predicting metal toxicity in sediments: a critique of current approaches

14
Predicting Metal Toxicity in Sediments: A Critique of Current Approaches Stuart L Simpson* and Graeme E Batley Centre for Environmental Contaminants Research, CSIRO Land and Water, Private Mailbag 7, Bangor, NSW 2234, Australia (Received 25 October 2005; Accepted 17 February 2006) ABSTRACT The ability to predict metal toxicity in sediments based on measurements of simple chemical parameters is not possible using currently available sediment-quality guidelines (SQGs). Past evaluations of available SQGs for metals indicated little difference in their predictive abilities; however, the scientific understanding of cause–effect relationships is progressing rapidly. Today, it is clear that they can be protective of benthic ecosystem health, but single-value SQGs will be ineffective for predicting the toxicity of metals in sediments. Recent exposure–effects models and the sediment biotic ligand model both indicate that a better approach would be to have SQG concentrations, or ranges, that are applied to different sediment types. This review indicates that significant improvements in laboratory and field-based measurements, better recording of parameters that influence metal toxicity in sediments, as well as quantification of the metal exposure routes and the relative contribution of dissolved and particulate sources to toxic effects are needed to improve the power of predictive models and the overall effectiveness of SQGs for metals. Simply exposing benthic organisms to contaminated sediments and reporting effects concentrations or thresholds based on particulate metal concentrations will provide little information to aid future SQG development. For all tests, careful measurement and reporting of concentrations of particulate metal-binding phases (e.g., sulfide, organic carbon, and iron phases), metal partitioning between porewater and sediments, and porewater pH are considered as minimum data requirements. When using metal-spiked sediments, much better efforts are required to achieve sediment properties that resemble those of naturally contaminated sediments. Our current understanding of metal toxicity indicates that considerably greater information requirements will be needed to predict sublethal and chronic effects of metals, because the toxic, metabolically available concentration of metals within an organism will fluctuate over time. Based on the review of exposure and effects models, along with improved measurement of metal exposure-related parameters, the measurement of the short-term uptake rate of metals into organisms is likely to improve future models. Keywords: Metals Toxicity Exposure pathway Predictive model Sediment-quality guideline INTRODUCTION The toxicity of metal contaminants in sediments to benthic organisms is dependent on the bioavailability of metals in both the water (via exposure to porewater, burrow water, or overlying water) and sediment phases (via ingestion of particles) and on the sensitivity of the organism to these metal exposures (Lee et al. 2000a; Eriksson-Wiklund and Sundelin 2002; Besser et al. 2003; Riba et al. 2003; Di Toro et al. 2005; Simpson 2005). Although the bioavailability of metals in the overlying water may be predicted using models such as the biotic ligand model (BLM; Paquin et al. 2002; Niyogi and Wood 2004), the bioavailability of metals in sediments is more complex. The latter is controlled by the following: 1) speciation (e.g., metal binding with particulate sulfide, organic carbon, and iron hydroxide phases); 2) sediment–water partitioning relationships; 3) organism phys- iology (e.g., uptake rates from waters and assimilation efficiencies [AEs] from particulates); and 4) organism feeding and other behavior (e.g., feeding selectivity and burrow irrigation; Luoma and Rainbow 2005; Simpson 2005). To predict the toxic effects of metals in sediments, it is important to consider the relative importance of the sediment and water-column compartments as sources of metals to benthic organisms (Warren et al. 1998; Hare et al. 2003; Luoma and Rainbow 2005; Simpson 2005). Because the burrowing and feeding behavior of benthic organisms varies greatly, the importance of dissolved and particulate metal exposure routes also varies. For benthic organisms that irrigate their burrows with the oxygenated water overlying the sediment and that feed on particles in this burrow irrigation water, their metals are more likely to come from the water column than from sediments (Munger and Hare 1997; Hare et al. 2001). For many polychaete worms , ingestion of sediments is the major pathway of metal exposure (Selck et al. 1998; Wang et al. 1999; Lee et al. 2001; Yan and Wang 2002). Amphipod and bivalve species may deposit-feed and/ or suspension-feed and are exposed to metals associated with ingested particles as well as with porewater, burrow water, and overlying water (Luoma et al. 1992; Fan and Wang 2001; Eriksson-Wiklund and Sundelin 2002; Griscom and Fisher 2002, 2004; King et al. 2005). Consequently, a one-size-fits- all approach to assessing species sensitivity to metals generally is not appropriate, and careful consideration of metal exposure routes is necessary. The relative contribution by each exposure route to metal toxicity will depend on both the metal concentration in each compartment (e.g., overlying water, porewater, and sediment) and the relative importance of each compartment for the individual organism. The ability to predict sediment metal toxicity is becoming increasingly important for the assessment of contaminated sediments and for the development of sediment-quality guidelines (SQGs) (Wenning et al. 2005). The most common approach has been to interpret metal toxicity only in terms of * To whom correspondence may be addressed: [email protected] Integrated Environmental Assessment and Management — Volume 3, Number 1—pp. 18–31 18 ȑ 2007 SETAC Review

Upload: csiro

Post on 24-Apr-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Predicting Metal Toxicity in Sediments: A Critique ofCurrent ApproachesStuart L Simpson* and Graeme E Batley

Centre for Environmental Contaminants Research, CSIRO Land and Water, Private Mailbag 7, Bangor, NSW 2234, Australia

(Received 25 October 2005; Accepted 17 February 2006)

ABSTRACTThe ability to predict metal toxicity in sediments based on measurements of simple chemical parameters is not possible

using currently available sediment-quality guidelines (SQGs). Past evaluations of available SQGs for metals indicated little

difference in their predictive abilities; however, the scientific understanding of cause–effect relationships is progressing

rapidly. Today, it is clear that they can be protective of benthic ecosystem health, but single-value SQGs will be ineffective for

predicting the toxicity of metals in sediments. Recent exposure–effects models and the sediment biotic ligand model both

indicate that a better approach would be to have SQG concentrations, or ranges, that are applied to different sediment

types. This review indicates that significant improvements in laboratory and field-based measurements, better recording of

parameters that influence metal toxicity in sediments, as well as quantification of the metal exposure routes and the relative

contribution of dissolved and particulate sources to toxic effects are needed to improve the power of predictive models and

the overall effectiveness of SQGs for metals. Simply exposing benthic organisms to contaminated sediments and reporting

effects concentrations or thresholds based on particulate metal concentrations will provide little information to aid future

SQG development. For all tests, careful measurement and reporting of concentrations of particulate metal-binding phases

(e.g., sulfide, organic carbon, and iron phases), metal partitioning between porewater and sediments, and porewater pH are

considered as minimum data requirements. When using metal-spiked sediments, much better efforts are required to achieve

sediment properties that resemble those of naturally contaminated sediments. Our current understanding of metal toxicity

indicates that considerably greater information requirements will be needed to predict sublethal and chronic effects of

metals, because the toxic, metabolically available concentration of metals within an organism will fluctuate over time. Based

on the review of exposure and effects models, along with improved measurement of metal exposure-related parameters, the

measurement of the short-term uptake rate of metals into organisms is likely to improve future models.

Keywords: Metals Toxicity Exposure pathway Predictive model Sediment-quality guideline

INTRODUCTIONThe toxicity of metal contaminants in sediments to benthic

organisms is dependent on the bioavailability of metals in

both the water (via exposure to porewater, burrow water, or

overlying water) and sediment phases (via ingestion of

particles) and on the sensitivity of the organism to these

metal exposures (Lee et al. 2000a; Eriksson-Wiklund and

Sundelin 2002; Besser et al. 2003; Riba et al. 2003; Di Toro et

al. 2005; Simpson 2005). Although the bioavailability of

metals in the overlying water may be predicted using models

such as the biotic ligand model (BLM; Paquin et al. 2002;

Niyogi and Wood 2004), the bioavailability of metals in

sediments is more complex. The latter is controlled by the

following: 1) speciation (e.g., metal binding with particulate

sulfide, organic carbon, and iron hydroxide phases); 2)

sediment–water partitioning relationships; 3) organism phys-

iology (e.g., uptake rates from waters and assimilation

efficiencies [AEs] from particulates); and 4) organism feeding

and other behavior (e.g., feeding selectivity and burrow

irrigation; Luoma and Rainbow 2005; Simpson 2005).

To predict the toxic effects of metals in sediments, it is

important to consider the relative importance of the sediment

and water-column compartments as sources of metals to

benthic organisms (Warren et al. 1998; Hare et al. 2003;

Luoma and Rainbow 2005; Simpson 2005). Because the

burrowing and feeding behavior of benthic organisms variesgreatly, the importance of dissolved and particulate metalexposure routes also varies. For benthic organisms thatirrigate their burrows with the oxygenated water overlyingthe sediment and that feed on particles in this burrowirrigation water, their metals are more likely to come from thewater column than from sediments (Munger and Hare 1997;Hare et al. 2001). For many polychaete worms, ingestion ofsediments is the major pathway of metal exposure (Selck etal. 1998; Wang et al. 1999; Lee et al. 2001; Yan and Wang2002). Amphipod and bivalve species may deposit-feed and/or suspension-feed and are exposed to metals associated withingested particles as well as with porewater, burrow water,and overlying water (Luoma et al. 1992; Fan and Wang 2001;Eriksson-Wiklund and Sundelin 2002; Griscom and Fisher2002, 2004; King et al. 2005). Consequently, a one-size-fits-all approach to assessing species sensitivity to metals generallyis not appropriate, and careful consideration of metalexposure routes is necessary. The relative contribution byeach exposure route to metal toxicity will depend on both themetal concentration in each compartment (e.g., overlyingwater, porewater, and sediment) and the relative importanceof each compartment for the individual organism.

The ability to predict sediment metal toxicity is becomingincreasingly important for the assessment of contaminatedsediments and for the development of sediment-qualityguidelines (SQGs) (Wenning et al. 2005). The most commonapproach has been to interpret metal toxicity only in terms of

* To whom correspondence may be addressed: [email protected]

Integrated Environmental Assessment and Management — Volume 3, Number 1—pp. 18–3118 � 2007 SETAC

Review

dissolved metals, namely those present in the porewater, withporewater–sediment partitioning models being used to predictsediment effects concentrations for various metals that can berelated to water-quality guidelines (Batley et al. 2005). Theequilibrium partitioning (EqP) model based on acid-volatilesulfide (AVS)/simultaneously extracted metals (SEM) iswidely regarded as being accurate for predicting the lack oftoxicity of the metals cadmium (Cd), copper (Cu), nickel(Ni), lead (Pb), and zinc (Zn) in laboratory-contaminated,metal-spiked sediments and in field-contaminated sediments,and it now forms the basis for EqP sediment benchmarks formetals proposed by the US Environmental Protection Agency(USEPA 2005). Di Toro et al. (2005) recently extended theAVS/SEM-based EqP approach by coupling it to a BLM withmore explicit consideration of particulate organic carbon(POC) as a metal-binding phase. This sediment BLM (sBLM)was demonstrated to have considerable potential for predict-ing metal toxicity associated with metal-spiked sediments.

The EqP approach can be contrasted with multiphaseexposure and effects models, which explicitly consider toxiceffects from both porewater and sediment-ingestion exposureroutes (Simpson 2005). The exposure–effects model (EEM)approach described by Simpson (2005) is derived from abioenergetic-based kinetic model that describes the rate ofassimilation of metals by benthic organisms from thedissolved and particulate phases. Toxicity occurs when themetal exposure exceeds a threshold value. Like the sBLM, theEEM uses an EqP approach to predict the exposure todissolved metals; however, the EEM also considers metalassimilation from ingested particles.

