TECHNICAL REPORTS: LANDSCAPE AND WATERSHED PROCESSES
Sources and Transformations of Nitrate from Streams Draining Varying Land Uses:
Evidence from Dual Isotope Analysis
Douglas A. Burns* U.S. Geological Survey
Elizabeth W. Boyer Pennsylvania State University
Emily M. Elliott University of Pittsburgh
Carol Kendall U.S. Geological Survey
Knowledge of key sources and biogeochemical processes that
affect the transport of nitrate (NO3−) in streams can inform
watershed management strategies for controlling downstream
eutrophication. We applied dual isotope analysis of NO3− to
determine the dominant sources and processes that affect
NO3− concentrations in six stream/river watersheds of different
land uses. Samples were collected monthly at a range of flow
conditions for 15 mo during 2004–05 and analyzed for
NO3− concentrations, δ15NNO3, and δ18ONO3. Samples from
two forested watersheds indicated that NO3− derived from
nitrification was dominant at baseflow. A watershed dominated
by suburban land use had three δ18ONO3 values greater than
+25‰, indicating a large direct contribution of atmospheric
NO3− transported to the stream during some high flows. Two
watersheds with large proportions of agricultural land use had
many δ15NNO3 values greater than +9‰, suggesting an animal
waste source consistent with regional dairy farming practices.
These data showed a linear seasonal pattern with a δ18ONO3:δ
15
NNO3 of 1:2, consistent with seasonally varying denitrification
that peaked in late summer to early fall with the warmest
temperatures and lowest annual streamflow. The large range of
δ 15NNO3 values (10‰) indicates that NO3− supply was likely
not limiting the rate of denitrification, consistent with ground
water and/or in-stream denitrification. Mixing of two or more
distinct sources may have affected the seasonal isotope patterns
observed in these two agricultural streams. In a mixed land use
watershed of large drainage area, none of the source and process
patterns observed in the small streams were evident. These
results emphasize that observations at watersheds of a few to a
few hundred km2 may be necessary to adequately quantify the
relative roles of various NO3− transport and process patterns
that contribute to streamflow in large basins.
Copyright © 2009 by the American Society of Agronomy, Crop Science
Society of America, and Soil Science Society of America. All rights
reserved. No part of this periodical may be reproduced or transmitted
in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system,
without permission in writing from the publisher.
Published in J. Environ. Qual. 38:1149–1159 (2009).
doi:10.2134/jeq2008.0371
Received 18 Aug. 2008.
*Corresponding author (daburns@usgs.gov).
© ASA, CSSA, SSSA
677 S. Segoe Rd., Madison, WI 53711 USA
T
he effects of human activities on the nitrogen (N) cycle
at regional and global scales is the focus of much research
and concern because humans have more than doubled N fluxes
and storage and the rates of many N cycling processes. This
human-induced acceleration of the N cycle is linked to myriad
environmental concerns, including soil acidification, tropospheric
ozone, acute ground water pollution, and estuarine eutrophication
(Galloway et al., 2003). At the regional scale, much work has
focused on the control of nitrate (NO3−) concentrations and
fluxes in riverine environments that range in scale from small
streams to large rivers (Boyer et al., 2002; Smith et al., 2005).
Major uncertainties remain in our understanding of how N from
various sources moves through landscapes to rivers and the extent
to which N cycling processes alter these sources during transport
(Schlesinger et al., 2006).
Dual isotope analysis of NO3− in surface waters can provide
meaningful insights into sources, sinks, and transport processes of
the N cycle that are difficult to obtain through more traditional
measurements of solute concentration and flux. Dual isotope studies of NO3− have proven useful for distinguishing sources such as
atmospheric deposition, human and animal waste, and fertilizer
and for tracing the effects of processes such as nitrification and
denitrification on NO3− pools in waters where measurements of
δ15NNO3 alone cannot distinguish among various sources and processes (Burns and Kendall, 2002; Mayer et al., 2002; Lehmann et
al., 2003; Anisfeld et al., 2007). For example, δ18ONO3 values in atmospheric deposition are quite high, generally greater than +60‰
(Elliott et al., 2007). These high values are believed to result from
reaction of atmospheric NOx with O3 of δ18O greater than +90‰
during formation of HNO3 in the atmosphere (Michalski et al.,
2003). Once atmospheric NO3− is deposited onto the landscape,
the N is readily taken up by biota where it may be mineralized and
nitrified to appear again as NO3−. Nitrifiers are believed to obtain
two oxygen atoms from H2O (δ18O, −25 to −5‰) and one from
O2 (δ18O ~ +23‰) (Amberger and Schmidt, 1987). As a result,
δ18ONO3 derived from nitrification in soils is generally in the range
of −10 to +10‰, and δ18ONO3 is a powerful tracer of the transforD.A. Burns, U.S. Geological Survey, Troy, NY; E.W. Boyer, Pennsylvania State Univ., State
College, PA. E.M. Elliott, Univ. of Pittsburgh, Pittsburgh, PA. C. Kendall, U.S. Geological
Survey, Menlo Park, CA.
Abbreviations: WWTP, wastewater treatment plant.
1149
mation of atmospherically derived NO3− in ecosystems (Burns
and Kendall, 2002; Sebestyen et al., 2008).
Dual NO3− isotope studies show distinct ranges for many N
sources and distinct evolution for many processes as well as overlapping ranges for some sources, indicating that NO3− isotopes
alone cannot always distinguish the principal sources to many
waters (Kendall, 1998; Kendall et al., 2007). However, many
investigations have used dual isotope data, sometimes combined
with other solutes and isotopes, to quantitatively determine
source contributions and/or rates of N cycling processes (Panno et al., 2006; Sebilo et al., 2006). In general, δ15NNO3 values
have proven more useful for distinguishing NO3− derived from
ammonium-bearing fertilizer, soil organic matter, and animal
manure in waters draining agricultural areas, whereas δ18ONO3
values have proven more useful for distinguishing NO3− derived
from nitrate-bearing fertilizer and atmospheric deposition of
NO3− (Kendall et al., 2007). Transformation processes often involve coupled fractionation; for example, denitrification often
results in a 1:2 change in δ18ONO3:δ15NNO3 of the residual pool in
waters, suggesting that fractionation of N is twice that of oxygen
in many environmental settings (Amberger and Schmidt, 1987;
Böttcher et al., 1990).
In the study described herein, we measured NO3− concentrations, δ15NNO3 values, and δ18ONO3 values in six streams and
rivers in New York of widely varying land uses that included
forested (with and without significant wetlands), suburban,
agricultural, and mixed land use. Our intent was to determine
the principal NO3− sources to stream water across these land use
settings. A secondary objective was to examine these data for
evidence of the influence of N cycling processes as a function of
land use. We hypothesized that the isotope data for the forested
watersheds would cluster within a range of δ15NNO3 values of
+3 to +10‰ as previously observed for NO3− derived from soil
nitrification, whereas the data from watersheds with agricultural
and suburban land use would have δ15NNO3 values greater than
+10‰, which would be indicative of an animal or human waste
source. Some samples collected at high flow in all watersheds
were expected to show δ18ONO3 values greater than +15‰, indicating a direct influence of atmospherically deposited NO3−.
