| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a U.S. Geological Survey, Water Resources Division, 345 Middlefield Road, MS 439, Menlo Park, CA 94025
b Dep. of Chemical Engineering and Materials Science, Univ. of California, Davis, CA 95616
* Corresponding author (jhduff{at}usgs.gov)
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
3 mg N L–1 beneath the ridge (80 m from the channel) to 0.01 to 1.0 mg N L–1 at wells 1 to 3 m from the channel. The Cl– concentrations and NO3/Cl ratios decreased toward the channel indicating NO3– dilution and biotic retention. In the bankside well transect parallel to the stream, two distinct ground water environments were observed: an alluvial environment upstream of a relict beaver dam influenced by stream water and a hillslope environment downstream of the relict beaver dam. Nitrate was elevated to levels representative of agricultural runoff in a third well transect located 5 m from the stream to assess the effectiveness of the riparian zone as a NO3– sink. Subsurface NO3– injections revealed transport of up to 15 mg N L–1 was nearly conservative in the alluvial riparian environment. Addition of glucose stimulated dissolved oxygen uptake and promoted NO3– retention under both background and elevated NO3– levels in summer and winter. Disappearance of added NO3– was followed by transient NO2– formation and, in the presence of C2H2, by N2O formation, demonstrating potential denitrification. Under current land use, most NO3– associated with local ground water is biotically retained or diluted before reaching the channel. However, elevating NO3– levels through agricultural cultivation would likely result in increased NO3– transport to the channel.
Abbreviations: BDOC, biologically available dissolved organic carbon • DOC, dissolved organic carbon • DO, dissolved oxygen • SRP, soluble reactive phosphorus • NO3/Cl, nitrate N/chloride ratio • NO3/Br, nitrate N/bromide ratio
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Numerous ground water studies of agricultural runoff (e.g., Peterjohn and Correll, 1984; Lowrance et al., 1984; Jacobs and Gilliam, 1985; Hill, 1996) reported that NO3– transported through riparian zones is effectively depleted, often over short distances (Hedin et al., 1998; Cey et al., 1999; Hill et al., 2000). The primary biological sinks for subsurface NO3– removal in natural riparian communities and near-stream wetlands are denitrification and plant uptake (Cooper, 1990; Groffman et al., 1992; Jordan et al., 1993; Pinay et al., 1993; Hill et al., 2000).
Nitrate processing in riparian ground water is controlled by the hydrogeomorphic setting (Phillips et al., 1993; Hamilton and Helsel, 1995). Interaction of ground water N with riparian biota depends on subsurface flow paths that intercept the shallow root zone and soils conducive for denitrification. These flow paths occur where a shallow impermeable sediment layer or aquiclude forces the shallow ground water into biotically active riparian habitats (Jacobs and Gilliam, 1985; Cooper, 1990; Jordan et al., 1993; Cey et al., 1999; Hill et al., 2000).
Biotic NO3– removal has been demonstrated at the upland riparian interface (Peterjohn and Correll, 1984; Jacobs and Gilliam, 1985; Cooper, 1990; Haycock and Pinay, 1993; Cey et al., 1999), the riparian stream interface (Hedin et al., 1998), and in peaty soil deposits, buried channel deposits, and riparian "hotspots" (Fustec et al., 1991; Haycock and Burt, 1993; Devito et al., 2000; Hill et al., 2000; McClain et al., 2003). Böhlke and Denver (1995), Hedin et al. (1998), Hill et al. (2000), and Böhlke et al. (2002) have demonstrated that the presence of electron donors (and redox zonation) can have a strong bearing on the location of the denitrification front of subsurface NO3– plumes and that reduction of NO3– can be coupled to oxidation of reduced iron and sulfur phases in place of organic C.
Other agricultural ground water studies have found that riparian zones were ineffective for lowering NO3– concentrations in ground water (e.g., Böhlke and Denver, 1995; Pinay et al., 1998; Puckett et al., 2002; Puckett and Hughes, 2005). These observations may in part be attributed to absence of shallow aquicludes, wherein N bypasses rooting zones and directly enters the stream (Cey et al., 1999; Hill et al., 2000). Nitrate processing is ineffective or underutilized in riparian ground waters because of low biologically available dissolved organic carbon (BDOC) (Hill et al., 2000), high O2 levels (Cey et al., 1999; Hill et al., 2000), and low NO3– levels (Cooper, 1990; Haycock and Burt, 1993; Hill et al., 2000).
