6 where the EQS is driven by the QS for the pelagic community. QS for the benthic community are stated to be under development and QS for mammalian and avian predators as well as for human health expressed in µg/l were larger at least by a factor of 10) than (the one for the pelagic community. In a recently available draft directive, only the Ni AA (as bioavailable Ni concentration in the water) and the MAC EQS are given, but no QS for biota (EC 2012). The Cu RAR (2008) aims to derive PNECs. An EU EQS fact sheet for Cu is not available because Cu is no priority substance 7 ). According to the EQS TGD (2011) Biomagnification of metals in aquatic organisms is rarely observed and, if it does occur, it usually involves the organo-metallic forms of metals (e.g. methyl mercury) However, the assessor should examine their potential to biomagnify or cause secondary poisoning in food chains, even for inorganic metal forms. It is especially important to look for evidence of organometallic species being formed in some compartments, or if the range over which homeostasis occurs is relatively small (e.g. selenium). Therefore, a useful first step is to review the information available for the metal in question in order to assess whether an in-depth secondary poisoning assessment is needed. 6 Nickel EQS dossier 2011, prepared by the Sub-Group on Review of the Priority Substances List (under Working Group E of the Common Implementation Strategy for the Water Framework Directive): nces/supporting_substances/eqs_dossiers/nickel_dossier_2011pdf/_en_1.0_&a=d 7 However Cu is one of the seven metals for which AA-EQS are laid down in the OGewV (2011, see Table 2). According to the ETOX database the QS for copper is based on a report by Schudoma, D. (1994): Ableitung von Zielvorgaben zum Schutz oberirdischer Binnengewässer für die Schwermetalle Blei, Cadmium, Chrom, Kupfer, Nickel, Quecksilber und Zink, Umweltbundesamt, Texte 52/94 (ETOX: Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 28

29 A lack of biomagnification should not be interpreted as lack of exposure or no concern for trophic transfer. Even in the absence of biomagnification, aquatic organisms can bioaccumulate relatively large amounts of metals and this can become a significant source of dietary metal to their predators... SCHER (2010) has concluded that chronic BLM (p. 74 of EQS TGD), i.e. which implicitly include the dietary exposure route, have been developed and validated for a number of species and metals. This fact together with dedicated studies demonstrating the rather small contribution of dietary metal to the overall toxicity value, suggests that this concern can be considered to be of minor importance. Current QS in Germany for As, Cr, Cu, Zn are defined as concentrations in suspended particulate matter (SPM) or sediment (see Table 2). Under the assumption of equilibrium partitioning between the different compartments, metal concentrations in SPM are monitored as a surrogate to concentrations in the water phase. They can be calculated from water concentrations using (water body specific) partitioning coefficients. The other way around, QS water could be expressed as QS SPM. With respect to the use of QS related to SPM the EQS TGD (2011) states: The overall EQS for water that is derived as described above is expressed as a dissolved concentration. Water column EQSs may also be expressed as a total (dissolved + particulate) concentration or concentration associated with SPM. In most cases the dissolved concentration will be preferred. However, for substances that are highly adsorbed to suspended matter the EQS might be based on suspended matter concentrations, which can be more appropriate for calculating substance fluxes in river systems. For such substances, this may be preferable to expressing the EQS as a total water concentration because this is dependent on the highly variable suspended matter concentration in water (which is a function of seasonality, turbidity and so on) and so may be highly uncertain. Emission controls are usually based on total concentrations in discharges too. When faced with such situations, the assessor should agree the preferred method of EQS expression/compliance assessment with policy makers or river basin managers. Which are requirements with respect to amount of data, distribution and taxonomic coverage? Following the EQS TGD (2011) quality standards in the context of the WFD are derived via the SSD approach, if possible. Therefore, valid toxicity data for at least 10, preferably more than 15 species, should be available. In addition, the species in the SSD should cover at least the following taxonomic groups (fish, a second family in the phylum Chordata, a crustacean, an insect, a family in a phylum other than Arthropoda or Chordata, a family in any order of insect or any phylum not already represented, algae, higher plants). For Ni, NOEC or EC10 values for 31 species could be used to construct the SSD, for Zn data for 189 species were available. The requirements for taxonomic coverage were fulfilled in both cases, except that for Zn no higher plant data were available (but data for a multicellular algae species). In both cases BLM were developed for algae (Pseudokirchneriella subspicatus), crustaceans (Daphnia magna) and fish (Onkorhynchus mykiss). For Ni also a BLM for Ceriodaphnia dubia is available. Thus, the full BLM cover plants, invertebrates and vertebrates as well as three trophic levels. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 29

30 There are no specific data requirements related to the consideration of bioavailability in the EQS TGD (2011). Moreover, application of other approaches than BLM, such as speciation models and chronic regression models (e.g. Cd hardness correction), are also possible. For the use of the BLM approach, the availability of the basic BLM for algae, Daphnia and fish seems to be the standard in the EU. Van Sprang & Oorts (2011) report that these BLM are implemented for Cu, Zn, Ni, Co and Mn in the EU and that BLM for Al and Pb are under development. In contrast to this, in the USA, BLM have been developed only for invertebrates and fish, and until now, only the acute BLM for Cu are implemented in the regulatory framework (van Sprang & Oorts 2011). 3. Monitoring of environmental metal concentrations and evaluation of compliance with QS QS compliance data have to be determined in a legal framework. The data should be measured precisely following documented protocols so that the measurement uncertainty and the variability in possible interpretations are low. Relevant steps for metal determinations and subsequent bioavailability assessment are: Sampling, filtration, sample stabilisation, sample storage, measurement of relevant metal species, measurement of additional parameters for metal bioavailability assessment (ph, DOC- and Caconcentrations). Experience from laboratories shows that especially the sample preparation (on site-filtration and acidification) is a critical step (see chapter 5; Busch et al. 2007). How is the bioavailability determined via BLM? Which parameters are necessary? It is not within the scope of the current project to describe the principles of BLM (reviews were cited above). Here only the usage of the provided so called user-friendly BLM sheet is discussed. It is intended that the user-friendly BLM tools mimic the outputs of full BLM in a precautionary way. However, they require relatively few inputs and can readily be used. One of these BLM was presented at a workshop organized by EU commission DG Environment, UK and Netherlands authorities, and industry stakeholders. The presented BLM was also provided as a Microsoft EXCEL sheet. The provided EXCEL sheet is the basis for the discussions in this report (EU BLM EXCEL sheet; Biomet Tool_Spreadsheet_Workshop_ xlsm). In November 2011 a new version was distributed via the developer s website (Bio-met_bioavailability_tool_- v1.4_ xlsm; user guide: Bio-met 2011). However, this file could not be tested fully (only test of basic performance and sensitivity). As input into the EU BLM EXCEL sheet the dissolved water concentrations of the three metals copper, nickel and zinc is necessary. These values are the concentrations which have to be determined according to the WFD EQS directive 2008/105/EC ( In the case of metals the QS refers to the dissolved concentration, i.e. the dissolved phase of a water sample obtained by filtration through a Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 30

31 0.45 μm filter or any equivalent pre-treatment. ; EC 2008). These are routinely determined by water monitoring authorities. For the calculation of the bioavailable metal fraction three further parameters are necessary, i.e. ph, DOC- and Ca concentrations (if the hardness information is provided in another form, e.g. as sum of Mg and Ca, this value can be converted into Ca concentrations by applying an Microsoft EXCELbased hardness converter provided together with the EU BLM tool: _Hardness conversion tool_biomet.xlsm; this file was provided during the EU workshop and is also available on the Biomet web pages at The DOC input into the BLM sheet should be site-specific. In the EU BLM workshop report (David et al. 2011) median concentrations from at least eight sampling occasions are recommended. The Ca data should also be analyzed as dissolved concentration. However, in the data set provided by the federal state Nordrhein-Westfalen (see below) the Ca concentrations were mainly reported as total Ca concentration. At least in some samples where both parameters were available differences of up to 50 % lower levels were observed for dissolved concentrations. The derivation of the bioavailable metal fractions is described in the manual available from the Biomet web pages (Bio-met 2011; see also Appendix for the information retrieved from the Bio-met website). This states that internally look-up tables are applied (see above). This results partly in step changes of output data (instead of steadily changing result functions). For example, calculations were performed only for a low number of fixed Ca concentrations (1, , 40, 80, 200 mg/l; Biomet 2011). However, according to the provided information (David et al. 2011), the BLM applied are based on the EU RAR for the respective metals which cover a larger range. For all three metals generic QS were derived (Table 8). The output data produced by the EU BLM EXCEL tools (see Figure 5 and Figure 6 for examples) are described below (the description is mainly for the Biomet Tool_Spreadsheet_ Workshop_ xlsm; June 2011). The output presentation of the new tool available since November 2011 is different in some details (Figure 6). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 31

32 Sample Number Date INPUT (MONITORING) DATA Measured Measured Measured Copper Nickel Zinc Conc Conc Conc (dissolved) (dissolved) (dissolved) [µg/l] [µg/l] [µg/l] ph DOC [mg/l] Ca [mg/l] Local EQS (dissolved) [µg/l] RESULTS (Copper) BioF Bioavailable Copper Conc [µg/l] A A A B B B B B B B B B B B B B C C Figure 5: Input and output data from the EU BLM EXCEL sheet (Biomet Tool_Spreadsheet_ Workshop_ xlsm version; June 2011). Output data are only presented for copper. The generic QS for copper is 1 µg/l. The grey shaded area shows the input data for dissolved metal concentrations and those for ph values, DOC- and Ca concentrations. Further explanations are given in the text. The data designated as EQS are here considered as QS (see comment in footnote 2, page 3). RCR Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 32

33 Sample Number Date INPUT (MONITORING) DATA Measured Nickel Conc (dissolved) [µg/l] Measured Copper Conc (dissolved) [µg/l] Measured Zinc Conc (dissolved) [µg/l] ph DOC [mg/l] Ca [mg/l] Local EQS (dissolved) [µg/l] RESULTS (Copper) BioF Bioavailable Copper Conc [µg/l] A A A B Y B Y B Y B Y B Y B Y B Y B Y B Y B B Y B Y B Y C Y C Y RCR Notes Figure 6: Input and output data from the EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm; available since end of November 2011). Output data are only presented for copper. The generic QS for copper is 1 µg/l. The grey shaded area shows the input data for dissolved metal concentrations and those for ph values, DOC- and Ca concentrations. Further explanations are given in the text. The data designated as EQS are here considered as QS (see comment in footnote 2, page 3). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 33

34 Further explanations to the EU BLM output presented in Figure 5 and Figure 6: Local QS (as dissolved metal concentration in µg/l; the data in the output sheet are designated as EQS although they can only be considered as QS and not as overall EQS according to WFD criteria; see comment in footnote 2, page 3): QS local is the calculated dissolved concentration of a metal that is equivalent to the QS bioavailabe considering the local water conditions at the respective site. In cases where this value is representing sensitive conditions (i.e., high bioavailability) the output value is flagged. If input data are outside the validated range of the BLM (i.e., ph values below or above the boundaries of validity for that specific metal), data are also marked in the output. BioF - is the ratio of the generic QS bioavailable divided by the local QS. This value is always 1 or less. When the value is 1, it is marked in red text and effectively means the metal in the specific water conditions at this site is 100% bioavailable and the site is relatively sensitive. Bioavailable metal concentrations are calculated by multiplying the dissolved metal concentrations for the site by the BioF. Through the use of a BioF, differences in (bio)availability are accounted for by adjustments to the monitoring data, but the (generic) QS remains the same. Bioavailable metal concentration (µg/l) this is the concentration of metal that is bioavailable at the site or waterbody. This value is calculated by multiplying the dissolved metal concentration for the site by the BioF. The risk characterization ratio (RCR) is calculated by dividing the dissolved metal concentration by the QS local. A value of > 1 indicates a potential risk. In these cases the respective cell is red and the text white (data are only calculated if dissolved metal concentrations are filled in). For coverage of water hardness the EU BLM EXCEL sheet uses as input parameter for the hardness only the Ca concentration. If hardness is expressed as combined Ca and Mg concentration (e.g., as degree German hardness = dh), then first a conversion to Ca concentration has to be performed. For this a Hardness Conversion Tool is provided as additional tool. The contribution of Mg to the total hardness is calculated assuming that the ratio between Ca and Mg concentrations is as described in a study by Peters et al. (2011). This may introduce an additional uncertainty. This assumption was not further analysed here. The limiting factor for the use of BLM will often be the availability of DOC data (see results of the interviews in the section Cost benefit analysis of chapter 5). However, in some cases information for other sampling locations within the same water body, or surrounding water bodies, may be used as substitute of site-specific data. In such cases it is recommended to select a relatively low percentile (e.g., the 25th percentile) of the data in order to ensure that the resulting value is sufficiently protective (EU BLM workshop report, David et al. 2011). DOC concentrations may also be estimated from UV absorbance data or from dissolved iron concentrations. These methods are less accurate, but may allow a preliminary assessment where DOC information is not available (EU workshop report, David et al. 2011). In the BLM workshop report such a method for estimating DOC data from Fe concentrations is described. However, as Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 34

