Sources and Transformations of Nitrate from Streams ... · in the Lisha Kill watershed, and about 4% urban land. Salmon Creek is a tributary to Cayuga Lake, one of the Finger Lakes

1149

Knowledge of key sources and biogeochemical processes that aff ect the transport of nitrate (NO

3−) in streams can inform

watershed management strategies for controlling downstream eutrophication. We applied dual isotope analysis of NO

3− to

determine the dominant sources and processes that aff ect NO

3− concentrations in six stream/river watersheds of diff erent

land uses. Samples were collected monthly at a range of fl ow conditions for 15 mo during 2004–05 and analyzed for NO

3− concentrations, δ15N

NO3, and δ18O

NO3. Samples from

two forested watersheds indicated that NO3− derived from

nitrifi cation was dominant at basefl ow. A watershed dominated by suburban land use had three δ18O

NO3 values greater than

+25‰, indicating a large direct contribution of atmospheric NO

3− transported to the stream during some high fl ows. Two

watersheds with large proportions of agricultural land use had many δ15N

NO3 values greater than +9‰, suggesting an animal

waste source consistent with regional dairy farming practices. Th ese data showed a linear seasonal pattern with a δ18O

NO3:δ

15NNO3

of 1:2, consistent with seasonally varying denitrifi cation that peaked in late summer to early fall with the warmest temperatures and lowest annual streamfl ow. Th e large range of δ 15N

NO3 values (10‰) indicates that NO

3− supply was likely

not limiting the rate of denitrifi cation, consistent with ground water and/or in-stream denitrifi cation. Mixing of two or more distinct sources may have aff ected 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. Th ese 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 NO

3− transport and process patterns

that contribute to streamfl ow in large basins.

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

The eff ects 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 fl uxes

and storage and the rates of many N cycling processes. Th is

human-induced acceleration of the N cycle is linked to myriad

environmental concerns, including soil acidifi cation, 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

fl uxes 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 diffi cult to obtain through more traditional

measurements of solute concentration and fl ux. Dual isotope stud-

ies of NO3− have proven useful for distinguishing sources such as

atmospheric deposition, human and animal waste, and fertilizer

and for tracing the eff ects of processes such as nitrifi cation and

denitrifi cation on NO3− pools in waters where measurements of

δ15NNO3

alone cannot distinguish among various sources and pro-

cesses (Burns and Kendall, 2002; Mayer et al., 2002; Lehmann et

al., 2003; Anisfeld et al., 2007). For example, δ18ONO3

values in at-

mospheric deposition are quite high, generally greater than +60‰

(Elliott et al., 2007). Th ese high values are believed to result from

reaction of atmospheric NOx with O

3 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

nitrifi ed to appear again as NO3−. Nitrifi ers 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 nitrifi cation in soils is generally in the range

of −10 to +10‰, and δ18ONO3

is a powerful tracer of the transfor-

Abbreviations: WWTP, wastewater treatment plant.

D.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.

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 pho-

tocopying, 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 ([email protected]).

© ASA, CSSA, SSSA

677 S. Segoe Rd., Madison, WI 53711 USA

TECHNICAL REPORTS: LANDSCAPE AND WATERSHED PROCESSES

1150 Journal of Environmental Quality • Volume 38 • May–June 2009

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 over-

lapping 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 (Pan-

no 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 in-

volve coupled fractionation; for example, denitrifi cation 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− concen-

trations, δ15NNO3

values, and δ18ONO3

values in six streams and

rivers in New York of widely varying land uses that included

forested (with and without signifi cant 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 infl uence 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

nitrifi cation, 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 fl ow in all watersheds

were expected to show δ18ONO3

values greater than +15‰, in-

dicating a direct infl uence 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 ba-

sins for which data on NO3− concentrations and loads combined

with δ15NNO3

and δ18ONO3

data do not always provide clear evi-

dence of the dominant processes (Mayer et al., 2002).

Materials and Methods

Study SitesTh e 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 Moun-

tains; and (vi) Buck Creek, a forested watershed with wetlands in

the Adirondack Mountains. Th e stream sampling sites are iden-

tifi ed on Fig. 1. Table 1 provides some physical and land cover

characteristics of each watershed, and Table 2 describes the geol-

ogy and soil drainage properties of each watershed.