In addition to the immediate objectives of this study, we believe
these data collected at small to medium size basin areas may help
improve models of N sources and processes in larger river basins for which data on NO3− concentrations and loads combined
with δ15NNO3 and δ18ONO3 data do not always provide clear evidence of the dominant processes (Mayer et al., 2002).
Materials and Methods
Study Sites
The study sites included (i) Lisha Kill, a watershed dominated
by suburban land use in eastern New York; (ii) Mohawk River, a
mixed land use watershed that drains a large watershed in eastern
and central New York; (iii and iv) Fall Creek and Salmon Creek,
watersheds with abundant agricultural land in central New York;
(v) Biscuit Brook, a forested watershed in the Catskill Mountains; and (vi) Buck Creek, a forested watershed with wetlands in
1150
the Adirondack Mountains. The stream sampling sites are identified on Fig. 1. Table 1 provides some physical and land cover
characteristics of each watershed, and Table 2 describes the geology and soil drainage properties of each watershed.
The Lisha Kill drains a largely suburban watershed with an
impervious area of 13.9%. Only 2.7% of the watershed is classified as high-density developed; the remaining developed land
is of low and medium density as well as open space typical of
suburban landscapes. The watershed includes some patches of
forested land and 29.2% wetland area (Table 1); nearly all the
wetland area is classified as woody wetland and is primarily
located in riparian areas. The watershed is underlain primarily
by dune sands and is excessively drained, the most permeable
of any soils in this study (Table 2). The stream is a second-order
tributary of the Mohawk River.
The Salmon Creek watershed consists of 72.0% agricultural
land (Table 1) of which 46.1% is cultivated crops, and the remainder is classified as pasture/hay. Dairy farming is the dominant agricultural activity in this region of central New York that
includes Salmon and Fall Creeks (Johnson et al., 2007). The
principal cultivated crop is corn, and the principal pasture crop
is alfalfa hay; both are used as cattle feed. This watershed contains 14.5% forested land, a proportion similar to that found
in the Lisha Kill watershed, and about 4% urban land. Salmon
Creek is a tributary to Cayuga Lake, one of the Finger Lakes.
Fall Creek is another tributary of Cayuga Lake, and agriculture
is the dominant land use in this watershed (44.6% of watershed
area; Table 1). Fall Creek has a lower proportion of agricultural
land in its watershed than does Salmon Creek and has different proportions of the two principal types of agricultural land.
In Salmon Creek about two thirds of the agricultural land is in
cultivated crops, whereas in Fall Creek about two thirds of the
agricultural land is in pasture/hay. The Fall Creek watershed is underlain by poorly drained soils, whereas those in Salmon Creek
are described as well drained (Table 2). Two wastewater treatment
plants discharge into Fall Creek, located in the Village of Dryden
and the Village of Freeville, which are permitted for 400,000 and
125,000 gallons/d, respectively (NYSDEC, 2004).
The Mohawk River (at Cohoes, NY) is the largest watershed considered in this study, with a drainage area of 8849 km2
and underlain by diverse bedrock types (Tables 1 and 2). This
mixed land use watershed was selected to determine whether
some of the patterns in NO3− isotopic content observed in the
smaller watersheds would be evident at a larger scale. The Mohawk River watershed is about 50% forested land, 25% agricultural land, and 7.5% urban land; these values are close to
the mean values for the other five sites. Dairy farming is the
principal land use practice on agricultural land, similar to the
two watersheds in central New York. There are several municipal wastewater treatment plants that discharge directly to the
river or to one of its principal tributaries (NYSDEC, 2004).
Biscuit Brook and Buck Creek are dominated by northern
hardwoods, including sugar maple (Acer saccharum), red maple (Acer rubrum L.), yellow birch (Betula alleghaniensis Britton), and American beech (Fagus grandifolia Ehrh.) and minor
amounts of conifers including balsam fir (Abies balsamea (L.)
Journal of Environmental Quality • Volume 38 • May–June 2009
Fig. 1. Map of New York State with hydrography and the locations of the six stream sampling sites.
Table 1. Physical and land cover/use characteristics of the six study stream watersheds. Land cover was determined from the 2002 National Land
Cover Data.† Footnotes describe the NLCD classes used for each category in the table.
Stream
Drainage area Average elevation Percent forested‡ Percent wetlands§ Percent urban¶ Percent agricultural#
km2
m
Lisha Kill
40
105
14.2
29.2
50.9
4.7
Mohawk River
8849
375
50.3
12.1
7.4
24.8
Salmon Creek
229
325
14.5
4.8
3.8
72.0
Fall Creek
327
415
34.0
7.4
6.1
44.6
Biscuit Brook
9.7
871
99.8
0.2
0
0
Buck Creek
3.2
651
69.4
28.4
0
0
† Data from http://www.mrlc.gov/mrlc2k_nlcd.asp.
‡ Sum of Classes C41, C42, and C43.
§ Sum of Classes C11, C90, and C95. All of the watersheds included <2% open water (C11).
¶ Sum of Classes C21, C22, C23, and C24.
# Sum of Classes C81 and C82.
Mill), red spruce (Picea rubens Sarg.), and eastern hemlock (Tsuga Canadensis (L.) Carrière). Additionally, Buck Creek has about
28% wetland area in its drainage, almost all of which is classified
as woody wetland (Table 1; NLCD Class C90). Both forested
watersheds are dominated by well drained soils (Table 2).
Sample and Field Data Collection
Stream water samples were collected approximately monthly at each stream for approximately 15 mo during 2004–05
and analyzed for NO3− concentrations and isotopes. Samples
were collected over a range of flow conditions because streams
commonly show flow-related changes in NO3− concentrations,
and several samples were collected at high flow when sources
may differ from those at low flow (Fig. 2A). For example, the
median probability that daily mean flow was exceeded for sam-
Burns et al.: Land Use and Nitrate Isotopes in Streams
Table 2. Geology and soil drainage properties of the six study stream
watersheds. Data reflect the dominant geology and soil type in
each watershed as described in the references cited in the table.
Stream
Lisha Kill
Mohawk River
Bedrock
type†
shale
–¶
Surficial
geology‡
sand dunes
glacial till
Soil drainage
properties§
excessively drained
moderately well
drained
well drained
poorly drained
well drained
well drained
Salmon Creek
shale
glacial till
Fall Creek
shale
glacial till
Biscuit Brook
sandstone exposed bedrock
Buck Creek
gneiss
glacial till
† Fisher et al. (1971).
‡ Cadwell et al. (1986).
§ Soil Survey Staff (2006)
¶ The Mohawk River has wide variation in bedrock type with sedimentary
rock in the central and southern parts of the basin and igneous and
metamorphic rock in the northern part of the basin.
1151
ples analyzed for NO3− concentrations was <0.50 at all sites,
and several samples were collected at all streams at daily mean
flow conditions that were exceeded <25% of the days based
on an analysis of the 2004–05 flow data at each stream (Fig.
2A). Only about half the samples collected in the forested settings were analyzed for NO3− isotopes because previous studies
at Biscuit Brook in the Catskills (Burns and Kendall, 2002)
and at Archer Creek in the Adirondacks (in a similar mixed
wetland-forested watershed <50 km from Buck Creek; Piatek
et al., 2005) documented the sources and seasonal patterns of
stream NO3− isotopes in these settings.