35 reference only a non-peer-reviewed poster is cited (Merrington et al. 2008). The suggested equations are (as cited in the EU workshop report, David et al. 2011): DOC (mg l -1 ) = * Fe (dissolved, mg l -1 ) r 2 = Eq. 1; log 10 (DOC, mg l -1 ) = log 10 (Fe, dissolved,mg l -1 ) r 2 = Eq. 2. It was not within the scope of the study to check the applicability of this approach. How are annual average data calculated from the original measurement values? All data from one year (e.g., monthly concentration data) are used for the mean value calculation (arithmetic mean). In cases of data below the respective limit of quantification (LOQ) special care has to be taken. Here the calculation of annual values from single has to follow the requirements of Directive 2009/90/EC (EU Commission 2009). For the calculation of mean values it is stated in Article 5: (1) Where the amounts of physico-chemical or chemical measurands in a given sample are below the limit of quantification, the measurement results shall be set to half of the value of the limit of quantification concerned for the calculation of mean values. (2) Where a calculated mean value of the measurement results referred to paragraph (1) is below the limits of quantification, the value shall be referred to as less than limit of quantification. If all necessary data (dissolved metal concentration, ph, Ca concentration, DOC) for each sampling date are available, for each sampling date a QS local is calculated (QS local is the calculated dissolved metal concentration equivalent to the site-specific QS bioavailable). Then, the single QS local are averaged for the whole year and a comparison is drawn to the averaged dissolved metal concentrations for risk characterization. Otherwise, all other parameters may be averaged for the respective period of the limiting parameter. E.g., if data for all parameters but DOC are available monthly and DOC only 3-monthly, the other parameters could be averaged also for 3-month periods and used for QS local calculations (if no large variations between the monthly data occur; i.e., appropriate outlier tests have to be performed). Then, the annual average from the four 3-months periods is calculated. However, if average values are used for the input parameters it is appropriate to consider how the data are summarized to provide the average values (EU BLM workshop report, David et al. 2011). Arithmetic mean values may be used for dissolved metal concentrations. However, some parameters follow a log-normal distribution and mean values may be an overestimation. Arithmetic means seem to be appropriate for ph (which is already log transformed), and Ca concentrations (if there are no large variations over the period of interest). However, it is recommended to apply median (50th percentile) DOC concentrations rather than arithmetic means because these will better represent the average concentration if the distribution of concentrations is log-normal (EU BLM workshop report, David et al. 2011). The arithmetic mean may result in an overestimation of the average DOC concentration. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 35

36 It is recommended to apply the same number of annual measurements at the selected sites for the BLM-based evaluation as with the current compliance monitoring system in Germany. This would allow a direct comparison of both approaches. Generally, the number of annual measurements should be higher (e.g., 12) if larger changes of dissolved metal concentrations or the additional parameters ph, Ca, and DOC occur throughout the year and less (e.g., four) if the site is known to show low variations for all parameters. How sensitive to the input parameters (ph, Ca, DOC) are the model results? The sensitivity of the site specific QS (designated as EQS local in the BLM tool) to the water conditions was analysed by two approaches: First, 500 parameter sets were created by selection ph, DOC and Ca concentrations randomly from the range of BLM validity for all three metals (ph: 6.5 8, Ca 5 87 mg/l) and the DOC range 1 17 mg/l as the typical range for EU water bodies according to the CU RAR (2008). Uniform distributions were assumed within the ranges and correlations between the water parameters were not considered. The QS calculated by the BLM tool were plotted against the single input parameters to show the effect of a single parameter including the variability of the other two parameters. The results indicate that DOC is the most important parameter -only for this parameter clear a clear trend overlaying the background variability of the two other parameters was found (Figure 7). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 36

37 Figure 7: Correlation of QS to parameters randomly selectedplots for The correlation coefficients were 0.76, 0.69 and 0.59 for Cu, Ni, Zn respectively. For Ni o positive correlation (0.20) was found for zinc, while for nickel the trend was decreasing (r=0.18). For copper the correlation was very low (r=0.02). Ca concentrations show no pronounced trend. In a second analysis, one parameter was systematic varied while the two others were kept constant. The results are presented in Figure Figure First, it was observed that the calculated QSlocal was not constant outside the defined boundaries for the three metals (e.g., the Cu QSlocal increased at Ca concentrations > 150 mg/l although the upper boundary is stated as 93 mg/l; see Figure 778, top). Since end of November 2011 a revised version of the BLM tool (Bio-met_bioavailability_tool_- v1.4_ xlsm; became available the tests were repeated and comparisons are shown in the following figures. The following differences were observed between the two versions: Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 37

38 ph dependence of Cu: the unsteady curve function for copper in the ph range for the selected test conditions was changed in the new version of the BLM tool. Now the QS local remains constant up to ph 6 and only then increases (Figure 8). ph dependence of Ni: the QS local range was changed. The previous version yielded for the selected test conditions QS local between about 3 and 9 µg/l; the new version yields QS local between 2 and 6 µg/l (Figure 8). Thus a higher number of exceedances has to be expected in comparison to the previous tool (relevant for the evaluations performed). DOC dependence of Ni: the QS local range was changed. The previous version yielded QS local between about 3 and about 90 µg/l for the selected test conditions; the new version yields QS local between 2 and about 80 µg/l (Figure 9 and Figure 10). Thus a higher number of exceedances has to be expected in comparison to the previous tool (relevant for the evaluations that have been performed in this study). Ca dependency of Ni: the new version of the BLM tool shows no Ca dependency of the QS local (constant output for the selected test conditions: 3.1 µg/l; Fehler! Verweisquelle konnte nicht gefunden werden.). For lower Ca concentrations this value is lower than calculated with the previous version of the tool (4 µg/l; Figure 12). Thus a higher number of exceedances has to be expected in comparison to the previous version of the tool (relevant for the evaluations performed). As a consequence of the revision of the EU BLM tool it is expected that especially the number of exceedances for Ni will increase (lower QS local, no protective function of Ca under certain conditions). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 38

39 Figure 8: top: ph dependence of QS local as calculated with the EU BLM EXCEL sheet (Biomet Tool_Spreadsheet_Workshop_ xlsm) provided at the EU BLM workshop in June 2011; bottom: ph dependence of QS local as calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November Boundaries are similar in both diagrams (see also Table 8). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 39

40 Figure 9: top: DOC dependence of QS local as calculated with the EU BLM EXCEL sheet (Biomet Tool_Spreadsheet_Workshop_ xlsm) provided at the workshop in June broad DOC range. bottom: DOC dependence of QS local as calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November broad DOC range. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 40

41 Figure 10: top: DOC dependence of QS local as calculated with the EU BLM EXCEL sheet (Biomet Tool_Spreadsheet_Workshop_ xlsm) provided at the workshop in June performance for low DOC concentrations. bottom: DOC dependence of QS local as calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November performance for low DOC concentrations. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 41

42 Figure 11: top: Ca concentration dependence of QS local as calculated with the EU BLM EXCEL sheet (Biomet Tool_Spreadsheet_Workshop_ xlsm) provided at the workshop in June broad Ca range. bottom: Ca concentration dependence of QS local as calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November broad Ca range. Boundaries are similar in both diagrams (see also Table 8). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 42

43 Figure 12: top: Ca concentration dependence of QS local as calculated with the EU BLM EXCEL sheet (Biomet Tool_Spreadsheet_Workshop_ xlsm) provided at the workshop in June performance for low Ca concentrations. bottom: Ca concentration dependence of QS local as calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November performance for low Ca concentrations. Boundaries are similar in both diagrams (see also Table 8). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 43

44 After contact with one of the developers of the EU BLM EXCEL sheet (G. Merrington, wca environment, UK) an additional BLM sheet was provided which is currently used by UK authorities. This UK BLM sheet uses a function for the calculation of the QS local for Cu, Zn, and Mn, and has no step changes in the output. The QS local is calculated applying a formula which uses ph, DOC, and Ca concentrations as variables (only for Ca concentrations < 1 mg/l a default value is applied). According to the information provided, the model for Cu will be revised. Thus only the respective Zn BLM of the UK BLM sheet was tested in this project (Mn was not in the scope of this project). The following formula is applied to calculate Zn QS local in the UK BLM sheet: Zn PNEC local = ((A * ph + (B)) * LN (Ca) + (C * ph + D)) * DOC^2 + ((E * ph + (F)) * LN (Ca) + (G * ph + (H))) * DOC + ((I * ph + J) * LN (Ca) + (K * ph + (L))) with A to L being numerical constants derived from a fitting of the original function with step changes to a steady function. As input ph values, Ca and DOC concentrations are required. In Figure 13 the dependence of the QS for Zn is shown for both tools. As to be expected, the UK-tool based on a fitted function results in smooth curves compared to the step patterns created by the user-friendly BLM tool. However, both tools give total different results for ph dependence (although absolute QS values are not quite different): the UK tool predicts a steady increase while the user-friendly tool based on look-up tables predicts a local minimum around ph 7 (Figure 13). However, it has to be considered that the UK tool is based on a UK dataset which is not the same as the dataset used in the EU report and for the user-friendly BLM tool. In addition, the EU report derives a PNECadd, aquatic and the user-friendly tool therefore a EQSadd while the UK tool has not considered background values (G. Merrington, pers. com. 2012). In the Zn EU RAR (2010) there is no figure for ph dependence of toxicity but in Heijerick et al. (2003), where the EC10 for D. magna shows an increase from ph 6 to 8.5. On the other hand, it is also said in this paper that ph dependence can be species-specific: Contradictory results have been reported on the effect of ph changes on metal toxicity. Like hardness ions, H + ions compete with trace metals for binding at cell surfaces.. Thus at low ph, more protons are available for binding and fewer metal ions will be bound to cell surfaces. An increase in ph not only reduces the proton concentration but also alters metal speciation distribution, leading to a decrease in free ion concentration. This metal fraction is generally considered to be the most bioavailable form for most metals. For fish some authors have reported increased acute Zn toxicity with a ph increase.. Schubauer- Berigan et al. (1993), reported the effect of three ph levels (6.3, 7.3, and 8.3) on the acute toxicity of five metals towards four invertebrate species. For Zn, the highest toxicity was observed at the highest ph with C. dubia, Pimephales promelas, and Hyalella azteca. Only for C. dubia could a continuous increase in toxicity be observed as a function of increasing ph. These results are not confirmed by the present study with D. magna: We noted a significant decrease in chronic zinc toxicity with increasing ph. The results of Schubauer-Berigan et al. (1993) are also not corroborated by Belanger and Cherry (1990), who reported a higher acute zinc toxicity for C. dubia at ph 6 compared to that at ph 9. (cited references see Heijerick et al. (2003)) Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 44

