The European Union Risk Assessment on Zinc and Zinc Compounds: The Process and the Facts

Transcription

1 Integrated Environmental Assessment and Management Volume 1, Number 4 pp Ó 2005 SETAC 301 The European Union Risk Assessment on Zinc and Zinc Compounds: The Process and the Facts Charles W.M. Bodar,*À Marja E.J. Pronk,` and Dick T.H.M. SijmÀ ÀNational Institute for Public Health and the Environment (RIVM), Expert Centre for Substances (SEC), `Centre for Substances and Integrated Risk Assessment (SIR), P.O. Box 1, 3720 BA Bilthoven, The Netherlands (Received 10 February 2005; Accepted 1 June 2005) ABSTRACT A risk assessment on zinc and zinc compounds was carried out within the framework of Council Regulation 793/93/EEC on Existing Chemicals. This risk assessment basically followed the European Union (EU) technical guidance documents (TGDs). These TGDs are built on the current knowledge on quantitative risk assessments, mainly for organic chemicals. This article describes the tailor-made approach for the zinc risk assessment. This work lasted almost a decade and involved the contributions of all EU member states and industry, who discussed the risk assessment during technical meetings. The risk assessment is initially based on scientific findings but is interrelated with pragmatic considerations. It follows a comprehensive approach, covering both environmental and human health. In the environmental part, new methodologies were developed to deal with the natural background of zinc, essentiality, speciation, and the use of species sensitivity distributions. The major results and the process of drawing conclusions of the risk assessment are outlined: potential environmental risks of zinc and zinc compounds may occur at local and regional scales in surfacewater, sediment, and soil. No potential health risks were identified for consumers and man indirectly exposed via the environment. For workers, potential health risks were identified only for zinc oxide and zinc chloride. Review Keywords: Zinc Risk assessment EC Regulation 793/93 INTRODUCTION Council Regulation 793/93/EEC (EC 1993) came into force in 1993 in the European Union (EU; see Table 1 for a full list of abbreviations). The underlying motive for this regulation was that the safe use of existing chemicals was still surrounded by many uncertainties. Existing chemicals are those that were on the market before The knowledge of the potential risks of existing chemicals is very often limited by large gaps in knowledge about (eco)toxicological properties and/or types of application (Allanou et al. 2000). This is in contrast to new chemicals, for which a notification procedure was introduced in the EU after The focus of EC Regulation 793/93 initially was on high production volume chemicals (HPVCs), i.e., existing chemicals with either a production or import in the EU larger than 1,000 tons/y. Since 1993 a number of risk assessments have been carried out for HPVCs, selected from the so-called EU priority lists. EU member states were appointed to take the lead in drafting a risk assessment report for one or more of the substances on the EU priority lists. Other member states, industry, and all parties involved are then invited to comment on these drafts. When the process of commenting and redrafting is completed, the Scientific Committee for Toxicity, Ecotoxicity, and the Environment (CSTEE; currently called the Scientific Committee for Human and Environmental Risks, or SCHER) will provide an opinion on the scientific quality of the report. Every EU risk assessment report within the framework of EC Regulation 793/93 formally ends with one of the following conclusions for each of the various protection goals: 1. there is need for further information and/or testing; * To whom correspondence may be addressed charles.bodar@rivm.nl 2. there is at present no need for further information and/or testing or for risk reduction measures beyond those which are being applied; or 3. there is a need for limiting the risks, and risk reduction measures that are already being applied shall be taken into account. It has to be decided if further information and/or testing are required to clarify the concern (conclusion 1) or if (further) risk reduction measures are necessary (conclusion 3). If risk of exposure is less than the hazard, the risk is considered to be negligible (conclusion 2). If a risk assessment report eventually indicates a potential risk, i.e., conclusion 3 for one or more endpoints, then a risk reduction strategy has to be implemented. Because it concerns an EU regulation, the legal consequences of the risk assessment are binding and directly applicable in all EU member states. In the case of zinc, it was The Netherlands (Cleven et al. 1993) that initiated the need for an international (EU) review process on the potential risks of zinc. For that reason, zinc metal, zinc oxide, zinc chloride, zinc distearate, zinc sulfate, and zinc phosphate (all HPVCs) were adopted on the second EU priority list (EC 1995). The EC regulation subsequently demands a comprehensive risk assessment for priority chemicals, covering both the environment and human health (consumer, worker, and man indirectly exposed via the environment). The main focus in this article is on the environmental part of the zinc risk assessment (EC 2004), because this section includes the main potential risks and new methodologies had to be introduced. The risk assessment for zinc and zinc compounds was the first in which a metal was assessed within the framework of EC Regulation 793/93. Such risk assessments are bound to the principles that are laid down in European Economic Communities (EEC) Regulation 1488/94 (EC 1994). The

