nena masthead
NENA Home Staff & Editors For Readers For Authors

Spatial Variation in Stream Water Quality in Relation to Riparian Buffer Dimensions in a Rural Watershed of Eastern New York State
Sean S. Madden, George R. Robinson, and John G. Arnason

Northeastern Naturalist, Volume 14, Issue 4 (2007): 605–618

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

 

Access Journal Content

Open access browsing of table of contents and abstract pages. Full text pdfs available for download for subscribers.



Current Issue: Vol. 30 (3)
NENA 30(3)

Check out NENA's latest Monograph:

Monograph 22
NENA monograph 22

All Regular Issues

Monographs

Special Issues

 

submit

 

subscribe

 

JSTOR logoClarivate logoWeb of science logoBioOne logo EbscoHOST logoProQuest logo

2007 NORTHEASTERN NATURALIST 14(4):605–618 Spatial Variation in Stream Water Quality in Relation to Riparian Buffer Dimensions in a Rural Watershed of Eastern New York State Sean S. Madden1,2,*, George R. Robinson1, and John G. Arnason3 Abstract - Studies of forested rural watersheds provide estimates of background contamination for comparison with streams and rivers in other settings. We performed a landscape analysis and measured major dissolved ions and benthic macroinvertebrates for a small rural watershed in Albany County, NY, to determine spatial variation in water quality. An estimated 73% of the surface cover is post-agricultural forest, with only 2.3% of the watershed covered by roads and other impervious surfaces. Although water quality was consistently high in most of the creek, we detected three relatively distinct zones separated by impoundments; zonation was most apparent in relative concentrations of major ions, less so with benthic macroinvertebrate community similarity. At ten sample stations, buffer size, measured as upstream land cover and distance to nearest road, did not correlate well with chemical water quality indicators. In particular, we found the highest levels of chloride, indicative of road-salt contamination, in areas of maximum forest buffer. Small feeder creeks that drain nearby roads may function as “leaks” in otherwise well-buffered watersheds with low road densities. Introduction Forested, rural watersheds supply much of the drinking water in the northeastern US, and also serve as reference study sites, useful for determining levels and effects of background contaminants, such as atmospheric pollution (e.g., Likens 2004). We studied the rural Ten Mile Creek watershed as a reference study area to examine relationships between riparian vegetation and water quality. In more urbanized regions, high levels of impervious surface cover are associated with degradation of stream water quality, and fine-scale variation can be detected in stream reaches with and without riparian buffers (Groffman et al. 2003, Paul and Meyer 2001, Roy et al. 2005, Walsh et al. 2005). However, even rural watersheds have the potential for spatial variability in water chemistry, in part because land cover in the region derives from a relatively fine-scale mosaic of land- and water-use history (Mladenoff et al. 1993, O’Neill et al. 1988). Rural streams also receive substantial amounts of non-point contamination, via atmospheric deposition, agricultural runoff, and highway deicing 1Program in Biodiversity, Conservation and Policy, University at Albany, 1400 Washington Avenue, Albany, NY 12222. 2Division of Fish, Wildlife and Marine Resources, New York State Department of Environmental Conservation, 625 Broadway, Albany, NY 12233. 3Department of Earth and Atmospheric Sciences, University at Albany, 1400 Washington Avenue, Albany, NY 12222. *Corresponding author - ssmadden@gw.dec.state.ny.us. 606 Northeastern Naturalist Vol. 14, No. 4 agents (Jones et al. 2000, Kaushal et al. 2005). With expanding incursions of urban fringes into rural lands throughout North America (Elvidge et al. 2004), impending land-cover change has prompted renewed interest in the buffering capacity of forest cover, driven by concerns for watershed conservation (Palmer et al. 2004, Theobald 2004). A common expectation in watersheds of all land-cover types is that riparian buffer width is positively correlated with stream water quality (Castelle et al. 1994, Clausen et al. 2000, Correll 2000, Fischer et al. 2000, Spruill 2000). Riparian zones sequester sediments and non-point source pollutants, slow surface fl ow, stabilize stream banks, moderate stream temperatures, and serve as important denitrifi cation sites (Correll 2000, Groffman et al. 2003). Low road densities and intact riparian forests are two of the characteristics that make rural watersheds attractive as reference sites, but anticipated landuse change may place greater demands on a reduced buffering capacity. We tested for spatial variation present in this rural watershed by examining stream water chemistry and benthic macroinvertebrate diversity in comparison to riparian buffer size and surrounding land cover. In