Th e Lisha Kill drains a largely suburban watershed with an

impervious area of 13.9%. Only 2.7% of the watershed is clas-

sifi ed as high-density developed; the remaining developed land

is of low and medium density as well as open space typical of

suburban landscapes. Th e watershed includes some patches of

forested land and 29.2% wetland area (Table 1); nearly all the

wetland area is classifi ed as woody wetland and is primarily

located in riparian areas. Th e watershed is underlain primarily

by dune sands and is excessively drained, the most permeable

of any soils in this study (Table 2). Th e stream is a second-order

tributary of the Mohawk River.

Th e Salmon Creek watershed consists of 72.0% agricultural

land (Table 1) of which 46.1% is cultivated crops, and the re-

mainder is classifi ed as pasture/hay. Dairy farming is the domi-

nant agricultural activity in this region of central New York that

includes Salmon and Fall Creeks (Johnson et al., 2007). Th e

principal cultivated crop is corn, and the principal pasture crop

is alfalfa hay; both are used as cattle feed. Th is watershed con-

tains 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 diff er-

ent 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. Th e Fall Creek watershed is un-

derlain 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).

Th e Mohawk River (at Cohoes, NY) is the largest water-

shed considered in this study, with a drainage area of 8849 km2

and underlain by diverse bedrock types (Tables 1 and 2). Th is

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. Th e Mo-

hawk River watershed is about 50% forested land, 25% agri-

cultural land, and 7.5% urban land; these values are close to

the mean values for the other fi ve sites. Dairy farming is the

principal land use practice on agricultural land, similar to the

two watersheds in central New York. Th ere are several munici-

pal 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 ma-

ple (Acer rubrum L.), yellow birch (Betula alleghaniensis Brit-

ton), and American beech (Fagus grandifolia Ehrh.) and minor

amounts of conifers including balsam fi r (Abies balsamea (L.)

Burns et al.: Land Use and Nitrate Isotopes in Streams 1151

Mill), red spruce (Picea rubens Sarg.), and eastern hemlock (Tsu-ga Canadensis (L.) Carrière). Additionally, Buck Creek has about

28% wetland area in its drainage, almost all of which is classifi ed

as woody wetland (Table 1; NLCD Class C90). Both forested

watersheds are dominated by well drained soils (Table 2).

Sample and Field Data CollectionStream water samples were collected approximately month-

ly at each stream for approximately 15 mo during 2004–05

and analyzed for NO3− concentrations and isotopes. Samples

were collected over a range of fl ow conditions because streams

commonly show fl ow-related changes in NO3− concentrations,

and several samples were collected at high fl ow when sources

may diff er from those at low fl ow (Fig. 2A). For example, the

median probability that daily mean fl ow was exceeded for sam-

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.

Table 2. Geology and soil drainage properties of the six study stream watersheds. Data refl ect the dominant geology and soil type in each watershed as described in the references cited in the table.

StreamBedrock

type†Surfi cial

geology‡Soil drainage properties§

Lisha Kill shale sand dunes excessively drained

Mohawk River –¶ glacial till moderately well drained

Salmon Creek shale glacial till well drained

Fall Creek shale glacial till poorly drained

Biscuit Brook sandstone exposed bedrock well drained

Buck Creek gneiss glacial till well drained

† 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.


ples analyzed for NO3− concentrations was <0.50 at all sites,

and several samples were collected at all streams at daily mean

fl ow conditions that were exceeded <25% of the days based

on an analysis of the 2004–05 fl ow data at each stream (Fig.

2A). Only about half the samples collected in the forested set-

tings 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 fi lling. Samples

were collected near the middle of the channel, except in the

Mohawk River, where water was collected by an automated

sampler through a fi xed line with an intake located close to

the stream bank. Samples were stored on ice in a cooler, trans-

ferred to a refrigerator (4°C) on returning from the fi eld, and

passed (generally within 24 h of collection) through a 0.4-μm

polycarbonate fi lter. 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). Th e proce-

dures for compositing these samples and a discussion of quality

assurance of results from a similar study of archived precipita-

tion samples can be found in Elliott et al. (2007). Th e 10 sites

chosen for analysis included all of the NADP/NTN sites in New

York that were operated throughout 2004–05 (see location in-

formation 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 signifi cantly diff erent 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 emis-

sions have isotope values that diff er from those of NO3− in wet

deposition (Pearson et al., 2000); however, the possible infl uence

of dry deposition of NOx on the isotopic composition of stream

NO3− could not be evaluated in the current study.

