Progress Report of the Ecological Committee on FIFRA Risk Assessment Methods: V. Terrestrial Effects Assessment
On this Page
- Lead Collaborators and Areas of Concern
- Test Suitability
- Intra-Specific Extrapolations
- Inter-Specific Extrapolations
- Population-Level Extrapolations
- Risk Characterization
The goal of the Terrestrial Effects Group of ECOFRAM is to provide insight into a practical and scientifically sound approach for translating a probabilistic assessment of dose to a probabilistic assessment of ecological risk at the level of individuals and populations. We have approached this goal by identifying five areas of concern in which to focus attention (see Figure). In general, we wish to define a distribution of risk that explicitly considers magnitudes of effect and the likelihood or probability of those effects. As a consequence, we wish to quantify variability and uncertainty. Equally important, we wish to identify those sources of variability and uncertainty that have the greatest impact on final conclusions. Even in the absence of intensive or extensive data, sensitivity analysis can identify those parameters whose variability or uncertainty have a large effect on final assessments of risk.
The Ecological Committee on FIFRA Risk Assessment Methods (ECOFRAM) was formed in June 1997. The Committee's purpose is to develop tools and processes within the FIFRA framework for predicting the magnitude and probabilities of adverse effects to non- target aquatic and terrestrial species resulting from the introduction of pesticides into their environment. A Terrestrial Effects Subgroup was formed to identify and discuss probabilistic methods for terrestrial effects assessments and develop recommendations for future use by EPA. In addition, they will identify information that must be developed in order to validate the proposed methods and to ensure that the proposed assessment process, if adopted by EPA, supports environmental decisions that are scientifically defensible. A report of the Subgroup's deliberations to date will be presented.
Lead Collaborators and Areas of Concern
|Name and Association||Area of Concern|
|Richard Bennett, Ecological Planning & Toxicology||Test Sensitivity|
|Michael Hooper, Texas Tech University||Intra-specific Extrapolations|
|Alain Baril, Canadian Wildlife Service||Inter-specific Extrapolations|
|Thomas Lacher, Texas A&M University||Population-level Extrapolations|
|Jennifer Shaw, Zeneca Ag Products||Risk Characterization|
The suitability of tests is dependent on how exposure is characterized. Because the ECOFRAM is working on models to characterize exposure as dose received (i.e., mg/kg or mg/kg/day) rather than as bioavailability (e.g., ppm in food items), the suitability of toxicity tests must be evaluated for their potential to provide information on the toxicity of the chemical relative to dose.
Acute oral tests
LD5Os provide a measure of acute toxicity in units of dose (mg/kg) that is pertinent to situations where active ingredients are ingested rapidly.
Because the test is designed to define the dose lethal to 50% of the population, ECOFRAM will be evaluating the suitability of using dose-response relationships to estimate other endpoints (e.g., LD5 or LD1O). The ECOFRAM also will evaluate the suitability of using abbreviated "up-down" tests to supplement data from definite acute oral tests.
The LC50 estimates the dietary concentration (ppm) that is toxic during a 5-day exposure and at least 3-day post-treatment period. One disadvantage is that the endpoint is reported as concentration on food rather than as the dose ingested (i.e., mg/kg/day). Because the exposure period is fixed at five days, it also is limited to providing a measure of effect at this arbitrary exposure period without consideration of differences in the temporal pattern of effects from different chemistries. The ECOFRAM will evaluate if and how the test can provide information on effects in relation to ingested dose and consider alternative designs to provide toxicity data suitable for a probabilistic assessment.
Avian reproduction tests provide an estimate of the dietary concentration (ppm) at which statistically significant effects are detected for a suite of parental and reproductive parameters. Two disadvantages are that
- the test is not designed to determine dose-response relationships, and
- the endpoints are reported as concentration of food rather than as dose ingested.
The ECOFRAM will evaluate if and how the test can provide data suitable for estimating effects on reproduction in a probabilistic assessment.
Probabilistic distributions of toxicity are only as reliable as the data on which they are based. A number of factors, intrinsic to the species and toxicity measurement process, contribute variability to estimates of toxicity. Identification of the sources of this variability make possible better data for use in probabilistic assessments, and assist in the extrapolation process that is necessary for under-represented species. We are examining the sources and repercussions of this variability.
LD50s and LD5s
Slope values of LD50 are just as important as the point estimates. Frequently, LD5 values are used as conservative estimates in toxicity distributions. As LD5s generally are not the focus in standard dose-response testing, extrapolations using slopes from LD50 determinations supply the data. Concern over these extrapolations arises because of
- high variability in slope estimates between and within laboratories for even a single chemical, and
- the increasing size of confidence intervals at low levels of toxicity.
Further, investigations into sub-lethal effects show that their distributions are as variable as are those for lethality determinations (i.e., normal, bimodal, binomial, and poisson distributions)
Life Stage Sensitivity
Within a species, life stage plays an important role in the level of sensitivity. Younger birds and mammals are generally more sensitive to organophosphorus insecticides than are adults. Alternatively, toxicity to organochlorine insecticides tends to increase with age. Little work has examined the sensitivity or resistance of older age groups. Other considerations (e.g., breeding or migrational status) likely will affect toxicity and are being evaluated by the ECOFRAM. Understanding differences in sensitivity and its mechanistic basis is important; it facilitates adjustment of toxicity distributions and can take into account most-sensitive life stages. Similarly, extrapolation from tested to non-tested species must proceed carefully, giving consideration to the life stage of the organism used in laboratory tests as well as that of the species to which it is being extrapolated.