2 302 Integr Environ Assess Manag 1, 2005 C.W.M. Bodar et al. technical guidance documents (TGDs) (EC 2003) serve as the practical application (cookbook) for conducting the EU risk assessment. These TGDs, however, are mainly built on the state-of-the-art of risk assessment for organic chemicals. The TGD does recognize that there are a number of fundamental differences between metals and organic chemicals, but only limited guidance is given on how to deal with these disparities. Performing the EU risk assessment for zinc and zinc compounds, therefore, meant breaking fresh ground as scientific and pragmatic solutions had to be found for a number of metal-specific issues. This article gives an overview of the selected approaches, highlighting the most important pitfalls. Specific attention will be paid to the natural background of zinc, essentiality, the added risk approach, speciation, bioavailability, and the application of statistical extrapolation methods in the effects assessment. For zinc, an extensive conclusion 1 research program was conducted to provide quantitative tools to incorporate the bioavailability of zinc in environmental media. The design, usefulness, and (non)applicability of this bioavailability correction has a central place in this article. The major results of the zinc risk assessment will be outlined, including the expected impact of this work on future EU risk assessments of other metals. Finally, we will discuss some dominant features of the lengthy process of realizing the zinc risk assessment within the EU. For the sake of completeness, the environmental part of the risk assessment currently has to go through a final written commenting round, mainly for editorial purposes. Subsequently, the SCHER will review the environmental risk assessment. The human health part was already discussed by the SCHER. GENERAL PRINCIPLES OF AN EU RISK ASSESSMENT General The structure of an EU risk assessment for priority chemicals is defined by a set of legal conditions. Regulation (EEC) 1488/94 (EC 1994) indicates that the risk assessment should proceed in the following sequence: hazard identification, dose (concentration) response (effect) assessment, exposure assessment, and risk characterization. The risk characterization comprises a quantitative comparison of the hazard identification/dose response assessment and the exposure assessment for the defined environmental and human protection goals (see below). Depending on the outcome of this comparison, every EU risk assessment within the framework of EC Regulation 793/93 formally ends up with 1 of the 3 conclusions (1, 2 or 3; see Introduction) for each of the various protection goals. Environment: The predicted environmental concentrations/ predicted no effect concentrations approach For the environment, the protection goals are aquatic ecosystems (including sediment), terrestrial ecosystems, top predators, micro-organisms in the sewage treatment systems, and atmosphere. The environmental risk characterization entails comparison of the predicted environmental concentrations (PEC) with the predicted no effect concentrations (PNEC) for the corresponding environmental compartment. If the PEC exceeds the PNEC, then it has to be decided whether further information and/or testing are required to clarify the concern or to refine the risk assessment (conclusion 1) or whether (further) risk reduction measures are necessary (conclusion 3). If the PEC is lower than the PNEC, then the risk is considered to be negligible (conclusion 2). Human health: The margin of safety approach The protection goals for the human species involve 3 distinct populations: workers, consumers, and man exposed indirectly via the environment. The human risk characterization is carried out by calculating a margin of safety (MOS); i.e., the ratio between the no-observed-adverse-effect level (NOAEL), or other no-effect or effect level, for a particular toxic endpoint (acute toxicity, irritation, corrosivity, sensitization, repeated dose toxicity, genotoxicity, carcinogenicity, and toxicity for reproduction) and the exposure estimate for the concerned population. The magnitude of the MOS determines whether there is a conclusion 1, 2, or 3. When the acceptability of the MOS is judged, account is taken of the various uncertainties in the extrapolation from experimental data to the human situation, in the available data set and in the exposure estimate under consideration. Decision process: Science and pragmatism As explained above, the Existing Substances Regulation (793/93/EEC) and Regulation (EEC) 1488/94 both form the legal basis for the risk assessment and risk management of existing substances. These regulations and the accompanying TGDs, however, do not exactly prescribe in detail how a risk assessment for every individual substance should be performed. Technical meetings have been established that are shared by the European Commission and in which all EU member states, industry, and interested parties participate to discuss the approach, progress, and conclusions of the risk assessment. Whereas one EU member state, the Rapporteur, is responsible for the first draft of a risk assessment report, it is the technical meeting that is responsible for drawing conclusions and the final report. Eventually, only member states have a voting right to draw the conclusions 1, 2, and 3. EU risk assessments are built on scientific principles, but, unarguably, they are interrelated with pragmatic considerations. Where does this pragmatism intrude into the risk assessment process? Principally, it boils down to the question of how uncertainties are dealt with during the decision process. Involved parties at the technical meetings often have different views on how much uncertainty is acceptable for them on certain risk assessment elements. In those conflicting circumstances, pragmatism (or conclusion 1) offers the way out without undermining a certain safety level. In addition to the decision process, it is emphasized that, strictly speaking, the TGD already contains several elements of built-in pragmatism. The TGD was originally developed by and extensively discussed with experts from member states, nongovernmental organizations (NGOs), and industry. Therefore, the TGD is, in fact, also a consensus outcome from a multistakeholder harmonization process (Bodar et al. 2003). It is very important to keep in mind that every EU risk assessment is a mixture of scientific and pragmatic elements, and the zinc risk assessment is, by no means, an exception. MAJOR ASSUMPTIONS FOR THE ZINC RISK ASSESSMENT General Zinc is present in the environment owing to natural processes, which results in a natural background concentration