45 Keeping in mind that that for the analysis here always only one parameter was varied while the others were fixed at ph of 7.6, 2.5 mg/l DOC and 40 mg/l Ca, respectively, the following patterns were found: ph dependence was totally different for the three metals: The QS for Zn show a minimum around ph 7, while the QS for Cu showed a maximum around ph 7.5 and the QS for Ni decreased over the full range of valid ph values. QS values outside the validity range of ph kept constant at the value of the boundary. For all metals the QS increased with increasing DOC with a clear step pattern. The Ni QS was only slightly affected by the DOC. The Ni QS was not affected by the Ca concentration. The QS for Zn showed an increase up to around 10 mg/l Ca, then a decrease to a local minimum in the Ca range of 15 to 25 mg/l, constant QS from around 30 to 60 mg/l and then a step around 60 mg/l Ca to constant QS values up to high Ca concentrations. Also the Cu QS showed an unexpected behaviour. There was a general trend of (stepwise) decrease within the validity range, but with a small local maximum. For Zn and Cu, the QS were not kept constant to the boundary values outside the validity range. Comparison with a BLM tool based on functions fitted to the results of the BLM and used by UK authorities revealed smooth curves of the QS dependence and a total different pattern for the ph dependence of the Zn QS (however, absolute QS derived were not quite different). Thus, the example analysis of the sensitivity of the QS against changes in the input parameter conducted here revealed that before a regular use of the BLM tool a more detailed analysis of the QS dependence on changes in the input parameters should be conducted. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 45

46 local QS (µg/l) local QS (µg/l) local QS (µg/l) ph dependence (2.5 mg/l DOC, 40 mg/l Ca) Zn QS local (UK BLM tool) Zn QS local (EU BLM tool) ph DOC dependence (ph 7.6, 40 mg/l Ca) Zn QS local (UK BLM tool) Zn QS local (EU BLM tool) mg/l DOC Ca dependence (2.5 mg/l DOC, ph 7.6) Zn QS local (UK BLM tool) Zn QS local (EU BLM tool) mg/l Ca Figure 13: Comparison of Zn results from the BLM sheet provided at the EU workshop (versin of June 2011; solid line) and the UK BLM EXCEL sheet provided by G. Merrington (dotted line). QS values are µg/l, and Ca and DOC concentrations mg/l. For Ca concentration < 1 mg/l a default QS of 10.9 µg/l Zn is applied (this corresponds to the applied generic QS; shown as red bar in the bottom diagram). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 46

47 How do the models perform at the system boundaries? The ph and Ca ranges of the applied BLM (EU BLM EXCEL sheet EU BLM; Biomet Tool_Spreadsheet_Workshop_ xlsm) are quite narrow. Special care has to be taken to evaluate the performance of the models outside these boundaries (for ph values and Ca concentrations boundaries are set because these parameters affect organisms also directly while no boundaries were set for DOC concentrations because DOC values found in waters do not affect organisms). According to the EU BLM workshop report (David et al. 2011) the input values are capped at the validated range (i.e., even at higher Ca concentrations or ph values the maximum values as given in Table 8 should be applied by the BLM tool). Several cases can be differentiated: A) The dissolved metal concentration is below the generic QS In this case the BLM application generally seems not necessary. Thus it is also not relevant whether the ph values and Ca concentrations are within the boundaries. B) An upper boundary is exceeded (e.g., ph or Ca concentration above upper range) value According to the EU BLM workshop report (David et al. 2011) the upper boundary for Ca is not set because of negative effects on organisms but because it is assumed that above a certain limit the protective role of Ca to the organisms (by competition with metal ions for e.g. gill binding sites) results in no further advantage. However, it has to be checked whether there are indications of negative effects of relevant hardness conditions on organisms (e.g., in the extreme salt-rich waters of the rivers of Sachsen-Anhalt, see below). C) Values are below lower boundaries (i.e., ph and/or Ca concentration below the lower range value) ph values below the lower boundary of the respective BLM (Table 8) may be critical for organisms. On the one hand H + is competing with metal ions for the biotic ligand, on the other hand it reduces also the binding of metal ions on DOC. It has also to be considered that each change of ph value may result in changes of the speciation of metals (stability of different inorganic species depending on ph). The BLM workshop report (David et al. 2011) states that significant changes in speciation around the lower ph limit for the BLM for Cu, Ni, and Zn are unlikely. This assumption should be confirmed by appropriate calculations with chemical speciation models (e.g., WHAM, Tipping et al. 1994). To deal with this situation the worst case assumption would be to assume that under these conditions the total dissolved metal concentration is bioavailable. This would result in a QS local similar to the generic QS. As consequence, outside the boundaries a little change of e.g. of the ph value could cause a large change of the QS local. This may be a problem when data are averaged or when sites are compared. If ph or Ca concentrations are outside the validity range of the BLM tool, options might be: - application of the generic QS (maximum bioavailability) as a worst case; Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 47

48 - consideration of natural background values (if appropriate); - application of the full BLM model for the respective metal; - use of biotests with local species to investigate effects (or of standard test species with water sampled from the respective sites (David et al. 2011). However, this options seems not purposeful (laborious, results may not be easy to interpret) The preconditions or iterative steps to prefer one or the other of these options are neither determined in the EQS TGD (2011) nor in the guidance on environmental risk assessment for metals and metal compounds (ECHA 2008b). In the latter guidance all possible options are listed without prioritisation. How relevant are conditions outside the boundaries for water bodies in Germany? The validity range of the BLM tool for the three metals is summarized in Table 8. For a worst case assessment in the following the validity range is defined as the range of ph and Ca concentration where the EU BLM tool can be applied for all three metals. For DOC no specific range of validity is defined since it is assumed that even higher DOC values do not affect organisms adversely. Table 8: Validity range of the BLM tool for Cu, Ni and Zn and the generic QS used in the tool. The total validity range is the range where all three BLM are valid. ph Ca [mg/l] DOC Generic QS [µg/l] Cu Not limited 1 Ni Not limited 2 Zn Not limited 10.9 Overlapping validity range Not limited - Relevance of the validity boundaries of the BLM tool for water bodies in Germany Data sets for the federal states (Bundeslaender) Nordrhein-Westfalen, Sachsen-Anhalt and Baden- Wuerttemberg were analysed to test how often monitoring data can be expected to fall outside the validity range (thus, at least for one metal, the BLM can not be applied). Examples of the dynamics of ph and Ca are given for some sampling sites. For a check if these ranges are sufficiently wide for typical water bodies included in the regular monitoring in Germany, example data sets kindly provided by LANUV (Nordrhein-Westfalen), LUBW (Baden-Wuerttemberg), and LHW (Sachsen-Anhalt) were analysed for the distribution of ph, Ca and DOC concentrations. The results are listed in Table 9. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 48

49 Table 9: Results of the comparison of monitoring data for ph, Ca and DOC and the validity ranges of the EU BLM tool (Version of June 2011) for Cu, Ni and Zn. Percentage values given are for the fraction of data which are outside the respective boundary. Nordrhein-Westfalen Sachsen-Anhalt Baden-Wuerttemberg number of samples , 92, ph range ph < % 0 % 1.3 % ph > % 40.5 % 44.5 % Ca range mg/l Ca < 5.0 mg/l 2.9 % 0 % 3.1 % Ca > 88.0 mg/l 15.9 % 100 % 42.8 % DOC range mg/l For Nordrhein-Westfalen in total data for samples from 891 sampling sites were available. Figure 14 shows a map of Nordrhein-Westfalen with the covered rivers. The number of samples at each site varied (partly up to more than 100). For all samples, only a few (< 1 % for ph, < 3 % for Ca) were below the validity range of the BLM. However, a higher percentage of the samples showed ph or Ca values above the upper limit of the validity range (e.g. in 24 % of the cases the ph was above the limit of the Zn BLM). DOC concentrations were in the range of < 1 µg/l to 55 mg/l. Examples for ph and Ca concentrations over time for a few example sampling sites of the rivers Rhine, Emscher and Lippe are shown in the following diagrams (Figure 17 - Figure 20). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 49

50 Figure 14: Map of the rivers Rhine, Lipper, Emscher and Wupper in Nordrhein-Westfalen considered in this evaluation (courtesy of LANUV Nordrhein-Westfalen). For Baden-Wuerttemberg, 2597 samples from 2007 to 2010 were available for evaluation. ph ranged from 4.7 to 9.0 with 1.3 % below and 44.5 % above the validity range. Ca concentrations ranged from 1.5 to 188 mg/l with 3.1 % of the samples below and 42.8 % above the validity range. DOC concentrations ranged from 0.3 to 16.1 mg/l. The results for the Neckar at the sampling station Rottweil are shown in Figure 21. The ph is sometimes above the validity range, but mostly for Zn only. Ca concentrations were often below the validity range. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 50

51 Figure 15: Map of the Neckar river in Baden-Wuerttemberg considered in this evaluation (this map was retrieved from the website For Sachsen-Anhalt monitoring data from 2010 and 2011 for six sites were provided (rivers Unstrut, Laucha, Salza, Schlenze, Bode and Saale). The sites were selected because of the known high salt concentrations in these rivers. For several samples, not all three parameters were measured. In total 126 ph values, 92 Ca concentrations and 48 DOC concentrations were available. ph values ranged from 6.8 to 8.6, Ca concentrations from 94 to 910 mg/l and DOC concentrations from 2.1 to 5 mg/l. Thus, all ph-values were above the lower validity limit of the BLM tool. However, 40 % of the ph values were above the upper limit of 8. All Ca concentrations were above the validity limit for Cu and Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 51

52 Ni, only for Zn, concentrations for a few samples were within the validity range. DOC values were generally low. An example for the river Saale is shown in Figure 22. Figure 16: Map of the river Saale in Sachsen-Anhalt considered in this evaluation. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 52

53 Figure 17: ph and Ca concentrations measured in the river Rhine at Bad Honnef (km 640). The shaded areas indicate the range where all three BLM (Cu, Ni, Zn) are valid. Figure 18: ph and Ca concentrations measured in the river Rhine at Kleve-Bimmen (km 865). The shaded areas indicate the range where all three BLM (Cu, Ni, Zn) are valid. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 53

54 Figure 19: ph and Ca concentrations measured in the river Emscher at the mouth into the river Rhine. The shaded areas indicate the range where all three BLM (Cu, Ni, Zn) are valid. Figure 20: ph and Ca concentrations measured in the river Lippe at Wesel. The shaded areas indicate the range where all three BLM (Cu, Ni, Zn) are valid. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 54

55 Figure 21: ph and Ca concentrations measured in the river Neckar, sampling site Rottweil. The shaded areas indicate the range where all three BLM (Cu, Ni, Zn) are valid. Figure 22: ph and Ca concentrations measured in the river Saale, sampling site The shaded areas indicate the range where all three BLM (Cu, Ni, Zn) are valid. In total, the example data indicate that there are considerable amounts of samples where the ph is above the validity range, at least for Zn. However, also for Cu, ph was often above the higher limit of 8.5. ph values below the lower validity limit of 6.5 were rarely found. Higher ph values correspond Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 55

