JEQ Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 5 April 2007
Published in J Environ Qual 36:664-680 (2007)
DOI: 10.2134/jeq2006.0084
© 2007 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Böhlke, J. K.
Right arrow Articles by Prestegaard, K. L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Böhlke, J. K.
Right arrow Articles by Prestegaard, K. L.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Böhlke, J. K.
Right arrow Articles by Prestegaard, K. L.
Related Collections
Right arrow Fate
Right arrow Nitrogen
Right arrow Geochemical Processes

TECHNICAL REPORTS

Landscape and Watershed Processes

Ground Water Stratification and Delivery of Nitrate to an Incised Stream under Varying Flow Conditions

J. K. Böhlkea,*, Michael E. O'Connell, deceaseda,b and Karen L. Prestegaardb

a USGS, 431 National Center, Reston, VA 20192
b Dep. of Geology, Univ. of Maryland, College Park, MD 20742

* Corresponding author (jkbohlke{at}usgs.gov)

Received for publication February 27, 2006.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Ground water processes affecting seasonal variations of surface water nitrate concentrations were investigated in an incised first-order stream in an agricultural watershed with a riparian forest in the coastal plain of Maryland. Aquifer characteristics including sediment stratigraphy, geochemistry, and hydraulic properties were examined in combination with chemical and isotopic analyses of ground water, macropore discharge, and stream water. The ground water flow system exhibits vertical stratification of hydraulic properties and redox conditions, with sub-horizontal boundaries that extend beneath the field and adjacent riparian forest. Below the minimum water table position, ground water age gradients indicate low recharge rates (2–5 cm yr–1) and long residence times (years to decades), whereas the transient ground water wedge between the maximum and minimum water table positions has a relatively short residence time (months to years), partly because of an upward increase in hydraulic conductivity. Oxygen reduction and denitrification in recharging ground waters are coupled with pyrite oxidation near the minimum water table elevation in a mottled weathering zone in Tertiary marine glauconitic sediments. The incised stream had high nitrate concentrations during high flow conditions when much of the ground water was transmitted rapidly across the riparian zone in a shallow oxic aquifer wedge with abundant outflow macropores, and low nitrate concentrations during low flow conditions when the oxic wedge was smaller and stream discharge was dominated by upwelling from the deeper denitrified parts of the aquifer. Results from this and similar studies illustrate the importance of near-stream geomorphology and subsurface geology as controls of riparian zone function and delivery of nitrate to streams in agricultural watersheds.

Abbreviations: CFC, chlorofluorocarbon • DIC, dissolved inorganic carbon • SERC, Smithsonian Environmental Research Center


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
NITRATE discharge rates are different from nitrate recharge rates in many agricultural watersheds, and they commonly exhibit large seasonal variations. Multiple processes controlling recharge-discharge imbalances at the watershed scale at various time scales are difficult to resolve, but they must be understood before predictive models of watershed nitrogen behavior will be acceptable. Relatively low average discharge fluxes of nitrate (NO3–) may be attributed to a variety of causes, including (i) assimilation or denitrification (microbial reduction of NO3– to N2 gas) in the stream channel or hyporheic zone (Duff et al., 1984; Seitzinger, 1988; Howarth et al., 1996; McMahon and Böhlke, 1996; Alexander et al., 2000), (ii) assimilation or denitrification by riparian plants or soils (Lowrance et al., 1984; Peterjohn and Correll, 1984; Jacobs and Gilliam, 1985; Hill, 1996), (iii) denitrification in upgradient aquifers (Mariotti et al., 1988; Postma et al., 1991; Böhlke et al., 2002), and (iv) delayed responses to increasing surface N loading in aquifers with long ground water residence times (Böhlke and Denver, 1995; Kauffman et al., 2001; Lindsey et al., 2003). Some aquifers transmit NO3– from recharge to discharge with relatively little modification, in which case it may be possible to predict annual stream NO3– loads from land use data (Gburek and Folmar, 1999); however, even in those situations there may be substantial seasonal variability in the concentrations.

Both positive and negative correlations have been observed between flow rates and NO3– concentrations in streams at various time scales. In agricultural watersheds, NO3– concentrations commonly are positively correlated with the logarithm of stream flow (Cohn et al., 1992), but the relations between flow and NO3– concentration vary greatly among different streams, and for a given stream they typically have large uncertainties (scatter) and they may vary depending on the time scale of interest. Decreases in NO3– concentrations during seasonal low-flow periods in warm months could be caused within the stream environment by assimilation or denitrification, both of which might be more active in summer than in winter. Alternatively, rising water levels in a watershed during high-flow periods could leach larger quantities of NO3– stored in the unsaturated zone and deliver it to streams in shallow flow systems, whereas low base flow derived from deeper ground water discharge may tend to be relatively NO3––free for various reasons including ground water age and denitrification (Lucey and Goolsby, 1993; Creed et al., 1996; Fenelon and Moore, 1998; Pauwels et al., 2001). Criteria for determining the relative importance of these different processes are not well developed and have not been applied to many different types of watersheds.

The purpose of the current study was to examine temporal relations between ground water hydrogeochemistry and surface water quality in a small agricultural watershed in the mid-Atlantic coastal plain where previous work indicated substantial net NO3– losses between recharge and discharge (Peterjohn and Correll, 1984; Correll et al., 1992). In this paper, we describe how vertical stratification of hydraulic and geochemical conditions in the aquifer may be translated into seasonal variations in stream base flow NO3– concentrations. Surface runoff events cause additional variations in stream chemistry that are different from the seasonal variations, but those events are not discussed in detail here.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
This study is focused on an incised first-order stream fed by a reactive stratified sedimentary aquifer, and provides important contrasts with some other geologic settings in the coastal plain and elsewhere. The approach included high-resolution vertical synoptic sampling of ground water between recharge and discharge areas, combined with time series measurements of ground water levels and of surface water flow and composition. Analyses included major element chemistry and dissolved gases, stable isotopes of N, C, and S, ground water dating with 3H, 3He, and chlorofluorocarbons (CFCs), and physical measurements including slug tests to determine aquifer hydraulic properties.

Study Site
Watersheds at the Smithsonian Environmental Research Center (SERC) have been studied intensively as models for agricultural contamination and riparian zone function (Correll et al., 1992). The SERC-109 watershed is part of the Rhode River basin near Edgewater Maryland on the west coast of Chesapeake Bay at 38°51' N, 76°32' W. The SERC-109 watershed has an overall area of 17 ha, of which approximately 60% is cultivated agricultural land (mainly corn) and 35% is forest (mainly in riparian lowlands) (Fig. 1) (Jordan et al., 1997). Average annual precipitation from 1972 to 1996 was 114 cm yr–1, while the average stream discharge during that period accounted for approximately 24 cm yr–1 (Correll et al., 1999). Mean monthly temperatures range from about 2°C (January) to 25°C (July) (Higman and Correll, 1982). Previous studies have shown that the average annual discharge flux of N from the SERC-109 watershed is substantially less than the combined annual N input from fertilizer application, N2 fixation, and atmospheric deposition (Correll, 1981; Peterjohn and Correll, 1984; Peterjohn and Correll, 1986; Correll and Weller, 1989). Much of the apparent N loss has been attributed to reduction of NO3– in ground water (by 85% in some years) during transport from upgradient cropland recharge areas into areas underlying the riparian forest (Peterjohn and Correll, 1984; Peterjohn and Correll, 1986). In addition, there is an overall positive correlation between NO3– and stream flow at each of several locations within the SERC-109 watershed along with a general decrease in the maximum concentrations from the headwaters to the basin outlet (Fig. 2). Large runoff events commonly result in short-term dilution of stream NO3– concentrations and disrupt the seasonal patterns (O'Connell et al., 1994).