The development of both the sBLM and EEM reliespredominantly on sediment toxicity data compiled fromstudies using metal-spiked sediments (Di Toro et al. 2005;Simpson 2005). Today, it is well recognized that inadequatesediment-spiking procedures will accentuate the partitioningof metals to the dissolved phase and shift the pathway formetal exposure from particulate to dissolved metals (Lee etal. 2000b, 2004; Simpson et al. 2004; Simpson 2005). Innaturally contaminated sediments, porewater metal concen-trations typically are in the sub- or low-lg/L range (especiallyin marine sediments), and exposure to particle-bound metalsmay be the major metal-exposure pathway (Luoma andRainbow 2005; Simpson 2005). For metal-spiked sedimentsthat have porewater and/or overlying water metal concen-trations that greatly exceed those of metal-contaminatedsediment environments (e.g., mg/L vs lg/L concentrations),toxicity thresholds determined in laboratory tests of thesesediments will not be widely applicable for developing SQGs.

Rainbow (2002) proposed that toxicity will occur when therate of metal uptake into the body of an organism exceeds thecombined rate of excretion and detoxification of metabol-ically available metal. Luoma and Rainbow (2005) showedthat the dynamic multipathway bioaccumulation model(DYM-BAM) could provide a unified explanation of chronicmetal bioaccumulation by a wide range of benthic organismsin a range of sediment environments. During chronicexposures, the metabolically available concentration (MAC)of a metal will fluctuate over time (not reaching a steadystate) as uptake rates (exposure sources) and efflux rates(excretion and detoxification) vary over time. Consequently,considerably greater information will be needed to predictsublethal and chronic effects of metals because of the need topredict, for all exposure sources, the uptake and detoxifica-

tion rates and to understand how these parameters influencethe MAC and the various chronic effects being assessed. Nocurrent models appear to address these issues adequately.

The purpose of the present review is to critique currentapproaches for predicting sediment metal toxicity. Limita-tions of both the data sets used for model development andthe predictive ability of the approaches are discussed. A goodpredictive model should predict the influence of sedimentproperties and organism physiology on toxic effects and,hence, how these factors will influence derived SQGs.Whereas many modeling approaches appear to be suitablefor progressing this science, the major current limitation is theavailability of good data sets. The requirements for generatingdata sets that will be applicable for evaluating a range ofmodeling approaches are outlined.

CONCENTRATION–RESPONSE MODELS FORPREDICTING SEDIMENT METAL TOXICITY

Ideally, SQGs should unequivocally distinguish betweensediments that cause biological effects and those that do not.However, in reality, the occurrence of biological effects doesnot show such a clearly delineated relationship (Batley et al.2005). A generalized concentration–response model (Figure1) has 3 distinct zones, comprising concentrations below thethreshold for effects, above the probable effects limit, and in atransition zone (TZ) between the two (Batley et al. 2005).Currently, greater confidence exists regarding our ability todefine the probable effects limit and threshold for effectszones; however, the TZ is poorly defined and may span morethan an order of magnitude of metal concentrations. This levelof uncertainty needs to be reduced, because the TZ alsoencompasses the concentration range of many contaminatedsediments that are of concern to regulators.

Effects data from toxicity tests are defined as statisticallysignificant relative to suitable control responses; consequently,the output is indicative of either effects or no effects. Factorsthat cause the overlap between effects and no-effects data arenumerous and include unaccounted for contributions fromuncharacterized chemicals or stressors, differences in bio-availability, differing responses among organisms, and errors inmeasurement of chemical and biological response parameters.Procedures exist for normalizing some of these factors (e.g.,

Figure 1. Generalized concentration–response relationship for contaminatedsediments. (Adapted with permission from Batley et al. 2005. Copyright 2005Society of Environmental Toxicology and Chemistry.)

Predicting Metal Toxicity in Sediments—Integr Environ Assess Manag 3, 2007 19

AVS and organic carbon), but no normalization procedureshave yet been successful for complete discrimination.

Outlined in the following sections are modeling approachescurrently being pursued for interpreting cause–effect data formetals in sediments.

AVS/SEM, SEM-AVS/fraction of organic carbon,and sBLM approaches

The AVS/SEM EqP theory predicts that when a molarexcess of AVS over SEM (Rsilver [Ag], Cd, Cu, Ni, Pb, Zn)exists, then these metals should not cause direct toxicity tobenthic organisms (Di Toro et al. 1990, 1992; Ankley et al.1996; Berry et al. 1996; USEPA 2000, 2005). Criticism of theAVS/SEM approach has been widespread regarding measure-ment-associated artifacts (Cooper and Morse 1998; Simpsonet al. 1998; O’Day et al. 2000; Simpson, Apte, et al. 2000).For example, in the 1 M HCl extraction used, CuS is almostinsoluble, NiS is only partially soluble, and PbS solubility iskinetically slow (Cooper and Morse 1998; Simpson et al.1998; Simpson, Apte, et al. 2000). The presence ofparticulate iron(III) phases (e.g., FeOOH and Fe(OH)3) thatdissolve during the AVS/SEM extraction procedure results inthe oxidative release of Cu from CuS, whereas the associatedsulfide is oxidized and not measured in the AVS fraction(Simpson et al. 1998). Nevertheless, because these metals willbe bound predominantly as sulfide phases when AVS/SEM .

1, the theory is sound.It was recognized that rather than a ratio, the difference

between SEM and AVS, which is indicative of the maximumexpected porewater metal concentrations, more accuratelyreflects the magnitude of metal bioavailability (Hare et al.1994). The inability of this difference ([SEM � AVS]) topredict toxicity was attributed to metal binding by othersediment components (e.g., organic carbon; Chapman et al.1998; Correia and Costa 2000; Besser et al. 2003). To accountfor binding by POC, the difference between SEM and AVScan be normalized to the fraction of organic carbon (foc) in thesediment when no toxicity is expected when [SEM � AVS]/foc , 1 (USEPA 2000).

The most recent extension of this procedure was thesBLM (Di Toro et al. 2005), in which a BLM was coupledwith a porewater–sediment partitioning model (based onWHAM V) (Tipping and Hurley 1992). It was used topredict the sediment concentration that is in equilibriumwith the biotic ligand effects concentration (Di Toro et al.2005). The sBLM model assumes that the residual metals ofthe SEM fraction that are in excess of the AVS metal-bindingcapacity are predominantly bound to the POC phases andthat other metal-binding phases in the sediments (e.g., ironhydroxide phases) can be ignored. The initial application ofthe sBLM to existing data sets indicated that the effects ofwater hardness, salinity, dissolved organic carbon (DOC),and competing ligands other than Hþ (pH) should have verylittle effect on the computed median effects concentrationon a sediment organic carbon (OC)–normalized basis(SEMoc ¼ [SEM � AVS]/foc, lmol/g OC). Only theporewater pH is important. Because the amount of metalbound to DOC and other ligands in the porewater is smallrelative to the amount of metal bound to POC, the DOCand other porewater ligands are ignored in the sBLM model.The reason for the minor effect of competing cations (e.g.,Ca2þ) other than Hþ is because of saturation of the POC-cation binding sites (Di Toro et al. 2005). The result of this

is that detailed computation of the porewater chemistrycould be bypassed.

The sBLM model assumes that the exposures from dietarysources (food and sediment ingestion pathways) do notcontribute to toxicity (Di Toro et al. 2005). This is the mostcontentious area with regard to metal toxicity in sediments(Munger and Hare 1997; Eriksson-Wiklund and Sundelin2002; Lee et al. 2004; Meyer et al. 2005; Simpson 2005;Simpson and King 2005) and will be discussed further later. Adiscussion of iron hydroxides as metal-binding phases andimplications for the sBLM by ignoring this phase also is givenlater.

The initial development and application of the sBLM reliedalmost entirely on data from metal-spiked sediment studies,for which porewater pH generally was not measured. On a log-concentration basis, the fit between measured acute andchronic effects and the SEMoc computed for the pH 6 to 9range (estimated) appears to be encouraging (Di Toro et al.2005). However, few data generally were available in the TZ(Figure 1), because the metal-spiked sediments generally hadeither an excess of AVS and negligible porewater metalconcentrations relative to the organism’s water-only effectsconcentration or very high (mg/L) porewater metal concen-trations as a consequence of [SEM�AVS] . 0 and inadequatemetal-spiking procedures resulting in low sediment–pore-water pH, possibly as low as pH 5 (Carlson et al. 1991; Casasand Crecelius 1994; Pesch et al. 1995; Berry et al. 1996;DeWitt et al. 1996, 1999; Sibley et al. 1996; Han et al. 2005).

Bioaccumulation and critical body residues

For nonessential and nonregulated metals, the expression oftoxicity on a body concentration basis often has been a betterindicator of biological effects than the use of total sedimentconcentrations (Borgmann 2000, 2003; Borgmann, Norwood,et al. 2001). The critical body residue (CBR)/lethal bodyconcentration (LBC) approach assumes that toxicity occurswhen a threshold body concentration (accumulation) isexceeded (regardless of the exposure pathway). For Cd, Ni,and Pb, relationships between sediment toxicity and CBRs ofthese metals were useful for predicting toxicity to thefreshwater amphipod Hyalella azteca (Borgmann 2003). WithH. azteca caged above the sediments and porewater, and withoverlying water Cd, Ni, and Pb concentrations comparable towater-only effects concentrations, the results indicated thatthe toxicity was caused by dissolved metals, not by metals inthe solid phase (Borgmann, Neron, et al. 2001). Copper andZn appeared to be regulated by H. azteca, and toxicity couldnot be related to body concentrations (Borgmann andNorwood 1997; Borgmann, Norwood, et al. 2001; Borgmann2003). There remain many difficulties associated with relatingbody concentration to the observed toxicity because of thedynamics of metal accumulation and associated removalprocesses (Rainbow 2002; Luoma and Rainbow 2005;Simpson and King 2005). Rainbow (2002) reviewed metalaccumulation strategies of aquatic invertebrates and con-cluded that body concentrations would be of limited value inexplaining metal toxicity. The deficiencies of the CBR/LBCapproach arise because 2 metal pools exist within organisms,1 that is metabolically active and 1 that is metabolicallyinactive. The internal sequestration of metals within organ-isms may rapidly change the bioavailability of ‘‘accumulated’’metals. Processes that detoxify the metabolically active metal,such as metal–metallothionein binding or metal granule

20 Integr Environ Assess Manag 3, 2007—SL Simpson and GE Batley

formation (Ahearn et al. 2004; Vijver et al. 2004; Amiard etal. 2006), greatly complicate the use of body concentrationdata for predicting effects.

Body concentrations of Cu in the amphipod Melitaplumulosa and the bivalve Tellina deltoidalis following 10-dwater and sediment exposures are shown in Figure 2(Simpson and King 2005). Although the mean body concen-trations (10 individuals, 3–4 replicates) provided goodrelationships with the Cu exposure concentration and couldbe used to define LBCs, the variability among individuals washigh. Past studies of metal accumulation using differentbenthic species and other metals have observed a similar highdegree of variability between individuals in laboratory-basedstudies (Borgmann and Norwood 1999; Borgmann, Neron,and Norwood. 2001; Borgmann 2003; Kahle and Zauke 2003)and for field-collected organisms (Tessier et al. 1984, 1993;Warren et al. 1998; Hare et al. 2001; Luoma and Rainbow2005). For M. plumulosa and T. deltoidalis, 10-d whole-sediment LC50 values for Cu were 1,300 and 1,020 lg/g,respectively, indicating median LBC values of approximately120 and 300 lg/g, respectively (Figure 2). Consequently, therespective median LBC values were observed in individualspresent in sediments having a wide range of whole-sedimentCu concentrations, making the use of LBCs for SQG purposesinappropriate. Furthermore, at high metal concentrations,increased mortality results in fewer organisms in sedimentsand decreases the precision of body concentration data. Thisvariability and increased uncertainty associated with bodyconcentration data cause difficulties in using these data fordefining the TZ region of Figure 1.

As for the AVS/SEM and sBLM model data sets (Di Toro etal. 2005), most of the data sets from which significant CBR–effects relationships have been derived are from metal-spikedsediments in which porewater or overlying water metalconcentrations are very high and provide the major contribu-

tion to toxic effects (Borgmann and Norwood 1997, 1999;Borgmann, Neron, et al. 2001).