56 to lower H + concentrations (less competition on, e.g., gills) and thus potentially increased bioavailability. The sensitivity analysis has shown this for Cu and Ni (at ph > 7.5; see metal-specific profiles in Figure 8). In contrast to this, for Zn an increase of the QS with higher ph has been found, suggesting that the QS calculated for the upper boundary is protective for higher ph values, too. For Ni und Zn it should be discussed if the QS calculated at the boundaries could be extrapolated to higher ph values or if the generic QS should be used as the most conservative approach. However, the last option could result in a large change of the QS if the ph is just above the boundary. Despite that the representativeness of the data set provided for this analysis is not analysed. The data show that at least several German water bodies reveal Ca concentrations above the validity range. For example in all samples available for Sachsen-Anhalt the Ca was higher than 88 mg/l (these rivers were chosen because of the high-salt conditions and are not representative for Germany). However, as higher hardness decreases the toxicity of the metals due to competition of the cations for the binding sites, the QS derived for the upper limit of the validity range should be conservative for these situations. It is suggested to apply calculations with the full BLM for the datasets from these sites since also high concentrations of sodium are present. In a few cases, for example in the river Neckar, Ca concentrations were below the lower limit of validity. In these cases, it may be appropriate to use the generic QS for a conservative assessment. 4. Potential implications of the use of BLM for emission and immission balances in the future Plant permissions refer to certain loads of substances emitted to the river. For the assessment of metal loads of waters the BLM are not applicable since they only consider concentrations in the water phase (at least not for those metals of which a large fraction is bound to suspended particulate matter). Thus total water concentrations of metals need to be determined in addition to dissolved metal concentration. What are the analytical requirements for monitoring for the application of the EU BLM tool? For the application of the user-friendly BLM tool the following measurements are required (see also chapter 3): dissolved metal concentrations in water (this step requires on site-filtration through a 0.45 µm membrane filter; see restraints described by Busch et al. 2007), ph (measured on site, since changes may occur during storage), dissolved Ca concentration (to be measured along with dissolved metal), DOC. Instead of Ca concentrations also hardness data can be applied. A separate tool allows the conversion of hardness (i.e., sum of Ca and Mg concentrations; Hardness-conversiontool_biomet_v xlsx, available at to a Ca concentration. However, a further uncertainty is introduced if the latter option is chosen. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 56

57 How many measurements are necessary to obtain plausible results? Monitoring data sets from German surface water measurement sites revealed in most cases only low variations between monthly samples. However, flood events may cause differing results (e.g. for sites with salt loads which is diluted by high water). In general, no different strategy as for a total water phase / suspended particulate matter compliance monitoring seems to be necessary (i.e. the same number of measurements as currently). Generally, the number of annual measurements should be higher (e.g., 12) if larger changes of dissolved metal concentrations or the additional parameters ph, Ca, and DOC occur throughout the year and less (e.g., four) if the site is known to show low variations for all parameters. How many measured concentrations have to be below the QS to consider the increased uncertainty? In the concept of the BLM tool-based assessment no special request is made regarding the number of values above or below the QS required for compliance. The decision on possible exceedances is not taken on the yes/no decisions on exceedances of the individual measurement data from one year. Instead, a mean QS for the year is calculated from the available (e.g., monthly or quarterly) data. Thus the normal procedures are applied (for technical specifications see Directive 2009/90/EC, EU commission 2009). For averaging of e.g. monthly QS local and dissolved metal concentration the usual procedures are applied. Alternatively, for each measurement a risk quotient (see Figure 5 and explanations in the text) is calculated and finally this value averaged over the year (this seems the best option if monthly data for all parameters including ph, DOC, and Ca are available). However, in cases with obviously differing results between samples from different time points the plausibility of measurement and data aggregation should be checked carefully. Are there dominant parameters for the different metals? In a recently provided user guide the development of the BLM tool is described (Bio-met 2011): Of all required input parameters, the key parameters driving the HC5 calculation were identified by means of a combination of sensitivity analysis and expert judgment. The average outcome across Ni, Cu and Zn was that: ph, DOC and Ca (or hardness) have a moderate to major impact on HC5 estimation. Magnesium [Mg], Sodium [Na], alkalinity, Dissolved Inorganic Carbon [DIC], iron [Fe] and aluminium [Al] have a low to moderate impact on HC5 estimation (depending on the metal of concern) but can be reasonably accurately calculated from Ca or ph. Temperature, potassium [K], sulphate [SO4] and chloride [Cl] have a negligible to low impact on HC5 estimation. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 57

58 DOC, ph and Ca were therefore selected as key input parameters. For the purposes of simulations, small steps were taken for ph (0.125 ph units from ph=6 to ph=8.5) and DOC (0.15 log DOC units from DOC= 0.1 mg/l to DOC=100mg/L). The following values for Ca were selected: 1, , 40, 80, 200 mg/l. The following values were selected for Na: 14.12, 20, 40, 80 mg/l. The user guide (Bio-met 2011) gives some further information on the applied correlations and internal rules which are partly based on a paper by Peters et al. (2011). Assessment factors of 1 to 5 are applied for the selected HC5 for each of the metals to calculate QS local values (based on extent and taxonomic diversity of the available ecotoxicity data). Generic QS were established for each of the metals based on HC5 values under conditions of reasonable maximum bioavailability. Although situations (sensitive sites) may occur which result in higher metal bioavailability, QS local predictions will not result in QS local values below the generic QS (in these cases the generic QS is applied). If an input value for ph or Ca is outside the validated range a QS local prediction using the best-fitting combination of validated ph, DOC and Ca values is applied. Some of the unexpected outcomes of the tool found during the sensitivity analysis conducted in chapter 3 may be explained by these facts. Validity of the EQS estimation outside the BLM boundaries For Cd it is well known, that hardness is the driving factor for the bioavailability and thus, the use of specific QS for defined hardness classes was considered as a simple and pragmatic approach to consider bioavailability (see Table 1). For Cu, Ni, Zn the sensitivity analysis revealed that obviously no single parameter alone can explain the variability sufficiently (see also chapter 2, section Validity of the user-friendly EU BLM tool, and Bio-met 2011). For the assessment of the influence of the measurement uncertainty two aspects have to be considered. The measurement uncertainty of the dissolved metal concentrations can be assessed directly since the values are used for comparison with the calculated QS local. Ca, DOC and ph uncertainty are acting indirectly by influencing the result of the QS local calculation (error propagation). Figure 23 - Figure 25 show the influence of uncertainties of ph, DOC and Ca concentrations on QS local calculations. For the conditions tested (reference: ph 7.6, 40 mg/l Ca, 2.5 mg/l DOC) the Cu QS is more strongly influenced by ph than the Ni and Zn QS. Uncertainty of DOC measurements influences mainly the QS for Cu and Zn. Uncertainty of Ca seems to be less important (probably to the step increase; however, uncertainty influence is especially high in the range of these step changes (not visible in the diagrams due to low number of test values). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 58

59 QS local µg/l QS local µg/l Local EQS Cu [µg/l] Local EQS Ni [µg/l] Local EQS Zn [µg/l] ph Figure 23: Influence of the variation of the ph values by + 5 % on the calculated EQS (as error bar). Calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_- v1.4_ xlsm) downloaded from the Biomet website in November Local EQS Cu [µg/l] Local EQS Ni [µg/l] Local EQS Zn [µg/l] DOC mg/l Figure 24: Influence of the variation of the DOC concentration by 30 % on the calculated EQS (as error bar). Calculated with the revised EU BLM EXCEL sheet (Biomet_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 59

60 QS local µg/l Local EQS Cu [µg/l] Local EQS Ni [µg/l] Local EQS Zn [µg/l] Ca mg/l Figure 25: Influence of the variation of the Ca concentration by 10 % on the calculated EQS (as error bar). Calculated with the revised EU BLM EXCEL sheet (Bio-met_bioavailability_ tool_- v1.4_ xlsm) downloaded from the Biomet website in November In Figure 26 a test calculation is presented where input data for monthly data were varied (see also chapter 3 for assumed measurement uncertainty; dissolved metal concentrations + 20 % for Cu; + 30 % for Ni, Zn, DOC; + 10 % for Ca; + 5 % for ph). For this test data set the changes had only a low influence on QS local for Zn (mainly generic QS). For copper the variation in the test dataset was in the range of factors of 2-3 (e.g., data from 3/2010 variation between 1.4 and 3.9 µg/l; data from 4/2010 between 1.8 and 4.4 µg/l). For nickel the variation in the test dataset was in the range of a factor of about 1.5 (e.g., data from 1/2010 variation between 3.8 and 5.3 µg/l; data from 5/2010 between 3.0 and 4.9 µg/l). A more thorough investigation with a check of performance in other concentration ranges was not possible within the scope of the project. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 60

61 Date INPUT (MONITORING) DATA Measured Measured Measured Copper Conc Nickel Conc Zinc Conc (dissolved) (dissolved) (dissolved) [µg/l] [µg/l] [µg/l] ph DOC [mg/l] Ca [mg/l] RESULTS (Copper) RESULTS (Nickel) RESULTS (Zinc) Local EQS (dissolved) [µg/l] Notes Local EQS (dissolved) [µg/l] Notes Local EQS (dissolved) [µg/l] 1/ Y 1/ Y 1/ Y 1/ Y Y 1/ Y 1/ Y 1/ Y 1/ Y 1/ Y 1/ Y 1/ Y 1/ Y 2/ / / / Y / / / Y 2/ / / Y 2/ / / Y 3/ Y 3/ Y 3/ Y Y 3/ Y 3/ Y 3/ Y 3/ Y 3/ / Y 3/ Y 3/ Y 4/ Y 4/ Y 4/ Y 4/ Y Y 4/ Y 4/ Y 4/ Y 4/ Y 4/ Y 4/ Y 4/ Y 4/ Y 5/ Y 5/ Y 5/ Y 5/ Y 5/ Y 5/ Y 5/ Y 5/ Y 5/ / Y 5/ Y 5/ Y 5/ Y 5/ Y 5/ Y 5/ Y Figure 26: Assessment of the influence of the measurements uncertainty - variation of input parameters. Data sets for five month were varied (only one: either dissolved metal concentrations or one of the parameters ph, DOC, Ca per dataset). Calculated with the Notes Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 61

62 revised EU BLM EXCEL sheet (Bio-met_bioavailability_tool_-v1.4_ xlsm) downloaded from the Biomet website in November What are the consequences for downstream sites with possible higher bioavailability? The potential influence at downstream sites is discussed for the case of data for the river Wupper (data kindly provided by LANUV Nordrhein Westfalen; Figure 27). Here the calculated QS local are compared with the (total) metal concentrations of downstream discharges. Some discharge concentrations of copper are above the QS local of the respective upstream sampling site. Most input concentrations of nickel are above the QS local of the respective upstream sampling site. At one site the Ni concentration is above the QS local. Most input concentrations of zinc are above the QS local of the respective upstream sampling site. At three sites the annual mean Zn concentration is above the QS local. However, comparisons can only be qualitative since input data are total metal data measured in effluents and dilutions are not considered (no data on ratio of river flow and input flow rates). Also the effect of possible additional inputs of DOC (from STPs) is not considered in the BLM calculations performed here (not in the scope of this project). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 62

63 Figure 27: Dissolved metal concentrations at sampling stations, total metal concentrations of effluents, and QS local calculated with the EU BLM tool (version of June 2011) for the river Wupper (Nordrhein-Westfalen, Germany). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 63

64 Figure 28: Annual average dissolved metal concentrations for Cu, Ni and Zinc at sampling stations and QS local calculated with the EU BLM tool (version of June 2011) for the river Neckar (Baden-Wuerttemberg, Germany). Data for The sampling sites are Rottweil, Börstingen, Kirchentellinsfurt, Deizisau, Poppenweiler, Besigheim, Kochendorf, Mannheim (first three letters are used as abbreviations). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 64

