This article addresses the regulatory issues associated with the application of recent data to support Registration, Evaluation, Authorisation and Restriction of Chemical substances (REACH) requirements in Europe and the use of metal-specific parameters by other countries to generate remediation values for metals in soil. The purposes of this article are to: 1) present approaches and advances developed over the last decade in Europe for the REACH regulation and proposed in Australia by the National Environment Protection Council, 2) review current US and Canadian regulatory practices on ecological soil cleanup values, and 3) evaluate the application of new scientific approaches, methods, and soil criteria development processes used in other countries. Integr Environ Assess Manag 2014;10:401–414. © 2013 SETAC
This article resulted from a workshop titled, “Ecological Soil Levels—Next Steps in the Development of Metal Clean-up Values,” which was held in September 2012, in Sundance, Utah. The goal of the workshop was to progress beyond using ecological soil screening levels (EcoSSLs) or similar screening values as a frequent and primary basis for making soil cleanup decisions, and toward the development of site-specific soil cleanup values. This advancement in the practice of soil ecological risk assessment and management can be achieved by providing environmental scientists and managers in the regulating and regulated community with a synthesis of the methods and processes that are available to facilitate movement from screening values to more transparent and technically defensible soil cleanup values that incorporate data on bioavailability, tools for normalizing toxicity thresholds, food-web considerations, and background. This contribution to the special series builds on the previous articles in this special series—articles that addressed the workshop background (Wentsel and Fairbrother this issue), methods to measure metal effects on invertebrates and plants (Checkai et al. this issue) and microbial communities (Kuperman et al. this issue), and wildlife metal exposure issues (Sample et al. this issue) and effects (Mayfield et al. this issue). It discusses potential regulatory issues associated with the development of ecological soil cleanup values (Eco-SCVs) for cleanup of contaminated sites within North America.
Ecological soil screening levels (EcoSSLs) were developed by the US Environmental Protection Agency (USEPA) in 2003 (http://www.epa.gov/ecotox/ecossl/). These values simplified the initial screening process at contaminated sites by identifying which of the most common soil pollutants are most likely to be causing ecological problems. Since then, EcoSSLs have been used by the USEPA Regional Offices and also by many states when conducting site assessments. Although the documentation clearly states that these values are not to be used as cleanup goals, it has become common practice to use them in this manner because of the lack of EPA-sanctioned, ecologically based cleanup concentrations. This is particularly problematic for many metals and other inorganics, because the EcoSSL values are not adjusted for site-related differences in soil types and do not include new principles of ecotoxicology that significantly influence the calculation of a safe soil concentration. In the nearly 10 years since the EcoSSLs were developed, there have been significant improvements in the test methods used to develop soil toxicity data, particularly for plants and soil invertebrates. This research was conducted in support of regulatory requirements of the European chemical registration program known as Registration, Evaluation, Authorisation and Restriction of Chemical substances (REACH 2006), and accounts for differences in bioavailability and organism response as a function of soil parameters. Additionally, new approaches to setting toxicity threshold values have gained acceptance, resulting in more accurate predictions.
The purposes of the present article are to: 1) present approaches and advances developed over the last decade in Europe for the REACH regulation and proposed in Australia by the National Environment Protection Council (NEPC), 2) review current US and Canadian regulatory practices on ecological soil cleanup values, and 3) evaluate the application of the new scientific approaches, methods, and soil criteria development processes used in other countries.
REGULATIONS AND CURRENT PRACTICES IN NORTH AMERICA
United States of America
USEPA metals framework
The USEPA Framework for Metals Risk Assessment (USEPA 2007) addresses the special attributes and behaviors of metals and metal compounds to be considered when assessing their human health and ecological risks. The purpose of the document was to present key guiding principles based on the unique attributes of metals (as differentiated from organic and organometallic compounds) and to describe how these metal-specific attributes and principles can then be applied in the context of existing EPA risk assessment guidance and practices. The document describes basic principles to be considered in assessing risks posed by metals and is intended to foster consistency in the application of these principles across the Agency's programs and regions when conducting these assessments.
The USEPA Framework put forward 5 principles that are more general, fundamental properties of metals that should be addressed and incorporated into all inorganic metals risk assessments:
- Metals are naturally occurring constituents in the environment and vary in concentrations across geographic regions.
- All environmental media have naturally occurring mixtures of metals, and metals are often introduced into the environment as mixtures.
- Some metals are essential for maintaining proper health of humans, animals, plants, and microorganisms.
- Metals, unlike organic chemicals, are neither created nor destroyed by biological or chemical processes, although these processes can transform metals from one species to another (valence states) and can convert them between inorganic and organic forms.
- The absorption, distribution, transformation, and excretion of a metal within an organism depends on the metal, the form of the metal or metal compound, and the organism's ability to regulate and/or store the metal.
Much of the REACH research efforts considered the principles above.
USEPA ecological risk assessment in the United States
To meet the requirements of a risk assessment for USEPA policy purposes under the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA, or Superfund), ecological risk assessment (ERA) is conducted at hazardous waste sites to determine whether there are acceptable or unacceptable levels of risk to ecological receptors. The Ecological Risk Assessment Guidance for Superfund (USEPA 1997) presents the process for assessing ecological risk at Superfund sites. The 8-step process described in the guidance is composed of 3 primary phases—screening-level ERA, refinement of preliminary contaminants of concern, and baseline risk characterization. If the final conclusion is that ecological risk at a site is unacceptable for 1 or more contaminant or contaminants, then the final task of the risk characterization is to develop an ecologically protective cleanup value for the site media (e.g., concentration in the soil) using literature-based and/or site-specific exposure–effects relationships developed during the ERA as the primary basis. The risk-based cleanup value is critical information needed by risk managers for remedy evaluation and final clean-up decisions for the hazardous chemicals that pose unacceptable risks. However, USEPA cannot conduct a resource-intensive risk assessment for every situation and decision, and the Agency must balance the uncertainty of the risk assessment with the resources available and the scope of the decision to be made.