The large body of literature from riparian ground water studies of agricultural runoff has demonstrated (i) a dependence on riparian interactions for reducing NO3– in agricultural runoff and (ii) variability in NO3– retention mechanisms throughout agricultural riparian ecosystems. Collectively, this body of literature shows the futility of utilizing any one approach to manage NO3– runoff. These studies demonstrate that strategies for effective NO3– management in agricultural landscapes benefit from multiple methods that promote ground water interaction with natural C and energy sources in riparianzones.
The purposes of this paper are to (i) examine the relationship between local ground water flows and NO3– transport in a natural, wooded-alluvial riparian zone of a headwater drainage and (ii) determine the controls on NO3– transformations at natural background and elevated NO3– levels representative of regional agriculture ground water. This study was conducted at the Shingobee River, near the headwaters of the Mississippi River, where 65% of the land use is forested or pasture. In size, riparian habitat, and geomorphology, the Shingobee River is representative of streams draining agricultural landscapes in the central Minnesota sand plains.
Site Description
The study was located in the headwaters of the Shingobee River, a second-order sand bed stream in north central Minnesota (Fig. 1a). The Shingobee headwaters area is located on a small topographic ridge that extends south from the much larger Itasca moraine. The ridge forms the divide between the Mississippi and Crow Wing River drainage (Winter and Rosenberry, 1997). The Shingobee River begins as a wetland seep in relatively young glaciated terrain and connects a series of lakes before entering the study area. Upstream of the study site, the river cuts into a regional aquifer resulting in significant ground water contribution to stream flow. This ‘old’ ground water has a recharge date pre-1945 (chlorofluorocarbon dating, Triska et al., 2002) and is relatively low in dissolved oxygen (DO) (<1 mg L–1), NO3– (<1 µg N L–1), and Cl– (chloride) (1.2 mg L–1), and high in NH4+ (ammonium) (350 µg N L–1) (Table 1).
|
|
0.15 m m–1). Stream water chemistry varies seasonally. Nitrate and NH4+ are 18 and 17 µg N L–1, respectively, in September and 92 and 345 µg N L–1 in January (Table 1). Chloride is approximately 0.85 mg L–1 year round. The study site was adjacent to a relict beaver dam, a common feature of the Shingobee headwaters area (Fig. 1b). The dam resulted in upstream sediment deposition. At the time of this study, the dam was approximately 0.5 m tall, but riparian wood deposits suggest the top was once 0.2 to 0.5 m higher. Locally derived ground water, the focus of this study, originates from recharge to the hillslope adjacent to the study site. This ‘new’ ground water is <8 yr old (Triska et al., 2002). Upland land use in this vicinity is open pasture with intermittent grazing. This local ground water flows to the channel above a till layer in a nearly pure sand unit. Local ground water is high in Cl– (4 mg L–1), NO3– (2 to 5 mg N L–1), and DO (>8 mg L–1), and low in NH4+ (<20 µg N L–1) (Table 1).
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Well Installation
Transect 1: Hillslope Wells
Transect 1 consisted of Shingobee headwaters wells SHW 1, 13, 12.5, 12, 11, and 10 going down slope toward the stream (Fig. 1b). The ridge well, SHW 1, was installed by truck-mounted drill rig in 1992 to a depth of 9 m and screened between 8.3 and 8.5 m. A till layer exists between 11.3- and 14.3-m depth (Winter et al., 2003), which is approximately 1 m below the elevation of the streambed. SHW 13 and 12 were installed with a portable gasoline-powered auger. The two wells closest to the stream, SHW 10 and SHW 11, were hand-augered through coarse gravel and cobble. SHW 12.5 was also installed by hand auger.
Transect 2: Bankside Wells
The bank wells (SHW 16 through 21) were installed with a gasoline-powered auger through sand to the water table. SHW 16 through 19 were located in sandy deposits with minimal gravel and cobble. Coarse gravel-cobble deposits were encountered in the vicinity of SHW 12, 20, and 21, similar to SHW 10 and 11. No significant organic matter deposits were observed during well installation. All SHW wells were constructed of 5-cm diam. polyvinyl chloride (PVC) pipe with wire-wound screen 0.2 m long with 0.25-mm slot size. The area around the well casings was backfilled with native material.