65 5. Cost-benefit analysis for application of BLM based QS for compliance monitoring Based on interviews with persons from institutions involved in the WFD monitoring it is discussed whether the BLM approach yields advantages over the currently applied monitoring and measuring approach. Due to project limitations the interviews could only be undertaken with persons from two institutions (two other institutions contacted were not able to respond during the duration of the project). It was requested to answer a questionnaire. After receipt of the questionnaire some aspects were discussed in more detail during phone calls. The following questions were covered: Are the required additional parameters (dissolved Ca concentration or hardness, ph, dissolved organic carbon) routinely measured parallel to the dissolved metal concentration (< 0.45 µm membrane-filtered)? It is assumed that the total metal determinations will be continued because they are required for mass flow calculations. From one state it was reported that metals are already determined as dissolved concentrations in compliance with the Directive 2008/105/EC. The additional parameters are determined routinely as they are integral part of the basic measurement program. Only at some sites Ca has to be measured additionally. Routinely Ca is measured in the non-filtered water (whole water phase). However, if dissolved metals are determined, Ca can be analyzed in parallel. Differences for both values (dissolved vs. total Ca) are observed especially in high water situations. From the other federal state it was reported that routinely only total metal (whole water phase) concentrations are determined. DOC and Ca are also not determined routinely while ph values are measured routinely on site. Dissolved metals are only analyzed in cases where the total metal concentrations are above the current QS (in these cases also the other parameters are determined in the filtered samples). The reasons for this approach are problems with high blank values upon on site filtration for dissolved metal analysis. This was also the outcome of a feasibility study on WFD monitoring practice (Busch et al. 2007). Since only two states were contacted, it is not known whether the common practice is different in other federal states. What additional cost are expected for the performance of additional measurements? The cost estimations from the two federal states were in the range between about for each additional working step or parameter (on-site filtration, dissolved metals, DOC; in total about ). These costs will be additional because sampling and measurements of whole water phase will continue due to necessity for load calculation and sewage treatment plant (STP) surveillance. Deciding to measure only dissolved or whole water parameters at selected sites would hinder statewide comparisons and results presentations. SPM sampling is performed routinely at a number of larger water bodies since some organic priority substances have to be analyzed in SPM. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 65

66 Which benefit is expected by application of a BLM as compared to the current use of QS (e.g., less exceedances, identification of currently not-considered sensitive sites, better characterization of sites which are assessed for discharges? Expected benefit as compared to the current regulations: - Basically a uniform (and easy to communicate) approach for all metals/metalloids by applying dissolved metal concentrations would be desirable (e.g., currently some metals are determined as dissolved concentrations; others like Cu, Cr, Zn are quantified in SPM or sediment). - Samplings of filtered water are expected to get better incorporated in a state-wide monitoring. They are advantageous compared to laborious sediment and SPM sampling (especially at sites with low sediments/spm). - It is expected that the assessment based on dissolved metal concentrations can be better integrated into monitoring and assessment systems. Generally it is assumed that results would be better to compare, more comprehensible and more practical for the implementation by authorities. However, it was emphasized that this assessment regards the BLM principles but not the practicability of the BLM tool-based compliance monitoring itself. - It would be seen as a success if the current differences in metal measurements (current QS partly for dissolved water, partly for SPM and sediment) were harmonized, and all QS would be derived using the BLM approach (on base of the EQS TGD 2011). However, since only for a limited number of metals validated BLM are available, this is not expected soon. It was recommended to consider this aspect for future revisions of relevant regulations. - An advantage may be that the derivation of geogenic background values of metals would be required in fewer regions since local QS will be exceeded at fewer sites (e.g., at sites with relevant DOC and Ca concentrations). Experience reveals that the water body specific derivation of metal backgrounds is difficult especially in industrialized regions. Moreover, an accepted approach for metal background derivations is currently missing. Possible drawbacks are: - One concern is that the safety margin is reduced by application of BLM since no assessment factors are applied on the generic QS. It was stated that the total metal determinations in water will still be required for mass flow assessments and compliance measurements for releases of waste water (emissions). The BLM approach seems only adequate for local immission assessments. How are the additional costs assessed as compared to the benefit for the EQS approach vs. the BLM? Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 66

67 Possible benefits are currently difficult to assess. The highest benefit is expected from the incorporation of geogenic background values in the assessment. However, just this step is currently not implemented and sufficient data are not available. Therefore, at first higher costs are expected because of more efforts for sampling and analyses. Regarding possible measures in case of QS exceedances the views differed. In one state Ni and Zn QS exceedances are expected at a few sites (water bodies with especially high waste water fraction or in historical mining regions). The additional costs for BLM-based assessments were assessed low as compared to higher tier requirements for measures planning in case of QS exceedances. In the other federal state no demand for considerable additional measures by non-compliance is expected upon BLM-based assessments. Should the BLM be incorporated into laboratory information management systems (LIMS) to allow automatic calculations of bioavailable metal concentrations? Is alternatively the use of a separate Excel BLM tool appropriate which allows the QS local calculations for a large number of sites? It was not considered as necessary to incorporate the BLM tool into a LIMS. However, on the level of the risk assessment it is required to incorporate the BLM tool into the currently used expert data base system which is applied to issue the assessments, documentations and reports. A separate BLM tool would not meet practical requirements. Therefore it seems important that the applied BLM algorithm can be easily incorporated in existing software (i.e., no look up tables but defined formulas, e.g. for certain parameter ranges). Which measurement uncertainties have to be considered for the relevant parameters (dissolved Ca concentration or hardness, ph, dissolved organic carbon)? These data are important for assessing the uncertainty of the BLM tool calculated bioavailable metal concentrations. Measurement uncertainties depend on the height of the respective value. For dissolved metal concentrations in the range of the generic QSs, they are up to 30 % (e.g., 20 % for copper or 30 % for Ni at a 0.5 µg/l level; 30 % for Zn at 10 µg/l), for Ca concentrations at a 100 mg/l level about 5-10%, and for ph values about 5% for the whole relevant range), and for DOC at a 1 mg/l level about 30% (at higher values about %). The reported measurement uncertainties do not cover the uncertainties from sampling and on site filtration. Since DOC is analyzed as difference between total carbon (TC) and total inorganic carbon (TIC) it is especially prone to high measurement uncertainties in the lower concentration range (determination by infrared detection of carbon dioxide; inorganic carbon is released after mineral acid addition and organic carbon after thermal decomposition). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 67

68 Which additional aspects have to be considered as important for the implementation of the bioavailability assessment of metals by BLM? Generally the tiered approach is appropriate: #1 comparison with generic QS, #2 QS local, and #3 consideration of geogenic backgrounds. The validity of the toxicological derivation has to be proven (for all biological quality components, including macrozoobenthos, algae etc.). It has to be proven that other uptake pathways can be neglected (i.e., that uptake via ingestion is not relevant). It has also to be proven that the BLM based approach is comparable protective as the current QS compliance testing with SPM. It also has to be confirmed that the derived QS local are sufficient to cover also other protection goals (e.g., sediment organisms, bioaccumulation/secondary poisoning). It was commented that the acceptance of the BLM could be improved if the boundaries of validity of the BLM for ph would be more appropriate. Since ph values of up to 9 are quite common, BLM should cover at least a range up to ph 9. If the BLM tools in the current versions were applied it had to be confirmed that they are sufficiently protective in this range. For the easy-to-use BLM tools it has to be confirmed that the simplifications do not put the results at risk (e.g., is the Ca concentration a sufficient descriptor of the total hardness?). One important practical aspect is the required on site filtration of the water samples for the dissolved metals measurements. Here experience is important to avoid problems with high blank values (which may be in the range of the generic QS). This was also the result of a study performed in Germany on the implementation of the WFD monitoring (Busch et al. 2007). There is some concern that it may be too early to implement the BLM assessment now since only BLM for three metals are currently validated. Moreover, only in very few EU member states (e.g. UK, The Netherlands) experiences with incorporation of BLM into WFD compliance monitoring and assessments are available. It would be preferred to do this step by step. A drawback seems that currently no clear technical requirements for the practical implementation are available which guarantee Europe-wide common implementation and assessment. Clear guidance is required on how to proceed in cases the model boundaries of parameters (ph, Ca) are exceeded, what number of annual measurements is required (e.g., four or 12 data per year for the different parameters, depending on variability), and how to assess exceedances (are only QS exceedances of the annual mean relevant or also those of single values throughout the year). An appropriate approach should be developed regarding the consideration of whole water phase concentrations due to the necessity of load calculations for water bodies. It was suggested to perform a method comparison or proficiency testing on the member state level for the aspects sampling, on site filtration, ph measurements. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 68

69 6. Final discussion Based on the analysis and discussion presented in the previous chapters and after consideration of additional literature the initial six main questions are answered as follows: I. Are BLM generally appropriate for the risk assessment of metals and metal compounds? The BLM approach is the state of the art approach to consider bioavailability in the risk assessment of metals. The approach was described and evaluated at several expert workshops, in the peer reviewed literature, and has been applied in risk assessment reports of the EU and USA. It is also recommended in the recent guidance document on deriving water quality standards under WFD (EQS TGD 2011). The concept was also evaluated by the SCHER (Scientific Committee on Health and Environmental Risks). In its Opinion 3.9 (SCHER 2010) it is stated: Our attention has been drawn to concerns about the use of BLMs in determining the level of protection of water bodies from metals. Does the SCHER agree that the use of BLMs as advised in the guidance provides sufficient protection from the potential effects of metals? The number of studies that describe and demonstrate the development, validation and application of BLMs for various species and metals has grown considerably during the past decade. As such, the amount of scientific information which demonstrates the reliable use of these models for regulatory purposes is at least comparable to that on other (exposure and effects) models or approaches currently used in the EU. It is the opinion of SCHER that the two main concerns stated in the European Workshop on Metals in the Environment report (Annex 5), i.e. (1) geochemical conditions not covered by the approaches and (2) the relative importance of other pathways other than the free ionic form of metals, are sufficiently documented in open literature and/or addressed in the TGD document. Indeed, the TGD clearly states that the models should only be applied within their development/ validation domains, thus elevating concern 1. Additionally, chronic BLMs (p. 74 TGD), i.e. which implicitly include the dietary exposure route, have been developed and validated for a number of species and metals. This fact together with dedicated studies demonstrating the rather small contribution of dietary metal to the overall toxicity value, suggests that concern 2 can be considered to be of minor importance. Consequently, SCHER is of the opinion that the use of BLMs as advised in the guidance provides sufficient protection from the potential effects of metals. The use of BLMs was also explicitly supported in the final draft for the Ni EQS dossier (INERIS 2010): An approach accounting for nickel bioavailability provides an ecologically and environmentally relevant metric by which to assess potential nickel risks. Using bioavailability in a regulatory context provides an evidence-based way to assess compliance, set discharge limits and, importantly, to prioritise and rank locations at potential risk (EC 2010). The explicit ecotoxicological premise on which the BLMs are constructed provides a direct link to the goals and aims of the WFD. However, also if BLM generally seem to be appropriate for the risk assessment of metals and metal compounds the suitability of the approach for compliance monitoring purposes and especially of the Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 69

70 user-friendly tool proposed at the EU workshop in June 2011 (David et al. 2011) is a different aspect and was analysed in the previous chapters. Within this project an evaluation whether the protection level achieved for German surface waters by application of BLM is comparable to that what is maintained with the current practice was only possible for Ni (for the 2660 data from Nordrhein-Westfalen 3.75 % of the dissolved Ni concentrations were above the current WFD EQS of 20 µg/l; data mainly from one river system). For Cu and Zn the current system is based on suspended particulate matter (SPM) data since the German Surface Water Ordinance (OGewV 2011) states SPM based EQS for these (and some other) metals. II. Have metal background values to be considered additionally? Metals are not only present in water bodies due to anthropogenic pollution but also due to geochemical processes (e.g., erosion and weathering of minerals, precipitation of long-range transported atmospheric loads from volcanic activity). Therefore, it is possible that QS derived from laboratory tests (with application of safety factors) can be below the natural background concentrations in some regions. Regarding metals, the EQS TGD (2011) recommends a tiered approach to deal with bioavailability and natural background concentrations (Figure 29). In the first conservative step, a generic QS is used (thus, maximum bioavailability and no background concentration is assumed). If bioavailability models (e.g., BLM) are available these are applied in the next step. If the resulting QS is below the background level the added risk approach (ARA) should be used in step 3. This tiered approach is also supported by the SCHER (2010). However, it should be kept in mind that the derivation of background values is still to be harmonized. A comparison of the total risk approach (TRA) and ARA for metals is given in Figure 30. While the application of the ARA per se is not difficult the biggest challenge for its EU-wide harmonized use will likely be the determination of the natural background levels and at an appropriate spatial scale. In the Zn RAR the PNEC derived is a PNEC add, aquatic. Thus, also Zn EQS calculated by the BLM tool is an EQS add. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 70

