David W. Blaha, Environmental Resources Management; Mary Dolan, M-NCPPC; Pam Rowe, MC DEP; Keith Van Ness, MC DEP; Carol Young, MC DEP; Kevin Groppe, ERM
Many studies have documented the effects of urbanization on the ecosystems of small streams and their associated riparian zones, but few have had a sufficiently large and accurate database to quantify the relationship between urbanization and stream quality as measured by fish and benthic Index of Biotic Integrity (IBI) scores. Montgomery County (MD) is located in the Piedmont province and in the rapidly growing Washington, D.C. metropolitan area. Using land use information and stream biological sampling data for 48 1st, 2nd, and a few small 3rd order streams, multivariate regression models were developed using benthic IBI, fish IBI, and a combined fish and benthic IBI score as dependent variables. Independent variables included impervious surfaces, various land use categories, road crossings of streams, wetlands, and forested riparian buffer. The strongest relationship was obtained using benthic IBI scores for first order streams. A strong relationship was also obtained using combined fish and benthic IBI scores. The impervious coverage of the watershed was the best predictor of stream quality, explaining 52 percent of the variability. This model will be used as a predictive tool for planning future development, contributing to effective planning for stream and riparian protection and prediction of cumulative impacts.
KEY TERMS AND PHRASES: watersheds, Index of Biotic Integrity, urbanization, cumulative impacts, impervious surfaces, wetlands, predictive model, Piedmont
In February 1998, the Maryland-National Capital Park and Planning Commission (M-NCPPC) initiated an update to the 1980 Potomac Subregion Master Plan. The Potomac Subregion encompasses approximately 66 square miles in southwest Montgomery County, Maryland. The Study Area comprises all or part of five watersheds that drain to the Potomac River.
At about the same time that the Potomac Subregion Master Plan process was beginning, the Montgomery County Department of Environmental Protection (DEP), in cooperation with M-NCPPC, completed the Countywide Stream Protection Strategy, or CSPS (February 1998), which was developed to provide an overall assessment of County stream conditions. The CSPS used comprehensive, countywide monitoring of biological life in streams (fish and benthic macroinvertebrates), in combination with stream channel habitat conditions and stream chemistry (e.g. dissolved oxygen, temperature, pH), to assess stream conditions (Van Ness et al, 1997). County staff used an Index of Biotic Integrity (IBI) approach (Karr et al, 1986), which is comprised of fish and benthic macroinvertebrates community metrics. These metrics are used to quantitatively compare a stream with the measured attributes of the biological community found in high quality reference streams. Based on these IBI scores, the CSPS rated the existing health of streams as Excellent, Good, Fair, and Poor, which is referred to as the stream's CSPS Rating.
In preparing information for the Potomac Subregion master planning process, the Montgomery County Planning Board recognized that existing stream quality and the density of development projected for this area would make an accurate analysis of the environmental effects of future development essential. Better tools for analyzing potential impact on stream quality were needed. The recent completion of the Countywide Stream Protection Strategy and an extensive Geographic Information System (GIS) database on land conditions (including paving, tree cover, wetlands, etc.) gave an opportunity to develop a more accurate tool. A statistical model relating stream conditions to land conditions could more accurately predict stream conditions that might result from the cumulative effects of different development scenarios.
As part of the CSPS, stream data was collected at over 300 monitoring stations. Six basic criteria were used to initially screen this data set for use in the Potomac Subregion model:
The final result of this screening process resulted in the selection of 48 monitoring stations and corresponding drainage areas for use in the statistical analysis. The data set was assessed for adequacy of sample size and normality of distributions. The screening process assured that complete information was available for the monitoring stations and that the stations are representative of the environmental and developmental characteristics of the Potomac Subregion.
The dependent variables that were analyzed include:
The selection of independent variables for analysis was primarily based on the technical literature that identified factors that affect stream health.
The data was analyzed using linear regression for the four dependent variables (CSPS Scores, Stream Rating, Benthic IBI, and Fish IBI) using all independent variables. The linear regression analysis was performed using both a program within MS-Excel and SAS. A separate K-cluster analysis was performed to better assess the relationship between stream condition and woodlands in the absence of a clear relationship using a simple least-square analysis on the entire data set.
