Environ. Sci. Technol. 2007, 41, 5014-5021
Use of Field Data to Support European Water Framework Directive Quality Standards for Dissolved Metals M A R K C R A N E , * ,† K E V I N W . H . K W O K , ‡ CLAIRE WELLS,§ PAUL WHITEHOUSE,§ AND GILBERT C. S. LUI⊥ Watts & Crane Associates, 23 London Street, Faringdon, Oxfordshire, SN7 7AG, Department of Ecology & Biodiversity, The Swire Institute of Marine Science, The University of Hong Kong, Pokfulam, Hong Kong, P.R. China, Environment Agency Science Group, Evenlode House, Howbery Park, Wallingford, Oxfordshire, OX10 8BD, and Department of Statistics and Actuarial Science, The University of Hong Kong, Pokfulam, Hong Kong, P.R. China
Quality standards (QS) for dissolved metals in freshwaters have been proposed under the European Water Framework Directive (WFD) and are based mainly upon laboratory ecotoxicity data. Uncertainties remain about laboratory-tofield extrapolation to establish QS that are neither overnor underprotective. Freshwater benthic macroinvertebrates are a group of organisms of known sensitivity to heavy metals. We analyzed a dataset from England and Wales of dissolved metal concentrations (cadmium, chromium, copper, iron, nickel, lead, and zinc) and associated benthic invertebrate community metrics, using piecewise regression, quantile regression, and information on metal concentrations consistent with good quality status. Analysis of these field data suggests that dissolved metal QS proposed under the WFD are similar to metal concentrations in rivers associated with unimpaired benthic invertebrate assemblages in England and Wales. The only exceptions to this are QS for iron and zinc, where use of relatively large assessment factors leads to standards that are substantially below concentrations associated with impaired invertebrate assemblages in the field.
Introduction Knowledge of the field effects of chemicals is a useful line of evidence when making risk assessment and management decisions (1, 2). This is because reliance on single species laboratory toxicity tests to predict the effects of chemical substances on populations and assemblages in complex natural environments necessarily involves a range of often untested assumptions (3). These assumptions include estimates of chemical bioavailability and organism sensitivity to individual chemicals as mediated through a complex * Correspondingauthorphone: +441367244311;e-mail: mark.crane@ wca-environment.com. † Watts & Crane Associates. ‡ The Swire Institute of Marine Science, The University of Hong Kong. § Environment Agency Science Group. ⊥ Department of Statistics and Actuarial Science, The University of Hong Kong. 5014
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 14, 2007
mixture of variable environmental stressors. The outcome of this is that although it is generally believed that estimates of risk based upon laboratory data and models are likely to be conservative, these estimates may be either conservative or liberal, depending upon the specific environmental conditions that apply at each natural site. Field data help to “ground truth” laboratory estimates and provide a warning of whether these estimates might be over- or underprecautionary. On 17 July 2006 the European Commission released a proposal for a Daughter Directive (4) to the Water Framework Directive (WFD) (5) to deal with the control of priority substances. The aim is to “ensure a high level of protection against risks to or via the aquatic environment stemming from...33 priority substances and certain other pollutants by setting environmental quality standards.” The focus of the Daughter Directive is on repealing the Dangerous Substances Directive (76/464/EEC and its Daughter Directives) and defining long term (annual average, AA) or short term (maximum allowable concentration, MAC) numerical water quality standards for freshwaters and saltwaters. The aim is to harmonize QS with the objectives and provisions of other Community legislation, particularly REACH (Registration, Evaluation and Authorization of Chemicals), the Pesticides Directive (91/414/EEC), the Directive on IPPC (Integrated Pollution Prevention and Control) and thematic strategies on the marine environment and on pesticides. In the Daughter Directive, AA QS are set for surface waters for all 33 priority substances. In addition to the QS for priority substances in the Daughter Directive, each member state of the European Union is also required to set QS for other “specific pollutants” using a similar methodology. Annex V of the WFD outlines the methodology to be used when deriving QS and states that, “...the standard thus derived should be compared with any evidence from field studies. Where anomalies appear, the derivation shall be reviewed to allow a more precise safety factor to be calculated.” Thus, when reliable and relevant evidence is available from a field study it can be used to increase or decrease the assessment factor applied either to the lowest laboratory no observed effect concentration (NOEC), or to the hazardous concentration for 5% of species (HC5) from a species sensitivity distribution. In the European Commission’s Technical Guidance Document (6) the term “field data” is used to describe the results from microcosm and mesocosm experiments as well as data from field monitoring studies. In this paper we use benthic invertebrate field monitoring data from England and Wales to derive effects thresholds for dissolved concentrations of the metals cadmium, chromium, copper, iron, nickel, lead, and zinc. We then compare these thresholds with proposed QS for these metals derived from laboratory toxicity tests to determine the extent to which the field data support the laboratory-based QS.