Figure 1
View larger version (33K):
[in this window]
[in a new window]

  		Fig. 1. Map of the SERC-109 watershed, showing locations of temporary weirs (W1–W4), the permanent weir (W109), and the upper basin transect near the stream headcut (Peterjohn and Correll, 1984; O'Connell, 1998).

 

Figure 2
View larger version (28K):
[in this window]
[in a new window]

  		Fig. 2. Relation between stream flow and concentration of NO3– at stream sites in the SERC-109 watershed (see Fig. 1 for sample locations).

 
This study was conducted mainly near the source of the SERC-109 stream (Fig. 1). Except during major runoff events, the headwaters of the stream emerged from macropores and seeps near the base of a 1.7-m high vertical headcut and flowed within an incised channel with a narrow streambed about 1.5 to 3 m below the level of an adjacent wooded floodplain. Of 27 active outflow macropores that were mapped within the SERC-109 watershed, 21 were along the first-order incised reach within 80 m of the headcut (O'Connell, 1998). Piezometer nests were installed along a transect approximately normal to the stream, from a point in the stream about 10 m downstream from the headcut, across the adjacent forested riparian zone, to an upgradient point within a cultivated field planted in corn (Fig. 1 and 3). This array of sampling points (the "upper-basin transect") yielded a vertical 2-dimensional section of the subsurface that was approximately parallel to the dominant direction of ground water flow at the upgradient end. At the downgradient end near the stream, it is possible that the direction of ground water flow deviated from the transect direction, but nevertheless the section is considered to include the major ground water types contributing to stream flow in the incised reach. Stream samples from the transect location and the nearby upper-basin weir (W1) are considered to have been relatively unaffected by in-stream processes such as biologic uptake or benthic reactions because these locations were close to the headcut where the surface water is fed locally by seepage and macropores, with minimal subsequent residence time in the exposed channel.


Figure 3
View larger version (35K):
[in this window]
[in a new window]

  		Fig. 3. Vertical section normal to the stream at the upper basin transect: (A) nested piezometer locations (HC1-HC9) with major features of the landscape and underlying geology; (B) logarithm of hydraulic conductivity at locations measured by slug tests.

 
Geologic Sampling and Hydraulic Measurements
Core samples were collected using a 2.6-cm diam. hand-augering device (AMS Inc., American Falls, Idaho) fitted with 35-cm plastic liners. Lithologic descriptions were recorded and total porosities were estimated by comparing dry and wet weights of intact core samples with known volumes. Piezometers for ground water sampling consisted of 2.5-cm i.d. plastic pipes with 10- to 25-cm screens at depths ranging from about 0 to 6 m below the land surface and 0 to 4 m below the water table (Fig. 3). Ground water levels were measured in the piezometers with an electronic water level indicator with relative precision of about ±0.3 cm. Hydraulic conductivities were measured in the field by performing slug tests in the piezometers (Bouwer and Rice, 1976; Bouwer, 1989). A limited number of samples were collected from suction lysimeters in the unsaturated zone. Temporary V-notch weirs were installed in the stream for the duration of the study at several locations in the stream (Fig. 1). Stream discharge was determined manually at temporary weirs by measuring the volume of water in a container held beneath the discharge for a known period of time. Flows from individual outflow macropores also were measured volumetrically.

Chemical and Isotopic Analyses
Water samples were filtered in the field (0.45 µm) for major element chemical analysis. Cation samples were acidified with HNO3 and analyzed by direct-coupled plasma emission spectroscopy (SpectraSpan V, ARL, Dearborn, MI). Anion samples were analyzed by ion chromatography (IC) (4000i, Dionex Corp., Sunnyvale, CA). Dissolved gas samples (unfiltered) were either collected in evacuated glass vessels and analyzed by mass spectrometry (Finnigan MAT 251, Thermo-Electron, Waltham, MA) for Ar, N2, and O2 (O'Connell, 1998) or collected in serum bottles without headspace and analyzed by gas chromatography (GC) (HP 6890, Agilent Technologies, Wilmington, DE) for Ar, N2, O2, and CH4 (USGS, 2006a). Dissolved gas data were used to determine air-water equilibration temperatures during recharge (Weiss, 1970) and to determine the amounts of N2 gas that might have resulted from denitrification (Vogel et al., 1981).

Isotopic analyses of N ({delta}15N), S ({delta}34S), and C ({delta}13C) were done to provide evidence for sources and sinks of NO3–, SO42–, and dissolved inorganic carbon (DIC) (USGS, 2007). For S isotope analysis, SO42– was precipitated as BaSO4, which was combusted with CuO to yield SO2 for mass spectrometry (Carmody et al., 1998). The SO42– fraction of the sediment S was measured by acid leaching and precipitation of SO42– with BaCl2, and the disulfide (FeS2) fraction of sediment S was measured by treating the leached samples with acidified CrCl2 solution under N2 to release H2S, which was collected as Ag2S and weighed (Canfield et al., 1986; Tuttle et al., 1986). For isotopic analysis of the sediment S fractions, the BaSO4 was heated with CuO and the Ag2S was heated with Cu2O (Robinson and Kusakabe, 1975) to produce SO2. The S isotope measurements were calibrated by analyses of reference materials with the following assumed {delta}34S values: NBS127 (+20.9{per thousand}), NBS123 (+17.1{per thousand}), IAEA-S1 (–0.3{per thousand}), and "Maine light" (–30.7{per thousand}), with reproducibilities of approximately ±0.2{per thousand} for aqueous SO42– and ±0.5{per thousand} for solid-phase S. For C isotope analyses, DIC was precipitated in the field as BaCO3 by addition of BaCl2 and NaOH, filtered in the laboratory under N2, and reacted with H3PO4 to produce CO2 for mass spectrometry (McCrea, 1950). The {delta}13C values were calibrated by analyses of NBS-19 carbonate reference material (–1.95{per thousand}) and have reproducibilities of approximately ±0.4{per thousand} or less. For N isotope analyses of NO3–, water samples were freeze-dried and the salts were baked in evacuated sealed tubes with Cu+Cu2O and CaO to produce N2 for mass spectrometry (Böhlke and Denver, 1995). Nitrate {delta}15N data were normalized to values of +4.7{per thousand} for IAEA-N3 and +180.0{per thousand} for USGS32 (Böhlke and Coplen, 1995). For N isotope analysis of dissolved N2, the headspace gas remaining after GC analysis was sealed into evacuated glass tubes with Cu+Cu2O and CaO and baked to remove O2, then admitted to the mass spectrometer and analyzed along with air samples ({delta}15N {equiv} 0.0{per thousand}) and with samples prepared from air-equilibrated water ({delta}15N = +0.65 ± 0.10{per thousand}). The reproducibilities of the N isotope analyses are approximately ±0.2{per thousand} for NO3– and ±0.1{per thousand} for N2.

Ground Water Dating
Ground water recharge dates and age gradients were estimated from measurements of CFCs, with limited comparison to 3H and 3He analyses. Samples to be analyzed for CFCs were extracted through Cu tubing with a peristaltic pump and collected in flame-sealed glass ampules under pure N2 headspace (Busenberg and Plummer, 1992). The CFCs were extracted in the laboratory by a purge-and-trap procedure and analyzed by GC (USGS, 2006b). Concentrations of CCl3F (CFC11), CCl2F2 (CFC12), and C2Cl3F3 (CFC113) were converted to equilibrium partial pressures at the local water table elevations (20 ± 10 m) at the equilibration temperatures indicated by dissolved gas data (average = 12°C), and the partial pressures then were compared with the atmospheric record to determine the apparent year of recharge (Plummer and Busenberg, 2000). The CFC11 and CFC113 concentrations apparently were altered slightly by contamination from the Tygon pump head tubing or other non-atmospheric source, so CFC12 was the only CFC compound used for the ground water dating.

