Regular issues
Special Issues

Northeastern Naturalist
    NENA Home
    Range and Scope
    Board of Editors
    Editorial Workflow
    Publication Charges

Other EH Journals
    Southeastern Naturalist
    Caribbean Naturalist
    Neotropical Naturalist
    Urban Naturalist
    Eastern Paleontologist
    Journal of the North Atlantic
    Eastern Biologist

EH Natural History Home

Judging a Brook by its Cover: The Relation Between Ecological Condition of a Stream and Urban Land Cover in New England
James F. Coles, Thomas F. Cuffney, Gerard McMahon, and Cornell J. Rosiu

Northeastern Naturalist, Volume 17, Issue 1 (2010): 29–48

Full-text pdf (Accessible only to subscribers.To subscribe click here.)


Site by Bennett Web & Design Co.
2010 NORTHEASTERN NATURALIST 17(1):29–48 Judging a Brook by its Cover: The Relation Between Ecological Condition of a Stream and Urban Land Cover in New England James F. Coles1,*, Thomas F. Cuffney2, Gerard McMahon2, and Cornell J. Rosiu3 Abstract - The US Geological Survey conducted an urban land-use study in the New England Coastal Basins (NECB) area during 2001 to determine how urbanization relates to changes in the ecological condition of streams. Thirty sites were selected that differed in their level of watershed development (low to high). An urban intensity value was calculated for each site from 24 landscape variables. Together, these 30 values reppresented a gradient of urban intensity. Among various biological, chemical, and physical factors surveyed at each site, benthic invertebrate assemblages were sampled from stream riffles and also from multiple habitats along the length of the sampling reach. We use some of the NECB data to derive a four-variable urbanintensity index (NECB-UII), where each variable represents a distinct component of urbanization: increasing human presence, expanding infrastructure, landscape development, and riparian vegetation loss. Using the NECB-UII as a characterization of urbanization, we describe how landscape fragmentation occurs with urbanization and how changes in the invertebrate assemblages, represented by metrics of ecological condition, are related to urbanization. Metrics with a strong linear response included EPT taxa richness, percentage richness of non-insect taxa, and pollution-tolerance values. Additionally, we describe how these relations can help in estimating the expected condition of a stream for its level of urbanization, thereby establishing a baseline for evaluating possible affects from specific point-source stressors. Introduction The likelihood that urbanization of a watershed will cause some degree of impairment to the receiving stream is generally acknowledged by land planners and ecologists, but ways in which an aquatic system responds and the level of urbanization at which responses occur are not always understood (cf. Karr and Chu 1999, Paul and Meyer 2001, Pickett et al. 2008, Schueler 1994, Walsh et al. 2005). In an urbanizing area, it may not be apparent which particular features of landscape alteration are the primary causes of ecological impairment to a stream. Alternatively, when biological, physical, or chemical characteristics are used to assess the condition of a stream, interpretations of the changes in these characteristics are more precise when ways in which they relate to urbanization are understood. 1US Geological Survey (USGS), 331 Commerce Way, Pembroke, NH 03275. 2USGS, 3916 Sunset Ridge Road, Raleigh, NC 27607. 3US Environmental Protection Agency Region 1 New England, 1 Congress Street, Suite 1100 (HBS) Boston, MA 02114. *Corresponding author - 30 Northeastern Naturalist Vol. 17, No. 1 USGS urban land-use studies From 2000 to 2004, the US Geological Survey (USGS) National Water- Quality Assessment (NAWQA) program conducted urban land-use studies to determine the effects of urbanization on stream ecosystems; these urban studies focused on nine major metropolitan areas across the US, each associated with a different physiographic region (Couch and Hamilton 2002, Tate et al. 2005). The main objective of the studies was to gain a better understanding of how urbanization affected stream ecosystems in a nationally consistent manner by investigating the relation of urban intensity to the biological, chemical, and physical condition of streams. The New England Coastal Basins (NECB) region is a designated USGS study area that includes most of Rhode Island and the eastern drainages of Connecticut, Massachusetts, New Hampshire, and southern Maine. The NECB study area includes metropolitan Boston, as well as some of the least-developed areas in the eastern US (Ayotte and Robinson 1997). Consequently, the NECB study area was one of the nine regions selected for an urban study. Details of the NECB urban study are described in a USGS Professional Paper (Coles et al. 2004), a summary of which is given below to provide the context for the current paper. (Reports on the NECB study unit that are referenced in this paper may be found online at nawqaweb.htm.) The NECB urban study was based on a network of sites consisting of 30 watersheds and their streams that were within 130 km from Boston, MA. All sites were within US Environmental Protection Agency (USEPA) Ecoregion 59 (Northeastern Coastal Zone; Omernik 1987) and, additionally, were within a select group of US Forest Service (USFS) ecological subsections (Keys et al. 1995). The purpose of restricting the network of sites by these ecologically defined areas was to help increase the certainty that differences among sites could be attributed to urbanization rather than to natural variability. The sites were further standardized by selecting third- to fifth-order streams and delineating sampling reaches that were about 150 m long with similar physical-habitat characteristics. The level of urbanization was quantified for the 30 sites using a gradient of urban intensity that collectively represented urbanizing watersheds by substituting spatial for temporal variability. The gradient of urban intensity was derived from 24 landscape variables that were related to different urban-associated changes in the watersheds. These variables were selected from datasets that characterized watersheds by three distinct categories: basin land cover, human demography, and basin infrastructure. The basis of selection was that a variable was correlated strongly to population density (Spearman’s |rho| > 0.600), but correlated relatively weakly to watershed size (|rho| < 0.400). The 24 variables were then equally weighted for use in the gradient of urban intensity, which was then standardized over the 30 study sites from 0 to 100 to represent the lowest to highest levels of urban intensity across the sites (McMahon and Cuffney 2000). 