Materials and Methods Field data were acquired from the Environment Agency of England and Wales (see Table 1 for number of sites and sample sizes). These data comprised spatially matched measurements of benthic macroinvertebrate family richness and dissolved metal concentrations for two sampling periods (Spring and Autumn) in 1995 (the only year for which matched data are currently available) from across all eight Environment Agency Regions in England and Wales. The chemical data were mean values for the 3 months before macroinvertebrate samples were taken, usually based on monthly sampling (i.e., n ) 3). Not all metals were measured at all sites, so the overall 10.1021/es0629460 CCC: $37.00
2007 American Chemical Society Published on Web 06/08/2007
TABLE 1. Number of Site and Sample Data Analyzed in This Study number of number of number of number of GQA sampling sites samples GQA A/B sites A/B samples Cadmium Chromium Copper Iron Nickel Lead Zinc
341 245 1438 253 467 527 291
428 417 2956 420 618 776 448
254 182 1125 163 357 346 202
304 278 1939 274 455 490 241
number of samples and the locations at which they were measured differs between the different metals. If the dissolved metal concentration data were lower or equal to the limit of detection we took half the detection limit as the concentration. Pearson correlations (not shown) between dissolved metal concentrations and 3-year average dissolved oxygen, biochemical oxygen demand and ammonia concentrations did not reveal any statistically significant correlations between metals and these three other important point source stressors, so other contaminants are unlikely to have confounded the results. The macroinvertebrate data were expressed as ecological quality indices (EQI) for the average score per taxon (ASPT). ASPT is a widely used metric in the UK that is calculated by dividing the biological monitoring working party (BMWP) score for the invertebrate families found at a site by the number of families contributing to that score (7). The EQI is calculated as the observed ASPT value divided by the ASPT value estimated by the RIVPACS (river invertebrate prediction and classification system) model used by the Environment Agency to predict reference conditions at a site in the absence of pollution (7). In addition to the EQI, Ephemeroptera (mayfly) and EPT (Ephemeroptera, Plecoptera [stonefly] and Trichoptera [caddis fly]) family richness were also used as biological effects metrics, as these families of aquatic insects are known to be sensitive to heavy metal toxicity (8, 9). The paired chemical concentration and macroinvertebrate metrics data were compared in two ways to determine whether there are dissolved metal thresholds beyond which adverse effects on macroinvertebrates occur. Data were plotted as log dissolved metal concentration against biological effects metrics to determine effects thresholds for (i) EQI scores, (ii) number of Ephemeroptera families and, (iii) number of EPT families. Effects thresholds were determined by using two statistical approaches: (1) Quantile regression, which is a technique that estimates multiple slopes for a set of data, and is specifically used by ecologists when only a subset of limiting factors are measured in a study (10). For plots of biological integrity against chemical concentration a high quantile (the 99th percentile in this analysis) could be interpreted as representing good quality at the intercept and it represents the response of the assemblages when the metal of interest is the most limiting. Quantile regression was applied to data sets using a loglinear model (log(y) ) a + b × concentration) using R (version 2.1.0; http://cran.r-project.org). The concentration at which this quantile declines by an agreed percentage (10% in this analysis, referred to throughout this paper as the EC1099%ile) can be estimated and used to define a threshold. In the current analysis, 2000 iterations were used to generate 95% confidence intervals (CI) for the quantiles. Quantile regressions could be fitted to Ephemeroptera and EPT data, but not to EQI data. (2) Piecewise regression, which is a “broken stick” model in which two or more lines are used to describe the data, joining at a “breakpoint” (11). For plots of biological integrity against chemical concentration such a breakpoint could be
