TECHNICAL REPORTS:TECHNICAL

LANDSCAPEREPORTS
AND WATERSHED PROCESSES

Baseflow Nitrate in Relation to Stream Order and Agricultural Land Use

Shujiang Kang and Henry Lin* The Pennsylvania State University
William J. Gburek and Gordon J. Folmar USDA-ARS
Birl Lowery University of Wisconsin–Madison

Management of agricultural nonpoint-source pollution

continues to be a challenge because of spatial and temporal
variability. Using stream order as an index, we explored the
K nowledge of the spatial distribution of key watershed variables
can enhance the understanding of watershed heterogeneity and
physical responses at different scales (Hornberger and Boyer, 1995).
distribution of nitrate concentration and load along the stream
network of a large agricultural watershed in Pennsylvania— Among diverse watershed variables, stream order is of particular
the East Mahantango Creek Watershed and two of its sub- interest in this study. Investigations on stream-ordering system
watersheds. To understand nitrate concentration variation in initiated by Horton (1945) and later refined by Strahler (1957)
the stream water contributed from ground water, this study led to the widely known Horton’s laws of drainage composition.
focused on baseflow. Impacts of agricultural land use area on
baseflow nitrate in the stream network were investigated. Nitrate
In essence, stream order is a measure of the position of a stream
concentration showed a general decreasing trend with increasing in a hierarchical tributary network. A watershed can be labeled
stream order based on stream order averaged values; however, with an order number corresponding to the stream order within
considerable spatial and temporal variability existed within each its boundary. Stream channel geometry based on stream order
snapshot sampling. Nitrate loads increased with stream order also presents insights of self-organizing of river network evolution,
in a power function because of the dominant effect of stream
flow rate over the nitrate concentration. Within delineated sub-
reflecting fractal characteristics, sediment transport, and landscape
watersheds based on stream orders, positive linear functions processes (Rodriguez-Iturbe and Rinaldo, 1997). Therefore, a
were found between agricultural land use area percentage and stream-ordering scheme provides a relative scale index that might
the baseflow nitrate concentration and between agricultural shed light on watershed processes. A fractal scaling relationship of
drainage area and the nitrate load. The slope of the positive stream number, stream length, and drainage area to stream order
linear regression between the baseflow nitrate concentration and
percent agricultural land area seems to be a valuable indicator of a
has been well recognized (Horton, 1945; Leopold et al., 1964;
watershed’s water quality as influenced by agricultural practices, Rodriguez-Iturbe and Rinaldo, 1997). However, the relationship
watershed size, and specific physiographic setting. Stream order between water quality and stream order is yet to be developed
seems to integrate, to a certain degree, the source and transport (Gupta and Waymire, 1998).
aspects of nonpoint-source pollution on a yearly averaged Limited efforts have been made to explore nutrient concentrations
basis and thus might provide a quick estimate of the overall
trend in baseflow nitrate concentration and load distribution
or loads in stream water as related to stream order. Few such relevant
along complex stream networks in agricultural watersheds. studies, as summarized herein, suggest that stream order is a poten-
tially valuable indicator of stream water quality. Böhlke and Denver
(1995) reported a decreasing nitrate concentration downstream dur-
ing baseflow in two agricultural watersheds in the Atlantic Coastal
Plain, USA. Meybeck (1998) studied water and particulate chemistry
(nutrients, major ions, and heavy metals) in the Seine watershed of
France at 10 key positions (from the first-order stream to the eighth-
order stream at the river mouth). He found that nitrate maximum
occurred in small agricultural streams, but phosphate maximum
was located at the most downstream stations. In studying the spatial
distribution of stream water composition, Hutchins et al. (1999) sug-
Copyright © 2008 by the American Society of Agronomy, Crop Science gested that the degree of the variability may show some relationship
Society of America, and Soil Science Society of America. All rights with stream order and land use. When attempting to demonstrate a
reserved. No part of this periodical may be reproduced or transmitted
in any form or by any means, electronic or mechanical, including pho-
link between perturbations to the terrestrial phosphorus (P) cycle, the
tocopying, recording, or any information storage and retrieval system, delivery of P to waters, and biological impact, Edwards et al. (2000)
without permission in writing from the publisher. indicated that the ecoregion concept combined with stream order
Published in J. Environ. Qual. 37:808–816 (2008).
would integrate spatial and temporal aspects of gradient, land use,
doi:10.2134/jeq2007.0011
Received 5 Jan. 2007. S. Kang and H. Lin, Dep. Crop and Soil Science, 116 ASI Building, The Pennsylvania State
*Corresponding author (henrylin@[Link]). Univ., University Park, PA 16802; W.J. Gburek and G.J. Folmar, Pasture System and Watershed
© ASA, CSSA, SSSA Management Research Unit, USDA-ARS, Building 3702, University Park, PA 16802; B. Lowery,
677 S. Segoe Rd., Madison, WI 53711 USA Dep. Soil Science, 1525 Observatory Dr, Univ. of Wisconsin, Madison, WI 53706.

