Assessing risks of metals added to soils in Europe and North America
Risk assessment of metals in soils is being evaluated all over the world using many different frameworks. In the European Union (EU), environmental risk assessments are being conducted for copper, zinc, nickel, cadmium, and lead in soils following the framework of the Existing Substances Regulation (ESR) (European Commission Council Regulation 793/73). The EU risk assessment process is based on the derivation of a single safe threshold value, the predicted no-effect concentration for soil organisms, which can be compared to a single predicted exposure concentration throughout Europe. Both the predicted exposure concentration and the predicted no-effect concentration are based on total concentrations, which for metals have proven to be inaccurate estimators of hazard and exposure.
Several international metal research organizations have set up extensive research projects to estimate the bioavailability of metals in soils across Europe as a function of major local physical and chemical variables. The ultimate objective is to develop models to predict the bioavailability and toxicity of metals in soil. The research performed in these projects will meet the needs of industry scientists and government regulators for models to predict where and when anthropogenic emissions of metals to soil will result in unacceptable soil toxicity. The models will allow appropriate adjustments to the predicted no-effect concentration for application to local or regional-scale risk assessments.
A large body of evidence indicates that a single threshold value based on total metal concentrations has little or no meaning for the protection of soil ecosystems. Indeed, metals are characterized by a set of properties that require additional specific considerations in environmental risk assessment. Metals occur as natural elements, and the large spatial variation in the natural background concentrations need to be considered. Some metals are necessary for life, and the relationship between essential nutritional levels and natural background levels must be understood. Finally, toxicity of metals in soils is related to the soil properties governing bioavailability of metals. Thus, establishing a single threshold value could lead to either over- or underestimation of risk to many terrestrial ecosystems, which would then result in incorrect risk management decisions. Thus, we need terrestrial models that can predict toxicity to soil organisms, taking into account local or regional conditions and the metal's unique properties.
The topic for this Special Issue of ET&C evolved from a symposium at the SETAC-Prague meeting in April 2004 titled “Metals in the Environment: From Fundamental Science to Applied Risk Assessment,” which focused on the EU risk assessment programs. Papers in this publication cover the three main areas of environmental risk assessment: chemical fate and exposure; effects to plants, soil invertebrates, and micro-flora; and risk characterization and models. Papers have been placed in these sections on the basis of their predominant topical area, although in some cases they would be suitable for other areas as well. They report the results of studies using a wide variety of approaches for estimating the risks associated with metals or metal salts added to soils. Approaches of increasing realism and complexity were used: testing plants grown hydroponically in manipulated chemistries in soil solutions, testing single species in a variety of natural soils, and, finally, multispecies or natural community investigations in the laboratory and in the field. The order of papers in each section is an approximate reflection of this progression from simple, artificial laboratory approaches through more realistic multispecies assessments in the field.
Soil chemistry, fate, and exposure
Total metal concentration is not a good basis to assess toxicity and predict risks. The first section in this issue brings together seven papers that identify potentially bioavailable metal species in soils, quantify the influence of physicochemical soil parameters on the available fraction and its fate, and propose analytical techniques to measure the available fraction in soil.
Understanding the speciation of metals in soil and the physicochemical parameters influencing speciation is the core issue in the process of defining and estimating bioavailability, assessing the long-term fate of metals in the terrestrial environment, and estimating more accurately the risks of metals in the terrestrial environment. Echevarria et al. noted how soil pH and different mineralogical phases influence availability of Ni to plants. Plants with different Ni-accumulation strategies were found to take up Ni from the same labile pool. Courchesne et al. highlight the importance of specific processes in the rhizosphere influencing the speciation and thus bioavailability of metals and recommend that the rhizosphere—the environment within the soil where plants take up metals—be considered in estimating risks to plants. Zhao et al. found that for Cu, the free Cu2+ activity alone did not explain variation in plant toxicity. They speculate that other ions affect the expression of toxicity, lending weight to the need for a terrestrial biotic ligand model.