Considerable uncertainty may be associated with the process of extrapolating toxicity endpoints across species. The ECOFRAM is addressing a number of questions related to the sources of this uncertainty.
Variability in sensitivity across species
Historical data on pesticide toxicity to a variety of bird species indicate that species differ widely in sensitivity to a particular pesticide. This variability is itself, pesticide-specific. Historical information on avian LC50s and LD50s provide insight to the uncertainty associated with these endpoints when extrapolating to species which have not been tested.
The representativeness of surrogate test species for extrapolation to other taxa has been debated frequently. Again, historical data provide insights regarding this issue. The sensitivity of avian test species compared to other species has been examined by a number of authors. These studies indicate some patterns in sensitivity (e.g., blackbirds are more sensitive than gallinaceous birds), but there are many exceptions to the observed patterns which suggest the need for caution when making inter-specific extrapolation.
Data requirements and tiers
0ne way to reduce uncertainty is to test a sufficient number of species to establish the parameters of the distribution of sensitivities. This often is a log-normal distribution. However, this approach may be wasteful of resources and birds if used in testing pesticides that pose little risk to species of concern. A tiered approach to testing is likely to be most efficacious. More test data should be required when initial estimates of risk or the uncertainties are high, whereas few studies should be required when the initial estimates of risk are low.
Accuracy of extrapolations based on body weight
Current extrapolation of toxicity data is done using a weight-based approach (i.e., same LD50 value expressed in mg/kg body weight for a 1 kg and a 30g bird). Historical data on LD50 tests show that small birds are more sensitive to cholinesterase inhibiting pesticides than predicted by weight alone.
Lethal endpoints, such as the LD50 and LC50, tend to be described across species by a log-normal distribution. Our familiarity and experience with these endpoints suggest that this variability can be described or quantified. The situation with non-lethal endpoints (e.g., behavior) may be quite different, especially when dealing with threshold or all-or-none responses.
An objective of ECOFRAM is to develop population-level projections of ecological effects. Such effects traditionally are modeled by treating all individuals as genetically, morphologically, and physiologically equal. Nonetheless, different age groups, sexes, body size classes, and even individuals can react differently to exposure to a toxicant. The ECOFRAM approach to probabilistic risk assessment explores several approaches for modeling risk at the population-level.
Age class structured models
Grouping individuals by age and sex can provide much better estimates of demographic parameters; combining this approach with estimates of differential toxicity will greatly improve risk projections. Major challenges include collecting vital rates for age and sex classes for the wide array of species under risk.
Stage and size structured models
Individuals of different body sizes can be differentially susceptible to exposure to environmental contaminants. Obtaining information on vital rates for different size classes is difficult; even accurately obtained, standard values lack estimates of variability in natural environs. Moreover, incorporation of estimates of environmental stochasticity represents additional challenges.
Individual based models
Individual based models, or physiologically based models, focus on physiological differences among individuals in both exposure and response. These models have been applied almost exclusively in aquatic systems. Although, individual-based models hold great heuristic promise, they are data-intensive and require information on the physiology of organisms that is often unavailable.
Spatially structured populations
Environmental contaminants are distributed heterogeneously across the landscape. Agricultural fields will contain more pesticides, for example, than do surrounding woodlands. Consequently, it is critical to include information on the density and reproductive output of species in different habitats, as well as the different levels of exposure in these habitats. This information should be included as a variation of source-sink modeling.
Risk characterization is the final stage of risk assessment at which results of exposure and effects analyses are integrated to evaluate the likelihood of adverse ecological effects occurring following exposure to a stressor. The ecological significance of the adverse effects should be discussed, including consideration of the types and magnitudes of effects, their spatial and temporal patterns, and the likelihood of recovery.
In the FIFRA regulatory process, the quotient method has been used in risk assessment for pesticides. A quotient of single values for exposure and effects is given, and if the quotient is ≥ 1, an adverse effect is considered likely to occur. Limitations of this approach include:
- a lack of information on the magnitude or probability of adverse effects,
- as the quotient approaches 1, an increased dependence on expert judgment is required, and
- use of single points that represent the most sensitive or conservative data in the estimate; other available data are ignored usually.
On the other hand, a quotient is a simple and efficient means of identifying pesticides that are likely to be very safe in the environment. Therefore the method could remain a preliminary screening option. However, for decision making, risk managers may require more refined risk assessments that describe the probability and magnitude of adverse effects. A suite of methods (see Table) may be the most effective way to provide flexibility to manage a diversity of pesticide scenarios in which a refined risk assessment is necessary.
An essential element of the risk characterization stage will be to analyze and summarize uncertainties. These will include uncertainty associated with natural stochasticity, parameter error, and model error from exposure and effects analyses and the risk characterization. In addition to uncertainty, the risk characterization will provide a discussion of the ecological significance of effects with particular emphasis on the magnitude and spatio-temporal extent of population-level impacts.