71 Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 71

72 Figure 29: Schematic description of the derivation of QS and the consideration of natural background values of metals. Scheme from EQS TGD (2011). Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 72

73 Figure 30: Schematic description of the derivation of PNEC and the consideration of natural background values of metals. Scheme from MERAG (2007). (Note that this scheme is for prospective risk assessment (PNEC) but the general principle is also valid for QS estimation. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 73

74 Programme of Measures Classification No further action necessary The tiered approach has been followed in an EQS setting proposed by the UK Environment Agency (2010) where the scheme shown in Figure 31 is given. 1. Comparison with generic (100% bioavailable) EQS FAIL Pass 2. Use of screening tool FAIL 3. Use of NiBLM FAIL Pass Pass 4. Consideration of local ambient background concentrations FAIL Pass 5. Remedial measures Figure 31: Tiered assessment scheme for implementing an EQS for zinc (UK Environment Agency 2010) III. Which concrete advantages result from consideration of bioavailability for the implementation of protection targets in comparison to the current approach? The EU BLM workshop report states the following advantages for the use of BLM (David et al. 2011): The use of BLM as state of the art approach should allow a more realistic assessment of the hazard potential resulting from metals in water bodies. Unnecessary risk mitigation measures may be avoided at sites where the water conditions reduce the bioavailability of metals (e.g., high DOC concentrations). Therefore more resources (for risk assessment and mitigation) would be available for the high risk sites. On the other hand, the approach may also result in identification of sites potentially affected which would have been overlooked by the current approach. David et al. (2011) report an example from UK sites. Many soft waters in Wales had been considered as the most sensitive waters for metal exposures as based on hardness-banded QS. However, since these waters often have relatively high dissolved organic carbon concentrations, the risk from metals is only limited. On the other hand, calcareous (high hardness) streams of southern England Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 74

75 were identified by applying the BLM approach as potentially at risk because these waters had very low DOC concentrations and high ph (high metal bioavailability). Further aspects are discussed in chapter 5 (section Which additional aspects have to be considered as important for the implementation of the bioavailability assessment of metals by BLM? ). IV. In which manner have the methodological boundary conditions to be formulated to come to comparable monitoring results throughout Europe? Because the BLM approach aims to a site specific assessment only the approach can be harmonized (e.g. the consideration of bioavailability of metals by the use of the same BLM tool). In the EU BLM workshop report (David et al. 2011), it is stated that those Member States that have trialled the BLM approach (France, The Netherlands and UK) reported that there is generally a significant reduction in the exceedances of copper QS, when compared with existing (often hardness-based) QS. For zinc the experience was that number of exceedances decreased, but not to such extent as for copper. An interesting result reported is that often the locations of exceedances change when bioavailability is considered in comparison to the previous assessment approach (see example above from UK). Detailed assessment reports from two European countries (Sweden and UK) were available. UK Peters et al. (2009) assess the BLM tool application in the UK as follows: This collaborative project has developed and tested a simple, user-friendly version of the copper BLM with the purpose of providing a rapid screening tool to fit into Environment Agency monitoring and assessment systems. This model is not intended to replace the existing BLM, but to deliver a method requiring quick, low resource input, high data throughput and rapid interpretation of monitoring data. This project effectively transforms BLMs from the preserve of researchers into practical and accessible tools for regulators and stakeholders. Several hydrometric area and waterbody based scenarios are used to examine the implications of using the potential EQSs and BLMs when compared with existing standards. Default input parameters have also been used in these scenarios and their performance relative to the use of matched data is assessed. Consideration has been given to the use of water column ambient metal background concentrations within a compliance regime. The use of metal background concentrations and the BLMs within a simple tiered approach is also assessed, and a road map is provided for embedding these approaches and tools within a regulatory framework. The added risk approach was originally adopted for performing generic, large scale risk assessments, but may not necessarily be appropriate for application to EQSs. A more suitable approach for the initial tiers of an assessment may be to include a small contribution from the ambient background Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 75

76 concentration in the generic predicted no-effect concentration (PNEC), for example the 5th percentile of dissolved metal concentrations taken from monitoring from the hydrometric area. Sweden Concerns have been raised that BLM may not be applicable in Swedish surface waters due to large differences in water chemistry between Sweden and those countries for which the models have mainly been developed and validated. In a recent study the applicability of BLM for assessing metal toxicity in Swedish waters has been evaluated by the Swedish IVL in close collaboration with industrial partners (Palm Cousins et al. 2009). The study covered the metals copper, zinc, nickel, cadmium, and chromium. It is summarized: The overall conclusions of the project are that a) BLM-models can and should be applied as a natural element when assessing metal risks in Swedish waters, preferably as a sub-step in a tiered approach; b) The chemical conditions of Swedish water bodies are generally in agreement with the BLM requirements concerning the two key parameters ph and DOC, when it concerns the metals Cu, Zn, and Ni; c) Swedish national monitoring programs should focus on routine measurements of dissolved metal concentrations, and also on measurements of DOC. If the currently used method is regarded as equivalent to filtration through a 0.45 µm filter, this should be clearly stated in the national monitoring databases In order to assess the model results, a validation exercise should be performed for soft waters regarding Zn, in particular for soft waters with a high ph-value. However, this assessment was not based on dissolved metal concentrations (partly total metal data applied, partly calculation with full BLM considering also particle bound metals). V. Are the BLM suitable for the protection of aquatic ecosystems and scientifically valid? At least in case of the application of full BLM as those derived during the EU risk assessment for existing substances (Cu, Zn, Ni), the models reflect the current scientific knowledge. Usage of the models allows the reduction of the variability of the ecotoxicity data from a large number of studies within a range of 2. However, whether the simplified BLM tools are similar successful has to be proven. Vink & Verschoor (2010) evaluated several BLM which were developed in recent years: Biotic ligand models (BLMs) for heavy metals were developed in past years as tools for water quality assessment, following the Water Framework Directive (WFD). The essence of BLMs is that chemical speciation is incorporated in the assessment of ecotoxicological risks for aquatic species. For WFD purposes, BLMs are recognized as useful concepts to determine site-specific risks, and are allowed as second-tier assessment method. Concepts have been made operational, resulting in tools that make the approach available to a larger audience. As a consequence, these tools are in general simplifications of the original, validated concept. The goal of this study is to test the performance of Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 76

77 available tools that aim at predicting risks, or related water quality assessment, based on the concept of biotic ligand modelling, in order to aid in the recommendation for second-tier assessment method for heavy metals in surface waters. The tools that were tested in this study are: 1) BLM EU-RAR Validated BLM models for Cu, Ni and Zn, used in the European (Voluntary) Risk Assessment Reports. Extended and operationalized by Deltares, NL. 2) BLM (HydroQual) A simplified tool (version 2.2.3) based on the BLM EU-RAR, using statistical functions. Developed by HydroQual, USA. 3) BLM (WCA) A simplified tool (version 8) based on the BLM EU-RAR which makes use of statistical functions to determine bioavailable fractions. Developed by WCA Environment ltd, UK. 4) Ni-BLM (ARCHE) A simplified method to calculate potential no effect concentrations for Ni. Version 10, developed by ARCHE, BE. 5) Transfer functions (STOWA) Statistical zero-order relations of HC5 values and DOC concentrations, based on local data sets. Published by STOWA, NL. For the five tools that were tested in this study, a large dataset of water composition measurements was composed that was used as input for all selected models. Data for water chemistry and metal concentrations were collected from the ibever National monitoring database. This database was supplemented with data from various monitoring programs of different local water managers. The database contained a grand total of 2575 records with geographical information and surface water compositions in the Netherlands. These data covered most of the water types that are described in the Water Framework Directive. The database was constructed to contain all parameters that are required to perform the chemical speciation calculations and the BLM modelling. Vink & Verschoor (2010) concluded that none of the tested models met the defined quality criteria for reliability (in relation to the full BLM) although agreements were found due to trend correlation. As demands for a routine implementation for WFD monitoring Vink & Verschoor (2010) identified: 1) the derivation of statistically sound transfer functions for the majority of WFD water types and a broad range of taxa; 2) the implementation of metal background consideration based on the added risk approach; and 3) the feasibility testing of the BLM approach for transition waters. In a recent paper Verschoor et al. (2011) report on an alternative approach. They assessed spatial and temporal variations of metal levels considering water-type specific sensitivities for a range of aquatic species in The Netherlands. Cu, Zn, and Ni BLM were used to normalize chronic NOEC data determined in test media into site-specific NOEC for about 370 sites. Then, site-specific SSD were constructed. Sensitivity of species (as NOEC) and of the ecosystem (as HC5) for Cu, Ni, and Zn revealed variations of up to 2 orders of magnitude. The authors report that the application of spacetime specific HC5 for Cu and Zn resulted in a reduction of sites at risk, but in an increase of sites at risks for Ni. It should be considered whether this site-specific approach via locally adapted SSD should be tested for German waters, too. By this means a validation of the user-friendly BLM tool could be Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 77

78 performed. However, a decision may be taken after further evaluations with a refined (final) version of the EU BLM tool are available. VI. Which definition of substance are applied for the BLM since they are metal-specific? BLM are cation specific. In principle, all substances which produce the respective metal cation are covered by the respective BLM (e.g., all Cu cation releasing compounds are covered by a Cu BLM). Consequently, for REACh registrations transformation/dissolution tests according to OECD monograph 29 (OECD 2001) are required for metal compounds which allow an assessment on metal ions transformed and dissolved under relevant environmental conditions. Also in water monitoring and effluents compliance checking not certain metal compounds are characterized but only the total concentrations of the respective metals (only in cases where different metal-species exhibit different toxicities these may be differentiated, e.g. Cr(III)/Cr(VI)). Thus the BLM approach seems to be compatible to the current risk assessment of metals as well as to the WFD compliance monitoring in surface waters. 7. Acknowledgements The authors thank the colleagues from the German federal states institutions which provided monitoring data sets, participated in the interviews on practical aspects of BLM implementation, or made valuable comments to the first draft of the report: Christiane Heiss, Umweltbundesamt, Dessau Markus Lehmann, Landesanstalt für Umwelt, Messungen und Naturschutz Baden- Wuerttemberg (LUBW), Karlsruhe Ulf Nilius, Petra Kasimir, Landesbetrieb für Hochwasserschutz und Wasserwirtschaft Sachsen-Anhalt (LHW), Magdeburg Dr. Friederike Vietoris, Dr. Jens Rosenbaum-Mertens, Dr. Brigitte von Danwitz, Dr. Dieter Busch, Landesamt für Natur, Umwelt und Verbraucherschutz NRW (LANUV), Düsseldorf 8. References Ahlf W, Heise S (2009): Incorporation of Metal Bioavailability into Regulatory Frameworks. Forschungsbericht UBA-FB Umweltbundesamt, Dessau-Rosslau Bio-met (2011): Bio-met Bioavailability Tool. User Guide (version 1.4, ). Available at (after registration) Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 78