Multipathway EEMs

With increasing evidence that exposure and effects can varywith dietary pathways, efforts are being made to develop andtest multiphase EEMs (Luoma and Rainbow 2005; Simpson2005). Simpson and King (2005) demonstrated that medianlethal concentration (LC50) data for short-term (4- to 10-d),water-only and whole-sediment exposures could be combinedwith bioenergetic-based kinetic models of exposure pathways(6- to 12-h metal uptake rate data) to explain the cause of Cutoxicity in whole-sediment toxicity tests using the amphipodM. plumulosa and the bivalve T. deltoidalis. In the lethalexposure concentration (LEC) approach, toxicity occurswhen the organism’s internal exposure to bioavailable copper(net uptake) exceeds a threshold value (e.g., LEC50, internalexposure concentration that causes 50% lethality). For theseorganisms, the LEC50 of copper was effectively the same forboth water-only and whole-sediment toxicity tests (Figure 3).For metals that are not regulated by organisms, the LEC isequivalent to the LBC (i.e., all the internal exposure isaccumulated). The LEC approach of Simpson and King(2005) was used to develop an exposure-effects model(EEM) for calculating metal effects concentrations for benthicorganisms in sediments with varying metal-binding properties(Simpson 2005). In the EEM, an organism’s internal exposureto metals was calculated as the sum of the internal exposurefrom all dissolved and particulate exposure routes. Thedissolved-phase exposure was calculated using an EqP-Kd

approach, and the particulate (ingestion)-phase exposure wascalculated using a relationship between the AE and theingestion rate of the particulate material (Simpson 2005).

Simpson (2005) developed EEMs for 9 benthic organismsand used these to predict the effects of sediment–water

Figure 2. Body concentrations of copper in the amphipod Melita plumulosa and bivalve Tellina deltoidalis following short-term water and sediment exposures.

Symbols represent the copper body concentrations measured at the end of the toxicity tests: 1 and 3 are results for individual organisms (in replicate tests),

and � is the mean of the replicate results for each test. Sediment tests were 10 d for both species. Water-only tests were 10 d for M. plumulosa and 4 d for T.

deltoidalis. Animals were depurated in clean seawater for 24 h before analyses. (Adapted with permission from Environ Sci Technol 2005, 39:837–843.

Copyright 2005 American Chemical Society.)

Predicting Metal Toxicity in Sediments—Integr Environ Assess Manag 3, 2007 21

partitioning (Kd) and Cu assimilation from ingested solids(AE) on toxic effects and to determine how these factors willinfluence derived SQG concentrations. For sediments withhigh Kd values, the contribution of the dissolved metalexposure pathway to the observed toxicity becomes negli-gible, and calculated LC50 values are mostly influenced bythe sediment exposure route. For sediments with low Kd

values, the uptake rates and LC50 values are influenced mostby a water exposure route and were similar to thosedetermined for water-only exposures. The modeling indicatedthat effects concentrations will be dependent on metalexposure from both water and particulate-ingestion exposureroutes and that improved mechanistic models of contaminantexposure, as influenced by both organism physiology andsediment properties, are needed to predict toxic effects insediments.

The major difficulties associated with the EEM approachare those of model calibration and prediction of effects onorganisms in the field. Varying filtration and feeding ratesbecause of water conditions or food availability will affect therate of exposure from water and food sources (Tran et al.2002; King et al. 2005). Although suitable methods exist forestimating Kd, methods for estimating AE and the organism’singestion rate require development (Simpson 2005). It also isto be expected that for some organisms, metals assimilatedthrough dissolved and particulate exposure pathways willcause toxic effects of different magnitudes, which need to beconsidered in future models.

Biodynamic models and predicting chronic effects

Biodynamic-based models (also known as bioenergetic- andtoxicokinetic-based models) show great promise for inter-preting metal accumulation by organisms with multipleexposure pathways (Thomann 1981; Wang and Fisher1999a; Vink 2002; Kahle and Zauke 2003; Luoma andRainbow 2005). Luoma and Rainbow (2005) showed that theDYM-BAM could provide a unified explanation of metalbioaccumulation by a wide range of benthic organisms in arange of sediment environments. The DYM-BAM uses metalconcentration data for dissolved and food (including sedi-ments) exposure sources and a biodynamic model for keyphysiological parameters that describe the metal influx andefflux rates for each organism from each metal and exposuresource to predict steady-state metal concentrations in benthicorganisms. Whereas model predictions of metal bioaccumu-lation were in close agreement with those in nature, extension

of the DYM-BAM for predicting toxic effects has the samechallenges as the EEM described by Simpson (2005), such asdetermining metal uptake and efflux rates from eachexposure source. As discussed, for metals that are regulatedby organisms (e.g., Cu and Zn), the rate of metal uptake(internal exposure) appears to be more important than thenet accumulation in determining effects thresholds (Simpsonand King 2005).

Rainbow (2002) proposed that toxicity will occur when therate of metal uptake into the body exceeds the combined ratesof excretion and detoxification of metabolically availablemetal. For the soil isopod Porcellio scaber, van Straalen et al.(2005) found that the rate of Zn accumulation was a superiorpredictor of toxicity compared to the Zn body concentration.Zinc toxicity to the isopod was attributed to dietary Znexposure, with mortality determined by the total Zn in thehepatopancreas, and growth reduction by the rate of Znaccumulation in the body. The authors reviewed applicabletoxicokinetic models, analyzing uptake rate data, and pro-posed the initial slope of the accumulation rate as a usefulindicator of bioavailability, but they acknowledged that thedynamics of metal bioavailability greatly complicated pre-dictions.

To use metal uptake rate data to predict toxic effects, thedynamics of the external metal exposure (e.g., depletion orfluctuation of the dissolved and particulate exposure sources),the dynamics of the concentration of metabolically availablemetal (internalized within the organism), and the site andmode of toxicity within the organism (e.g., specific cellularfunction effects) need to be understood. The rate of metaldetoxification of the accumulated metal within the organisms(e.g., metal–metallothionein binding and metal granuleformation) cannot yet be quantified. Without knowledge ofmetal detoxification rates, it is not possible to determine ormodel the effects of the true MACs of metals within anorganism. It also is likely that metal detoxification rates willvary for metals assimilated via different exposure routes (e.g.,dissolved vs particulate).

In the EEM approach, the total metal assimilated duringacute exposures (the short-term internal exposure) wasconsidered as metabolically available, regardless of whetherit was subsequently rendered nonmetabolically availablethrough excretion and detoxification processes. Duringchronic exposures (as might be expected for steady-statemetal accumulation), the MAC of metals will fluctuate asorganisms move within their environment and, hence, as the

Figure 3. Calculated copper exposures of Melita plumulosa and Tellina deltoidalis during 10-d water-only and whole-sediment toxicity tests. The exposure wascalculated as the organism’s internal exposure to copper (net assimilation, net uptake) in lg/g tissue dry weight. Diagonal lines represent the calculatedexposure for water-only (dashed) and whole-sediment (solid) toxicity tests. LEC50¼ internal exposure concentration that causes 50% lethality. The 10-d effectsconcentrations for sediments had Kd (Cu)¼ 53 104 L/kg, and dissolved copper concentrations were below the lowest observable effects concentration (LOEC)of the 10-d water-only exposures. (Adapted with permission from Environ Sci Technol 2005, 39:837–843. Copyright 2005 American Chemical Society.)

22 Integr Environ Assess Manag 3, 2007—SL Simpson and GE Batley

influx rates vary over time. A chronic effects threshold willdepend on the specific chronic effects being assessed; forexample, the chronic effects threshold for growth rate effectsmay be greater than the chronic effects threshold forreproductive effects. A chronic exposure and effects modelshould determine if the chronic effects threshold is exceededduring any stage of the chronic exposure. Because theorganism’s exposure history seldom will be known for chronicexposures, some estimates will be needed. Extensive data setscurrently are not available for better development of kinetic-based exposure and effects models.

Based on the above discussion, the extension of the DYM-BAM and EEM to the prediction of chronic effects appears tobe very challenging. Because of the dynamics of the MAC ofmetals, the flux of metals into organisms (uptake rate datafrom short-term exposures) rather than total concentrationsin organisms currently appears to be the best approach forquantifying the toxic effects of metals in benthic organisms(Rainbow 2002; Simpson 2005; van Straalen et al. 2005).

In contrast to the data used for the AVS/SEM and CBRmodels (Borgmann 2003; Di Toro et al. 2005), the data setsused for developing the DYN-BAM were based on naturallycontaminated sediments, and the EEM used metal-spikedsediments with low (lg/L range) dissolved (porewater andoverlying water) metal concentrations. Consequently, for theDYN-BAM and EEM, the effects of the porewater metalexposure were not exaggerated to the same extent as the dataused for the AVS/SEM, sBLM, and CBR models, and theimportant contribution of ingested sediment as an exposurepathway is clear.

Empirical models

Most current SQGs are based on empirical models derivedfrom matching chemistry and toxicity data. They typicallydefine concentrations associated with low and high proba-bilities of biological effects (Long, Field, et al. 1998; Long,MacDonald, et al. 1998; Field et al. 1999, 2002; Long et al.2000; MacDonald et al. 2000; Smith et al. 2003). A fulldiscussion of these approaches is provided by Batley et al.(2005). A major limitation of empirical models is that thedata are confounded by the co-occurrence of contaminants forwhich the observed toxicity typically is ascribed equally to allcontaminants. In natural settings, metal contamination oftenco-occurs with other contaminants, such as hydrocarbons(Long et al. 2000; Field et al. 2002; Smith et al. 2003;Spadaro et al. 2006). The use of logistic regression models forpredicting toxicity is an improvement on earlier empiricalmodels (Field et al. 1999, 2002; Smith et al. 2003). In thelogistic regression models approach, rather than specifyingthreshold values below which sediment toxicity is unlikely tooccur, concentration–response relationships are used toestimate the extent to which the probability of toxicityincreases as the contaminant concentrations increase. Inassessment terms, co-occurrence of contaminants is not anissue if the same contaminants always co-occur; however, itwill mean that at least some of the derived guidelines will beconservative. In principle, this might be good, but theconsequences for management in overestimating hazard canbe costly.

Several studies have evaluated the effectiveness of SQGsfor predicting effects (Long, Field, et al. 1998; Long,MacDonald, et al. 1998; O’Conner et al. 1998; MacDonaldet al. 2000; Shine et al. 2003). Shine et al. (2003) used

receiver operating characteristic curves, which statisticallyassess the discriminatory power of diagnostic tests, to evaluatethe effectiveness of a range of SQGs for metals. The receiveroperating characteristic curves were used to evaluate theextent to which the SQG threshold could classify a toxicsample as toxic and a nontoxic sample as nontoxic. Withrespect to specificity and sensitivity, the SQG approaches didnot differ in overall effectiveness. This analysis indicated thatsignificant improvements in SQG approaches (predictivemodels) would be needed to enhance overall SQG effective-ness.

EFFECTS DATA FOR METAL-CONTAMINATEDSEDIMENTS

Because, as discussed above, the interacting effects ofmultiple co-occurring contaminants complicate the interpre-tation of the observed organism response for each contami-nant, the preferred approach for developing cause–effectbased SQGs is determining species sensitivity to an individualmetal or specific groups of metals (in isolation from nonmetalcontaminants; Borgmann 2003; Di Toro et al. 2005; Simpson2005). Consequently, artificially metal-contaminated sedi-ments, created by spiking single or multiple metal salts intosediments, form the majority of sediment metal toxicity datasets used for developing predictive models (Borgmann 2003;Di Toro et al. 2005; Simpson 2005).

