Debate and Commentary
Ecological risk assessment issues identified during the U.S. environmental protection agency's examination of risk assessment practices
Recently, the U.S. Environmental Protection Agency examined its current risk-assessment principles and practices. As part of the examination, aspects of ecological risk-assessment practices were reviewed. Several issues related to ecological risk assessment were identified, including the use of organism-level versus population-level attributes to characterize risk, the possible opportunities associated with the increased use of probabilistic approaches for ecological risk assessment, and the notion of conservatism in estimating risks. The agency examination provides an understanding of current practices and is intended to begin a dialogue in which the risk assessment community can engage in addressing the identified issues to improve and enhance ecological risk assessment.
In early 2004, the U.S. Environmental Protection Agency (U.S. EPA) staff published a report, “An Examination of EPA Risk Assessment Principles and Practices” (the staff paper), that presented current U.S. EPA risk assessment principles and practices (USEPA 2004). The staff paper is not a guidance document, but rather a perspective of U.S. EPA risk assessment practices that is intended to open a dialogue about how the risk assessment community can move forward to clarify, and where appropriate, improve risk assessment practices. Issues concerning specific risk assessment practices are identified with the desire to collectively address the issues and strengthen risk assessment overall.
The staff effort was initiated in early 2003 when senior U.S. EPA management requested that U.S. EPA staff examine current risk-assessment practices and make recommendations for improvement, including identification of issues that need addressed. Further, the U.S. Office of Management and Budget (USOMB) published in the Federal Register a request for views on a number of important issues pertaining to the practice of risk assessment in the federal government (USOMB 2003). The USOMB received many public comments, many of them on EPA risk assessment practices. The staff effort took advantage of these comments during the examination.
Most of the emphasis in the examination and in the public comments focused on human health assessment. However, many of the principles and practices apply to ecological risk assessments as well. The staff paper presented these considerations for ecological risk assessment and identified several issues that could be addressed. While there are a number of issues in ecological risk assessment that are ripe for ongoing discussion and research, including, e.g., the use of risk assessment methods in the Water Quality Standards Program, invasive species, and integration of ecological and economic assessments, three issues are discussed here to illustrate the potential for dialogue within the scientific community. Foremost among these issues is a concern about the use of organism- (individual-) level attributes to assess ecological risk versus a population-level assessment. Also, the staff paper recommended addressing uncertainty in risk assessments by encouraging the examination of opportunities to use probabilistic analyses in risk assessments, including ecological risk assessments. Another issue identified was the concern by some that risk estimates in ecological assessments may be overly conservative.
ECOLOGICAL RISK ASSESSMENT
In 1983, the National Academy of Sciences published Risk Assessment in the Federal Government: Managing the Process (commonly referred to as the Red Book) (NRC 1983). The U.S. EPA integrated the principles of risk assessment detailed in the Red Book into its practices. Subsequently, U.S. EPA published a series of guidelines for conducting risk assessments (e.g., in 1986 for cancer, mutagenicity, chemical mixtures, and developmental toxicology, and in 1992 for estimating exposures). These guidances focused on human health assessments. Tools and methods for conducting ecological risk assessments evolved through the 1980s and resulted in the Agency's first statement of principles for ecological risk assessment in the Framework for Ecological Risk Assessment in 1992 (Norton et al. 1992; USEPA 1992). Later in the 1990s, the U.S. EPA published guidances for ecological risk assessments in the Ecological Risk Assessment Guidance for Superfund (USEPA 1997a) and ultimately the Guidelines for Ecological Risk Assessment (USEPA 1998). These guidances describe techniques for conducting single-species, chemical-based risk assessments, but also describe techniques for assessing risks to ecosystems from multiple stressors and multiple endpoints.
