We analyzed the relation of the amount and spatial pattern of land cover with stream fish communities, in-stream habitat, and baseflow in 47 small southeastern Wisconsin, USA, watersheds encompassing a gradient of predominantly agricultural to predominantly urban land uses. The amount of connected impervious surface in the watershed was the best measure of urbanization for predicting fish density, species richness, diversity, and index of biotic integrity (IBI) score; bank erosion; and base flow. However, connected imperviousness was not significantly correlated with overall habitat quality for fish. Nonlinear models were developed using quantile regression to predict the maximum possible number of fish species, IBI score, and base flow for a given level of imperviousness. At watershed connected imperviousness levels less than about 8%, all three variables could have high values, whereas at connected imperviousness levels greater than 12% their values were inevitably low. Connected imperviousness levels between 8 and 12% represented a threshold region where minor changes in urbanization could result in major changes in stream condition. In a spatial analysis, connected imperviousness within a 50-m buffer along the stream or within a 1.6-km radius upstream of the sampling site had more influence on stream fish and base flow than did comparable amounts of imperviousness further away. Our results suggest that urban development that minimizes amount of connected impervious surface and establishes undeveloped buffer areas along streams should have less impact than conventional types of development.

No abstract

We analyzed data from 79 watersheds in an undegraded U.S. ecoregion to identify key environmental factors that explained stream fish assemblage patterns and to evaluate the relative influence of environmental factors operating at different spatial scales. A few key factors from the watershed, reach, and riparian scale explained a significant amount of the variance in fish attributes. Three environment–fish associations were evident. Top carnivores and intolerant cold-water fishes were associated with relatively narrow, deep, high-gradient, cold streams with strong groundwater inputs. Tolerant cyprinids occurred in small streams with low groundwater input, low dissolved oxygen, and abundant macrophytes. A diverse assemblage (Cyprinidae, Catostomidae, Centrarchidae, Percidae) existed in warm, wide streams in watersheds dominated by lacustrine sand and gravel geology and abundant wetlands and lakes. Overall, reach-scale variables directly explained the most, watershed-scale variables less, and riparian-scale variables the least variation in fish attributes. Watershed and riparian variables had indirect connections with fishes through their direct influence on reach variables. In conjunction with findings from more degraded regions, we conclude that the relative influence of reach-scale variables on fishes are greatest in undegraded areas and that direct effects of watershed-scale variables are increasingly important as human modifications of the landscape increase.

The index of biotic integrity (IBI). developed from information on the structure, composition, and functional organization of fish assemblages, is used to assess the health of aquatic ecosystems. We analyzed two large statewide data sets on stream fish assemblages to develop and test a version of the IBI for application to Wisconsin coldwater streams (maximum daily mean water temperature usually <22°C). This new IBI is needed because fish assemblages in Wisconsin coldwater streams differ significantly from those in warmwater streams (maximum daily mean temperature >24°C), for which an IBI already exists. High‐quality coldwater streams have few species, with salmonids and cottids dominating. and lack many of the taxonomic groups that are important in high‐quality warmwater streams. In contrast, high‐quality warmwater streams have numerous species, and cyprinids, catostomids, centrarchids, and percids typically dominate. Environmental degradation often causes an increase in species richness in coldwater fish assemblages, the opposite of what occurs in warmwater assemblages, as a small number of coldwater species are replaced by a larger number of more tolerant eurythermal and warmwater species. The new coldwater IBI has five metrics: (1) number of intolerant species, (2) percent of all individuals that are tolerant species, (3) percent of all individuals that are top carnivore species, (4) percent of all individuals that are native or exotic stenothermal coldwater or coolwater species, and (5) percent of salmonid individuals that are brook trout Salvelinus fontinalis. No regional or stream‐size adjustments in metric scoring criteria are needed. Relative coldwater IBI scores and ratings of stream sites throughout Wisconsin closely match independent rankings of environmental quality on the basis of physical habitat and water quality of the sites. Variation in IBI scores within and among years is gencrally low. The new coldwater IBI is not appropriate for coolwater streams (typical maximum summer daily mean temperature 22–24°C).

We compared watershed land‐use and fish community data between the 1970s and 1990s in 47 small streams in southeastern Wisconsin. Our goal was to quantify effects of increasing urbanization on stream fishes in what had been a predominantly agricultural region. In the 43 test watersheds, mean surface coverage by agricultural lands decreased from 54 percent to 43 percent and urban lands increased from 24 percent to 31 percent between 1970 and 1990. Agriculture dominated the four reference watersheds, but neither agriculture (65–59 percent) nor urban (4.4–4.8 percent) land‐uses changed significantly in those watersheds during the study period. From the 1970s to the 1990s the mean number of fish species for the test stream sites decreased 15 percent, fish density decreased 41 percent, and the index of biotic integrity (IBI) score dropped 32 percent. Fish community attributes at the four reference sites did not change significantly during the same period, although density was substantially lower in the 1990s. For both the 1970s and 1990s test sites, numbers of fish species and IBI scores were positively correlated with watershed percent agricultural land coverage and negatively correlated with watershed urban land uses, as indexed by percent effective connected imperviousness. Numbers of fish species per site and IBI scores were highly variable below 10 percent imperviousness, but consistently low above 10 percent. Sites that had less than 10 percent imperviousness and fewer than 10 fish species in the 1970s suffered the greatest relative increase in imperviousness and decline in species number over the study period. Our findings are consistent with previous studies that have found strong negative effects of urban land uses on stream ecosystems and a threshold of environmental damage at about 10 percent imperviousness. We conclude that although agricultural land uses often degrade stream fish communities, agricultural land impacts are generally less severe than those from urbanization on a per‐unit‐area basis.