1986), whereas recovery from the effects of a catchment-wide fire or a massive dam failure takes tens to hundreds of years (Minshall et al., 1983, 1989; Richards and Minshall, 1992). The demise of salmonid populations in the Pacific Northwest, referred to earlier, is symptomatic of a region- or basinwide loss of ecological integrity within the present century.

Variations in activity, condition, distribution, and abundance (hence, recruitment and/or mortality) of aquatic organisms across seasons are common. This is to be expected in strongly seasonal (temperate) environments, but such variations are found even in the tropics (Covich, 1988). Temporal variation in biotic responses should be accounted for in biological assessments of environmental conditions or determinations of change, but frequently it is not. Consideration of seasonal differences is especially important when comparing data from different locations or for the same area over time. For assessment of long-term trends, samples must be collected at the same general time within a season. However, in comparative studies among sites or years, sampling times should be determined on the basis of cumulative temperatures (e.g., number of degree-days) rather than specific calendar date. Temperature often is the primary factor responsible for seasonal variations in the temperate parts of the world. Even when light is the primary factor responsible, temperature is a reasonable surrogate because both are strongly influenced by the amount of solar radiation reaching the surface of a water body. This implies the need for continuous long-term records of temperature from the waters in question and/or the development of reliable regressions between air temperature or solar radiation and water temperature for specific localities and habitats of interest.

Recently, there has been a paradigm shift from a belief in the dominance of equilibrium processes in ecology to one that emphasizes the importance of nonequilibrium processes (e.g., Harris, 1986; Botkin, 1990; Reice, 1994). Previously, the dynamics within ecological levels of organization, from populations through ecosystems, were viewed as being controlled primarily by processes that were density dependent and tended toward equilibrium conditions. The present view, whose implications have yet to be fully appreciated by most ecologists and resource managers, is that these same dynamics are controlled largely by processes that are density independent and of a nonequilibrium type. Consequently, they are believed to be heavily dependent on random (stochastic) forces and hence to be disturbance driven and variable rather than constant. Though reality probably lies somewhere between these two extremes, the latter currently