The 8-step process for conducting ERAs under Superfund in the United States is shown in Figure 1. The screening-level ERA is encompassed in Steps 1 and 2 and includes the use of soil screening values, including the EPA EcoSSLs (USEPA 2005). The screening step requires comparison of abiotic media concentrations to screening levels, to determine whether ecological threats are negligible or the process should continue to a more detailed ecological risk assessment (Steps 3–7). It should be noted that, if screening levels do not exist for a given contaminant, the contaminant is not excluded from the baseline ERA. Step 3, problem formulation, sets the conditions and assumptions under which the baseline ERA will be completed for the assessment endpoints (AEs), which are explicit expressions of the actual environmental values (e.g., ecological receptors) that are to be protected. Assessment endpoints are integral to the ecological risk assessment and generally focus on survival, growth, and reproduction of receptors. The conceptual site model encompasses the various exposure routes to be evaluated in the risk assessments. Within this critical step, the guidance describes the refinement of preliminary contaminants of concern, which can allow for consideration of scientifically derived modifying factors related to contaminant bioavailability in media such as soil (e.g., pH, cation exchange capacity, soil type). Completion of the problem formulation in Step 3 includes specifying the measurements that are needed to evaluate the risk questions for the AEs (i.e., measurement endpoints, or MEs). Step 4 involves producing the work plan for obtaining data and other necessary information (literature-based and/or site-specific) that will be needed to complete the ERA. Step 5 consists of verifying the field sampling design, if site-specific data collection is planned, and the investigation is conducted in Step 6. Risk characterization in Step 7 involves analysis of the exposure and effects data to produce a risk range for site-specific decision making.
Site-specific information from Superfund ERAs can then be used to make risk management decisions (Step 8). An integral component of such decision making is establishing the cleanup value, the development of which should consider the risk ranges that were characterized (in Step 7) for the AEs selected for the site. EPA evaluates ecological risks on assessment endpoints (AEs), and this focus allows the Agency to make definitive cleanup decisions for protection of the environment based on particular trophic positions or receptor groups on a site-specific basis. Future land use scenarios, urban and non-urban, play a role in cleanup decisions. The nonurban scenario includes both preserved lands, such as parks and natural lands, and unpreserved lands. For the AEs in the ERA, the threshold for effect is described as a range that bounds the threshold for estimated adverse ecological effects, given the uncertainty inherent in the data and models used. The lower bound of the threshold would be based on consistent conservative assumptions and no-observed-adverse-effect level (NOAEL) toxicity values. The upper bound would be based on observed impacts or predictions that ecological impacts could occur. This upper bound would be developed using consistent assumptions, site-specific data, lowest-observed-adverse-effect levels (LOAELs) toxicity values, or an impact evaluation (USEPA 1997, 1998). Therefore, the risk range for the assessment endpoints, in most cases, is expected to be reported as either the NOAEL-to-LOAEL risk range, or an effect concentration (EC)x-lower to ECx-upper risk range.
An important focus of the AEs is that the necessity for action to protect a particular environmental value must be readily communicated to the public, to avoid incurring the so-called “So What?” response. (Why spend millions of dollars to clean a site to protect mice when I am trying to get rid of mice in my house?) Thus, an effective risk communication program must effectively “sell” the need for expenditures of resources resolve these unacceptable risks. It is worth noting that different regulatory programs have different risk assessment and communication goals. Programs such as EPA's pesticide registration program or wastewater treatment oversight seek to minimize the risks associated with releases before they occur, whereas programs such as Superfund conduct risk assessments after a release to the environment has occurred, seeking to ameliorate the impacts of such releases.
Since the development of EcoSSLs by USEPA (2005) for use in the screening-level ERA, it is apparent that the additional research, some of which was used in other programs such as REACH (Europe) and the Australian National Environment Protection Council (NEPC), provides additional data and methods that could be useful to North American risk assessors and risk managers for the development of site-specific soil cleanup values. No synthesis of the methods and data has been published that would provide an informational resource to allow for more widespread incorporation of the methods and data into ERA practice that remains consistent with the Metals Framework (USEPA 2007) and EPA guidance.
Canadian Soil Quality Guidelines
The Canadian Council of Ministers of the Environment (CCME) describes a tiered framework through which soil quality guidelines for a specific site are developed (Figure 2) (CCME 2006). The national framework incorporates the following 3 principles:
- A consistent risk-based approach to evaluate and set priorities for remediation of contaminated sites
- A tiered approach to assessment and remediation with generic national criteria (or guidelines) and guidance on site-specific objectives
- Equal protection of human health and the environment
Tools for each decision point are available—notably, CCME (2006) for derivation of environmental and human health guidelines, and CCME (1996a, 1996b) for guidance on risk assessment and development of site-specific remediation objectives (SSROs), respectively. Although a national policy and generic guidelines do exist, provinces and territories can and do develop jurisdiction-specific guidelines and implementation policies.
In Canada, contaminated soils are generally under provincial or territorial jurisdiction, with the exception of national parks and federal government properties. As a result, the approaches to addressing soil contamination vary to some extent across the country. However, some efforts have been made to harmonize approaches across the country, particularly through the CCME. CCME is an intergovernmental body represented by all provincial and territorial environment ministries, as well as Environment Canada. One area in which CCME is active is the development of environmental quality guidelines, including soil quality guidelines that can be adopted either directly or in modified form by member agencies.
The Canadian primary guideline document is a living document (CCME 1999 and updates). These guidelines include soil quality guidelines that were derived using methods described in CCME (2006), but some existing guidelines have not been updated to reflect changes in this latest version of the protocol. Guidelines presented in CCME (1999 and updates) are not federal regulatory numbers, although they are applied for the assessment and remediation of sites under federal jurisdiction (e.g., government properties and national parks). Provinces or territories often include or refer to these guidelines in their own regulations, however, which are applied to all other contaminated sites.
Some provinces may modify the CCME (2006) methods and/or numeric values (CCME 1999 and updates) to derive their own guidelines. For example, Alberta Environment (2010a, 2010b), Government of British Columbia (2011), and Ontario Ministry of Environment (2011) describe provincial soil quality guidelines. However, the subsequent discussion herein addresses CCME (2006) only.
The underlying objective of Canadian soil quality guidelines is to “provide a healthy functioning ecosystem capable of sustaining the current and likely future uses of the site by ecological receptors and humans” (CCME 2006). The guideline framework is shown in Figure 2. The implementation of this process varies among jurisdictions, but in general, the 3 tiers include (CCME 2006):
- Adoption of generic guidelines (CCME 1999 and updates) as remediation objectives
- Limited modification of generic guidelines to calculate a site-specific remediation objective
- Use of site-specific risk assessment
Generic soil quality guidelines in Canada vary with land use. Four land uses (agricultural, residential-parkland, commercial, and industrial) are defined by CCME (2006); provinces may use different or additional land uses (e.g., natural areas) (Alberta Environment 2010a). Within a specific land use, there are 2 major receptor groups: human and ecological. Each land use has a defined scenario with associated receptors, exposure pathways, and level of protection; as a result, the pathways associated with each receptor group vary with land use (Table 1). The ecological receptor pathways are presented below. Human pathways are not discussed further, as they are beyond the scope of this article.
|Soil contact||Soil nutrient cycling processes, soil invertebrates, crops/plants, livestock/wildlife||Soil nutrient cycling processes, soil invertebrates, plants, wildlife||Soil nutrient cycling processes, soil invertebrates, plants, wildlife|
|Soil and food ingestion||Herbivores, secondaryb and tertiaryb consumers||Herbivores,b secondary,b and tertiaryb consumers|
|Ingestion of contaminated water||Livestock|
|Contact with contaminated water||Freshwater life, crops (irrigation)||Freshwater life||Freshwater life|
A guideline is estimated for each pathway within each of the land uses shown in Table 1. The lowest number for all pathways within a land use becomes the generic guideline. The metal guidelines are based on total concentrations, and not “added risk” relative to concentrations as is done in the “added risk” approach used in Australia. If risk-based guidelines are lower than the Canada-wide “typical” background concentration, the latter is adopted as the guideline. Provisions are available to estimate a site-specific background concentration.