Streamfl ow data for these streams/rivers were obtained by

gages operated by the U.S. Geological Survey at fi ve 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 re-

lationship was developed between mean daily discharge at Fall

Creek and Salmon Creek from July 2006 through September

2007 (r2 = 0.83; p < 0.001). Th is relationship was used to pre-

dict a mean daily discharge at Salmon Creek.

Analytical MethodsSamples collected at Biscuit Brook, Buck Creek, Lisha Kill,

and Mohawk River were analyzed for NO3− concentrations by

ion chromatography according to methods described in Law-

rence et al. (1995). Quality assurance/quality control methods

and results are described in biannual reports available at http://

ny.water.usgs.gov/publications. Th e data quality objective for

analytical precision is ±10%, and these objectives are generally

achieved for >90% of analyses of quality assurance samples (Lin-

coln et al., 2006). Th e samples collected at Fall Creek and Salm-

on Creek were also analyzed by ion chromatography at the labo-

ratory 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 N

2O 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 identifi ca-

tion 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 (‰) = sample standard

standard

(R) (R)

(R)

− × 1000 [1]

where R denotes the ratio of the heavy to light isotope (e.g., 15N/14N or 18O/16O).

Samples were corrected using international reference stan-

dards USGS34, USGS35, and N3 (Böhlke et al., 2003); linear-

ity 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 pre-

cision for δ15N was 0.3‰ based on an internal reference stan-

dard that was analyzed at least once per 40 samples.

Data AnalysisStreamfl ow data from the Lisha Kill were separated into a

basefl ow and a direct runoff component using a local mini-

mums method of daily fl ow as calculated by the Base Flow

Index program (http://www.usbr.gov/pmts/hydraulics_lab/

twahl/bfi /). Th ese 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 denitrifi cation in seasonal NO3− evolution.

Th ese independent variables were chosen because they are among

the dominant environmental controls on denitrifi cation rates for

which data were available in this study. Th e 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


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).

Th e lowest concentrations at Salmon Creek were found in ear-

ly- to mid-summer, but values otherwise varied little through-

out the year.

Fall Creek, with the second greatest proportion of agricul-

tural 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. Th e mean NO

3− 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 grow-

ing seasons (0.9 mg L−1 as NO3−–N) and higher concentra-

tions in the November–April dormant season (1.3 mg L−1 as

NO3−–N), similar to Salmon Creek, although the seasonal de-

clines 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 aff ected

NO3− concentrations in the stream. Combined discharge from

these two plants was provided by New York State (Dept of En-

vironmental Conserv., unpublished data), and the average con-

tribution to streamfl ow 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 (NY-

SDEC, 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 NO

3− 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− con-

centrations were about 0.5 to 1 mg L−1 as N during the non-

growing season (November–April) despite the wide diff erences

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. Th e suburban Lisha Kill showed greater varia-

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 modifi cation of the

fi gure shown in Kendall et al. (2007). Some general patterns and

diff erences among the sites are immediately evident on examining

Fig. 3A and 3B. Th e 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). Th e 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). Th e higher δ18ONO3

values obtained with the

denitrifi er method compared with the previously common closed-

tube AgNO3 method have also been noted by previous investiga-

tors (Revesz and Böhlke, 2002; Kendall et al., 2007).

Fig. 2. Samples collected and analyzed for nitrate concentrations in the six study streams during 2004–05. (A) Flow duration curve with sample fl ow exceedances marked. (B) NO

3−–N stream

concentrations as a function of date.


Nearly all the stream isotope data lie within the overlapping

fi elds defi ned 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). Th ese data and the likely

reasons for the observed patterns are discussed in greater de-

tail 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 WatershedsTh e isotope data from Biscuit Brook and Buck Creek are con-

sistent with nitrifi cation of soil N as the dominant source to these

streams similar to previous fi ndings in these two regions (Burns

and Kendall, 2002; Piatek et al., 2005). Th ese data do not show

evidence of a large direct contribution of atmospheric NO3− to

these streams, similar to many previous studies in forested wa-

tersheds (Kendall et al., 1995; Ohte et al., 2004; Sebestyen et

al., 2008). Th e 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 denitrifi cation may

have aff ected the evolution of the NO3− pool that is ultimately

transported in the stream. Th ese 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 denitrifi ed NO3− pool. Th is 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 eff ect on stream N fl uxes (McHale et al., 2004).