2010 J.F. Coles, T.F. Cuffney, G. McMahon, and C.J. Rosiu 31 The gradient of urban intensity derived for the NECB study was referenced in Coles et al. (2004) as an a priori urban-intensity index (UII), but it was used primarily to identify relations between urbanization and biological, physical, and chemical characteristics of streams. It was not designed to assign an urban-intensity value to watersheds within the NECB area beyond the original study sites. However, because the a priori UII had been shown to effectively characterize urbanization, it was relatively simple to derive an index to measure the urban intensity for other watersheds in the NECB area. Deriving the NECB urban intensity index (NECB-UII) required two steps. First, the a priori urban gradient was cumbersome with 24 variables, but this number could be reduced by removing the less-relevant and redundant variables so that the NECB-UII would use only a few variables and still function effectively. Second, because the a priori urban gradient was scaled to function only with the original 30 study sites, formulation of the NECB-UII would require that the relation be expressed in a mathematical model for use in quantifying urbanization at other sites. Quantifying relations to urbanization In a review of different approaches to rural land-use planning and conservation, Theobald et al. (2005) described the need for developing environmental indicators that are applicable to local planning efforts. USEPA (2005) guidance for managing nonpoint source pollution in urbanizing areas acknowledges that indicators for monitoring urban effects vary by region and need to be specified. The guidance further defines ecological measures needed for effective management of nonpoint pollution, including characterizations of watershed conditions, and reliable biological, physical, and chemical indicators for use in monitoring the effects of urbanization. Objectives of our paper are to use data from the NECB urban study to (1) determine the general landscape changes that are related to urbanization in the region, (2) derive an NECB-UII to quantify urbanization in the watersheds, and (3) identify invertebrate-based metrics of ecological condition that are related to urbanization. Additionally, a procedure is described to estimate the expected ecological condition of a stream for a given level of urban intensity, based on previously defined relations between urbanization and invertebrate indicator metrics in the region. This approach would have applications in ecological assessments of streams where it would be difficult to assign impacts to a specific disturbance when compared to the overall effects from watershed development (Rosiu and Coles 2005). For example, when an ecological assessment is conducted at a site, invertebrateassemblage metrics from the site could be compared to values expected for the level of urbanization. Professional judgment could then be used to decide if differences between actual and expected values infer a degraded stream condition caused by a local disturbance (e.g., point source). 32 Northeastern Naturalist Vol. 17, No. 1 Methods The 30 sites for the New England Coastal Basins (NECB) urban study were selected during 1999 to represent a gradient of urban intensity from low to high values, while ensuring that in-stream natural features were constant among sampling reaches. Criteria for a sampling reach were that the stream was free-flowing for at least 150 m with some riffles, had no sign of recent human disturbance, and had well-defined banks with at least 50% mature vegetation cover. With metropolitan Boston MA, as the primary urban area, the sites collectively represented a gradient of urbanizing watersheds (Fig. 1, Table 1). Biological assemblages were sampled from August 1 to September 1, 2000. Semi-quantitative samples of aquatic invertebrates were collected Figure 1. Site locations for the New England Coastal Basins (NECB) urban gradient study and their association with the USEPA Level III Ecoregion and the USFS Ecological Subsection. 2010 J.F. Coles, T.F. Cuffney, G. McMahon, and C.J. Rosiu 33 with a Slack sampler (0.25-m2 area, 425-micron net) from five riffle areas in each sampling reach, which were composited and designated as the richest targeted habitat (RTH) sample. A qualitative multihabitat (QMH) invertebrate sample was also collected with a 212-micron mesh dipnet by sampling various microhabitats along the 150-m sampling reach (Cuffney et al. 1993). The RTH data were expressed as relative abundances. Additional taxa identified in the RTH sample but not found in the QMH sample were included in Table 1. Location of the urban land-use sites, area of their watersheds, and values of the New England Coastal Basins urban intensity index (NECB-UII). Site number corresponds to site location on Fig. 1. WA = watershed area (km2), UII = NECB-UII value. USGS Site Site Station code # Site name and location WA UII Maine 01072540 LIME 1 Little River near Lebanon 45.8 13.6 01072650 GREA 2 Greatworks River near North Berwick 60.2 23.4 New Hampshire 01072845 ISIN 3 Isinglass River at Batchelder Road near center 59.4 18.7 Strafford 01072904 BELL 4 Bellamy River at Bellamy Road near Dover 68.5 28.2 01073260 LAMP 5 Lamprey River at Cotton Road near Deerfield 83.1 14.6 01073458 NORT 6 North River at Rt 152 near Nottingham 74.9 17.5 010734833 LINH 7 Little River at Cartland Road at Lee 52.2 19.3 01089743 LSUN 8 Little Suncook River at Blackhall Road at Epsom 101.4 23.2 01090477 BLAB 9 Black Brook at Dunbarton Road near Manchester 53.7 16.9 01094005 BABO 10 Baboosic River at Bedford Road near Merrimack 73.0 24.1 010965852 BEAV 13 Beaver Brook at North Pelham 121.7 49.7 Massachusetts 01095220 STIL 11 Stillwater River near Sterling 78.7 21.0 01096544 STON 12 Stony Brook at School Street at Chelmsford 107.7 43.4 01096710 ASSA 14 Assabet River at Allen Street at Northborough 76.4 51.0 01096945 ELIZ 15 Elizabeth Brook off White Pond Road near Stow 48.5 28.4 01097270 FORT 16 Fort Pond Brook at River Road near