Leaching of metals through soil is another process determining the fate of metals in soil, and Bertling et al. assessed the influence of pH and soil organic matter on potential leaching of Cu through soils.
The risk of metals in soil impacting flora and fauna has traditionally been assessed using laboratory tests with soils spiked with relatively high metal concentrations. Several publications have reported on the reduced toxicity of metals in field—contaminated soils versus freshly spiked soils in laboratory tests. However, the mechanistic basis of this reduction in toxicity is poorly understood. Ma et al. and Crout et al. investigated the decline of the labile fraction of several metals over both short and long time periods using isotopic dilution techniques. Parameters were identified governing the attenuation process, and models are proposed to predict this process.
The availability of good models to predict and easy-to-use analytical techniques to monitor the bioavailable fraction are indispensable tools for the implementation of the acquired knowledge in the assessment of actual metal risk. Bertling et al. demonstrate how laboratory experiments in combination with validated models could be used to predict leaching of metals through soil.
Ponizovky et al. were able to predict the Cu activity in soil solutions for a wide range of soils with the WHAM IV (Windermere humic acid model version IV) model using either bulk soil properties or soil solution properties. Zhang et al. and Zhao et al. found diffusive gradients in thin films to be a promising technique to assess available metals to plants in a wide range of soils, while Echevarria et al. found diethylene triamine pentaacetate extraction to be an easy-to-run tool.
Traditional extraction and analytical techniques have their limitations. Promising new techniques using isotopes assist in further unraveling the fate and availability of metals in soil (Ma et al., Crout et al., and Echevarria et al.).
Effects of metals on plants
Susceptibility to metal toxicity is not an intrinsic property of the interaction of plant species with a metal but varies greatly with the lability of the form in which the metal is added and is modified extensively by the chemical and physical properties of different soils. For instance, Zhao et al. found that toxicities varied by as much as 116-fold across 18 European soils for a single plant species. Thus, predicting the toxicity or uptake of an element by plant species or setting protective toxicity thresholds is likely to be a complex, site-specific process. The papers in this section demonstrate many of the important factors controlling the bioavailability and toxicity of metals to plants that need to be considered in risk assessment.
Five studies tested metal toxicity to plant root growth in a variety of natural soils with widely varying physical chemical properties (Zhao et al. and Rooney et al.—18 European soils; Ginocchio et al.—three Chilean soils; Dayton et al. and Brad-ham et al.—21 U.S. soils; Feisthauer et al.—three Canadian soils). In contrast, two of the studies used soil solutions extracted from natural soils as their toxicity test medium (Voigt et al. and Temminghoff et al.). Ginocchio et al. also examined the lability and bioavailability of Cu delivered to soils in the form of common copper products or Cu-containing wastes— common scenarios encountered in risk assessment.
Remarkably, despite the wide variety of soils or test conditions, these studies were quite consistent in identifying a discrete set of factors as major influences on metal bioavailability and toxicity. Most of the studies showed that toxicity was controlled by pH, organic matter/organic carbon, and/or cation exchange capacity at soil pH, which depends on both organic matter and pH. Dayton et al. used path analysis, a statistical method used to separate the individual effects of multiple-correlated variables contributing to an observed correlation to identify organic carbon and cation exchange capacity as controls on Pb solubility and bioavailability across all 21 soils in their study. Zhao et al., Voigt et al., and Temminghoff et al. examined metal speciation to identify the best predictor of bioavailability and toxicity. In addition to estimating free metal ion activity, Zhao et al. and Temminghoff et al. used Diffusive Gradients in Thin Films to estimate the labile or bioavailable pool of metal for plants (see also Zhang et al.). Ginocchio et al., Zhao et al., Voigt et al., and Temminghoff et al. concluded that total metal was generally a poor predictor of metal bioavailability or toxicity.