The introduction of metals into sediments initiates aplethora of transformations to existing sediment chemistry,and equilibration times are slow and dependent on a largevariety of different sediment properties (Simpson, Rosner, etal. 2000; Simpson et al. 2004). For determining speciessensitivity to metals, possibly the most important factor indetermining the valid use of data is the partitioning of metalsbetween the porewaters and the sediment particles (Kd ¼[sediment metal]/[porewater metal]). Added metals willdisplace iron(II) and other cations into the porewaters, andboth the residual added metal and the displaced metalshydrolyze, resulting in decreased pH and increased redoxpotential (Simpson, Rosner, et al. 2000; Simpson et al. 2004).The greater the concentration of added metal, the greaterthese effects will be, and it is not uncommon to measureiron(II) concentrations of more than 100 mg/L in theporewaters of metal-spiked sediments (Simpson et al. 2004;Han et al. 2005), whereas in natural sediments, iron(II)concentrations often are in the range of 1 to 20 mg/L (Kromet al. 2002; Simpson et al. 2004).

Spiking of sediments in an inert atmosphere (e.g., undernitrogen) can minimize oxidation of iron(II) displaced fromFeS (AVS) and other phases, but the displaced iron(II) may beoxidized during mixing with other sediment phases (e.g.,MnO2) (Simpson and Batley 2003). For sulfidic sediments,added metals that bind to sulfide (AVS) will not be hydro-lyzed. However, once the sulfide phase is exhausted (i.e.,SEM/AVS . 1), considerable hydrolysis of added metalsoccurs, and the pH decreases (Simpson, Rosner, et al. 2000).The relatively large pH-buffering capacity of seawater orhighly alkaline freshwaters is insignificant compared to thebuffering capacity of the sediment solid phase. The extent ofpH decrease thus will depend on the buffering capacity of thesediment and the properties of the metal that is added (e.g.,Cu will cause greater effects than Zn; Simpson et al. 2004).For large metal additions, the pH may drop below 5 (Simpsonet al. 2004).

Predicting Metal Toxicity in Sediments—Integr Environ Assess Manag 3, 2007 23

The final concentrations of metals in porewaters willdepend on a large number of factors, with the most importantbeing the types and concentrations present in the sediments ofmetal-binding phases, such as particulate sulfide (AVS),organic carbon, and iron, and on the porewater pH (Besseret al. 2003; Lee et al. 2004; Simpson et al. 2004). Particle sizewill affect the accessibility of the different metal-bindingphases, and high DOC concentrations may lead to an increasein porewater concentrations (although not necessarily inbiologically available forms; Besser et al. 2003). It is wellaccepted that with an excess of AVS over SEM, metalconcentrations in porewaters will remain low (Di Toro et al.1990, 1992; Casas and Crecelius 1994; Ankley et al. 1996;Berry et al. 1996; Lee et al. 2000b; Simpson, Rosner, and Ellis2000; USEPA 2005). When [SEM � AVS] . 0, sedimentphases other than AVS will become important in metalbinding (Oakley et al. 1981; Lion et al. 1982; Millward andMoore 1982; Luoma and Davis 1983; Tessier and Campbell1987; Tessier 1992; Stumm and Morgan 1996; Tessier et al.1996; Rivera-Duarte and Flegal 1997; Markwiese and Colberg2000; Trivedi and Axe 2000). Metal binding by POC and ironhydroxide phases is very pH-dependent, with the pHdependence of metal-adsorption isotherms being greatest inthe pH 5 to 7 range (Lion et al. 1982; Millward and Moore1982; Stumm and Morgan 1996), and Kd values typicallydecreasing substantially as pH decreases (Tessier 1992;Mahony et al. 1996; Stumm and Morgan 1996; Tessier etal. 1996; Trivedi and Axe 2000). Consequently, in metal-spiked sediments for which no precautions are taken toneutralize the pH effects associated with the hydrolysis ofadded metals by the addition of base, the metal-bindingcapacity of POC and iron phases often is easily exceeded. Asdiscussed above, in natural metal-contaminated sediments,porewater metal concentrations generally are in the sub- tolow-lg/L range (Tessier 1992), but poor metal-spikingprocedures frequently result in high-mg/L metal concentra-tions once SEM/AVS . 1 (Kemp and Swartz 1988; Carlson etal. 1991; Casas and Crecelius 1994; Pesch et al. 1995; Berry etal. 1996; DeWitt et al. 1996; Bat et al. 1998; Borgmann andNorwood 1999; Simpson, Rosner, and Ellis 2000; Hoss et al.2001; Kuhn et al. 2002; Han et al. 2005).

Equilibration rates of metal-spiked sediments vary consid-erably and are dependent on sediment and metal properties(Lee et al. 2004; Simpson et al. 2004). Results of both short-term laboratory experiments and long-term field experimentsindicate that following metal spiking, porewater metalconcentrations decline slowly and may take many months toreach equilibrium with the sediments (Hare et al. 1994;Warren et al. 1998; Hare et al. 2001; Lee et al. 2004; Simpsonet al. 2004). Schlekat and Luoma (2000) used the bivalveMacoma baltica to illustrate how short equilibration times formetal-spiked sediments may erroneously overemphasize theporewater exposure route. Porewater metals equilibrate fasterin sediments with high concentrations of metal-binding sites(e.g., particulate sulfide, organic matter, and iron hydroxidephases) and large surface areas (e.g., fine, silty sediments) thanin sandy sediments with low binding capacities. Equilibrationrates are faster at higher pH and slower at cooler temperatures(Simpson et al. 2004). The equilibration rate for porewatermetals is faster than that observed for weak acid-extractablemetals (i.e., those metals weakly bound to the sediments)(Simpson et al. 2004). Very long equilibration times may berequired to reduce the bioavailability of particulate metals

(that may be ingested by organisms) because of the slow re-equilibration rates for adsorbed metals or metal precipitates.However, regardless of equilibration time, if the pH decreasessignificantly because of metal-spiking procedures, porewatermetals concentrations will remain unrealistically high inmetal-spiked sediments.

Using concentration–response models (Figure 1) for theprediction of toxicity, sediments with [SEM� AVS] , 0 havemetal concentrations less than the threshold for effects. Formetal-spiked sediments with [SEM � AVS] . 0, the vastmajority of the available data are well into the probableeffects limit zone because of the high porewater metalconcentrations (Figure 4). This is quite evident in the sBLMmodel, in which the onset of toxicity occurs at SEMoc ’ 100lmol/g OC, regardless of the metal (Di Toro et al. 2005). Inother words, the porewater metal toxicity threshold isexceeded so greatly that the differences in toxicity of thevarious metals (Cd, Cu, Ni, Pb, and Zn) are lost. Data in theTZ (see TZ in Figure 1) are the most relevant for thedevelopment of models that accurately predict the onset ofmetal toxicity. The threshold for effects concentrations willexist for both porewater and sediment-bound metals, andaccording to the EEM of Simpson (2005), toxicity may occurin the TZ because of exposure from both these sources. Basedon most of the currently available data, the probable effectslimit zone will be dominated by porewater exposure effects.However, as better data sets become available, effects causedby sediment-bound metals are expected to be observed morecommonly at lower concentrations near the TZ (Simpson2005).

Figure 4 is compiled from the data sets used for the sBLMdevelopment, in which effects were reported for single metalsspiked into sediments and in which both porewater metalconcentrations and AVS/SEM were reported (Carlson et al.1991; Casas and Crecelius 1994; Pesch et al. 1995; Berry et al.1996; DeWitt et al. 1996, 1999; Sibley et al. 1996). Exceptfor sediments affected by low acidity, porewater metalconcentrations rarely exceed 100 lg/L (Carignan et al.1985; Tessier 1992; Brumbaugh et al. 1994; Ingersoll et al.1994; Tessier et al. 1996; Rivera-Duarte and Flegal 1997;Simpson, Rochford, et al. 2002; Simpson, Pryor, et al. 2002;Teasdale et al. 2003). In region A, porewater metal concen-trations are in the region of environmental realism. In regionB, porewater metal concentrations range from high to veryhigh and are at the limit of environmental realism. In regionC, porewater metal concentrations are extremely high and arewell beyond the limits of environmental realism. In region C,the low porewater pH (caused by hydrolysis of unboundmetals) greatly decreases metal binding to sediment particles,and pH is the major factor controlling metal partitioning.Because the porewater pH was either not measured or notreported in any of the studies from which the sBLM data werecollated, it is difficult to judge the state of equilibriumbetween the spiked metals and the sediments. Regardless ofthe use of the AVS/SEM-based models, region C in Figure 4would be expected to be classified as toxic because of the veryhigh and easily measurable porewater metal concentrations.

Like the AVS/SEM (sBLM) data sets, most of the betterrelationships between CBRs and sediment metal concentra-tions were based on metal-spiked sediment data sets with veryelevated porewater metal concentrations. For example,reduced survival of H. azteca in metal-spiked sediments wasa result of high dissolved metal concentrations in porewaters

24 Integr Environ Assess Manag 3, 2007—SL Simpson and GE Batley

and overlying waters (typically hundreds of lg/L of Cu, Ni,Pb, or Zn; Borgmann and Norwood 1997, 1999; Borgmann,Neron, and Norwood 2001) rather than of effects from thesediment particles. For much of this literature, effectsthresholds were derived in terms of particulate metalconcentrations. However, this is unreasonable, because thetoxic effects almost certainly resulted from dissolved metals inthe porewater and overlying water that were beyond theregion of environmental realism.

Fortunately, because of the exaggerated porewater expo-sure route, SQGs for metals derived from both the AVS/SEMand CBR data sets probably are quite conservative, unless it isshown that the exposure from particulate metal sourcescauses toxic effects at very low concentrations. To confusematters, however, several studies indicate that effects fromparticulate metals may occur at very low total metalconcentrations (Costa et al. 1998; Marsden and Wong 2001;Marsden 2002). For the amphipod species Gammarus locosta,Costa et al. (1998) determined an LC50 for Cu of 6.8 mg/kgdry weight for sediments in which porewater concentrationsof Cu were still less than half the LOEC in the water-onlyexposure (50 lg/L).

DIETARY EXPOSUREThe most significant difference between the models used

for predicting metal toxicity in sediments is the treatment ofthe exposure from dietary sources (food and sedimentingestion pathways). The sBLM model assumes that dietarysources do not contribute to toxicity (Di Toro et al. 2005), andthe CBR/LBC approaches do not consider exposure routes(Borgmann 2003). However, the EEM approach explicitlyconsiders dietary exposure and effects (Simpson 2005).

The significance of the dietary exposure pathway incontributing to metal exposure, accumulation, and toxiceffects has been highlighted for many years (Luoma 1989;Luoma et al. 1992; Luoma and Fisher 1997; Munger and Hare1997; Lee and Luoma 1998; Schlekat and Luoma 2000;Eriksson-Wiklund and Sundelin 2002; Fan et al. 2002; Yan

and Wang 2002; Hare et al. 2003; Griscom and Fisher 2004;Besser et al. 2005; King et al. 2005; Meyer et al. 2005).Luoma et al. (1992) showed that whereas exposure fromdissolved selenium was insufficient to explain seleniumaccumulation by the deposit-feeding bivalve Macoma balth-ica, ingestion and assimilation of selenium from sedimentparticulates could explain the accumulation. In a study of therelative importance of water and food as Cd sources to thepredatory insect Chaoborus punctipennis, Munger and Hare(1997) found that direct uptake from water probably wasunimportant. Chen and Maher (1999) observed that metalsolubilization in the organism’s gut generally is biphasic,initially rapid, then slower, and that steady-state equilibriumis unlikely to be reached within the gut passage time ofingested sediment particles. Chen et al. (2002) proposed thatmetal contaminants solubilized in digestive fluids may inhibitdigestive enzyme activities in the guts of marine benthicinvertebrates. For the estuarine amphipod Monporeia affinis,Eriksson-Wiklund and Sundelin (2002) showed that metals inthe sediments (food), rather than metals in the porewaters,were the main source of metal accumulation. Fan and Wang(2003) found that the extraction of metals by gut digestivefluids from 3 contaminated coastal sediments was notcorrelated with either SEM or [SEM � AVS].

