During the past decade, ecological risk assessments (ERAs) have increased in frequency, scope, and importance in determining cleanup activities at Superfund, Resource Conservation and Recovery Act (RCRA), and other contaminated sites. The increase in ERA activities has been spurred both by the availability of a formalized process and guidance for performing ERAs and by the recognition that contaminant levels derived to be protective of human health may not be similarly protective of ecological receptors. In the United States, the U.S. Environmental Protection Agency (USEPA) presented its framework for ERA in 1992, followed by guidance for Superfund sites (USEPA 1997) and agency-wide guidelines (USEPA 1998). Supplemental guidance has been presented in periodic ECO Updates, including the 2001 Update (USEPA 2001b) that described the role of screening-level ERAs in refining contaminants of concern.
Concurrent with the increase in the number of ERAs has been a proliferation of published compilations of ecological screening values (SVs) and the methodologies used in their derivation. SVs are used in screening-level ERAs to identify contaminants of potential concern and to determine whether further site investigation and assessment is needed. SVs are generally based on minimal effect concentrations, such as no observed adverse effect levels (NOAELs), and are not intended to serve as site-specific clean-up levels. In most cases, however, if concentrations of site-related contaminants do not exceed media-based SVs, then an absence of exceedances can serve as a defensible basis for no further action. Further, application of SVs in the early phase of a baseline or detailed ERA can substantially reduce the number of contaminants of potential concern and limit the scope of sampling and analyses needed to adequately evaluate impacts and characterize ecological risks (USEPA 2001b).
Ecological SV compilations are available from a variety of sources, including U.S. state and federal agencies, national laboratories, and governments of other countries. Although the individual SVs from these sources are generally readily available, the technical basis for SV derivation varies substantially, and in many cases, the technical derivations are not readily discernable. There have been only limited previous reports evaluating SV methodologies [see e.g., Attachment 1–1 of (USEPA 2003)], with the majority focused on general approaches used to develop SVs for sediment (Jones et al. 1997; Chapman and Mann 1999; Wenning and Ingersoll 2002).
In this article, a synopsis is presented of the technical basis and content of 13 compilations of SVs in surfacewater, sediment, soil, and tissue that are commonly used in ERAs in the United States and internationally. Media SVs based on food chain risks to wildlife are reviewed, but compilations of dietary toxicity reference values (TRVs; mg/kg diet or mg/ kg.d) for wildlife were not evaluated, unless SVs were reported for environmental media (e.g., mg/kg soil, rather than mg/kg diet). Media-based ecological SV compilations and technical support documents were selected that (1) represented a spectrum of the compilations available from state, regional, national, and international agencies; (2) included the range of available methodologies used to develop media-based SVs for surfacewater, sediment, and soil; (3) were widely used or were relied upon by other reviewed compilations; and (4) had sufficient information to discern their technical basis. The first 5 SV compilations discussed below were derived independently, the remaining 8 SV compilations are presented in general rank order of the independence of the SVs (i.e., higher independence to lower independence).
SOURCES OF ECOLOGICAL SCREENING VALUES
National ambient water quality criteria
The USEPA (2002a) has developed a compilation of acute and chronic National Ambient Water Quality Criteria (AWQC) for freshwater and saltwater (Table 1). The chronic AWQC value is the criterion continuous concentration, which is the lowest among the final chronic value (toxicity to aquatic animals), final plant value (toxicity to aquatic plants), or final residue value (toxicity to wildlife from chemical bioaccumulation in aquatic organisms) (USEPA 1985). Final chronic values were derived to be protective of the majority of aquatic species (e.g., 95%) using a statistical estimation of the cumulative probability of 0.05 in acute toxicity values for aquatic organisms determined from species-sensitivity distributions, divided by an acute-to-chronic ratio of toxicity. A secondary procedure was to develop chronic values from the geometric mean of NOAEL and lowest observed adverse effect level (LOAEL) values from chronic toxicity tests (USEPA 1985).
AWQC were independently derived and have been periodically updated extensively, with the last reported update in November 2002. A principal strength of the AWQC is the extensive database used in the development of the SVs, which spans multiple aquatic families and genera. A significant limitation is that, for the majority of listed chemicals, the AWQC do not consider bioaccumulation and toxicity to wildlife. Sediment and soil SVs are not components of the AWQC, and AWQC development considered only North American species.
The Netherlands maximum permissible concentrations
The Netherlands (Crommentuijn et al. 2000a, 200b) has established maximum permissible concentrations (MPCs) for over 160 compounds in surfacewater, sediment, and soil that evolved from previous Dutch standards, including target and intervention values (MHSPE 1994; Friday 1998) (Table 2). The MPCs were based on toxicity to sediment, surfacewater, or soil-dwelling organisms and were independently derived and extensive updates were reported in 2000 (Crommentuijn et al. 2000a, 2000b). Additionally, MPCs based on bioaccumulation and wildlife toxicity were determined for 21 organochlorine compounds that were considered to have the potential for “secondary poisoning” of wildlife. MPCs protective of wildlife were calculated from the ratio of a no effect dietary value (mg/kg food) and a bioaccumulation factor, multiplied by a correction factor (0.2 to 0.32) for the caloric content of food (Trass 2001).