Several of the studies have made major strides in elucidating the competitive and complexing mechanisms in soils that control metal bioavailability. They identify phenomena at the plant root that seem to be analogous to that of the “biotic ligand” identified as the primary site of metal toxic action in a variety of aquatic organisms, such as invertebrates, fish, and algae. Zhao et al. noted that toxicities associated with free Cu ion activity correlated with pH (i.e., an apparent protective effect of H+), consistent with the competition at biotic ligand binding sites in the aquatic biotic ligand model (BLM). Both Voigt et al. and Temminghoff et al. identified metal-root complexes or root adsorption as controlling uptake or toxicity, consistent with the BLM concept of a discrete number of binding sites on the biotic ligand. Voigt et al. also noted the competitive effect of Ca at the root, also consistent with the BLM construct; they tested an existing BLM as a predictor of toxicity and noted good predictions. In their multimetal experiment, Temminghoff et al. identified apparent competition among metals in two groups for the apparent metal adsorption sites on the plant root that were specific for each metal group. These studies strongly suggest that the aquatic BLM could be modified and refined to predict metal toxicity to a plant species based on the physical and chemical properties of a soil. This would be an extremely useful tool both for setting toxicity thresholds in soils and for risk assessment.
Effects of metals on soil invertebrates
This section contains nine papers that evaluate the toxicity or bioavailability of metals to soil-dwelling invertebrates. Natural soils may contain more than 1,000 species of invertebrates, with up to 100,000/m2 being detritivores and decomposers, such as springtails (Collembola—nsect)and mites, and as many as 50,000/m2 larger invertebrates, such as earthworms, snails, pot worms (enchytraeids), and other insects, such as burrowing beetle grubs and fly larvae. Detritivores feed on decaying plant and animal materials, and long-term sources of ingestion of metals could potentially lead to indirect toxicity by accumulated metals passed through the food chain to higher trophic organisms (e.g., avian). Without detritivores to decompose dead plant and animal materials, our soils would be layered in organic debris and lose their fertility. Thus, the potential risk to soil receptors posed by metals in soils warrants consideration.
Unlike the suite of organisms used to evaluate aquatic toxicity in laboratory studies, the number of soil invertebrates used in standard laboratory studies (e.g., the Organization for Economic Cooperation and Development and the American Society for Testing and Materials) to evaluate the hazards of chemicals in soils is limited to a few. Aquatic toxicity methods are well developed; culturing organisms used in the laboratory tests and understanding physiological mechanisms of metal uptake by and toxicity to aquatic organisms have a history of more than 50 years.
Only a few soil organisms, limited to one or two trophic levels, have been used in laboratory studies to measure the direct toxicity of metals in soils. Earthworms have been the most commonly used organisms to assess the toxicity of chemicals in soils. These organisms have been used in short-term contact (dermal) studies and in chronic 28- to 62-d ingestion-route-of-exposure studies that include determining growth, reproduction, and hatching of cocoons.
Six of the nine papers (Bradham et al., Inouye et al., Spurgeon et al., Kuperman et al., Rombke et al., and Feisthauer et al.) in this section report laboratory toxicity findings using earthworms, and a seventh paper (Vijver et al.) reports the distribution of nine metals in field-collected earthworms to determine the mechanisms of detoxifying metals. The spring-tail appears to be the second most tested invertebrate in soil toxicity studies and was used in three studies (Kuperman et al., Rombke et al., and Feisthauer et al.) along with the earthworm. Adding another trophic level, the scarcely used snail (herbivore), was tested in two toxicity studies (de Vaufleury et al. and Scheifler et al.). The latter involved a soil-plant-snail food chain. De Vaufleury et al. found that Cd-contaminated food was sixfold more toxic to snails than Cd via soil contamination.