The metal AE from different sediment phases will dependon the organism’s physiology (e.g., gut passage time and gutchemistry) as well as on the properties of the sediment phase(Wang et al. 1995; Wang and Fisher 1999b; Fan and Wang2001; Fan et al. 2002; Griscom et al. 2002; Griscom andFisher 2004; Simpson and King 2005). For the 2 bivalves M.balthica and Mytilus edulis, the AEs of Ag, Cd, and Coassociated with particles of differing geochemistry (e.g.,sulfide, organic carbon, and iron) varied by greater than anorder of magnitude (Griscom and Fisher 2004). For metal-contaminated sediments with varying sediment properties,Fan et al. (2002) found no statistically significant relationshipsbetween the AEs for Cd, Cr, and Zn of the clam Ruditapesphilippinarum and the concentrations of SEM or [SEM �

Figure 4. Concentration–response data used as the basis for acid-volatile sulfide (AVS)/simultaneously extracted metals (SEM) and sBLM model development.

Symbols represent sediments in which AVS/SEM . 1 (filled symbols) and AVS/SEM , 1 (open symbols) for the metals cadmium (� and ¤), copper (n and m),

nickel (� and *), lead (n and u), and zinc (n and u). Data are from Carlson et al. (1991), Casas and Crecelius (1994), Berry et al. (1996), DeWitt et al. (1996,

1999), Pesch et al. (1995), and Sibley et al. (1996).

Predicting Metal Toxicity in Sediments—Integr Environ Assess Manag 3, 2007 25

AVS]. The metal uptake for sediment particles is greatlyinfluenced by food quality, with metal uptakes rates fromsediments increasing in the presence of algae (Lee and Luoma1998; King et al. 2005). For sediments with low porewatermetal concentrations, the importance of the AE of differentsediment components becomes increasingly important, andincorporating this information in predictive EEMs is verychallenging (Simpson 2005).

A substantial amount of literature highlights the signifi-cance of dietary exposure routes, but it still is commonlyobserved that the significance is missed when organismsensitivity is assessed using metal-spiked sediments. AlthoughKemp and Swartz (1988) found that Cd bioavailability toamphipods was predictable by EqP approaches, close exami-nation of their data reveals that the porewater Cd concen-trations were more than 1,000-fold those found incontaminated estuarine waters and, consequently, that therelative importance of the sediment exposure pathways wasdiluted. As exemplified in Figure 4, this overexaggeration ofthe water exposure route is very common.

NONSULFIDE BINDING INFLUENCING METALTOXICITY

Both POC and iron oxyhydroxide phases are wellrecognized for their strong binding of metals in sediments(Luoma and Davis 1983; Tessier and Campbell 1987; Tessier1992; Tessier et al. 1993; Trivedi and Axe 2000). Numerousstudies have shown the importance of POC at reducing metaltoxicity in sediments, with normalization of metal effectsconcentrations to the fraction of organic carbon in sedimentsimproving the predictions of metal toxicity (Mahony et al.1996; Correia and Costa 2000; Hoss et al. 2001; Besser et al.2003). Few studies directly implicate iron as the major metal-binding phase that reduces metal toxicity in sediments,because the chemistry of metal-binding iron oxyhydroxidephases is considered to be much more complex than that ofmetal-binding POC phases (Burdige 1993; Tessier et al. 1993;Stumm and Morgan 1996; Williamson et al. 1999; Simpsonand Batley 2003).

The chemistry of particulate and dissolved forms of iron insediments is both complex and dynamic (Burdige 1993;Canfield et al. 1993; Tessier et al. 1993; Williamson et al.1999; Kristensen 2000; Simpson and Batley 2003). In manysediments, the cycling between reduced and oxidized formsof iron strongly influences metal partitioning (Burdige 1993;Kristensen 2000; Simpson and Batley 2003). The oxicfraction of silty sediments usually extends to depths of 2 to5 mm (Williamson et al. 1999; Kristensen 2000). At greaterdepths, the sediment becomes suboxic, containing mixturesof oxic solid phases (e.g., iron and manganese (hydr)oxides)in equilibrium with reduced dissolved phases (e.g., iron(II)and manganese(II)) (Simpson and Batley 2003). Once theeasily reducible iron and manganese (hydr)oxide phases havebeen depleted, bacteria reduce sulfate to sulfide, and thesolubility of many metals is controlled by the solubility ofmetal sulfide phases (Di Toro et al. 1992; Burdige 1993). Inmany sediments, redox zonation is not clear, and oxic(oxyhydroxides) and anoxic (sulfide) phases coexist (Burdige1993; Williamson et al. 1999; Simpson, Apte, et al. 2000).Depth distributions indicate that iron(III) hydroxide phasesand porewater iron(II) coexist in most surface sediments andthat an individual iron atom typically is oxidized and reducedhundreds of times before ultimate burial (Canfield et al.

1993; Williamson et al. 1999). The burrowing activity ofbenthic organisms can cause sediment resuspension andmixing of previously redox-stratified sediments with oxy-genated overlying waters, thereby altering metal sediment–water partitioning and speciation in the dissolved phase(Riedel et al. 1997; Simpson et al. 1998; Williamson et al.1999; Kristensen 2000; Simpson, Rosner, et al. 2002; Ciutatand Boudou 2003). As discussed above, the AE of metals,including iron, from particulate phases will vary greatly withparticle geochemistry (Fan and Wang 2001; Fan et al. 2002;Griscom and Fisher 2004).

In many sediments, it might be expected that theconcentrations of POC, particulate iron, and the fractions offine particles may be correlated and, therefore, that only 1 ofthese parameters may need to be included in an EEM.However, data for natural sediments (Figure 5) indicate thatthis commonly is not the case (Spadaro et al. 2006). Thedistribution of metal contaminants among the differentparticle size fractions may be similarly heterogeneous.

It is expected that both POC and iron oxyhydroxide phasesplay an important role in metal binding in most sediments,particularly the more oxic surface sediments with whichorganisms interact and feed (Tessier et al. 1993, 1996; Hare etal. 1994; Simpson and Batley 2003), and sediment particlesize also influences the bioavailability of these metal phases.Careful measurement of both these parameters is expected toimprove interpretation and allow more accurate modeling ofeffects data.

Porewater metal complexation and dynamics

The initial sBLM model calculations of Di Toro et al.(2005) indicated that DOC and competing ligands, other thanHþ (pH), should have very little effect on the computedmedian effects concentration on a POC-normalized basis(SEMoc). Besser et al. (2003) observed that the toxicity of Cdand Cu in porewater to Hyalella organisms decreased withincreasing humus concentrations in the sediments, and thiseffect was attributed to the formation of nonbioavailable Cu–organic complexes. Hoss et al. (2001) observed that thetoxicity of Cd in porewater to the nematode Caenorhabditiselegans increased as POC concentrations in the sedimentsincreased, and they attributed this finding to increasedconcentrations of Cd–organic complexes that were bioavail-able in the gut of the nematodes. These contrasting resultsindicate that further research is needed to understand betterthe nature of POC and porewater DOC.

As discussed earlier, metal-spiking of sediments oftencauses high concentrations of iron(II) in the porewaters(often .100 mg/L; Simpson et al. 2004; Han et al. 2005). Toour knowledge, no studies have reported the effects that highporewater iron(II) concentrations on the bioavailability ofother porewater metals. For sediment-dwelling organisms,exposure pathways become even more complicated when therelative importance of exposure from overlying water, pore-water, and burrow water needs to be considered (Griscom andFisher 2002). Because of varying irrigation and respirationrates of organisms in sediment burrows, concentrations ofdissolved oxygen and oxygen-sensitive ions (e.g., iron(II))within the burrow water and surrounding porewater willfluctuate according to the organism’s behavior (Kristensen2000; Wenzhofer and Glud 2004). Consequently, in thesuboxic waters of organism burrows and the surroundingporewaters, the dissolved concentrations of metal ions that

26 Integr Environ Assess Manag 3, 2007—SL Simpson and GE Batley

bind to iron hydroxide phases will fluctuate as the equilib-rium position of the iron(II)/iron(III)–(oxy)hydroxide redoxcouple shifts (Kristensen 2000; Simpson and Batley 2003).Equilibrium partitioning theories do not account for thedynamics that affect metal bioavailability in suboxic sedi-ments. Although many metal sulfide phases are relativelystable in the presence of oxygen (Simpson, Apte, et al. 1998,2000), a better understanding is needed of how organismactivity within sediments (including burrows) affects concen-trations of dissolved metals in the associated waters. Fewstudies have considered the importance of porewater colloids,but a significant amount of metal present in the filterableporewater fraction (typically, ,0.45 lm) likely is colloidal(Wang and Guo 2000; Cantwell and Burgess 2001).

DEVELOPING BETTER EFFECTS MODELS AND SQGSThe literature contains numerous examples of situations in

which the existing models and SQGs appear to be suitable forpredicting toxic effects of contaminants to benthic organisms(Shine et al. 2003; Word et al. 2005). However, as theemphasis of sediment-quality assessment moves furthertoward prediction of chronic effects, the effects models willrequire better treatment of the sediment factors that modifytoxicity and better distinction between effects from a range ofdifferent exposure pathways. The sBLM proposed by Di Toroet al. (2005) is elegant in its simplicity, and its initialapplication indicates great improvement compared with pastEqP models for predicting the effects of metals in sediments.The EEM developed by Simpson (2005) illustrates a novelattempt at modeling the exposure and effects of metals fromboth dissolved and particulate exposure pathways on sedi-ment biota. However, predictions based on the EEM indicatethat changes in metal uptake rate from the various exposure

routes are influenced greatly by organism behavior (e.g.,

feeding) and that adequate parameterization of the EEM is

difficult. Both the sBLM and EEM indicate that the develop-

ment of greatly improved SQGs will be dependent on

improved measurement and reporting of key parameters

affecting causality.

Measurement parameters

Poor documentation of procedures and the omission of

carefully measured sediment parameters greatly limit the use

of most current data sets for wider modeling applications. A

number of physicochemical parameters are crucial for

modeling metal partitioning and metal toxicity in sediments,

and a larger number of parameters may be useful for

improving the accuracy of predictive models. Parameters that

are considered to be crucial for development of the sBLM are

AVS, SEM, POC, and pH. However, if values for parameters

such as pH and porewater metal concentrations are outside

the realms of environmental realism, then the data sets may

be of little relevance. To support the development of EEMs,

then along with those 4 parameters, the porewater metal

concentrations (including iron and manganese) and the

particulate iron and manganese concentrations of the sedi-

ment should be reported as well. Furthermore, recent

literature should be reviewed regularly to consider any

‘‘new’’ parameters that are being used in model develop-

ment. For organisms that ingest sediments, sediment particle

size, the metal concentrations associated with the fine-

particle-size fraction, and the accessibility of the metal in the

organism’s gut (i.e., the metal AE) also will be key

parameters. However, beyond the use of radiotracer experi-

ments, methods for determining AEs for metals associated

Figure 5. Concentration of particulate iron, organic carbon, and percentage of particles less than 63 lm in size for 60 marine sediments from Sydney, Australia.Analyses were as follows: Particulate organic carbon (POC) by a high-temperature total organic carbon (TOC) analyzer, loss-on-ignition (LOI) at 3758C followinginorganic carbon removal, Total iron (u) by hot microwave digestion in 2:1 HCl:HNO3, and SEM-iron (¤) by 1 M HCl (30 min, cold). SEM-Fe¼ the iron concentrationin sediment determined by a 30-min extraction with 1 M HCl. (Adapted with permission from Spadaro et al. (2006). Copyright 2005 CSIRO Australia.)

Predicting Metal Toxicity in Sediments—Integr Environ Assess Manag 3, 2007 27

with different food sources remain to be developed (Wangand Fisher 1999b).