Th e δ15NNO3

values are similar to those reported previ-

ously 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). Th ese data suggest that the soil source

box represented in fi gures such as that in Kendall et al. (2007)

ought to extend to a δ15NNO3

value of about 0‰.

Large diff erences 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 farm-

ing or fi re. Th ese results may seem puzzling because in the case of

precipitation, values obtained with the older closed-tube meth-

ods were less than those obtained with the denitrifi er method,

whereas in the case of stream water, the older values are higher

than recent values from the same or similar watersheds. Th ese

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. Th erefore, the

results reported here are consistent with a high bias from such a

contaminant source. However, we are not certain why these new

results diff er from past studies in similar environments, and pos-

sible 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).

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 NO

3− derived by nitrifi cation

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.


Suburban WatershedTh e isotope data from the suburban Lisha Kill generally lie

within a range consistent with a nitrifi ed soil N source but could

also be aff ected by a waste source as well (Fig. 3B). Th e δ15NNO3

values of the Lisha Kill are higher than those from the two forest-

ed 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 nitrifi ed source and waste source overlap

in this range, the isotope data are not defi nitive 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.

Th e data in Fig. 3B show that four samples collected in the

suburban Lisha Kill lie outside the range of δ18ONO3

values ex-

pected from soil or waste sources, suggesting a signifi cant 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 mem-

ber and the median value of δ18ONO3

of +5.7‰ in stream wa-

ter 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). Th ese large direct contributions of atmospheric NO3−

have been observed only during high fl ow events and tend to be

transient in nature, with the dominant NO3− source at basefl ow

quickly reimposing its isotopic signature in stream water soon

after peakfl ow conditions have waned.

To further explore these changes in the NO3− isotope signa-

ture, the δ18ONO3

values at Lisha Kill are shown as a function of

stream runoff , event fl ow, and NO3− concentration (Fig. 4). A

statistically signifi cant relationship between stream runoff and

δ18ONO3

is evident (r2 = 0.35; p = 0.025) (Fig. 4A), but this rela-

tionship is highly leveraged by the sample collected at the high-

est fl ow 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.

Th e Lisha Kill watershed consists of about 14% impervious

area with storm sewers that discharge directly into the stream or

into small ephemeral tributaries. Th is network of paved surfaces

and storm sewers may allow atmospheric NO3− to bypass soil

contact whenever a rainfall or snowmelt event occurs in the wa-

tershed regardless of the size of the event and the gross amount

of streamfl ow. To evaluate this rapid runoff eff ect, we calculated

direct runoff with the Base Flow Index program and found that

this quantity was signifi cantly 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 . Addi-

tionally, the three samples with the highest δ18ONO3

values were

collected when the direct runoff component was greater than

approximately 50% of total streamfl ow. 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 lever-

aging 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. Th e Lisha Kill also contains abundant ripar-

ian wetlands, and these landscape features have previously been

shown to greatly aff ect 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). Th e highest δ18ONO3

values were associ-

ated with some of the lowest NO3− concentrations measured.

Th ese 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, stormfl ow has

a tendency to dilute NO3− concentrations relative to basefl ow,

but this relationship is not strong. Other studies in residential

suburban settings have shown a similar tendency for dilution

of stream NO3− concentrations during stormfl ow (Hook and

Yeakley, 2005; Kim et al., 2007), although the storm response

can be highly variable and depends on the relative concentra-

tions of NO3− in ground water and precipitation, the relative

amount and connectivity of impermeable surfaces, the pres-

ence of septic systems, and the seasonality with respect to fertil-

izer application among other factors (Silva et al., 2002; Hook

and Yeakley, 2005; Poor and McDonnell, 2007).

Agricultural WatershedsMost 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 esti-

mates of the dominant stream NO3− sources identifi ed 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‰,


higher than the range previously reported for waters aff ected by ni-

trifi cation 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

val-

ues greater than +9‰, outside the range previously reported for

NO3− derived by nitrifi cation of soil organic matter, further sug-

gesting the dominance of a direct waste source.