the process, we found significantly higher levels of chloride contamination in areas of maximum forest buffer width. In addition, concentrations of major dissolved ions revealed potential contamination not evident from biotic sampling. Methods Site description The headwaters of Ten Mile Creek are located at approximately 600 m in elevation in the Partridge Run Wildlife Management Area of Albany County, NY. The creek flows south into Greene County, where it joins Catskill Creek, a major tributary of the Hudson River. The creek drains a predominately well-forested and post-agricultural watershed (57.7 km2) with no current industrial point sources or urban land use. The watershed supports a low population density, but rural Albany County is projected to experience significant population growth and associated suburban development within the next ten years (Capital District Regional Planning Commission 2005). The study area comprises the uppermost 10 km of creek, draining 26.9 km2 of the watershed where most of the creek is bordered by the 75-year-old Edmund Niles Huyck Preserve and Biological Research Station (http://www.huyckpreserve.org/). Within the study area, the creek passes through two impoundments—Lincoln Pond (4 ha) and Lake Myosotis (44 ha)—and the Hamlet of Rensselaerville, which receives drinking water from the latter impoundment. In the study area, the bedrock geology consists of sandstones, siltstones, and shales of the Lower Hamilton Group. Surfi cial geology consists of thin (<1 m) glacial till deposits in upland areas with thicker (up to 30 m) glacial outwash deposits of sand and gravel in stream channels and low-lying areas. As a result of glaciation and erosion from past agriculture, the soils are generally shallow, 2007 S.S. Madden, G.R. Robinson, and J.G. Arnason 607 poorly draining silt loams. Mineral soils are acidic, in the range of pH 3.5 in old hemlock stands to 6.0 in deciduous forests (Wyman 1988). Except for miscellaneous records from private wells, there is little information on groundwater. Within the town of Rensselaerville, wells in bedrock yield an average of 0.5 L s-1, whereas wells in unconsolidated deposits yield 1.3 L s-1 (Bagdon Environmental Associates 1989). It is likely that groundwater fl ow in bedrock is fracture-controlled, occurring along joints and bedding planes, whereas porous fl ow occurs through unconsolidated sand and gravel deposits. Groundwater recharge would be expected to be higher in areas overlain by unconsolidated sand and gravel compared to areas overlain by relatively impermeable till. While the Ten Mile Creek catchment appears well forested today, it is far from pristine and has a history of timber harvesting, milling, and industry (Wyman 1988). Some active agriculture is still present in parts of the watershed, but most fi elds are in mid- to late stages of succession to closed-canopy forest. Although some remnant old-growth forest remains on steep slopes in the valley, most forest cover is second and third growth. Housing is dispersed and on the fringe of rural and natural lands except for concentrated settlement in the Hamlet of Rensselaerville (2000 US Census population = 528). Ten sample sites were chosen within the study area (Fig. 1). Site selection was designed to capture potential spatial variation, including sampling above and below the impoundments and above and below the Hamlet of Rensselaerville. Two sites were clustered at outlets from Lake Myosotis, one to sample spillway overfl ow and the other to sample a deep-water outlet, which converge approximately 50 m downstream. Landscape analysis A geographic information system (GIS) database was developed using ESRI ArcGIS 8.3TM to conduct watershed mapping and analysis. Digital Figure 1. Ten Mile Creek Study Watershed, Albany County, NY. The study area is the northern half of the 57.7-km2 Ten Mile Creek watershed. Sampling sites are numbered 1 through 10 and were grouped into sections to test for spatial variability. Two small tributaries (Tributary 1 and Tributary 2) were also sampled as part of the study. 608 Northeastern Naturalist Vol. 14, No. 4 ortho-images were downloaded from the New York State GIS Clearinghouse (www.nysgis.state.ny.us) to provide a base map for the impervious surface and land-use analysis. A watershed boundary for Ten Mile Creek was obtained from the Capital District Regional Planning Commission (T. Fabozzi, Capital District Regional Planning Commission, Albany, NY, pers. comm). Land use and land cover within the watershed was divided into 14 categories. Each category was hand-digitized over orthophotographic layers, and the surface area for each category was calculated. Interpretations were corrected using others sources, including USGS and NYS Department of Transportation base maps, satellite imagery, researcher’s maps from the E.N. Huyck Preserve, and ground reconnaissance. All impervious and semipervious surface categories within the watershed boundary were summed to