the Pacific Northwest as a result of impacts on tourism, fisheries, and so forth, they are not in a position to weigh the ethical values of conserving a species like the Northern Spotted Owl against various economic values associated with continued loss of its habitat.

One of the few generalizations that can be derived from the past two decades of ecological research is that few generalizations can be made about the dynamics of ecological systems. This conclusion represents a significant change from the beliefs of researchers in the 1950s and 1960s, when studies suggested that ecological systems functioned in a state of dynamic equilibrium and that communities were tightly co-evolved assemblages of species. Considerable effort was devoted to the elaboration of underlying rules that would transform ecology into a predictive science. Today, there are relatively few general theories that researchers would be willing to apply to communities or to ecological interactions in a predictive manner. This change has come about as scientists came to realize that most communities appeared to be non-equilibrium systems with non-linear dynamics (Botkin 1990, Buzas and Culver 1994).

With few predictive theories at hand, considerable emphasis must be placed on empirical studies of specific systems to develop the ability to assess the consequences of change in those systems. Our ability to evaluate the impacts of any given perturbation depends strongly on the availability of long-term data on the particular system and in particular on the availability of experimental information bearing on the change being assessed. Even where scientists can predict the direct consequences of a specific perturbation on certain species (for example, the impact of a pesticide on a particular species), we are unable to predict the secondary effects of those changes on the structure and function of the ecosystem without much more detailed information.

The status of our knowledge about the impacts of human-caused changes differs depending on the type of perturbation involved:

Our ability to predict the direct effect of a number of different classes of toxic chemicals on various wildlife populations is quite good. For many families of pesticides, for example, we have a good understanding of the mechanism by which the chemical works, and can predict which species might be affected by related chemicals and how they might be affected. Moreover, laboratory studies can often be used to identify particularly threatening chemicals relatively quickly and cheaply.

More worrisome, however, are chemicals that do not exhibit direct toxic effects on plants and animals but may have chronic effects. For example, a growing body of evidence suggests that a broad array of "estrogenic" chemicals may be influencing the reproductive physiology of many wildlife species (Colborn