808
flow velocity, and water quality. Alexander et al. (2000) reported Many previous studies of watershed water quality have focused
that nitrogen loss rate in streams declined rapidly with increasing on headwater streams of small catchments or on specific monitor-
channel size in their SPARROW analysis of 374 monitoring sta- ing sites in large river basins. Sampling schemes were generally
tions in the USA, suggesting that surface water travel time through snapshots (e.g., one-time sampling), which may not interpret
smaller tributaries (lower-order streams) is a major determinant of the spatial and temporal variations of water quality from a whole
their downstream transport to marine systems. Meybeck (2001) re- watershed perspective over a longer period (Wayland et al., 2003).
ported that daily sediment transport decreased within higher-order Understanding nutrient variability and their relations to land use
stream in a French river. A marked upstream–downstream gradient in different orders of streams may provide insights regarding water
of contamination was observed that showed major discrepancies quality spatial and temporal variation in stream networks. There-
between least sensitive elements (cobalt, barium, arsenic) and most fore, the objectives of this study were (i) to investigate the quantita-
sensitive elements (cadmium, mercury, zinc) in the river sediment. tive relationship between stream order and stream nitrate concen-
Peterson et al. (2001) found that headwater streams played a dis- tration and load during baseflow and (ii) to examine agricultural
proportionately large role in nitrogen (N) transformation on the land use impacts on stream nitrate in a large agricultural watershed
landscape despite their relatively small dimensions. The importance (from the first-order to the fifth-order streams) through a year-
of lower-order streams in contributing nitrate to stream channel round baseflow monitoring. The overall yearly trend reported in
network has also been reported by Gburek and Folmar (1999a) this study provides an assessment of the snapshot measure, when
and Lin et al. (2001). linked to stream order and agricultural land use area, in estimating
Landscape variables affect water quality in stream and nutrient concentration and load in different sizes of watersheds.
ground waters by altering chemical loads, sediment transport,
and hydrological processes (Karr and Schlosser, 1978; O’Neil Materials and Methods
et al., 1997; Basnyat et al., 1999). These variables could explain
a high percentage of total variation in N, P, and sediment yield
Study Watershed
(Hunsaker et al., 1992; Bolstad and Swank, 1997; Jones et The East Mahantango Creek Watershed is located in the
al., 2001). For example, 87% of the variation in total nitrate Ridge and Valley Physiographic Province of east-central Pennsyl-
yield could be explained by landscape metrics (O’Neil et al., vania, USA (Fig. 1). The drainage basin area is 423.6 km2, with
1997). The most significant indicators of the landscape were an altitude dropping from 550 m in the east to 110 m in the
the amount of agricultural land use, riparian forests, and atmo- west. The main geological formations are sandstones with inter-
spheric nitrate deposition in a watershed (Jones et al., 2001). bedded sedimentary rocks (Berg and Dodge, 1981). The climate
Land use is a landscape variable that is largely controlled by hu- is temperate and humid with an average annual precipitation of
man interests but is also often dictated by topography, soil con- 1090 mm. Agricultural land use dominates in this watershed,
ditions, and climate. Changes of land use by human activities with 57% for crops, 35% for forests, and 5% for pastures. Point-
could lead to higher nitrate concentration in streams or aquifers source pollution was insignificant in this watershed owing to the
than that of natural background. High nitrate loss to surface absence of a large sewage system and other point sources (Gburek
and ground waters caused by high percentage of agricultural and Folmar, 1999a; Lindsey et al., 2001). Within this watershed,
land use have been widely reported (e.g., Boyer and Pasquarell, a sub-watershed, called WE-38, has served as a research water-
1996; Gburek and Folmar, 1999a; Lin et al., 2001; Coulter et shed for the USDA-Agricultural Research Services’ Pasture Sys-
al., 2004). The percentage of cropland in a watershed has been tem and Watershed Management Research Unit since the 1960s.
suggested to linearly correlate with nitrate concentration in Another sub-watershed adjacent to WE-38, labeled as PA-01,
stream water (Gburek and Folmar, 1999a; Schilling and Libra, was selected in this study for comparison purpose with the WE-
2000; Lin et al., 2001; Schilling and Spooner, 2006). 38. Further details of the watershed under study can be found in
Ground water and surface water are fundamentally in- Pionke et al. (1996) and Lindsey et al. (2001).
terconnected with multiple hydrological processes (Winter
et al., 1998). Nonpoint source pollutants that infiltrate into Stream Order System, Sub-watershed Delineation, and
deeper soil or enter streams can be carried into aquifers by the Watershed Land Use
exchange process between ground water and surface water or Derived from the USGS national hydrography datasets
vadose zone water. Vice versa, pollutants contained within (1:24,000), the stream network was used to determine the
ground water discharge can have a significant impact on sur- stream orders for all stream segments within the watershed
face water quality (Winter et al., 1998; Harvey et al., 2003; using the Horton-Strahler method (Strahler, 1957). A total
Vanni et al., 2001). During baseflow (also called “low flow,” of 218 first-order streams, 48 second-order streams, 11 third-
i.e., when ground water discharge dominates stream flow), order streams, two fourth-order streams, and one fifth-order
high nitrate concentration resulting from agricultural land use stream were identified in the entire stream network of the
can be detected from surface water quality monitoring (Boyer East Mahantango Creek Watershed.
and Pasquarell, 1996; Vanni et al., 2001). Gburek and Folmar Eleven quads (7.5 minutes) of the 10-m resolution digi-
(1999a) reported that high nitrate discharge in shallow fracture tal elevation model of the USGS national elevation dataset
layer in the East Mahantango Creek Watershed significantly were processed (including format conversion, emerging, and
leads to high nitrate concentration in streams during baseflow.