create total percent of impervious surfaces for the watershed. A powerline corridor bisecting a northern portion of the watershed was categorized as impervious, because it includes compacted roadbeds and because the vegetation had recently been cut back. Riparian buffer composition and minimum buffer width At each mapped water-collection point, a 16-ha polygon was created that established a zone 200 m on each side of the point and 400 m upstream. Riparian buffer composition was measured as the percentage of forest versus impervious surface within that upstream zone. A separate buffer characteristic, minimum buffer width, was measured as the straight-line distance to the nearest impervious surface upgradient of each sample point. Sampling sites at outlets of the two impoundments were excluded, because their source waters derive from a much broader landscape than their immediate surroundings. Riparian buffer composition and minimum width were regressed against water-quality variables to test for potential relationships. Water sampling and analysis From May 2003 through April 2004, water samples were collected the fi rst week of every month from the ten sampling sites along Ten Mile Creek. Samples were collected without consideration of prior storms or fl ow conditions. Stream temperature and dissolved oxygen were measured, and a 100-ml water sample was collected. Two sites were inaccessible for one or two sampling periods, so the dataset reported contains 117 samples. In the laboratory, water samples were passed through a 0.45-μm Millipore fi lter, and pH was measured using a Thermo Orion pH Electrode (Model 915 7BN). Filtered samples were analyzed for concentrations of major anions (fl uoride, chloride, nitrate, phosphate, and sulfate) and cations (sodium, potassium, ammonium, magnesium, and calcium), using Dionex DX-120 and ICS-90 ion chromatograph systems run with ChromeleonTM software. Protocols followed US EPA method 300.1 (EPA 1997). Climate data, including precipitation and departure from normal precipitation, were obtained for the sample period from the National Climatic 2007 S.S. Madden, G.R. Robinson, and J.G. Arnason 609 Data Center (www.ncdc.noaa.gov/oa/ncdc) maintained by the National Oceanographic and Atmospheric Administration. The Albany Airport weather station (42°45'N, 73°48'W) was used as the climate data source. Although there are several weather stations closer to the study area, Albany Airport had the most complete climate records available for May 2003 through April 2004. Stream invertebrate sampling and analysis We collected stream macroinvertebrates near fi ve of the ten watersampling sites in both early July and early August. Sample locations were selected for similar depth, fl ow, and shade characteristics; three sites were chosen because of their use in previous New York State Department of Environmental Conservation (NYS DEC) monitoring (Abele et al. 1998). Traveling kick samples were used, following protocols of the Stream Biomonitoring Unit of the NYS DEC (Bode at al. 2002). Macroinvertebrates in each sample were counted and identifi ed to order and family to calculate four water-quality indices modifi ed from Bode et al. (2002): Family richness (the number of families identifi ed in each sample), EPT richness (the number of families representing the three orders Ephemeroptera, Plecoptera, and Trichoptera), family biotic index (a measure of the tolerance of the families represented based on established values for sensitivity [Hilsenhoff 1988]), and percent model affi nity (comparing sample community structure to a model non-impacted community). Data from these four indices were converted to a 10-scale scoring regime modifi ed from Bode et al. (2002) for family-level indices (A.J. Smith and R. Bode, NYS DEC, Stream Biomonitoring Unit, Albany, NY, pers. comm.), and samples from July and August were averaged. Once converted, the four indices were then averaged to calculate what Bode et al. (2002) describes as a biological assessment profi le for each sample site, which measures whether water quality at a site is severely, moderately, slightly, or non-impacted. In addition, invertebrate composition among sites was compared on the basis of taxonomic similarity at the family level. Our prediction was that samples would cluster within each of three reaches (above the impoundments, below impoundments, and below the Hamlet). Statistical methods To determine temporal trends, data from all sample sites were pooled and means were calculated for water chemistry variables for each month and months grouped by season (sampling occurred the fi rst week of every month; winter = Jan–Mar, spring = Apr–Jun, summer = Jul–Sept, and autumn = Oct–Dec). Single-factor ANOVA analyses were conducted for all water-chemistry data to test for seasonal variation with season as a factor, with post-hoc Bonferroni tests for pairwise comparison. Spatial trends were determined by plotting the sample-site means in relation