If sufficient aquatic toxicity data were available, then a surfacewater MPC was derived using statistical extrapolation of NOAELs to be protective of 95% of aquatic species, similar to the AWQC procedure. The secondary procedure was to derive the MPC from the LOAEL in aquatic toxicity studies and an uncertainty factor determined from the study duration and endpoint. Surfacewater MPCs were applicable to dissolved concentrations of chemicals. As with the AWQC, a principal strength of the MPCs was the reliance on toxicity databases encompassing a broad range of families and genera. There was no adjustment of the MPC for water hardness or other water chemistry parameters, which differed from the AWQC (USEPA 2002a) procedure and reduces potential site specificity.
MPCs for the protection of benthic invertebrates were derived using equilibrium partitioning (EqP) by extrapolating a sediment concentration from the surfacewater MPC and an assumed 10% sediment organic carbon content. The sediment was also assumed to be composed of 25% clay. MPCs for metals were based on empirically derived sediment-water partition coefficients rather than organic carbon that may not have broad applicability. Advantages of EqP-based SVs include accounting for chemical bioavailability determined by sediment composition and a mechanistic basis for causally linking concentration and effects (Wenning and Ingersoll 2002). Limitations of the EqP-based SVs include the implicit assumption that porewater was the predominant route of exposure to the receptor and that the method was limited to nonionic organics.
Soil MPCs were derived from 1 of 3 approaches: (1) the preferred method of statistical extrapolation using toxicity data for plants, soil invertebrates, and microbes; (2) an LOAEL and an uncertainty factor determined from the study duration and endpoint; or (3) EqP. The EqP-based derivation for soil SVs used the surfacewater MPC rather than a more relevant toxicity value for soil organisms. MPCs were based on either a 10% soil organic carbon for organic compounds or empirically derived soil-water partition coefficients for metals and metalloids. MPCs for metals and metalloids accounted for background concentrations by considering the added risk from concentrations that exceed natural soil concentrations in The Netherlands (Crommentuijn et al. 2000b). Thus, applicability of the soil MPCs to other geographic locations requires evaluation in the risk assessment.
Table Table 1.. USEPA (2002a) ambient water quality criteria (AWQC)a
|Surfacewater||Freshwater: 38||Aquatic organism||Contact||a|
| ||Saltwater: 33||Wildlife||Boaccumulation||a|
Table Table 2.. The Netherlands (Crommentuijn 2000a, 2000b) maximum permissible concentrations (MPCs)a
|Surfacewater||166||Aquatic organism||Contact||s, l|
| || ||Wildlife||Bioaccumulation||w|
| || ||Wildlife||Bioaccumulation||w|
|Soil||162||Plant, invertebrate||Contact||s, l, e|
| || ||Wildlife||Bioaccumulation||w|
USEPA ecological soil screening levels
The USEPA Office of Emergency and Remedial Response (USEPA) 2003 has derived ecological soil screening levels (SSLs) for 9 chemicals (Table 3). Separate values were presented for plants, soil invertebrates, birds, and mammals, if adequate quantity and quality of data were available for the evaluated chemicals. Invertebrate and plant SSLs were computed from the geometric mean toxicity values equivalent to a 10 to 20% effect level or from the maximum acceptable toxicant concentration calculated from NOAELs and LOAELs. The SSL derivation for wildlife considered only those birds and mammals with high soil contact (vole, shrew, dove, woodcock) and key terrestrial predators (weasel, hawk). Bioaccumulation models were used to determine bioaccumulation factors in highly exposed forage and prey. A TRV was derived from the geometric mean of the NOAELs or the highest NOAEL below the lowest of the LOAELs for mortality, growth, or reproduction. The SSL was computed from the TRV and the bioaccumulation factor. Separate SSL values were presented for invertebrate, plant, birds, and mammals where data were considered to be adequate for deriving a SSL.
The SSLs were independently derived, and updates and development of SSLs for additional chemicals have been in progress. The advantages of the SSLs include their derivation based on a process of comprehensive literature review that includes a rigorous review of quality and applicability of toxicity studies. The limitation of the SSL method has been the extensive time, cost, and data needed to develop a relatively few SVs
Integrative sediment toxicity compilations
Several integrative sediment toxicity compilations have been developed from statistical interpretations of the incidence of biological effects on benthic invertebrates and chemical concentrations in sediment. The effects range-low (ER-L) and effects range-medium (ER-M) SVs (Long et al. 1995) developed from the biological effects database for sediments of the National Oceanic and Atmospheric Administration (NOAA) have been the most commonly used in ERAs. Biological effects have been determined from (1) laboratory toxicity tests of spiked (added chemical) sediments, (2) field surveys of benthic invertebrate abundance and diversity, (3) laboratory tests of field-collected sediments, and (4) levels estimated from EqP and predicted porewater toxicity. The effects range-low and effects range-medium SVs were calculated as the 10th (low) and 50th (medium) percentiles on an ordered list of chemical concentrations in sediment that co-occurred with a biological effect and have been interpreted as an approximate 10% and 50% probability of toxicity (O'Connor 2004).