Measurement of the metal uptake rates by benthicorganisms for short-term exposures will provide useful datafor refining biodynamic models to predict effects (Rainbow2002; Simpson 2005; van Straalen et al. 2005). Measurementsof internal metal sequestration will improve our under-standing of the dynamics of the MAC of metals withinorganisms and of how the impairment of specific cellularfunction leads to quantifiable effects (Ahearn et al. 2004;Vijver et al. 2004).

Future challenges for effects models

The influence of organism behavior provides furtherdifficulties for predictive models, and perhaps the mostchallenging is selective feeding that depends on the nutritionalvalue of the sediment. Selective feeding behavior of benthicorganisms (feeding rate and particle selection) influences theuptake and toxic effects of metals in sediments (Munger andHare 2000; Lee et al. 2001; de Haas et al. 2004; King et al.2005). For the amphipod M. plumulosa, the presence of algaeincreased the rate of uptake of both Cu and Cd fromsediments and indicated that M. plumulosa fed more rapidly,but not necessarily selectively, when algae were present withsediments (King et al. 2005). The effect of varying feedingbehavior was not considered in the current models, but innatural sediments, many organisms may selectively ingest, forexample, organic-rich fine particles that have higher contam-inant concentrations than nonselected particles (e.g., the sandcomponent of the sediment).

Sediments are heterogeneous, and the speciation ofparticulate metals varies greatly over small spatial scales.Therefore, partitioning of contaminants is easily disrupted byanimal activity (Zhang et al. 2002; Ciutat and Boudou 2003;Simpson and Batley 2003). The spatial variability, hetero-geneity, and localized partitioning of metals within micro-environments, which exist in most sediments, eventually maylimit the accuracy of EqP relationships for predicting pore-water metal concentrations (Besser et al. 1996; Grabowski etal. 2001; Zhang et al. 2002; van Griethuysen et al. 2003). Theexistence of thermodynamic EqP may be rare in sedimentsthat are continually disrupted by animal activity at smallspatial (lm to mm) and temporal (s to min) scales (Forbes1999; Simpson and Batley 2003). Understanding how thesefactors affect species sensitivity requires further study todetermine whether assumptions of heterogeneity and con-taminant dynamics can be ignored or require greater consid-eration (Forbes 1999).

Although the additivity of the effects of metals appears tobe accounted for in the AVS/SEM and CBR approaches,multiple-metal effects are not always additive (Hagopian-Schlekat et al. 2001). Dealing with the effects additivity ofmetals from multiple exposure pathways adds greatercomplexity to effects prediction (Simpson 2005). However,for the development of better EEM, these issues probably areminor compared to those regarding the dynamics of metalpartitioning to POC and iron hydroxide phases.

Developing better SQGs

Despite the uncertainties discussed above, using SQGs asscreening or trigger values ensures that they serve some usefulpurpose in sediment-quality assessment, particularly in

association with other lines of evidence as part of a weight-of-evidence approach (Chapman et al. 2002). Better models willsimply assist in refining the trigger thresholds. Using suitabledata for a range of benthic test organisms, it will be possible touse species sensitivity distributions to derive SQGs that areprotective of a given percentage of sediment species, ascurrently is done for water-quality guidelines. The difference,however, will be that because the exposure concentration willbe dependent on sediment characteristics that controlcontaminant binding, such species sensitivity distributionswill need to be derived separately for a range of sedimenttypes, as indicated in Figure 6. As discussed by Simpson(2005), EEMs indicate that the ‘‘position’’ of species withinthese distributions may move as the relative importance ofexposure pathways changes with varying sediment properties.How well cumulative frequency distributions (e.g., thoseshown in Figure 6) will collapse to form a single distributionwhen normalized on the basis of sediment properties (e.g.,AVS, SEM, POC, iron, and pH) may depend on the organismsensitivity to metals from each exposure pathway and onchanges in the relative importance of each exposure pathwayfor different sediment types. Regardless of the SQG approachthat is used, careful scrutiny should be given to allexperimental procedures used in estimating toxicity thresh-olds or effects concentrations before such values are adoptedin SQGs.

Acknowledgment—We would like to thank Jenny Stauber(CSIRO), Kay Ho (USEPA), and 2 anonymous reviewers forconstructive comments on this manuscript.

ReferencesAhearn GA, Mandal PK, Mandal A. 2004. Mechanisms of heavy-metal

sequestration and detoxification in crustaceans: A review. J Comp Physiol B

174:439–452.

Amiard JC, Amiard-Triquet C, Barka S, Pellerin J, Rainbow PS. 2006. Metal-

lothioneins in aquatic invertebrates: Their role in metal detoxification and their

use as biomarkers. Aquat Toxicol 76:160–202.

Ankley GT, Di Toro DM, Hansen DJ, Berry WJ. 1996. Technical basis and proposal

for deriving sediment-quality criteria for metals. Environ Toxicol Chem

15:2056–2066.

Bat L, Raffaelli D, Marr IL. 1998. The accumulation of copper, zinc, and cadmium by

the amphipod Corophium volutator (Pallas). J Exp Mar Biol Ecol 223:167–184.

Figure 6. Conceptual species sensitivity distributions (SSDs) based on acumulative frequency model representative of toxic effects in sediments withvarying sediment properties. Symbols represent different classes of sediment-dwelling biota, such as bacteria (diamond), algae (circle), mysids (hexagon),amphipods (cross), bivalves (4-sided star), polychaete worms (triangle), snails(5-sided star), and crabs (square).

28 Integr Environ Assess Manag 3, 2007—SL Simpson and GE Batley

Batley GE, Stahl RG, Babut MP, Bott TL, Clark JR, Field LJ, Ho KT, Mount DR, Swartz

RC, Tessier A. 2005. Scientific underpinnings of sediment-quality guidelines.

In: Wenning RJ, Batley GE, Ingersoll CG, Moore DW, editors. Use of sediment-

quality guidelines and related tools for the assessment of contaminated

sediments. Pensacola (FL): Society of Environmental Toxicology and Chemistry.

p 39–120.

Berry WJ, Hansen DJ, Mahony JD, Robson DL, Di Toro DM, Shipley BP, Rogers B,

Corbin JM, Boothman WS. 1996. Predicting the toxicity of metal-spiked

laboratory sediments using acid-volatile sulfide and interstitial water normal-

izations. Environ Toxicol Chem 15:2067–2079.

Besser JM, Ingersoll CG, Giesy JP. 1996. Effects of spatial and temporal variation of

acid-volatile sulfide on the bioavailability of copper and zinc in freshwater

sediments. Environ Toxicol Chem 5:286–293.

Besser JM, Brumbaugh WG, May TW, Ingersoll CG. 2003. Effect of organic

amendments on the toxicity and bioavailability of cadmium and copper in

spiked formulated sediments. Environ Toxicol Chem 22:805–815.

Besser JM, Brumbaugh WG, Brunson EL, Ingersoll CG. 2005. Acute and chronic

toxicity of lead in water and diet to the amphipod Hyalella azteca. Environ

Toxicol Chem 24:1807–1815.

Borgmann U. 2000. Methods for assessing the toxicological significance of metals

in aquatic ecosystems: Bioaccumulation-toxicity relationships, water concen-

trations, and sediment-spiking approaches. Aquat Ecosyst Health Manag

3:277–289.

Borgmann U. 2003. Derivation of cause-effect based sediment quality guidelines.

Can J Fish Aquat Sci 60:352–360.

Borgmann U, Norwood WP. 1997. Toxicity and accumulation of zinc and copper in

Hyalella azteca exposed to metal-spiked sediments. Can J Fish Aquat Sci

54:1046–1054.

Borgmann U, Norwood WP. 1999. Assessing the toxicity of lead in sediments to

Hyalella azteca: The significance of bioaccumulation and dissolved metal. Can

J Fish Aquat Sci 56:1494–1503.

Borgmann U, Neron R, Norwood WP. 2001. Quantification of bioavailable nickel in

sediments and toxic thresholds to Hyalella azteca. Environ Pollut 111:189–

198.

Borgmann U, Norwood WP, Reynoldson TB, Rosa F. 2001. Identifying cause in

sediment assessments: Bioavailability and the Sediment-Quality Triad. Can J

Fish Aquat Sci 58:950–960.

Brumbaugh WG, Ingersoll CG, Kemble NE, May TW, Zajicek JL. 1994. Chemical

characterization of sediments and porewater from the upper Clark Fork River

and Milltown Reservoir, Montana. Environ Toxicol Chem 13:1971–1983.

Burdige DJ. 1993. The biogeochemistry of manganese and iron reduction in

marine sediments. Earth-Sci Rev 35:249–284.

Canfield DE, Thamdrup B, Hansen JW. 1993. The anaerobic degradation of

organic matter in Danish coastal sediments: Iron reduction, manganese

reduction, and sulfate reduction. Geochim Cosmochim Acta 57:3867–3883.

Cantwell MG, Burgess RM. 2001. Metal-colloid partitioning in artificial interstitial

waters of marine sediments: Influences of salinity, pH, and colloidal organic

carbon concentration. Environ Toxicol Chem 20:2420–2427.

Carignan R, Rapin F, Tessier A. 1985. Sediment porewater sampling for metal

analysis: A comparison of techniques. Geochim Cosmochim Acta 49:2493–

2497.

Carlson AR, Phipps GL, Mattson VR, Kosian PA, Cotter AM. 1991. The role of acid-

volatile sulfide in determining cadmium bioavailability and toxicity in

freshwater sediments. Environ Toxicol Chem 10:1309–1319.

Casas AM, Crecelius EA. 1994. Relationship between acid volatile sulfide and the

toxicity of zinc, lead, and copper in marine sediments. Environ Toxicol Chem

13:529–536.

Chapman PM, Wang FY, Janssen C, Persoone G, Allen HE. 1998. Ecotoxicology of

metals in aquatic sediments: binding and release, bioavailability, risk assess-

ment, and remediation. Can J Fish Aquat Sci 55:2221–2243.

Chapman PM, McDonald BG, Lawrence GS. 2002. Weight-of-evidence issues and

frameworks for sediment quality (and other) assessments. Hum Ecol Risk

Assess 8:1489–1515.

Chen Z, Mayer LM. 1999. Sedimentary metal bioavailability determined by the

digestive constraints of marine deposit feeders: Gut retention time and

dissolved amino acids. Mar Ecol Prog Series 176:139–151.

Chen Z, Mayer LM, Weston DP, Bock MJ, Jumars PA. 2002. Inhibition of digestive

enzyme activities by copper in the guts of various marine benthic

invertebrates. Environ Toxicol Chem 21:1243–1248.

Ciutat A, Boudou A. 2003. Bioturbation effects on cadmium and zinc transfers

from a contaminated sediment and on metal bioavailability to benthic

bivalves. Environ Toxicol Chem 22:1574–1581.

Cooper DC, Morse JW. 1998. Extractability of metal sulfide minerals in acidic

solutions: Application to environmental studies of trace metal contamination

within anoxic sediments. Environ Sci Technol 32:1076–1078.

Correia AD, Costa MH. 2000. Effects of sediment geochemical properties on the

toxicity of copper-spiked sediments to the marine amphipod Gammarus

locusta. Sci Total Environ 247:99–106.

Costa FO, Correia AD, Costa MH. 1998. Acute marine sediment toxicity: A

potential new test with the amphipod Gammarus locusta. Ecotoxicol Environ

Saf 40:81–87.

de Haas EM, Paumen ML, Koelmans AA, Kraak MHS. 2004. Combined effects of

copper and food on the midge Chironomus riparius in whole-sediment

bioassays. Environ Pollut 127:99–107.

DeWitt TH, Swartz RC, Hansen DJ, McGovern D, Berry WJ. 1996. Bioavailability

and chronic toxicity of cadmium in sediment to the estuarine amphipod

Leptocheirus plumulosus. Environ Toxicol Chem 15:2095–2101.