At present, there is no ability to modify guidelines for the soil contact pathway, and only limited options exist to modify guidelines for the soil and food ingestion pathways, because there are no site-specific parameters in the relevant models. Therefore, in practice, the only options are to apply the generic guidelines as remediation objectives or to conduct a detailed site-specific risk assessment. Specific metal-relevant pathways are discussed below.
Direct soil contact guidelines for the protection of plants and soil invertebrates are derived using laboratory and field plant and invertebrate toxicological data from the scientific literature that meets specified criteria. A hierarchy of approaches can be used depending on the quality and quantity of available data. The preferred approach involves ranking EC25 values (or the closest available EC values) from toxicity studies. The 25th percentile of this ranked distribution is used to establish the guideline value for agricultural and residential-parkland land uses, based on the assumption that only minimal effects would be observed at this level. The 50th percentile of the distribution of EC25 values is used to establish the commercial and industrial guidelines, because the primary focus of the commercial and industrial guidelines is ornamental vegetation, and a low level of effects is considered acceptable. Uncertainty factors can be applied, depending on the quantity and quality of data (CCME 2006).
A similar approach is used to calculate guidelines for the protection of nutrient and energy cycling processes, although at this time, the guidelines are generally calculated using LOEC values (CCME 2006). Preference is given to toxicity data showing effects on nitrification and N fixation; if additional data are necessary to meet minimum data quality and quantity requirements (at least 10 data points from 3 different studies, with at least 2 nitrification and 2 N-fixation endpoints), then C cycling and N mineralization studies can be included as well. Guidelines for this pathway have been calculated for only a small number of substances at this time. Historically, this guideline was used as a “check” to adjust the guideline calculated based on plants and invertebrates; the current CCME (2006) protocol treats this as a separate pathway, but professional judgment is used to determine whether this value should be used as a guideline value based on the quality of the available data.
Soil and food ingestion guidelines are calculated based on grazing herbivores, except in the case of persistent substances with a strong tendency to bioaccumulate or biomagnify, for which secondary and tertiary consumers are also evaluated using bioconcentration factors. There is also a provision for other wildlife receptors to be considered if there is strong evidence that a particular receptor may be especially sensitive to the contaminant, and if sufficient data are available to calculate a meaningful guideline based on that receptor. The guidelines are calculated based on the species considered to be most threatened, as determined by the ratio of exposure to toxicity limits; the minimum data requirements include toxicity data for at least 2 mammalian species (including, where possible, a grazing herbivore) and 1 avian species. The lowest-effects dose is used, modified by an uncertainty factor if necessary. Bioavailability from food and soil can be considered if there are sufficient data, but in practice, the bioavailability factor is generally assumed to be one (CCME 2006).
The implementation of soil quality guidelines falls under provincial or territorial jurisdiction, and these guidelines can be used as screening or cleanup land use specific guidelines. Particularly when dealing with metals, there are limited options for modification short of a full site-specific risk assessment. A bioavailability adjustment factor is included in the soil guidelines for food and ingestion, but no guidance is provided on how it should be estimated on a site-specific basis. The only consideration of bioavailability in direct soil contact guidelines is a provision that if, based on best professional judgment regarding the nature of the toxicity data, the studies represent unusual bioavailability conditions, an uncertainty factor may be used to adjust the estimated percentile of the toxicity database, thereby reducing the cleanup level required under the soil contact guideline. No provision is made for the low bioavailability condition where a guideline might be increased defensibly. The only recourse to not using generic guidelines for metals is, in many cases, detailed site-specific risk assessments or risk management.
For direct soil contact pathways, risk assessment often involves costly and time-consuming site-specific chronic plant and invertebrate toxicity testing. In most jurisdictions, the minimum requirements for site-specific toxicity tests are a suite of at least 3 plant and 2 invertebrate tests that meet the same requirements as for a generic CCME guideline. In practice, this results in the generic guidelines being applied for all but the largest contaminated sites. For soil and food ingestion, approaches are available for calculating guidelines using site-appropriate receptors (Environment Canada 2010a, 2010b, 2012).
Differences between Canada and US guidance
- Canadian guidelines integrate human and ecological receptors in the development of soil quality guidelines, whereas USEPA EcoSSLs focus on ecological entities that need to be protected and the receptor appropriate to the assessment endpoints. Microbial processes are considered as a specific pathway, even though these are not always calculated.
- In Canada, the soil contact pathway includes plants and invertebrates, as well as, potentially, nutrient and energy cycling. If that pathway is the most sensitive within a land use, the associated pathway-specific guideline will become the guideline for that land use and will be used as a screening or remediation value, because these pathways are considered surrogates for overall ecosystem function and land-use capability. In the United States, screening and cleanup values are generally driven by higher trophic level evaluations.
- In Canada, soil quality guidelines use 4 generic land uses for human and ecological risk assessments (residential-parkland, agricultural, commercial, and industrial), each with an associated exposure scenario and level of protection. The USEPA uses multiple exposure scenarios for human health risk assessment, but generally uses only 2 scenarios for ecological assessment—urban and nonurban.
- Wildlife is considered in Canadian soil quality guidelines, but this consideration is generally limited to exposure of grazing herbivores to metals. The USEPA approaches wildlife and birds much more specifically, both as assessment endpoints and as specific receptors.
- In general, the Canadian guidelines are more focused on protecting the ecosystem as a whole, rather than on specific endpoints or receptors, whereas the USEPA process tends to compartmentalize the risk assessment based on assessment endpoints and receptors associated with these endpoints.
NEW DATA AND METHODS
Overview of new data and methods developed over the last decade and its application in Europe
In Europe, the REACH () regulation (EC) No. 1907/2006, implemented in 2008, requires that all producers and importers of chemicals assess the risks to human health and the environment of their products in Europe and recommend measures to manage any risks that may occur. Assessments have to be made at the European, regional, and local scales. The terrestrial environment is one of the compartments to be assessed and has 2 protection targets: 1) organisms living in soil and directly exposed to soil, which includes plants, invertebrates, and microorganisms; and 2) higher organisms, including wildlife and birds, that can be affected through secondary poisoning.