A total of 12 different regression analyses were conducted. For each of the 4 dependent variables (CSPS Scores, Stream Rating, Benthic IBI, and Fish IBI), separate analyses were performed for all streams (n=48 monitoring stations), just 1st order streams (n=16 monitoring stations), and just 2nd/3rd order streams (n=32 monitoring stations). Initial regression analysis indicated that separating the data set by stream order strengthened some of the relationships. Since impervious coverage was such an important variable, explaining at least 52 percent of the variation in dependent variables, various iterations of this variable were tested using linear, natural log, and exponential relationships to see if they would reveal more about the relationship of this important variable to stream condition and to other variables, such as wetlands and riparian buffer quality.
In general, the strongest relationships were obtained for the Benthic IBI dependent variable, including 1st Order Streams (r2=0.88), 2nd Order Streams (r2=0.81), and for all streams (r2=0.71). These correlation coefficients are much stronger than the Fish IBI values for 1st Order Streams (r2= 0.62), 2nd Order Streams (r2=0.39), and all streams (r2=0.39). This difference is attributable to several possible factors. First, benthic invertebrates are less mobile than fish and are unable to move in response to habitat degradation. Fish, on the other hand, can temporarily leave a stream or can satisfy various life stage requirements elsewhere. Second, habitat impacts, especially sedimentation, affect benthic macroinvertebrates life requisites more directly and immediately, whereas fish are affected more indirectly or seasonally through feeding and reproductive mechanisms (Berkman et al., 1986). As a result, fish communities may have a lag or delay factor in reflecting habitat degradation.
The CSPS Score and Stream Rating dependent variables were both based on Fish and Benthic IBI scores. The CSPS Score dependent variable performed better than the Stream Rating for 1st order streams and similarly for 2nd order and all streams. The Stream Rating generalizes these IBI scores into four ordinal groups (i.e. excellent, good, fair, and poor), whereas the CSPS Scores maintain the more robust semi-continuous values (i.e. scores possibly ranging from 20 - 100).
It appears that benthic invertebrates in 1st order streams are affected more by localized surface and subsurface hydrology (imperviousness and wetlands) and inputs of sediment (from cropland) and nutrients (from cropland and septic systems). Benthic invertebrates in larger 2nd and 3rd order streams are more affected by in-stream processes and channel hydraulics, which are indirectly affected by imperviousness.
This approach of developing statistical models that relate stream conditions with land use variables can be an effective tool in developing sound land use plans that protect streams and associated riparian areas.
Imperviousness was by far the best predictor of stream conditions. It explained between 52 to 66 percent of the variability within the data set and was included in nearly all the regression models. At the extremes (below 6 percent and over approximately 28 percent), imperviousness strongly influences stream conditions, overwhelming the effects of other variables. At the extreme high end of imperviousness, it is possible to partially overcome the effects of imperviousness through aggressive restoration strategies. The use of extraordinary Best Management Practices (BMPs) is also expected to partially offset the effects of very high imperviousness, and data is currently being collected to analyze the effects of extra BMPs.
The number of road crossings per acre of drainage area was a significant factor for explaining variability of data. It negatively correlates with stream conditions. May (1998) also identified road crossings of streams as the potentially most damaging fragmentation of riparian corridors. Road crossings not only fragment the riparian corridor from a habitat perspective, they function as a direct conduit for stormwater runoff to the stream, which carries a wide variety of pollutants such as oils, grease, heavy metals, and road salt.
The percentage cropland in the drainage area inversely correlated with stream condition. As the percentage of cropland increases, the stream conditions decrease. These results are generally consistent with other research. For example, Rabeni, and Boyle (1986) found benthic invertebrate correlated closer than fish with stream quality in agriculture areas of Missouri. This has been attributed to the more direct and immediate impacts of sediment on benthic invertebrate habitat requirements. Many collector-filterers lose their attachment to the substrate as sedimentation increases.