As the staff paper presents, an ecological risk assessment is a flexible process for organizing and analyzing data, information, assumptions, and uncertainties in order to evaluate the likelihood of adverse ecological effects from environmental stressors (USEPA 2004). An ecological risk assessment includes an initial planning step, problem formulation, analysis of stressors and effects, and risk characterization (USEPA 1998). Problem formulation involves defining and specifying the issue under consideration by evaluating goals and selecting assessment endpoints, preparing a conceptual model, and developing an analysis plan. The analysis phase evaluates exposure to stressors and the relationship between stressor levels and ecological effects. Risk characterization includes both quantitative estimation of risk and a qualitative description of risk, consistent with EPA's Risk Characterization Policy and Handbook (USEPA 2000).
Important advances in risk assessment evolved from the development of the ecological risk assessment guidelines. For example, the interface among risk assessors, risk managers, and interested parties during the planning step at the beginning of the risk assessment is now a critical part of risk assessment. Problem formulation explicitly outlines what the important assessment endpoints are, how they are measured, and what unacceptable risk levels are. These two phases provide the boundaries of the risk assessment, i.e., what is necessary to address and what will not be addressed. Additionally, communication of risk at the end of the assessment process is critical to ensure that the results of the assessment can be used appropriately to support risk management decisions. Effective communication and transparency of the risk assessment process is an important risk assessment practice that the staff paper highlights and endorses. These advances in the ecological risk assessment approach are now being applied to improve the human health risk assessment approach and risk assessment in general.
Organism-level and population-level assessment approaches
The U.S. EPA recently published guidance on developing ecological assessment endpoints (USEPA 2003a) that analyzed the rationale for selecting various levels of biological organization as endpoints for risk-management decision making. The decision to use organism-level or population-level endpoints in assessing ecological risk should be made in the problem-formulation stage of an ecological risk assessment. The use of organism-level attributes is well supported by policy and precedent and by U.S. courts (see section 6.2 in USEPA 2004). The protection of organism-level attributes is generally interpreted as occurring in a population or community context (USEPA 2003a). That is, increased mortality or decreased fecundity or growth of organisms in an assessment population or community is assumed to be significant, even with no other outward demonstration that a population- or community-level property is affected.
However, the assessment focus is expanding to populations, communities, and critical habitat and ecosystem functions due to broader resource management questions posed by international trade agreements, harmonization of national standards, marine and coastal resource-protection laws, and resource-allocation litigation. The U.S. EPA has been active in applying and developing population-level methods for use in ecological risk assessments, e.g., the formation of the Ecological Committee on Federal Insecticide, Fungicide, and Rodenticide Act Risk Assessment Methods, with subsequent activities including the incorporation of probabilistic tools and methods into the development of models to evaluate the effects of pesticides to terrestrial and aquatic species (http://www.epa.gov/oppefedl/ecorisk/index.htm). The U.S. EPA scientists also were active in a Society of Environmental Toxicology and Chemistry workshop, held in August 2003, on the use of population-level methods in ecological risk assessment. In that workshop, a framework was developed to guide assessors in the use of population-level methods.
Predicting effects at higher ecological organization levels may be difficult. The use of population-level methodologies will include different parameters that will have different uncertainty issues than with organism-level assessments. Such uncertainty issues may include incomplete information on density-dependent effects, especially for long-term predictions (Barnthouse 1993), and nonlinearities resulting from qualitative changes in the causes and mechanisms of system response at different levels of organization (Munns 2002). As described by Munns and Mitro (forthcoming), “Most population-level attributes, including abundance attributes, are determined by the vital rates (births, deaths) of individuals within the population, as well as the rates of migration into and out of the population.” Effects on vital rates and on population dynamics can be linked by correlating responses between different levels of biological organization, or by mechanistic modeling of causal relationships (see Maltby et al. 2001 for a conceptualization and discussion of these linkages). For example, models that integrate the effects of stressors on survivorship and fecundity can extrapolate from organism-level effects to expected population-level responses (Munns and Mitro forthcoming). However, experience indicates that the many unverifiable assumptions involved in this approach may lead to very different outcomes from dueling models that cannot easily be resolved. Those assumptions, e.g., as regards the spatial distributions of receptor organisms, generally arise from an incomplete understanding of underlying biological processes. Field studies can generate data to parameterize population models, evaluate model outputs, or directly assess the effects of contaminants at a site (Munns and Mitro forthcoming). However, there are inherent difficulties in measuring characteristics of natural populations in the field as compared with laboratory settings, where it is possible to control the test population.