Although the principles of the risk assessment and the minimum data requirements are specified in the regulation, extensive technical guidance documents were developed describing the risk assessment methodology, the data requirements, how to assess the quality and relevance of data, and how to derive safe limit values. Guidance documents are available at http://echa.europa.eu/guidance-documents/guidance-on-reach. Dedicated guidance on the terrestrial compartment is given in “Guidance on information requirements and chemical safety assessment Chapters R.7c and R10” (ECHA 2008, 2012).
An important element in the ecological risk assessment is the derivation of safe limit values or predicted no-effect concentrations (PNECs) for each compartment. For the soil compartment, 2 PNECs are derived, 1 to protect soil-dwelling organisms (PNECsoil) and 1 to protect predators from secondary poisoning (PNECoral).
If risks are identified, the risk assessor can either refine the exposure and effects assessment or propose risk management measures to reduce exposure levels below the derived PNEC. Users of the substance have to implement these measures or prove that their exposure values are below the PNEC. The PNECs are generic values that are applicable across Europe and for all land uses, including pristine environments. No differentiation in protection level is made among different land uses. PNEC values are derived in a tiered approach, starting at a precautionary level. The tiered approach includes:
- Applying the equilibrium partitioning method using the PNECwater and Kd of the substance, in the case where no terrestrial ecotoxicity data are available
- Dividing the lowest EC50 or EC10/NOEC value by a large assessment factor, when only a minimal data set is available to conduct a more detailed and accurate assessment
- Using the species sensitivity distribution approach and applying a small assessment factor on the 95% protection level (HC5) to address remaining uncertainties, when more extensive and robust data are available
PNEC values may not fall below the background concentration when based on total metal concentrations. When this is the case, a PNEC added is derived, which is the maximum allowed concentration on top of the ambient background concentration.
For several metals, even those with a considerable amount of data, the above method resulted in generic PNEC values in the background concentration range, implying that risks exist at background concentrations. Such an outcome can be explained by the fact that important aspects for metals, such as variations in ambient background concentrations, differences in bioavailability, and therefore toxicity, in different soil types, and laboratory-to-field differences (aging and leaching processes), were not considered.
It was generally recognized that the above factors would have to be addressed to reach more realistic risk assessments. Because insufficient information was available to quantify these factors in risk assessment, extensive research projects were undertaken to fill these data gaps. Through the REACH () regulation, an extensive set of new, high-quality data on the ecotoxicity of metals in soils has been created, and for a number of metals, models have been developed that consider bioavailability, laboratory-to-field differences (aging-leaching processes), and background concentrations.
Additional toxicity tests were performed to meet the minimum requirements for the application of a species sensitivity distribution (SSD) under REACH, which requires more than 10 species from more than 8 different taxonomic groups. Furthermore, to be able to assess the influence of soil properties on the toxicity of a number of metals on soil-dwelling organisms, standard tests with plants, microorganisms, and invertebrates were performed in 7 to 19 typical European soils. These data were used to develop regression models to predict the toxicity of metals in soils to soil-dwelling organisms (Smolders et al. 2009).
Research has indicated that toxicity of metals is less pronounced in the field than in laboratory experiments. These laboratory-to-field differences are attributed to the irrelevance of the laboratory test conditions to field conditions: in laboratory toxicity tests, metals are added to soils as soluble salts in relatively large amounts, and tests are started after a short equilibration period of about a week. In such tests, toxicity is not only influenced by the soluble metal ion, but also by the considerable amount of salts added to the soil (significantly affecting the pH and ionic strength of the porewater). In the field, metals often enter the soil gradually over a considerable amount of time, in forms that are less soluble or bioavailable than salts (e.g., aerial deposition). Over time, the bioavailability of soluble metals in soil reduces through an aging process (i.e., slow equilibration processes). The bioavailability of metals in the field is therefore much lower than in laboratory experiments.
Research to quantify the aging effects included laboratory tests with plants, invertebrates, and microorganisms repeated over time (1–18 months) after soils had been spiked with metal salts. Where single-metal-contaminated field gradients were available, the toxicity of the field-contaminated soil was compared with the toxicity of the corresponding control soil freshly spiked with the same amount of metal as a salt.
The salt effect was assessed by comparing results of ecotoxicity tests with leached and unleached soils, whereby the former soils where freshly spiked and leached, removing salts from the porewater, before performing the ecotoxicology tests. A leaching-aging (LA) factor was then derived to take into account the effects of aging in the field and leaching of salt.
Background concentrations and soil properties
Initially, high quality and consistent data were lacking background concentrations of metals in soils. In cooperation with Eurogeosurveys, a European monitoring project (GEMAS) was set up to collect soil samples in agricultural soils and to determine the background concentration and soil properties that influence the bioavailability of metals in soils (clay content, pH, cation exchange capacity [eCEC], and percent organic matter [OM]). The availability of this geo-referenced data set allowed for location-specific bioavailability corrections and accurate assessments at local, regional, and continental scales.
Table 2 presents an overview of the data and models available for a number of metals: Zn, Cu, Ni, Co, Mo, and Pb. It should be noted that, for many metals, such models or extensive data sets have not been developed, generally because no risks were identified by the more precautionary assessment factor method, or because anthropogenic emissions are very small in comparison to the background concentrations (e.g., Al, Fe, and Mn).
|Metal||Toxicity data (n)||Data normalized with||L/A factor||Background data (n)|
|Zn||171||eCEC, background Zn, pH||3||4500|
|Cu||252||eCEC, %clay, %OC, pH||2||4500|
|Ni||173||eCEC||1–3 (increasing as a function of pH)||4500|
|Co||141||eCEC||1.1–3.5 (increasing as a function of pH)||4500|
Figure 3 describes the approach that was followed to derive soil type and site-specific PNEC values, taking into account bioavailability, leaching-aging, and ambient background concentrations. In the first step, data are screened for quality and relevance. Next, added NOEC/EC10 values derived in the laboratory using freshly spiked soils are converted to field values by multiplying the added NOEC/EC10 by the LA factor. Background values of the soils tested are then added to the normalized NOEC/EC10 values, to obtain total soil concentrations. These values are subsequently normalized to conditions (pH, organic matter content, cation exchange, etc.) of a specific soil using the bioavailability regression models that are relevant to the species-endpoint. Where multiple data are available for the same endpoint, the geometric mean for the endpoint is derived. If data are available for multiple endpoints for the same species, the most sensitive endpoint is retained. The SSD is then constructed using the retained species values, and the HC5 value is estimated. Finally, an assessment factor ranging from 1 to 5 is applied on the HC5 to address any remaining uncertainties. The advantage of using soil-type-specific PNEC values instead of a generic PNEC value is that the former is more accurate and will lead to less under-prediction and overprediction of risks.