DeWitt TH, Hickey CW, Morrisey DJ, Nipper MG, Roper DS, Williamson RB, Vandam

L, Williams EK. 1999. Do amphipods have the same concentration-response to

contaminated sediment in situ as in vitro? Environ Toxicol Chem18:1026–1037.

Di Toro DM, Mahony JD, Hansen DJ, Scott KJ, Hicks MB, Mayr SM, Redmond MS.

1990. Toxicity of cadmium in sediments: The role of acid-volatile sulfide.

Environ Toxicol Chem 9:1487–1502.

Di Toro DM, Mahony JD, Hansen DJ, Scott KJ, Carlson AR, Ankley GT. 1992. Acid-

volatile sulfide predicts the acute toxicity of cadmium and nickel in sediments.

Environ Sci Technol 26:96–101.

Di Toro DM, McGrath JA, Hansen DJ, Berry WJ, Paquin PR, Rooni M, Wu KW,

Santore RC. 2005. Predicting sediment metal toxicity using a sediment biotic

ligand model: Methodology and initial application. Environ Toxicol Chem

24:2410–2427.

Eriksson-Wiklund AK, Sundelin B. 2002. Bioavailability of metals to the amphipod

Monoporeia affinis: Interactions with authigenic sulfides in urban brackish-

water and freshwater sediments. Environ Toxicol Chem 21:1219–1228.

Fan WH, Wang WX. 2001. Sediment geochemical controls on Cd, Cr, and Zn

assimilation by the clam Ruditapes philippinarum. Environ Toxicol Chem

20:2309–2317.

Fan WH, Wang WX. 2003. Extraction of spiked metals from contaminated coastal

sediments: A comparison of different methods. Environ Toxicol Chem

22:2659–2666.

Fan WH, Wang WX, Chen JS. 2002. Geochemistry of Cd, Cr, and Zn in highly

contaminated sediments and its influences on assimilation by marine bivalves.

Environ Sci Technol 36:5164–5171.

Field LJ, Macdonald DD, Norton SB, Severn CG, Ingersoll CG. 1999. Evaluating

sediment chemistry and toxicity data using logistic regression modeling.

Environ Toxicol Chem 18:1311–1322.

Field LJ, Macdonald DD, Norton SB, Ingersoll CG, Severn CG, Smorong D,

Lindskoog R. 2002. Predicting amphipod toxicity from sediment chemistry

using logistic regression models. Environ Toxicol Chem 21:1993–2005.

Forbes TL. 1999. Understanding small-scale processes controlling the bioavail-

ability of organic contaminants to deposit-feeding benthos. In: Gray JS,

Ambrose Jr W, Szaniawska A, editors. Biogeochemical cycling and sediment

ecology. Kluwer-NATO Science Partnership, Subseries 2. Environmental

Security 59:125–136.

Grabowski LA, Houpis JLJ, Woods WI, Johnson KA. 2001. Seasonal bioavailability

of sediment-associated heavy metals along the Mississippi River floodplain.

Chemosphere 45:643–651.

Griscom SB, Fisher NS. 2002. Uptake of dissolved Ag, Cd, and Co by the clam

Macoma balthica: Relative importance of overlying water, oxic porewater, and

burrow water. Environ Sci Technol 36:2471–2478.

Griscom SB, Fisher NS. 2004. Bioavailability of sediment-bound metals to marine

bivalve mollusks: An overview. Estuaries 27:826–838.

Griscom SB, Fisher NS, Aller RC, Lee BG. 2002. Effects of gut chemistry in marine

bivalves on the assimilation of metals from ingested sediment particles. J Mar

Res 60:101–120.

Hagopian-Schlekat T, Chandler GT, Shaw TJ. 2001. Acute toxicity of 5 sediment-

associated metals, individually and in mixture, to the estuarine meiobenthic

harpacticoid copepod Amphiascus tenuiremis. Mar Environ Res 51:247–

264.

Predicting Metal Toxicity in Sediments—Integr Environ Assess Manag 3, 2007 29

Han JB, Ma DY, Quan X, Wang JY, Yan QL. 2005. Bioavailability of zinc in the

sediment to the estuarine amphipod Grandidierella japonica. Hydrobiologia

541:149–154.

Hare L, Carignan R, Huerta-Diaz MA. 1994. A field study of metal toxicity and

accumulation by benthic invertebrates: Implications for the acid-volatile

sulfide (AVS) model. Liminol Oceanogr 39:1653–1668.

Hare L, Tessier A, Warren L. 2001. Cadmium accumulation by invertebrates living

at the sediment-water interface. Environ Toxicol Chem 20:880–889.

Hare L, Tessier A, Borgmann U. 2003. Metal sources for freshwater invertebrates:

Pertinence for risk assessment. Human Ecological Risk Assessment 9:779–793.

Hoss S, Henschel T, Haitzer M, Traunspurger W, Steinberg CEW. 2001. Toxicity of

cadmium to Caenorhabditis elegans (Nematoda) in whole sediment and

porewater: The ambiguous role of organic matter. Environ Toxicol Chem

20:2794–2801.

Ingersoll CG, Brumbaugh WG, Dwyer FJ, Kemble NE. 1994. Bioaccumulation of

metals by Hyalella azteca exposed to contaminated sediments from the upper

Clark Fork River, Montana. Environ Toxicol Chem 13:2013–2020.

Kahle J, Zauke GP. 2003. Bioaccumulation of trace metals in the Antarctic

amphipod Orchomene plebs: Evaluation of toxicokinetic models. Mar Environ

Res 55:359–384.

Kemp PF, Swartz RC. 1988. Acute toxicity of interstitial and particle-bound

cadmium to a marine infaunal amphipod. Marine Environ Res 26:135–153.

King CK, Simpson SL, Smith SV, Stauber JL, Batley GE. 2005. Short-term

accumulation of Cd and Cu from water, sediment, and algae by the amphipod

Melita plumulosa and the bivalve Tellina deltoidalis. Mar Ecol Prog Ser

287:177–188.

Kristensen E. 2000. Organic matter diagenesis at the oxic/anoxic interface in

coastal marine sediments, with emphasis on the role of burrowing animals.

Hydrobiologia 426:1–24.

Krom MD, Mortimer RJG, Poulton SW, Hayes P, Davies IM, Davison W, Zhang H.

2002. In-situ determination of dissolved iron production in recent marine

sediments. Aquat Sci 64:282–291.

Kuhn A, Munns W, Serbst J, Edwards P, Cantwell M, Gleason T, Pelletier M, Berry

WJ. 2002. Evaluating the ecological significance of laboratory response data

to predict population-level effects for the estuarine amphipod. Ampelisca

abdita. Environ Toxicol Chem 21:865–874.

Lee BG, Luoma SN. 1998. Influence of microalgal biomass on absorption efficiency

of Cd, Cr, and Zn by 2 bivalves from San Francisco Bay. Limnol Oceanogr

43:1455–1466.

Lee JS, Lee BG, Luoma SN, Choi HJ, Koh CH, Brown CL. 2000a. Influence of acid-

volatile sulfides and metal concentrations on metal bioavailability to marine

invertebrates in contaminated sediments. Environ Sci Technol 34:4517–4523.

Lee JS, Lee BG, Luoma SN, Choi HJ, Koh CH, Brown CL. 2000b. Influence of acid-

volatile sulfides and metal concentrations on metal partitioning in contami-

nated sediments. Environ Sci Technol 34:4511–4516.

Lee JS, Lee BG, Yoo H, Koh CH, Luoma SN. 2001. Influence of reactive sulfide (AVS)

and supplementary food on Ag, Cd, and Zn bioaccumulation in the marine

polychaete Neanthes arenaceodentata. Mar Ecol Prog Ser 216:129–140.

Lee JS, Lee BG, Luoma SN, Yoo H. 2004. Importance of equilibration time in the

partitioning and toxicity of zinc in spiked sediment bioassays. Environ Toxicol

Chem 23:65–71.

Lion LW, Altmann RS, Leckie JO. 1982. Trace-metal adsorption characteristics of

estuarine particulate matter: Evaluation of contributions of Fe/Mn oxide and

organic surface coatings. Environ Sci Technol 16:660–666.

Long ER, Field LJ, MacDonald DD. 1998. Predicting toxicity in marine sediments

with numerical sediment-quality guidelines. Environ Toxicol Chem 17:714–

727.

Long ER, MacDonald DD, Cubbage JC, Ingersoll CG. 1998. Predicting the toxicity

of sediment-associated trace metals with simultaneously extracted trace

metal: Acid-volatile sulfide concentrations and dry weight-normalized

concentrations: A critical comparison. Environ Toxicol Chem 17:972–974.

Long ER, MacDonald DD, Severn CG, Hong CB. 2000. Classifying probabilities of

acute toxicity in marine sediments with empirically derived sediment-quality

guidelines. Environ Toxicol Chem 19:2598–2601.

Luoma SN. 1989. Can we determine the biological availability of sediment bound

trace elements? Hydrobiologia 176/177:379–396.

Luoma SN, Davis JA. 1983. Requirements for modeling trace-metal partitioning in

oxidized estuarine sediments. Mar Chem 12:159–181.

Luoma SN, Fisher NS. 1997. Uncertainties in assessing contaminant exposure from

sediment. In: Ingersoll CG, Dillon T, Biddinger GR, editors. Ecological risk

assessment of contaminated sediment. Pensacola (FL): Society of Environ-

mental Toxicology and Chemistry. p 211–237.

Luoma SN, Rainbow PS. 2005. Why is metal bioaccumulation so variable?

Biodynamics as a unifying concept. Environ Sci Technol 39:1921–1931.

Luoma SN, Johns C, Fisher NS, Steinberg NA, Oremland RS, Reinfelder JR. 1992.

Determination of selenium bioavailability to a benthic bivalve from particulate

and solute pathways. Environ Sci Technol 26:485–491.

MacDonald DD, Ingersoll CG, Berger TA. 2000. Development and evaluation of

consensus-based sediment-quality guidelines for freshwater ecosystems. Arch

Environ Contam Toxicol 39:20–31.

Mahony JD, Di Toro DM, Gonzalez AM, Curto M, Dilg M, De Rosa LD, Sparrow LA.

1996. Partitioning of metals to sediment organic carbon. Environ Toxicol

Chem 15:2187–2197.

Markwiese JT, Colberg PJS. 2000. Bacterial reduction of copper-contaminated

ferric oxide: Copper toxicity and the interaction between fermentative and

iron-reducing bacteria. Arch Environ Contam Toxicol 38:139–146.

Marsden ID. 2002. Life-history traits of tube-dwelling corophioid amphipod,

Paracorophium excavatum, exposed to sediment copper. J Exp Mar Bio Ecol

270:57–72.

Marsden ID, Wong CHT. 2001. Effects of sediment copper on a tube-dwelling

estuarine amphipod, Paracorophium excavatum. Mar Freshwater Res

52:1007–1014.

Meyer J, Adams W, Brix K, Luoma S, Mount D, Stubblefield W, Wood C. 2005.

Toxicity of diet-borne metals to aquatic organisms. Pensacola (FL): Society of

Environmental Toxicology and Chemistry. 303 p.

Millward GE, Moore RM. 1982. The adsorption of Cu, Mn, and Zn by iron

oxyhydroxide in model estuarine solutions. Water Res 16:981–985.

Munger C, Hare L. 1997. Relative importance ofwater and food as cadmium sources

to an aquatic insect (Chaoborus punctipennis): Implications for predicting

cadmium bioaccumulation in nature. Environ Sci Technol 31:891–895.

Munger C, Hare L. 2000. Influence of ingestion rate and food types on cadmium

accumulation by the aquatic insect Chaoborus. Can J Fish Aquat Sci 57:327–

332.

Niyogi S, Wood CM. 2004. Biotic ligand model, a flexible tool for developing site-

specific water-quality guidelines for metals. Environ Sci Technol 38:6177–

6192.

