Focus on Exposure-Response Relationships, and Complex Forms will come Naturally

Glenn W. Suter II, Ph.D.

National Center for Environmental Assessment

U.S. Environmental Protection Agency

26 W. Martin L. King Dr., MC-117

Cincinnati, OH 45268 USA

Tel: 513-569-7808

Fax: 513-569-7475


In his thoughtful explication of the implications of hormesis in ecotoxicology and ecological risk assessment, Dr. Chapman attributes the neglect of this phenomenon to the lack of familiarity with the phenomenon, the lack of low exposure levels in tests, the added cost of testing for hormetic effects, and the lack of appropriate statistics. While all of these reasons are real enough, I believe that they are less important than a more fundamental problem. Ecotoxicologists and ecological risk assessors have not, in general, been concerned with exposure-response relationships. Instead conventional practice has focused on defining regulatory bright lines, safe levels, threshold values, etc. Few publications in ecotoxicology examine the form of the exposure-response relationship, and few ecological risk assessments do more than estimate the risk of exceeding a threshold value. Until we routinely base our toxicology and assessment on exposure-response relationships, we will not notice or care about hormesis or other peculiarities in the form of such relationships.

Given this premise, the problem is not simply to accommodate hormesis, but to accommodate the estimation of effects through exposure-response modeling. The statistical and logical flaws in the substitution of statistical significance thresholds for effects levels have been well described if not well appreciated by practitioners (Suter 1996). However, the inadequacy of toxicological thresholds as bases for decision making is less well known. In particular, thresholds can not serve to indicate the consequences of actions relative to alternative actions or inactions. Are effects of soil contaminants sufficient to justify the damage done by remedial actions? Are short-lived pesticides with severe acute effects better than more persistent pesticides with less toxicity? Are risks from pesticides worse than risks of converting more land to tilled agriculture? These sorts of questions cannot be properly answered without estimating effects. Even if risk-risk comparisons are not made, most regulatory or remedial actions involve some comparison of costs and benefits. These comparisons also require estimation of effects. Any estimation of effects requires a model of the relation of response to exposure.

Once risk assessment and management are based on risks of effects rather than risks of exceeding a threshold, the door is opened to serious consideration of the form of exposure-response models. For ecological risks, this immediately leads to the use of convex exposure-response models (i.e., stimulatory effects prior to toxic effects as exposure increases). That is because, as Dr. Chapman points out, major sources of ecological risk include nutrients and habitat disturbances which are beneficial at low levels. This phenomenon was presented as a law of ecology, the subsidy-stress gradient, by Eugene Odum (1971). Hormesis differs only in terms of its degree of acceptance by the scientific community which is low due to poorly defined mechanisms. That is not a trivial problem, but it is a technical problem and therefore solvable through research. Hence, in ecological risk assessment, toxicological hormesis could be simply one of a number of phenomena that make convex exposure-response curves commonplace.

Adoption of this view could allow ecological risk assessors to avoid sterile arguments about the definition of hormesis and its reality as a general phenomenon. Rather we can focus on two real technical problems. The first is defining exposure-response relationships within the range of real world exposures. The second is understanding the mechanisms underlying the observed functional forms. Only when convex exposure-response models are well defined and replicated and the mechanisms that cause the stimulation are as well demonstrated for chemical contaminants as for nutrients, fire, etc., will the ecological hormesis debate yield consensus. Those who believe that hormesis is real and important in ecotoxicology should begin to study chemicals that are representative of important mechanisms of action. If they demonstrate hormesis, I believe that they will find that ecological risk assessors and managers are more open to the phenomenon than human health risk assessors and managers.


Odum, E. P. 1971. Fundamentals of Ecology. W. B. Saunders Co., Philadelphia.

Suter, G. W., II. 1996. Abuse of hypothesis testing statistics in ecological risk assessment. Human and Ecological Risk Assessment 2:331-349.

1The views expressed in this paper are those of the author and do not necessarily reflect the views or policies of the U.S. Environmental Protection Agency.