User-friendly tools to derive the PNECsoil
Application of the bioavailability models and LA factors to the ecotoxicity data can be complex and time consuming, so user-friendly tools were developed (http://www.arche-consulting.be/metal-csa-toolbox/soil-pnec-calculator). These tools integrate the data and models for a range of metals, allowing assessment of risks in a minimum amount of time at any location where information on soil properties is available.
The assessment of substance toxicity to wildlife and birds (secondary poisoning) is assessed through the pathway soil → earthworms → earthworm-eating birds and/or wildlife.
For the assessment of secondary poisoning, a conservative approach is followed whereby the PNECoral is derived based on the lowest ECx value for effects of oral intake of the metal, divided by an assessment factor, the metal in the food is assumed to be 100% bioavailable, a constant bioaccumulation factor for uptake in earthworms across all soils is assumed, and the diet of the organism is assumed to be composed of 100% earthworms (50% from local and 50% from regional area). For several metals, this approach leads to PNECoral values below the background concentrations. Research to address this was focused primarily on making adjustments for bioavailability of the metal in food to replace the default of 100% bioavailability.
For certain essential elements such as Cu and Zn, secondary poisoning was not considered relevant, because data indicated that uptake was regulated (homeostasis) and toxicity occurs first in the lower part of the food chain before any higher levels would be affected.
Overview of new data and methods developed over the last decade and its application in Australia—Development of Australian Ecological Soil Criteria
In 2011, the Australian National Environment Protection Agency published several draft reports updating their procedures to assess site contamination (NEPC 2009, 2011a, 2011b). They developed ecological investigation levels (EILs) that act as a Tier 1 assessment of the toxicity of soil metals to plants, soil invertebrates, and soil microbial processes. When EILs are exceeded, further investigation is triggered through higher tiers of risk assessment and site-specific data.
The strengths of the EIL method are that it incorporates many of the advancements in the assessment of metal and/or soil chemistry and the effects of metals on soil ecological endpoints. These strengths include:
- Use of SSDs to estimate a concentration protective of a selected percentage of species
- Estimation of EILs and soil quality guidelines (SQGs) for different land uses
- Consideration of bioavailability to, depending on appropriate data, derive soil-specific EILs
- Addresses aging and leaching for aged soil contamination
- Addresses the ambient background soil concentration issues
Development of EILs
The details of EIL development can be found in the Australian draft publications (NEPC 2011a, 2011b) and other manuscripts from this workshop. Herein, we briefly review regulatory and environmental policy issues associated with the EIL development. Key aspects of the NEPC (2011a) report are that the literature was evaluated using a data quality process, LOEC and EC30 values were used, the toxicity data set was normalized for endpoints and soil chemistry, an SSD process was used, and key soil parameters, (e.g., pH and CEC) were identified and matrices developed that presented predicted toxicity values versus ranges for the key soil parameters. In addition, the aging of metal contaminants in soils was addressed. In total, the report presents an approach to addressing the key bioavailability issues for metal effects on soil biota. The authors reviewed the toxicity literature and used data quality criteria to finalize the toxicity database. Expert judgment was used in developing toxicity data conversion factors and adjusting endpoints. Then the toxicity data set was normalized to soil parameters of: 6.0 pH, 10 cmol/kg CEC, 10% clay content of the soil, and 1% organic matter. Normalization relationships used open-literature publications that estimated empirical relationships between the toxicity data for a single contaminant and a given species, and the soil chemistry properties where the tests were conducted. For each metal assessed, the key soil parameters that affected the bioavailability and toxicity of the metal were identified. For example, the key parameters for the metals were pH and CEC for Zn; pH, CED, and clay for Cu; CEC for Ni; and pH and clay for Mo. A normalization process was used to minimize the effect of soil parameters on the toxicity data, so the resulting toxicity data would reflect the effects of metal contamination on test species. These normalized data were then used to generate an SSD for plants, soil invertebrates, and soil microbial processes.
Various studies have shown that, over time, metals in soil become less toxic due to binding to soil particles and soil chemistry reactions. Aging of metals in soils was addressed using the results from a study by Smolders et al. (2009). They conducted studies to determine aging and/or leaching factors (LAs) for Zn2+ (3), Cu2+ (2), Ni2+ (1–3), Co2+ (1.1–3.5), Pb2+ (4.2), and Cd2+. These were based on toxicity measures in a variety of European field and freshly spiked soils.
The Australian report (NEPC 2011a) estimated a percentage of species protection from the SSDs that varies with land-use category and the reference soil (pH 6.0, etc.) to produce a range of added contaminant levels (ACLs) for the metals that were assessed. Table 3 summarizes that information.
Expert judgment and environmental policy in the NEPC process
Because of the limited availability of soil toxicity data for a number of substances, various conservative assumptions, assessment factors, and conversion factors were used. This section presents an overview of the primary assumptions and factors.
In the Australian method, toxicity data were grouped to increase the usable data available. For example, toxicity data that cause a 20% to 40% effect, and LOEC data, were assumed to be equivalent and were referred to as EC30 and LOEC data. For toxicity data reported as ranges, usually the lowest value of the range was used for a conservative estimate of toxicity.
The LOEC/EC30 toxicity data were normalized as outlined in Heemsbergen et al. (2009). Geometric means for each toxic endpoint (e.g., mortality, reproduction, seedling emergence) for each species were calculated, and the lowest geometric mean was selected for the SSD to represent the sensitivity of each species and/or microbial process. Chronic NOEC, LOEC, EC/LC50, and maximum acceptable toxicant concentration (MATC) values were used by applying conversion factors to convert them to LOEC/EC30 values (Warne 2001).
For background, the authors used an added concentration, above background approach and expressed all toxicity data as an added concentration. Measured ambient background concentrations (ABCs) were used if available, but if a reliable background concentration was not available, then various estimation methods could be used (Hamon et al. 2004).
USE OF NEW DATA AND MODELS FOR THE DERIVATION OF US CLEANUP GOALS (TIER 2) AND TIER 2 GUIDELINES (CANADA)
REACH () has created a wealth of new high-quality data on toxicity of metals to soil-dwelling organisms and, for a number of metals models, allowing the derivation of site-specific limit values. The following discussion provides examples and descriptions of the methods that can be incorporated into ERA practices in a manner that is consistent with USEPA and Canadian ERA guidance, and can provide an improved database for potential use in the development of cleanup values at the end of the ERA.
The REACH database for plants, soil invertebrates, and microbial communities incorporates the primary experimental design and soil chemistry issues. High-quality standardized protocols to conduct toxicity tests with metals that address husbandry, test design, and measurement of putative soil toxicity modifying factors (such as pH, CEC, particle size) have resulted in improved guidance documents (NEPC 2011a, 2011b) for assessment of the effects of metals on biota. Checkai et al. (this issue) review REACH and NEPC data and approaches and put forward an approach to generate soil cleanup values.