Samples were collected by rinsing a 250-mL polyethylene
bottle three times with stream water before filling. Samples
were collected near the middle of the channel, except in the
Mohawk River, where water was collected by an automated
sampler through a fixed line with an intake located close to
the stream bank. Samples were stored on ice in a cooler, transferred to a refrigerator (4°C) on returning from the field, and
passed (generally within 24 h of collection) through a 0.4-μm
polycarbonate filter. A 100-mL aliquot of each sample was then
shipped on ice to the U.S. Geological Survey Menlo Park, CA
Stable Isotope and Tritium Laboratory for isotope analysis, and
the remaining sample was stored at 4°C until NO3− analysis.
Monthly samples of precipitation were pooled from weekly
samples collected during 2004 and 2005 by wet only collectors
at 10 National Trends Network sites operated by the National
Atmospheric Deposition Program (NADP/NTN). The procedures for compositing these samples and a discussion of quality
assurance of results from a similar study of archived precipitation samples can be found in Elliott et al. (2007). The 10 sites
chosen for analysis included all of the NADP/NTN sites in New
York that were operated throughout 2004–05 (see location information at http://nadp.sws.uiuc.edu/sites/sitemap.asp?state =
ny). Nitrate isotope data for dry deposition were not available. A
previous study at Biscuit Brook found that neither δ18ONO3 nor
δ15NNO3 values in throughfall (often considered a surrogate for
wet plus dry deposition) were significantly different than those
in wet deposition, suggesting similar isotope values among wet
and dry deposition in rural forested settings (Burns and Kendall,
2002). Despite a lack of data for oxidized nitrogen species in dry
deposition, indirect measurements, such as δ15N in the tissue of
mosses collected near roadways, suggests that vehicle NOx emissions have isotope values that differ from those of NO3− in wet
deposition (Pearson et al., 2000); however, the possible influence
of dry deposition of NOx on the isotopic composition of stream
NO3− could not be evaluated in the current study.
Streamflow data for these streams/rivers were obtained by
gages operated by the U.S. Geological Survey at five of the six
study streams (see data and information at http://waterdata.
usgs.gov/ny/nwis/sw). Salmon Creek had no operating stream
gage until July 2006. To estimate daily discharge data for the
2004–05 study period at Salmon Creek, a linear regression relationship was developed between mean daily discharge at Fall
Creek and Salmon Creek from July 2006 through September
2007 (r2 = 0.83; p < 0.001). This relationship was used to predict a mean daily discharge at Salmon Creek.
1152
Analytical Methods
Samples collected at Biscuit Brook, Buck Creek, Lisha Kill,
and Mohawk River were analyzed for NO3− concentrations by
ion chromatography according to methods described in Lawrence et al. (1995). Quality assurance/quality control methods
and results are described in biannual reports available at http://
ny.water.usgs.gov/publications. The data quality objective for
analytical precision is ±10%, and these objectives are generally
achieved for >90% of analyses of quality assurance samples (Lincoln et al., 2006). The samples collected at Fall Creek and Salmon Creek were also analyzed by ion chromatography at the laboratory of E.W. Boyer at the State University of New York College
of Environmental Science and Forestry in Syracuse, NY.
A denitrifying bacteria, Pseudomonas aureofaciens, was used
to convert 20 to 60 nmoles of NO3− into gaseous N2O before
isotope analysis (Casciotti et al., 2002; Sigman et al., 2001).
Samples were analyzed for δ15N and δ18O in duplicate using
a Micromass Instruments IsoPrime Continuous Flow Isotope
Ratio Mass Spectrometer (use of brand names is for identification purposes only and does not imply endorsement by the U.S.
Government) and values are reported in parts per thousand (‰)
relative to atmospheric N2 and Vienna Standard Mean Ocean
Water, for δ15N and δ18O, respectively, using the equation:
δ (‰) =
(R)sample −(R)standard
× 1000
[1]
(R)standard
where R denotes the ratio of the heavy to light isotope (e.g.,
15
N/14N or 18O/16O).
Samples were corrected using international reference standards USGS34, USGS35, and N3 (Böhlke et al., 2003); linearity and instrument drift were corrected using internal standards.
Sample replicates had an average standard deviation (σ) of 0.2‰
for δ15N (n = 204) and 0.7‰ for δ18O (n = 204). Analytical precision for δ15N was 0.3‰ based on an internal reference standard that was analyzed at least once per 40 samples.
Data Analysis
Streamflow data from the Lisha Kill were separated into a
baseflow and a direct runoff component using a local minimums method of daily flow as calculated by the Base Flow
Index program (http://www.usbr.gov/pmts/hydraulics_lab/
twahl/bfi/). These hydrograph separations were performed
only for days when samples were collected, and the resulting
data were used to assist with interpretation of δ18ONO3 values.
Multivariate stepwise linear regression was used to explore
the role of NO3− availability, stream runoff, and temperature on
δ15NNO3 values at Fall and Salmon Creeks where isotope data
supported a role for denitrification in seasonal NO3− evolution.
These independent variables were chosen because they are among
the dominant environmental controls on denitrification rates for
which data were available in this study. The air temperature data
for this analysis were obtained from the Freeville, NY National
Weather Service site (http://climod.nrcc.cornell.edu).
Pairwise statistical comparisons of isotope data from various
sites were performed with a t test if the data passed tests for
Journal of Environmental Quality • Volume 38 • May–June 2009
normality and equal variance or with the Mann–Whitney rank
sum test (Mann and Whitney, 1947) if the data failed either of
these tests.
Results and Discussion
Stream NO3− Concentrations
Salmon Creek, the watershed with the highest proportion
of agricultural land, had the highest NO3− concentrations
throughout the sampling period, with values that ranged from
about 2 to 7 mg L−1 as NO3−–N (mean, 4.5 mg L−1; Fig. 2B).
The lowest concentrations at Salmon Creek were found in early- to mid-summer, but values otherwise varied little throughout the year.
Fall Creek, with the second greatest proportion of agricultural land, had the second highest NO3− concentrations among
the six streams/rivers with values that ranged from about 0.6 to
1.8 mg L−1 as NO3−–N. The mean NO3− concentration in Fall
Creek during the study was 1.1 mg L−1, close to the value of
1.16 mg L−1 reported for Fall Creek during 2000 through 2005
by Johnson et al. (2007). Fall Creek showed a general tendency
for lower NO3− concentrations during the May–October growing seasons (0.9 mg L−1 as NO3−–N) and higher concentrations in the November–April dormant season (1.3 mg L−1 as
NO3−–N), similar to Salmon Creek, although the seasonal declines from late spring to mid-summer were not as sharp and
distinct as those of Salmon Creek.
Discharge from the two wastewater treatment plants
(WWTPs) in Fall Creek was not likely to have greatly affected
NO3− concentrations in the stream. Combined discharge from
these two plants was provided by New York State (Dept of Environmental Conserv., unpublished data), and the average contribution to streamflow at Fall Creek during the study was 0.35
± 0.21% (1 SD of the mean). Domestic sewage typically has
total N concentrations of 21 to 42 mg L−1, which are reduced
to about 1 to 5 mg L−1 through biological treatment (Sedlak,
1991), such as that found in the largest of the two plants (NYSDEC, 2004). Assuming an average WWTP discharge of total
N into Fall Creek of 5 mg L−1 and complete conversion of this
waste N to NO3− (fairly liberal estimates), the load of NO3− in
Fall Creek from WWTP discharge averaged about 1.6 ± 0.9%
of the total annual load, with a high monthly value of 3.2%.