3 EU Risk Assessment on Zinc Integr Environ Assess Manag 1, of zinc in all environmental compartments, including organisms. Zinc is also an essential element. Chemical and biological processes will furthermore affect the speciation of zinc in the environment and humans. These 3 aspects have fundamental implications for the exposure and effect assessment of zinc and, thus, for its risk characterization. Below we explain how these issues were dealt with in the current EU zinc risk assessment; however, zinc in the form of nanoparticles falls outside the scope of the EU risk assessment. Natural background of zinc Environment The TGDs do not provide detailed information on how to deal with (essential) elements, such as zinc, that have a natural background concentration in the environment. The added risk approach (Struijs et al. 1997) has therefore been introduced in the zinc risk assessment. In this approach, both the PEC and the PNEC are determined on the basis of the added amount of zinc, resulting in an added predicted environmental concentration (PEC add ) and an added predicted no effect concentration (PNEC add ), respectively. The use of the added risk approach implies that, in principle, only the anthropogenic amount of a substance in the environment (i.e., the amount added to the natural background concentration) is considered to be relevant for the risk assessment of that substance. Implicitly, a possible contribution of the natural background concentration to the adverse ecotoxicological effects is almost ignored. Although the natural background will be (partially) bioavailable, it is assumed that it will not lead to adverse effects. On the contrary, for an essential element such as zinc, part of the natural background must be bioavailable because it will provide organisms with sufficient essential metals. For zinc, Crommentuijn et al. (2000) studied the theoretical influence of the bioavailable fraction of the natural background of zinc in water and soil on the magnitude of the PNEC add, which they then called maximum permissible addition (MPA). They showed that bioavailability of the natural background of zinc in water had a negligible effect on the MPA. A high bioavailable natural background of zinc in soil could significantly affect the MPA. The added risk approach is a method that essentially could be used for all naturally occurring substances, including organic chemicals. In the environmental exposure assessment, the use of the added risk approach implies that the PEC add values have been calculated from zinc emissions resulting from anthropogenic activities. Thus, the PEC add is the anthropogenic part of the zinc concentration in the environment. By focusing only on the anthropogenic part of zinc, the problem of the variety of natural background concentrations of zinc over the different geographic regions is eliminated. In the environmental effects assessment, the use of the added risk approach implies that the PNEC add has been derived from ecotoxicity data that are based on the added zinc concentration in the tests. Thus, the PNEC add is the maximum permissible addition to the background concentration. From the background concentration (C b ) and the PNEC add, the PNEC can be calculated as PNEC ¼ C b þ PNEC add. Finally, in the environmental risk characterization, the use of the added risk approach implies the evaluation of the PEC add -to-pnec add ratios. In case measured environmental concentrations are used in the risk characterization, the background concentration has to be subtracted from that measured environmental concentration, resulting in a PEC add - to-pnec add ratio. Human health Because of the natural occurrence of zinc in the environment, humans (and animals) have a background intake of zinc via food, drinking water, and ambient air. In the human exposure assessment, this intake has been dealt with by creating a separate scenario for this natural exposure, as opposed to the exposure scenarios resulting from the anthropogenic use of zinc. In the human effects assessment, it was assumed that the background intake levels of zinc in experimental animal studies would be the same for humans. Essentiality Zinc is an essential element, which implies that organisms, including humans, will have a minimum requirement for zinc that supplies their needs, and a maximum concentration above which zinc is toxic. The minimum requirement is necessary because zinc plays an essential role in organisms (IPCS 2001). The range between the minimum and maximum is often called the window of essentiality (Hopkin 1993). Organisms have evolved mechanisms to supply their needs independently of the external concentration by regulating an essential element to a constant internal level (so-called homeostasis) (Rainbow and Dallinger 1993). However, when an organism is exposed to such a low level of an essential element that it no longer can regulate its internal concentration to cope with its needs, effects resulting from deficiency may occur. The use of the added risk approach in the environmental risk assessment implies that there is no risk for deficiency at the PNEC (PNEC ¼ C b þ PNEC add ), because the PNEC add derived in this approach is defined as the multiple permissible addition to the background concentration. The background concentration in a given ecosystem is partly bioavailable and provides the organisms in that ecosystem with sufficient essential metals, thus contributing to the existing biodiversity under prevailing natural conditions. Speciation and bioavailability of zinc Environment In addition to zinc being an essential element and occurring in the environment at a natural background, chemical speciation of zinc in the environment may be relevant for biological processes and, thus, for the assessment of potential risks. The speciation of zinc in the various environmental compartments will be briefly discussed below. We will also discuss how the various forms of zinc may affect bioavailability and toxicity to organisms. Finally, the assumptions that are made with respect to speciation and bioavailability, as well as the way it is implemented in the risk assessment, will be explained. Water Zinc in freshwater or seawater exists in suspended and dissolved forms and is distributed over a number of chemical species. In freshwater, zinc is mainly present in the dissolved form (i.e., as hydrated ions) in complexes of organic ligands (humic and fulvic acids), as zinc oxy ions, and when adsorbed to solid particulate matter (Cleven et al. 1993). In the generic Dutch fresh surfacewater, about 25% of the total zinc concentration is in the dissolved form and 75% is adsorbed to particulate matter (Cleven et al. 1993). Under anaerobic conditions and in the presence of sulfide ions, precipitation of zinc sulfide may occur.