South Acton 53.7 36.6 01097476 SUDB 17 Sudbury River at Concord Street at Ashland 89.6 40.3 01101500 IPSW 18 Ipswich River at South Middleton 115.3 63.7 01102345 SAUG 19 Saugus River at Saugus Ironworks at Saugus 60.4 96.7 01102500 ABER 20 Aberjona River (head of Mystic River) at 65.1 10.9 Winchester 011032058 CHAR 21 Charles River at Maple Street at North Bellingham 54.4 55.2 01105000 NEPO 22 Neponset River at Norwood 84.9 57.3 01105500 ENEP 23 East Branch Neponset River at Canton 72.9 66.6 01105581 MONA 24 Monatiquot River at River Street at Braintree 71.2 80.0 01106468 MATF 25 Matfield River at North Central Street at East 79.8 93.1 Bridgewater 01109000 WADE 26 Wading River (head of Threemile River) near Norton 113.4 39.3 01109595 MIDD 27 Middle River off Sutton Lane at Worcester 124.7 57.6 01110000 QUIN 28 Quinsigamond River at North Grafton 66.2 80.9 01112262 MILL 29 Mill River at Summer Street near Blackstone 73.7 31.5 Connecticut 01193340 BLAL 30 Blackledge River above Lyman Brook near North 49.2 24.8 Westchester 34 Northeastern Naturalist Vol. 17, No. 1 the QMH dataset. The final QMH dataset, therefore, expressed overall assemblage richness along the sampling reach. Addition data collected included fish and invertebrate-assemblage data, habitat data, water-chemistry data, and stream-stage and water-temperature data monitored continuously over about a year. These data are not discussed in detail, but it is important to specify that there was relatively little variance among sites in the habitat data. This result confirms that the studydesign objective for heterogeneity among sampling reaches was successful, and that ecological condition of the streams could be related to watershed features with greater certainty. See Coles et al. (2004) for further details on these data. Selecting variables for the NECB-UII The 24 landscape variables originally used in the NECB urban gradient were evaluated by principal component analyses (PCA) to identify a smaller set of variables that could define urbanization. By the nature of PCA, a variable strongly associated with urbanization would have a strong loading value (factor score on the first ordination axis) and a high fit statistic (correlation along the first ordination axis). Additionally, variables in the reduced set were categorized by specific but different aspects of urbanization: basin land cover, human demography, and basin infrastructure. We considered that including variables representing each of these categories was important for the NECB-UII so that it would function as a proxy for many urban-related changes. Therefore, the NECB-UII would be a more comprehensive representation of urban intensity than any single variable and likely be more responsive to various disturbances. Variables selected for the NECB-UII required scaling coefficients before they could be expressed collectively in a model used for calculating the urban intensity of watersheds. The scaling coefficient for each variable was its regression-line slope that was determined from the first-order (linear) equation of the scaled values (0–100) over the raw values for the original 30 study sites. The scaling coefficients could then be used in the NECB-UII to determine the urban intensity for any site by multiplying values of the landscape variables by their scaling coefficients, then finding the average of the products. Identifying the effects of urbanization The general change to the landscape from urbanization in the region was evaluated using correlations of the NECB-UII to land-cover patch statistics. Patch in this context represents a discrete area of homogeneous land cover that is differentiated by discontinuities with the adjoining patches. Patch density (the number of patches in a given area) analysis was done using FRAGSTATS (McGarigal and Marks 1995) to determine the extent that the landscape fragments as forested and developed land change in relation to urban intensity. 2010 J.F. Coles, T.F. Cuffney, G. McMahon, and C.J. Rosiu 35 To evaluate the relations of invertebrate RTH (quantitative) and QMH (qualitative) assemblages to urbanization, each dataset was first analyzed with correspondence analyses (CA), and the resulting first-axis site scores were correlated with the NECB-UII. Invertebrate metrics that were previously found to respond to urban intensity (i.e., |rho| > 0.7; Coles et al. 2004) were correlated with the NECB-UII. Included were metrics of abundance, richness, diversity, dominance, functional groups, and tolerance, which are commonly used in stream bioassessments (Barbour et al. 1999, Davis and Simon 1995, Rosenberg and Resh 1993). The USGS Invertebrate Data Analysis System (IDAS; Cuffney 2003) was used to resolve taxa ambiguities in the invertebrate data and to calculate the invertebrate metrics. Tolerance and functional-group metrics were calculated in IDAS with attribute data from the USEPA Rapid Bioassessment Protocol (Barbour et al. 1999). The multivariate ordinations were done with CANOCO 4.0 (ter Braak and Smilauer 1998). The correlations, regressions and scatterplots were done with SYSTAT 8 (SPSS 1998). Spearman correlation coefficients > 0.7 (|rho|, absolute value) were considered ecologically relevant. Results Deriving the urban intensity index PCA results indicated how strongly each of the 24 landscape variables contributed to the a priori urban gradient. First-axis loading values from the PCA ranged from 0.535 to 0.977, and fit statistics ranged from 0.286 to 0.955. Of the variables, seven had loadings and fit statistics that were 0.9 or greater and these were deemed candidates for the UII (Table 2). Candidate variables were judged as to how they were related to fundamentally different landscape changes associated with urbanization. This process was Table 2. Variables from the a priori urban gradient that were identified as candidate variables for the New England Coastal Basins urban-intensity index (NECB-UII). Loading values and fit statistics are results from PCA ordination axis-1, and indicate the relation of each variable to the urban gradient. Variables with an “*” were used in the NECB-UII. NLCD, National Land Cover Data. Loading Fit Variable Description value statistic ROADDEN* Road density in watershed [road length (km)/watershed 0.975 0.950 area (km2)] pBUF_2 Percentage of stream buffer (120 m outward from banks) 0.948 0.900 in developed land cover