Kang et al.: Baseflow Nitrate in Relation to Stream Order & Land Use 809
Fig. 1. Location of the East Mahantango Creek Watershed within the Chesapeake Bay Basin, its sub-watersheds of WE-38 and PA-01, and stream
sampling sites used in this study. Numbers shown in the two sub-watersheds illustrate the stream order numbering system.

smoothing) using ArcGIS 8.2 (Environmental Systems Re- sites of lower-order streams were located in the two focused sub-
search Institute, Redlands, CA). With the hydrology extension watersheds, WE-38 and PA-01. The other sites for higher-order
in ArcGIS 8.2, sinks were found and filled first, and then flow streams were distributed along the main channel network. A total
directions were generated based on the filled digital elevation of 17 samplings were conducted at these 31 sites during baseflow
model data. Outlets of different order streams were used to from May 2003 to April 2004. To monitor nitrate concentration
delineate the sub-watersheds based on stream order. in stream water contributed from ground water, this study focused
Land use coverage for the entire East Mahantango Creek on baseflow sampling. No significant thawing or precipitation
was obtained from the USGS National Land Cover Dataset, within 3 d is required for baseflow status in the East Mahantango
which was based on the Landsat 5 TM imagery (1992) with Creek Watershed, based on an empirical relationship developed
a spatial resolution of 30 m (Vogelmann et al., 2001). Agri- by Linsley et al. (1975): D = 0.827A0.2, where D is the number of
cultural land use percentage (including crop and fallow land, days between the storm peak and the end of overland flow, and A
pasture, and hay) was calculated as total agricultural land area is the drainage area in square kilometers. In our study watershed,
within a sub-watershed divided by total sub-watershed area. A = 423.6 km2, and D = 2.77 d. In this study, baseflow samplings
were conducted during a period when there was no significant
Water Quality Monitoring and Analysis thawing or precipitation within at least 4 d. Real-time flow data at
The protocols from the National Field Manual for the Collec- a USGS gauge near the outlet of the watershed (located at the fifth-
tion of Water Quality Data (USGS, 1997) were followed for water order stream) were used to assist in judging baseflow conditions in
sampling in this study. Approximately biweekly sampling was the watershed (Fig. 2), where almost all sampling was conducted
conducted to monitor baseflow stream nitrate at 31 representative when the stream flow rate at the watershed outlet was below ap-
sites distributed throughout the watershed (Fig. 1). The monitoring proximately 10 m3 s−1. To monitor nitrate loads more closely at
sites included nine first-order streams, eight second-order streams, different order streams, four gauging stations within the East
six third-order streams, four sites at two fourth-order streams, and Mahantango Creek were used in our sampling: the FD-36 gauge
four sites along the fifth-order stream channel (Fig. 1). Due to the on a first-order stream, the WE-38 gauge on a third-order stream,
similarity of topography, geology, and land use in the lower-order the Klingertown gauge on a fourth-order stream, and the USGS
streams in the East Mahantango Creek Watershed, most sampling gauge located on a fifth-order stream (Fig. 1). Nitrate loads were