4 304 Integr Environ Assess Manag 1, 2005 C.W.M. Bodar et al. In the zinc risk assessment report (EC 2004), the initial correction for bioavailability is based on the assumption that only the dissolved fraction is potentially bioavailable for aquatic organisms. The dissolved fraction is operationally defined as zinc in water that passes a 0.45-lm sieve. The larger particulate matter bound zinc, thus, is not relevant for the risk assessment. Following the assumption that the entire dissolved fraction is bioavailable would then lead to PEC add - to-pnec add ratios. 1 and to a conclusion 3 for many situations in the risk assessment. Because this assumption was scientifically challenged, i.e., not all but only a fraction of the dissolved zinc concentration in water would be bioavailable, the technical meeting decided to draw a conclusion 1. Studies under the conclusion 1 program were subsequently designed and conducted to quantify the effects of abiotic parameters on bioavailability and toxicity in water. Background concentration Adaptation to natural background levels and probably also to test conditions was shown in studies in which the sensitivity to zinc was affected up to a factor of 2 to 3 over a 3 orders of magnitude range of background concentrations (Muyssen and Janssen 2001, 2002). However, this refers to only a few individual studies, and even fewer studies showed the relationship between background concentration of zinc in water and its influence on toxicity on an ecosystem scale. In European freshwaters, background concentrations for zinc, expressed as dissolved concentrations, vary from approximately 1 to 40 lg/l. The risk assessment showed that there is no clear relationship between background zinc concentration in water and toxicity, based on a meta-analysis of data from the various relevant ecotoxicological studies, in which background concentrations varied in the same range as found in European freshwaters, and toxicity varied over 3 orders of magnitude (Figure 1). Therefore, background concentration may affect the sensitivity or tolerance of individual organisms, but this effect is relatively small compared with the larger variety of toxicity observed in multiple species and is most likely overshadowed by other influences. ph, hardness, and dissolved organic carbon The mitigating effect of ph, hardness, and dissolved organic carbon (DOC) on the aquatic toxicity of zinc was initially investigated in the process of drafting the risk assessment. However, for each of these individual abiotic parameters, there was much controversy on whether or not they individually mitigated toxicity (see Meyer 1999). Furthermore, recent insights on the speciation of metals in water, the binding of metals to biotic ligands and the resulting influence on aquatic toxicity led the technical meeting to decide to focus a conclusion 1 study on the development of biotic ligand models (BLMs). The technical meeting set conditions for this conclusion 1 work: the BLMs had to be developed under EU-related water conditions and had to be studied with chronic toxicity as an endpoint. BLMs have been proposed as a tool to evaluate quantitatively the manner in which water chemistry affects the speciation and biological availability of metals in aquatic systems. This is an important consideration because it is the bioavailability and bioreactivity of metals that control their potential to cause adverse effects. The BLM approach has gained widespread interest among the scientific, regulated, and regulatory communities because of its potential for use in developing water quality criteria and in performing aquatic risk assessment for metals. The BLM does this in a way that considers the influences important to site-specific water quality (Paquin et al. 2002). Figure 1. Relationship between the background concentration of zinc in water and toxicity expressed as the no observed effect concentration (NOEC). Figure 2 shows the BLM concept. Free zinc ions (Zn 2þ ) bind to the biotic ligand of organisms, which may be transport sites and/or toxic action sites. The concentration of zinc bound to the biotic ligand is directly proportional to the toxic effect and is independent of the physicochemical characteristics of the test medium. The chemical activity of Zn 2þ is, however, reduced by binding to organic (e.g., DOC) and inorganic ligands that reduce the bioavailability and, thus, reduce the toxicity. Inorganic ligands include OH and CO 2 3. The concentrations of these ligands are increased at increasing ph and alkalinity of the test medium, respectively. Cations in solution can compete with zinc for the biotic ligand, which also reduces bioavailability to the biotic ligand and, thus, reduces toxicity. The speciation of Zn 2þ, including the binding of Zn 2þ to DOC, is calculated by the WHAM V model (Tipping 1994), which is an integral part of the BLM method (Hydroqual 2002). Despite these promising possibilities of the BLM for metals in general, its applicability for the zinc risk assessment was initially considered as not sufficient. The main reason was that, until then, reliable BLMs for chronic effects were lacking for zinc, taking into account chronic (rather than acute) toxicity for all 3 determining taxonomic groups (i.e., fish, algae, and daphnids). The technical meeting concluded, however, that the possibility of developing an adequate BLM addressing these shortcomings was worth investigating for addressing bioavailability of zinc in water. The conclusion 1 research program was, therefore, launched during the risk assessment process, focusing on the bioavailability/blm issue in surfacewater. The conclusion 1 program provided information on the relationship between ph and chronic toxicity of zinc to Pseudokirchneriella subcapitata (algae), Daphnia magna (crustacean), and Oncorhynchus mykiss (rainbow trout) in a ph range of 5.5 to 8.5 (De Schamphelaere et al. 2003). In addition, chronic toxicity was studied in a hardness range between 24 and 250 mg/l as calcium carbonate (CaCO 3 ) (De Schamphelaere et al. 2003), and DOC was varied when taking a series of natural waters to test chronic toxicity in the same 3 organisms. In the analysis above, data were selected for only those European situations in which mean hardness is between 24 and 250 mg/l as CaCO 3. Because in some of the Scandinavian countries, and in other parts of Europe, hardness may be lower than 24 mg/l as CaCO 3, the effect of hardness on