Additionally, septic systems serve many rural areas of the Fall
and Salmon Creek watersheds and likely contribute additional
NO3− to streams that originate from human waste (Genesee/
Finger Lakes Regional Planning Council, 2001).
Among the other four streams/rivers studied, NO3− concentrations were about 0.5 to 1 mg L−1 as N during the nongrowing season (November–April) despite the wide differences
in land use/landscape cover among these watersheds (Fig.
2B). During the May through October growing season, lower
NO3− stream concentrations were evident in the two streams
(Biscuit Brook and Buck Creek) that drain forested or mixed
forested/wetland landscapes. In contrast, NO3− concentrations
remained fairly constant throughout the sampling period at the
Mohawk River. The suburban Lisha Kill showed greater varia-
Burns et al.: Land Use and Nitrate Isotopes in Streams
Fig. 2. Samples collected and analyzed for nitrate concentrations in
the six study streams during 2004–05. (A) Flow duration curve
with sample flow exceedances marked. (B) NO3−–N stream
concentrations as a function of date.
tion in NO3− concentrations than the Mohawk but less than
that of the two forest-covered watersheds.
General Source Indications from NO3− Isotope Data
Isotope source diagrams such as shown in Fig. 3 indicate the
ranges of δ15NNO3 and δ18ONO3 values reported for various sources
in published studies, and here we use a slight modification of the
figure shown in Kendall et al. (2007). Some general patterns and
differences among the sites are immediately evident on examining
Fig. 3A and 3B. The precipitation data generally lie within the
range reported in previous studies (Elliott et al., 2007; Kendall
et al., 2007), with a mean δ15NNO3 value of –0.8 ‰ (SD 2.7)
and mean δ18ONO3 value of +77.8‰ (SD 8.1). The δ18ONO3 value
is slightly higher than the mean value of +72.0‰ reported for
bulk deposition, throughfall, and snowpack samples collected in
the Adirondacks during 2001–02 (Piatek et al. (2005), about 50
km from Buck Creek) and much higher than the mean value of
+46.0‰ in wet deposition in the Catskills during 1995–97 (Burns
and Kendall, 2002). The higher δ18ONO3 values obtained with the
denitrifier method compared with the previously common closedtube AgNO3 method have also been noted by previous investigators (Revesz and Böhlke, 2002; Kendall et al., 2007).
1153
Fig. 3. Dual NO3− isotope source plot with ranges reported for
various NO3− sources as summarized in Kendall et al. (2007)
for (A) agricultural watersheds and (B) suburban and forested
watersheds. The soil N box represents NO3− derived by nitrification
and is completely encompassed by the waste NO3− box. The area
where these two sources plus the fertilizer and rain NH4+ box
overlap is shaded in gray cross hatch.
Nearly all the stream isotope data lie within the overlapping
fields defined by a soil or human/animal waste source with two
exceptions: (i) four samples from the suburban Lisha Kill had
δ18ONO3 values greater than +20‰ (Fig. 3B), and (ii) about
70% of the samples from Fall and Salmon Creeks had δ15NNO3
values greater than +9‰ (Fig. 3A). These data and the likely
reasons for the observed patterns are discussed in greater detail below. In general, there is a progression from the lowest
δ15NNO3 values in the two forested streams, the next highest in
the suburban stream and mixed land use large river, and the
highest values in the two agricultural streams.
Forested Watersheds
The isotope data from Biscuit Brook and Buck Creek are consistent with nitrification of soil N as the dominant source to these
streams similar to previous findings in these two regions (Burns
and Kendall, 2002; Piatek et al., 2005). These data do not show
evidence of a large direct contribution of atmospheric NO3− to
these streams, similar to many previous studies in forested wa-
1154
tersheds (Kendall et al., 1995; Ohte et al., 2004; Sebestyen et
al., 2008). The isotope values from the two streams overlap to
a great extent and are essentially indistinguishable (Fig. 3B); the
mean δ18ONO3 and δ15NNO3 values at Biscuit Brook were +2.8
and +1.9‰, respectively, whereas those at Buck Creek were
+3.8 and +1.9%. Because the Buck Creek watershed includes
>25% wetland area, there is a possibility that denitrification may
have affected the evolution of the NO3− pool that is ultimately
transported in the stream. These NO3− isotope data, however,
provide no evidence to support this possibility. None of the
samples show noticeably higher isotope values, and the data do
not follow a δ18ONO3:δ15NNO3 line with a slope of 1:2 that might
indicate a denitrified NO3− pool. This result is not surprising
because wetlands in the Adirondack region do not always show
a strong hydrologic connection to the stream; wetland soils are
commonly dense with low hydraulic conductivity and may have
little effect on stream N fluxes (McHale et al., 2004).
The δ15NNO3 values are similar to those reported previously for streams that drain forested watersheds in these two
regions of New York. Burns and Kendall (2002) reported a
mean δ15NNO3 value of +2.3‰ for Biscuit Brook and another
nearby stream in the Catskills, and Piatek et al. (2005) reported
a mean value of +1.2‰ for a stream in the Adirondacks near
Buck Creek. About half of the δ15NNO3 values from the current
study are lower than would be indicated by a soil N source
end member that extends only to a value of about +2‰, as in
Kendall et al. (2007). These data suggest that the soil source
box represented in figures such as that in Kendall et al. (2007)
ought to extend to a δ15NNO3 value of about 0‰.
Large differences were found among δ18ONO3 values from the
current investigation and those from previous investigations in the
same or nearby streams in the Catskills and Adirondacks. Burns
and Kendall (2002) reported a mean δ18ONO3 value of +17.7‰,
and Piatek et al. (2005) reported a mean value of +10.4‰, both
substantially higher than the values reported here. Additionally,
Williard et al. (2001) reported a mean δ18ONO3 value of +13.2‰
for nine forested watersheds that did not have a history of farming or fire. These results may seem puzzling because in the case of
precipitation, values obtained with the older closed-tube methods were less than those obtained with the denitrifier method,
whereas in the case of stream water, the older values are higher
than recent values from the same or similar watersheds. These
results, however, are consistent with a study that showed O from
CO2 produced by combustion in closed glass tubes can exchange
with O in the glass and that the contaminant O likely has a
δ18O value in the range of +15 to +20‰ (Revesz and Böhlke,
2002), which would lower the δ18ONO3 value in precipitation
but increase the δ18ONO3 value in stream water. Therefore, the
results reported here are consistent with a high bias from such a
contaminant source. However, we are not certain why these new
results differ from past studies in similar environments, and possible explanations include natural variation, incomplete removal
of DOC from the samples analyzed with the older method and
hence contamination of the Ag2O with O from the DOC, and
the lack of international standards for normalizing δ18O values
(Revesz and Böhlke, 2002; Kendall et al., 2007).