79 Busch D, Furtmann K, Schneiderwind A, Zyuzina I, Reupert R, Sielex K (2007): Einfluss von Probenahme und Probenvorbereitung auf die Ergebnisse bei der Bestimmung ausgewählter prioritärer Stoffe nach der Wasserrahmenrichtlinie. Landesumweltamt NRW, Düsseldorf, im Auftrag des Umweltbundesamtes, Dessau. Forschungsbericht , UBA-FB Cu RAR (2008): Voluntary risk assessment reports - Copper and Copper Compounds. European Copper Institute, Brussels, Belgium. /substance/429/search/ /term David, M., et al. (2011): Workshop on Metal Bioavailability under the Water Framework Directive: Policy, Science and Implementation of Regulatory Tools. Workshop Report, June 2011 (upload date ). ility&vm=detailed&sb=title EC (2008): Directive 2008/105/EC of the European Parliament and of the Council of 16 December 2008 on environmental quality standards in the field of water policy, amending and subsequently repealing Council Directives 82/176/EEC, 83/513/EEC, 84/156/EEC, 84/491/EEC, 86/280/EEC and amending Directive 2000/60/EC of the European Parliament and of the Council. Official Journal of the European Union, L 348/84, EC (2009): Guidance document No 19 - Guidance on surface chemical water monitoring under the. Water Framework Directive. Common Implementation Strategy for the Water Framework Directive (2000/60/EC). Technical Report European Commission, Office for Official Publications of the European Communities, Luxembourg. toringpdf/_en_1.0_&a=i EC (2012): Proposal for a Directive of the European Parliament and of the Council amending Directives 2000/60/EC and 2008/105/EC as regards priority substances in the field of water policy. COM(2011) 876 final. 2011/0429 (COD). Brussels, ECHA (2008a): Guidance on information requirements and chemical safety assessment. Chapter R.10: Characterisation of dose [concentration]-response for environment. European Chemicals Agency (ECHA). Helsinki, Finland. _08. Accessed on ECHA (2008b): Guidance on information requirements and chemical safety assessment Appendix R : Environmental risk assessment for metals and metal compounds. European Chemicals Agency (ECHA). Helsinki, Finland. Accessed on Emans H, Plassche E, Canton J, Okkerman P, Sparenburg P (1993): Validation of some extrapolation methods used for effect assessment. Environ Toxicol Chem 12, EQS TGD (2011): Technical Guidance for Deriving Environmental Quality Standards. Common Implementation Strategy for the Water Framework Directive (2000/60/EC). Guidance Document No: 27. Final version 8.0 after formatting. Prepared by EU, Member States and stakeholders. Version of July 2011 EU Commission (2009): Commission Directive 2009/90/EC of 31 July 2009 laying down, pursuant to Directive 2000/60/EC of the European Parliament and of the Council, technical specifications for chemical analysis and Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 79

80 monitoring of water status. Official Journal of the European Union, L 201/36, EU REACH Directive (2006): Regulation (EC) No 1907/2006 of the European Parliament and of the Council of 18 December 2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), establishing a European Chemicals Agency, amending Directive 1999/45/EC and repealing Council Regulation (EEC) No 793/93 and Commission Regulation (EC) No 1488/94 as well as Council Directive 76/769/EEC and Commission Directives 91/155/EEC, 93/67/EEC, 93/105/EC and 2000/21/EC. EU WFD (2000): Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000 establishing a framework for Community action in the field of water policy. Official Journal of the European Communities, L 327/1, Heijerick, D.G., C.R. Janssen and W.M. De Coen The combined effects of hardness, ph, and dissolved organic carbon on the chronic toxicity of Zn to D. magna: Development of a surface response model. Arch. Environ. Toxicol. 44, INERIS (2010): Ni EQS dossier. Draft version. Maltby L, Blake N, Brock TCM, Van den Brink PJ (2005): Insecticide species sensitivity distributions: importance of test species selection and relevance to aquatic ecosystems. Environ Toxicol Chem 24, MERAG (2007): Metals Environmental Risk Assessment Guidance. Fact Sheet no Incorporation of bioavailability for water, soils and sediments January. ICMM, Eurometaux, EURAS, and Defra (UK Department for Environment, Food and Rural Affairs), January Merrington G, Peters A, Brown B, Delbeke K, van Assche F, Sturdy L, Waeterschoot H, Batty J. (2008): The use of biotic ligand models in regulation: the development of simplified screening models and default water parameters. Paper presented at SETAC World Congress, Sydney, August 3-7th Meyer JS, Clearwater SJ, Doser TA, Rogaczewski MJ, Hansen JA (2007): Effects of Water Chemistry on Bioavailability and Toxicity of Waterborne Cadmium, Copper, Nickel, Lead, and Zinc to Freshwater Organisms. SETAC, Pensacola, FL, USA, 328 p. Ni EU RAR (2008): European Union Risk Assessment Report - Nickel and Nickel Compounds. Office for Official Publications of the European Communities, Luxembourg. Niyogi S, Wood CM (2004): Biotic Ligand Model, a Flexible Tool for Developing Site-Specific Water Quality Guidelines for Metals. Environ Sci Technol 38, OECD (2001): Chemicals Testing Monograph No. 29: Guidance Document on Transformation/ Dissolution of Metals and Metal Compounds in Aqueous Media. OECD Environment Directorate, Environment, Health and Safety Division, Paris, France. OGewV (2011): Verordnung zum Schutz der Oberflächengewässer (Oberflächengewässerverordnung - OGewV ) vom 20. Juli (German Surface water Ordinance). Bundesgesetzblatt Jahrgang 2011 Teil I Nr. 37, Juli Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 80

81 Okkerman PC, Vanderplassche EJ, Emans HJB, Canton JH (1993): Validation of Some Extrapolation Methods with Toxicity Data Derived from Multiple Species Experiments. Ecotoxicol Environ Safety 25, Palm Cousins A, Jönsson Å, Iverfeldt A (2009): Testing the Biotic Ligand Model for Swedish surface water conditions - a pilot study to investigate the applicability of BLM in Sweden. Report no. B1858. IVL, Stockholm. Paquin P R, Santore R C, Wu K B, Kavvadas C D, Di Toro D M (2000): The biotic ligand model: a model of the acute toxicity of metals to aquatic life. Environ Sci Policy 3, S175-S182 Paquin P R, Gorsuch J W, Apte S, Batley G E, Bowles K C, Campbell P G C et al. (2002): The biotic ligand model: a historical overview. Compar. Biochem Physiol Part C 133, 3-35 Peters A, Merrington G, Brown B (2009): Using biotic ligand models to help implement environmental quality standards for metals under the Water Framework Directive Science. Report SC080021/SR7b. Environment Agency, Bristol, UK. Peters A, Merrington G, de Schamphelaere K, Delbeke K (2011): Regulatory consideration of bioavailability for metals: simplification of input parameters for the chronic copper biotic ligand model. Integr Environ Assess Manag 7, Posthuma L, Suter GW, Traas TP. editors. (2002). The use of species sensitivity distributions in ecotoxicology. Boca Raton (FL): Lewis. 587 p. SCHER (2010): Opinion on the Chemicals and the Water Framework Directive: Technical Guidance for Deriving Environmental Quality Standards; Scientific Committee on Health and Environmental Risks (SCHER). October Schlekat CE, Van Genderen E, De Schamphelaere KAC, Antunes PMC, Rogevich EC,. Stubblefield WA (2010): Cross-species extrapolation of chronic nickel Biotic Ligand Models. Sci Total Environ 408, Tipping, E., 1994: "WHAM - A Chemical Equilibrium Model and Computer Code for Waters, Sediments and Soils Incorporating a Discrete Site/Electrostatic Model of Ion Binding by Humic Substance", Comp Geosci 20, UK Environment Agency (2010): Zn fact Sheet. UK Environment Agency. October Van den Brink PJ, Blake N, Brock TCM, Maltby L Predictive value of species sensitivity distributions for effects of herbicides in freshwater ecosystems. Human Ecolog Risk Assess 12 (4), Vink J, Verschoor A (2010): Biotic Ligand Models: availability, performance and applicability for water quality assessment. Report Deltares, Utrecht, NL. van Sprang P, Oorts K (2011): The use of Biotic Ligand Models in environmental risk assessments: an European perspective (...versus USA). Platform presentation during the SETAC North America meeting, November 2011, Boston, USA Van Tilborg, 2002 // Heijerick, 2006 // Euro-Ecole, 2002 Table p. 18 Zinc EU RAR (2010): European Union Risk Assessment Report Zinc Metal. Office for Official Publications of the European Communities, Luxembourg. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 81

82 Appendix Online documentation of the bio-met tool, accessed on February Scientific Development of the Biotic Ligand Model Regulatory authorities all over the world have developed or are developing methods to derive Environmental Quality Standards (EQS) and to evaluate the potential environmental risks of metals and man-made substances. EQS derivation and the risk assessment of metals in the water compartment is still predominantly based on total or dissolved concentrations (Bergmann and Dorward-King 1997, Janssen et al. 2000). However, it has been extensively demonstrated that neither total nor dissolved concentrations of metals are good predictors of their potential effects on ecosystems. Indeed, the physicochemisty of water, described by characteristics such as dissolved organic carbon (DOC), ph and hardness can modify the toxicity of metals by several orders of magnitude. The effect of water physicochemistry on metal toxicity is related to bioavailability. A metal is considered bioavailable when it is free for uptake by an organism and can react with its metabolic machinery, which may result in a toxicity response (Newman and Jagoe 1994, Campbell et al. 1988). The main idea behind bioavailability is that the toxic effect of a metal does not only depend on the total (or dissolved) concentration of that metal in the surrounding environment, but also on the complex interaction between physicochemical and biological factors. In other words, the same total or dissolved metal concentration does not result in the same degree of toxic effect under all environmental conditions. Overview of the Biotic Ligand Concept. (reproduced from If bioavailability is not taken into account any EQS for water based on total or dissolved concentrations of metal may be under-protective for one type of surface water and over-protective for another. Within the context of sustainable development, neither over nor under-protection is desirable as the former will result in increased societal costs involved with emission reduction, whereas the latter may result in harm to aquatic life and biodiversity. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 82

83 Despite the large body of evidence on the effect of water chemistry on metal toxicity generated over the past 10 to 20 years few regulatory systems have taken this into account. This is mainly due to a lack of quantitative tools. However, the recently developed Biotic Ligand Model (BLM) has gained interest from the academic, industrial and regulatory communities as this (conceptual) model is able to predict metal toxicity by integrating the most important effects of water chemistry. As such, the development of the BLM can be regarded as a milestone in the ecological risk assessment of metals. Although the foundations were laid in the early 1970 s, the reason for the current success of the BLM is that, for the first time, a model was able to integrate all available, interdisciplinary, knowledge on metal bioavailability into a generalized, visually attractive and easy-to-handle computerized framework. The BLM was originally developed to predict acute (short-term) toxicity to fish (Di Toro et al. 2001, Paquin et al. 2002) based on a combination of existing chemical, toxicological, biological and physiological data. More recently, and specifically in Europe, the focus has shifted toward predicting chronic toxic effects (i.e. not only effects on survival but also on growth and reproduction) to aquatic organisms belonging to different trophic levels such as algae, invertebrates and fish. This is indeed necessary if the BLM is to be used in the framework of EU risk assessment or EQS setting for metals in the WFD. References Bergman HL, Dorward-King EJ Reassessment of metals criteria for aquatic life protection. SETAC Press, Pensacola, FL, USA. Janssen CR, De Schamphelaere K, Heijerick D, Muyssen B, Lock K, Bossuyt B, Vangheluwe M, Van Sprang P Uncertainties in the environmental risk assessment of metals. Human and Ecological Risk Assessment 6: Newman MC, Jagoe CH Ligands and the Bioavailibility of Metals in Aquatic Environments. In: Hamelink JL, Landrum PF, Bergman HL, Benson WH (eds.) Bioavailability: Physical, Chemical and Biological Interactions. Lewis Publishers, Boca Raton, U.S.A. Campbell PGC, Lewis AG, Chapman PM, Crowder AA, Fletcher WK, Imber B, Luoma SN, stokes PM, Winfrey M Biologically available metals in sediments (NRCC No ), National Research Council Canada, Ottawa. Di Toro DM, Allen HE, Bergman H, Meyer JS, Paquin PR, Santore CS Biotic ligand model of the acute toxicity of metals. 1. Technical Basis. Environmental Toxicology and Chemistry 20: Paquin PR, Gorsuch JW, Apte S, Batley GE, Bowles KC, Campbell PGC, Delos CG, Di Toro DM, Dwyer RI, Galvez F, Gensemer RW, Goss GG, Hogstrand C, Janssen CR, McGeer JC, Naddy RB, Playle RC, Santore RC, Schneider U, Stubblefield WA, Wood CM, Wu KB. 2002a. The biotic ligand model: a historical overview. Comparative Biochemistry and Physiology C 133:3-36. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 83

