INC. Nickel Producers Environmental Research Association NICKEL INDUSTRY SUMMARY OF HIGHER TIER DATA AND THEIR RELEVANCE TOWARD THE NI EQS UNDER THE WATER FRAMEWORK DIRECTIVE January 31 st, 2011 Comments provided by: The Nickel Institute Brussels, Belgium and Nickel Producers Environmental Research Association (NiPERA) Durham, NC USA
Assessment of the margin of safety between the EQS and new higher tier aquatic data for nickel. Executive Summary Summary Paper February 2011 The purpose of this communication is to introduce a series of documents that provide new higher tier information on laboratory to field extrapolation of nickel (Ni) toxicity to aquatic organisms. These new data were generated in response to the conclusions from the Existing Substances Risk Assessment of Nickel (EU RAR) completed in 2008. Specifically, the conclusions identified the relative uncertainty associated with the derivation of the aquatic Predicted No Effect Concentration (PNEC) in relation to the absence of field and/or mesocosm data. The absence of these data in the EU RAR led to the Technical Committee for New Existing Substances agreeing that in order to account for the relative uncertainties an assessment factor of 2 should used in the PNEC derivation process. The new higher tier data presented here are used in the PNEC derivation process to now place the laboratory-based ecotoxicity studies into a real-world context. A four month community level mesocosm study with nickel was undertaken by the Fraunhofer Institute and completed in July 2010. The physico-chemical conditions of the mesocosm resulted in high nickel bioavailability. The organisms in the mesocosms included phytoplankton, zooplankton, epiphytic invertebrates and importantly, as identified in the EU RAR as the most sensitive species, the snail Lymnaea stagnalis. Numerous endpoints were examined and no effects were seen below 24 µg Ni L -1. The NiBLMs were used to predict the bioavailability-based HC5(50%) for this study and the values range from 4.2 6.8 µg Ni L -1, some 3.5 times lower than the mesocosm NOEC. The field-based analysis was undertaken on stream data from the Environment Agency of England and Wales for sites with matched benthic invertebrate and water chemistry data. Water Framework Directive (WFD) benthic community metrics under a range of nickel exposures were compared to expected metrics as determined using the RIVPAC system. The ratio of observed to expected community metrics are used to define ecological status under the WFD. The benthic macroinvertebrates included nickel sensitive snail species identified from the EU RAR. The bioavailable Ni exposure concentration was estimated at each site and at the boundary of good to moderate ecological status. The community-based metrics, such as N-Taxa, showed that the boundary of good to moderate status was five times the EQS. Individual sensitive taxa (including the snails) also showed a margin of safety of two over the EQS. This paper provides a summary of the new higher tier data from the mesocosm and the field analysis. Additional lines of evidence are discussed, along with the possible implications of these new data in relation to the margin of safety to aquatic communities provided by the Ni EQS and the assessment factor on the PNEC. The full reports of the studies described below and a paper undertaking a weight of evidence analysis are currently on CIRCA. 2
Background The Existing Substances Risk Assessment of Nickel (EU RAR) 1 concluded that there was relative uncertainty around the aquatic HC5(50%) derived using a species sensitivity distribution (SSD) and accounting for bioavailability due to the lack of field and/or mescosm data. This relative uncertainty led to the use of an assessment factor of 2 on the HC5(50%). However, it should be noted that the Ni EU RAR has the largest chronic aquatic dataset (193 individual chronic values, 31 species, 19 families) of any of the substances considered under the Existing Substances Regulation (793/93/EEC). Additionally, four chronic Biotic Ligand Models (BLM) (two invertebrates, one alga, and one fish) are available to normalize the Ni ecotoxicity database. Finally, the approach for using the BLMs (i.e., using the most stringent invertebrate BLM in cases where an option exists) results in an added element of precaution; specifically, the HC5(50%) is 1.5 times lower in high bioavailability waters (high ph, low dissolved organic carbon) than would be achieved if the less stringent BLM was used. Since the completion of the