Journal of Environmental Quality • Volume 38 • May–June 2009
Suburban Watershed
The isotope data from the suburban Lisha Kill generally lie
within a range consistent with a nitrified soil N source but could
also be affected by a waste source as well (Fig. 3B). The δ15NNO3
values of the Lisha Kill are higher than those from the two forested streams (p < 0.001), with values that range from about +4 to
+8‰, which suggests the possibility of a waste source. Because
the source boxes for a nitrified source and waste source overlap
in this range, the isotope data are not definitive in identifying a
likely waste source. Residences and buildings within the Lisha
Kill are all connected to WWTPs that discharge outside of the
watershed, so a large NO3− source from waste as might be found
in residential areas with septic systems (Burns et al., 2005) was
not expected in the Lisha Kill. Nonetheless, numerous sources of
N are commonly found in residential areas, including fertilizers
applied to lawns and golf courses, leaky sewer lines, and waste
from pets (Paul and Meyer, 2001). Previous investigations of the
Lisha Kill have found measurable concentrations of insecticides
and herbicides that are commonly applied to residential lawns
(Phillips et al., 2002); this same vector of transport would be
expected to transport NO3− associated with fertilizer and animal
waste to the stream.
The data in Fig. 3B show that four samples collected in the
suburban Lisha Kill lie outside the range of δ18ONO3 values expected from soil or waste sources, suggesting a significant direct
contribution of atmospheric NO3− to the stream (one of these
lies within the range for NO3−–bearing fertilizer). Application
of a simple two end-member mixing model, in which the mean
atmospheric δ18ONO3 value of +77.8‰ represents one end member and the median value of δ18ONO3 of +5.7‰ in stream water represents the other, indicates that the three samples with
δ18ONO3 values greater than +25‰ have a direct atmospheric
contribution to stream NO3− that ranges from about 40 to 53%.
Several previous isotope investigations have shown that direct
atmospheric NO3− may be dominant in stream water in forested
(Burns and Kendall, 2002; Ohte et al., 2004; Sebestyen et al.,
2008) and urban watersheds (Silva et al., 2002; Anisfeld et al.,
2007). These large direct contributions of atmospheric NO3−
have been observed only during high flow events and tend to be
transient in nature, with the dominant NO3− source at baseflow
quickly reimposing its isotopic signature in stream water soon
after peakflow conditions have waned.
To further explore these changes in the NO3− isotope signature, the δ18ONO3 values at Lisha Kill are shown as a function of
stream runoff, event flow, and NO3− concentration (Fig. 4). A
statistically significant relationship between stream runoff and
δ18ONO3 is evident (r2 = 0.35; p = 0.025) (Fig. 4A), but this relationship is highly leveraged by the sample collected at the highest flow conditions during the snowmelt of 2005. If this sample
is removed and the regression re-calculated, a relationship was
not apparent (p = 0.552) between these two variables.
The Lisha Kill watershed consists of about 14% impervious
area with storm sewers that discharge directly into the stream or
into small ephemeral tributaries. This network of paved surfaces
and storm sewers may allow atmospheric NO3− to bypass soil
Burns et al.: Land Use and Nitrate Isotopes in Streams
contact whenever a rainfall or snowmelt event occurs in the watershed regardless of the size of the event and the gross amount
of streamflow. To evaluate this rapid runoff effect, we calculated
direct runoff with the Base Flow Index program and found that
this quantity was significantly related to δ18ONO3 values on the
day of sample collection (Fig. 4B). Although the r2 value of 0.34
was similar to that obtained for the relationship of δ18ONO3 to
bulk stream runoff (Fig. 4A), the relationship with direct runoff
was not highly leveraged like that of bulk stream runoff. Additionally, the three samples with the highest δ18ONO3 values were
collected when the direct runoff component was greater than
approximately 50% of total streamflow. However, there were
three other samples collected when similar high proportions of
direct runoff were present but for which δ18ONO3 values were
quite low. Although the hydrograph separation method slightly
improved the interpretation of these data by removing the leveraging in the relationship, these results also indicate that stream
NO3− in this suburban setting does not always consist of a high
proportion of atmospheric NO3− when direct runoff proportions
are high. Other factors, such as rainfall intensity, season, and the
timing of fertilizer application, likely play a role in the watershed
NO3− response, but we were unable to quantify these factors in
the current study. The Lisha Kill also contains abundant riparian wetlands, and these landscape features have previously been
shown to greatly affect stream and solute runoff in suburban
settings through storage that varies with season and antecedent
moisture conditions (Burns et al., 2005).
A weak inverse relationship between NO3− concentrations
and δ18ONO3 values (r2 = 0.23; p = 0.086) was observed in the
Lisha Kill (Fig. 4C). The highest δ18ONO3 values were associated with some of the lowest NO3− concentrations measured.
These data contrast with those in forested settings where the
highest δ18ONO3 values are typically found in the samples with
the highest NO3− concentrations, which are often associated
with spring snowmelt (Burns and Kendall, 2002; Ohte et al.,
2004). In the suburban setting of the Lisha Kill, stormflow has
a tendency to dilute NO3− concentrations relative to baseflow,
but this relationship is not strong. Other studies in residential
suburban settings have shown a similar tendency for dilution
of stream NO3− concentrations during stormflow (Hook and
Yeakley, 2005; Kim et al., 2007), although the storm response
can be highly variable and depends on the relative concentrations of NO3− in ground water and precipitation, the relative
amount and connectivity of impermeable surfaces, the presence of septic systems, and the seasonality with respect to fertilizer application among other factors (Silva et al., 2002; Hook
and Yeakley, 2005; Poor and McDonnell, 2007).
Agricultural Watersheds
Most of the samples collected at Fall and Salmon Creeks
showed high δ15NNO3 values that suggest a large NO3− source from
animal/human waste in these watersheds, consistent with estimates of the dominant stream NO3− sources identified in previous
studies in these watersheds (Likens and Bormann, 1974; Johnson
et al., 1976; Johnson et al., 2007). All of the samples collected at
Fall and Salmon Creeks had δ15NNO3 values greater than +5‰,
1155
Fig. 4. δ18ONO3 values in the suburban Lisha Kill watershed as a
function of (A) stream runoff, (B) percent event flow, and (C)
NO3− concentrations.
higher than the range previously reported for waters affected by nitrification of NH4+ fertilizer (Kendall et al., 2007), suggesting that
fertilizer is not a dominant source of NO3− in these watersheds or
that a fertilizer isotope signal is altered by mixing with a source
with higher δ15N values and/or biogeochemical processes before
stream sample collection. Fertilizer N imports into the Fall Creek
watershed are estimated to be about 6 kg ha−1 yr−1, less than half
the annual animal manure return of 14 kg N ha−1 yr−1 (Johnson
et al., 2007). Additionally, 56% of samples collected at Fall Creek
and 94% of samples collected at Salmon Creek had δ15NNO3 values greater than +9‰, outside the range previously reported for
NO3− derived by nitrification of soil organic matter, further suggesting the dominance of a direct waste source.
