Developing a statistical support system for environmental hazard evaluation

Estimating the hazard or risk to both human health and the environment has been based almost exclusively on single species toxicity tests low in environmental realism and without validation of their accuracy in more complex systems. While this may be quite appropriate for humans in a large variety of circumstances, there is no substantive body of direct experimental evidence indicating that precise predictions of harm from hazardous materials can be extrapolated from single species laboratory tests (or even multispecies laboratory tests) to the more complex highly variable natural systems. Now added to the hazardous chemical assessment problem is the accidental or deliberate release of genetically engineered microorganisms into the environment that have the additional capability of multiplying and expanding their numbers and also transferring genetic information to other organisms. This paper focuses entirely on hazard evaluation for organisms other than humans, namely predicting the potential risk or probability of harm to natural systems based on laboratory toxicity testing using single species. Not only will the basic risk assessment strategy itself be examined but also the question of determining the statistical reliability of various extrapolations from one level of biological organization to another. ‘For every complex problem, there is a simple, direct solution ... and it is invariably wrong!’ H. L. Mencken