FIGURE 1. Association between dissolved cadmium concentration and (a) EQI, (b) EPT, (c) number of mayfly families. Solid line represents fitted piecewise regression. interpretedasatoxicitythreshold.ThenumberofEphemeroptera and EPT families, and the dissolved metal concentrations, were log transformed before the analysis. A log-linear model was used to fit the three sets of data (SAS version 9.1.3; SAS Institute Inc., North Carolina). Two pieces of log-linear functions were fitted over the dataset by the method of leastsquares and the breakpoint was determined at the concentration where the sum of squared errors was at a minimum. After the position of the breakpoint was determined, the parameters of the two pieces of log-linear lines were derived. In the current analysis 2000 iterations were used to generate 95% CI (the 2.5 and 97.5 percentiles) for the regression parameters and the breakpoint. VOL. 41, NO. 14, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5015
FIGURE 2. Association between dissolved chromium concentration and (a) EQI, (b) EPT, (c) number of mayfly families. Solid line represents fitted piecewise regression and dashed line represents fitted quantile regression. A complementary approach was also used on the same data set. The UK Technical Advisory Group (UKTAG) on the Water Framework Directive has recently published a report on UK Environmental Standards and Conditions (12), which deals with the derivation of standards for Biochemical Oxygen Demand, dissolved oxygen, total ammonia, pH, salinity, and nutrients. The approach adopted by UKTAG was to focus on sites in the UK that achieved Class A and B in the general quality assessment (GQA) that is routinely undertaken, as these classes are equivalent to “good quality” sites under the WFD. Concentrations of determinands of interest were then collated for these sites, and the 90th percentile concentration was calculated. The rationale for this is that some sites may 5016
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 14, 2007
FIGURE 3. Association between dissolved copper concentration and (a) EQI, (b) EPT, (c) number of mayfly families. Solid line represents fitted piecewise regression and dashed line represents fitted quantile regression. have been misclassified as Class A or B when they were actually of lower quality. Taking the 90th percentile is, therefore, a precautionary approach for establishing the maximum concentration of a determinand that is still consistent with a site achieving Class A or B. We used the same approach for the data on dissolved metals. All Class A and B sites in the data set were selected and dissolved metal concentrations at the 90th, 95th, and 99th percentiles were identified, along with the maximum value recorded at any site.
Results Associations between dissolved metal concentration and EQI, EPT, or number of mayfly families are shown
FIGURE 4. Association between dissolved iron concentration and (a) EQI, (b) EPT, (c) number of mayfly families. Solid line represents fitted piecewise regression and dashed line represents fitted quantile regression. in Figure 1 (cadmium), Figure 2 (chromium), Figure 3 (copper), Figure 4 (iron), Figure 5 (nickel), Figure 6 (lead), and Figure 7 (zinc). The cadmium data were mostly distributed between three values (0.1, 0.2, and 0.5 µg Cd L-1), which reflect different limits of detection in different analytical laboratories. Because of this, quantile regression could not be used successfully to analyze any of the cadmium data. Figure 8 compares (i) piecewise regression breakpoints and the EC1099%ile of Ephemeroptera, EPT, and EQI scores with (ii) the 90th, 95th, and 99th percentiles and maximum concentration at class A/B sites and (iii) proposed European WFD quality standards. Table 2 summarizes the basis of these WFD quality standards.
FIGURE 5. Association between dissolved nickel concentration and (a) EQI, (b) EPT, (c) number of mayfly families. Solid line represents fitted piecewise regression and dashed line represents fitted quantile regression.