Because of the low hydraulic conductivities of the sediments at the upper basin transect, some piezometers did not yield water at a rate high enough to support continuous pumping. Water level recovery times for piezometers screened in the deeper parts of the section ranged from less than an hour to more than a day, which could be long enough to permit air-water exchange. The effects of atmospheric equilibration were tested in the piezometer with the slowest recovery rate (HC1 at 12.5 m elevation). The piezometer was emptied, then allowed to recover for 48 h. The water column then was pumped (~50–100 mL min–1) with the Cu tube inlet within a few cm of the bottom (Fig. 4), and aliquots were collected until the piezometer was emptied again. The CFC12 data indicate that the first three aliquots were not altered greatly by exchange with air in the piezometer, whereas the fourth (last) aliquot apparently was altered. This experiment indicates that ground water entering the recovering piezometer moved upward from the top of the screened interval by piston flow with little mixing. The first water to enter the piezometer after purging remained near the top of the water column and equilibrated partially with air, whereas later inflow remained relatively isolated from air. Subsequently during drawdown while samples were taken, the piston flow was reversed in the piezometer so the air-contaminated water reached the sampling tube last (Fig. 4). Ground water ages in the slowly recovering piezometers, therefore, were derived from CFC12 analyses of early aliquots from the bottoms of the piezometers, as were the concentrations of Ar and N2.


Figure 4
View larger version (16K):
[in this window]
[in a new window]

  		Fig. 4. Design, results, and interpretation of a pump test for collection of CCl2F2 (CFC12) in a piezometer with slow recovery (HC1 at 12.5 m) (see Fig. 3 for location): (A) the piezometer was purged and then allowed to recover for 48 h by inflow of ground water through the screen at the bottom of the tube; (B) water was pumped out through a Cu tube from the bottom of the piezometer and samples were taken in the sequence 1 to 4 until the piezometer was empty; (C) analyses indicate that the first three samples were uniformly undersaturated with respect to atmospheric CFC12, whereas the last sample was somewhat nearer to saturation. Gas exchange between ground water and air in the piezometer during the recovery stage may have been limited to the top of the water column (Sample #4, the first to enter the piezometer during recovery).

 
Samples for 3H-3He dating were collected from two piezometers that yielded water at high enough rates to be pumped continuously with the peristaltic pump. The 3H samples were collected in glass bottles and analyzed by liquid scintillation counting after electrolytic enrichment (Thatcher et al., 1976). The He isotope samples were collected in Cu tubes that were crimp-sealed in the field and the He and Ne were extracted for mass spectrometric analysis at Lamont-Doherty Earth Observatory Noble Gas Laboratory (Ludin et al., 1998). The age of each ground water sample (the difference between the time of recharge and time of sampling) was assumed to be equal to the time indicated by decay of 3H to 3He in a closed system, after adjustments for atmospheric gas contributions and excess terrigenic He with an assumed 3He/4He ratio of 2 x 10–8 (Schlosser et al., 1998).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Geology and Hydraulic Properties
The chemical and hydraulic properties of the ground water flow system in the upper basin area of the SERC-109 watershed are controlled in part by a horizontal stratigraphic boundary and a subparallel redox boundary within the tertiary sediments that underly the watershed. Core data and local mapping indicate that the upper basin transect is underlain by the Miocene Calvert Formation (above 14.5 m elevation), the Eocene Nanjemoy Formation (about 0–14.5 m elevation), and the Marlboro Clay (below about 0 m elevation) (Glaser, 1971; Shay, 1993; O'Connell, 1998) (Fig. 3). The Marlboro Clay is considered to be the lower boundary of the surficial ground water flow system, though the Nanjemoy Formation also includes some relatively impermeable units (Glaser, 1971). Near the upper basin transect, the stream was incised to about the elevation of the contact between the Nanjemoy and Calvert Formations (Fig. 3). Unweathered parts of the Nanjemoy Formation consist of greenish gray fine sand and silt with a large fraction of glauconite. The upper part of the Nanjemoy Formation includes a mottled zone with both brownish and greenish-gray components. The lower boundary of the mottled zone dips toward the stream, where it is at an elevation slightly lower than the stream bottom. The upper boundary of the mottled zone is in the lower part of the Calvert Formation, which mainly consists of fine-grained sand and diatomaceous silt and may contain reworked glauconitic material at the base (Glaser, 1971). The Calvert Formation above the mottled zone is weathered and brownish in color. The soils and saturated sediments of the Calvert and Nanjemoy Formations at the site are non-calcareous (Peterjohn and Correll, 1984; Jarriel, 1994). Pyrite commonly is present in unweathered parts of the glauconitic Nanjemoy Formation (Rabenhorst and Fanning, 1989) (present study).

The mottled zone largely corresponds to the parts of the Nanjemoy Formation and lower Calvert Formation that are within the range of annual variation of the water table. Though the lower boundary of the mottled zone was below the lowest water table elevation observed during this 3-yr study, it is plausibly within the range of variation over longer time scales. The position of the mottled zone, therefore, may be attributed to incomplete weathering and oxidation of Fe-bearing minerals including glauconite by reaction with oxygenated ground water and/or unsaturated zone air within the zone of variable saturation.

Hydraulic conductivities measured within the Nanjemoy Formation and the lower parts of the Calvert Formation were of the order of 10–9 to 10–6 m s–1 and generally decreased downward (Fig. 3). Measured total porosities in this part of the section were between about 40 and 60% (Shay, 1993; O'Connell, 1998). In shallower sediments above and within the upper part of mottled zone, beneath the riparian forest, is a network of macropores (~0.1 to >2 cm in diam.) that facilitated delivery of ground water to the stream during much of the year. The highest density of macropores was in the almost vertical face of the stream headcut. Large macropores near the top of the saturated zone were excavated and occupied by crawfish (Winston, 1994; O'Connell, 1998). Large macropores within the unsaturated zone included root casts and worm holes, and some were observed to have formed by ground water outflow during storms. Owing in part to the distribution of macropores in this horizon, total porosities typically range from around 50 to 80% and hydraulic conductivities are of the order of 10–7 to 10–3 m s–1 (Shay, 1993; Jarriel, 1994; O'Connell, 1998). Therefore, the riparian forest near the headcut of this deeply incised first-order stream is underlain by a relatively shallow (~0–2 m), higher conductivity layer that dips toward the streambed and a deeper, lower conductivity layer.

Ground Water Flow Patterns and Age Gradients
Vertical hydraulic head gradients (Fig. 5A) indicated potential for ground water recharge across most of the transect including the riparian forest under most conditions throughout the year. Horizontal head gradients consistently indicated ground water flow toward the stream, while gradients indicating strong vertical components of discharge were measured only beneath the stream. The steepness of the water table gradient adjacent to the stream increased with a rise in water table elevation. As a result, the horizontal head gradients driving flow toward the stream in the upper part of the section were higher during high-flow conditions in winter and spring and lower during summer and fall.


Figure 5
View larger version (28K):
[in this window]
[in a new window]

  		Fig. 5. Hydraulic heads and apparent ground water ages in the upper basin transect: (A) hydraulic heads in April 1996; (B) CCl2F2 (CFC12) apparent ages in July 1996; (C) CFC12 apparent age gradients in July 1996, with additional data from "other" sites in the upper basin area (not all from the same time period). Heavy dashed lines in (C) indicate approximate limits of apparent age gradients and recharge rates (for porosity of 0.5).