Other sediment SV compilations include threshold and probable effect levels (MacDonald et al. 1996; Jones et al. 1997), “apparent effects thresholds” (Barrick et al. 1989), and lowest and severe effect levels (MOE 1993). The sediment SVs were independently derived and have had only limited and infrequent updates. Buchman (1999) tabulated SVs from several of the sediment toxicity compilations in screening quick reference tables (SQuiRTs) from threshold effect levels indicative of a low probability of adverse effects to apparent effects thresholds indicative of a high probability of effects (Table 4). The last reported update of the screening quick reference table compilation was in October 1999.
Table Table 3.. USEPA (2003) ecological soil screening levels (SSLs)a
| || ||Wildlife||Bioaccumulation||h|
Table Table 4.. NOAA (Buchman 1999) screening quick reference table (SQuiRT) valuesa
|Sediment||Freshwater: 52 Saltwater: 80||Benthic invertebrates||Contact||i|
|Soil||66||Aquatic habitat||Not specified||y, z|
Chapman and Mann (1999) critically reviewed several integrative sediment toxicity compilations and noted concerns including (1) variability in the degree of conservatism of the SVs, (2) no direct consideration of bioaccumulation and food chain effects, (3) variability in bioavailability not incorporated, (4) confounding effects of contaminant mixtures, and (5) uncertain ability to predict toxicity. Wenning and Ingersol (2002) also noted the need to discriminate the causal effects of contaminants in sediment effects databases from noncontaminant stressors such as grain size and habitat modification. Despite concerns, Chapman and Mann (1999) recommended that sediment compilations were appropriate as SVs but should only be used in more comprehensive phases of ERA as part of weight-of-evidence assessments.
Canadian environmental quality guidelines
The Canadin Council of Ministers of the Environment (CCME) (1999) developed a compliation of environmental quality guidelines (EQGs) in surfacewater, sediment, and soil (Table 5). CCME also developed tissue residue guidelines for aquatic organism tissue concentrations of selected bioaccumulative chemicals, which are protective of piscivorous wildlife. Tissue-based EQCs were derived only for selected chemicals that were both bioaccumulative (e.g., accumulation factor ≥5,000) and persistent, and were based on toxicity data for sensitive life stages and species of wildlife that may feed on contaminated aquatic organisms. The EQG was calculated as a tissue concentration in aquatic prey (mg/kg wet weight) from the geometric mean of NOAELs and LOAELs and species-specific ingestion and body weight parameters. Tissue-based criteria were an attribute of the EQGs that have been generally unavailable in SV compilations. A general strength of the EQGs was that they were independently derived and have been periodically updated, with the last reported update in July 2002.
Separate freshwater and marine surfacewater quality guidelines were derived from the LOAEL determined in a review of toxicity studies with aquatic organisms. An EQG was calculated by dividing the LOAEL for the most sensitive species and life stage by an uncertainty factor determined by the study duration and toxicity endpoint (e.g., a smaller uncertainty factor for a chronic life cycle study than for an acute test). Limitations included a strong reliance on uncertainty factors, more limited applicability to a broad range of species than AWQC or MPCs, and wildlife toxicity not being incorporated into the SV development.
Marine and freshwater sediment quality guidelines were based on toxicity to benthic invertebrates. The EQGs were developed as both interim sediment quality guidelines (threshold for effects) and probable effects levels. Screening values were based on statistical interpretations of databases on the incidence of biological effects and chemical concentrations in sediment (Long et al. 1995; MacDonald et al. 1996) and, thus, have the limitations noted above. Wildlife toxicity was not considered in the derivation of the sediment EQGs, which was a common limitation among sediment SV compilations.
Table Table 5.. Canadian environmental quality guidelines (EQGs) (CCME 1999; updated 2002)a
|Surfacewater||Freshwater: 96 Saltwater: 21||Aquatic organism||Contact||l|
|Sediment||Freshwater: 32 Saltwater: 32||Benthic invertebrate||Contact||i|
| || ||Wildlife||Bioaccumulation, soil ingestion||k|
The soil quality guidelines were determined by the intended land use, with agricultural uses generally having the lowest EQGs. The EQG for agricultural land was determined as the lowest soil value protective of adverse effects from (1) direct contact to plants, invertebrates, and microbes; (2) ingestion of soil and plants by grazing wildlife and livestock; or (3) human exposures. The direct contact value was derived from available soil toxicity data, and the ingestion value was derived from toxicity to animals and bioaccumulation in plants. Consideration of wildlife exposures was an attribute of the soil EQGs but could lead to a variable level of conservatism if receptors and pathways were present or absent at a site. It is worth noting that the protection of microbes, which are not typically identified as a receptor in ERA, can be considered a limitation of some of the SVs.
Oak Ridge National Laboratory preliminary remediation goals
Oak Ridge National Laboratory (ORNL) has developed a compilation of preliminary remediation goals (PRGs) for chemicals in surfacewater, sediment, and soil (Table 6). The ORNL PRGs (Efroymson, Suter, et al. 1997) were derived from the lowest value for either the direct toxicity to aquatic or soil organisms or from dietary toxicity to wildlife of bioaccumulative chemicals. The PRGs were generally based on LOAELs, rather than NOAELs and presented as toxicological benchmarks developed in other ORNL compilations (Sample et al. 1996; Suter and Tsao 1996; Efroymson, Will, Suter 1997; Efroymson, Will, Suter, Wooten 1997). General strengths of the ORNL PRGs and toxicological benchmarks include the rigor of their derivation, a comprehensive review of toxicity databases, consideration of multiple receptors and pathways, and transparency in their technical basis. A general limitation is their basis on older (1996 and 1997) compilations of environmental and toxicity data that have not been updated.
