This study describes the benthic invertebrate communities of 20 benchmark streams in agricultural areas of eastern Wisconsin. Streams with minimal adverse effects from human activity were selected from four agricultural areas with differing surficial deposits and bedrock types (relatively homogeneous units, or RHU's). Most aquatic invertebrate orders were well represented in the 20 benchmark stream samples; 217 species and 151 genera within 56 families were identified. Diptera was the best represented order (96 species), followed by Trichoptera (42 species) and Ephemeroptera (26 species). Diptera were the most abundant organisms in terms of numbers of individuals in the sample (28 percent of the total) followed by Trichoptera (25 percent) and Ephemeroptera (13 percent). Nine species of freshwater mussels were found, but only in 5 of the 20 benchmark streams.

Community measures were calculated for the following: total number of individuals; number of species; number of families; Margalef's diversity index; percent dominant family; percent Ephemeroptera-Plecoptera-Trichoptera (EPT); ratio of EPT to Chironomidae; percent shredders; ratio of scrapers to collectors-gatherers-filterers; Hilsenhoff's Biotic Index; Hilsenhoff's family level biotic index; and mean tolerance value. The S AS statistical software package was used for calculations of variance and correlations, normality checks, and principal components analysis of these measures and to find relations between benthic-invertebrate data and environmental-setting, habitat, and water-quality data.

Coefficients of variation within the RHU's were as great or greater than those for all 20 streams for most measures and RHU's. The specific taxa assemblages present at the sites did not show distinct differences between RHU's or similarities within the RHU's. The covariance and the Kruskal-Wallis tests showed that the benthic invertebrate measures were not related to RHU. These results all indicate that the combined effect of the RHU variables (bedrock geology, texture of surficial deposits, and land use/land cover) were not elemental in describing invertebrate communities in the study-area streams.

A principal components analysis (PCA) was done on the 20 benchmark streams which used the invertebrate population measures as variables. A three-dimensional ordination plot of these components revealed that 18 of the 20 streams could be divided into three groups relative to stream size, available habitat, and water quality. The three classifications of streams include large, warmer streams with slight pollution; deep, mixed-water streams with minimal pollution; and small, cold, pristine headwater streams. The two streams not defined by the three PCA groupings were not suitable to represent benchmark conditions. One site lacked suitable quality habitat or sufficient nutrients to support a healthy population of invertebrates, causing low measures of diversity. The other site appeared to be affected by sedimentation and low flows.

The classification groupings did not show any significant relations to percentage agricultural land use. Percentage of agricultural land use varied greatly within each group and the means for each group were similar. All streams in this study had some level of protection from agricultural practices in their basins. Although the intensity of agriculture is known to be a factor causing deterioration of invertebrate populations in past studies, the finding in this study indicated that the level of protection the stream received and other factors such as environmental setting and habitat could be more important to benthic invertebrates than the percentage of agriculture in the basin.

Information gathered from these benchmark streams can be used as a regional reference for comparison with other streams in agricultural areas, based on communities of aquatic biota, habitat, and water quality.