Discussion Historically, most field monitoring data for aquatic systems focused on the measurement of chemical concentrations alone (e.g., ref 13). However, biomonitoring approaches have become more popular, especially those that involve the monitoring of benthic macroinvertebrate assemblages (14). The presence of a biological species, or group of species, known to be sensitive to a particular type of pollutant provides strong evidence that the chemical is either not present or not bioavailable (1). There is considerable evidence to show that benthic macroinvertebrates are sensitive to metal toxicity (8, 9, 15, 16), although they may not always be the most sensitive species to specific metals (e.g., see Table 2 entries VOL. 41, NO. 14, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5017
FIGURE 6. Association between dissolved lead concentration and (a), (b) EPT, (c) number of mayfly families. Solid line represents fitted piecewise regression and dashed line represents fitted quantile regression. for Cu, Pb, and Zn), and this must be considered when using the data presented in this paper as an additional line of evidence in quality standard-setting. The results from field monitoring studies have the additional advantage of direct relevance and easy communication to the public (17). One of the main criticisms of field biomonitoring approaches is that it is difficult to establish what the “reference condition” should be, which is the structure of the assemblage that would occur in the absence of anthropogenic stressors. In the UK, this problem has been addressed through the development of the river invertebrate prediction and classification (RIVPACS) system, which allows prediction of the 5018
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 14, 2007
FIGURE 7. Association between dissolved zinc concentration and (a) EQI, (b) EPT, (c) number of mayfly families. Solid line represents fitted piecewise regression and dashed line represents fitted quantile regression. macroinvertebrate assemblage which should occur at a particular site, based upon natural physical and chemical water quality parameters (7). The EQI metric used to summarize the data takes the reference condition into account. Currently, WFD QS are likely to be derived based on single species laboratory toxicity tests. Either the lowest value or a percentile of a species sensitivity distribution (usually a fifth percentile, or HC5) is selected and an assessment factor (AF) is applied to calculate a PNEC. It is the size of this AF (usually between 1 and 100) that might be modified upward or downward by field data according to Annex V of the WFD. The analyses presented in this paper show that currently proposed QS for dissolved metals could usefully include information from benthic macroinvertebrate field data. These field data, when used as an additional line of evidence with laboratory data, suggest that standards for dissolved metals
FIGURE 8. Comparison of piecewise regression breakpoints (open symbols) and EC1099%ile (closed symbols) for Ephemeroptera, EPT, and EQI scores, 90th, 95th, 99th percentiles and maximum concentrations at class A/B sites (as vertical lines on the dashed line), and proposed European Water Framework Directive Quality Standards (WFD QS) for seven different metals. Error bars represents 95% CI of piecewise regression breakpoints and EC1099%ile. should be in the following range, based upon the 95% confidence intervals around piecewise regression breakpoints and EC1099%ile values, and the 90th percentile concentrations found at Class A and B sites in England and Wales: cadmium 0.2-0.5 µg L-1, chromium 1-2 µg L-1, copper 2-4 µg L-1, iron 43-250 µg L-1, nickel 0.6-7 µg L-1, lead 0.6-2.5 µg L-1, and zinc 20-27 µg L-1. The proposed WFD QS for cadmium, chromium, and copper fall within, or are very close to, these ranges. The QS
for lead and nickel are high because interim values have been proposed by the European Commission until risk assessments for these metals have been completed. Based on the data available so far it is very likely that the final QS for lead and nickel will be close to the field-based thresholds identified in this paper. The proposed QS for iron and zinc are lower than the thresholds suggested by the statistical analyses presented in this paper. However, the laboratory NOECs on VOL. 41, NO. 14, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5019
TABLE 2. Basis for Proposed Water Framework Directive (WFD) Annual Average (AA) Quality Standards (QS)a proposed WFD QS (AA) µg L-1
metal Cd
basis for proposed WFD QS (AA)
0.08-0.25 HC5 from SSD, normalized for water hardness. (depending on water hardness) The most sensitive taxa in the SSD were invertebrates. 3.4 (Cr VI) Cr (VI) QS based on an HC5 from an SSD with an AF of 3. The most sensitive taxa in the SSD were invertebrates. 4.7 (Cr III) Cr (III) QS based on the most sensitive species (the invertebrate Daphnia magna) NOEC of 47 µg L-1 with an AF of 10. 8.2 HC5 from an SSD without an AF (because of the large number of data in the SSD). The most sensitive taxa in the SSD were invertebrates (lowest NOEC ) 4 µg L-1) and fish (lowest NOEC ) 2.2 µg L-1). 16 21-day NOEC of 160 µg L-1 for reproduction of the invertebrate Daphnia magna (lowest value) and an AF of 10. 20 The substance data sheet for Ni recommends a maximum permissible addition (MPA) of 1.7 µg Ni L-1 to background levels (e.g., the background of 2.1 µg L-1 used in the EU risk assessment, which leads to a QS of 3.8 µg L-1). The MPA of 1.7 µg L-1 is based on an HC5 of 5.1 µg L-1 from an SSD, with application of an AF of 3. The most sensitive taxa in the SSD were invertebrates. The WFD Daughter Directive recommends an interim value of 20 µg L-1 only until the Ni EU risk assessment has been finalized. 7.2 The substance data sheet for Pb recommends an MPA of 2.1 µg Pb L-1 to background levels (e.g., the background of 0.2 µg L-1 used in the EU risk assessment, which leads to a QS of 2.3 µg L-1). The MPA of 2.1 µg L-1 is based on an HC5 of 6.4 µg L-1 from an SSD, with application of an AF of 3. The most sensitive taxa in the SSD were fish, but invertebrates were poorly represented, with only one insect species (the midge Chironomus tentans) The WFD Daughter Directive recommends an interim value of 7.2 µg L-1 only until the Pb EU risk assessment has been finalized. 7.8 HC5 from an SSD with an AF of 2. The most sensitive taxa in the SSD were algae (lowest NOEC ) 16 µg L-1) and invertebrates (>2-fold less sensitive, with lowest NOEC ) 37 µg L-1).