5 EU Risk Assessment on Zinc Integr Environ Assess Manag 1, Figure 2. Summary of the concept of the biotic ligand model. toxicity was separately addressed for soft waters. The conclusion 1 program integrated all the results on the ph-, hardness-, and DOC-dependent toxicity and developed multivariate BLMs for each of the 3 organisms for bioavailability correction of the dissolved zinc concentrations in water (De Schamphelaere et al. 2003; Heijerick et al. 2003). The BLMs were further validated using different surfacewaters that are representative of the observed variation of physicochemical parameters in EU surfacewaters. The BLMs reduced the variation in toxicity (owing to differences in zinc bioavailability) from up to a factor 100 to a factor of 2 for all organisms studied, for both laboratory and field waters. This provided a good scientific basis for the incorporation of the BLMs into the zinc risk assessment. The BLMs are used to take into account bioavailability of zinc in surfacewaters, by correcting the PEC add s, where appropriate (BioF water PEC add, where BioF water is the bioavailability factor for water). In the zinc risk assessment, BLMs were applied to all 3 aquatic species (i.e., algae, Daphnia, and fish) for each relevant local site or a region X with known specific abiotic conditions (hardness, ph, DOC, etc.). Subsequently, the technical meeting decided that the most conservative value of the 3 possible BioF water values (i.e., the smallest correction for bioavailability) be selected. Then, the bioavailability-corrected PEC add and the risk quotient (RCR) were calculated: RCR ¼ PEC add, bioavailable / PNEC add. If no sufficient site or region-specific information on the abiotic parameters is available, then no bioavailability correction is possible. To further take into account some uncertainty in the various parameters, both average and rather worst-case settings of the abiotic conditions have been used in the zinc risk assessment. Table 2 shows that the BioF water generally ranges between 0.4 and 1 for a number of European regional surfacewaters. Because the BioF water for either fish or algae dominated the bioavailability correction, the BioF water values for Daphnia are not presented in Table 2. Sediment Adsorption to suspended matter and bed sediment is an important factor for the behavior and speciation of zinc in aquatic systems. Several phosphates, hydroxides, clay minerals, and organic matter are important for the adsorption of zinc in (an)aerobic waters (Cleven et al. 1993). Under anaerobic conditions, zinc may be precipitated in the sediment as zinc sulfide. It is often stated that the current approach of using wet or dry weight normalized metal concentrations has poor predictive power. Current discussions conclude that the mitigating effect of bioavailability of metals in anaerobic sediment should be taken into account. Metal speciation is highly affected by the acid-volatile sulfide (AVS) content in those anaerobic sediments. If the ratio of simultaneously extracted metals (SEM):AVS is smaller than 1, then the metals would not be bioavailable and would not cause any acute deleterious effects (Swartz et al. 1985; Allen et al. 1993). At a ratio above 1, effects could be expected. However, the molar ratio may not be a suitable predictor of potential effects, because the ratio gives no information on the absolute amount of SEM present in excess of AVS. Hence, the molar difference, SEM AVS, is a more suitable predictor of potential effects: [SEM] [AVS]. D.M. DiToro, J. McGrath, D.J. Hansen, and W.J. Berry (HydroQual, Mahwah, NJ, USA, unpublished data) suggested the use of (SEM AVS) / f oc as a basis for assessing the toxicity of metals in sediment, where f oc is the organic carbon content of the sediment. Shine et al. (2003) have recently evaluated zinc toxicity in sediments in the United States. The primary goal of their study was to compare different approaches and models used to estimate the toxicity of metals in sediments. The focus of the evaluation was on the extent to which a method was able to correctly classify a toxic sample as toxic and a nontoxic sample as nontoxic. The results of the SEM:AVS model evaluation showed that this approach has a very high sensitivity (96%); i.e., there is a high probability that the model correctly classifies a nontoxic sample as nontoxic and is regarded as protective of the environment. The SEM:AVS model provides a low positive predictive power of 55%. The latter means that in a large number of cases, exceeding the SEM:AVS ratio does not result in any observed toxic effects. This is not surprising because both the SEM:AVS threshold of 1 and SEM AVS threshold of 0 are not intended to predict toxicity but to foretell its absence. Thus, there seem to be many studies that indicate that the SEM:AVS concept may be useful for assessing the toxicity of zinc and other metals. Using an AVS approach could potentially take into account bioavailability of zinc in (anaerobic) sediments. Furthermore, not correcting for AVS would lead to many situations in the zinc risk assessment in which the traditional PEC add -to-pnec add ratios would exceed unity and, thus, would automatically lead to a conclusion 3. The other side of the picture is that there still are a number of critical comments on the SEM:AVS concept, which could limit its use. In some cases, uptake of metals was observed under conditions of SEM:AVS, 1 (Ankley 1996), which suggests that metals would be bioavailable. In addition, the experimentally determined SEM values may underestimate the actual concentration of