Relation of urbanization to stream habitat and geomorphic characteristics in nine metropolitan areas of the United States

The relation of urbanization to stream habitat and geomorphic characteristics was examined collectively and individually for nine metropolitan areas of the United States?Portland, Oregon; Salt Lake City, Utah; Denver, Colorado; Dallas?Forth Worth, Texas; Milwaukee?Green Bay, Wisconsin; Birmingham, Alabama; Atlanta, Georgia; Raleigh, North Carolina; and Boston, Massachusetts. The study was part of a larger study conducted by the U.S. Geological Survey from 1999 to 2004 to examine the effects of urbanization on the physical, chemical, and biological components of stream ecosystems. The objectives of the current study were to determine how stream habitat and geomorphic characteristics relate to different aspects of urbanization across a variety of diverse environmental settings and spatial scales. A space-for-time rural-to-urban land-cover gradient approach was used. Reach-scale habitat data and geomorphic characteristic data were collected once during low flow and included indicators of potential habitat degradation such as measures of channel geometry and hydraulics, streambed substrate, low-flow reach volume (an estimate of base-flow conditions), habitat complexity, and riparian/bank conditions. Hydrologic metrics included in the analyses were those expected to be altered by increases in impervious surfaces, such as high-flow frequency and duration, flashiness, and low-flow duration. Other natural and human features, such as reach-scale channel engineering, geologic setting, and slope, were quantified to identify their possible confounding influences on habitat relations with watershed-scale urbanization indicators. Habitat and geomorphic characteristics were compared to several watershed-scale indicators of urbanization, natural landscape characteristics, and hydrologic metrics by use of correlation analyses and stepwise linear regression.

Habitat and geomorphic characteristics were related to percentages of impervious surfaces only in some metropolitan areas and environmental settings. The relations between watershed-scale indicators of urbanization and stream habitat depended on physiography and climate, hydrology, pre-urban channel alterations, reach-scale slope and presence of bedrock, and amount of bank stabilization and grade control. Channels increased in size with increasing percentages of impervious surfaces in southeastern and midwestern metropolitan areas regardless of whether the pre-existing land use was forest or agriculture. The amount of enlargement depended on annual precipitation and frequency of high-flow events. The lack of a relation between channel enlargement and increasing impervious surfaces in other metropolitan areas was thought to be confounded by pre-urbanization hydrologic and channel alterations. Direct relations of channel shape and streambed substrate to urbanization were variable or lacking, probably because the type, amount, and source of sediment are dependent on the phase of urbanization. Reach-scale slope also was important for determining variations in streambed substrate and habitat complexity (percentage of riffles and runs). Urbanization-associated changes in reach-scale riparian vegetation varied geographically, partially depending on pre-existing riparian vegetation characteristics. Bank erosion increased in Milwaukee?Green Bay and Boston urban streams, and bank erosion also increased with an increase in a streamflow flashiness index. However, potential relations likely were confounded by the frequent use of channel stabilization and bank protection in urban settings. Low-flow reach volume did not decrease with increasing urbanization, but instead was related to natural landscape characteristics and possibly other unmeasured factors. The presence of intermittent bedrock in some sampled reaches likely limited some geomorphic responses to urbanization, such as channel bed erosion. Results from this study emphasize the importance of including a wide range of landscape variables at m