Th e relative contributions of animal vs. human waste to

stream NO3− in Salmon and Fall Creeks is not known and can-

not 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

activities contributed about 61% in the mid-1970s. Salmon

Creek, with a higher proportion of agricultural land, less ur-

ban 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 manage-

ment changes in Fall Creek since the 1970s that include an

increase in population, addition of a WWTP in Freeville, in-

creased dairy production, and improved manure management

practices, agricultural activities are still believed to be the prin-

cipal 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 rela-

tionship between δ18ONO3

and δ15NNO3

values at Fall and Salm-

on Creeks (Fig. 3A). Th e 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 denitrifi ed,

a relationship supported by empirical evidence and theoreti-

cal kinetics and generally considered unambiguous evidence of

denitrifi cation in fresh waters (Böttcher et al., 1990; Chen and

MacQuarrie, 2005; Panno et al., 2006). Th e infl uence of deni-

trifi cation would likely be least evident under high fl ow condi-

tions when rapid runoff processes might transport atmospheric

NO3− and freshly nitrifi ed manure sources to these streams.

On this basis, the highest fl ow samples were removed from the

data set (those with a probability of exceedance of <0.15 based

on 2003–05 fl ow data at Fall Creek). Th e 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 fi rst-order

kinetics and laboratory reaction rate constants for denitrifi ca-

tion (Olleros, 1983; Chen and MacQuarrie, 2005). Th e infl u-

ence of denitrifi cation seems to persist through the growing

(May–October) and dormant (November–April) seasons, al-

though the variability about the regression line is greater in the

dormant season, as might be expected when the infl uence of

denitrifi cation 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 signifi cant [p >

0.05] for Salmon Creek).

Th ese regressions indicate that δ15NNO3

values tended to

increase as stream runoff decreased and air temperature in-

creased and are consistent with a seasonally varying infl uence

of denitrifi cation that was greatest at warm temperatures when

streamfl ow was lowest (and in the case of Fall Creek when

NO3− concentrations were lowest).

Th e large range of δ15NNO3

values observed in these streams

suggests a large fractionation associated with denitrifi cation

(10‰), indicative of so-called “riparian” denitrifi cation in which

the supply of NO3− was not likely limiting the rate of denitri-

Fig. 4. δ18ONO3

values in the suburban Lisha Kill watershed as a function of (A) stream runoff , (B) percent event fl ow, and (C) NO

3− concentrations.


fi cation, whereas low fractionation values would suggest that

the rate of denitrifi cation was limited by NO3− supply, which

is characterized by much lower fractionation values (Sebilo et

al., 2003; Lehmann et al., 2003). Th ese data cannot identify

whether denitrifi cation was primarily occurring during ground

water transport or within the stream environment; however, re-

sults of a previous study at Fall Creek could not fi nd signifi cant

in-stream NO3− sinks, suggesting that the ground water environ-

ment may be the primary zone of denitrifi cation in this water-

shed (Johnson et al., 1976). Additionally, some denitrifi cation

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 denitrifi cation and may also originate from

variable amounts of mixing of soil NO3− with denitrifi ed water or

by algal NO3− uptake (Kendall et al., 2007). Th e 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 diff erent

δ15NNO3

values and NO3− concentrations is only linear when iso-

tope values are plotted as a function of 1/NO3− concentration

(Kendall, 1998), whereas isotope fractionations observed in bio-

logical 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 indi-

cate that these data are consistent with mixing and denitrifi cation

in Fall and Salmon Creeks; neither process can be defi nitively

ruled out based on the relationships evident in these data (Fig.

6). Th e 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). How-

ever, highly signifi cant statistical relationships are present for both

streams in each of the plots. Mixing with another source may

aff ect Fall Creek to a greater extent than Salmon Creek because

of the greater proportion of forested land and slight infl uence of

wastewater NO3− discharge in Fall Creek.

Mixing of two or more NO3− sources and denitrifi cation are

not necessarily mutually exclusive, and both processes may be af-

fecting the seasonal evolution of NO3− in these streams. Ground

water, with seasonally varying amounts of denitrifi cation, 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 denitrifi cation and mixing

of waters from diff erent sources aff ected the concentrations and

isotopic composition of NO3− in these two agricultural water-

sheds is not possible in this study and requires a combination of

additional chemistry, tracer, and hydrometric data.