The ORNL PRGs and toxicological benchmarks have been widely used in ERAs. The primary surfacewater PRGs were the AWQC, and secondary chronic or Tier II values were derived if an AWQC value was not available (Suter and Tsao 1996). The secondary water quality values have less-stringent data set requirements for development than AWQC (i.e., fewer families of aquatic organisms). A few screening values were based on the ORNL piscivorous wildlife value (Sample et al. 1996) determined for either mink or a piscivorous bird. The piscivorous wildlife value was derived from an accumulation factor and species-specific LOAELTRV and from drinking water exposure (Sample et al. 1996). Sediment PRGs were determined from the ORNL compilation by Jones et al. (1997) on the toxicity to benthic invertebrates. The PRG was determined from the lowest value from either EqP or reported threshold values determined from statistical interpretations of databases on the incidence of toxicity and biological effects and chemical concentrations in sediment (Long et al. 1995; MacDonald et al. 1996). Wildlife toxicity was not considered in the sediment PRGs, which was a common limitation for sediment SV compilations. Soil PRGs were the lowest of the SVs for plants, soil invertebrates, or wildlife. The invertebrate value was determined from the toxicity to earthworms (Efroymson, Will, Suter 1997), and the plant value was based on toxicity to terrestrial plants (Efroymson, Will, Suter, Wooten 1997). Plant and invertebrate PRGs were generally derived from the 10th-percentile LOAEL determined by rank ordering of the chemical and receptor-specific LOAELs. Wildlife values were derived from the pathway-specific accumulation factors (e.g., soil to plant, soil to small mammal) and species-specific LOAELTRV for 6 species including predators (Sample et al. 1996).
In general, Efroymson, Will, Suter (1997) and Efroymson, Will, Suter, Wooten 1997 had low confidence in the soil toxicity values used in deriving the ORNL PRGs because of the limited number of studies and endpoints that formed the basis of the SVs and reliance on bioaccumulation-toxicity models for wildlife, which did not account for soil properties (Efroymson, Suter, et al. 1997).
USEPA Region 5 ecological screening levels
USEPA Region 5 (USEPA 1999) has developed a comprehensive compilation of ecological screening levels (ESLs) for surfacewater, sediment, and soil, for those analytes listed in the Resource Conservation and Recovery Act Appendix IX (Table 7). The ESLs only address freshwater ecosystems and were derived specifically for USEPA Region 5, which includes states bordering the Great Lakes. Surfacewater ESLs were determined from the lowest value determined from state water quality standards within USEPA Region 5, AWQC or other aquatic toxicity values, or food chain exposures to wildlife. The wildlife ESLs were based on food chain modeling of mink and kingfisher and NOAEL TRVs. Sediment ESLs were primarily based on EqP and secondarily on other sediment toxicity thresholds (MacDonald et al. 2000). The majority of soil ESLs were derived using NOAEL TRVs for wildlife and a food chain model for the masked shrew and, secondarily, the meadow vole. A few soil ESLs were based on toxicity to plants or earthworms.
Table Table 6.. Oak Ridge National Laboratory (ORNL) (Efroymson, Suter, et al. 1997) preliminary remediation goals (PRGs)a
|Surfacewater||107||Aquatic organism||Contact||a, o|
| || ||Wildlife||Bioaccumulation drinking water||o|
|Sediment||95||Benthic invertebrate||Contact||e, i|
| || ||Wildlife||Bioaccumulation||o|
Table Table 7.. USEPA Region 5 (USEPA 1999; updated 2003) ecological screening levels (ESLs)a
|Surfacewater||151||Aquatic organism||Contact||r, a, q|
| || ||Wildlife||Bioaccumulation||m, r|
|Sediment||147||Benthic invertebrate||Contact||e, b|
| || ||Wildlife||Bioaccumulation||e|
| || ||Wildlife||Bioaccumulation||v|
In general, USEPA Region 5 ESLs represent a combination of both independently derived and dependent values, with an extensive update reported in August 2003. The SVs were derived using a diversity of methods, including reliance on state water quality values that may not be toxicity based, resulting in varying conservatism and, in some cases, an uncertain technical basis. Other limitations included extensive use of professional judgment in applying uncertainty factors where data were limited (e.g., extrapolation from acute to chronic toxicity) (USEPA 1999). General strengths include SVs for a broad range of chemicals, many that have not been available in other compilations. The lowest SV determined from either toxicity to aquatic or terrestrial organisms or to wildlife was used as the media-specific ESL, which results in a high degree of conservatism (Table 7).