EU RAR in 2008, the Nickel Industry has performed a field-based analysis, a mesocosm study, and a weight-of-evidence analysis to reduce the uncertainty in the PNEC derivation process. These new data were not considered by the Danish Environment Protection Agency in the development of the proposed Ni Environmental Quality Standard (Ni EQS) as they were not available. This new information is relevant to the determination of an appropriate Ni EQS, and because of the importance in aligning the approaches taken for the Water Framework Directive (WFD) and the REACH process, it is vital that this information be considered in the current cycle of Ni EQS determination. Otherwise, updates to the REACH Generic Exposure Scenarios performed by the Ni REACH Consortia may differ from the Ni EQS under the WFD, as the REACH process obligates the Ni REACH Consortia to report the outcome of these studies. Harmonizing the two approaches therefore makes sense from the policy, administrative and technical perspectives. The aquatic effects assessment within the EU RAR was of critical importance because it served as the technical foundation for the Ni EQS under the WFD and the Generic Exposure Scenarios under REACH. The new data are in a highly refined state that informed bodies, like the SCHER, will be able to readily grasp. While recent informal meetings have raised questions concerning the details of the findings, the major implications of the data are clear: they substantially reduce uncertainty in the Ni Aquatic Effects Assessment. In addition, according to the EQS TGD, a review of the RAR PNEC is warranted when new, potentially critical, ecotoxicity data (i.e. sensitive species or endpoints) has become available since the publication of the RAR or when the uncertainty analysis of a proposed EQS becomes relevant to the practical implementation of a compliance regime. These new data provide new insight on the major area of uncertainty analysis identified in the Ni Aquatic Effects Assessment within the EU RAR. Each line of new evidence is summarized below. A discussion section then provides a consideration of the potential implications of these data on the relative uncertainty and safety margin of the current Ni EQS. 1. Community level study with nickel in aquatic Mesocosm The Fraunhofer Institute at Schmallenberg, Germany, performed a community-level study with nickel in aquatic mesocosms, and three documents describing the results of this study are available on CIRCA. The three documents include: 1 http://ecb.jrc.ec.europa.eu/documents/existing-chemicals/risk_assessment/report/nickelreport311.pdf 3
1. Community Level study with nickel in aquatic microcosms : The main report describing the methods, community-based response, and overall conclusions; 2. Community level study with nickel in aquatic microcosms - Appendix B - Analytical Report : The full analytical results of the study; and, 3. Validation of the BLM normalized HC5(50%) for Ni against the responses observed in a 4-month microcosm experiment : This document was prepared by K. De Schamphelaere to calculate the BLM-normalized HC5(50%) that would be predicted for a system with the same water quality parameters as was observed in the Fraunhofer mesocosm systems. The exposure portion of the Fraunhofer study was initiated in October 2009, and it ended in February, 2010, for a total exposure duration of 16 weeks. The study was performed indoors using 750 L test chambers. Nickel concentrations ranged from 6 to 96 µg Ni L -1, using two replicate test chambers per Ni concentration in addition to four control replicates. The freshwater community introduced to the test chambers included phytoplankton, zooplankton, and epiphytic invertebrates. Importantly, organisms shown to be sensitive in the Aquatic Effects Assessment of Ni from the EU RAR were included in the test community, including cladocerans (several Daphnia species) and the snail Lymnaea stagnalis. Nickel was added throughout the experiment to maintain a steady dissolved Ni concentration, and thorough and frequent chemical analyses were made over the course of the study. Briefly, chemistry results indicated that ranges of water chemical parameters of relevance to Ni bioavailability were as follows: ph = 8.37 to 8.69; hardness = 154 to 207 mg CaCO 3 L -1 ; and, dissolved organic carbon (DOC) = 3.1 to 5.3 mg L -1. Given the high ph and the relatively low DOC concentrations, this combination of water chemistry parameters would be reflective of a system showing high Ni bioavailability. A