pBUF_4* Percentage of stream buffer (120 m outward from banks) 0.950 0.903 in forest land cover pNLCD_2* Percentage of watershed in NLCD “level 1” category: 0.970 0.940 developed pNLCD_4 Percentage of watershed in NLCD “level 1” category: forest 0.977 0.955 pNLCD_21 Percentage of watershed in low-intensity residential 0.958 0.917 POP99DEN* Population density, 1999, people per hectare 0.953 0.909 36 Northeastern Naturalist Vol. 17, No. 1 useful for avoiding strong variable redundancies as well as defining variables that would contribute to a comprehensive urban-intensity index. First excluded was pNLCD_21 because it represents a highly specific land-cover category that was inclusive with the general category pNLCD_2. Variables for developed and forested land cover had a strong inverse relation (r2 = 0.954). Therefore, the land-cover associations of pBUF_2 to pBUF_4 and of pNLCD_2 to pNLCD_4 were redundant, and consequently, only one variable from each of these couplets would be considered a candidate for the NECB-UII. Each candidate variable was standardized over its range, and various combinations of the variables were evaluated for the NECB-UII. Based on the two-tiered procedure to identify variables by PCA that strongly related to the a priori urban gradient, and that also represented specific aspects of urbanization, variables selected were ROADDEN, pBUF_4, pNLCD_2, and POP99DEN. The NECB-UII derived from these four variables was strongly related to the a priori urban gradient (r2 = 0.966), and by advantage of the four variables, would relate to the extents of infrastructure, loss of stream buffer, watershed development, and human predominance on the landscape. Additionally, when 21 invertebrate metrics were correlated to the NECB-UII and to the a priori urban gradient, the coefficients were overall slightly stronger with the NECB-UII (average |rho| = 0.860 and 0.849, respectively). At this stage of development, the NECB-UII was still scaled only among the study sites, as was the a priori urban gradient. However, by regressing the absolute values against the scaled values of each variable, scaling coefficients were determined so that the NECB-UII could be expressed by the following equation: NECB-UII = ([(ROAD*12.5) + (BUFF%*1.74) + (DEV%*1.53) + (POP *8.09)]*0.25), where ROAD = road density (road length [km] / watershed area [km2]), BUFF% = percentage stream buffer not in forest land cover (NLCD level 1), DEV% = percentage watershed in developed land cover (NLCD level 1), and POP = population density, people per hectare (US census data). A consequence of this equation for the NECB-UII is that the urbanintensity values across the 30 study sites change from 0 to 100 to a range from 13.6 to 111. For example, the least urbanized site in the study (LIME; Table 1) has a NECB-UII value of 13.6 instead of zero, because a zero value would now represent a hypothetical site with no development. Fragmentation of the landscape from urbanization The predominant effect of urbanization on the landscape of the region was the general loss of forest cover with encroaching development (Coles et al. 2004). However, this shift was not a simple one-to-one parcel replacement of forested land by developed land, as determined when the changes in patch densities of the two land-use categories were compared. 2010 J.F. Coles, T.F. Cuffney, G. McMahon, and C.J. Rosiu 37 There was a linear increase in patch density (number of patches per area) of forested land with increased urban intensity (rho = 0.941, Fig. 2a). Concurrently, the average patch size decreased with urban intensity (rho = -0.975), indicating that forests are lost from urbanization as they fragment into smaller patches. Patch density of developed land shows a different response pattern (Fig. 2b). Over the urban-intensity gradient, the average patch size of developed land increases consistently (rho = 0.952), but the number of the patches increases only to a NECB-UII value of about 40 (rho = 0.872). Above this value, patch density decreases with urban intensity (rho = -0.859). These results suggest that urbanization of a forested watershed begins with a few isolated small communities and that the number of communities increases with urbanization only to a certain level. As urban intensity continues to increase, communities appear to be coalescing into fewer but larger towns and cities, which results in a decrease in the number of developed-land patches. Forested land, however, continues to become more fragmented with increasing urban intensity, possibly due in part to concurrent increase in road density. Invertebrate assemblage relations to urbanization There were 294 invertebrate taxa identified in the samples collected from the 30 sites. The CA ordinations of invertebrate data resulted in primaryaxis eigenvalues of 0.350 for the RTH data and 0.324 for the QMH data, indicating that both datasets had strong patterns of assemblage structure. Furthermore, site scores from the RTH and QMH data ordinations were strongly correlated to the NECB-UII (RTH, |rho| = 0.893; QMH, |rho| = 0.909) and they were relatively linear over the responses (Fig. 3A, B). Several of the invertebrate metrics of structure and function that are commonly used in biomonitoring programs were correlated to the Figure 2. The relations between patch density of land cover and urban intensity (NECB-UII). (A) Patches of forest vs. NECB-UII, rho = 0.941. (B) Patches of developed land vs. NECB-UII, rho = 0.872 for the increasing response and rho = –0.859 for the decreasing response. 