mation on physical and chemical characteristics of water quality, biological community structure, and habitat. Of the many variables it collects, TVA has selected five indicators for evaluating reservoir health: dissolved oxygen, chlorophyll-a, sediment quality, benthic macroinvertebrates, and fish community. Stream evaluation is based on four aquatic ecosystem indicators: nutrient concentrations, sediment quality, benthic macroinvertebrates, and fish community.
Ecological health evaluations depend on the ability to discriminate between good and poor conditions for each indicator. This is more easily done for streams because they offer relatively unaltered reference sites that can be examined to define ''good" conditions for each indicator. For example, various indices of biotic integrity for fish and benthic stream communities compare results at monitoring locations with conditions at reference sites (Karr et al., 1986; Kerans et al., 1992). But reservoirs are man-made alterations of natural streams; thus, no "reference reservoirs" exist for comparison. They require an alternative approach to reference conditions.
Scoring criteria for dissolved oxygen and chlorophyll-a in reservoirs are based on a conceptual model that TVA developed over several years from its experience in evaluating biological systems in reservoirs. The model for dissolved oxygen criteria for a reservoir is complicated by the combined effects of flow regulation and potential oxygen depletion in the hypolimnion (deep water). TVA's scoring criteria consider dissolved oxygen levels both in the water column and near the bottom of the reservoir. For chlorophyll-a, TVA's experience is that, below a threshold level, primary production is not sufficient to support an active, biologically healthy food chain. However, chlorophyll-a concentrations above a higher threshold result in undesirable eutrophic conditions.
For the benthic macroinvertebrate and fish community indicators, TVA based its scoring criteria on a statistical examination of multiple years of data from TVA reservoirs. All previously collected TVA reservoir data for a characteristic of a selected community (e.g., number of taxa, total abundance) are ranked and divided into good, fair, and poor groupings. The current year's results are compared with these groupings and scored accordingly. This approach is valid if the database is sufficiently large and covers the full spectrum of good-to-poor conditions.
The sediment-quality scoring criteria use a combination of sediment toxicity to test organisms and sediment chemical analyses for ammonia, heavy metals, pesticides, and polychlorinated biphenyls (PCBs).