multitude of species-specific and community-level metrics were examined (> 150 ecotoxicity endpoints). Exposure up to 24 µg Ni L -1 resulted in no ecologically adverse effects on phytoplankton, periphyton, zooplankton and snails. Exposures of 48 and 96 µg Ni L -1 gave clear effects on phytoplankton (namely Cryptophyceae), and on snails. As shown in Table 1 below, the overall studyspecific NOEC was concluded to be 24 µg Ni L -1 (with the LOEC being 48 µg Ni L -1 ). Table 1: Effect classification for the different Ni treatment levels in the microcosm study 1: No effects: 2: Slight and transient (one sampling date) effects, 3: Pronounced temporary effects, 5: Pronounced effect over more than 8 weeks or until/at the end of the study Nominal Ni concentration [µg/l] 6 12 24 48 96 Phytoplankton community 1 1 1 5 5 Phytoplankton population 1 1 2 5 5 Periphyton 1 1 1 5 + 5 + Zooplankton community 1 1 1 5 5 Zooplankton population 1 1 1 5 5 + Snails 1 1 1 5 5 Meiobenthos 1 1 1 5 5 In order to compare these results with the bioavailability-based approach that was developed in the Ni EU RAR and used in the EQS Dossier, the relevant water chemistry data were used by researchers at Ghent University to determine a bioavailability-based HC5(50%). Because the ph of the system went beyond the higher boundary of ph used in the development of Ni BLMs, the researchers extrapolated ph relationships to the observed ph. Because water chemistry varied over time, a range of predicted HC5(50%) values from 4.2 (based on water chemistry at the beginning of the test) to 6.8 µg Ni/L (based 4
on water chemistry at the end of the test) was determined. This range is at least 3.5 times lower than the observed study-specific NOEC. There are different approaches for analyzing the sensitive snail abundance data. Spearman rank correlation can be used to determine relationships between snail abundance and time, with the hypothesis that a decrease in snail abundance over time within a Ni-added treatment would indicate an adverse effect of Ni on the snail population. The conclusion from this analysis would suggest that the NOEC for the snail abundance endpoint from the mesocosm should be 6 µg Ni L -1, and not 24 µg Ni L -1 as stated in the Fraunhofer report. However, it is likely that Spearman rank correlation is not appropriate for such an analysis as correlation does not establish causality, meaning that while the outcome indicates that snail abundance decreased over time, it does not identify Ni exposure as the causative factor. In addition, correlation analysis assumes that the observations were not biased. As the Fraunhofer report indicates, the beginning numbers of snails was not the same across all microcosm chambers. Four large snails (Lymnaea stagnalis) were added to each microcosm, but snails also occurred in the sediment that was added to the microcosms, which led to uneven numbers of snails being added to the microcosms at the beginning of the experiment. Also, snail abundance was estimated over the course of the study by counting numbers of snails on one glass wall per microcosm. Definitive counts were only made at the end of the study. A recent re-analysis of the snail abundance data throws further doubt upon the validity of the Spearman rank correlation analyses in that that snail abundance in one of the control microcosms (of which there were four) was unusually high. Specifically, 179 snails were counted in this control microcosm (microcosm #2), compared with 6, 25, and 5 from the other three control microcosms. If this microcosm is considered to be an outlier, the analysis changes drastically. Mean numbers of snails at the end of the test actually increase from the control treatment (mean = 12 snails) to the 6 µg Ni L -1 treatment (mean = 17 snails) to the 12 µg Ni L -1 treatment (mean = 22 snails). The intention of measuring snail abundance is reasonable, given the sensitivity demonstrated by L. stagnalis in laboratory tests. However, snail abundance failed to be a robust and independent variable 2 ; for this reason, only the most basic statistical analysis is appropriate. The only robust evidence-based conclusion to be drawn from the snail data is that the 48 and 96 µg Ni L -1 treatments showed a clear and unambiguous impact on snail abundance. In summary, the vast majority of the analyses support a NOEC of 24 µg Ni L -1 from the analysis, which is at least 3.5 times greater than the bioavailability normalized HC5(50%) for the chemical characteristics of this water. Given the precautionary steps that are built into the determination of the HC5(50%), these results reduce the uncertainty surrounding the ability of the HC5(50%) to be protective of real world ecosystems. 