Australian guidelines and investigation levels
The Australian government has established surfacewater and sediment quality guidelines (ANZECC and ARMA-CANZ 2000) and soil investigation levels (NEPC 2003) as SVs (Table 8). The SVs were a combination of both independently derived and dependent values and have been periodically updated. Surfacewater values were presented as trigger values for different levels of protection (80 to 99% of aquatic species). The screening values were derived independently using the Netherlands approach of statistical analysis of aquatic toxicity data. The surfacewater SVs were developed using data for a broader range of species data than AWQC; thus, ERAs should evaluate the regional applicability of the SVs. The interim sediment quality guidelines were the same values as those identified by NOAA (Long et al. 1995) as effects range-low and -medium SVs. Interim ecological investigation levels for urban soil were determined from unspecified phytotoxicity and soil properties (NEPC 2003) and had the limitations of relatively few values and an uncertain technical basis. The Australian SVs also had the general limitation common to many SV compilations, that is, that wildlife toxicity was not incorporated into the derivation.
USEPA ecotox thresholds
The USEPA Office of Solid Waste and Emergency Response (USEPA 1996) has developed an initial list of “ecotox thresholds” (ETs) in surfacewater and sediment (Table 9) for application in risk assessments at U.S. “Superfund” hazardous waste sites. Separate surfacewater SVs were listed for freshwater and saltwater, based primarily on chronic AWQC values reported before 1996. The ETs also included secondary or Tier II water quality values that have less-stringent data-set requirements than AWQC (i.e., fewer families of aquatic organisms). Sediment ETs were primarily derived using EqP, with the exception of a few chemicals that rely on NOAA effects range-low values (Long et al. 1995) to represent the ET. The ETs have not been updated since they were published in January 1996 and, thus, are likely limited by their dependence on possibly outdated toxicity values and AWQC. Wildlife risks were not considered in the ETs, which was a common limitation of the majority of SV compilations.
Table Table 8.. Australian surfacewater and sediment quality guidelines (ANZECC and ARMACANZ 2000) and soil investigation levels (NEPC 2003)
|Surfacewater||Freshwater: 81 Saltwater: 27||Aquatic organism||Contact||s|
Table Table 9.. USEPA ECO Update (USEPA 1996) ecotox thresholds (ETs)a
|Surfacewater||Freshwater: 65 Saltwater: 17||Aquatic organism||Contact||a, t|
|Sediment||52||Benthic invertebrate||Contact||e, i|
State of Oregon screening-level values
The Oregon Department of Environmental Quality (ODEQ 2001) developed a comprehensive compilation of screening level values (SLVs) in surfacewater, sediment, and soil (Table 10). Separate freshwater SLVs were compiled for 3 categories of receptors: aquatic organisms, birds, and mammals, which is a substantial strength because receptor-pathway–specific risks can be screened. The SLVs represent a combination of both independently derived and dependent values that were last updated in November 2001 (Table 10). The primary SLVs for aquatic organisms were based on chronic AWQC (April 1999 version). The ORNL values (Suter and Tsao 1996) were used if AWQC were not available. Surfacewater SLVs for birds were based on the NOAELs for robins computed for drinking water–only exposures from ORNL values (Sample et al. 1996). Surface-water SLVs for mammals were based on NOAELs for mammals (although the species was not specified) and drinking water–only exposures from ORNL values (Sample et al. 1996). General limitations of the surfacewater SLVs include reliance on outdated AWQC and uncertainty in the level of protection for aquatic-dependent wildlife.
Separate SLVs for sediment were compiled based on freshwater, marine, and bioaccumulation considerations. The marine and freshwater sediment SLVs were based on toxicity to benthic invertebrates and determined from SVs listed in the NOAA screening quick reference tables (Buchman 1999); in a few cases, benthic invertebrate toxicity values from other sources were also used. Bioaccumulation SLVs for sediment were derived from combined food and drinking water exposure to piscivorous wildlife (Sample et al. 1996) and predicted porewater concentrations (primarily from EqP). The bioaccumulation SLV was the lowest of the values determined for mink or great blue heron. The limitations of the ODEQ sediment SLVs are similar to those discussed earlier for integrative sediment SVs (Chapman and Mann 1999).
Separate soil SLVs were compiled for plants, invertebrates, birds, and mammals. The invertebrate and plant SLVs were based on ORNL (Efroymson, Will, Suter 1997; Efroymson, Will, Suter, Wooten 1997) summaries of the toxicity to earthworms and plants, as discussed earlier. ORNL values for soil microbes were used as the SLV for soil invertebrates when earthworm values were not available (Efroymson, Will, Suter 1997), but differences in sensitivity between microbes and earthworms may lead to variable levels of conservatism in the SVs. The soil SLVs for birds and mammals were based on NOAELs from Sample et al. (1996) and exposure from only incidental soil ingestion: robin (20% soil) and an unspecified mammal (10% soil). Bioaccumulation in prey was not considered in the soil SLV, which leads to uncertainty in screening for food chain risks.