84 Regulatory use of the Botic Ligang Model for Risk Assessment and EQS Derivation Biotic Ligand Models have been adopted under several European Regulatory Regimes. These are the Water Framework Directive (WFD) and the Existing Substance Regulation (ESR). EU Water Framework Directive EQS Derivation The Technical Guidance for deriving Environmental Quality Standards (EQS) under the Water Framework Directive (WFD) allows for bioavailability to be taken into account during the derivation of EQS for metals. The technical guidance can be downloaded directly from the EU Circa system by following this link. Nickel is currently classified as a priority substance under the WFD. As such a harmonised EQS has been set across the EU. A revised EQS for nickel, that takes account of bioavailability using BLMs, has been developed (based in part on the EU Risk Assessment conducted as part of the ESR). This EQS is currently undergoing final technical agreement. Copper and zinc are not identified as priority substances under the WFD. However, they can be identified by individual Member States as Specific Pollutants. EU Risk Assessment (RAR) under the Existing Substance Regulation (ESR) Zinc Zinc metal and five zinc compounds i.e. zinc oxide, zinc chloride, zinc sulphate, zinc phosphate and zinc distearate were prioritized as second priority substances under EU Regulation EEC/793/93 in September This implied that a full risk assessment for Zn needed to be carried out following the guidelines detailed in the Technical Guidance Document (TGD) on Risk Assessment for New and Existing Substances (EU, 2003). During this assessment, it became apparent that both chemical and biological processes may considerably affect the speciation of zinc, its bioavailability and therefore its toxicity. Strong recommendations were therefore formulated by the Dutch Rapporteur1 to develop chronic bioavailability tools i.e. Biotic Ligand Models (BLM), to evaluate quantitatively the manner in which water chemistry affects the speciation and biological availability of zinc in aquatic systems. These tools were further used for the derivation of normalized exposure concentrations i.e. Predicted Environmental Concentrations (PEC), and therefore further propagated in the risk characterization of the risk assessment. The draft final European Union Risk Assessment Report (EU RAR) on zinc (environmental part) was published in After the draft EU RAR was closed by the TCNES2 process in 2006, the SCHER3 provided on 29th November 2007 its opinion and concerns on the scientific quality of the report. The zinc RAR report can be downloaded from ESIS by following this link. Nickel The EU Commission listed nickel metal and nickel sulphate as 3rd priority substances in 1996 due to concerns expressed by the Danish government (D-EPA) about potential skin sensitisation (nickel allergic rashes). In 2000 nickel chloride, nickel nitrate and nickel carbonate were nominated on the 4th priority list. The Danish Environmental Protection Agency (D-EPA) was nominated by the EU Commission as Rapporteur for the nickel risk assessment. Similarly to the EU RAR on zinc, strong recommendations were formulated to develop chronic BLM tools for nickel in freshwater ecosystems. For nickel, the developed bioavailability tools were further used on the effects side for the derivation of normalized Predicted No Effect Concentrations (PNEC) and further propagated in the risk characterization of the risk assessment. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 84

85 The final European on nickel (environmental part) has become available in After the draft EU RAR was closed by the TCNES2 process in 2008, the SCHER3 provided on 13th January 2009 its opinion and concerns on the scientific quality of the report. The nickel RAR report can be downloaded from ESIS by following this link. Copper In 2000, the copper industry initiated a voluntary risk assessment (VRA). The assessment process was agreed with the Italian Government s Instituto Superiore di Sanita, acting as the review country on behalf of the European Commission and the EU Member States. Copper followed the same approach as nickel and has developed bioavailability tools that further used on the effects side for the derivation of normalized Predicted No Effect Concentrations (PNEC). In 2005, on behalf of the European Copper Institute, Italy submitted the draft risk assessment for review by the European Commission and EU Member States. In 2008, after three years of detailed analysis and improvement, this review process was completed. The final European on copper (environmental part) has become available in After the draft EU RAR was closed by the TCNES2 process in 2008, the SCHER3 adopted the risk assessment report at its 27th plenary on 13 January The voluntary copper RAR can be downloaded from the ECHA website by following this link. 1: The Rapporteur takes responsibility for the compilation of a dossier and subsequent report for consideration and approval by the TC NES before it is passed to the CSTEE for endorsement and the Article 15 Committee for publication as EU policy and any necessary risk reduction measures. 2: The TCNES is an EU Commission working group comprising technical experts representing Member States Competent Authorities (CAs) and chaired by a representative of the EU Commission. Their opinion on risk assessment reports is developed by consensus at TCNES. 3: EU Scientific Committee for Human and Environmental Risks (SCHER): Independent Scientific Committees providing the Commission with the scientific advice it needs when preparing policy and proposals relating to the environment. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 85

86 Development of the bio-met bioavailability tool (user-friendly BLM) Background The currently available software tools for undertaking Biotic Ligand Model calculations are data-demanding (more than 10 physico-chemical input parameters are required to run the models) and time-consuming (around one minute per sample/site). For some metals, the available software tools are also insufficiently user-friendly (for example for Ni and Zn, a combination of WHAM chemical speciation software and excel calculations are currently required). Software tools also require the installation of executable files on a user s PCs (which may not be acceptable on some corporate networks).these drawbacks are significant barriers to the regulatory acceptance and implimentation of Biotic Ligand Models for routine use in metals risk assessment in Europe and elsewhere. To address these barriers a user-friendly Biotic Ligand Model (bio-met bioavailability tool) has been developed as part of the bio-met initiative. The biomet bioavailability tool reports results obtained from Full Biotic Ligand Models, but has the advantages of only requiring three input parameters, being able to process large numbers of samples quickly and does not require installation on users PCs. Principle The basic approach behind the bio-met bioavailability tool is a large database of more than 20,000 different combinations of key input parameters (ph, Dissolved Organic Carbon [DOC] and Calcium [Ca] concentrations) and corresponding HC5 (Hazardous Concentration at 5% assuming a lognormal Species Sensitivity Distribution) calculations for Cu, Zn and Ni, using the full BLM. The database then serves as a lookup table. The physicochemistry of new sites are compared to existing simulations in the lookup table and the minimum HC5 of the two best matching lookup table entries is selected for local EQS derivation (after application of an additional assessment factor which varies between the metals). Development steps In the development of the bio-met bioavailability tool the following steps were taken: 1) Of all required input parameters, the key parameters driving the HC5 calculation were identified by means of a combination of sensitivity analysis and expert judgment. The average outcome across Ni, Cu and Zn was that: - ph, DOC and Ca (or hardness) have a moderate to major impact on HC5 estimation. - Magnesium [Mg], Sodium [Na], alkalinity, Dissolved Inorganic Carbon [DIC], iron [Fe] and aluminium [Al] have a low to moderate impact on HC5 estimation (depending on the metal of concern) but can be reasonably accurately calculated from Ca or ph. - Temperature, potassium [K], sulphate [SO4] and chloride [Cl] have a negligible to low impact on HC5 estimation. DOC, ph and Ca were therefore selected as key input parameters. For the purposes of simulations, small steps were taken for ph (0.125 ph units from ph=6 to ph=8.5) and DOC (0.15 log DOC units from DOC= 0.1 mg/l to DOC=100mg/L). The following values for Ca were selected: 1, , 40, 80, 200 mg/l. The following values were selected for Na: 14.12, 20, 40, 80 mg/l. Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 86

87 2) The next step was to select and/or derive the relationships to calculate the low to moderate impact parameters: Relationship Reference Mg (mg/l) = 10^( * log10(ca) ) Peters et al., 2011 Na (mg/l) = 10^( * log10(ca) ) Peters et al., 2011 Alkalinity (mg CaCO3/L) = 10^( * log10(ca) ) (for Zn, Ni) Peters et al., 2011 Alkalinity = 10^(1.0665*pH ) (for Cu) Calculated from De Schamphelaere & Janssen, 2004 Fe and Al are calculated based on speciation De Schamphelaere The negligible to low impact parameters were set at following reasonable worst-case values: temperature = 5 C., K = 25 mg/l, SO4 = 100 mg/l, Cl = 160 mg/l. 3) More than 20,000 simulations of different combinations of ph, DOC and Ca were simulated to calculate HC5 using the full BLM software. Metal Software reference Zn Hydroqual BLM version 2.12 and ARCHE semi-automatic spreadsheet processing script version Cu Hydroqual BLM version 2.12 and ARCHE automatic spreadsheet processing script version 1.3 Ni WHAM version and ARCHE semi-automatic spreadsheet processing script 4) The BLM models were developed and validated for relevant ranges for each physicochemical parameter. Metal ph Ca (mg/l) Zn Ni Cu When the user inputs a value for ph or Ca outside its validated range a prediction using the best-fitting combination of validated ph, DOC and Ca values is returned. 5) Additional assessment factors of between 1 and 5 were then applied to the selected HC5 for each of the metals to calculate Local EQS values. The size of the assessment factor reflects the relative uncertainty of the HC5 predictions to protect aquatic communities in the environment. The size of the assessment factor applied to the EQS is the based on the extent and taxonomic diversity of the available ecotoxicity data for each of the three metals. 6) Generic EQS (bioavailable) were also established for each of the metals based on HC5 values under conditions of reasonable maximum (worst-case) bioavailability. Local EQS predictions are returned as equal to the generic EQS under sensitive conditions. As such, Local EQS are not derived below the generic EQS. References Peters A, Merrington G, de Schamphelaere K, Delbeke K Regulatory consideration of bioavailability for metals: simplification of input parameters for the chronic copper biotic ligand model. Integrated Environment Assessment and Management, 7(3), De Schamphelaere KAC, Janssen CR Effects of dissolved organic carbon concentration and source, ph and water hardness on chronic toxicity of copper to Daphnia magna. Environmental Toxicology and Chemistry, 23(5), Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 87

88 Validation of bio-met biovailability tool (user-friendly BLM) Bio-met bioavailability tool (user-friendly BLM): Performance against Full Biotic Ligand Models (Version th of October, 2011) To evaluate the performance of the bio-met bioavailability tool against the full BLMs for copper, nickel and zinc, individual monitoring data points from different surface waters from UK (>900 sites) and The Netherlands (> 500 samples) were used to estimate HC5 using the full BLM/WHAM models and using the bio-met bioavailability tool. To achieve a perfect correspondence between the models, all data points should be on the 1:1 diagonal line. Many points above the 1:1 line would suggest that the bio-met bioavailability tool was lessconservative than the full BLM. Many points below the 1:1 line would suggest that the bio-met bioavailability tool was more precautionary than the full BLM. In addition to model predictions, the original BLM validation graphs for Daphnia magna describing the performance of the full BLM versus the observed toxicity in laboratory tests are also given for comparison purposes. Bio-met bioavailability tool (user-friendly BLM) validation results For all metals considered (i.e. Cu, Ni and Zn), the predictability of the full BLM by the user-friendly BLM is as or even more accurate (i.e. < factor of 2) and sufficiently conservative compared to the predictability of observed field toxicity by the full BLM. Therefore, the user-friendly BLM is sufficiently accurate. References Deleebeeck NME, De Schamphelaere KAC, Heijerick DG et al The acute toxicity of nickel to Daphnia magna: Predictive capacity of bioavailability models in artificial and natural waters. Ecotoxicology and Environmental Safety, 70(1), De Schamphelaere KAC, Janssen CR Effects of chronic dietary copper exposure on growth and reproduction of Daphnia magna. Environmental Toxicology and Chemistry, 23(8), De Schamphelaere KAC, Lofts S, Janssen CR Bioavailability models for predicting acute and chronic toxicity of zinc to algae, daphnids and fish in natural surface waters. Environmental Toxicology and Chemistry 24: Fraunhofer IME: Feasibility of BLM application 2nd DRAFT 88