O’Conner TP, Daskalakis KD, Hyland JL, Paul JF, Summers JK. 1998. Comparisons of

sediment toxicity with predictions based on chemical guidelines. Environ

Toxicol Chem 17:468–471.

O’Day PA, Carroll SA, Randall S, Martinelli RE, Anderson SL, Jelinski J, Knezovich JP.

2000. Metal speciation and bioavailability in contaminated estuary sediments,

Alameda Naval Air Station, California. Environ Sci Technol 34:3665–3673.

Oakley SM, Nelson PO, Williamson KJ. 1981. Model of trace-metal partitioning in

marine sediments. Environ Sci Technol 15:474–480.

Paquin PR, Gorsuch JW, Apte SC, Batley GE, Bowles KC, Campbell PGC, Delos CG,

Di Toro DM, Dwyer RL, Galvez F, Gensemer RW, Goss GG, Hogstrand C,

Janssen CR, McGeer JC, Naddy RB, Playle RC, Santore RC, Schneider U,

Stubblefield WA, Wood CM, Wu KB. 2002. The biotic ligand model: a

historical overview. Comp Biochem Physiol 133C:3–35.

Pesch CE, Hansen DJ, Boothman WS, Berry WJ, Mahony JD. 1995. The role of acid-

volatile sulfide and interstitial water metal concentrations in determining

bioavailability of cadmium and nickel from contaminated sediments to the

marine polychaete Neanthes Arenaceodentata. Environ Toxicol Chem 14:129–

141.

Rainbow PS. 2002. Trace-metal concentrations in aquatic invertebrates: Why and

so what? Environ Pollut 120:497–507.

Riba I, Garcia-Luque E, Blasco J, Delvalls TA. 2003. Bioavailability of heavy metals

bound to estuarine sediments as a function of pH and salinity values. Chem

Spec Bioavail 15:101–114.

Riedel GF, Sanders JG, Osman RW. 1997. Biogeochemical control on the flux of

trace elements from estuarine sediments—water column oxygen concen-

trations and benthic infauna. Estuarine Coastal Shelf Sci 44:23–28.

Rivera-Duarte I, Flegal AR. 1997. Porewater gradients and diffusive benthic fluxes

of Co, Ni, Cu, Zn, and Cd in San Francisco Bay. Croat Chem Acta 70:389–417.

Schlekat CE, Luoma SN. 2000. You are what you eat: Incorporating dietary metals

uptake into environmental quality guidelines for aquatic ecosystems. SETAC

Globe March-April:38–39.

Selck H, Forbes VE, Forbes TL. 1998. Toxicity and toxicokinetics of cadmium in

Capitella sp. I: Relative importance of water and sediment as routes of

cadmium uptake. Mar Ecol Prog Ser 164:167–178.

30 Integr Environ Assess Manag 3, 2007—SL Simpson and GE Batley

Shine JP, Trapp CJ, Coull BA. 2003. Use of receiver operating characteristic (ROC)

curves to evaluate sediment-quality guidelines for metals. Environ Toxicol

Chem 22:1642–1648.

Sibley PK, Ankley GT, Cotter AM, Leonard EN. 1996. Predicting chronic toxicity of

sediments spiked with zinc: An evaluation of the acid-volatile sulfide model

using a life-cycle test with the midge Chironomus tentans. Environ Toxicol

Chem 15:2102–2112.

Simpson SL. 2005. An exposure-effect model for calculating copper effect

concentrations in sediments with varying copper-binding properties: A

synthesis. Environ Sci Technol 39:7089–7096.

Simpson SL, Batley GE. 2003. Disturbances to metal partitioning during toxicity

testing of iron(II)-rich estuarine porewaters and whole sediments. Environ

Toxicol Chem 22:424–432.

Simpson SL, King CK. 2005. Exposure-pathway models explain causality in whole-

sediment toxicity tests. Environ Sci Technol 39:837–843.

Simpson SL, Apte SC, Batley GE. 1998. Effect of short term resuspension events on

trace metal speciation in polluted anoxic sediments. Environ Sci Technol

32:620–625.

Simpson SL, Apte SC, Batley GE. 2000. Effect of short-term resuspension events

on the oxidation of cadmium, lead, and zinc sulfide phases in anoxic estuarine

sediments. Environ Sci Technol 34:4533–4537.

Simpson SL, Rosner J, Ellis J. 2000. Competitive displacement reactions of

cadmium, copper, and zinc added to a polluted, sulfidic estuarine sediment.

Environ Toxicol Chem 19:1992–1999.

Simpson SL, Rochford L, Birch GF. 2002. Geochemical influences on metal

partitioning in contaminated estuarine sediments. Mar Freshwater Res 53:9–

17.

Simpson SL, Pryor ID, Mewburn BR, Batley GE, Jolley D. 2002. Considerations for

capping metal-contaminated sediments in dynamic estuarine environments.

Environ Sci Technol 36:3772–3778.

Simpson SL, Angel BM, Jolley DF. 2004.Metal equilibration in laboratory-

contaminated (spiked) sediments used for the development of whole-

sediment toxicity tests. Chemosphere 54:597–609.

Smith EP, Robinson T, Field LJ, Norton SB. 2003. Predicting sediment toxicity using

logistic regression: A concentration–addition approach. Environ Toxicol Chem

22:565–575.

Spadaro DA, Simpson SL, Chariton AA. 2006. Relationships between metal

contamination and physical properties of sediments of estuarine waters near

Sydney, NSW, Australia. CSIRO Energy Technology Report ET/IR 851R, Bangor,

Australia. 45 p.

Stumm W, Morgan JJ. 1996. Aquatic chemistry: Chemical equilibria and rates in

natural waters. New York (NY): John Wiley. 1022 p.

Teasdale PR, Apte SC, Ford PW, Batley GE, Koehnken L. 2003. Geochemical cycling

and speciation of copper in waters and sediments of Macquarie Harbor,

Western Tasmania. Estuar Coastal Shelf Sci 57:475–487.

Tessier A. 1992. Sorption of trace elements on natural particles in oxic

environments. In: Buffle J, van Leeuwen HP, editors. Environmental particles,

Vol 1. Boca Raton (FL): Lewis. p 425–453.

Tessier A, Campbell PGC. 1987. Partitioning of trace metals in sediments:

Relationships with bioavailability. Hydrobiologia 149:43–52.

Tessier A, Campbell PGC, Auclair JC, Bisson M. 1984. Relationships between the

partitioning of trace metals in sediments and their accumulation in the tissues

of the freshwater mollusk Eppiptio complanta in a mining area. Can J Fish

Aquat Sci 41:1463–1472.

Tessier A., Couillard Y, Campbell PGC, Auclair J-C. 1993. Modeling Cd partitioning

in oxic lake sediments and Cd concentrations in the freshwater bivalve

Anodonta grandis. Limnol Oceanogr 38:1–17.

Tessier A, Fortin D, Belzile N, Devitre RR, Leppard GG. 1996. Metal sorption to

diagenetic iron and manganese oxyhydroxides and associated organic matter:

Narrowing the gap between field and laboratory measurements. Geochim

Cosmochim Acta 60:387–404.

Thomann RV. 1981. Equilibrium model of fate of microcontaminants in diverse

aquatic food chains. Can J Fish Aquat Sci 38:280–296.

Tipping E, Hurley MA. 1992. A unifing model of cation binding by humic

substances. Geochim Cosmochim Acta 56:3627–3641.

Tran D, Boudou A, Massabuau J-C. 2002. Relationship between feeding-induced

ventilatory activity and bioaccumulation of dissolved and algal-bound

cadmium in the Asiatic clam Corbicula fluminea. Environ Toxicol Chem

21:327–333.

Trivedi P, Axe L. 2000. Modeling Cd and Zn sorption to hydrous metal oxides.

Environ Sci Technol 34:2215–2223.

[USEPA] US Environmental Protection Agency. 2000. Science Advisory board

review of an integrated approach to metals assessment in surface waters and

sediments. EPA-SAB-EPEC-00-005, Washington DC: Office of Water. 45 p.

[USEPA] US Environmental Protection Agency. 2005. Procedures for the derivation

of equilibrium partitioning sediment benchmarks (ESBs) for the protection of

benthic organisms: Metal mixtures (cadmium, copper, lead, nickel, silver, and

zinc). EPA-600-R-02-011. Washington DC: Office of Research and Develop-

ment. 104 p.

van Griethuysen C, Meijboom EW, Koelmans AA. 2003. Spatial variation of metals

and acid-volatile sulfide in floodplain lake sediment. Environ Toxicol Chem

22:457–465.

van Straalen NM, Donker MH, Vijver MG, van Gestel CAM. 2005. Bioavailability of

contaminants estimated from uptake rates into soil invertebrates. Environ

Pollut 136:409–417.

Vijver MC, van Gestel CAM, Lanno RP, van Straalen NM, Peijnenburg WJGM. 2004.

Internal metal sequestration and its ecological relevance: A review. Environ Sci

Technol 38:4705–4711.

Vink JPM. 2002. Measurement of heavy-metal speciation over redox gradients in

natural water-sediment interfaces and implications for uptake by benthic

organisms. Environ Sci Technol 36:5130–5138.

Warren LA, Tessier A, Hare L. 1998. Modeling cadmium accumulation by benthic

invertebrates in situ: The relative contributions of sediment and overlying

water reservoirs to organism cadmium concentrations. Limnol Oceanogr

43:1442–1454.

Wang WX, Fisher NS. 1999a. Delineating metal accumulation pathways for

marine invertebrates. Sci Tot Environ 238:459–472.

Wang WX, Fisher NS. 1999b. Assimilation efficiencies of chemical contaminants in

aquatic invertebrates: A synthesis. Environ Toxicol Chem 18:2034–2045.

Wang WX, Guo LD. 2000. Influences of natural colloids on metal bioavailability to

2 marine bivalves. Environ Sci Technol 34:4571–4576.

Wang WX, Fisher NS, Luoma SN. 1995. Assimilation of trace elements ingested by

the mussel Mytilus edulis: effects of algal food abundance. Mar Ecol Prog Ser

129:165–176.

Wang WX, Stupakoff I, Fisher NS. 1999. Bioavailability of dissolved and sediment-

bound metals to a marine deposit feeding polychaete. Mar Ecol Prog Ser

178:281–293.

Wenning RJ, Batley GE, Ingersoll CG, Moore DW. 2005. Use of sediment-quality

guidelines and related tools for the assessment of contaminated sediments.

Pensacola (FL): Society of Environmental Toxicology and Chemistry. 783 p.

Wenzhofer F, Glud RN. 2004. Small-scale spatial and temporal variability in

benthic O2 dynamics of coastal sediments: Impact of fauna activity. Limnol

Oceanogr 49:1471–1481.

Word JQ, Albrecht BB, Anghera ML, Baudo R, May SM, Di Toro DM, Hyland JL,

Ingersoll CG, Landrum PF, Long ER, Meador JP, Moore DW, O’Connor TP, Shine J.

2005. Predictive ability of sediment quality guidelines. In: Wenning RJ, Batley

GE, Ingersoll CG, Moore DW, editors. Use of sediment-quality guidelines and

related tools for the assessment of contaminated sediments. Pensacola (FL):

Society of Environmental Toxicology and Chemistry. p 121–161.

Williamson RB, Wilcock RJ, Wise BE, Pickmere SE. 1999. The effect of burrowing

by the crab Helice crassa on the chemistry of intertidal muddy sediments.

Environ Sci Technol 18:2078–2086.

Yan QL, Wang WX. 2002. Metal exposure and bioavailability to a marine deposit-

feeding sipuncula, Sipunculus nudus. Environ Sci Technol 36:40–47.

Zhang H, Davison W, Mortimer RJG, Krom MD, Hayes PJ, Davies IM. 2002.

Localized remobilization of metals in a marine sediment. Sci Total Environ

296:175–187.

Predicting Metal Toxicity in Sediments—Integr Environ Assess Manag 3, 2007 31