Causal Assessment: Tools, Techniques and Results
Barbara Washburn, California Office of Environmental Health Hazard Assessment
California is current evaluating the use of causal assessment to identify causes of biological impairment as part of its effort to establish biological objectives for wadeable streams, so the session at the SETAC North America 33rd Annual Meeting in Long Beach, California, session on tools, techniques and results of causal assessment was very timely.
Three of the talks presented case studies on the use of the United States Environmental Protection Agency’s (USEPA’s) stressor identification guidance. Stressor identification is a weight-of-evidence approach, based on epidemiological principles, for identifying causes of biological impairment. One of the strengths of this approach is that chemical, physical and biological stressors can be evaluated together. Criteria used to assess the role of various stressors include evaluating spatial and temporal co-occurrence (of the stressor and effect), existence of a causal pathway, a stressor-response relationship as well as others. These criteria are then used to determine if existing data strengthened or weakened the possibility that a potential cause might be responsible for impairment. The output from this process is a list of stressors most likely associated with the impairment. Data gaps can also be identified. The speakers addressed the application of the stressor identification methodology to impairments on streams in Northern and Southern California as well as the Central Coast.
Another complementary presentation described the use of Bayesian networks for causal assessment. Bayesian networks describe probabilistic relationships among variables of interest. They are composed of a series of nodes or variables, a set of links representing the causal relationships between the nodes (conditioning influences), and a set of probabilities and uncertainties, one for each node, that identify the belief that a node will be in a particular state given the state of the "parent" nodes. They are usually presented graphically.
Bayesian networks can be used to estimate the probability that a stressor causes an observed response. One example related to the assessment of possible adverse effects of pesticide applications on fish and insects. Probabilities were assigned to high, moderate or low rates of application, distance from the waterway, drop size distribution and wind speed. These estimates were then used to calculate the categorical probability of effects on fish and wildlife (i.e., probabilities of no effects, mild, moderate….extirpated).
Both the USEPA’s stressor identification guidance and the Bayesian network approach are suitable for actively engaging stakeholders to build and refine models and review results. Both approaches support simultaneous evaluation of multiple stressors (physical, chemical, landscape, etc.) and are designed to improve the decision-making process. Overall, this session provided a good overview and suggested significant practical application.
Author’s contact information: email@example.com
Return to the Globe