New York State Department of Environmental Conservation criteria
The New York State Department of Environmental Conservation (NYSDEC 1998, 1999) has developed a comprehensive compilation of SVs for surfacewater (water quality standards or guidelines) and sediment (sediment criteria), ecological soil criteria for 3 chemicals, and tissue residue criteria (Table 11). Separate surfacewater criteria were listed for freshwater and saltwater (NYSDEC 1998), which were primarily based on chronic AWQC and Tier II water quality values. Surfacewater values protective of wildlife were determined from USEPA (2002b) or were computed from fish-flesh criteria from Newell et al. (1987) and food chain bioaccumulation factors determined by professional judgment assuming an approximately 10% lipid content in prey (USEPA 2002b). Sediment criteria for organics were primarily based on EqP using the New York State surfacewater value or AWQC and were expressed per g of organic carbon (NYSDEC 1999). Sediment SVs for metals were low-effect and high-effect levels from several integrative compilations discussed above (CCME 1999). Sediment wildlife values were determined from EqP using water quality values protective of wildlife. Soil criteria were only reported for PCB, lead, and cadmium based on bioaccumulation in prey and wildlife toxicity. The New York State fish-tissue residue criteria were based on Newell et al. (1987), who identified acceptable levels for 17 organochlorine compounds that would be protective of piscivorous wildlife. The tissue criterion was calculated as a tissue concentration in aquatic prey (mg/kg wet weight) from NOAELs, uncertainty factors, and species-specific ingestion and body weight parameters.
Table Table 10.. Oregon Department of Environmental Quality (ODEQ 2001) screening level values (SLVs)
|Surfacewater||Freshwater: 172||Aquatic organism||Contact||a, o|
| || ||Wildlife||Drinking water||o|
|Sediment||Freshwater: 52 Saltwater: 75 Bioaccumulation: 59||Benthic invertebrate||Contact||n, b|
| || ||Wildlife||Bioaccumulation drinking water||o|
| || ||Wildlife||soil ingestion||o|
Table Table 11.. New York State Department of Environmental Conservation water quality, sediment, soil, and tissue criteria (Newell et al. 1987; NYSDEC 1998, 1999)
|Surfacewater||Freshwater: 75 saltwater: 37 wildlife: 6||Aquatic organism||Contact||a, t|
| || ||Wildlife||Bioaccumulation||x|
|Sediment||Freshwater: 57 saltwater: 42 wildlife: 12||Benthic invertebrate||Contact||e, i|
| || ||Wildlife||Bioaccumulation||w|
Overall, the SVs were a combination of both independently derived and dependent values that were, in part, determined from professional judgment, including consideration of the variability in bioaccumulation. A general limitation in the SVs was a less-than-transparent use of uncertainty factors and professional judgment that results in uncertainty in the level of conservatism. Strengths and limitations of EqP-based sediment SVs were discussed above in association with the Netherlands MPCs. Development of separate SVs in media and tissue protective of wildlife was a general strength of the New York SVs. The SVs have received limited updates as the compilation is expanded to include additional chemicals or to include revisions to existing SVs.
USEPA Region 4 ecological screening values
USEPA Region 4 (USEPA 2001a) compiled a comprehensive list of ecological screening values (ESVs) in surfacewater, sediment, and soil (Table 12). Separate surfacewater ESVs were compiled for freshwater and saltwater aquatic organisms based on primarily chronic values from an older version of AWQC (April 1999). If AWQC were not available, then ESVs were derived from an unspecified LOAEL toxicity value and uncertainty factor. For a few chemicals, the ESV was based on an unknown technical basis, including unspecified wildlife toxicity or “marketability” of fish. Sediment ESVs were determined from reported toxicity thresholds derived from statistical interpretations of databases on the incidence of biological effects and chemical concentrations in sediment (Long et al. 1995; MacDonald et al. 1996), or the practical quantitation limit of the analytical method if it exceeded the toxicity value. Soil ESVs were the ecological screening values reported by Friday (1998). These ESVs were the lower of soil quality guidelines from either (1) a 1997 Canadian compilation now superseded by the CCME (1999) and subsequent updates or (2) 1997 or earlier compilations from The Netherlands (Friday 1998).
The ESVs were a combination of both independently derived and dependent values. Recent updates have including expanding the compilation to include soil SVs, with the last reported update in November 2001. However, the ESVs are largely outdated and do not reflect more recent updates in the compilations from which they derive a substantial number of SVs. Additional limitations of some of the ESVs have included an uncertain technical basis or derivation based on analytical quantitation limits rather than toxicity.
State of Texas ecological benchmarks
The Texas Natural Resources Conservation Commission (TNRCC 2001) compiled a comprehensive list of ecological benchmarks (EBs) in surfacewater, sediment, and soil (Table 13). Separate surfacewater EBs were listed for freshwater and saltwater that were derived from a variety of sources, including Texas water quality values and chronic AWQC. Secondarily, EBs were derived from acute toxicity data, Tier II water quality values (Suter and Tsao 1996), and USEPA Region 4 values (USEPA 2001). Separate EBs were listed for freshwater and saltwater sediment, and were from the NOAA (Long et al. 1995) and MacDonald et al. (1996) compilations. Separate soil EBs were compiled for plants and invertebrates using a 2000 draft of the SSLs (USEPA 2003) and ORNL (Efroymson, Will, Suter 1997; Efroymson, Will, Suter, Wooten 1997) summaries of the toxicity to earthworms and plants. The majority of surfacewater EBs, as well as the sediment and soil EBs, were dependent on other compilations, many of which are now outdated. Of additional concern, the surfacewater SVs were derived using several different methodologies, including state water quality values, resulting in varying conservatism and, in some cases, an uncertain technical basis. The protection of wildlife was not considered in any of the EBs.