The applicability of the soil parameter ranges used by REACH and EcoSSL efforts is important for remediating ecological risk assessments. If the site-specific soil parameters fall outside the REACH database test range, then we would recommend validating their use under those specific conditions before applying them to a given site. Relevance of the data, including soil properties and species tested, for North American soils will need to be verified for exceptions. If soil properties or species tested are not found in North America, these data should be excluded when soil criteria calculations are conducted in North America.
The SSD concept
Species sensitivity distributions have been used in the assessment of pesticides and are effective for using available toxicity data to estimate protective chemical concentrations (Posthuma et al. 2002). Under REACH () in Europe, the PNEC values derived using the SSD approach aim to protect at least 95% of the species living in pristine environments. NOEC and EC10 values are used to build the SSD. The data and models that have been created can be used as well to set different protection targets. Australia and Flanders-Belgium, for example, have used these in the derivation of soil type and land-use-specific limits. Australia has validated the models for their soils and used them to set EILs.
Toxicity test measurement endpoints within a narrow range of effects for a specific contaminant can be presented as a frequency histogram. The histogram represents a range or distribution of sensitivities of the species to that contaminant. If the frequency histogram is cumulative (i.e., the y-axis now represents the cumulative number of species affected at a specific concentration), this cumulative distribution function is known as an SSD. The SSD can be used to generate an environmental quality guideline (EQG) by choosing which fraction of the species should be protected and predicting the associated concentration. This is illustrated in Figure 4B.
The SSD approach to estimating EQGs supersedes other approaches that typically involve applying an uncertainty factor to the lowest-effect concentration in the collection of toxicity test measurement endpoints. Species sensitivity distributions are generally regarded as an improvement to the uncertainty factor approach, because the criticisms of using uncertainty factors are obviated, undue reliance on the extreme value in the data set is obviated, and an EQG is not driven by a search for the most sensitive taxa/measurement endpoint. The SSD approach allows for defensible and transparent changes in protection levels (e.g., cross-reference to 25th vs 50th percentile choices for agricultural, residential, and parkland vs industrial and commercial land uses). Finally, the SSD approach rewards increased data collection efforts with a demonstrable increase in precision, whereas the uncertainty factor approach does not.
Use of SSDs for soil-dwelling organisms
The additional methods and data on metals toxicity and bioavailability that have been generated over the past decade, especially for microbes, plants, and invertebrates, provide additional support for a potential increased use of SSDs in ecological risk assessment, and the selection of ecologically protective soil cleanup values by risk managers. The toxicity databases that are now available can be used to calculate effects levels (e.g., ECx) on which risk ranges can be developed and cleanup levels selected from within the range. Two options for the development of a risk range are shown in Figure 4. In the first example (Figure 4A), the risk range is developed from the toxicity database by generating the SSDs for 2 point-estimate effects concentrations (ECx) of interest, an upper and lower ECx value, for making bounding decisions. Then, a probability level (percentile) for protection of species is chosen, and the corresponding concentrations from the SSDs define the lower and upper bounds of the risk range. In Figure 4B, a single SSD is developed from the data for a selected ECx, and then 2 percentile levels are selected for defining the risk range off of the SSD curve. Factors that can be considered in the selection of the ECx level of interest include statistical considerations such as the confidence interval on the ECx estimates and minimum significant differences expected for population level responses, risk assessor's and/or manager's evaluation of the practice of using EC20 as a substitute for a NOAEC, risk assessor's and/or manager's evaluation of the practice of using EC50 as a substitute for a MATC, and land-use assumptions (Canada, United States). Factors that can be considered in selecting the percentile level for protection of species include evaluation of clustering patterns of test species (surrogates for assessment endpoints) on the SSD, consideration of habitat quality at a site, and other judgments that challenge the risk assessor and manager to select reasonable levels of species protection relative to the ECx values used to derive the SSDs.
One option not commonly applied in ERA—but used in human health risk assessment—is the use of benchmark dose models to ascertain population effect levels.
If a sufficient body of data is available for a number of species, the SSDs can be generated, and the HC5 can be used to broadly capture the range of possible effects concentrations for all species. These models can be used in a site-specific manner relevant to the species of concern (e.g., a site-specific SSD). Low effect concentrations such as EC/IC10s (effect concentration/inhibition concentration corresponding to a 10% effect) may fall within the range of allowable control variability for standardized toxicity tests. A guideline driven by an effect concentration that could arise from the uncertainty inherent in test measurement is not defensible. Because of the formula used to predict effect concentrations, 50% predicted effect concentrations are predicted most precisely, and effect concentrations moving away from 50% in either direction are estimated less precisely. Consequently, EC/IC10s are predicted less precisely than EC/IC20s (Chapman et al. 1996; de Bruijn and Hof 1997).
Wildlife toxicity and secondary poisoning
Bioavailability of metals in food plays an important role in assessing the toxicity of metals to wildlife and birds. REACH () addressed secondary poisoning of metals through a soil to the earthworm-to-bird pathway. The use of uncertainty factors by EU regulators has made the results of this process near or below background for most of the metals evaluated, so this REACH activity is not constructive for application in North America. In the United States and Canada, direct toxicity to wildlife is assessed, and it is recognized that biomagnifications of inorganic metals across 2 (or 3) trophic levels generally does not occur (USEPA 2007).
The workshop wildlife toxicity work group put forward several recommendations for improvement of wildlife ERAs (Mayfield et al. this issue) that could be used in the United States and Canada. Site-specific information can be used to refine the wildlife assessment endpoints used in ERAs. This can be done by focusing on specific species of concern, understanding their population dynamics in refinement of what toxicity data are most relevant, and segregation of data into the foraging guild that best represents the species of concern. Sample et al. (this issue) stated that improved realism in site-specific wildlife ecological soil criteria values (Eco-SCVs) could be achieved by obtaining more realistic estimates for diet composition, bioaccumulation, bioavailability and/or bioaccessibility, soil ingestion, spatial aspects of exposure, and target organ exposure. The use of bioavailability information on the metals in the food and soil relevant to the receptor, and the source and type of contamination, are important to this refinement (e.g., bioavailability of a Ni-salt is much higher than that of Ni from slag). These considerations allow for site-specific or metal-specific bioavailability adjustments.