810 Journal of Environmental Quality • Volume 37 • May–June 2008

calculated by multiplying flow rate and nitrate concentration. For
each sampling, grab samples were collected from all 31 sites within
a period of 4 h. We used a specially designed bailer for sampling
mixed water samples in large streams. Samples were kept on ice
until analysis in the laboratory. After water samples were filtered to
autosampler vials through 0.45-μm filters, nitrate was analyzed by
the Ion Chromatograph (Dionex Company, Sunnyvale, CA). Au-
tosampler vials were set for wash with mill-Q water, and five stan-
dard concentrations (0.1, 1, 5, 10, and 15 mg L−1 nitrate N) were
run to develop a standard curve. To ensure quality assurance and
quality control, about 10% of samples were randomly replicated
for analysis. The standard concentrations after a mill-Q water wash
were also randomly inserted to a sample queue.
Nitrate and flow data collected during a sustained baseflow
on 13 July and 14 July 1998 for a large portion of the East Ma-
hantango Creek (Lindsey et al., 2001) were re-analyzed with Fig. 2. Hydrograph at the USGS gauge near the outlet of the East
Mahantango Creek Watershed and the sampling dates (dots)
stream order in this study. All 26 sampling sites by Lindsey et al. from May 2003 and May 2004. The dashed line approximates the
(2001) were located at the upstream of the Klingerstown gauge, baseline of the baseflow in this watershed.
including five first-order streams, nine second-order streams,
and Urban (1990), Böhlke and Denver (1995), and Lindsey et
nine third-order streams, and three fourth-order streams. Wading
al. (1997). Third, higher nitrate concentration from lower-order
measurement with a pygmy current meter and a Flo-Mate model
streams would be diluted downstream in higher-order streams.
2000 portable water flowmeter (Marsh-Mcbirney, Inc., Freder-
A fourth reason is nitrate depletion during in-stream transport
ick, MD) were used for flow measurements in that study (Lind-
through processes such as denitrification in the riparian and hypo-
sey et al., 2001). Five synoptic baseflow water quality samplings
rheic zones and nitrate consumption by phytoplankton (Cooper,
in the WE-38 conducted from July 1997 to May 1999, plus one
1990; Burns, 1998; Grimaldi and Chaplot, 1999). Thus, increased
collected in September 1990 (Gburek and Folmar, 1999a), were
in-stream processes in higher-order streams would further reduce
included for further analysis in this study to investigate their rela-
nitrate concentration downstream during the baseflow periods.
tionships to stream order.
Significant variations of nitrate concentration in different order
streams were noticeable (Fig. 3). The reasons for such variation
Results and Discussion include (i) a high degree of land use variability within each stream
Nitrate Concentration in Relation to Stream Order order (Fig. 4 and 5), particularly in headwater areas, where two
during Baseflow extreme scenarios were encountered in this study (i.e., 94.14%
agricultural use and 92.21% forestry in some first-order sub-water-
Averaged nitrate concentration during baseflow over the year-
round approximately biweekly sampling showed a general declin-
ing trend with increasing stream order in the East Mahantango
Creek Watershed, although considerable variation existed within
each sampling (Fig. 3). Lower-order streams (e.g., the first- and
second-order streams) tended to have higher nitrate concentrations
than higher-order streams (e.g., the fourth- and fifth-order streams)
on the basis of stream order averaged values. One major reason
for such a higher nitrate concentration in the low-order streams
was higher discharge from shallow ground water that contained
higher nitrate concentration caused by higher agricultural land
use in headwater catchments. This observation was supported by
the studies of Pionke and Urban (1985) and Gburek and Urban
(1990). In contrast, several factors combined to contribute to the
lower nitrate concentrations in higher-order streams. First, there is
a general trend of reduced agricultural land use area in the higher-
order sub-watersheds of the East Mahantango Creek Watershed,
although the difference among the five stream orders was not
statistically significant (Fig. 4). This general trend would imply
reduced sources of nitrate to streams in higher-order sub-water-
sheds. Second, discharge of ground water in higher-order sub-wa- Fig. 3. Averaged nitrate concentration in relation to stream order in the
East Mahantango Creek Watershed during baseflow from May
tersheds tends to come from deeper aquifers that have lower nitrate 2003 to April 2004. The dots represent 17 sampling sets, with
concentration, as suggested by Pionke and Urban (1985), Gburek dashed lines indicating 95% confident interval over the averages.