metals (Cooper and Morse 1998), whereas the AVS values from pooled sediment samples may overestimate the actual AVS concentration in the top, aerobic sediment layer (Van den Berg et al. 1998). Furthermore, the SEM/AVS approach may neglect the role of dietary exposure. Through diet, metals are not necessarily taken up via the porewater but, additionally, directly from the sediment. By being aware of these potential shortcomings of the AVS approach and recognizing that otherwise many sediments could be mistakenly characterized as showing an unacceptably high risk quotient, the technical meeting decided that a conclusion 1 program should be established to further validate the AVS approach. A long-term field study was then conducted in European freshwater sediments for this purpose. The study focused on determining the tolerable zinc concentrations to benthic macroinvertebrate communities and on determining whether there is a relationship with AVS content (Burton et al. 2005). Results of the field evaluations at 4 EU test sites clearly showed that total zinc concentrations in

6 306 Integr Environ Assess Manag 1, 2005 C.W.M. Bodar et al. Table 1. List of abbreviations Abbreviation AF AVS BioF BLM C b CEC CSTEE DOC EC EC50 EEC EU EUSES f oc HC5 HPVC IPCS LC50 LOAEL LOEC MOS MPA NGO NOAEL OECD PEC PEC add PNEC PNEC add RCR SCHER SEM STP TGD TM WHO Term Assessment factor Acid volatile sulfide Bioavailability factor Biotic ligand model Background concentration (natural) Cation exchange capacity Scientific Committee for Toxicity, Ecotoxicity, and the Environment Dissolved organic carbon European Commission Median effect concentration European Economic Community European Union European Union System for the Evaluation of Substances Fraction of organic carbon Hazardous concentration for 5% of the species High production volume chemical International Programme on Chemical Safety Median lethal concentration Lowest observed adverse effect level Lowest observed effect concentration Margin of safety Maximum permissible addition Nongovernmental organization No observed adverse effect level Organization for Economic Co-operation and Development Predicted environmental concentration Predicted environmental concentration (added value, without natural background) Predicted no effect concentration Predicted no effect concentration (added value, without natural background) Risk characterization ratio Scientific Committee for Human and Environmental Risks Simultaneously extracted metals Sewage treatment plant Technical guidance documents Technical meeting (since 2003 called Technical committee of new and existing chemical substances, or TC-NES) World Health Organization sediments showed no relationship to benthic effects. The field evaluations confirmed the validity of the AVS model (i.e., further support could be found for the SEM:AVS approach) however the technical meeting concluded, contradicting the researchers, that no validation was found for the [(SEM AVS) / f oc ] approach. The technical meeting also recognized that some gaps in scientific knowledge still remained. For example, the significance of transport of metals into the food

7 EU Risk Assessment on Zinc Integr Environ Assess Manag 1, Table 2. Characteristics of several European Union regional waters for which bioavailability factors (BioF water ) are calculated with biotic ligand models for algae and fish (examples from risk assessment) a Region DOC (mg/l) ph Hardness (as mg CaCO 3 /L) BioF water a Meuse River (Belgium/Netherlands) 10P 50P 10P 50P 90P 10P 50P 90P algae 10/90 algae 50/50 fish 10/10 fish 50/ German rivers Elbe Main Mosel Flanders region (Belgium) Walloon Provinces region (Belgium) Netherlands region Rhine Meuse region (France) Rhone Med. Region (France) a BioF water values in bold represent the values that are used in the zinc risk assessment for the predicted environmental concentration, added value without natural background (PEC add ) correction for average and realistic worst-case conditions, respectively. Average conditions (50/50) make use of the 50th percentile value of dissolved organic carbon (DOC) and the 50th percentile values of the inorganics concentrations. Realistic worst-case conditions make use of the 10th percentile value of DOC and the 90th percentile values of the inorganics concentrations for algae, and the 10th percentile value of DOC and the 10th percentile values of the inorganics concentrations for fish. web from sediment ingestion, from the ingestion of contaminated benthos, or from the capability of organisms, such as polychaetes, to actively extract substances from sediments remained unanswered (Ankley 1994; Ankley et al. 1996; Mayer et al. 1996). Because both the wet or dry weight normalized PNEC approach and the AVS approach seem to have merits, as well as limitations, the following 2-tiered approach is used in the risk assessment of zinc in sediment. Under the 1st tier, the region- or site-specific risk of zinc in the sediment is assessed, Table 3. Results of the zinc environmental effects assessment for the various technical guidance document protection goals Protection goal Lowest NOEC/5th SSD a Assessment factor PNEC add b Surfacewater (lg/l; dissolved) 15.6 (5th SSD) (NOEC) Surfacewater, soft water conditions (lg/l; dissolved) c 7.8 (PNEC water) 2.5 d 3.1 Sediment (mg/kg dry wt) 74 (NOEC) 2 37 Soil micro-organisms (mg/kg dry wt) 27 (5th SSD) (NOEC) Soil, plants/invertebrates 52 (5th SSD) 2 26 (mg/kg dry wt) 32 (NOEC) Micro-organisms, sewage treatment plant (lg/l; dissolved) 5200 (EC50) a For water and soil, both lowest NOEC and 5th percentile of log-normal species sensitivity distribution (SSD; median confidence value) are given for comparison. b Values in bold indicate the overall predicted concentration (PEC), added value without natural background (PEC add ) that is used in the risk characterization. c Soft water is defined as,24 mg/l calcium carbonate (CaCO 3 ). d Average ratio between soft water and medium hardwater toxicity for algae, daphnids, and fish.