38 Northeastern Naturalist Vol. 17, No. 1 NECB-UII. Of approximately 100 metrics tested, those based on taxa richness generally showed a stronger response than metrics based strictly on abundance. These results indicated that qualitative information from the QMH data was as relevant as the RTH abundance data in characterizing relations of invertebrate assemblages to urbanization. This finding is consistent with the results from the correlations of ordination site scores to the NECB-UII (Fig. 3A, B) that indicated the RTH and QMH data were strongly related to urbanization. Total taxa richness was negatively correlated with the NECB-UII (rho = -0.901) and with a linear response (Fig. 4A). Similarly, EPT taxa richness was negatively correlated with the NECB-UII (rho = -0.902), even though the number of EPT taxa accounted for less than half of the total taxa in every sample (Fig 4B). Conversely, the percentage richness of non-insect taxa increased linearly with increasing urban intensity (rho = 0.904; Fig. 4C). The percentage richness of mollusks plus crustaceans (Mol + Crus), the dominant non-insect taxa, also increased linearly with increasing urban intensity (rho = 0.901; Fig. 4D). Functional feeding-group metrics generally were not related as strongly to urbanization as did metrics based on taxa structure. An exception was richness of gatherer-collector taxa, which was negatively correlated with the NECB-UII (rho = -0.800; Fig. 4E), and showed a “response-loss” threshold at a NECB-UII value of about 50. A threshold such as this one indicates that values of the metric are predictable only over a segment of the urban gradient, which in this case was from low to moderate levels of intensity. The average taxa tolerance, an average of pollution-tolerance values assigned to taxa at a site, was linear and had a strong positive correlation with the NECB-UII (rho = 0.892; Fig. 4F). This result indicated a gradual shift in the invertebrate assemblages from sensitive to tolerant taxa with increasing urban intensity. Figure 3. The relation of invertebrate site scores to the NECB-UII. The site scores were derived from Correspondence Analysis ordinations of the invertebrate data: (A) RTH, quantitative sample from riffle habitats (|rho| = 0.893). (B) QMH, qualitative sample from multiple habitats |rho| = 0.909). 2010 J.F. Coles, T.F. Cuffney, G. McMahon, and C.J. Rosiu 39 Although richness-based metrics generally had somewhat stronger correlations to the NECB-UII than abundance-based metrics, hence the usefulness of the QMH data, the quantitative RTH data were essential for Figure 4. Relations of invertebrate taxa-richness metrics to the NECB-UII. These metrics were based on the QMH (qualitative multihabitat) data. Spearman rho values for each correlation: (A) total richness = -0.901, (B) EPT richness = -0.902, (C) percentage richness of non-insects = 0.904, (D) percentage richness of mollusks plus crustaceans = 0.901, (E) gatherer-collector richness = -0.800, (F) average taxa tolerance = 0.892. 40 Northeastern Naturalist Vol. 17, No. 1 indices of diversity, evenness, and dominance. Shannon-Wiener diversity index (SHANDIV) was negatively correlated with the NECB-UII (rho = -0.867; Fig. 5A). In addition, the percentage abundance of the five most dominant taxa (DOM5) was positively correlated with the NECB-UII (rho = 0.870; Fig. 5B). A regression between SHANDIV and DOM5 showed they were inversely related (r2 = 0.980), and suggests the response of SHANDIV was influenced by the five dominant taxa. These responses furthermore indicated that diversity and evenness of the taxa strongly decreased with urbanization. Discussion The advantages of an urban-intensity index Each of the four variables used in the NECB-UII represents a separate aspect of an urbanizing watershed, and when combined in the NECB-UII, functions as an effective proxy for various other changes associated with urbanization. For example, population density has often been used as an indirect measure of urban intensity, and accordingly, its use in the NECB-UII is indicative of the general impact of human presence on the landscape. In this regard, population density represents human activities in the watershed that can affect stream condition. However, a concern in using a single variable to characterize urbanization is that too much reliance is placed on one indicator to respond to many changes that can cause stream impairment; a multimetric index overcomes this uncertainty in part by relying on several variables that function comprehensively to quantify urbanization. Road density was used to characterize the infrastructure associated with many of the human activities that contribute to urbanization, and has been used as a surrogate for impervious surface (May et al. 1997). Of the 24 landscape variables used in the a priori urban gradient, road density Figure 5. Relations of selected RTH invertebrate metrics to the NECB-UII. Spearman rho values for the correlations: (A) Shannon-Wiener diversity = -0.867, (B) percentage abundance of dominant 5 taxa = 0.870. 2010 J.F. Coles, T.F. Cuffney, G. McMahon, and C.J. Rosiu 41 generally correlated most strongly with site scores from the biological, chemical, and physical datasets (Coles et al. 2004). Contaminants from sources such as atmospheric deposition, vehicular traffic, and applications of de-icing agents can accumulate on roadways and flush into streams, thereby affecting water quality (Buckler and Granato 1999, Forman et al. 2002, Granato and Smith 1999). Biological assemblages can be directly affected by these water-quality changes, but also by many of the physical factors associated with roadways, such as increased sediment loads and flow modifications to streams (Angermeier et al. 2004, Forman and Alexander 1998, Wood and Armitage 1997). Percentage of watershed area in general developed land cover (NLCDLevel I) was used to characterize the general change from a natural landscape (typically forested for the region) to residential, commercial, and industrial centers. Although land-cover classifications that represent more specific land use (NLCD-Level II) were available, it was previously reported that correlations between landscape change and biological, physical, and chemical endpoints were usually stronger with the more comprehensive Level I classifications (Coles et al. 2004). The percentage of stream buffer in forested cover was used to characterize the loss of riparian forest that often is associated with urbanization. The loss of forested cover along the riparian buffer was previously reported to be associated with declining fish and invertebrate assemblages (Coles et al. 2004), and alterations in this zone probably affect stream condition more directly than other land-cover variables. Although loss of forested cover was closely related to increase in developed land cover for the watershed, the loss of riparian forest helped to “weight” these disturbances by their proximity to the stream. The importance of managing the natural riparian corridor was recognized through legislation in Massachusetts as the Rivers Protection Act of 1996 (Commonwealth of Massachusetts 1996), which protects areas extending from the mean annual high-water line on banks