Possible probabilistic approaches
The U.S. EPA is examining opportunities to use probabilistic approaches in risk assessments, including ecological assessments, building on its “Policy for Use of Probabilistic Analysis in Risk Assessment” (USEPA 1997b, 1997c). Many of its stated principles for selecting input data and distributions, evaluating variability and uncertainty, and presenting results of the analysis are critical—particularly in light of technical improvements in probabilistic approaches, a proliferation of new methods, and enhancements to analyze and present results. The analysis must be planned with a risk management perspective in mind so that it supports the risk management decision process. It is recommended that EPA work to identify best practices for probabilistic risk assessment, to apply appropriate applications of the principles through case studies, and to communicate these practices and new developments to potential users.
EPA's experience using probabilistic risk assessment approaches has mostly been to characterize the prospective or retrospective distribution of stressors for registration, permitting, or remediation purposes. These approaches have also been applied to estimate residues and to characterize anticipated exposures when stressors are released. In some instances, effects models have been converted for probabilistic risk estimations; however, the models are very complex and the toxicity data bases are limited, so there are many challenges that need to be addressed in data extrapolation and model evaluation. One challenge is dealing with defaults, missing data, and varying time and spatial scales associated with input data. Fortunately, interest and effort in these approaches are increasing. For example, following a dialogue with internal and external experts in the field of ecological risk assessment and probabilistic methods, the Environmental Fate and Effects Division of the U.S. EPA's Office of Pesticide Programs is currently developing methodology to use probabilistic risk assessments in appropriate situations (e.g., see http://www.epa.gov/oppefedl/ecorisk/index.htm).
“Conservatism” in ecological risk assessments
An issue identified for risk assessments in general was the concern expressed in some of the comments submitted to USOMB that many estimates of risk may be considered overly conservative (i.e., overestimate risk). This concern is highlighted in the staff paper for ecological risk assessments, particularly as it relates to the development of toxicity reference values. In the early stages of a risk assessment (e.g., for screening purposes), a degree of conservatism is usually sought to ensure against an underestimation of risk. In many instances, these screening-level assessments are followed by a more detailed definitive assessment. Definitive assessments strive to be as realistic as possible, replacing “conservative” assumptions with best estimates of exposures, effects, and associated uncertainties. As an example, Ecological Soil Screening Levels, designed for screening purposes, are based on generic, relatively conservative toxicity reference values that are not site specific (USEPA 2003b). In a definitive assessment, the use of site-specific information instead of generic default values to derive appropriate exposure concentrations is encouraged. For example, the degree to which spatial and temporal use of habitats by receptor species should be considered is an important question. To elaborate, if we know that fish move throughout (over different habitats of) a distinct management area, exposure concentrations should be derived using data from the entire area. In the case of a large river system with tributaries that may be relatively clean compared with the main channel, the decision as to whether or not to include tributaries in the assessment typically depends on how contaminated they are and how they are used by the receptor species.
This paper examines some ecological risk assessment issues that were the subject of the U.S. EPA's examination of its risk-assessment principles and practices and of public comment. As stated in the staff paper (USEPA 2004), the U.S. EPA “attempts to use all information available in an objective, realistic, scientifically balanced way to make decisions.” Nonetheless, these are complex issues, and opportunities to improve ecological risk assessment practices are welcome. With this in mind, further discussion with the scientific community is encouraged on these and other ecological risk assessment issues.
The authors wish to thank Wayne Munns and Barbara Klieforth for reviewing and providing comments on this manuscript. The manuscript also benefited from discussions with Glenn Suter and Michael Griffith.
Disclaimer—This manuscript has been reviewed in accordance with U.S. EPA policy and approved for publication and distribution. The views presented are the authors' opinions and do not constitute U.S. EPA policy. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.