The relative contributions of animal vs. human waste to
stream NO3− in Salmon and Fall Creeks is not known and cannot be determined from isotope and concentration data alone;
however, Johnson et al. (1976) estimated that human waste
(treated sewage plus septic waste) contributed about 17% of
the annual export of NO3− in Fall Creek and that agricultural
1156
activities contributed about 61% in the mid-1970s. Salmon
Creek, with a higher proportion of agricultural land, less urban land, and a lower population density than Fall Creek, is
likely to show an even greater dominance of animal waste as
a source of stream NO3−. Despite land use and land management changes in Fall Creek since the 1970s that include an
increase in population, addition of a WWTP in Freeville, increased dairy production, and improved manure management
practices, agricultural activities are still believed to be the principal source of stream NO3− (Johnson et al., 2007), although
improved animal waste management since the 1990s may have
reduced this source relative to that of human waste.
A distinctive characteristic of these data is the linear relationship between δ18ONO3 and δ15NNO3 values at Fall and Salmon Creeks (Fig. 3A). The slope of this relationship is 0.46 (r2 =
0.54; p < 0.001) (Fig. 5A), which is very close to the 1:2 slope
expected for a pool of NO3− that has been partially denitrified,
a relationship supported by empirical evidence and theoretical kinetics and generally considered unambiguous evidence of
denitrification in fresh waters (Böttcher et al., 1990; Chen and
MacQuarrie, 2005; Panno et al., 2006). The influence of denitrification would likely be least evident under high flow conditions when rapid runoff processes might transport atmospheric
NO3− and freshly nitrified manure sources to these streams.
On this basis, the highest flow samples were removed from the
data set (those with a probability of exceedance of <0.15 based
on 2003–05 flow data at Fall Creek). The resulting r2 value
increased to 0.78, and the slope increased slightly to 0.51 (Fig.
5B), even closer to the 1:2 slope expected based on first-order
kinetics and laboratory reaction rate constants for denitrification (Olleros, 1983; Chen and MacQuarrie, 2005). The influence of denitrification seems to persist through the growing
(May–October) and dormant (November–April) seasons, although the variability about the regression line is greater in the
dormant season, as might be expected when the influence of
denitrification should be at a minimum (Fig. 5C and 5D).
Multivariate regression with NO3− concentrations, stream
runoff, and air temperature as potential independent variables
yielded the following relationships:
Fall Creek δ15NNO3 = −1.388 NO3− conc. – 0.606 Runoff +
0.0679 Air Temp, r2 = 0.87
Salmon Creek δ15NNO3 = –0.904 Runoff + 0.197 Air
Temp, r2 = 0.84 (NO3− conc. was not significant [p >
0.05] for Salmon Creek).
These regressions indicate that δ15NNO3 values tended to
increase as stream runoff decreased and air temperature increased and are consistent with a seasonally varying influence
of denitrification that was greatest at warm temperatures when
streamflow was lowest (and in the case of Fall Creek when
NO3− concentrations were lowest).
The large range of δ15NNO3 values observed in these streams
suggests a large fractionation associated with denitrification
(10‰), indicative of so-called “riparian” denitrification in which
the supply of NO3− was not likely limiting the rate of denitri-
Journal of Environmental Quality • Volume 38 • May–June 2009
fication, whereas low fractionation values would suggest that
the rate of denitrification was limited by NO3− supply, which
is characterized by much lower fractionation values (Sebilo et
al., 2003; Lehmann et al., 2003). These data cannot identify
whether denitrification was primarily occurring during ground
water transport or within the stream environment; however, results of a previous study at Fall Creek could not find significant
in-stream NO3− sinks, suggesting that the ground water environment may be the primary zone of denitrification in this watershed (Johnson et al., 1976). Additionally, some denitrification
may have occurred in other environments, such as manure piles
and lagoons or ponds that capture dairy runoff.
Systematic linear variation of δ15NNO3 and δ18ONO3 values is
not always the result of denitrification and may also originate from
variable amounts of mixing of soil NO3− with denitrified water or
by algal NO3− uptake (Kendall et al., 2007). The dominance of
mixing vs. fractionation on variations in N isotopic content can
sometimes be determined through contrasting plots of δ15NNO3
vs. NO3− concentration. Mixing of two solutions with different
δ15NNO3 values and NO3− concentrations is only linear when isotope values are plotted as a function of 1/NO3− concentration
(Kendall, 1998), whereas isotope fractionations observed in biological systems can be approximated as Rayleigh fractionations,
which are exponential and only linear when isotope values are
plotted as a function of ln NO3− concentration. Such plots indicate that these data are consistent with mixing and denitrification
in Fall and Salmon Creeks; neither process can be definitively
ruled out based on the relationships evident in these data (Fig.
6). The Fall Creek data appear more strongly linear (r2 = 0.59; p <
0.001) than those of Salmon Creek (r2 = 0.49; p = 0.001) in the
mixing plot (Fig. 6A), whereas the data from Salmon Creek have
a stronger linear relationship (r2 = 0.62; p < 0.001) than those of
Fall Creek in the fractionation plot (r2 = 0.46; p = 0.002). However, highly significant statistical relationships are present for both
streams in each of the plots. Mixing with another source may
affect Fall Creek to a greater extent than Salmon Creek because
of the greater proportion of forested land and slight influence of
wastewater NO3− discharge in Fall Creek.
Mixing of two or more NO3− sources and denitrification are
not necessarily mutually exclusive, and both processes may be affecting the seasonal evolution of NO3− in these streams. Ground
water, with seasonally varying amounts of denitrification, is
likely providing NO3− to these streams from agricultural parts of
the landscape, but forested and urban parts of the landscape are
also likely contributing NO3− to these streams, as estimated by
Johnson et al. (1976) for Fall Creek in the 1970s. A quantitative
determination of the extent to which denitrification and mixing
of waters from different sources affected the concentrations and
isotopic composition of NO3− in these two agricultural watersheds is not possible in this study and requires a combination of
additional chemistry, tracer, and hydrometric data.
Mixed Land Use Watershed
The Mohawk River watershed contains a mixture of forested,
agricultural, and urban land use in proportions that are approximately equal to the mean values among the other five study wa-
Burns et al.: Land Use and Nitrate Isotopes in Streams
Fig. 5. δ15NNO3 vs. δ18ONO3 in the agricultural Fall and Salmon Creek
watersheds for (A) all data, (B) low flow, (C) growing season, and
(D) dormant season.
tersheds. The drainage area of the Mohawk watershed, however,
exceeds that of the others in the study by one to three orders of
magnitude. This large size likely explains why the NO3− isotope
data from the Mohawk River do not show clear indications of
a waste source, such as in Fall Creek and Salmon Creek, or an
indication of unaltered NO3− from precipitation, as in the Lisha
Kill. Mixing of NO3− from multiple sources combined with long
travel times and greater opportunity for alteration through biogeochemical processes may explain why large rivers such as the
Mohawk do not show clear indications of sources such as were
found in the smaller rivers and streams in this study (Mayer et
al., 2002). Our results suggest that a drainage area of <500 km2
may be most appropriate for identifying sources and processes
that affect the evolution of NO3− in riverine environments.