 
The importance of macropore discharge to total stream flow is illustrated by several sets of discharge measurements made under varying flow conditions (Table 1). These are considered minimum values because they are sums of measured macropore outflows and small macropores may have been overlooked. During low base flow in August 1994, the combined flow from 14 macropores in the incised reach of the stream accounted for about half the total stream flow at the mid-basin weir. Similarly during high base flow in February 1998, the combined discharge from 27 macropores accounted for about a third of the stream flow. The number of actively discharging macropores in the incised reach of the stream increased as the water levels increased beneath the watershed.


View this table:
[in this window]
[in a new window]

  		Table 1. Measured contributions of macropore discharge in the incised reach to total stream flow at W3 under varying flow conditions.

 
Although the conditions of the aquifer and the scale of sampling were not ideal for dissolved gas measurements, the apparent ground water ages derived from CFC12 concentrations increased systematically downward beneath the field and the riparian forest, consistent with ground water stratification under distributed recharge in both areas (Fig. 5B and 5C). At two sites where CFC12 and 3H-3He methods were compared, both methods indicated that the ground waters were of the order of decades old, but the CFC12 ages were significantly older than the 3H-3He ages (Table 2). This discrepancy was not evaluated in detail, but it may be related in part to a difference in gas confinement during recharge, to partial degradation of CFC12, or to mixing young and old water in the samples. Schlosser and others (1989) showed that vertical dispersion or diffusion could cause significant loss of tritiogenic 3He from ground water while it was being produced if vertical velocities near the water table were less than about 0.2 to 0.5 m yr–1. The problem of confinement should be less severe for CFCs, which have lower diffusion coefficients and shallow ground water concentrations that are close to equilibrium with air (Cook and Solomon, 1997). In this case, both methods might yield apparent ages that are younger than the recharge ages because of back-exchange with unsaturated zone air, but in the absence of contamination or degradation, the CFC12 ages would be more nearly correct than the 3H-3He ages. If the CFC12 were partially degraded, then the apparent CFC ages would be too large. One sample (HC1 at 1.9 m) with an apparent CFC12 age that was anomalously old in comparison to others nearby had an anomalously high concentration of CH4 (750 µmol L–1, compared with <2 µmol L–1 elsewhere), which could indicate conditions favorable for CFC degradation (Plummer and Busenberg, 2000). In either case, the 3H-3He ages are considered to be minimum values and they confirm the CFC data indicating ground water travel times of the order of decades in the deeper parts of the sampled transect below the mottled zone.


View this table:
[in this window]
[in a new window]

  		Table 2. Comparison of CCl2F2 (CFC12) and 3H-3He dating of ground waters from a recharge area (U12P) and a discharge area (FP3I) (see Fig. 1 for locations).{dagger}

 
Apparent vertical velocities derived from CFC data in the deeper part of the flow system beneath the low water table position were approximately 0.04 to 0.1 m yr–1, corresponding to recharge rates of around 2 to 5 cm yr–1, given an average porosity of 0.5 (Fig. 5B and 5C). The estimated recharge rate of the deeper ground water reservoir is low in comparison to those of many surficial unconsolidated aquifers in the region, which commonly are of the order of 20 to 30 cm yr–1 (Cushing et al., 1973; Dunkle et al., 1993). This contrast is presumed to result from the relatively low hydraulic conductivity in the Nanjemoy Formation and lower Calvert Formation. Vertical velocities in the upper part of the saturated zone during times of moderate to high water table elevation appear to be distinctly higher, though not well defined because of uncertainties in the CFC ages in the 1990s and the transient volume of water in this zone. Substantially higher recharge rates and fluxes in the upper part of the flow system are consistent with higher average hydraulic conductivities in the Calvert Formation above the redox transition and with the presence of a macropore network beneath the riparian forest.

Ground waters discharging to the stream near the upper-basin transect ranged from 0 to more than 30 yr in age (Fig. 5B). Samples collected from two macropores in 1994–1995 had CFC12 concentrations ranging from about 2.47 to 2.77 pmol L–1, indicating apparent ages of around 0 to 4 yr assuming average equilibrium temperatures of 12°C. Stream samples from five locations in 1994–1995 had CFC12 concentrations ranging from about 1.80 to 2.50 pmol L–1, with an average of 2.15 pmol L–1, indicating apparent ages of around 4 to 12 yr (average 8) assuming equilibration at 12°C. For stream samples collected in May 1995, the apparent ages would be less if calculated using the measured water temperatures (15–18°C), but they still indicate a substantial fraction of the discharge mixture was not modern. Stream water at the upper basin transect in November 1994 had 2.25 pmol L–1 of CFC12, which would indicate an apparent age of about 7 yr at 12°C or 9 yr at the measured stream temperature of 9°C. The CFC concentrations in macropore and stream samples are likely to have been affected to varying degrees by mixing and air-water exchange; nevertheless, the limited macropore data are consistent with a high proportion of shallow (younger) ground water from the relatively permeable part of the flow system, whereas the stream data may indicate additional contributions from deeper (older) parts of the flow system.

Chemical and Isotopic Gradients in Ground Water and Aquifer Sediments
Concentrations of NO3– were relatively high across the upper basin transect near the water table, but decreased abruptly downward within the mottled zone near the contact between the Nanjemoy and Calvert Formations (Fig. 6). Beneath the field and much of the riparian forest, SO42– concentrations increased downward as NO3– concentrations decreased, while the pH values and alkalinities remained low. The deeper ground waters in the section contained substantial amounts of excess N2 (for example, Fig. 7), which is interpreted to be a product of denitrification. These results indicate that the ground waters were stratified chemically and that the oxidation state decreased downward beneath both the field and the riparian forest. The concentrations of excess N2 in the deeper samples (where NO3– was not present) typically were less than the concentrations of NO3– near the water table, indicating NO3– concentrations in recharge may have changed by as much as a factor of 3 (e.g., from around 200 µmol L–1 in the 1960s and 1970s to about 600 µmol L–1 in the 1990s).


Figure 6
View larger version (53K):
[in this window]
[in a new window]

  		Fig. 6. Distributions of selected solutes (A through F) in the SERC-109 upper basin transect, sampled 5 Apr. 1996. Additional NO3– data are given for selected samples from late April 1996 (italics). Concentrations are in µmol L–1. Zero values are <10 for NO3– and <50 for HCO3–.

 

Figure 7
View larger version (24K):
[in this window]
[in a new window]

  		Fig. 7. Dissolved gas data for piezometer nest HC9 beneath the cultivated field, indicating effects of denitrification: (A) concentrations of Ar and N2, with solid curves indicating aqueous gas concentrations in equilibrium with humid air for a range of temperatures (10–20°C) and addition of either 0 or 2 cm3 (STP) L–1 unfractionated air, and with shading indicating approximate concentrations assumed for recharge in the denitrified samples; (B) isotopic composition of dissolved N2 gas, showing the interpreted value for the denitrification component.

 
The relation between denitrification and the stratigraphic and geochemical features of the tertiary marine sediments is illustrated by a ground water profile at HC9 beneath the agricultural field at the upgradient end of the transect (Table 3; Fig. 8). Similar features were observed beneath the riparian forest at HC7 and HC1 (O'Connell, 1998). In the oxic zone of the HC9 profile, the {delta}15N values of NO3– were between +3 and +6{per thousand}, typical of ground water NO3– beneath fertilized agriculture at this site and elsewhere in the region (Böhlke, 2003). In the suboxic zone where NO3– was not detected, the {delta}15N values of the excess N2 were about +5{per thousand} (Fig. 7), consistent with complete denitrification of NO3– similar to that still present above the denitrification zone. Coexisting NO3– and excess N2 at the 14.8 m elevation both had similar {delta}15N values, which would be inconsistent with isotopic fractionation associated with partial denitrification. These data are interpreted to indicate that the 20-cm piezometer screen at 14.8 ± 0.1 m yielded a mixture of undenitrified and denitrified water from both sides of a redox boundary. Elevated {delta}15N[NO3–] values were measured in some other ground water profiles at various downgradient locations within the SERC-109 watershed (including +18{per thousand} at HC1, 14.3 m), indicating that partially denitrified waters were present locally (Fig. 9).