Experimental Well Array
The experimental well array was located at the intersection of Transects 1 and 2, near the boundary between hillslope and alluvial deposits (Fig. 1c). The experimental wells consisted of 1.9-cm diam. stainless steel drivepoints screened over 23 cm and driven 2.3 m into the alluvium with a fence post driver. A shallow cobble-gravel layer was encountered during installation. The injection well was a 2-m length of 1-cm stainless steel tubing with a sharp point driven 1.8 m into the sediments. The bottom of the tubing contained four vertical slits 0.025 cm wide and 1 cm long equally spaced around the periphery. We estimate that the slits on the injection well were 0.4 m above mid-screen of the receiving wells. Of ten experimental wells installed, preliminary injections revealed tracer at only four. Wells 1 and 3 were on a nearly straight line passing through the injection well (Fig. 1b). Wells 2 and 4 were to the left of the line, but in the path of the tracer.
Sampling and Chemical Methods
Well SHW 1 was sampled 20 times for nutrients between September 1992 and 1998. All other wells were sampled five to six times from 1996 through 1998. Before collecting samples from Transects 1 and 2, the wells were pumped an equivalent of three well volumes using a battery operated peristaltic pump. In the experimental well array, a peristaltic pump was directly connected to 1.3-cm diam. Teflon tubing that ran down the steel pipe and attached to a barbed fitting on the screened drive point. A minimum of three drivepoint volumes were withdrawn before collecting a sample.
Two wells, F-94 (28.5 m deep) and F-197 (60 m deep), located near SHW 1, were sampled five times from 1996 through 1998 to determine nutrient concentrations deep in the aquifer. These wells were installed for another study (Winter and Rosenberry, 1997). Two to 3 h of pumping with a submersible pump were required to evacuate a single well volume.
Dissolved oxygen and conductivity samples were collected in 60-mL biological oxygen demand (BOD) bottles overfilled from the bottom with approximately three bottle volumes then sealed. Dissolved oxygen was measured using a stirring BOD probe, after which conductivity was measured at room temperature.
Nutrient samples were filtered (0.45-µm membrane), in line, into 60-mL polyethylene bottles. Bottles were rinsed three times with filtered water before collecting the sample. All nutrient samples were frozen. Chloride, bromide (Br–), and NO3– were determined by ion chromatography. Nitrite (NO2–) and NH4+ were determined colorimetrically on an AutoAnalyzer (Bran & Luebbe, Germany).
Nitrous oxide (N2O) was determined from 4 mL of water equilibrated with helium in a 10-mL serum bottle. Headspace samples were withdrawn and N2O was analyzed on a 63Ni electron capture gas chromatograph equipped with a Poropak R column. The oven and detector temperatures were 60°C and 300°C, respectively.
Piezometric Water Levels
The piezometric water levels of seven wells surrounding the alluvial deposit were measured with a chalked steel measuring tape on six dates between 1996 and 1998. The top of the well casings were sited with a transit level. For each survey, the water levels were normalized to well SHW 12 (assigned 0 m) located near the transition between hillslope and alluvial deposits. Contours of the water level surface for the six dates were combined to produce a general water table map at the intersection of the two well transects for the study period. The data were fitted with a mathematical function (radial basis, thin-plate spline) to the nearest input points (Surfer 8, Golden Software).
Experimental Ground Water Injections
For the experimental ground water injections, 25 L of water was pumped from the injection well over 1 h, amended with solutes, and returned over 4 to 7 h (Table 2). Chloride and Br– were alternated as conservative tracers, allowing a second injection before the vestiges of the first injection cleared the well field.
|
|
RESULTS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
4.1 mg N L–1, n = 2). On the other sampling dates, it was nearly constant at 2.8 mg N L–1 (Table 3). Nitrate in SHW 1 exceeded NO3– in precipitation by as much as one order of magnitude (Table 1) and NO3– in deep ground water by two to three orders of magnitude (Wells F-94 and F-197, Table 3). Nitrate decreased between SHW1 and SHW 13, the next closest well 57 m down slope. Nitrate levels at SHW 13 averaged 0.7 mg N L–1, a mean decrease of approximately 70%. The conservative anion Cl– also decreased approximately 30% between the wells from a mean of 4.1 mg L–1 to 2.6 mg L–1 (Table 3). The mean NO3/Cl ratio decreased from 667 to 286 µg mg–1 indicating NO3– disappearance relative to Cl– (Table 3). At SHW 12.5, an additional 9 m down slope (66 m from SHW 1), both mean NO3– and Cl– levels were further reduced. The mean NO3/Cl ratio decreased to 105 µg mg–1, comparable to SHW12, 8 m closer to the stream.