to location. We used a semi-variogram analysis to assess spatial autocorrelation for 610 Northeastern Naturalist Vol. 14, No. 4 measurements with high variance among sample sites. Chloride ion concentrations exhibited the highest variance, which reached a threshold (sill) at an inter-sample distance (lag) below 1 km, the approximate mean distance between adjacent sample sites (Table 1). Sodium and chloride ion concentrations were converted to molar equivalents and compared for all pooled water samples using linear regression. Cluster analysis was used to test for spatial variability in the invertebrate community among the fi ve sample sites on Ten Mile Creek. We tested for similarity among all site pairs, using the Curtis-Bray Index, CN = 2jN / (aN + bN), where aN = total number of individuals in site A; bN = total number of individuals in site B; jN = the sum of the lower of the two abundances recorded for species found in both areas (Magurran 1988). To assess spatial pattern in water chemistry, we used discriminant function analysis (DFA) based on the 12 monthly samples at each site for all ten anions and cations. As with invertebrates, the prediction was that sample sites would cluster into three sections, above the main impoundment (I), below the impoundments (II), and below the hamlet (III) (Fig. 1). DFA uses eigenanalysis to identify factor-loading scores for each ion using one or more canonical axes, and to generate spreads of data points corresponding to sample site/period combinations. We used a jackknife iteration option in SYSTAT®11 to estimate probabilities that data points in each of the three pre-determined clusters were correctly assigned (expressed as percent of points nearest the correct centroid for each section). Linear regressions were used to test for relationships between concentrations of selected ions and riparian buffer width and composition. Sodium concentration was regressed on chloride concentration (both in mEq) to test for a potential contribution from de-icing road salt. All statistical analyses were performed with SYSTAT®11 software. Table 1. Impervious surface, forest cover, and the next largest land-cover classifi cation expressed as a percentage of total area of upstream buffer zone (16 ha) and minimum buffer (distance from sample point to nearest upstream impervious surface). Stream km is measured from an estimate of the stream headwater (0.0 km). Sample sites immediately below impoundments are excluded. % (%) Min Stream Stream impervious % forest buffer Site km section cover cover Next largest land cover width (m) 1 0.88 1 2.25 96.76 Wetland (0.99%) 3.2 2 2.57 1 2.03 92.66 Wetland (3.98%) 234.8 3 4.86 1 0.00 80.99 Old fi eld (19.01%) 382.4 5 5.71 1 3.22 92.50 Open water (4.28%) 183.1 8 7.62 2 0.77 91.39 Residential lawn (5.73%) 139.4 9 8.15 3 14.25 54.14 Residential lawn (31.39%) 35.9 10 9.29 3 0.00 97.61 Old fi eld (1.06%) 207.8 2007 S.S. Madden, G.R. Robinson, and J.G. Arnason 611 Results Landscape and riparian buffer characteristics Major land cover of the watershed consisted of mixed deciduous forest (50.17%), coniferous forest (22.86%), agricultural fi eld (9.76%), successional old fi eld (9.85%), open water (1.69%), wetland (1.29%), and meadow (1.07%). Lawns, paved roads, dirt roads, powerlines, rooftops, and parking lots each represented <1% of land cover. Together, these impervious surfaces covered only a small fraction (2.3%) of the study area. Most sample sites had relatively large buffer zones, with little nearby impervious surface and high forest cover (Table 1). Percent impervious surface in the 16-ha buffer zones were all below 3.22%, with the exception of site 9 (below the hamlet) which had 14.25% impervious surface cover. Site 3 and site 10 had 0.0% impervious surface cover in their respective buffer zones. Minimum buffer width is low near the headwaters and at site 9 (below Hamlet of Rensselaerville) due to close proximity of roads. Minimum buffer width was negatively but weakly correlated with percent impervious surface (R = -.517, N = 7, p = .197); a less stringent measure of width would yield a stronger relationship. Seasonal and spatial variation in water chemistry Precipitation during the period of sampling was 24.9 mm greater than normal. Most of the water chemistry sampling dates had precipitation within the seven days prior to sampling, indicating that conditions in the creek were likely above basefl ow. Seven of the sampling months had precipitation in the form of snow in the seven days prior to sampling. Most variation in pH was seasonal, ranging from a mean of 7.1 in winter (January–March) months to 7.8 in summer (June–September). Seasonal variation was signifi cant overall (ANOVA F3,113 = 30.34, p < 0.001), and winter pH was signifi cantly lower than all other seasons (Bonferroni test at 95% level). Phosphate and nitrate concentrations were relatively low (Table 2). Seasonal variation in chloride concentrations was small when all sample sites were