View this table:
[in this window]
[in a new window]

  		Table 3. Chemical and isotopic data and ground water ages for the vertical profile beneath the cultivated field at HC9 (see Fig. 3 for location). DIC, dissolved inorganic carbon.{dagger}

 

Figure 8
View larger version (38K):
[in this window]
[in a new window]

  		Fig. 8. Vertical variation of chemical and isotopic data at piezometer nest HC9, sampled 19–26 Apr. 1996 (Table 3). Data for excess N2 were derived from measurements of total N2 (Fig. 7). Measured values for selected sources of N (open circles) and S (open diamonds) are shown for comparison. X° and X indicate samples related by the reaction given in Eq. [4].

 

Figure 9
View larger version (27K):
[in this window]
[in a new window]

  		Fig. 9. Variation of {delta}15N with concentration of NO3– in ground water and stream water under varying flow conditions. All data are from the SERC-109 watershed (Fig. 1). Values of {delta}15N in the stream were relatively constant over a wide range of concentrations at high base flow and low base flow, but deviated toward precipitation values briefly during a storm event.

 
The downward abrupt decrease of NO3– in the HC9 profile was accompanied by an increase in the concentration of SO42– and a decrease in the {delta}34S value of the SO42– (Fig. 8). Unoxidized parts of the Nanjemoy Formation had variable but substantial amounts of extractable solid Fe disulfide (FeS2), which is presumed to be mainly pyrite on the basis of regional mineralogic analyses of the Nanjemoy Formation and other glauconitic sediments (Otton, 1955; Drobnyk, 1965). Core material from the HC9 piezometer nest at 13.7-m elevation had FeS2–S = 1.3 g kg–1 with {delta}34S = –36{per thousand}. Two other sediment samples from different locations in the SERC-109 watershed had total S = 0.04 and 0.01 g kg–1 with {delta}34S = –25 and –19{per thousand}, respectively (O'Connell, 1998). Above the denitrification zone at HC9, {delta}34S[SO42–] values were around 0 to 2 {per thousand}, slightly lower than those of atmospheric deposition (+3 to +5{per thousand}) and fertilizer (+12{per thousand}). Below the denitrification zone, {delta}34S[SO42–] values were as low as –31 {per thousand}, approaching that of FeS2 in the sediment at the same locality (–36 {per thousand}). Several other samples of reduced ground water at the SERC site had {delta}34S[SO42–] values less than –20 {per thousand}, ranging down to –33{per thousand} (O'Connell, 1998). Dissolved inorganic C was present mainly as neutral CO2 in the low-pH ground waters throughout the profile at HC9, it did not change systematically in concentration, and it had {delta}13C values that were essentially constant at –15 ± 1{per thousand}. These results indicate that O2 reduction and denitrification were accompanied by production of SO42– from oxidation of FeS2, but apparently not with substantial production of alkalinity from oxidation of organic C.

In a small part of the section near the stream, Fe concentrations and alkalinities were relatively high, and SO42– concentrations were low (Fig. 6), consistent with Fe3+ and SO42– reduction. Concentrations of Cl were relatively high near the water table in the same general area. This reduced ground water may have come from a different source area not included in the recharge area of the transect, perhaps having entered the section from an upstream direction, or it may indicate reduction of Fe and SO42– by organic C within the aquifer downgradient from the zone of denitrification. Small and variable concentrations of NO3– detected in some of the deep piezometers (Fig. 6A) were associated with unusually low {delta}15N[NO3–] values (1–2{per thousand}) along with substantial amounts of excess N2. These data seem to be anomalous and may represent artifacts, such as NO3– production within the piezometers, or vertical leakage in the piezometer nests.

Chemical and Isotopic Variations in Discharge
The concentration of NO3– in the stream near the upper basin transect varied seasonally between about 0 and 450 µmol L–1 and was positively correlated with stream flow under most flow conditions except storm events (Fig. 2 and 10) (O'Connell, 1998). Concentrations of NO3– in discharging macropores typically were equal to or higher than the stream values (Fig. 10C). Farther downgradient in the SERC-109 watershed, stream NO3– concentrations generally were lower and less variable than in the upper basin area (Fig. 2). Although the stream NO3– loads increased downstream, the data indicate that the incised reaches received more NO3– in discharge per unit watershed area than did the non-incised reaches. Other constituents that had relatively high concentrations in the upper incised reaches of the SERC-109 stream in comparison to the lower reaches of the SERC-109 stream or to the SERC-110 stream in a nearby forested watershed include Mg, Ca, Cl, and possibly Sr, all of which are plausibly considered to have been enriched in ground water as a result of agricultural activities (Denver, 1989; Böhlke and Denver, 1995; O'Connell, 1998; Böhlke and Horan, 2000).


Figure 10
View larger version (21K):
[in this window]
[in a new window]

  		Fig. 10. Partial records in the upper basin area for (A) stream flows, (B) water table elevations in a near-stream piezometer, (C) stream and macropore NO3– concentrations, and (D) stream sulfate concentrations. The light curve in (A) indicates daily stream flow values at W109, divided by 10 (T. Jordan, personal communication, 2002).

 
The {delta}15N values of NO3– in the stream during most flow conditions were relatively constant despite large variations in the NO3– concentration (Fig. 9). The average {delta}15N[NO3–] value in the stream (+4.9 ± 1.3{per thousand}) was similar to those of oxic ground waters and macropore waters. These results indicate that most of the base flow NO3– was delivered to the stream in shallow oxic ground water, and that partially denitrified ground water with isotopically fractionated NO3– was not a major component of discharge. The average {delta}15N value of NO3– at the SERC-109 watershed outlet was indistinguishable from the average value in the upper basin area, indicating that (i) isotopically fractionated NO3– was not a major component of discharge in other parts of the watershed, and (ii) NO3– was not fractionated isotopically by in-stream denitrification.

In contrast, the {delta}34S[SO42–] values in the stream exhibited large variations, ranging from slightly less than that of atmospheric SO42– to much lower values approaching those of sedimentary sulfide (Fig. 11). The higher {delta}34S values occurred during higher flow periods and the lower values occurred during low base flow. By analogy with the ground water transect data, these results are interpreted to indicate that the stream SO42– was derived largely from surficial sources (atmospheric deposition ± fertilizer) during high flow and largely from deeper ground water sources (oxidation of FeS2) during low flow. Stream SO42– concentrations and {delta}34S values generally were inversely related, as in some of the ground water profiles (Fig. 8), except for some late summer base flow samples that had low SO42– concentrations and negative {delta}34S values.


Figure 11
View larger version (20K):
[in this window]
[in a new window]

  		Fig. 11. Relation between water table elevations in a near-stream piezometer nest and (A) stream flows, (B) stream NO3– concentrations, and (C) stream values of {delta}34S[SO42–] (data were collected at irregular intervals between January 1993 and May 1996).

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
Ground Water Flow and Transport of Agricultural Nitrate
Delivery of water and NO3– from the fields to the upper incised parts of the SERC-109 stream may be considered in the context of several major flow regimes controlled by the interaction of climatic variables with geologic and geomorphic features of the watershed (Fig. 12). Low base flow occurred in late summer and fall when the water table was in the lower part of the mottled zone near the contact between the Calvert Formation and underlying Nanjemoy Formation and discharge from less permeable deeper parts of the flow system was relatively important. Seasonally high base flow occurred in winter and spring when the water table was up within the weathered parts of the Calvert Formation and discharge from more permeable shallow parts of the flow system were relatively important. Storm ("event") flow was relatively short-lived (hours to days) and included saturated overland flow and infiltration-excess overland and subsurface flow (O'Connell, 1998).