2. Assessment of the Effects of Nickel on Benthic Invertebrates in the Field Peters et al. (2010) 3 report on a field-based evaluation of the biological effects of potential Ni exposures conducted using monitoring data for benthic macroinvertebrates and water chemistry parameters for streams in England and Wales. Benthic macroinvertebrate and water chemistry data were from Environment Agency water quality monitoring sites. Observed benthic community metrics were compared to expected community metrics under reference conditions using RIVPACS (River Invertebrate Prediction and Classification System). The ratio of observed (O) to expected (E) community metrics (i.e., O/E) is 2 Snail populations varied with primary production and abundance at the start of the experiment. 3 A. Peters, M. Crane, P. Simpson, and G. Merrington. 2010. Assessment of the Effects of Nickel on Benthic Invertebrates in the Field. Final report. Prepared for the Nickel Producers Environmental Research Association, Durham, NC, USA. 5
referred to as the Ecological Quality Index (EQI) and is used to define ecological status. In order to evaluate relationships between Ni concentrations and benthic community metrics, bioavailable Ni concentrations were also calculated for each site. The bioavailable Ni concentrations were calculated using the Ni Screening Tool, as described in the EQS Dossier. A limiting effect from nickel on the 90 th percentile of the maximum achievable ecological quality was derived for the WFD classification metric number of BMWP scoring families (N-taxa). A concentration above which only 10% of sites would be able to achieve Good Ecological Status was derived at bioavailable nickel exposures of 10.3 g l -1 (95% confidence interval 9.0 to 18.2). These data show that a bioavailable nickel concentration would need to be over 10.3 g l -1 for there to be a change in the ecological status classification of a site. If this threshold were expressed as total dissolved Ni for the bioavailability scenarios considered in the EU RAR they would range between 18-90 g total dissolved Ni l -1. Snail abundance was also analysed as a subset of the community as laboratory ecotoxicity testing suggested that they may be a sensitive fraction of the ecosystem. Snail abundance was also expressed relative to reference conditions. From the average of assessments based on spring and autumn data for snails, a threshold above which O/E values of 0.9 are achieved in 10% of the samples was derived as 3.9 g l -1. Therefore, for one of the most sensitive groups indentified in the laboratory testing bioavailabile Ni concentrations need to exceed 3.9 g l -1 before changes in snail abundance are likely to occur in the field. As above, if this snail threshold were expressed as total dissolved Ni for the bioavailability scenarios considered in the EU RAR they would range between 6.8-34 g total dissolved Ni l -1. Nickel exposure concentrations co-vary with the concentrations of other stressors in the dataset, and high concentrations of nickel are also associated with elevated concentrations of other contaminants such as copper, zinc, chromium, iron, arsenic, phosphate and ammonia. The effect of additional toxicants which co-vary in exposure with nickel is to reduce the maximum achievable ecological quality and results in more stringent thresholds being derived that account for mixed exposures to contaminants. 3. Weight of evidence A weight-of-evidence analysis (entitled Lab-to-field Assessment Factor Evaluation for the Nickel Surface Water PNEC ) was performed by A. Fairbrother (Exponent) and D. De Forest (Windward). This review examined several additional lines of evidence, in addition to the above mesocosm and field studies, to review residual uncertainty in derivation of the HC5 and determine the margin of safety of the derived EQS. Data analysis was performed on 632 sites where Ni water concentrations were measured by the Forum of European Geological Surveys (FOREGS); just 7 (1%) had an ambient Ni concentration exceeding the HC5 and 44 (7%) had an ambient Ni concentration exceeding the HC5 divided by 2 (4.6 µg L -1 ). There is no apparent pattern to the 44 sites that exceeded the HC5 divided by 2; they are scattered around Europe and have no consistent water type. Deleebeck et al. (2008) supported this observation, reporting that a Ni concentration (6.0 µg