Mixed Land Use WatershedTh e Mohawk River watershed contains a mixture of forested,

agricultural, and urban land use in proportions that are approxi-

mately equal to the mean values among the other fi ve study wa-

tersheds. Th e drainage area of the Mohawk watershed, however,

exceeds that of the others in the study by one to three orders of

magnitude. Th is 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 bio-

geochemical 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 aff ect the evolution of NO3− in riverine environments.

SummaryDual isotope data revealed varying sources and processes that

aff ect NO3− concentrations among six streams/watersheds of dif-

ferent land uses in New York. A suburban watershed with no

septic or wastewater infl uence showed NO3− concentrations only

slightly greater than those of two forested watersheds but slightly

higher δ15NNO3

values that may refl ect some infl uence from a

Fig. 5. δ15NNO3

vs. δ18ONO3

in the agricultural Fall and Salmon Creek watersheds for (A) all data, (B) low fl ow, (C) growing season, and (D) dormant season.


waste source. Th ree high fl ow 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 re-

fl ected 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 farm-

ing in the watersheds. Th ese isotope data showed a linear rela-

tionship of δ18ONO3

:δ15NNO3

of about 1:2, consistent with vari-

able amounts of denitrifi cation that likely occurred outside of

the stream environment. Th e eff ects of denitrifi cation appeared

greatest (based on multivariate regression of δ15NNO3

values)

when air temperature was warmest and streamfl ow was low-

est during summer and early fall, coincident with the lowest

stream NO3− concentrations. Th e NO

3− 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 fi ve

study watersheds, showed neither the high δ18ONO3

values of the

suburban stream nor the high δ15NNO3

values and 1:2 sloping line

of the agricultural streams. Th is fi nding 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. Th e small to

medium drainage area scale in streams dominated by one form of

land use is likely the most eff ective scale for distinguishing the role

of N sources and hydrologic and biogeochemical processes on the

transport of NO3− in riverine environments.

AcknowledgmentsTh is work was supported by a grant from the New York State

Energy Research and Development Authority. Th e authors

thank Elizabeth Nystrom, Heather Golden, and Michael

Brown for help with sample collection and Gregory Lawrence

and Michael McHale for sharing data.

ReferencesAmberger, 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 denitrifi cation 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 diff erentiate NO

3− 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. Eff ects 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. Surfi cial 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 denitrifi er method. Anal. Chem. 74:4905–4912.

Chen, D.J.Z., and K.T.B. MacQuarrie. 2005. Correlation of δ15N and δ18O in NO

3− during denitrifi cation 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 NO

x 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. Th e nitrogen cascade. Bioscience 53:341–356.

Genesee/Finger Lakes Regional Planning Council. 2001. Cayuga Lake Restoration and Protection Plan. Available at www.cayugawatershed.org (verifi ed 1 Mar. 2009).

Hook, A.M., and J.A. Yeakley. 2005. Stormfl ow dynamics of dissolved organic carbon and total dissolved nitrogen in a small urban watershed. Biogeochemistry 75:409–431.

Fig. 6. (A) Mixing plot, 1/NO3− concentration and δ15N

NO3. (B) Isotope

fractionation plot, ln NO3− concentration, and δ15N

NO3 for Fall and

Salmon Creeks.


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 fl uxes: 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, Th e 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 washed-off 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 denitrifi cation 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. Nadelhoff er, 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. Th iemens. 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.

31:L21506, doi:10.1029/2004GL020908.

Olleros, T. 1983. Kinetic isotope eff ects 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 denitrifi cation 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. Traffi c 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. Th e eff ects 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 diff erent combustion techniques. Anal. Chem. 74:5410–5413.

Schlesinger, W.H., K.H. Reckhow, and E.S. Bernhardt. 2006. Global change: Th e 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 nitrifi cation and denitrifi cation 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 denitrifi cation 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/ (verifi ed 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.

Documents

Sources and Transformations of Nitrate from Streams ... · in the Lisha Kill watershed, and about 4% urban land. Salmon Creek is a tributary to Cayuga Lake, one of the Finger Lakes