of perennial streams. Landscape changes related to urbanization The most notable landscape change related to urbanization in the region was in how developed land increased as forested land was lost. It appeared that urbanization resulted in a steady loss of forests by fragmenting the natural landscape with patches of development. Fragmentation of forests continued with urbanization, but beyond moderate levels of urban intensity (NECB-UII > 40), the number of developed patches began decreasing as they apparently coalesced into increasingly larger urban centers. Impervious surface area (ISA) is closely associated with urbanization and has been used to measure urban intensity in different regions (Schueler 1994). Values of ISA were not available at the time of our study, but establishing the relation between the NECB-UII and ISA was considered important so that our results could be compared to studies that use ISA as a surrogate for urbanization. Subsequent to our study, estimated values of impervious42 Northeastern Naturalist Vol. 17, No. 1 surface data were derived by NOAA from mid-1990s satellite night-light observations, 1992 land-cover data, and 2000 Census road-density data (Elvidge et al. 2004, NOAA 2006). The NOAA-based ISA values for our sites were used in a regression with the NECB-UII, and the resulting relation was determined: percentage ISA = 0.382 * NECB-UII -4.27 (r2 = 0.982, P < 0.001)). Although a strong linear relation was confirmed, the regression line intersects the NECB-UII axis at 11.2 (x-intercept value), which may indicate that the relation is nonlinear over low values (Fig. 6). Stream condition indicators of urbanization The overall results of this study showed that urbanization was associated with declining ecological conditions, which is consistent with results reported in other studies (Kennen 1999, May et al. 1997, Morley and Karr 2002, Walsh et al. 2005). Among the responses in the invertebrate-assemblage data, changes in taxa richness appeared more important than abundance. This finding was most strongly supported by the qualitative QMH data, where the CA site scores and the metrics such as EPT richness and pollution tolerance of taxa were strongly correlated to the NECB-UII. In relating the effects of urbanization to invertebrate assemblages, a presumed ecological “coupling” exists across spatial scales such that watershed development causes changes to water quality that will ultimately affect assemblages at the stream-reach level (Picket et al. 2008). A gradient of water quality was described in Coles et al. (2004) that was PCA-derived from water-chemistry data collected at the 30 study sites. Among water-quality constituents that increased with urbanization were specific conductance, total Kjeldahl nitrogen (TKN), total phosphorus, and total pesticides. Subsequent analyses showed that the NECB-UII was strongly correlated to this water-quality gradient (rho = -0.927), indicating that water quality declined with urbanization. The water-quality gradient was also strongly correlated to invertebrate metrics such as EPT richness (rho = 0.878), so it is likely that water quality was an intermediate factor between urbanization and the invertebrate response. Figure 6. Relation between derived values of impervious surface area (ISA) and the NECB-UII. 2010 J.F. Coles, T.F. Cuffney, G. McMahon, and C.J. Rosiu 43 Interpreting responses to urbanization Changes in the invertebrate assemblages generally began as soon as conditions departed from background (minimal urban), which suggests that they did not show a resistance to initial impairment as has been observed in other studies (Cuffney et al. 2000). This result may indicate that the biological communities are highly sensitive to watershed change, or that urban intensity at “background conditions” was already at a level where impairment would occur. The lowest urban-intensity value in the study (NECB-UII = 13.6) indicates some degree of ecological disturbance may have occurred in even the least urbanized watersheds in the region. Additionally, if the relation between ISA and NECB-UII holds true (described above), a threshold of disturbance at some critical ISA value was not apparent in how the invertebrate assemblages responded over very low ISA values. This result also indicates that the ISA level at which ecological disturbances first occur is less than the sometimes-cited values of 10–15 percent (cf. Arnold and Gibbons 1996, Booth and Jackson 1997, Booth et al. 2004, Klein 1979, May et al.1997, Schueler 1994). Although variables did not show an initial resistance to urbanization, we observed examples where the response was linear from low to moderate urban intensities, but which then declined or even reached an exhaustion phase at higher urban intensities. An example of this response-loss type of threshold was observed in gatherer-collector richness for the invertebrates (Fig. 4e), and has been described for taxa richness in benthic diatoms and abundance of insectivorous fish (Coles et al. 2004). Response-loss thresholds occurred at about the same level of urban intensity (NECB-UII around 50), which was just above the level where the number of developed land patches changed from increasing to decreasing with urbanization (Fig. 2b). It was not discernable from the study, however, whether these thresholds indicated a sensitivity of biological assemblages to effects from specific landscape changes at that urban intensity, such as coalescing of developed patches into urban centers. Because thresholds are points along a response gradient where the underlying relation changes abruptly, it is important that they be recognized when assessing or monitoring aquatic resources. For example, if urban intensity were to increase from mid- to high levels, an assessment of changing stream condition would be different if based on the gatherer-collector richness (Fig. 4E) compared to the average taxa tolerance (Fig. 4F). Although the use of metrics that show a threshold response might be considered less effective for assessing ecological condition, this information is still very useful for identifying the level of a disturbance below which certain taxa can survive. This situation could occur where the protection of particular species is important, and the threshold value would indicate the urban intensity where extirpation would be expected. Compared to