pooled (range of seasonal means: 7.1 to 13.3 mg l-1; ANOVA F3,113 = 0.89, p = 0.46), but spatial variation was apparent, with increasing concentrations downstream (Fig. 2). Table 2. Mean concentrations (mg l-1), concentration ranges (maximum = Max; minimum = Min; standard deviation = Sd), and standard error (SE) of major ions for all sample locations and sampling periods pooled (N = 117). “-” = not detectable within our calibration settings. F- Cl- NO3 - PO4 3- SO4 4- Na+ NH4 + K+ Mg2+ Ca2+ Mean 0.02 10.23 0.51 0.02 4.73 7.49 0.23 1.09 1.84 14.41 Sd 0.06 14.34 1.14 0.05 3.48 7.57 0.31 0.78 0.49 4.49 Min - 0.28 - - - 0.81 - 0.18 0.12 0.38 Max 0.54 85.17 9.47 0.28 38.43 42.87 2.94 6.30 3.50 33.37 SE 0.01 1.33 0.11 0.01 0.32 0.70 0.03 0.07 0.05 0.42 612 Northeastern Naturalist Vol. 14, No. 4 Spatial variation was also evident for ion concentrations using DFA (Fig. 3). Much of the multivariate pattern is explained by Factor 1, which is strongly correlated with sodium and chloride, as well as several other cations. As with chloride alone (Fig. 2), the three sections are quite distinct, supporting our predictions that differences would be detectable below the main impoundment and below the hamlet. However, it is interesting that chloride levels were highest at the furthest downstream site, more than 1 km south of the hamlet. Two small streams enter Ten Mile Creek between the hamlet and site 10 (Fig. 1). Although we only collected water-chemistry data on the tributaries from three sampling events, Cl concentrations were consistently greater in the tributaries than at site 10 (Table 3). Figure 2. Mean chloride ion concentrations per sample site along Ten Mile Creek from twelve monthly samples. Error bars represent one Sd. Figure 3. Discriminant-function diagram and standardized canonical discriminantfunction loadings for the 10 major ions collected for 12 months at ten sampling sites along Ten Mile Creek. Ellipses represent 95% confi dence intervals around centroids (mean x-y value per section ± 5%) for data points assigned to each section. Jack-knifed estimates indicate high levels of confi dence in clustering (95% correct for Section 1, 88% for Section 2, and 83% for Section 3). Eigenvalues were 6.58 (Factor 1) and 0.24 (Factor 2). 2007 S.S. Madden, G.R. Robinson, and J.G. Arnason 613 Sources of chloride ion were probably not confined to NaCl. Regressing Na on Cl concentrations produced a slope <1:1 for May–October samples (regression: N = 59, mEq Na = 0.082 + 0.876 [ ±0.038] mEq Cl), and even lower slope for November–April samples (regression: N = 58, mEq Na = 0.104 + 0.784 [±0.024] mEq Cl). Although cation exchange with soils and sediments may explain some of the imbalance, other components of deicing salts (MgCl2 and CaCl2) may have contributed. In addition, calcium chloride is applied to gravel roads and driveways during warm months to reduce dust in the region (R. Wyman, E.N. Huyck Preserve, Rensselaerville, NY, pers. comm.). Spatial variation of invertebrate communities Twenty-nine families of benthic macroinvertebrates, representing 11 orders, were found among the fi ve sample sites. All biotic indices were in high (unimpaired) ranges and values for three sites (Sites 8, 9, and 10) were in the same ranges as previously reported (Table 4; Abele et al. 1998). In contrast to the pattern of increasing downstream chloride contamination, no spatial pattern was evident. However, when examined on the basis of taxonomic composition, sites clustered into three somewhat distinct groups (Fig. 4), corresponding to the pattern observed for major ions (Fig. 3). Buffer properties compared to water quality Our main objective, to examine associations between riparian buffer properties and water-quality indicators, was diffi cult to meet because low Table 4. Benthic invertebrate indices from two pooled samples taken at fi ve sites. Family Rich = family richness; EPT Rich = family richness in the orders Ephemera, Plecoptera, and Trichoptera; Family BI = family biotic index; PMA = percent model affi nity; and BAP = biological assessment profi le. Water quality impact (WQI) for the combined indices is assigned one of four general categories from severe to none (Bode et al. 2002). Sites with asterisks are in proximity to past monitoring locations for the NYS DEC Stream Biomonitoring program and previous values for these indicators are included in parentheses (Abele et al. 1998). NYS DEC data represents species-level indices for richness and biotic index. Index Site 3 Site 5 Site 8* Site 9* Site 10* Family richness 10.0 6.5 6.3 (6.8) 7.8 (6.2) 8.1 (7.1) EPT richness 8.3 7.3 7.0 (5.5) 7.3 (5.5) 7.9 (7.3) Family BI 8.3 8.4 6.9 (6.7) 6.3 (6.8) 7.6 (8.2) PMA 7.7 7.9 8.4 (5.3) 7.3 (4.8) 7.9 (7.3) BAP 8.6 7.5 7.1 (6.1) 7.2 (5.8) 8.0 (7.5) WQI none none/slight none/slight none/slight none Table 3. Chloride concentrations (ppm) measured in two small tributaries to Ten Mile Creek and compared to sample sites above and below where the tributaries