Table Table 12.. USEPA Region IV (USEPA 2001a) ecological screening values (ESVs)
|Surfacewater||Freshwater: 112 Saltwater: 90||Aquatic organism||Contact||a, l, c|
|Sediment||40||Benthic invertebrate||Contact||i, p|
| || ||Wildlife||Bioaccumulation soil ingestion||f|
Variation in SV methodologies
Several methodologies have been used to develop media-based SVs for screening for the presence or absence of ecological risk, with the majority of SVs developed through 1 of 9 general approaches (Table 14). The majority of ecological SVs have been based on an assessment of toxicity to either aquatic or terrestrial organisms and, less frequently, to wildlife from either drinking water exposures or food chain bio-accumulation (Table 14). Surfacewater SVs were primarily derived for the protection of aquatic organisms using 2 general approaches: (1) a statistical assessment of toxicity values using species sensitivity distributions, or (2) extrapolation of an LOAEL determined from more limited toxicity data using an uncertainty factor. The surfacewater SVs in several compilations were also based on toxicity to wildlife from drinking water exposure or chemical accumulation in aquatic prey (Table 14). Sediment SVs were primarily derived for the protection of benthic invertebrates using 2 approaches: (1) statistical interpretations of databases on the incidence of biological effects and chemical concentrations in sediment, such as the NOAA effects range-low values; or (2) EqP-derived values based on a surfacewater SV. Soil SVs were derived using several approaches and were usually based on the lowest value determined from soil toxicity to terrestrial plants, invertebrates, wildlife, and sometimes microbes. Soil SVs based on toxicity to wildlife were determined from modeled incidental soil ingestion or chemical accumulation in terrestrial organisms.
An emerging area has been the development of tissue-based SVs. Of the 13 SV compilations reviewed, only CCME (1999) and the State of New York (Newell et al. 1987) reported tissue-based SVs that were to protect piscivorous wildlife for a limited number of bioaccumulative chemicals.
Conservatism, uncertainty, and limitations in SVs
In general, the most conservative SV methodologies were those that derived SVs based on no effect levels of toxicity to wildlife, in addition to aquatic or soil organisms, and selected the lowest value (e.g., USEPA Region 5 ESLs) (Table 15).
Table Table 13.. Texas Natural Resources Conservation Commission (2001) ecological benchmarks (EBs)
|Surfacewater||Freshwater: 189 Saltwater: 108||Aquatic organisms||Contact||r, a, q, o|
|Sediment||Freshwater: 43 Saltwater: 34||Benthic invertebrates||Contact||i|
|Soil||63||Plants, invertebrates||Contact||u, o|
Table Table 14.. Comparison of screening value (SV) methodologies and compilations with independently derived SVs and SVs dependent upon other compilationsa
|Species sensitivity distributions||Applicable to broader range of species||Large data requirement||Water||AWQC, MPC, Ausc,||PRG, ESL, ET, SLV, NYd, ESV, EB|
| || || ||Soil||MPC, PRG||ESV|
|Chronic toxicity||Limited data requirements||Broad species applicability uncertain||Water||AWQC, MPC, EQG, PRG, Ausc, ET, ESV, EB, ESL||SLV|
| || || ||Sediment||ESL, SLV|| |
| || || ||Soil||MPC, SSL, EQG, PRG, ESL, Ausc||SLV, ESV, EB|
|Wildlife toxicity/ bioaccumulation factor or diet exposure model||Limited data requirements||Broad species applicability uncertain||Water||AWQC, MPC, PRG, ESL, NYd||SLV, NYd|
| || || ||Sediment||MPC, NYd||SLV|
| || || ||Soil||MPC, SSL, EQG, PRG, ESL, NYd||SLV, ESV|
|Equilibrium partitioning (EqP)||Causal linkage and mechanistic basis||Only considers porewater exposure||Sediment||MPC, PRG, ESLe, ET, NYd|| |
| || || ||Soil||MPC|| |
|Statistical association||Effects of field samples||Causation uncertain, complex mixtures||Sediment||NOAAf, EQG||PRG, Ausc, ET, SLV, NYd, EB|
|Quantitation limit||Considers analytical limitations||Not toxicity-based; changes with analytical method||Sediment||ESV|| |
|State water quality standards||Greater regional specificity||Technical basis less certain||Water|| ||ESL, EB|
|Wildlife toxicity||Limited data requirements||Broad species applicability uncertain||Tissue||EQG, NYd|| |
|Background levels||Considers natural background||Limited to metals, spatial variability||Water||MPC|| |
| || || ||Sediment||MPC|| |
| || || ||Soil||MPC|| |
Table Table 15.. Comparison of screening values (SVs) for 4 chemicals from 13 compilationsab
|AWQC (USEPA 2002a)||NAd||NA||NA||NA||0.25||9||0.056||0.014||NA||NA||NA||NA|
|MPC (Crommentuijn et al. 2002a, 2002b)||30||73||067||Ce||0.42||1.5||0.018||C||1.6||40||0.05||C|
|SSL (USEPA 2003)||NA||NA||NA||NA||NA||NA||NA||NA||0.38–140||NA||0.00028–0.0016||NA|
|NOAA (Long et al. 1995)||1.2||34||NA||0.023||NA||NA||NA||NA||NA||NA||NA||NA|
|EQG (CCME 1999)||0.60||35.7||0.003||0.012||0.017||2–4||NA||NA||1.4||63||NA||0.7|
|ORNL (Efroymson, Suter, et al. 1997)||4.2||78||0.0043||0.18||1.1||12||NA||0.019||4||60||NA||0.37|