variables showing a threshold response, variables that indicate long linear responses are often favored when assessing the 44 Northeastern Naturalist Vol. 17, No. 1 ecological condition of streams because they are more predictable over a disturbance gradient. Examples from our study (all based on qualitative data) were EPT taxa richness, percentage richness of non-insect taxa, and the average taxa tolerance (Table 3). Each of these was derived from a combination of at least three single-parameter variables (i.e., multiple invertebrate taxa), which emphasizes that multimetric indices often more effectively indicate a continuous response over a disturbance gradient than a single variable can indicate. Predicting effects from urbanization EPA Region 1 (New England) recently initiated a project through the Regional Applied Research Effort (RARE) of the Office of Science Policy (USEPA 2007a) to evaluate the relations between land-cover variables and biological responses, using data from the NECB urban gradient study and the New England Wadeable Streams Survey (USEPA 2006, 2007b). In addition, the biological responses investigated for this RARE project are to be evaluated in the context of the Biological Condition Gradient (BCG), a descriptive model in which a suite of ecological attributes are predicted to change in response to increasing levels of stressors (Davies and Jackson 2006). The results are expected to determine the extent that ecological responses, such as those we have described, are applicable to sites across the region (N.E. Detenbeck, US EPA, June 2008 pers. comm.). Results from our study may help in understanding the underlying relations between urbanization and ecological response variables. It is anticipated that our findings will be useful in assigning values of the expected ecological condition for streams, given the level of urbanization in their watersheds. For example, for a degraded stream segment undergoing restoration, the expected condition after recovery could depend in large part on urban intensity above the site. Using the relations between invertebrate metrics and the NECB-UII (Fig. 4, Table 3), if the urban intensity of the watershed was NECB-UII = 20, it would be reasonable to expect an EPT richness of around 20 to 30 after restoration. Achieving EPT richness in this range would be unrealistic, however, if the urban intensity was around NECB-UII = 80. In this case, it is more likely that EPT values below 10 would be achieved. Therefore, the ability to predict the condition at a site by accounting for “background” effects of urbanization is expected to Table 3. Regression statistics for selected invertebrate metrics that have a strong linear relation to urbanization as represented by the NECB-UII. Metrics are based on qualitative multihabitat (QMH data), which characterizes assemblage richness (taxa presence/absence). Invertebrate Probability Regression Y Figure metric R2 value slope intercept in text EPT richness 0.743 <0.001 -0.294 30.70 4b Non-insect % richness 0.837 <0.001 0.342 4.45 4c Average taxa tolerance 0.824 <0.001 0.022 3.92 4f 2010 J.F. Coles, T.F. Cuffney, G. McMahon, and C.J. Rosiu 45 improve the precision of stream assessments, especially when evaluating effects from specific stressors. Acknowledgments The authors thank the many USGS colleagues who participated in the fieldwork and the extensive data-collection efforts of the original study; special thanks go to Karen Beaulieu of the USGS Connecticut Water Science Center, who coordinated most of the fieldwork, kept track of the data, and assisted in the analysis. We greatly appreciate the efforts of Mary C. Freeman (USGS), Cathy M. Tate (USGS), and Gary Kleppel (University of Albany, SUNY) for technical reviews of the manuscript. Support from EPA Region 1 was provided to J.F. Coles to serve as technical advisor on the RARE project described above and to produce a draft of this paper. We recognize Naomi Detenbeck (EPA/ORD Laboratory, Narragansett, RI) for her efforts in using data from the NECB urban study in conjunction with regional EPA data to investigate possible causes of impairments to aquatic systems in the region. Literature Cited Angermeier, P.A., A.P Wheeler, and A.E. Rosenberger. 2004. A conceptual framework for assessing impacts of roads on aquatic biota. Fisheries 29:19–29. Arnold, C.L., and C.J. Gibbons. 1996. Impervious surface coverage: The emergence of a key environmental indicator. American Planners Association Journal 62:243–258. Ayotte, J.D., and K.W. Robinson. 1997. National water-quality assessment program: New England Coastal Basins. US Geological Survey Fact Sheet FS-060-97. Available online at Accessed January 2010. 4 pp. Barbour, M.T., J. Gerritsen, B. D. Snyder, and J.B. Stribling. 1999. Rapid bioassessment protocols for use in streams and wadeable rivers: Periphyton, benthic macroinvertebrates, and fish (2d Edition). US Environmental Protection Agency, Office of Water, EPA 841-B-99-002, Washington, DC. Booth, D.B., and C.R. Jackson. 1997. Urbanization of aquatic systems: Degradation thresholds, stormwater detection, and the limits of mitigation. Journal of the American Water Resources Association 33(5):1077–1090. Booth, D.B, J.R. Karr, S. Schauman, C.P. Konrad, Christopher, S.A. Morley, M.G. Larson, and S.J. Burges. 2004. Reviving urban streams: Land use, hydrology, biology, and human behavior. Journal of the American Water Resources Association 40:1351–1363. Buckler, D.R., and G.E. Granato. 1999. Assessing biological effects from highway runoff constituents. US Geological Survey Open-File Report 99-240, Northborough, MA. Coles, J.F., T.F. Cuffney, G. McMahon, and K.M. Beaulieu. 2004. The effects of urbanization on the biological, physical, and chemical characteristics of coastal New England streams. US Geological Survey Professional Paper 1695, Northborough, MA. Commonwealth of Massachusetts. 1996. 310 Code of Massachusetts Regulations (CMR) 10.00, chapter 258. 46 Northeastern Naturalist Vol. 17, No. 1 Couch, C., and P. Hamilton. 2002. Effects of urbanization on stream ecosystems. US Geological Survey Fact Sheet 042-02, Reston, VA. Cuffney, T.F. 2003. User's manual for the National Water-Quality Assessment program invertebrate data analysis system (IDAS) software—Version 3. US Geological Survey Open-File Report 03-172, Raleigh, NC. Cuffney, T.F., M.E. Gurtz, and M.R. Meador. 