Figure 12
View larger version (40K):
[in this window]
[in a new window]

  		Fig. 12. Schematic diagram of ground water flow regimes contributing to stream flow under different flow conditions in the incised upper basin area of the SERC-109 watershed. Correlated variations in flow and NO3– concentrations in the stream are related to vertical stratification of hydraulic conductivity and redox status of the aquifer sediments.

 
The distribution and isotopic composition of NO3– in ground water and surface water at the SERC site indicates that it was largely derived from agricultural (cropland) recharge (Correll et al., 1992; Jordan et al., 1997; present study). Recharge beneath the riparian forest apparently contained some NO3– brought in by overland flow from upgradient fields (O'Connell, 1998), but not as much as recharge beneath the fields. Combining the results of ground water dating with analyses of NO3– and estimates of excess N2 (Fig. 5 and 8), it appears that the initial NO3– concentrations of recharging ground waters beneath the agricultural fields may have increased over the last few decades or that the deeper ground waters may have had mixed land uses in the recharge area. Therefore, even in the absence of denitrification or other NO3– removal processes, discharge from the deeper parts of the flow system might have had lower NO3– concentrations than discharge from the shallower parts of the flow system.

In addition to possible age-dependent variation in recharging NO3– concentrations, at least two other factors had important effects on seasonal or long-term variations in stream NO3– concentration in the upper basin area: (i) denitrification removed NO3– completely from the deeper (older) ground water below an elevation of around 15 ± 1 m, and (ii) contrasting hydraulic properties caused relatively young ground water from the shallow flow system (>15 ± 1 m elevation) to be delivered to the stream much more rapidly than the older deeper ground water when water levels were high. The estimated recharge rate of the deeper flow system (2–5 cm yr–1) is only about 2 to 4% of the average annual precipitation rate and 8 to 21% of the average annual stream flow in the SERC-109 watershed (Correll et al., 1999). These water balance estimates indicate that the shallower flow system dominates stream flow on an annual basis. Young ground-water age fractions commonly are relatively abundant even in discharge from homogeneous aquifers with smooth age distributions (Zuber, 1986; Cook and Böhlke, 2000), but this pattern may be exaggerated where hydraulic conductivities decrease downward as in the upper-basin transect, where the thickness of the "shallow" flow system (0–2+ m) contributing most of the flow was only a small fraction of the total aquifer thickness (possibly as much as 15 m).

Measurements indicate that both water and NO3– from the shallow flow system were delivered to the stream largely by macropore discharge during moderate to high flow conditions (Table 1). For example, when 27 measured macropores contributed approximately a third of the total stream discharge from the incised reach in February 1998, they contributed more than half the total stream NO3– load. Stream NO3– concentrations were lower at locations farther down in the watershed (Fig. 2), where the stream was not incised and macropores were active discharge sites under a more limited range of flow conditions (O'Connell, 1998). The combined evidence from the outflow macropore measurements, downgradient stream variations, and ground water analyses indicates that the total NO3– load discharged from the SERC-109 watershed during base flow on an annual basis was largely derived from macropore discharge in the incised upper reaches of the stream. This high-NO3– discharge was diluted throughout the watershed by discharge of low NO3– ground water components that were denitrified at depth in the glauconitic sediments, denitrified near the surface in organic-rich saturated lowlands, or recharged within the riparian forest, which covered more area in the lower elevations.

Lithogeochemical Control of Denitrification
Microbial reduction of O2 and NO3– could be coupled with oxidation of a variety of electron donors in the aquifer including organic carbon, sulfide, or ferrous iron phases, for example:

Formula 1[1]

Formula 2[2]

Formula 3[3]
where C could represent various organic compounds and FeO could represent components of Fe oxides, carbonates, or silicates, including glauconite. Glauconite in marine sediments of the mid-Atlantic coastal plain can have Fe2+/Fe3+ ratios of around 0.35 (Fanning et al., 1989), and its abundance in the deeper parts of the upper basin transect could make it a potential source of electrons for denitrification. Syngenetic or diagenetic pyrite formed on or beneath the seafloor is common in coastal plain glauconitic sediments including the Nanjemoy Formation (Otton, 1955; Drobnyk, 1965; Rabenhorst and Fanning, 1989). In addition, late authigenic pyrite may be present in the aquifer where SO42– reduction has occurred. Although organic C was not analyzed in this study, other studies indicate that C may be present as scattered lignitic particles or as cryptic coatings on grains in the glauconitic sediments (Drobnyk, 1965).

Because the initial NO3– concentrations were different in the older (denitrified) and younger (undenitrified) ground waters, it is not possible to determine redox reaction stoichiometry by comparing directly the compositions of old and young waters along a flow path. This problem may be circumvented in some situations by comparing ground waters of similar ages that have taken separate paths and encountered contrasting geochemical environments (Böhlke and Denver, 1995; Böhlke et al., 2002), but this approach is not possible in the HC9 profile. Instead, it was assumed that the initial concentrations of redox-sensitive species in the water sampled at 13.9-m elevation (below the denitrification zone) were similar to those at 16.0-m elevation (above the denitrification zone) but with a lower initial NO3– concentration given by the excess N2 abundance in the denitrified sample. Given these assumptions, it is possible to account for reduction of O2 and NO3– between 16.0 and 13.9 m with FeS2 as the only electron donor in a stoichiometrically balanced equation that matches the observed profile (Table 3; Fig. 8):

Formula 4[4]

This reaction is supported by the S isotope results, which indicate that the SO42– in the denitrified sample (with {delta}34S = –31{per thousand}) is a mixture of surface-derived SO42– ({delta}34S = 0 to +12{per thousand}) and SO42– derived from oxidation of sediment sulfide (FeS2 with {delta}34S = –36{per thousand}). Because this reaction yields excess protons, and the sediments near the redox boundary are noncalcareous, the ground waters remain acidic through the profile. The lack of alkalinity production and the constant values of {delta}13C through the profile (Fig. 8) support the interpretation that organic C was not an important electron donor for reduction of O2 or NO3– upgradient from this site. Glauconite containing Fe2+ may have contributed to the reaction, releasing SiO2 (Fig. 6E), but this does not appear to be required as a major electron source.

Recharge of agricultural NO3– may be viewed as an anthropogenic weathering agent that has accelerated natural oxidative reactions in watersheds underlain by glauconitic sediments in the mid-Atlantic coastal plain (Böhlke and Denver, 1995). The concentration of FeS2 in the reactive sediment probably is variable, but if the sample from HC9 is representative of that part of the watershed (FeS2–S = 1.3 g kg–1), this could represent a potential electron supply of 410 mol e– m–3 of aquifer. If FeS2 oxidation were limited by downward advection of electron acceptors in 2 to 5 cm yr–1 of recharge, then the corresponding velocity of the oxidation front would be ~0.009 to 0.022 cm yr–1 for the O2 and NO3– reactant concentrations in Eq. [4]. That is, large amounts of NO3– can be removed and large amounts of SO42– can be produced without causing the redox boundary in the sediment to move substantially. Denitrification in pyritic aquifers is an indirect anthropogenic source of SO42– contamination in ground water (Kölle et al., 1985; Böhlke et al., 2002) that can affect the interpretation of watershed responses to changes in atmospheric SO42– deposition (Mann et al., 1998).