Cr
Cu Fe Ni
Pb
Zn
a Proposed WFD QS are taken from the Europa website (Cd, Ni, Pb) or unpublished draft UK reports (Cr, Cu, Fe, Zn): http://eur-lex.europa.eu/ LexUriServ/site/en/com/2006/com2006_0397en01.pdf (Daughter Directive on Priority Substances). http://forum.europa.eu.int/Public/irc/env/wfd/ library?l)/framework_directive/i-priority_substances/supporting_background/substance_sheets&vm)detailed&sb)Title (Priority Substance data sheets). SSD ) Species Sensitivity Distribution; AF ) Assessment Factor; HC5 ) Hazardous Concentration for 5% of species.
which the QS are based are similar to the field-derived thresholds. There are limitations to the data analyzed in this study. Rapid Biomonitoring/Bioassessment Protocols such as the one used to generate the benthic macroinvertebrate data provide qualitative information on species richness and no, or rather limited, information on abundance. There is also no within-site replication, as pooled samples are taken from each site. The overall effect of this is likely to be some reduction in the sensitivity of the biological metrics. The chemical analyses suffer from the flaw that either limits of detection or reporting protocols apparently differed between analytical laboratories, which led to a clumping of reported values around certain concentrations. This is unlikely to have a large influence on the statistical analyses. For essential metals (Cu, Fe, Ni, Zn, and Cr) the breakpoint from piecewise regression could theoretically correspond to a change between deficiency and an optimal zone, or between the optimal zone and the toxic domain. The plotted data do not suggest such a phenomenon, or it could be masked by intrinsic data variation. It is important to note that the current study looked at changes in diversity at the family level and encompasses many different invertebrate species. It is likely that these species have different requirements for each essential metal, thus a sharp parabola pattern would not be observed. The quantile regression results can be compared with the piecewise regression breakpoint as an adequacy check. With the exception of Cr and Zn, all the metals have EC10 values lower than, or similar to, the breakpoint, suggesting that the breakpoint is indeed capturing the change between optimal and toxic domains. The analyses presented in this paper could be expanded in several ways. The effects of hardness, pH, and other water quality parameters on the apparent effects of dissolved metals could be examined. Unfortunately, dissolved organic carbon is not routinely measured in the UK during monitoring programs, so it is unlikely that these data could be used directly to help develop biotic ligand modeling approaches. 5020
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 14, 2007
The potential effects of several metals at the same site could also be examined through use of mixture toxicity models. It would also be possible to examine any influence of season (Spring or Autumn) on metal concentrations and associated effects on benthic macroinvertebrates. Finally, with further local information, it might be possible to distinguish between anthropogenic inputs and background concentrations to determine whether “added risk” approaches to metal risk assessment and QS derivation (18) are protective of benthic macroinvertebrates.
Acknowledgments We thank the Environment Agency of England and Wales for sponsoring this project, Kenny Leung for useful discussions about appropriate statistical approaches, Anders Bjorgesæter for sharing technical information on quantile regression, and three anonymous referees for their helpful comments on the first draft of the paper. KWHK was partially supported by the University Grants Committee of the Hong Kong Special Administration, China, through the Area of Excellence Scheme (project no. AoE/P-04/2004).
