Development of Macroinvertebrate Biocriteria for Streams of Minnesota's Lake Superior Watershed

Abstract

Genus-level macroinvertebrate data from reference and disturbed streams were used to develop and test biocriteria for 1st - 3rd order streams in Minnesota’s Lake Superior Watershed (LSW). Fifteen metrics, most used elsewhere, were investigated for utility. Five metrics failed because of high correlation or inability to differentiate disturbed streams. Ten metrics were combined into a multimetric index. Metric values were scored relative to metric biocriteria according to EPA protocols and summed for each stream The minimum reference stream score defined the index biocriterion. Eleven urban or agricultural/rural stream scores were compared to the index biocriterion to test its ability to reveal impairment in test streams. Reference and disturbed stream scores were statistically different (p < 0.01). Urban streams were better separated from the reference condition (p < 0.001) than Ag/Rural streams (p < 0.01). The index biocriterion detected impairment in eight of the disturbed streams, while two streams scored within the safety buffer where a judgement of impairment was uncertain but possible. One stream scored slightly above the biocriterion. Several metrics which were useful elsewhere were also effective here, while others were not. The locally-tailored multimetric index and associated biocriteria developed here were effective in assessing stream ecosystem health in appropriate areas of the LSW. GIS analysis of subwatershed land use/cover showed that even moderate percentages of developed + hay/pasture/grass land covers (12-15%) and developed + hay/pasture/grass + roads (15-17%) resulted in some streams scoring as impaired. Therefore, LSW streams appear to be relatively fragile environments requiring careful watershed management. Methodologies involving chironomid inclusion, sample processing, sample size, and reach location were investigated. Chironomid abundance varied among Reference streams by more than an order of magnitude, though % chironomids varied less. Thus, inclusion of family-level chironomid data may increase variability in metrics involving abundance. A high percentage of taxa were “rare” at levels of <2 and <4 individuals per replicate; 34.0 and 48.0 % respectively for Reference streams and 44.2 and 61.2 for Disturbed streams. Thus, subsampling may strongly influence richness metric values, suggesting whole LSW samples should be processed for maximum assessment effectiveness. A richness/area curve showed continual increase through 5 replicates, thus, collecting fewer samples may reduce biocriteria effectiveness in the LSW if samples are composited. Metric scores were often significantly different in the three Kimball Cr. sample reaches, thus reach location may be an important consideration in developing biocriteria in the LSW.