Role of the Riparian Zone in Nitrate Transport
Forested riparian zones commonly are considered to impede transport of NO3––bearing ground water from agricultural recharge areas to surface water bodies. In contrast, the riparian zone along this incised headwater stream, with locally steep hydraulic gradients and macropore permeability, apparently was an efficient transmitter of NO3– from upgradient sources to the stream during much of the year. An earlier conceptual model for the lower part of the SERC watershed includes substantial reduction of ground water NO3– from upgradient sources during transport beneath the riparian forest (Correll and Weller, 1989). In that model, NO3– removal occurred in vertically bounded regions where ground water from upgradient fields encountered reducing conditions related to the presence of the uncultivated riparian forests. Our data for the upper basin area are in agreement with the general observation that denitrification occurs in the subsurface, but they indicate a substantially different model for riparian zone function. Detailed vertical sampling beneath the upper basin transect indicates that denitrification occurs throughout the landscape and is not limited to areas beneath the riparian forest. Denitrification beneath forest and cropland is largely controlled by the mineralogy of marine sediments that were deposited and altered diagenetically in Eocene and Miocene time and partially weathered subsequently. Detailed vertical sampling has not been done all the way across the cornfield toward the ground-water divide, but it is reasonable to suppose that the thickness of the oxic ground-water wedge continues to increase along with the water-table elevation with increasing distance from the stream (Fig. 12). If so, then our results could be consistent with the finding of Peterjohn and Correll (1984) in other parts of the watershed that the concentrations of NO3– and O2 decreased toward the stream in wells with long screened intervals that intersect varying proportions of deep and shallow ground water. Our data illustrate the importance of vertical (rather than horizontal) geochemical gradients, and stream-channel incision (rather than riparian habitats or soils), as major controls of redox reactions within the saturated zone. Although the presence of the riparian forest was not linked with NO3– loss in the transect leading to the incised stream reach, the forest is in a good position to limit both recharge and discharge of NO3– even where it does not actively reduce the NO3–, because ground water flow paths from distant agricultural recharge areas are more likely to encounter denitrifying conditions at depth in the aquifer.

The patterns of flow and reaction beneath the riparian zone at the SERC-109 upper basin transect provide a useful contrast to the patterns documented with similar techniques in a glacial outwash sand plain in central Minnesota (Böhlke et al., 2002). Both sites exhibit similar vertical gradients in redox conditions and NO3– concentrations beneath agricultural fields owing to reaction of recharging contaminated meteoric water with reducing aquifer sediments including FeS2; however, the fate of the shallow oxic ground water wedge differs in the respective riparian zones. At the Minnesota site, shallow oxic NO3––bearing ground water from the upland flows beneath a semi-confining organic-rich valley-filling unit under a low elevation riparian wetland, resulting in an inverted redox gradient with anoxic ground water overlying oxic ground water. In this case, additional O2 reduction and denitrification occur during upward transport through a relatively wide discharge area in the valley-filling sediments, and the stream is relatively NO3– free. Although inverted redox gradients of this type may exist elsewhere in the SERC watershed, they were not identified at the SERC-109 upper basin transect where the forested riparian zone, with a relatively thick unsaturated zone and high permeability, was an area of recharge and of rapid transmission of NO3–rich ground water from upgradient sources to the stream. The contrast between the redox configurations and fate of NO3– in these settings illustrates the importance of near-stream geomorphology and sedimentation history for riparian zone function.

Seasonal Variations in Stream Nitrate and Sulfate
Streams draining agricultural watersheds commonly exhibit seasonal variations in NO3– concentrations that are positively correlated with stream flow. Several different processes or watershed characteristics could be responsible for seasonal variation in stream NO3– concentrations, for example (i) higher rates of denitrification in warm months; (ii) higher rates of assimilation of nitrogen by plants in warm months; (iii) higher rates of NO3– leaching from soils when plants are not growing and when the rate of ground water recharge is high. Correll and Weller (1989) attributed such a correlation to variation in the contact time between NO3––bearing agricultural ground water and denitrifying soils beneath the riparian forest. In their conceptual model, breakthrough of higher NO3– concentrations during colder months was attributed to relatively high horizontal flow rates throughout the aquifer that permit streamward migration of kinetically controlled redox zones.

Our data indicate that the major redox boundary affecting NO3– transport in the upper basin transect is sub-horizontal, and that the vertical component of ground water flow played a critical role in the fate of NO3– between recharge and discharge (Fig. 12). The overall positive correlation between stream flow and NO3– concentration can be related directly to the position of the water table with respect to the aquifer redox boundary (Fig. 11). When water levels were low throughout the watershed in late summer and fall, the water table near the stream was in the lower part of the mottled zone of the sediments and the low permeability saturated zone contributing to low base flow was almost entirely reduced (low O2 and NO3–). When water levels were higher throughout the watershed in winter and spring, the water table near the stream was in the high permeability oxidized zone, permitting NO3––rich ground water to flow rapidly to the stream beneath the riparian forest during high base flow. Seasonal variations in stream NO3– concentration can be attributed to variations in the relative proportions of oxic NO3––bearing ground water and suboxic denitrified ground water discharging from shallower and deeper parts of the flow system, respectively. The transmission of NO3– to the stream during high flow conditions was facilitated by the relatively high hydraulic conductivities of the macropore-rich shallow riparian sediments and by the high elevation of the surficial riparian soils above the saturated zone near the incised stream reach. In contrast to the concentration patterns described above, storm responses may include substantial dilution of NO3– at high flow followed by high rates of NO3– discharge in the post-storm recession, resulting in substantial short-term deviations from the correlations shown in Fig. 2, 9, and 11 (O'Connell et al., 1994; O'Connell et al., 1997; O'Connell, 1998).

Effects of the hydraulically and chemically layered ground water flow system on stream flow and NO3– concentrations described in this study may be similar to the effects of tile drainage systems in other agricultural settings. For example, Fenelon and Moore (1998) describe a small watershed in Indiana where tile drains delivered NO3––rich shallow ground water to a stream during high base flow conditions, whereas during low base flow conditions, drain flow ceased and the stream was fed by discharge of deeper reduced ground waters containing little or no NO3–. The seasonality of NO3– delivery through the riparian macropore network to the incised stream in the upper-basin area of the SERC-109 watershed can be interpreted as a more natural analog of such a system.

Our results also may be compared with the effects of layered flow systems in fractured rock aquifers in other geologic settings. For example, in areas of the Mahantango watershed on metamorphic rocks in Pennsylvania, USA, the shallow part of the ground water flow system has relatively high permeability because of weathering and a higher density of open fractures near the land surface (Schnabel et al., 1993; Lindsey et al., 2003). Stream concentrations of NO3– and SO42– at Mahantango were positively correlated with stream flow in part because the deeper flow system was recharged beneath mixed-use and non-agricultural land and the shallower flow system contained more agricultural recharge (Schnabel et al., 1993). In addition, the deeper flow system contains a larger fraction of ground water that was old enough to have been recharged when agricultural chemical application rates were lower (Lindsey et al., 2003). Seasonal variations in stream NO3– concentrations were relatively subtle at Mahantango because denitrification was not as widely distributed in the aquifer there. The positive correlation between flow and SO42– concentrations at Mahantango is qualitatively similar to variations in forested watersheds that have been attributed to seasonal storage and release of SO42– in soils (Lynch and Corbett, 1989), whereas variations in stream SO42– concentrations at SERC-109 are complicated by the combined effects of seasonal storage in soils, major aquifer sources of SO42–, and localized SO42– reduction. In a small watershed overlying fractured metamorphic rocks in France, Pauwels and others (2001) describe seasonal variations in stream NO3– concentrations that are linked in part to denitrification and sulfide oxidation at depth, which is more like the situation in the upper basin of the SERC-109 coastal plain site.