Ideally, the most representative and sensitive species should be used in toxicity studies, and an effort is under way to identify the most sensitive soil invertebrates in laboratory studies. Once identified, extrapolation using laboratory data from these organisms can be attempted in the field. By protecting the most sensitive species, presumably, the population or community will be protected; however, often a single “most sensitive” species is not found in all soil environments. Some invertebrate populations exposed in the field become tolerant. Soil chemistry (e.g., pH, organic matter and clay content, and weathering and aging processes) influences metal bioavailability and toxicity to organisms. Again, the risk to soil fauna based on total metal may not accurately predict toxicity.
By using path analysis, it was found that soil pH was the most important factor affecting Pb toxicity to earthworms. In a companion paper to the Bradham et al. article, Dayton et al. uses path analysis and also found that soil pH was the most important factor affecting Pb toxicity to lettuce. Spurgeon et al. demonstrated that soil metal concentration and pH influenced metal accumulation and toxicity. Field toxicity studies using invertebrates are rare; however, site-specific studies can be conducted using methods presented by Kuperman et al. and Feisthauer et al. From an environmentally relevant point of view, it is more desirable to use natural soils in standardized tests than artificial soils such as the one specified by the Organization for Economic Cooperation and Development. Rombke et al. tested Zn toxicity in nine natural soils and found it to be more toxic to earthworms and springtails by a factor of four than when tested in artificial soil. Large data gaps exist for the toxicity of most metals to soil invertebrates in laboratory studies. The most studied metals in soil invertebrate toxicity tests are Cu, Ni, Pb, and Zn, as shown by data on these metals in seven of the nine papers in this section. The paper by Kuperman et al. includes data for three less frequently studied metals—Sb, Ba, and Be—on three ecologically relevant invertebrates, and Inouye et al. conducted a study with W on earthworms.
Effects of metals on microflora
The ecotoxicological risks of metal-contaminated soils pose potentially harmful effects not only to plants, animals, and humans but also to soil microflora. Soil microorganisms play an essential role in the edaphon, the aggregate of organisms in the soil with the exception of plant parts like roots with respect to their abundance (often >1012 organisms/kg soil), diversity (>3,000 species), and biomass (about 200–2,200 mg biomass C/kg soil) . By mineralizing organic matter and by maintaining important nutrient cycles, microbial populations contribute substantially to the sustainable soil fertility. Thus, permanent damage to the soil microflora population should be avoided.
Elevated metal concentrations in soils—both essential and nonessential—may impair microorganismsand microbial functions, such as nitrification, mineralization of organic matter, and degradation of xenobiotics. Numerous studies have dealt with toxic effects of metals on soil microflora [2,3]. Compared to other ecotoxicological tests, soil microbial activities are often recognized to be among the most sensitive endpoints. However, the huge variation in toxicity thresholds (> 1,000-fold for a single metal ) and the lack of widely accepted critical metal concentrations in soils for microbial processes must be acknowledged. A critical evaluation of recent and older publications reveals that although the potential reasons for this variation were named in many cases, the influencing parameters have not been quantified.
Indeed, an appropriate risk assessment should include a multiplicity of factors that govern the microbial toxicity of metals in soils. Differences exist in the acute and chronic inherent toxicity of metal. Each metal interacts differently with soil sorbents, depending on soil pH, redox conditions, organic matter, texture, and so on. Finally, increasing sorption and immobilization over time, as well as adaptation processes, facilitates detoxification.
Three papers deal with the microbial toxicity of metals from different viewpoints reflecting the current state of the art. The first paper of Oorts et al. presents a comparative study with Cu and Ni in freshly spiked soils. The Cu and Ni toxicity thresholds varied 19- to 90-fold among soils. Their findings indicate a strong influence of soil parameters on Cu and Ni sorption, proving that total metal content in soils is an insufficient measure, and clearly validate the ambivalent role of pH. Raising pH decreases metal solubility (via sorption) but increases the toxicity of the soluble species.
Data obtained with freshly spiked soils have been used extensively in assessing the risks of metals in soils. However, in a second paper, Oorts et al. systematically compare Cu toxicity in freshly spiked soils to that of field-contaminated soils with a 6- to 80-year-old Cu input. On the basis of identical total Cu contents, fresh additions of the metal reveal a much higher microbial toxicity than in old field contamination. The authors recommend leaching and equilibrating freshly spiked soils.