8 308 Integr Environ Assess Manag 1, 2005 C.W.M. Bodar et al. Table 4. NOAELs and LOAELs for use in human risk characterization a Toxic endpoint Zinc metal Zinc oxide Zinc distearate Zinc chloride Zinc sulphate Zinc phosphate Acute toxicity NEoC a NEoC NEoC NEoC Oral 1,100 1,260 mg/kg bw (LD50) ab,2,000 mg/kg bw (LD50) Inhalation 1,975 mg/m 3 (LC50) NEoC Dermal NEoC NEoC Irritation NEoC NEoC NEoC Skin NEoC Corrosive NEoC Eyes NEoC Corrosive Severely irritating Respiratory tract NEoC NEoC 1,975 mg/m 3 (LOAEL) NEoC For ultrafine particles in fumes 5 mg/m 3 NEoC 4.8 mg/m 3 (LOAEL) c NEoC (LOAEL) b Corrosivity NEoC NEoC NEoC Corrosive NEoC NEoC Sensitization NEoC Repeated dose toxicity Systemic effects 50 mg Zn 2þ /d (0.83 mg/kg bw/d) (NOAEL) a Genotoxicity NEoC Carcinogenicity NEoC Reproductive toxicity NEoC a LC50 ¼ lethal dose or concentration causing 50% mortality; NOAELs ¼ No observed adverse effect level; LOAELs ¼ lowest observed adverse effect level; NEoC ¼ no endpoint of concern; bw ¼ body weight. b Metal fume fever. c Respiratory tract irritation. based on the ratio of the PEC add and the PNEC add (i.e., based on wet or dry weight normalized zinc concentrations in the sediment). If the ratio is less than 1, then no potential risk can be assumed (conclusion 2). If the ratio is higher than 1, then a potential risk can be assumed and the 2nd tier should be followed. Background zinc concentrations in the sediment (taking into account AVS) as well as the contribution of those metals that bind more to AVS than zinc itself need to be investigated. If SEM-AVS, 0, then no potential risk is assumed (conclusion 2). If SEM-AVS. 0, then the excess zinc should be compared to the PNEC add, both expressed as either wet or dry weight concentrations. If the ratio is less than 1, then no potential risk (conclusion 2) can be assumed. If the ratio is higher than 1, then a potential risk (conclusion 3) is indicated. If no sufficient site- or region-specific information on the abiotic parameters is available, then only a default bioavailability correction of 0.5 for the PEC add sediment is used (see Table 5). The value of 0.5 is a compromise between available quantitative data on the ratio between total and AVScorrected concentrations of zinc in sediment in several European anaerobic sediments and, on the other hand, the qualitative information on the possible binding of zinc to oxides in more aerobic sediments. Soil In soil, zinc interacts with various reactive soil surfaces, such as soil organic matter, amorphous soil oxides (aluminum, iron, manganese), and clay minerals. Zinc in soil is distributed between the following fractions (IPCS 2001; Van Riemsdijk 2001): 1. Dissolved in porewater (which includes many zinc species); 2. Exchangeable, bound to soil particles; 3. Exchangeable, bound to organic ligands (of which a small part in the dissolved fraction and the major part in the solid fraction); 4. Present in secondary clay minerals and metal oxides or hydroxides; and 5. Present in primary minerals. Lock and Janssen (2003) showed that the comparative toxicity of a zinc salt, zinc metal powder, and zinc oxide to Eisenia fetida (earthworm), Enchytraeus albidus (potworm), and Folsomia candida (springtail) was similar and could be attributed to that of the zinc ion. Zinc compounds and zinc metal can be assumed to be converted into the zinc ion that subsequently may interact with various soil surfaces. These findings support the approach in the risk assessment in which irrespective of which zinc (i.e., zinc metal or salt) is emitted into the environment, all concentrations are recalculated to the zinc ion. Current discussions in the literature conclude that toxicity observed in the laboratory are generally higher than that observed under field conditions. In addition, it is often found that toxicity depends on soil type or soil properties. These latter 2 discussions are related to possible mitigating effects of

9 EU Risk Assessment on Zinc Integr Environ Assess Manag 1, Table 5. Risk characterization of zinc in sediment for a number of European Union regions. In Step 1, the measured PEC is corrected for the natural background concentration, resulting in a corresponding PEC add. In Step 2, the PEC add is corrected for bioavailability, using either the SEM/AVS approach or the generic bioavailability correction factor of 0.5. The results of the 2nd tier risk characterization are given in Step 3. PECs refer to 90th p values unless stated otherwise. SEM/AVS ¼ simultaneously extracted metal/acid volatile sulfide; PEC ¼ predicted environmental concentration; PEC add ¼ PEC, added value without natural background; me ¼ median; av ¼ average; mx ¼ maximum. Region Rhine (Lobith, Netherlands) PEC (mg/kg dry wt) Step 1 (Natural background correction) a PEC add Step 2 (Bioavailability correction) SEM/AVS data: no PEC add corrected (default factor 0.5) SEM/AVS data: yes Step 3 (Risk characterization) PEC add /PNEC add (corrected) Germany, various rivers Elbe 1,696 1, Main Mosel 1, Mulde 3,230 3,090 1, Rhein Saale 2,519 2,379 1, Spree 1, Sweden Northern Sweden 150 (me) Southern Sweden 240 (me) France Artoie, Picardie 1,200 1, Rhin, Meuse 1,908 1, Seine, Normandie Loire, Bretagne Flanders, Belgium b 268 c Specific data on SEM/AVS for 200 sampling points 12% of 200 sampling points where excess zinc/pnec add. 1 Meuse, Belgium Dave Vise Hollandsch Diep, 293 (av) Netherlands 4,003 (mx) 3,863 1, Dordtsche Biesbosch, 1,131 (av) Netherlands 2,802 (mx) 2,662 1, a A generic natural background correction of 140 mg/kg dry wt is assumed, unless otherwise noted. b The natural background concentration is 69 mg/kg dry wt. c Without using SEM/AVS method, the PEC add /PNEC add amounts to 7 (first tier). bioavailability of zinc in soil, which initially were not taken into account in the risk assessment of zinc. Subsequently, the technical meeting decided that a conclusion 1 study be carried out to study the effects of aging or laboratory-to-field differences, as well as the effects of soil properties on zinc bioavailability and toxicity to terrestrial flora and fauna. It is often assumed that aging may play a significant role in soil, affecting the bioavailability of zinc to terrestrial