Another recommendation relevant to refinement of toxicity reference values (TRVs) used for ERA and development of cleanup values is to use tools that incorporate complete dose-response functions, with a focus on using dose metrics (e.g., ECx), rather than continuing to rely on less reliable toxicity estimates (e.g., NOAELs and LOAELs). One option that is not commonly applied in ERA, but is used in human health risk assessment, is the use of benchmark dose models to ascertain population effect levels. If a sufficient amount of data is available for a number of species, SSDs can be generated and the HC5 can be used to broadly capture the range of possible effects concentrations for all species. These models can be used in a site-specific manner relevant to species of concern (e.g., a site-specific SSD). Other models are available for species extrapolation of toxicity data (e.g., mechanistic information, adverse outcome pathways, physiologically based toxicokinetic models, etc.) to further address wildlife AEs. Consideration should be given to the development and use of tissue-specific toxicity values and the use of alternative lines of evidence (e.g., field data) to refine and adjust toxicity values and develop a weight-of-evidence approach for species of concern. Finally, toxicity values could be bracketed (e.g., through 10% and 20% population estimates) for derivation of acute, subacute, and subchronic TRVs.
Mayfield et al. (this issue) recommended the following next steps:
- Use of bioavailability information on the metal in the food and soil relevant to the species, the site, and the source of contamination (e.g., bioavailability of a Ni-salt in food is much higher than that of Ni-oxides in food)
- Develop procedures for wildlife toxicity studies intended for TRV development
- Establish a multistakeholder expert panel to develop standard dose-response curves and TRVs for use in both screening-level and baseline ERAs
Use of data on soil processes and soil microbial toxicity testing in the United States
The soil microbial processes workgroup in the workshop provided several recommendations for incorporating microbial endpoints. Since the derivation of EcoSSLs in the United States, the science of soil processes and soil microbial toxicity testing for incorporation into ERA protocols has moved forward. Microorganisms are essential for sustaining soil fertility, structure, nutrient cycling, and groundwater purification, and test data on these species are expected to be useful within the context of the toxicity database for soil-dwelling organisms. Recent work with metals such as Cu, Ni, Zn, and Mo has shown that the toxic-effects concentrations observed for microorganisms are generally within the range shown by plants and invertebrates (Kuperman et al. this issue). In an SSD context, for those metals investigated, including soil processes does not substantially change the HC5 when these processes are incorporated into the models. By combining the data for these soil-dwelling groups of organisms (microorganisms, plants, and invertebrates), the statistical power of SSDs that can be developed from these data will be improved. In North America, soil processes and microbial endpoints can be used in Canada for soil contaminant management, whereas in the US Superfund Program, microbial endpoints are not selected for driving cleanup decisions, and therefore, soil cleanup values are not developed based on such endpoints. However, the improvements to the SSD relationships from incorporation of the microbial data can improve the confidence (at this point, it is an “optical” confidence, i.e., we are using more phyla and, hence, are more certain that the number is “more” able to support an ecosystem) in soil cleanup values and decisions at Superfund sites that pertain to plant and soil invertebrate assessment endpoints. Therefore, in this way the microbial data would serve as supplemental and auxiliary data, and potentially would play a role in applying a weight-of-evidence approach to the ERAs conducted under Superfund.
PROCESS TO PROVIDE METHODS AND GUIDANCE ON THE DEVELOPMENT OF SOIL CLEANUP VALUES
Use of new data and methods in Canada
For any of these new data or procedures to be incorporated into site-specific guideline modification, specific guidance documents would be needed. At the guideline modification level, the allowable approaches are generally limited and restrictive. The guidance documents would need to specify the acceptable procedures, data sources, and any limitations on their use. Full site-specific risk assessment is less restricted, and the incorporation of new data and methods is more easily accomplished at this level.
Regulatory acceptance would also be required. Soil assessment and remediation generally falls under provincial and/or territorial jurisdiction in Canada, with the exception of national parks and other federal government properties, and although there have been some efforts to harmonize soil quality guidelines, particularly through CCME, there remain a variety of different sets of soil quality guidelines and regulations on soil quality guideline implementation across Canada.
There is value in having options for assessing the ecological toxicity of metals, looking beyond the generic guidelines without having to resort to a full site-specific risk assessment. For the direct soil contact guidelines, being able to use REACH methods to modify guidelines to account for site-specific bioavailability, as well as addressing leaching and aging issues, would result in more scientifically defensible assessments. This would allow for a relatively straightforward site-specific adjustment of soil contact guidelines at contaminated sites that are not large enough to warrant a detailed site-specific risk assessment, potentially reducing remediation of metals-contaminated soils.
There is also the potential to use the new approaches developed by REACH, Australia, and others to improve the derivation of generic CCME guidelines. Specifically, SSDs should be considered for use in estimating percentiles of the soil contact toxicity database (as used in CCME  for water quality guidelines) rather than quantile estimators (as used in CCME ). It may also be appropriate to normalize the soil toxicity data for a set of specific soil quality characteristics, which would improve the ability to adjust for site-specific bioavailability considerations.
For the REACH database on toxicity to soil and plants to be used in Canada, the data would need to be evaluated for relevance to Canadian sites. In particular, the relevance of the species used in testing, as well as the range of soils, should be assessed. Also, the REACH process uses EC10/NOEC values, whereas for Canadian soil quality guidelines, EC25 (or at least effects concentrations in the range of 20–30) are preferred, so the raw data may need to be used to generate new endpoints.
REACH and Australian methods also include consideration of microbial toxicity beyond the nitrification and N fixation data used for Canadian guidelines. The relevance and applicability of these data should be evaluated further to determine whether they should be considered in generic guideline development. Use of functional endpoints such as nitrification is already incorporated into Canadian guideline development as a nutrient and energy cycling check, but the additional available data from REACH and Australia could be readily incorporated.
The database on which the REACH PNEC values were based, and the database used to inform the Australian Ecological Investigation Levels, hold a significant amount of information that both the Canadian Government and US Government could mine for information to supplement the CCME guideline values and assist in further development of soil cleanup values. CCME would need to evaluate the utility of the information for inclusion into the CCME soil guidelines. The established process to include these data in the 2 databases that support Canadian and American regulations and guidance needs to be followed. Additionally, the data will need to be evaluated for quality. The Canadian process currently has a limited suite of options for a refinement step after the comparison to CCME soil guidelines for the ecological risk investigation, but they could be modified. The Canadian Government might also find value in the use of information from the databases to inform the later stages of site-specific ecological risk assessments, including the development of values for CCME site-specific risk assessments.
Improved evaluation of mammalian and avian exposure could be of benefit if bioavailability/bioaccessibility of metals in soil and/or bioaccumulation in plants can be linked to soil properties, and this could lead to a mechanism for site-specific modification of wildlife and livestock soil and food ingestion guidelines. Improved exposure assessment will also be of use in more detailed site-specific risk assessment. Consideration of a wider range of mammalian and avian receptors, following procedures used in the United States, Europe, and Australia, could improve generic guidelines or be used for the development of ecoregion-specific guidelines.