Consideration of SWM and forested riparian buffers aids considerably in understanding stream conditions. There is clearly a dynamic relationship among these three variables, especially for moderate levels of imperviousness ranging from 12 to 28 percent, where SWM and forested riparian buffers can offset to some extent the adverse effects of increasing imperviousness.
Wetlands, and specifically palustrine emergent wetlands, were positively correlated with stream condition and were a significant factor in explaining the variability within the data set for the CSPS, Benthic IBI, and Fish IBI regression analyses. Wetlands were especially significant for 1st order streams where they were the third best predictor of stream conditions (behind imperviousness and septic systems) and for Fish IBI 2nd/3rd Order and All Streams.
Other studies (May et al., 1997) have also identified the importance of riparian wetlands in maintaining Benthic IBI scores. These wetlands may function to modify hydrologic changes resulting from urbanization, stabilize stream banks or trap sediment that adversely affect aquatic organisms, produce organic material/detritus that serves as a food source for aquatic insects and fish, and provide a more diverse habitat.
The fact that wetlands were more strongly correlated with stream condition in 1st order rather than 2nd order streams also makes sense in that their relative functional contribution is greater in smaller streams where there is a greater stream/wetland interface. The relatively strong correlation with Fish IBI in large streams may reflect their habitat value.
These results indicate a clear correlation between increasing percent woodlands and improving CSPS Scores up to approximately 30 percent woodlands. For streams with more than 30 percent woodlands in their drainage area, a direct correlation with CSPS Scores is not as clear. It should be noted, however, that, at least for this data base, all streams with over 35 percent wooded drainage areas were in good or excellent condition and other factors appear to more strongly influence CSPS Scores.
The density of septic systems within the drainage area was positively correlated with stream conditions, especially for 1st order streams where it was the second best predictor of stream condition behind imperviousness for CSPS, Stream Rating, and Benthic IBI. While several studies have indicated nutrient enrichment can result in increases in fish biomass and the number of fish collected, there may be a threshold effect where direct nutrient enrichment can adversely affect aquatic communities. A second possible explanation is that septic systems are common in rural areas and are simply correlated with low impervious surfaces. More study of in this area is needed.
Arnold, Jr., and C. L., C. J. Gibbons, 1996. Impervious Surface Coverage, The Emergence of a Key Environmental Indicator. APA Journal, Spring 1996, pp. 243 - 258.
Berkman, H. E., C. F. Rabeni, and T. P. Boyle, 1986. Biomonitors of Stream Quality in Agricultural Areas: Fish versus Invertebrates. Environmental Management, Vol. 10, No. 3, pp. 413 - 419.
Karr, J. R., K. D., Fausch, Pl L. Angermeier, P. R. Yant, and I. J., Schlosser, 1986. Assessing Biological Integrity in Running Waters, A Method and Its Rational. Illinois Natural History Survey, Special Publication 5.
Klein, R., 1979. Urbanization and Stream Quality Impairment. American Water Resources Association. Water Resources Bulletin. 15(4).
May, C. W., 1999. The Cumulative Effects of Urbanization on Small Stream Watersheds in the Puget Sound Lowland Ecoregion, as presented at the symposium Watershed Management: Moving From Theory to Implementation, in Denver Colorado, Water Environment Federation, May 3 through 6, 1998.
May, C. W., R. R. Horner, J. R. Karr, B. W. Mar, and E. B Welch, 1997. Effects of Urbanization on Small Streams in the Puget Sound Lowland Ecoregion. Watershed Protection Techniques, Vol. 2, No. 4, June 1997, pp. 483 - 494.
Montgomery County Department of Environmental Protection in Cooperation with the Maryland-National Capital Park and Planning Commission, 1998. Montgomery County Countywide Stream Protection Strategy. February 1998.
Schueler, T., 1994. The Importance of Imperviousness. Watershed Protection Techniques, Vol 1, No. 3, Fall 1994, pp. 100 - 111.
Van Ness, K., K. Brown, M. Haddaway, D. Marshall, D. Jordahl, 1997. Montgomery County Water Quality Monitoring Program; Stream Monitoring Protocols. Watershed Management Division, Montgomery County Department of Environmental Protection.