A basically different way of deriving threshold values for metals in soils is presented by Sauvé. He used the field-contaminated soils from the same site in Denmark as Oorts et al., who detected no significant microbial inhibitions in these samples (>800 mg Cu/kg) using applied classical laboratory test procedures. Based on the observation that elevated Cu contamination slows (microbial) organic matter degradation, Sauvé et al. derive a dose-response curve that relates total soil Cu to the content of soil organic matter. This integrative measurement represents a rather sensitive ecotoxicological end-point. Thresholds derived this way are highly valuable for risk assessment because they are based on old field contamination and summarize chronic effects over long exposure periods. The technique will not be applicable for all toxicants but provides an excellent opportunity to “calibrate” or validate more artificial routine tests and proposed safe threshold values.
Risk characterization and models
The final section contains papers that address the refinement of exposure and effects assessment for better risk characterization of metals. Significant scientific developments in the refinement of exposure and effects metals data increase the ecological relevance and reduce the uncertainty in risk characterization. The developments, driven in no small part by the EU Existing Substances Regulation 793/93/EC, have contributed to the understanding of the effects of soil chemistry on ecotoxicology, source description, and prediction of metal availability. This understanding will improve the characterization of the level and likelihood of effects from exposure to metals.
However, risk characterization, whatever the scale or scenario, is only as good as the exposure and effects data; it is also essential to be able to compare exposure and effects data in common dimensions. Heijerick et al. describe a probabilistic refinement of the terrestrial exposure to Cu providing worst-case predicted exposure concentrations for a range of scenarios, greatly aiding risk characterization on a regional or continental scale. The significant advantage of this work and the accompanying effects study (Rooney et al.) is that generic scenarios at a regional scale or above can be used to communicate high-level policy decisions and actions.
Prospective risk characterization is also needed for a range of purposes including planning, risk mapping, and standard setting (e.g., EU Integrated Pollution, Prevention and Control legislation begun in 1996). The critical loads work presented here (Hall et al.) draws together exposure and effects data to provide a potentially powerful, regional-scale tool for metals risk characterization, again with clear regulatory application.
In the area of retrospective risk characterization, such as in compliance assessment or contaminated land assessment, advances have been most significant for metals that are relatively data rich. Not surprisingly, these metals lend themselves most usefully to probabilistic risk assessment methods and to risk characterization as carried out for ESR. However, for compliance assessment (especially for data-poor metals), deterministic assessment must be used because compliance determination generally requires comparison to a single value, not a range (cf. Kapustka et al.). Methods and refinements are least advanced in this form of assessment, as many regulators and risk practitioners are still very reliant on assessment factors and conservative assumptions in metal risk limit or soil standard determination. Limited metals data and even less understanding of such data present future challenges for characterizing risks in the terrestrial environment in a robust but proportionate manner. Kuperman et al. have demonstrated a broadly applicable approach to terrestrial toxicity testing of relatively data-poor metals to derive ecological screening values for ecological risk assessment.
Environmental regulators generally, in our experience, are conservative toward the assessment of ecological risks from metals and their management. Often this conservatism is criticized as being a sign of reluctance to adopt new approaches or techniques for risk assessment, but it is in some ways symptomatic of the gulf that exists between the researcher and regulator communities. The development and application of the terrestrial biotic ligand model (Antunes et al.) is a sound example of how this gulf may be bridged—a research tool that has been tested in a regulatory context. The balanced discussion in this paper demonstrates the regulatory relevance of the terrestrial biotic ligand model and highlights not only the significant potential of the model's use but also its current weaknesses. Balanced, informed, and proportionate risk management depends on relevant and accurate exposure and effects integration in risk characterization. A common linkage with the papers in this section is the attempt to realistically fulfill this integration.