Summary
Dual isotope data revealed varying sources and processes that
affect NO3− concentrations among six streams/watersheds of different land uses in New York. A suburban watershed with no
septic or wastewater influence showed NO3− concentrations only
slightly greater than those of two forested watersheds but slightly
higher δ15NNO3 values that may reflect some influence from a
1157
suburban stream nor the high δ15NNO3 values and 1:2 sloping line
of the agricultural streams. This finding suggests that in mixed
land use watersheds at large basin scales, the complex mixture of
NO3− sources combined with additional travel time may obscure
the process- and source-level information that can be observed in
NO3− dual isotope patterns at smaller drainage scales. The small to
medium drainage area scale in streams dominated by one form of
land use is likely the most effective scale for distinguishing the role
of N sources and hydrologic and biogeochemical processes on the
transport of NO3− in riverine environments.
Acknowledgments
This work was supported by a grant from the New York State
Energy Research and Development Authority. The authors
thank Elizabeth Nystrom, Heather Golden, and Michael
Brown for help with sample collection and Gregory Lawrence
and Michael McHale for sharing data.
References
Fig. 6. (A) Mixing plot, 1/NO3− concentration and δ15N NO3. (B) Isotope
fractionation plot, ln NO3− concentration, and δ15NNO3 for Fall and
Salmon Creeks.
waste source. Three high flow samples from the suburban stream
had δ18ONO3 values greater than +25‰, indicating that about
50% of the NO3− was from a direct atmospheric source and reflected some bypassing of soil contact via impermeable surfaces
and the network of storm sewers in the watershed.
Two of the streams with large proportions of agricultural
land had the highest NO3− concentrations in this study and a
source signature consistent with animal waste from dairy farming in the watersheds. These isotope data showed a linear relationship of δ18ONO3:δ15NNO3 of about 1:2, consistent with variable amounts of denitrification that likely occurred outside of
the stream environment. The effects of denitrification appeared
greatest (based on multivariate regression of δ15NNO3 values)
when air temperature was warmest and streamflow was lowest during summer and early fall, coincident with the lowest
stream NO3− concentrations. The NO3− in these streams may
also have been transported from two or more distinct sources,
but this possibility could not be fully evaluated with these data
and would require additional hydrometric and/or tracer data.
Nitrate isotope data from the Mohawk River, a basin of nearly
9000 km2 that drains agricultural, urban, and forested land in
proportions approximately equal to the average of the other five
study watersheds, showed neither the high δ18ONO3 values of the
1158
Amberger, A., and H.-L. Schmidt. 1987. Natürliche isotopengehalte von
nitrate als indikatoren für dessen herkunft. Geochim. Cosmochim. Acta
51:2699–2705.
Anisfeld, S.C., R.T. Barnes, M.A. Altabet, and W. Taixing. 2007. Isotopic
apportionment of atmospheric and sewage nitrogen sources in two
Connecticut rivers. Environ. Sci. Technol. 41:6363–6369.
Böhlke, J.-K., S.J. Mroczkowski, and T.B. Coplen. 2003. Oxygen isotopes
in nitrate: New reference materials for 18O:17O:16O measurements and
observations on nitrate-water equilibration. Rapid Commun. Mass
Spectrom. 17:1835–1846.
Böttcher, J., O. Strebel, S. Voerkelius, and H.-L. Schmidt. 1990. Using isotope
fractionation of nitrate-nitrogen and nitrate-oxygen for evaluation of
microbial denitrification in a sandy aquifer. J. Hydrol. 114:413–424.
Boyer, E.W., C.L. Goodale, N.A. Jaworski, and R.W. Howarth. 2002.
Anthropogenic nitrogen sources and relationships to riverine nitrogen
export in the northeastern USA. Bigeochemistry 57/58:137–169.
Burns, D.A., and C. Kendall. 2002. Analysis of δ18O and δ15N to differentiate
NO3− sources in runoff at two watersheds in the Catskill Mountains of
New York. Water Resour. Res. 38:1051, doi:10.1029/2001WR000292.
Burns, D.A., T. Vitvar, J.J. McDonnell, J. Hassett, J. Duncan, and C. Kendall.
2005. Effects of suburban development on runoff generation in the
Croton River basin, New York, USA. J. Hydrol. 311:266–281.
Cadwell, D.H., et al. 1986. Surficial geologic map of New York. N.Y. State
Museum Geological Survey, Map and Chart Ser. 40, Albany, NY.
Casciotti, K.L., D.M. Sigman, M.G. Hastings, J.K. Böhlke, and A. Hilkert.
2002. Measurement of the oxygen isotopic composition of nitrate in
seawater and freshwater using the denitrifier method. Anal. Chem.
74:4905–4912.
Chen, D.J.Z., and K.T.B. MacQuarrie. 2005. Correlation of δ15N and δ18O
in NO3− during denitrification in groundwater. J. Environ. Eng. Sci.
4:221–226.
Elliott, E.M., C. Kendall, E.W. Boyer, D.A. Burns, S.D. Wankel, D.J. Bain, K.
Harlin, T.J. Butler, and R. Carlton. 2007. An isotopic tracer of stationary
source NOx emissions across the midwestern and northeastern United
States. Environ. Sci. Technol. 41:7661–7667.
Fisher, D.W., Y.W. Isachsen, and L.V. Rickard. 1971. Generalized tectonic–
metamorphic map of New York, N.Y. State Museum and Science Serv.,
Map and Chart Ser. 15.
Galloway, J.N., J.D. Aber, J.W. Erisman, S.P. Seitzinger, R.W. Howarth,
E.B. Cowling, and B.J. Cosby. 2003. The nitrogen cascade. Bioscience
53:341–356.
Genesee/Finger Lakes Regional Planning Council. 2001. Cayuga Lake
Restoration and Protection Plan. Available at www.cayugawatershed.org
(verified 1 Mar. 2009).
Hook, A.M., and J.A. Yeakley. 2005. Stormflow dynamics of dissolved
organic carbon and total dissolved nitrogen in a small urban watershed.
Biogeochemistry 75:409–431.
Journal of Environmental Quality • Volume 38 • May–June 2009
Johnson, A.H., D.R. Bouldin, E.A. Goyette, and A.M. Hedges. 1976. Nitrate
dynamics in Fall Creek, New York. J. Environ. Qual. 5:386–391.
Johnson, M.S., P.B. Woodbury, A.N. Pell, and J. Lehmann. 2007. Land-use
change and stream water fluxes: Decadal dynamics in watershed nitrate
exports. Ecosystems 10:1182–1196.
Kendall, C. 1998. Tracing nitrogen sources and cycles in catchments. p.
519–576. In C. Kendall and J.J. McDonnell (ed.) Isotope tracers in
catchment hydrology. Elsevier, Amsterdam, The Netherlands.
Kendall, C., E.M. Elliott, and S.D. Wankel. 2007. Tracing anthropogenic
inputs of nitrogen to ecosystems. p. 375–449. In R.H. Michener and K.
Lajtha (ed.) Stable isotopes in ecology and environmental studies, 2nd
ed. Wiley-Blackwell Publishing, Oxford, UK.
Kendall, C., D.H. Campbell, D.A. Burns, J.B. Shanley, S.R. Silva, and C.C.Y.
Chang. 1995 Tracing sources of nitrate in snowmelt runoff using the
oxygen and nitrogen isotopic compositions of nitrate. p. 339–347. In
K.A. Tonnessen, M.W. Williams, and M. Trantner (ed.) Biogeochemistry
of seasonally snow-covered catchments. I.A.H.S. Publication 228,
Wallingford, UK.
