est. Strang and Sage demonstrate how such measurements have helped to document improvements in the conditions of the rivers they studied.
Recently, the concepts of indicator species and species diversity have been elaborated with the development of so-called multimetric biotic indices. These measures assess the overall condition of an ecosystem through simultaneous use of a variety of metrics. One such index is Karr's (1981) IBI. Karr (1992) argues that the multivariate nature of natural systems dictates that effective measures of ecosystem condition be based upon a variety of relevant biological attributes but that, to be usable, a comprehensive measure cannot require data for an endless number of system properties. The IBI "represents a synthesis of a dozen distinct hypotheses about the relationship between attributes of biological systems under varying influence from human society" (Karr, 1992, p. 233). The IBI is now widely used in North America and Europe to assess the condition of stream fish communities. Karr's original IBI has been modified to apply the same approach to other organisms, such as stream invertebrates. Carriker (this volume) and Yoder and Rankin apply Karr's approach to assess the conditions of streams and reservoirs managed by the Tennessee Valley Authority and the Ohio Environmental Protection Agency.
Any efforts to limit human environmental impacts should include two goals. The first should be to minimize the undesirable environmental impact per unit of human activity. The second should be to ensure that the cumulative impact of all human activities is compatible with the persistence of all critical ecosystem conditions and processes. Profound uncertainties will complicate efforts to achieve these goals, but they are appropriate targets.
Environmental performance measures are key tools in identifying opportunities to move toward the first of these goals. Measures of ecosystem condition are vital in charting progress toward the second goal. However, these two sets of tools will be most useful if they can be used to accurately predict the consequences of human activity on affected ecosystems. To achieve this, they must be refined and coordinated such that predictions can be based on the product of the environmental impact per unit of an activity multiplied by the scale of that activity. Such information would be useful for distinguishing acceptable and unacceptable impacts. Users of performance and condition measures should examine the potential for finding or developing relationships between the two types of indicators. Ideally, it will eventually become possible to measure environmental performance in terms that can be related directly to the consequences for affected ecosystems. In essence then, the environmental impact of a given activity could