Variability in stream chemistry in relation to urban development and biological condition in seven metropolitan areas of the United States, 1999-2004

Abstract

Beginning in 1999, the U.S. Geological Survey National Water Quality Assessment Program investigated the effects of urban development on stream ecosystems in nine metropolitan study areas across the United States. In seven of these study areas, stream-chemistry samples were collected every other month for 1 year at 6 to 10 sites. Within a study area, the sites collectively represented a gradient of urban development from minimally to highly developed watersheds, based on the percentage of urban land cover; depending on study area, the land cover before urban development was either forested or agricultural. The stream-chemistry factors measured in the samples were total nitrogen, total phosphorus, chloride, and pesticide toxicity. These data were used to characterize the stream-chemistry factors in four ways (hereafter referred to as characterizations)—seasonal high-flow value, seasonal low-flow value, the median value (representing a single integrated value of the factor over the year), and the standard deviation of values (representing the variation of the factor over the year). Aquatic macroinvertebrate communities were sampled at each site to infer the biological condition of the stream based on the relative sensitivity of the community to environmental stressors. A Spearman correlation analysis was used to evaluate relations between (1) urban development and each characterization of the stream-chemistry factors and (2) the biological condition of a stream and the different characterizations of chloride and pesticide toxicity. Overall, the study areas where the land cover before urban development was primarily forested had a greater number of moderate and strong relations compared with the study areas where the land cover before urban development was primarily agriculture; this was true when urban development was correlated with the stream-chemistry factors (except chloride) and when chloride and pesticide toxicity was correlated with the biological condition. Except for primarily phosphorus in two study areas, stream-chemistry factors generally increased with urban development, and among the different characterizations, the median value typically indicated the strongest relations. The variation in stream-chemistry factors throughout the year generally increased with urban development, indicating that water quality became less consistent as watersheds were developed. In study areas with high annual snow fall, the variation in chloride concentrations throughout the year was particularly strongly related to urban development, likely a result of road salt applications during the winter. The relations of the biological condition to chloride and pesticide toxicity were calculated irrespective of urban development, but the overall results indicated that the relations were still stronger in the study areas that had been forested before urban development. The weaker relations in the study areas that had been agricultural before urban development were likely the results of biological communities having been degraded from agricultural practices in the watersheds. Collectively, these results indicated that, compared with sampling a stream at a single point in time, sampling at regular intervals during a year may provide a more representative measure of water quality, especially in the areas of high urban development where water quality fluctuated more widely between samples. Furthermore, the use of "integrated" values of stream chemistry factors may be more appropriate when assessing relations to the biological condition of a stream because the taxa composition of a biological community typically reflects the water-quality conditions over time.