Layered flow systems like these are in contrast to many thicker, more permeable alluvial and coastal plain aquifers that have smoother age distributions, older median ages, and commonly (but not always) more uniform redox conditions (Böhlke and Denver, 1995; Szabo et al., 1996). Discharge from the layered systems is more likely to vary seasonally and to be dominated on an annual basis by relatively young ground water, whereas the more uniform flow systems may deliver NO3– to streams with less seasonal variation, but with more substantial long-term delay with respect to changes in agricultural practices (Böhlke, 2002).

Contrasting Isotope Effects in Nitrate and Sulfate
Stable isotope fractionation effects commonly are used as evidence for denitrification in ground water and surface water (Mariotti et al., 1988; Kellman and Hillaire-Marcel, 1998; Groffman et al., 2006). In the SERC-109 watershed, there was essentially no systematic variation in the {delta}15N[NO3–] values in the stream, even though the concentration variations are attributed largely to ground water denitrification, and despite evidence locally for isotopic fractionation of NO3– undergoing denitrification in ground water (Fig. 9). Lack of seasonal isotopic variation in the stream is interpreted to indicate that fractionated partially denitrified ground water NO3– was only a minor component of the total ground water NO3– discharging to the stream. Because the thickness of the subsurface denitrification zone containing isotopically fractionated NO3– was small compared with the thicknesses of the oxic zone containing unfractionated NO3– and the anoxic zone containing no NO3–, the effect of denitrification on the isotopic composition of discharge is similar to that of dilution. This hypothesis is consistent with the pattern of ground water stratification in the upper basin transect (Fig. 6 and 8).

In contrast to {delta}15N[NO3–], values of {delta}34S[SO42–] in the stream varied substantially with variations in discharge from denitrified and undenitrified parts of the ground water flow system (Fig. 11). The {delta}34S values of stream SO42– were higher (more like those of surficial S sources) when water levels were high and lower (more like those of sedimentary sulfide) when water levels were low. The SO42– concentrations were not as well correlated with stream flow as were the {delta}34S values. This discrepancy may be attributed in part to variations in the amounts of SO42– from surficial sources as well as to localized SO42– reduction. Whereas Eq. [4] (Fig. 8) would seem to imply an inverse relation between NO3– and SO42– in discharge, this relation would be opposed by either larger contributions of atmospheric SO42– at high flow or more SO42– reduction at low flow. Microbial SO42– reduction generally results in an increase of the {delta}34S values of the residual SO42– (Chambers and Trudinger, 1979) but this isotope effect may have been minimized in discharge in the same way that the N isotope effect of denitrification was minimized, by having only a minor fraction of the total discharge coming from relatively thin zones of incomplete reaction. Furthermore, because the difference in the isotope source signatures between surficial and sedimentary S is relatively large (~40{per thousand}), substantial loss and isotopic fractionation of SO42– could have occurred without altering the isotopic source signatures beyond recognition.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
We compared spatial and temporal patterns in ground water flow and chemistry with seasonal variations in stream discharge and chemistry to determine the effects of channel morphology and subsurface features on riparian zone function and delivery of agricultural NO3– to an incised stream. Ground water ages, hydraulic heads, and chemical compositions were stratified vertically beneath a cultivated field and an adjacent riparian forest. The ground water data indicate that NO3– losses along ground water flow paths leading to the stream were not the result of interactions with riparian forest vegetation or soils, but instead were caused primarily by interaction with FeS2 in buried marine sediments below a surficial weathered horizon. Glauconitic marine sediments are widely distributed beneath the mid-Atlantic coastal plain region of North America and may have an important role in reducing NO3– transmission through agricultural watersheds elsewhere in the region. The relative roles of sulfides, other ferrous minerals, and organic carbon as electron donors may vary locally, depending on the conditions attending deposition, diagenesis, and weathering of the sediments.

As a result of the normal vertical gradients in ground-water redox conditions and hydraulic conductivities, seasonal variations in stream NO3– concentrations were directly correlated with water table elevations near the stream. During low base flow in summer and fall, when the water table was low throughout the watershed, stream NO3– concentrations were low because discharge to the incised stream consisted largely of old ground waters that were denitrified by reducing sediments at depth. During high base flow in winter and spring, stream NO3– concentrations were high because discharge was mainly younger ground water that was delivered to the incised stream from the zone of oxic ground water that extended beneath the riparian forest soils and above the reducing marine sediments. Stream {delta}34S[SO42–] values were directly correlated with flow because of variations in the proportions of atmospheric (high {delta}34S) and sediment (low {delta}34S) sources of S in the shallow and deep flow systems, respectively. Stream {delta}15N[NO3–] values were relatively constant because they generally represented mixtures of undenitrified and completely denitrified ground water components. Ground waters following shallow flow paths through the riparian zone were transmitted to the stream rapidly and without substantial NO3– loss largely through a network of outflow macropores.

Seasonal correlations between stream NO3– concentrations and regional water levels are observed in many different watersheds, but the temporal patterns and underlying causes may differ greatly from place to place. The relation between ground water stratification and seasonal variation of flow and NO3– concentration in this agricultural coastal-plain system apparently has analogs in other agricultural areas with layered hydraulic and geochemical properties, such as tile-drained fields and fractured rock aquifers with permeable weathering zones. These situations are different from many coastal plain and alluvial aquifers with more uniform hydraulic properties and lower hydraulic gradients near the stream, where seasonal variation in flow and chemistry may be less pronounced. Also, the normal recharging redox stratification (oxidized over reduced) beneath the riparian forest in the incised part of the SERC-109 stream valley is in marked contrast to the inverted gradients (reduced over oxidized) that exist beneath some riparian wetlands. The latter types of systems are more likely to prevent NO3– from discharging under a wider range of flow conditions, which they may do in some other parts of the SERC watershed.


   ACKNOWLEDGMENTS
 
Much of the work summarized here was done as part of a Ph.D. dissertation by Mike O'Connell, who died unexpectedly while this paper was being prepared. Mike's friendship and intellect are sincerely missed. This study was supported by the U.S. National Science Foundation (EAR-9220170), the U.S. Geological Survey (USGS) National Research Program, the Maryland Water Resources Research Center, and the Geological Society of America Penrose Foundation. D. Correll, T. Jordan, D. Weller, and others at the Smithsonian Environmental Research Center provided access to the site, assistance in the field, background data, and helpful discussions. Analytical support at USGS was provided by the CFC laboratory (L.N. Plummer, E. Busenberg), tritium laboratory (R.L. Michel), stable isotope laboratory (T.B. Coplen), and chemistry laboratory (M. Doughten). Assistance with sampling and analysis was provided by R. Carmody, G. Casile, T.B. Coplen, T. Councell, C. Gwinn, J. Hannon, J. Jarriel, J. Mann, J. Wayland, and P. Widman. J. Denver and T. Jordan provided helpful comments on an early version of the manuscript.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 J. Environ. Qual.Home page
C. T. Green, L. J. Puckett, J. K. Bohlke, B. A. Bekins, S. P. Phillips, L. J. Kauffman, J. M. Denver, and H. M. Johnson
Limited Occurrence of Denitrification in Four Shallow Aquifers in Agricultural Areas of the United States
J. Environ. Qual., May 1, 2008; 37(3): 994 - 1009.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Böhlke, J. K.
Right arrow Articles by Prestegaard, K. L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Böhlke, J. K.
Right arrow Articles by Prestegaard, K. L.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Böhlke, J. K.
Right arrow Articles by Prestegaard, K. L.
Related Collections
Right arrow Fate
Right arrow Nitrogen
Right arrow Geochemical Processes

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Soil Science Society of America Journal Journal of Plant Registrations The Plant Genome