Kang et al.: Baseflow Nitrate in Relation to Stream Order & Land Use 811
 percentage in lieu of detailed separation of row cropland from
other agricultural land uses can provide an easier prediction of
stream nitrate concentration without the constraint of detailed
land use area distribution (which may be difficult to obtain, es-
pecially for large watersheds). In addition, fallow and pasture/hay
lands are potential nitrate source areas and therefore should be
considered for predicting nitrate in the watershed.
We observed the slope value of the positive linear regression
between baseflow nitrate concentration and agricultural land area
percentage to decrease with increasing watershed size (Fig. 5). Fig-
ure 5a includes the dataset collected from all five orders of streams
during this study, and Fig. 5b contains a separate sampling set by
Lindsey et al. (2001) that was limited to within the fourth-order
streams in the East Mahantango Creek Watershed and another
sampling set by Gburek and Folmar (1999a) for the WE-38 only
(the first- to third-order streams). The slopes of the three regres-
sions shown in Fig. 5 decreased from 0.15 for the third-order
WE-38 to 0.12 for the fourth-order sub-watershed and to 0.10 for
the entire fifth-order watershed. Such a trend suggests that agricul-
Fig. 4. Agricultural land use percentage in relation to stream order in
the East Mahantango Creek Watershed. Error bars indicate 1 SD.
tural land area percentage tends to have more pronounced impact
on stream water quality in lower-order stream areas. As watershed
sheds); (ii) varying manure application rates from dairy and poultry size (and stream order) increases, the watershed system “damping”
farms within this watershed, as discussed by Gburek and Folmar tends to increase. Schilling and Libra (2000) summarized linear
(1999a); (iii) different stream flow regimes occurring during season regressions of nitrate concentration and row cropland percentage
shifting or after large storm events, as suggested by Pionke and for different sizes of watersheds (from 0.1 to 237,000 km2) in the
DeWalle (1994) and Pionke et al. (1999); and (iv) baseflow nitrate midwestern USA and found that the slope of nitrate concentra-
may exhibit a diurnal pattern, as reported by Scholefield et al. tion versus row crop percentage decreased from 0.14 to 0.07 with
(2005) for a small, moorland-fed river in the UK, where the ampli- increasing watershed size. Another interesting observation in this
tude of nitrate daily cycle was about 30% of the mean value. Be- study is that, if we force the intercept of the linear regression to be
cause of such noticeable spatial and temporal variability, this paper that of the forestry watershed average (0.47 in this study [see Fig.
emphasizes the general trend of baseflow nitrate level averaged by 5a], which could be interpreted as the watershed’s background ni-
stream orders, particularly from a yearly averaged perspective. trate concentration), then the slope for all three regressions shown
in Fig. 5 for the Mahantango Creek Watershed would be almost
Nitrate Concentration in Relation to Agricultural Land identical (i.e., ?0.10). This suggests that the slope of nitrate con-
Use Area centration (mg L−1) vs. agricultural land area (%) relationship can
Nitrate concentration was positively related to agricultural serve as a comparative indicator of nitrate output during baseflow
land area percentage in a nearly linear function in this study (Fig. among different agricultural watersheds within similar climatic
5). This result was in agreement with previous studies reported regimes. Besides watershed size (and stream order), other factors,
by Gburek and Folmar (1999b) and Lindsey et al. (2001). Simi- such as hydrogeology, geomorphology, and the type of agriculture,
lar linear relationships between stream nitrate concentration influence this slope value. For example, Boyer and Pasquarell
and agricultural land use area have also been reported in other (1995) reported the slope of linear regression of median nitrate
states of the USA, such as West Virginia (Boyer and Pasquarell, concentration with percent agricultural land area to range from
1995), Kentucky (Taraba et al., 1996), and Iowa (Schilling and 0.19 to 0.25 in Karst springs in an extensively grazed area in West
Libra, 2000). Gburek and Folmar (1999b) developed a simple Virginia, suggesting that animal agriculture has a greater impact
model for the WE-38 using weighted land uses to predict nitrate per percent increase in agricultural land area in Karst watershed.
concentration in the stream: NO3–N (mg L−1) = [1 (%forest) + Averaging over the entire year of our monitoring dataset, base-
8 (%crop rotation) + 10 (%pasture) + 20 (%corn) + 30 (%ani- flow nitrate concentration in this study showed a linear relationship
mal)]/100. Agricultural land use area percentages in this equation with agricultural land area (y = 0.69 + 0.10x; r2 = 0.55) and stream
(including rotation, pasture, row crops, and animal) were domi- order (y = 7.90 − 0.69x; r2 = 0.36), but with opposite trends (Fig.
nant variables influencing water quality. Lindsey et al. (2001) 3 and 5). This could be partially explained by an overall decreasing
simplified the above model using only three land use variables of trend of agricultural land area percentage with increasing stream
forest, row crop, and pasture. Our regression equations shown in order in the watershed studied (Fig. 4). With added links to in-
Fig. 5 contained an overall percentage of agricultural land area stream processes and travel distance, stream order seems to reflect,
within each sub-watershed that included cropland, fallow, and to a certain degree, a combined effect of source and transport as-
pasture/hay lands. Using such an overall agricultural land area pects of nonpoint-source pollution, hence achieving a higher year-
ly-averaged regression r2 (0.76) with baseflow nitrate concentration

812 Journal of Environmental Quality • Volume 37 • May–June 2008

than that with agricultural land area percentage alone
(r2 = 0.55). We are aware that in some other areas
agricultural land use area percentage may increase
with increasing stream order because of the tendency
of steeper landforms (poorly suited to agriculture)
to occur closer to watershed divides (i.e., in lower-
order stream areas). Thus, the observed stream order
relationships in this study may not hold everywhere.
Nevertheless, the generic nature of a stream-ordering
scheme would allow a wider application of stream
order in examining watershed processes in relation
to water quality in different portions of a hierarchical
stream tributary network.

