Impact Assessment Using the Before-After-Control-Impact (BACI) Model: Concerns and Comments

The effect of a change in an ecosystem can often be assessed through the use of a statistical model that incorporates the change. A sensible approach for assessing the effects of an industrial or power plant on the aquatic environment is to sample the environment both before and after the plant starts operation and test for a change in some biologically relevant parameter. To improve sensitivity, samples may be taken at a control site as well as at sites receiving the plant effluent. While this provides a powerful means for assessing effects, the implementation of the design is important and subsequent analysis of the collected data depends on proper implementation. Problems such as trends in the measurements, failure to meet the assumptions of the model, irregular sampling, confounding factors, and changes in the habitat can influence results, as we illustrate using a long-term impact assessment of a power plant on fish populations. In long-term studies, it may be difficult to separate effects due to the plant from effects due to other sources. Sound design requires both a good statistical model and an understanding of the underlying biological processes (what to measure) and careful planning (how to measure it well).