L -1 ) at one of six locations in Belgium, France, and The Netherlands not influenced by anthropogenic inputs was greater than the HC5 divided by 2. In contrast, the background Ni concentrations were never greater than the HC5 alone. Two published modelling studies determined that Ni is similar to all chemicals in level of protection afforded by the HC5. Emans et al. (1993) addressed whether an HC5 from an SSD based on NOECs from single species tests is an accurate estimation of NOECs from multispecies tests or field observations. They used a database of over 300 multispecies studies, including metals (aluminum, cadmium, copper, and zinc), and concluded that single-species toxicity data can be used to derive safe values for the aquatic ecosystem. De Vries et al. (2010) examined the level of protection afforded by the HC5 of a 6
hypothetical chemical to the biodiversity of aquatic communities through the use of empirical models. Specifically, they examined whether the HC5 derived from a standard SSD is equally protective of communities regardless of the relative density of robust or sensitive species at the time of the application of a stressor. They concluded that the HC5 level is a protective threshold for changes in biodiversity in 99.6% of the cases for chemicals for which direct effects are dominant, and that changes in concentrations at and below the HC5 would not be detectable, regardless of the type of toxicant effect considered. We now know that ecosystems are in a constant state of flux and divergence from a given state is considered a common event (Hobbs and Morton 1999). In response, many regulatory bodies have set maintenance or enhancement of biodiversity as a primary goal of environmental management (e.g., EU 2000). Because quantifying the biodiversity of a community is dependent on spatial and temporal scales, the determination of whether a biological community continues to be self-sustaining requires a focus on the core species while potentially ignoring the recruiting species (see definitions from Magurran 2007). However, it is not possible to a priori define core species for all ecosystems so a generic EQS or PNEC should be sufficiently conservative to be protective of most species most of the time, but does not need to afford protection to all (Stephan et al. 1985). This suggests that an HC5, which is protective of at least 95% of the species, should be sufficiently conservative to preserve system integrity, including the resiliency needed for continual adaptation to changing environmental conditions. This would be especially true for the Ni HC5(50%), which is based on the largest chronic aquatic dataset considered in the EU. 4. Implications for the margin of safety on the EQS The current HC5 for Ni in freshwater is based on the 5 th percentile of chronic EC10 values and NOECs available for 31 species, including algae, higher plants, invertebrates, fish, and amphibians. Because the HC5 is derived using the BLM, it also accounts for site-specific factors that influence bioavailability, thereby helping to ensure that the HC5 is not unduly under- or over-conservative. Given the protectiveness of the Ni HC5 when compared to mesocosm- and field-based effects data, we conclude that the uncertainty within the aquatic effects assessment is sufficiently understood, and that the HC5 provides an adequate level of protection for freshwater systems. References cited De Vries P, Smit MGD, van Dalpsen JA, De Laender F, Karman CC. 2010. Consequences of stressor-induced changes in species assemblage for biodiversity indicators. Environ Toxicol Chem. 29:1868-1876. Deleebeck NME, De Schamphelaere KAC, Janssen CR. 2008. A novel method for predicting chronic nickel bioavailability and toxicity to Daphnia magna in artificial and natural waters. Environ Toxicol Chem 27:2097-2107. Emans HJB, Plassche JVD, Canton JH, Okkerman, PC, Sparenburgs PM. 1993. Validation of some extrapolation methods used for effect assessment. Environ Toxicol Chem 12:2139-2154. European Union (EU). 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. OJEC L327/1-72. Hobbs RJ, Morton SR. 1999. Moving from descriptive to prescriptive ecology. Agroforest Syst 45:43-55. Magurran AE. 2007. Species abundance distributions of time. Ecol Lett 10:347-354. Stephan CE, Mount DI, Hansen DJ, Gentile JH, Chapman GA, Brungs WA. 1985. Guidelines for deriving numerical national water quality criteria for the protection of aquatic organisms and their uses. US Environmental Protection Agency, Washington DC. PB85-227049 7