Nitrate Loads in Relation to Stream Order

and Agricultural Land Use Area
Based on stream flow and water quality data
collected in the 1990s, nitrate loads increased with
increasing stream order in a power function (Fig.
6). Seasonal effects changed total loads of nitrate
but did not change the power function relation-
ship between nitrate loads and stream order (Fig.
6b). Nitrate loads obtained from our four gauging
stations (representing four different order streams)
from May 2003 to April 2004 also showed a pow-
er function relationship with stream order (Fig. 7).
Such a relationship was apparently dominated by
the impact from stream flow rate or, equivalently,
drainage area because they both had a power func-
tion relationship with stream order (Fig. 7b). Be-
cause nitrate loads were calculated as the product
of flow rate and nitrate concentration and because
stream flow rate varied much more significantly
from the first-order to the fifth-order streams than
that of baseflow nitrate concentration, the load
distribution along the stream network was con- Fig. 5. Nitrate concentration in relation to agricultural land use in the East Mahantango
trolled by the flow rate rather than the nutrient Creek Watershed. (a) Average ±1 SD over the entire year’s bi-weekly sampling
from May 2003 to April 2004 at the 31 monitoring sites. (b) Data collected
concentration. Lower nitrate loads in headwater during a sustained baseflow for a large portion of the East Mahantango Creek
streams were due to lower flow with smaller drain- Watershed and a synoptic baseflow sampling collected in the WE-38 in 1998.
age area, although nitrate concentration tended to (Data from Lindsey et al., 2001; Gburek and Folmar, 1999a.)
be higher than that in higher-order streams in the hierarchical tributary network that might enable quick prediction
watershed in this study. of relative nutrient concentration, load, stream flow, and landscape
In terms of contributing area of agricultural land and nitrate features across different sizes of watersheds.
loads, their relationship is nearly linear for the East Mahantango
Creek Watershed as a whole and for its sub-watershed WE-38 (Fig. Summary
8). The power function relationship of drainage area and stream or- The results of this study suggest that stream order, as indi-
der, as suggested in the Horton’s laws (Horton, 1945), implies that cated in earlier work, might quantitatively integrate the effects
a linear relationship between agricultural drainage area and nitrate of source and transport factors controlling diffuse pollution.
loads is expected because nitrate loads and stream order also display Stream order in this study showed the general trend of baseflow
a power function relationship in this watershed. Compared with nitrate concentration and load distribution along the tributary
drainage area, stream order is simpler to determine and can provide stream network in the East Mahantango Creek Watershed,
well defined sub-watershed area and the location in the complex USA. An overall inverse relationship was observed between
stream network, whereas drainage area is more continuous and baseflow nitrate concentration and stream order using stream
varies considerably depending on the targeted stream segment or order averaged values; however, considerable spatial and tem-
sampling site. In addition, stream order is a simple variable in the poral variations existed within each stream order during each

Kang et al.: Baseflow Nitrate in Relation to Stream Order & Land Use 813
Fig. 6. Nitrate loads in different order streams in (a) the East Mahantango Creek Watershed and (b) the WE-38 sub-watershed. Error bars indicate
1 SD. The regression analysis in (a) was performed with averaged nitrate load for each stream order. (Data from Lindsey et al., 2001; Gburek
and Folmar, 1999a.)

snapshot sampling, owing to the variability in agricultural land concentration, load, stream flow, and landscape features across
area, manure application rate, stream flow regime change, and different sizes of watersheds. Further studies are needed to
seasonal or diurnal fluctuations. The general decreasing trend of explore nonpoint-source pollutant distribution in various wa-
baseflow nitrate concentration with increasing stream order is tershed stream networks and how in-stream processes may alter
explained by the reduced nitrate sources (e.g., decreased agricul- such distribution. Although watershed specificity might exist,
tural land area percentage) and increased in-stream processes, the generic nature of a stream-ordering scheme might provide
including dilution effect. In contrast, a power function relation- wider implications for enhanced understanding of watershed
ship was observed between nitrate loads and stream order in processes as related to water quality along a hierarchical stream
the study watershed, owing to the dominant effect of stream tributary network.
flow rate that increased with stream order in a power function This study found that the slope of the linear regression
manner. Nitrate loads can also be estimated by the contribution between baseflow nitrate concentration (mg L−1) and percent
area of agricultural land use through a simply linear function. agricultural land area (%) is a useful indicator of the degree
Nevertheless, compared with drainage area, the simpler index of agricultural impacts on water quality in a given watershed.
of stream order might enable quick estimate of relative nutrient Such a slope varies with watershed size (and stream order) and

