Land use versus instream habitat: the response of invertebrate metrics

M.G.S. Wood and J.D. Allan. School of Natural Resources and Environment, University of Michigan, Ann Arbor, MI 48109-1115 USA

Land cover, in-stream habitat and underlying natural variation potentially influence macroinvertebrate assemblages. We assessed the utility of these different groups of variables in identifying principal factors influencing metrics of aquatic invertebrate communities. Focusing on two catchments of southeast Michigan, 46 sites were sampled for invertebrates in the spring and fall during 1999-2001 using 250 μm D-nets. Regression analyses using all environmental factors generally attributed variation in invertebrate measures to habitat or land cover. However, a tiered regression approach, which tested sets of variables in a pre-determined order, indicated that surficial geology accounted for 4 to 31% of the variance when tested first. Addition of measures of land cover usually improved model fit, indicating that anthropogenic land use was an important predictor of invertebrate metrics. Percent land cover in riparian buffers proved to be superior to sub-catchment land cover as well as landscape pattern metrics. However, measures of instream habitat explained a substantial proportion of variance not explained by land cover metrics. In summary, a number of richness metrics were highly responsive to human activity, particularly to declines in natural land cover and stream habitat quality. Furthermore, tiered regression analyses helped to determine the utility of different sets of environmental variables.