enter for the same sample period. Sample site August 2003 December 2003 February 2004 Site 9 12.55 6.47 15.20 Tributary 1 79.98 88.53 16.60 Tributary 2 53.51 104.67 69.19 Site 10 23.86 18.87 34.27 614 Northeastern Naturalist Vol. 14, No. 4 variability in the macroinvertebrate indices offered low statistical power. Chloride ion concentrations were quite variable, but means were poorly correlated with minimum buffer width (N = 7, R = 0.105), forest cover (N = 7; R = 0.110), or total impervious cover (N = 7; R = 0.149). Variability in other major ions was likewise not correlated with buffer properties. Discussion Along a continuum of water quality, forested headwaters contribute relatively pure water that gets degraded downstream by changing land use and new inputs (Groffman et al. 2003, Paul and Meyer 2001). For this rural watershed, we predicted that impoundments and a small residential community would produce measurable variation in an otherwise homogeneous stream. Surrounding land cover, however, was an inadequate predictor of variation, primarily due to unanticipated “leaks” in the putative forest buffer. The highest levels of sodium and chloride were found in the stream reach with the highest percentage of forest cover and maximum distance from impervious surfaces (site 10). Likely sources for sodium and chloride in the stream water are 1) discharge of naturally saline groundwater directly into the hyporheic zone, and 2) road salt in surface runoff or shallow groundwater. Without data on groundwater chemistry, it is not possible to rule out the fi rst source. However, elevated sodium chloride concentrations measured in two small streams draining roadways and entering Ten Mile Creek just above site 10 are consistent with a road-salt source. Furthermore, chloride concentrations peaked during winter months, when deicing salts are applied. This contamination was not revealed in biotic indices derived from macroinvertebrate collections, although there was evidence of differentiation in taxonomic composition. Non-point road runoff of chloride into streams has been most often associated with urban and suburban watersheds, but it is a growing concern in rural watersheds as well (Kaushal et al. 2005). It is particularly interesting that chloride levels at particular locations remained elevated throughout the year, a pattern noted elsewhere (Kaushal et al. 2005). Lovett et al. (2005) and Bastviken et al. (2006) found that forest soils can act as both sinks and sources for chloride, although all the conditions that regulate the retention or release of chloride are not well understood. Figure 4. Additive-cluster diagram based on Curtis-Bray similarity matrix for invertebrate family (N = 29) composition and relative abundance per sampling site. Sites 3 and 5 are above Lake Myosotis (Section I), Site 8 is above the Hamlet of Rensselaerville (Section II), and Sites 9 and 10 are below the hamlet (Section III). 2007 S.S. Madden, G.R. Robinson, and J.G. Arnason 615 While groundwater inputs provide possible explanations for the observed spatial variation in chloride we observed, in rural Ten Mile Creek, non-point runoff seems more likely. The fact that much of our sampling occurred under conditions above basefl ow and during a wet year makes it diffi cult to generalize, but surface runoff seems to be a path for Cl to reach streams throughout the year. Concentrations increased in the vicinity of the hamlet, but continued to increase and reach their highest levels well downstream (>1 km), in a reach with the widest forest buffers. We traced these consistently higher chloride levels to two small feeder creeks with direct contact to nearby roads. Although the watershed has very low road densities, these small tributaries may be serving as direct conduits for non-point-source pollutants. Jackson and Jobbagy (2005) have suggested that a Na:Cl ratio of less than 0.65:1 can be used to infer that the dominant path for Cl reaching streams is through the soil. Our ratios for Ten Mile Creek were much closer to 1:1, indicating that surface fl ow may dominate. Biotic effects of chronic low-level chloride contamination in streams are poorly documented. Work with acute doses has identified sensitive aquatic organisms, but at concentrations well above our highest mean estimates (e.g., caddisfly acute mean value = 4039 mg/L, almost 50 times our highest Cl concentration; reviewed in USEPA 1988). Stream contamination is often reflected in the composition of algal and invertebrate communities (Audette 2004, Demers and Sage 1990, Dickman and Gochnauer 1978, Dussart 1984, Metzeling 1993), and benthic macroinvertebrates are considered relatively sensitive (USEPA 1988), but direct tests of long-term chronic effects of low-level exposures in fluvial systems are rare. Our results for macroinvertebrate indices, which closely matched previous reports from the same locations (Abele et al. 1998), were not correlated with the spatial pattern of significant