814 Journal of Environmental Quality • Volume 37 • May–June 2008

Fig. 7. (a) Nitrate loads and (b) flow rate (both in log scale) at the Fig. 8. Nitrate loads in relation to the contribution area of agricultural
four stream gauge stations used in this study. The four stream land use within the East Mahantango Creek Watershed and its
gauges are labeled in Fig. 1. The FD-36 gauge is at the outlet sub-watershed WE-38. (Data from Lindsey et al., 2001; Gburek
of a first-order stream, the WE-38 gauge is at the outlet of a and Folmar, 1999a.)
third-order stream, the Klingerstown gauge is at a fourth-order
stream, and the USGS gauge is at a fifth-order stream outlet. contamination in two agricultural watersheds, Atlantic Coastal Plain,
Maryland. Water Resour. Res. 31:2319–2339.
the type of watershed and its related landscape features, such Bolstad, P.V., and W.T. Swank. 1997. Cumulative impacts of land use on
as the type of agriculture, hydrogeology, and climate. water quality in a southern Appalachian watershed. J. Am. Water
Resour. Assoc. 33:519–533.
Acknowledgments Boyer, D.G., and G.C. Pasquarell. 1995. Nitrate concentrations in Karst
Springs in an extensively grazed area. Water Resour. Bull. 31:729–736.
This study was supported by the USDA-NRI Watersheds Boyer, D.G., and G.C. Pasquarell. 1996. Agricultural land use effects on
Program under Grant #2001-35102-11593. We thank Mr. nitrate concentrations in a mature karst aquifer. Water Resour. Bull.
Terry Troutman and his staff in the East Mahantango Creek 32:565–573.
Burns, D.A. 1998. Retention of NO3-N in an upland stream environment: A
Watershed for their assistance in the field sampling. The mass balance approach. Biogeochemistry 40:73–96.
comments from three anonymous reviewers helped improve Cooper, A.B. 1990. Nitrate depletion in the riparian zone and stream channel
the quality of this paper. of a small headwater catchment. Hydrobiologia 202:13–16.
Coulter, C.B., R.K. Kolka, and J.A. Thompson. 2004. Water quality in
agricultural, urban, and mixed land use watersheds. J. Am. Water
References Resour. Assoc. 40:1593–1601.
Alexander, R.B., R.A. Smith, and G.E. Schwarz. 2000. Effects of stream Edwards, A.C., H. Twist, and G.A. Codd. 2000. Assessing the impact of
channel size on the delivery of nitrogen to the Gulf of Mexico. Nature terrestrially derived phosphorus on flowing water systems. J. Environ.
403:758–761. Qual. 29:117–124.
Basnyat, P., D.T. Lawrence, M.F. Kathryn, and B.G. Lockaby. 1999. Gburek, W.J., and J.B. Urban. 1990. The shallow weathered fracture layer in
Relationships between landscape characteristics and nonpoint source the near-stream zone. Ground Water 28:875–883.
pollution inputs to coastal estuaries. Environ. Manage. 23:539–549. Gburek, W.J., and G.J. Folmar. 1999a. Flow and chemical contributions to
Berg, T.M., and C.M. Dodge. 1981. Atlas of preliminary geologic quadrangle streamflow in an upland watershed: A baseflow survey. J. Hydrol. 217:1–18.
maps of Pennsylvania. 4th ser., Map 61, scale approx. 1:62,500. Gburek, W.J., and G.J. Folmar. 1999b. Patterns of contaminant transport in
Pennsylvania Geological Survey, Middletown, PA. a layered fractured aquifer. J. Contam. Hydrol. 37:87–109.
Böhlke, J.K., and J.M. Denver. 1995. Combined use of groundwater dating, Grimaldi, C., and C. Chaplot. 1999. Nitrate depletion during within-stream
chemical, and isotopic analyses to resolve the history and fate of nitrate transport: Effects of exchange processes between stream water, the