10 310 Integr Environ Assess Manag 1, 2005 C.W.M. Bodar et al. organisms. After addition of a metal to a soil, it may be argued that part of the metal will be incorporated into the soil matrix (almost) irreversibly and poorly available for organisms. It is emphasized, however, that aging in soil is not only a timerelated process, as the word suggests, but also the sum of various processes involved, such as leaching, incorporation in soil matrices, effects of changing redox conditions, and changing minerals. The related assumption is that the fraction of zinc that is incorporated in the soil matrix, which in turn may be the result of aging processes, is less or not bioavailable (EC 2004). Based on various studies from the literature, the risk assessment on zinc showed that if zinc has been in soil for more than 1 y, the actual bioavailable concentration was only one third or less of that of freshly added zinc (Smit et al. 1997; Smolders, McGrath et al. 2003). Most emissions of zinc to soil addressed in the zinc risk assessment should have been gradually built up over years. Therefore, the technical meeting decided that the PEC add s in the risk assessment be divided by a factor of 3 to correct in a generic and pragmatic way for the aging aspects. An additional correction for bioavailability for the soil type is subsequently evaluated in the risk assessment of zinc based on the assumption that soil parameters may further affect bioavailability. Background concentration of zinc Adaptation to natural background levels may influence the sensitivity to zinc. However, few studies have looked at the relationship between background concentration of zinc in soil and its influence on toxicity. The risk assessment showed that there is no clear relationship between background zinc concentration in soil and toxicity, based on a meta-analysis of data from various sources. However, some recent studies showed weak, but still significant, relationships between 2 microbial endpoints, i.e., substrate induced respiration (SIR) and potential nitrification rate (PNR), and background concentration of zinc in 15 different European soils (Lock et al. 2003; Smolders et al. 2004). Therefore, the technical meeting concluded that there is a basis for using the background concentration of zinc to correct for bioavailability in soil for microbial endpoints, but not for other endpoints such as terrestrial invertebrates and plants. The relationships between background concentration and the effects on microbial endpoints were used to convert all the data on microbial effects in the ecotoxicity database to background-normalized data. Subsequently, the 5th percentile value of the species sensitivity distribution (HC5) was determined (see Effects Assessment). The ratio between the HC5 of the normalized data and the HC5 of the raw data was used as a bioavailability factor for soil microbial toxicity (BioF soil ). ph and cation exchange capacity Normalization of zinc concentrations in soil to ph may be applied because ph significantly affects the porewater concentration of zinc (McLaughlin and Smolders 2001). The basic physicalchemical idea of this ph normalization is that the resulting porewater concentrations would form a better basis for interpreting ecotoxicity data (Janssen et al. 1997). At present, evidence for porewater-related uptake of metals is present only for a limited number of plant species and for microorganisms. Other soil-dwelling organisms such as invertebrates are probably exposed via a combination of uptake routes, such as porewater, food, and direct ingestion of soil (Lock and Janssen 2001). Uptake by soil organisms and bioavailability from the soil is not, or is only poorly, related to ph for zinc. The cation exchange capacity (CEC) of a soil is a measure of how much cations, including heavy metals, can be kept from the soil solution. CEC and ph often are collinear; when CEC increases, so does ph, and vice versa. Recent studies (Lock et al. 2003; Smolders et al. 2004) have shown significant and strong relationships between CEC (and ph) of a soil and its influence on zinc toxicity for invertebrates and plants. These studies investigated zinc toxicity in 15 European soils. Based on these data, it is concluded that the storage capacity of a soil for cations does have an important mitigating effect on the bioavailability and toxicity of zinc for invertebrates and plants. Porewater concentration Biological availability is often thought to be comparable to chemical availability, as in the case of uptake of metals by plants (Römkens and Groenenberg 2001). Considerable debate exists as to whether just the solution fraction, the porewater concentration, or the free metal ion activity is relevant to plant uptake (Parker and Pedler 1997). In addition, several studies showed that for soil invertebrates, dietary metal exposure might also be an important route of uptake under environmentally relevant conditions (Lock and Janssen 2001, 2003; Lock et al. 2003). Thus, concentrations of metals in the porewater of the soil or the metal activity in the soil solution lack sufficient basis for risk assessment purposes. Clay and organic matter content Few studies have explicitly examined the relationship between organic matter content or clay content in soil and its influence on the toxicity of zinc. The risk assessment showed that there is no clear relationship between either clay or organic matter content and toxicity, based on a meta-analysis of data from various sources. From this information, no consistent conclusion can be drawn that clay content or organic matter content are sole modulating factors for terrestrial species and microbe-mediated processes. Bioavailability of zinc in soil is, thus, clearly not a single function of the speciation of zinc in soil. A clear relationship between a chemically defined available concentration in the soil solution and the real, biological availability as experienced by plants, invertebrates, micro-organisms, etc., can at the moment not be provided. The technical meeting, however, concluded that there is a scientific basis to correct the PECs for aging (laboratory-to-field differences), as well as for soil properties, to take into account the bioavailability of zinc in soil. For microbial endpoints, it is the background concentration of zinc that determines the microbial sensitivity to zinc; for plants and invertebrates, it is ph and CEC that significantly affect bioavailability and toxicity. These 3 soil parameters are, therefore, used to correct the PEC add for bioavailability, in addition to the aging correction of a factor of 3. Because terrestrial microbial processes and terrestrial plants and invertebrates need to be protected, the approach to correct for bioavailability by taking soil properties into account was performed in a conservative way. This meant that the smallest correction from either the microbial-related equations or the plants- and invertebrates-related equations was used to correct for bioavailability of zinc in the soil. The range of these corrections for soil properties varied from negligible for sandy soil to 1.5 for river clay. The most conservative bioavailability factors (BioF soil ) are then derived to correct the PEC add. The overall terrestrial PEC add correction thus becomes PEC add, bioavailable ¼ PEC add /(3 BioF soil ). If no sufficient site- or region-specific