Distinguishing natural from anthropogenic regulation of algae in aquatic ecosystems

            Assessments of human effects on ecological systems can be improved greatly by accounting for natural variability in conditions.  In this study, assessments of human effects on diatom species composition and inferred nutrient conditions were distinguished from natural variation in 119 streams throughout the lower peninsula of Michigan to determine sensitivity of assessments to statistical approach.  Diatom assemblages, water chemistry, and map-based land use land cover attributes were determine for each stream.  Diatom species composition was characterized with a trophic diatom index, which also inferred nutrient concentrations.  The trophic diatom index was based on nutrient requirement traits and the relative abundances of each species observed in a stream.  The trophic diatom indices in the 119 streams were then related to 37 climatic, hydrologic, soils, and land use variables with three statistical approaches to develop a model that would predict the natural trophic diatom index in naturally different stream types.  One approach was traditional forward stepwise regression and the second was forward stepwise regression of residuals from preceding models.  The third approach was classification and regression tree.  All approaches indicated the importance of riparian land use and some natural factor related to hydrologic conductivity (e.g. percent coarse soils or a Darcy groundwater indicator) for predicting the trophic diatom index for a stream.  Effects of natural factors and anthropogenic activities on the diatom index were approximately equal in this region.  Deviations in the trophic diatom index from expected natural condition were calculated for each stream and then related to likely effects on invertebrates and fish assemblages that would be caused by anthropogenically elevated nutrient conditions.  Deviations in results among these analyses indicated the importance of scientific interpretation as well as political and social factors in their application by government regulators and environmental managers.