ABSTRACT
This chapter suggests that correlations of laboratory toxicity and field impairment are necessary, but are not sufficiently rigorous tests of validity for predictive methods. The Artificial Substrate Microcosm (AS-M) is similar in both approach and physical and temporal scale to conventional toxicity tests, but the community-level endpoints monitored in AS-M is more comparable to those used to evaluate natural systems. In predictive applications, the validity of a toxicity test depends on the ability to use test results to predict environmental outcome in natural systems with a degree of precision adequate to management needs. Misinformation about cost, replicability, standardization, and the uncharismatic nature of microbes remain as obstacles to microscale community-level toxicity tests. The extrapolation approach based on tolerance distributions is much more appealing than the use of arbitrary application factors, but it has a number of uncertainties and assumptions, both statistical and ecological.