Literature Cited (1) Clements, W. H.; Newman, M. C. Community Ecotoxicology; Wiley, Chichester, UK. 2002. (2) ECETOC. Workshop on availability, interpretation and use of environmental monitoring data 20-21 March 2003; ECETOC Workshop report no. 1; European Centre for Ecotoxicology and Toxicology of Chemicals, Brussels, Belgium. 2003. (3) Cairns, J., Jr. The myth of the most sensitive species. Bioscience 1986, 36, 670-672. (4) EC. Proposal for a Directive of the European Parliament and of the Council on Environmental Quality Standards in the Field of Water Policy and Amending Directive 2000/60/EC. COM(2006) 397 final, 17 July 2006, Brussels, Belgium. 2006. (5) EC. Directive 2000/60/EC of the European Parliament and of the Council establishing a framework for the Community action in the field of water policy. Brussels, Belgium. 2000. (6) EC. Technical Guidance Document on Risk Assessment in Support of Commission Directive 93/67/EEC on Risk Assessment for new
(7) (8)
(9) (10) (11) (12) (13)
notified substances, Commission Regulation (EC) No 1488/94 on Risk Assessment for existing substances, and Directive 98/8/EC of the European Parliament and of the Council concerning the placing of biocidal products on the market. European Commission Joint Research Centre EUR 20418 EN/2. 2003. Wright, J. F.; Sutcliff, D. W.; Furse, M. T. Assessing the Biological Quality of Fresh Waters, RIVPACS and other Techniques; The Freshwater Biological Association: Ambleside, UK. 2000. Clements, W. H.; Cherry, D. S.; van Hassell, J. H. Assessment of the impact of heavy metals on benthic communities at the Clinch River (Virginia): evaluation of an index of community sensitivity. Can. J. Fish. Aquat. Sci. 1992, 49, 1686-1694. Clements, W. H.; Carlisle, D. M.; Lazorchak, J. M.; Johnson, P. C. Heavy metals structure benthic communities in Colorado streams. Ecol. Appl. 2000, 10, 626-638. Cade, B. S.; Noon, B. R. A gentle introduction to quantile regression for ecologists. Front. Ecol. Environ. 2003, 1, 412420. Toms, J. D.; Lesperance, M. L. Piecewise regression: a tool for identifying ecological thresholds. Ecology 2003, 84, 2034-2041. UKTAG. UK Environmental Standards and Conditions (Phase 1). UK Technical Advisory Group on the Water Framework Directive, final report (SR1-2006), August 2006, www.wfduk.org. Leslie, H.; Kotterman, M.; Leonards, P. Monitoring Base: Collation and Evaluation of Monitoring Programmes and Measured Environmental Concentration Data on Organic Chemicals in European Aquatic Environments; Report C079/04 to CEFIC, Netherlands Institute for Fisheries Research:Ymuiden, The Netherlands. 2004.
(14) Fore, L. S.; Karr, J. R.; Wisseman, R. W. Assessing invertebrate responses to human activities: evaluating alternative approaches. J. North Am. Benthol. Soc. 1996, 15, 212-231. (15) Malmqvist, B.; Hoffsten, P. Influence of drainage from old mine deposits on benthic macroinvertebrate communities in Central Swedish streams. Water Res. 1999, 33, 2415-2423. (16) Beasley, G.; Kneale, P. E. Investigating the influence of heavy metals on macroinvertebrate assemblages using partial canonical correspondence analysis (pCCA). Hydrol. Earth Syst. Sci. 2003, 7, 221-233. (17) Liess, M.; Brown, C.; Dohmen, P.; Duquesne, S.; Hart, A.; Heimbach, F.; Kreuger, J.; Lagadic, L.; Maund, S.; Reinert, W.; Streloke, M.; Tarazona, J. Effects of Pesticides in the Field; SETAC Press: Pensacola, FL. 2005. (18) Struijs, J.; van de Meent, D.; Peijnenburg, W. J. G. M.; van den Hoop, M. A. G. T.; Crommentuijn, T. Added risk approach to derive maximum permissible concentrations for heavy metals: how to take natural background levels into account. Ecotoxicol. Environ. Safety 1997, 37, 112-118. (19) Leung, K. M. Y.; Bjorgesæter, A.; Gray, J. S.; Li, W. K.; Lui, G. C. S.; Wang, Y.; Lam, P. K. S. Deriving sediment quality guidelines from field-based species sensitivity distributions. Environ. Sci. Technol. 2005, 39, 5148-5156.
Received for review December 12, 2006. Revised manuscript received April 16, 2007. Accepted May 1, 2007. ES0629460
VOL. 41, NO. 14, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5021