differences in dissolved chloride. Differences in community composition were consistent with patterns of major-ion concentrations, but many other ecological factors could be driving these rather subtle taxonomic differences. Enhancement and restoration of riparian buffers are important tools for protecting and improving stream water quality (Charbonneau and Resh 1992, Hunter et al. 1999, Purcell and Resh. 2002, Riley 1998). Ordinances designed to reduce non-point source pollution along waterways across the United States require riparian buffers that range in width from 6–60 m, with a median requirement of ±30 m (Castelle et al. 1994, Correll 2000, Fischer et al. 2000, Lee et al. 2004, Semlitsch and Bodie 2003). Slope and vegetation structure are key considerations, as well as the types and levels of potential contamination. We found evidence for road-salt contamination in the most highly buffered reach studied, and this pattern was consistent over 12 months. Unless this watershed is unique in the northeast, riparian buffers, traditionally assumed to function as barriers for diffuse pollution coming off the landscape, may not be fully effective at limiting road-salt contamination from reaching streams. 616 Northeastern Naturalist Vol. 14, No. 4 Acknowledgments We thank Richard Wyman for logistic support and consultation, and Jeena Madden, Laura Audette, Benjamin Dittbrenner, David Newman, and Rebecca Shirer for assistance with field sampling. Robert Bode and A.J. Smith assisted with invertebrate collection methods and identifications. Support was provided by the UAlbany Biodiversity, Conservation, and Policy Program, The Polgar Fellowship Program of the Hudson River Foundation, New York State Department of Environmental Conservation Contract C302745, and the E.N. Huyck Preserve and Biological Field Station. This research represents partial fulfillment of M.Sc. degree requirements for S. Madden. Literature Cited Abele, L.E., R.W. Bode, and M.A. Novak. 1998. Biological stream assessment: Tenmile Creek, 1997 Survey. New York State Department of Environmental Conservation, Albany. NY. Audette, L.C. 2004. The ecological restoration of an urban stream corridor, Patroon Creek, Albany, NY. M.Sc. Thesis. State University of New York at Albany, Albany, NY. Bastviken, D., P. Sanden, T. Svensson, C. Staglberg, M. Magounakis, and G. Oberg. 2006. Chloride retention and release in a boreal forest soil: Effects of soil-water residence time and nitrogen and chloride loads. Environmental Science and Technology 40:2977–2982. Bagdon Environmental Associates. 1989. Town of Rensselaerville comprehensive land-use plan. Prepared by Bagdon Environmental Associates, Inc, 3 Normanskill Boulevard, Delmar, NY. 151 pp. Bode, R.W., M.A. Novak, L.E. Abele, D.L. Heitzman, and A.J. Smith. 2002. Quality assurance work plan for biological stream monitoring in New York State. New York State Department of Environmental Conservation, Albany, NY. Available online at http://www.dec.state.ny.us/website/dow/bwam/sbuqa02.pdf. Accessed April 10, 2007. Capital District Regional Planning Commission. 2005. Draft effects of alternative development scenarios in the Capital District. Available online at http://www. cdrpc.org. Accessed April 10, 2007. Castelle, A.J., A.W. Johnson, and C. Conolly. 1994. Wetland and stream buffer size requirements: A review. Journal of Environmental Quality 23:878–882. Charbonneau R., and V.H. Resh. 1992. Strawberry Creek on the University of California, Berkeley campus: A case history of urban stream restoration. Aquatic Conservation: Marine and Freshwater Ecosystems 2:293–307. Clausen, J.C., K. Guillard, C.M. Sigmund, and K.M. Dors. 2000. Water-quality changes from riparian buffer restoration in Connecticut. Journal of Environmental Quality 29:1751–1761. Correll, D.L. 2000. The current status of our knowledge of riparian-buffer waterquality functions. In P.J. Wigington and R.L. Beschta (Eds.). American Water Resources Association International Conference on Riparian Ecology and Management in Multi-Land Use Watersheds Proceedings. Portland, OR. Demers, C.L., and R.W. Sage, Jr. 1990. Effects of road-deicing salt on aquatic invertebrates in four Adirondack streams. Pp. 245–252, In F.M. D’Itri (Ed.). Chemical Deicers and the Environment. Lewis Publishers, Chelsea, MI. 584 pp. 2007 S.S. Madden, G.R. Robinson, and J.G. Arnason 617 Dickman, M.D., and M.B. Gochnauer. 1978. Impact of sodium chloride on the microbiota of a small stream. Environmental Pollution 49:369–373. Dussart, G.B.J. 1984. Effects of motorway runoff on the ecology of stream algae. Water Pollution Control 83:409–415. 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 85:1. Fischer, R.A., C.O. Martin, and J.C. Fischenich. 2000. Improving riparian buffer strips and corridors for water quality and wildlife. In P.J. Wingington and R.L. Beschta. American Water Resources Association International Conference on Riparian Ecology and Management in Multi-Land Use Watersheds Proceedings. Portland, OR. Groffman, P.M., D.J. Bain, L.E. Band, K.T. Belt, G.S. Brush, J.M. Grove, R.V. Pouyat, I.C. Yesilonis, and W.C. Zipperer. 