1993. Methods for collecting benthic invertebrate samples as part of the National Water Quality Assessment Program. US. Geological Survey Open File Report 93-406, Raleigh, NC. Cuffney, T.F., M.R. Meador, S.D. Porter, and M.E. Gurtz. 2000. Responses of physical, chemical, and biological indicators of water quality to a gradient of agricultural land use in the Yakima River Basin, Washington. Environmental Monitoring and Assessment 64:259–270. Davies, S.P., and S.K. Jackson. 2006. The biological condition gradient: A descriptive model for interpreting change in aquatic ecosystems. Ecological Applications 16(4):1251–1266. Davis, W.S., and T.P. Simon. 1995. Biological Assessment and Criteria: Tools for Water Resource Planning and Decision Making. Lewis Publishers, Boca Raton, fl. Elvidge, C.D., C. Milesi, J.B. Dietz, B.T. Tuttle, P.C. Sutton, R. Nemani, and J.E. Vogelmann. 2004. US constructed area approaches the size of ohio. Eos, Transactions of the American Geophysical Union 85, 24, 233. Forman, R.T.T., and L.E. Alexander. 1998. Roads and their major ecological effects. Annual Review of Ecology and Systematics 29:207–231. Forman, R.T.T., D. Sperling, J.A. Bissonette, A.P. Clevenger, C.D. Cutshall, V.H. Dale, L. Fahrig, R. France, C.R. Goldman, K. Heanue, J.A. Jones, F.J. Swanson, T. Turrentine, and T.C. Winter. 2002. Road Ecology: Science and Solutions. Island Press, Washington, DC. Granato, G.E., and K.P. Smith. 1999. Estimating concentrations of road-salt constituents in highway-runoff from measurements of specific conductance. US Geological Survey Water-Resources Investigations Report 99-4077, Northborough, MA. Karr, J.R., and E.W. Chu. 1999. Restoring Life in Running Waters: Better Biological Monitoring. Island Press, Washington, DC. Kennen, J.G. 1999. Relation of macroinvertebrate community impairment to catchment characteristics in New Jersey streams. Journal of the American Water Resources Association 36(4):939–955. Keys, J.E., Jr., C.A. Carpenter, S.L. Hooks, F.G. Koenig, W.H. McNab, W.E. Russell, and M.L. Smith. 1995. Ecological units of the eastern United States—First approximation: Atlanta, GA (map [scale 1:3,500,000] and booklet of tables). US Department of Agriculture, Forest Service. Available online at http://www.fs.fed. us/rm/ecoregions/. Accessed January 2010. 80 pp. Klein, R.D. 1979. Urbanization and stream quality impairment. Water Resources Bulletin 15:948–963. May, C.W., R. Horner, J.R. Karr, B. Mar, and E. Welch. 1997. Effects of urbanization on small streams in the Puget Sound Lowland Ecoregion. Watershed Protection Techniques 2(4):483–494. 2010 J.F. Coles, T.F. Cuffney, G. McMahon, and C.J. Rosiu 47 McGarigal, K., and B.J. Marks. 1995. FRAGSTATS: Spatial pattern analysis program for quantifying Landscape Structure. US Department of Agriculture, Forest Service, Pacific Northwest Research Station. General Technical Report PNW-GTR-351. Available online at Accessed January 2010. McMahon, G., and T.F. Cuffney. 2000. Quantifying urban intensity in drainage basins for assessing stream ecological conditions. Journal of American Water Resources Association 36(6):1247–1261. Morley, S.A., and J.R. Karr. 2002. Assessing and restoring the health of urban streams in the Puget Sound Basin. Conservation Biology 16(6):1448–1509. National Oceanic and Atmospheric Administration (NOAA). 2006. Impervious surface area of the United States. Available online at lca/ccap.html. Accessed July 2006. Omernik, J.M. 1987. Ecoregions of the conterminous United States. Annals of the Association of American Geographers 77(1):118–125. Paul, M.J., and J.L. Meyer. 2001. Streams in the urban landscape. Annual Review of Ecology and Systematics 32:333–365. Pickett, S.T.A., M.L. Cadenasso, J.M. Grove, P.M. Groffman, L.E. Band, C.G. Boone, W.R. Burch, C.S.B. Grimmond, J. Hom, J.C. Jenkins, N.L. Law, C.H. Nilon, R.V. Pouyat, K. Szlavecz, P.S. Warren, and M.A. Wilson. 2008. Beyond urban legends: An emerging framework of urban ecology, as illustrated by the Baltimore Ecosystem Study. Bioscience 58(2):139–150. Rosenberg, D.M., and V.H. Resh. 1993. Freshwater Biomonitoring and Benthic Macroinvertebrates. Chapman and Hall, Inc., Routledge, New York, NY. Rosiu, C.J, and J.F. Coles. 2005. Standardizing sediment risk characterization on the basis of urban intensity of the watershed. Poster presented at EPA Science Forum, May 18, 2005. Collaborative Science for Environmental Solutions, Washington, DC. Schueler, T. 1994. The importance of imperviousness. Watershed Protection 1(3):100–111. SPSS. 1998. SYSTAT 8.0 Statistics. SPSS Inc., Chicago, IL. Tate, C.M., T.F. Cuffney, G. McMahon, E.M.P. Giddings, J.F. Coles, and H. Zappia. 2005. Use of an urban-intensity index to assess urban effects on streams in three contrasting environmental settings. Pp. 291–315, In L.R. Brown, R.M. Hughes, R. Gray, and M.R. Meador (Eds.). Effects of Urbanization on Stream Ecosystems. American Fisheries Society, Symposium 47, Bethesda, MD. ter Braak, C.J.F. and P. Smilauer. 1998. CANOCO manual and user’s guide to CANOCO for windows. Software for CANOCO community ordination (Version 4). Microcomputer Power, Ithaca, NY. Theobald, D.M., Thomas Spies, J.D. Kline, Bruce Maxwell, N.T Hobbs, and V.H. Dale. 2005. Ecological support for rural land-use planning. Ecological Applications 15(6):1906–1914. US Environmental Protection Agency (USEPA). 2005. National management measures to control nonpoint source pollution from urban areas. USEPA, Office of Wetlands, Oceans, and Watersheds, Washington, DC. EPA-841-B-05-004. USEPA. 2006. Wadeable Streams Assessment: A Collaborative Survey of the Nation’s Streams. USEPA, Office of Wetlands, Oceans, and Watersheds, Washington, DC. EPA-841-B-06-002. 48 Northeastern Naturalist Vol. 17, No. 1 USEPA. 2007a. Regional Applied Research Effort (RARE) Program homepage. Available online at Accessed 7 June 2007. USEPA. 2007b. The New England Wadeable Stream Survey (NEWS): Development of common assessments in the framework of the biological condition gradient. Available online at pdf. Accessed October 2007. Walsh, C.J, A.H. Roy, J.W. Feminella, P.D. Cottingham, P.M. Groffman, and R.P. Morgan. 2005. The urban-stream syndrome: Current knowledge and the search for a cure. Journal of the North American Benthological Society 24:706–723 Wood, P.J., and P.D. Armitage. 1997. Biological effects of fine sediment in the lotic environment. Environmental Management 21:203–217.