|ESL (USEPA 1999)||0.99||31.6||0.0019||0.060||0.15||1.6||0.00007||0.00012||0.022||5.4||0.0024||0.00033|
|Ausf (ANZECC 2000; NEPC 2003)||1.5||65||0.02||0.023||0.2||1.4||NA||NA||3||100||NA||NA|
|ET (USEPA 1996)||1.2||34||0.052||0.023||1.0||11||0.062||0.19||NA||NA||NA||NA|
|SLV (ODEQ 2001)||0.003–0.6||10–36||0.003–0.004||0.034||2.2–10,000||9–340,000||0.056–600||0.014–270||4–125||50–390||0.3–3||4–40|
|NYg (NYSDEC 1998, 1999)||0.60||16||0.008–0.09||0.014–0.19||2.7||3.4–4.8||0.056||0.00012||1.0||NA||NA||43|
|ESV (USEPA 2001a)||1||18.7||3.3||33||0.66||6.5||0.0019||0.014||NA||NA||NA||NA|
|EB (TNRCC 2001)||0.60||35.7||0.003||0.007||0.60||7||0.002||0.001||4.0–20||50–100||NA||40|
The level of conservatism and uncertainty, as well as the receptors and pathways considered, varied substantially between and often within an SV compilation (Table 14). The ecological SVs were generally conservative because they were based on minimal or no effect concentrations, such as NOAELs, and implicit assumptions of high receptor exposures. Also, the majority of SV compilations did not consider natural background levels of metals, although The Netherlands MPCs were a notable exception. In some cases, the level of conservatism of the SV was unknown because wildlife exposures were not considered in the SV derivation (e.g., most sediment SVs) or only a limited number of relevant exposure pathways were included in the derivation (e.g., media ingestion but not diet). Also, key receptors were often excluded because of an absence of toxicity data (e.g., soil SVs based on toxicity to invertebrates but not to plants).
Variation in the approaches used to develop SVs, as well as variable conservatism and uncertainty, has resulted in a range in media-based SVs that span several orders of magnitude for individual chemicals (Table 15). The largest variation in SVs between different compilations is attributed generally to inclusion or exclusion of wildlife toxicity in the SV derivation. For example, inclusion of wildlife toxicity for persistent organic pollutants generally resulted in lower SVs than those derived only for the protection of aquatic or terrestrial organisms. All of the SV compilations relied on several approaches to derive SVs for a specific medium, resulting in a variable level of uncertainty and conservatism (Table 14). A general limitation of many SVs was a less-than-transparent use of uncertainty factors and professional judgment and insufficient available information to readily verify the accuracy and validity of each SV. The AWQC, SSLs, and ORNL PRGs were the exception because they included a detailed presentation of the toxicity database used to develop the SVs and a transparent method for their development. The SVs in several compilations were derived from outdated values or older toxicity data (e.g., USEPA Region 4 ESVs [USEPA 2001a] Texas EBs [TNRCC 2001,] and USEPA ETs [USEPA 1996]).
RECOMMENDATIONS FOR THE FUTURE
Careful selection of SVs used in ERAs to screen ecological risks is recommended because the available SV compilations and methodologies are, in general, associated with varying degrees of conservatism and are not consistent with regard to exposure pathways and receptors considered in the SV derivation. In selecting an SV, ecological risk assessments should consider the technical basis, strengths and limitations, conservatism, and uncertainty discussed above, as well as the applicability and relevancy to a particular contaminated site or risk assessment application.
For example, SVs based on wildlife toxicity should be selected only if food chain pathways may be complete for that medium. Other selection criteria should include the quality and relevance of the environmental and toxicity data used to derive the SV. Ecological risk assessments should focus SV selection on those compilations based on large numbers of independently derived SVs, rather than relying on SVs representing secondary compilations, often representing varied sources and quality of supporting information.
We recommend that future efforts focus on developing media-based SVs protective of wildlife that are separate from those intended to be protective of aquatic organisms or terrestrial organisms. This would allow general receptor-based screening in addition to the media-based screening that has been currently available in only a few SV compilations, such as the New York environmental conservation criteria (NYSDEC 1998, 1999). Ecological SVs should remain conservative so that the absence of an exceedance can be interpreted as the absence of significant risk. An exceedance of an SV should trigger a more refined risk assessment.
Finally, the technical basis and data sets used to derive SVs in the majority of compilations should be made more transparent and accessible to ecological risk assessors. It is clear that, at the present time, uncertainty in SVs and applicability to different classes of receptors represent large and unavoidable limitations in the practice of ERA in the United States and possibly elsewhere.