US applications: Refinement of assumptions in the process of establishing COPCs
The algorithms for inclusion of soil factors in the refinement of contaminants of potential concern in the EPA 8-step ERA process are applicable to specific assessment endpoints (e.g., the normalization of Mo total concentrations to pH and clay is used for evaluating toxicity to soil invertebrates [van Gestel et al. 2011]). Because of the great potential for more immediate and widespread use of normalizing factors (Table 2) when evaluating ecological risk from bioavailable metals, we recommend that a compendium of this research be developed. The authors anticipate that risk assessors would benefit from a handbook style (e.g., USEPA Wildlife Exposure Factors Handbook) (USEPA 1993) in which they could look up the following for each metal or metalloid of concern: the normalization parameters; methods for measuring the parameter and, where possible, the recommended method; the normalization equations and, where possible, the recommended equation to use; any limitations on the use; the assessment endpoint (species or group of organisms or function of concern); and the supporting data, including available links. Additionally, identifying the data that are needed to adjust the toxicity data for site-specific chemistry before collection of soils for analysis will ensure the use of the normalizing functions. Risk assessors should be able to look up what metal normalization parameters are needed, methods for measuring the parameter, the normalization equations, and the recommended equations to use for different assessment endpoints. Because of the variability in standardized toxicity tests for soil biota, it is difficult to calculate precisely low-effect values (such as the concentration where 10% of the test population would be affected). Therefore, an effect concentration ranging from approximately 20% to 50% is often recommended for all endpoints. To address this challenge for US ecological risk assessors, using these new databases and methods available from REACH and Australia could lead to a change in the risk assessor's and/or manager's evaluation of the practice of using EC20 as a substitute for a NOAEC and using EC50 as a substitute for a maximum acceptable toxicant concentration (MATC). Compiling data into species-sensitivity distributions for subsequent selection of appropriate protection levels is highly encouraged.
1. Use the available data to augment data used in North America
The authors recommend that Canada and the United States take advantage of the REACH database and use their accomplishments as leverage to improve the scientific support of cleanup limits or guidance for ecological risk assessors and managers. A summary of the soil toxicity data, including NOEC/EC10/EC50 values developed for REACH, is available on the European Chemicals Agency (ECHA) web site (http://echa.europa.eu/web/guest/information-on-chemicals/registered-substances). Where industry is the data owner, data can be obtained from the relevant consortia. If different toxicity parameters are to be used (e.g., EC20 values), these can be derived from the raw data. Finally, microbe data needs to be evaluated relative to plants and invertebrates in the United States, and the data's potential use should be considered further.
2. Re-estimate the trophic magnification factor models using REACH and North American data where possible
The EPA Metals Framework (USEPA 2007) stated that inorganic metal compounds rarely biomagnify across 3 or more trophic levels and that for soil invertebrates and most plants, metal BAFs are typically less than 1 and usually are based on the total metal in soil and tissue that do not account for bioavailability differences. Recent evidence indicates that soil properties, other than metal concentrations, can influence trace metal accumulation in plants and animals and bioaccessibility of metals in soil (Sample et al. this issue). The application of site specific soil chemistry values, regression models developed to derive soil type or site-specific soil limit values, and aging-leaching factors to correct for changes in toxicity and bioaccessibility over time will improve site specific trophic bioaccumulation estimates (Smolders et al. 2009; Sample et al. this issue). The application of site specific soil chemistry values, regression models developed to derive soil type or site-specific soil limit values, and aging-leaching factors (Smolders et al. 2009) to correct for changes in bioavailability and toxicity over time is expected to improve site specific trophic bioaccumulation estimates.
3. Make adjustments for land use
In Canada, this adjustment is made merely by choosing different percentiles of the soil contact effects database and/or changing exposure pathways or exposure parameters within a specific pathway. In the United States, although land-use considerations (e.g., residential, industrial, agricultural, urban, nonurban) are important for understanding human exposures, for ecological receptors, the site habitat conditions expected under these given land-use categories can also inform the selection of appropriate effects levels or probabilities of interest (see Figure 4) for site-specific risk assessment.
4. The REACH data provide additional support for a potential increased use of SSDs in the ecological risk assessment process and selection of ecologically protective soil cleanup values by risk managers
Efforts to use the available database to generate SSDs for use in North America would provide the additional advantage of reducing uncertainties in the ecological risk assessment of metals in soils. The toxicity databases that are now available could be used to calculate effects levels (e.g., EC10, EC20, EC30…ECx) on which site-specific risk ranges can be developed and cleanup levels selected from within the range (see options discussed in the SSD discussion above and Figure 4). Once the SSDs are generated, risk managers could focus on the more difficult question of the percentile of species to be protected. The SSDs would then be used to find the soil concentrations that correspond to the fractional protection level or levels selected. This approach would be an improvement over existing approaches that typically involve applying an uncertainty factor to the lowest single-species effect concentration that is available from toxicity studies. Factors that should be considered for regulatory application of SSDs include the following:
- Selection of the ECx level
- Statistical considerations such as the confidence interval on the ECx estimates and minimum significant differences expected for population level responses
- Risk assessor and risk manager evaluation of the practice of using an EC20 as a substitute for a NOAEC, and similarly, EC50 as a substitute for a MATC
- Selection of the percentile level for protection
- Where the species represent the assessment endpoints used in the site-specific ERA cluster in the SSD (if they, in fact, cluster)
- The level of species protection selected relative to the ECx values used to derive the SSDs; these are often likely going to be judgment calls, but any technical considerations leading to the selection should be explained
- Habitat availability and quality on site at present and anticipated in the future
Recent improvements in soil ecotoxicity, bioavailability determinations, and risk assessment for metals that were developed primarily under the European REACH regulation provide a basis for moving beyond using EcoSSLs or similar screening values as a frequent and primary basis for making soil cleanup decisions. These new techniques and approaches can reduce uncertainties in ERAs, and are expected to provide the technical means to move North American risk assessors and regulatory scientists toward the development of site-specific soil cleanup values. The next steps needed to include the following: 1) compile a database of REACH, EIL, and other sources of existing and new data for use by North American risk assessors; 2) develop technical guidance, fact sheets, and white articles that specify acceptable procedures and data sources, and describe any limitations on the use of the methods; 3) hold soil ERA seminars, training sessions, and discussions that include risk assessors from federal and state regulatory agencies; and 4) select current projects where application of these methods could be demonstrated, and share the results widely. We look forward to engaging in this path forward.
The authors recognize and thank the workshop sponsors: Rio Tinto, Copper Development Association, International Zinc Association, Nickel Producers Environmental Research Association, North American Metals Council, Exponent, US Army Environmental Center, Vale Canada Limited, and International Molybdenum Association.
All views or opinions expressed herein are solely those of the authors and do not necessarily represent the policy or guidance of any other public or private entity. No official endorsement is implied or to be inferred. Mention of any specific brand or model of instrument or material does not constitute endorsement by any of the agencies.