Kim, L.-H., S.-O. Ko, S. Jeong, and J. Yoon. 2007. Characteristics of washedoff pollutants and dynamic EMCs in parking lots and bridges during a
storm. Sci. Total Environ. 376:178–184.
Lawrence, G.B., T.A. Lincoln, D.R. Horan-Ross, M.L. Olson, and L.A.
Waldron. 1995. Analytical methods of the U.S. Geological Survey’s New
York District Water-Analysis Laboratory. US Geol. Surv. Open-File Rpt.
95–416. Troy, NY, 78 p.
Lehmann, M.F., P. Reichert, S.M. Bernasconi, A. Barbieri, and J.A.
McKenzie. 2003. Modeling nitrogen and oxygen isotope fractionation
during denitrification in a lacustrine redox-transition zone. Geochim.
Cosmochim. Acta 67:2529–2542.
Likens, G.E., and F.H. Bormann. 1974. Linkages between terrestrial and
aquatic ecosystems. Bioscience 24:447–456.
Lincoln, T.A., D.A. Horan-Ross, M.R. McHale, and G.B. Lawrence. 2006.
Quality-assurance data for routine water analyses by the U.S. Geological
Survey Laboratory in Troy, New York—July 1999 through June 2001.
U.S. Geol. Surv. Open-File Rpt. 2006–1246, Reston, VA.
Mann, H.B., and D.R. Whitney. 1947. On a test of whether one of two
random variables is statistically larger than the other. Ann. Math. Stat.
18:50–60.
Mayer, B., E.W. Boyer, C. Goodale, N.A. Jaworski, N. Van Breemen, R.W.
Howarth, S. Seitzinger, G. Billen, K. Lajtha, K. Nadelhoffer, D. Van
Dam, L.J. Hetling, M. Nosal, and K. Paustian. 2002. Sources of nitrate
in rivers draining sixteen watersheds in the northeastern U.S.: Isotopic
constraints. Biogeochemistry 57/58:171–197.
McHale, M.R., C.P. Cirmo, M.R. Mitchell, and J.J. McDonnell. 2004.
Wetland dynamics in an Adirondack forested watershed. Hydrol.
Processes 18:1853–1870.
Michalski, G., Z. Scott, M. Kabiling, and M.H. Thiemens. 2003. First
measurements and modeling of Δ17O in atmospheric nitrate. Geophys.
Res. Lett. 30, doi:10.1029GL017015.
NYSDEC. 2004. Descriptive data of municipal wastewater treatment plants
in New York State. New York State Department of Environmental
Conservation, Div. of Water, Bureau of Water Compliance, Albany, NY.
Ohte, N., S.D. Sebestyen, J.B. Shanley, D.H. Doctor, C. Kendall, S.D.
Wankel, and E.W. Boyer. 2004. Tracing sources of nitrate in snowmelt
runoff using a high-resolution isotopic technique. Geophys. Res. Lett.
Burns et al.: Land Use and Nitrate Isotopes in Streams
31:L21506, doi:10.1029/2004GL020908.
Olleros, T. 1983. Kinetic isotope effects of the enzymatic splitting of arginine
and nitrate; a contribution to the explanation of the reaction mechanism.
Dissertation, Technical University Munchen-Weihenstephan.
Panno, S.V., K.C. Hackley, W.R. Kelly, and H.-H. Hwang. 2006. Isotopic
evidence of nitrate sources and denitrification in the Mississippi River,
Illinois. J. Environ. Qual. 35:495–504.
Paul, M.J., and J.L. Meyer. 2001. Streams in the urban landscape. Annu. Rev.
Ecol. Syst. 32:333–365.
Pearson, J., D.M. Wells, K.J. Seller, A. Bennett, A. Soares, J. Woodall, and M.J.
Ingrouille. 2000. Traffic exposure increases natural 15N and heavy metal
concentrations in mosses. New Phytol. 147:317–326.
Phillips, P.J., D.A. Eckhardt, D.A. Freehafer, G.R. Wall, and H.H. Ingleston.
2002. Regional patterns of pesticide concentrations in surface waters of
New York in 1997. J. Am. Water Resour. Assoc. 38:731–745.
Piatek, K.B., M.J. Mitchell, S.R. Silva, and C. Kendall. 2005. Sources of
nitrate in snowmelt discharge: Evidence from water chemistry and stable
isotopes of nitrate. Water Air Soil Pollut. 165:13–35.
Poor, C.J., and J.J. McDonnell. 2007. The effects of land use on stream nitrate
dynamics. J. Hydrol. 332:54–68.
Revesz, K., and J.K. Böhlke. 2002. Comparison of δ18O measurements in nitrate
by different combustion techniques. Anal. Chem. 74:5410–5413.
Schlesinger, W.H., K.H. Reckhow, and E.S. Bernhardt. 2006. Global
change: The nitrogen cycle and rivers. Water Resour. Res. 42:W03806,
doi:10.1029/2005WR004300.
Sebestyen, S.D., E.W. Boyer, J.B. Shanley, C. Kendall, D.H. Doctor, G.R.
Aiken, and N. Ohte. 2008. Sources, transformations, and hydrological
processes that control stream nitrate and dissolved organic matter
concentrations during snowmelt at an upland forest. Water Resour. Res.
44, W12410, doi: 10.1029/2008WR006983.
Sebilo, M., G. Billen, B. Mayer, D. Billiou, M. Grably, J. Garnier, and A.
Mariotti. 2006. Assessing nitrification and denitrification in the Seine
River and estuary using chemical and isotopic techniques. Ecosystems
9:564–577.
Sebilo, M., G. Billen, M. Grably, and A. Mariotti. 2003. Isotopic composition
of nitrate-nitrogen as a marker of riparian and benthic denitrification at
the scale of the whole Seine River system. Biogeochemistry 63:35–51.
Sedlak, R.L. (ed.) 1991. Phosphorus and nitrogen removal from municipal
wastewater: Principles and practice. Second ed. Lewis Publishers, Boca
Raton, FL.
Sigman, D.M., K.L. Casciotti, M. Andreani, C. Barford, M. Galanter, and J.K.
Böhlke. 2001. A bacterial method for the nitrogen isotopic analysis of
nitrate in seawater and freshwater. Anal. Chem. 73:4145–4153.
Silva, S.R., P.B. Ging, R.W. Lee, J.C. Ebbert, A.J. Tesoriero, and E.L. Inkpen.
2002. Forensic applications of nitrogen and oxygen isotopes in tracing
sources in urban environments. Environ. Foren. 3:125–130.
Smith, S.V., D.P. Swaney, R.W. Buddemeier, M.R. Scarsbrook, M.A.
Weatherhead, C. Humborg, H. Eriksson, and F. Hannerz. 2005. River
nutrient loads and catchment size. Biogeochemistry 75:83–107.
Soil Survey Staff. 2006. Natural Resources Conservation Service, U.S. General
Soils Map (STATSGO2) for New York. Available at http://soildatamart.
nrcs.usda.gov/ (verified 1 Mar. 2009).
Williard, K.W.J., D.R. DeWalle, P.J. Edwards, and W.E. Sharp. 2001. 18O
separation of stream nitrate sources in mid-Appalachian forested
watersheds. J. Hydrol. 252:174–188.
1159