Kang et al.: Baseflow Nitrate in Relation to Stream Order & Land Use 815
 hyporheic and riparian zones. Water Air Soil Pollut. 124:95–112. Peterson, B.J., W.M. Wollheim, P.J. Mulholland, J.R. Webster, J.L. Tank, E.
Gupta, V.K., and E.C. Waymire. 1998. Spatial variability and scale invariance Marti, W.B. Bowden, H.M. Valett, A.E. Hershey, W.H. McDowell,
in hydrological regionalization. p. 88–135. In G. Sposito (ed.) Scale W.K. Dodds, S.K. Hamilton, S. Gregory, and D.D. Morrall. 2001.
dependence and scale invariance in hydrology. Cambridge, UK. Control of nitrogen export from watersheds by headwater streams.
Harvey, J.W., M.H. Conklin, and R.S. Koelsch. 2003. Predicting changes Science 292:86–90.
in hydrologic retention in an evolving semi-arid stream. Adv. Water Pionke, H.B., and D.R. DeWalle. 1994. Streamflow generation on a small
Resour. 26:939–950. agricultural catchment during autumn recharge: I. Nonstormflow
Hornberger, G.M., and E.W. Boyer. 1995. Recent advances in watershed period. J. Hydrol. 163:1–22.
modeling. Rev. Geophys. 33(suppl.):949–957. Pionke, H.B., W.J. Gburek, and A.N. Sharpley. 1996. Flow and nutrient
Horton, R. 1945. Erosion development of streams and their drainage export patterns for an agricultural hill-land watershed. Water Resour.
watersheds; hydrophysical approach to quantitative morphology. Geol. Res. 32:1795–1804.
Soc. Am. Bull. 56:275–370. Pionke, H.B., W.J. Gburek, R.R. Schnabel, A.N. Sharpley, and G.F.
Hunsaker, C.T., D.A. Levine, S.P. Timmins, B.L. Jackson, and R.V. O’Neil. Elwinger. 1999. Seasonal flow, nutrient concentrations, and loads
1992. Landscape characterization for assessing regional water quality. p. patterns in stream flow draining an agricultural hill-land watershed. J.
997–1006. In D.H. McKenzie et al. (ed.) Ecological indicators. Elsevier Hydrol. 220:62–73.
Appl. Sci., New York. Pionke, H.B., and J.B. Urban. 1985. Effects of agricultural land use on ground
Hutchins, M.G., B. Reynolds, B. Smith, G.N. Wiggans, and T.R. Lister. water quality in a small Pennsylvania watershed. Ground Water 23:68–80.
1999. Evaluation and interpretation of regional and site-specific Rodriguez-Iturbe, I., and A. Rinaldo. 1997. Fractal river basins, chance and
hydrochemical database for water quality assessment. Hydrol. Earth self-organization. Cambridge Univ. Press, Cambridge, UK.
Syst. Sci. 3:565–580. Schilling, K.E., and R.D. Libra. 2000. The relationship of nitrate
Jones, B.K., A.C. Neale, M.S. Nash, R.D. Wan Remortel, J.D. Wickman, K.H. concentrations in streams to row crop land use in Iowa. J. Environ.
Riitters, and R.V. O’Neill. 2001. Predicting nutrient and sediment loads Qual. 29:1846–1851.
to streams from landscape metrics: A multiple watershed study from the Schilling, K.E., and J. Spooner. 2006. Effects of watershed-scale land use change
United States Mid-Atlantic Region. Landscape Ecol. 16:301–312. on stream nitrate concentrations. J. Environ. Qual. 35:2132–2154.
Karr, J.R., and I.J. Schlosser. 1978. Water resources and the land-water Scholefield, D., T. Le Goff, J. Braven, L. Ebdon, T. Long, and M. Butler.
interface. Science 201:229–233. 2005. Concerted diurnal patterns in riverine nutrient concentrations
Leopold, L.B., M.G. Wolman, and J.P. Miller. 1964. Fluvial processes in and physical conditions. Sci. Total Environ. 344:201–210.
geomorphology. Freeman, San Francisco. Strahler, A.N. 1957. Quantitative analysis of watershed geomorphology. Eos
Lin, H., R. Cook, and B. Shaw. 2001. Nitrate relationships between stream 38:912–920.
baseflow, well water, and land use in the Tomorrow-Waupaca watershed. Taraba, J.L., J.S. Dinger, L.V.A. Sendlein, and G.K. Felton. 1996. Land
The Scientific World 1:187–193. use impacts on water quality from small agricultural watersheds in
Lindsey, B.D., C.A. Loper, and R.A. Hainly. 1997. Nitrate in ground water Kentucky. ASAE Paper No. 96-2087. ASAE, St. Joseph, MI.
and stream baseflow in the Lower Susquehanna River watershed, USGS. 1997. National Field Manual for the Collection of Water Quality Data.
Pennsylvania and Maryland. USGS Water Resources Investigations USGS Techniques of Water Resources Investigations, Book 9, Chapter
Rep. 97-4146. USGS, Washington, DC. A1–A9 (published from 1997–1998). USGS, Washington, DC. Available
Lindsey, B.D., W.J. Gburek, and G.J. Folmar. 2001. Watershed scaling effect at [Link] (verified 15 Oct. 2007).
on base flow nitrate, valley and ridge physiographic province. J. Am. Vanni, M.J., W.H. Renwick, J.L. Headworth, J.D. Auch, and M.H. Schaus.
Water Resour. Assoc. 37:1103–1117. 2001. Dissolved and particulate nutrient flux from three adjacent
Linsley, R.K., M.A. Kohler, and J.L.H. Paulhus. 1975. Hydrology for agricultural watershed. Biogeochemistry 54:85–114.
engineers. McGraw-Hill, New York. Vogelmann, J.E., S.M. Howard, L. Yang, C.R. Larson, B.K. Wylie, and
Meybeck, M. 1998. Man and river interface: Multiple impacts on water N. Van Driel. 2001. Completion of the 1990s National Land Cover
and particulates chemistry illustrated in the seine river watersheds. Data Set for the Conterminous United States from Landsat Thematic
Hydrobiologia 374:1–20. Mapper Data and Ancillary Data Sources. Photogramm. Eng. Remote
Meybeck, M. 2001. River sediment transport and quality: From space and Sens. 67:650–652.
time variability to management. Houille Blanche-Revue International Wayland, K.G., D.T. Long, D.W. Hyndman, B.C. Pijanowski, S.W.
de l’Eau 6–7:34–43. Woodhams, and S.K. Haack. 2003. Identifying relationships between
O’Neil, R.V., C.T. Hunsaker, K.B. Jones, K.H. Riitters, J.D. Wickham, P.M. baseflow geochemistry and land use with synoptic sampling and R-
Schwartz, I.A. Goodman, B.L. Jackson, and W.S. Baillargeon. 1997. mode factor analysis. J. Environ. Qual. 32:180–190.
Monitoring environmental quality at the landscape scale. Bioscience Winter, T.C., J.W. Harvey, O.L. Franke, and W.M. Alley. 1998. Ground
47:513–519. water and surface water, a single resource. Circ. 1139. USGS,
Washington, DC.

816 Journal of Environmental Quality • Volume 37 • May–June 2008