Methods and models for single stressors are combined and refined to diagnose multiple sources 
of impairment. The latter stage includes the development of tools both to allocate cause among 
multiple additive stressors and to diagnose significant interactive effects among stressors. The 
first stage in diagnosing significant stressor interactions will involve the use of generic ecosystem 
models to perform sensitivity analyses to determine the probability of observing significant 
interactions among stressor classes over realistic ranges of loadings or stressor levels (e.g., 
Bartell et al. 1984, Mitsch and Reeder 1991, Hanratty and Stay 1994, EPA 2000b). Ultimately 
these tools will be incorporated into a decision-support system. 
Step 4. Develop forecasting approaches. 
This step builds upon the development of multi-stressor methods and models to include 
forecasting techniques to project the response of aquatic ecosystems to load reductions and/or 
watershed restoration activities into the future. Forecasting methods will be particularly 
important in protecting large, complex, unique resources (e.g., Great Lakes, Gulf of Mexico, 
Chesapeake Bay) for which costs of restoration are large, interactions are involved, and lag times 
between an event and the eventual system response must be taken into consideration. 
Development of forecasting techniques will also allow NHEERL to be proactive in defining 
potential shifts in causes of impairment, and to anticipate future threats to the environment. 
Activities in this area will be coordinated with NRMRL. 
Outputs from this research path will require collaborative efforts with NERL to develop 
improved loading models for TMDLs that include components predicting biological respcmses, 
and the development of appropriate exposure metrics to improve monitoring designs. In 
addition, classification frameworks and other tools developed here will be coordinated with 
research on prioritization of watershed restoration activities, prediction of recovery paths, and 
assessment of the success of remediation actions currently under way within NRMRL. 
Methods and models developed under diagnostics research also will feed into the diagnostic logic 
flow sequence described in the EPA SI document for analysis of data for a weight-of-evidence 
approach (EPA 2000c). Potential points of influence on the SI overall process are presented in 
Figure 13. 
In Figure 14, the research products (APMs) from the critical path are connected to the 
State/Tribal implementation stages for both the TMDL and watershed restoration processes for 
impaired surface waters, and then merged into an integrated process. The first APG, 
development of a conceptual framework, feeds directly into the problem formulation aspect of 
the diagnostic process and provides the basis for developing a decision-support system for the 
diagnostic process. The second APG connects directly to diagnosing the primary cause of 
impairment. The third APG provides research products which support three areas of the 
combined TMDL/Watershed Restoration path: Diagnostic and Condition-based Monitoring and 
Assessment, Allocation of Causes and Interactions among Multiple Stressors, and Confirmation 
of Diagnosis with Uncertainty Evaluation. The fourth goal supports diagnostic model 
development for forecasting and predictive approaches for the evaluation of remediation options. 
143 