2003. Down by the riverside: Urban riparian ecology. Frontiers in Ecology and the Environment 1:15–321. Hilsenhoff, W.L. 1988. Rapid fi eld assessment of organic pollution with a familylevel biotic index. Journal of the North American Benthological Sociey 7:65- 68. Hunter, J.C., K.B. Willett, M.C. McCoy, J.F. Quinn, and K.E. Keller. 1999. Prospects for preservation and restoration of riparian forests of the Sacramento Valley, California, USA. Environmental Management 24:65–75. Jackson, R.B., and E.G. Jobbagy. 2005. From icy roads to salty streams. Proceedings of the National Academy of Sciences 102:14487–14488. Jones, B.K., A.C. Neale, M.S. Nash, R.D. Van Remortel, J.D. Wickham, K.H. Ritters, and R.V. O’Neill. 2000. Predicting nutrient and sediment loadings to streams from landscape metrics: A multiple watershed study from the United States Mid-Atlantic Region. Landscape Ecology 16:301–312. Kaushal, S.S., P.M. Groffman, G.E. Likens, K.T. Belt, W.P. Stack, and V.R. Kelly. 2005. Increased salinization of fresh water in the northeastern United States. Proceedings of the National Academy of Sciences 102:13517–13520. Lee, P., C. Smyth, and S. Boutin. 2004. Quantitative review of riparian buffer width guidelines from Canada and the United States. Journal of Environmental Management 70:165–180. Likens, G.E. 2004. Some perspectives on long-term biogeochemical research from the Hubbard Brook ecosystem study. Ecology 85:2355–2362. Lovett, G.M., G.E. Likens, D.C. Buso, C.T. Driscoll, and S.W. Bailey. 2005. The biogeochemistry of chlorine at Hubbard Brook, New Hampshire, USA. Biogeochemistry 72:191–232. Magurran, A.E. 1988. Ecological Diversity and Its Measurement. Princeton University Press, Princeton, NJ. Metzeling, L. 1993. Benthic macroinvertebrate community structure in steams of different salinities. Australian Journal of Marine and Freshwater Research 44:335–351. Mladenoff, D.J., M.A. White, J. Pastor, and T.R. Crow. 1993. Comparing spatial pattern in unaltered old-growth and disturbed forest landscapes. Ecological Applications 3:294–306. O’Neill, R.V., J.R. Krummel, R.H. Gardner, G. Sugihara, B. Jackson, D.L. DeAngelis, B.T. Milne, M.G. Turner, B. Zygmunt, S.W. Christensen, V.H. Dale, and R.L. Graham. 1988. Indices of landscape pattern. Landscape Ecology 1:153–162. 618 Northeastern Naturalist Vol. 14, No. 4 Palmer, M., E. Bernhardt, E. Charesky, S. Collins, A. Dobson, C. Duke, B. Gold, R. Jacobson, S. Kingsland, R. Kranz, M. Mappin, M.L. Martinez, F. Micheli, J. Morse. M. Pace, M. Pascual, S. Palumbi, O.J. Reichman, A. Simmons, A. Townsend, and M. Turner. 2004. Ecology for a crowded planet. Science 304:1251–1252. Paul, M.L., and J.L. Meyer. 2001. Streams in the urban landscape. Annual Review of Ecology and Systematics 32:333–365. Purcell, A.H., C. Friedrich, and V.H. Resh. 2002. An assessment of a small urban stream restoration project in northern California. Restoration Ecology 10:685–694. Riley, A.L. 1998. Restoring Streams in Cities: A Guide for Planners, Policymakers, and Citizens. Island Press, Washington, DC. Roy, A.H., M.C. Freeman, B.J. Freeman, J.J. Wenger, W.E. Ensign, and J.L. Meyer. 2005. Investigating hydrologic alteration as a mechanism of fi sh-assemblage shifts in urbanizing streams. Journal of the North American Benthological Society 24:656–678. Semlitsch, R.D., and J.R. Bodie. 2003. Biological criteria for buffer zones around wetlands and riparian habitats for amphibians and reptiles. Conservation Biology 17:1219–1228. Spruill, T.B. 2000. Statistical evaluation of effects of riparian buffers on nitrate and groundwater quality. Journal Environmental Quality 29:1523–1538. Theobald, D.M. 2004. Placing exurban land-use change in a human-modifi cation framework. Frontiers in Ecology 2:139–144. US Environmental Protection Agency (EPA). 1988. Ambient water quality criteria for chloride. Bulletin EPA 440/5-88-001. Offi ce of Water Regulations and Standards, Criteria and Standards Division, Washington, DC. US Environmental Protection Agency (EPA). 1997. US EPA Method 300.1, Determination of inorganic anions in drinking water by ion chromatography, D.P. Hautman and D.J. Munch, Revision 1.0 (Revision of J.D. Pfaff, US EPA Method 300.0, 1993). National Exposure Research Laboratory, Offi ce of Research and Development, US Environmental Protection Agency, Cincinnati, OH. Walsh, C.J., A.H. Roy, J.W. Feminella, P.D. Cottingham, P.M. Groffman, and R.P. Magan II. 2005. The urban stream syndrome: Current knowledge and the search for a cure. Journal of the North American Benthological Society 24:706–723. Wyman, R.L. (Ed.). 1988. Occasional paper number 1: History of research and a description of the biota and ecological communities of the Edmund Niles Huyck Preserve and Biological Research Station. Edmund Niles Huyck Preserve and Biological Research Station, Rensselaerville, NY.