|
Water temperatures at SHW 1 and 13 exhibited minimal seasonal variation (Table 4) and were consistent with ground water temperatures in the Shingobee River Headwaters Area (D.O. Rosenberry, personal communication, 2005). In contrast, alluvial Wells SHW 10 and 11 exhibited seasonal variations that were consistent with seasonal stream water variations. Stream water temperature was 0.7°C in January 1998 and 18.0°C in September 1998. Winter and summer temperatures at well SHW 12 were between stream water and ground water.
|
The mean NO3/Cl ratios at SHW 16 and 17 were 504 and 640 µg mg–1, similar to SHW 1, although mean concentrations of both ions were 70 to 80% lower than in SHW 1. Mean NO3/Cl ratio at SHW 18 was only 3 µg mg–1. In contrast, the NO3/Cl ratio was 269 µg mg–1 at SHW 19 although both SHW 18 and 19 were highly oxygenated and close to each other. At SHW 20 and 21 on the alluvial terrace, NO3/Cl ratios were 18 and 121 µg mg–1. Seasonal variation in the NO3/Cl ratio at SHW 20 and 21 were similar to stream water (high in winter, low in summer, data not presented).
Ammonium and SRP were typically <10 µg L–1 in alluvial and bankside wells along Transect 2 (Table 3). Bankside wells downstream of the alluvial deposit were always highly aerobic. Wells on the alluvial terrace had lower mean DO, commonly around 1.0 mg L–1 during summer (Table 3).
Temperature in bankside wells downstream of the relict beaver dam varied little seasonally and tended to reflect the hillslope wells (Table 4). Temperature at wells on the alluvial terrace (SHW 20 and 21) were similar to stream water and varied seasonally.
Piezometric Water Levels in Transects 1 and 2
The piezometric surface of the water table in Transect 1 dropped 4.65 m between SHW 1 and SHW 13 (data not shown). The slope was relatively steep (0.08 m m–1) but not uncommon in the Shingobee Lake area because the river and lake are deeply incised relative to the land surface and the sand is fine to medium grade (D.O. Rosenberry, personal communication, 2005). The piezometric surface of the water table in the wells surrounding the alluvial terrace was flat (0.0004 m m–1) compared with the hillslope but indicated, in general, that hillslope water from SHW 13, up-gradient water from SHW 21, and lateral stream water from SHW 10 converged between SHW 11 and SHW 12, located near the transition between hillslope and alluvial deposits (Fig. 2). The convergence of the piezometric water surface varied on individual dates the measurements were made but the general trends were the same.
|
|
|
|
45 µg mg–1) at the beginning of Injection 2, which included glucose, NO3–, and Cl–. As the tracer plume passed each well, the NO3/Cl ratio decreased (Fig. 4b). The addition of glucose in Injection 2 also lowered DO in all wells (Fig. 4c). The greatest decline was observed in Well 3. The onset of the decrease in the NO3/Cl ratios coincided with the DO minima in Wells 1 and 2 but occurred before the minimum in Well 3. Declines in DO were not observed in the absence of glucose in any injection. With C2H2 added to the injectate solution during Injection 2, NO2– and N2O formation accompanied NO3– depletion in all wells as indicated by a decrease in the NO3/Cl ratio and simultaneous increase in the NO2–N/Cl– and N2O-N/Cl– ratios (Fig. 5). In Well 1, the decline in the NO3/Cl ratio occurred near the tail of the tracer cloud approximately 104 h after it first arrived. In Well 2, the steepest decline in the NO3/Cl ratio occurred shortly after the peak tracer concentration passed, approximately 96 h after it first arrived. In Well 3, the decline in the NO3/Cl ratio occurred shortly after the tracer arrived.
|
|
|
|
DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Ground water NO3– concentrations beneath the ridge of the hillslope (SHW 1) exceeded all down slope locations, deep ground water (F-94 and F-197), and precipitation. Between 1992 and 1998, NO3– ranged from 1.9 to 5.7 mg N L–1 in 20 samples from SHW 1. Nitrate levels above 3 mg N L–1 in the north central forested area of Minnesota indicate possible human input (Madison and Brunett, 1984) and probably reflect recent anthropogenic land use near the site such as grazing. But even the highest NO3– concentration in SHW 1, 5.7 mg N L–1 in June 1994, is low relative to ground water impacted by fertilized row crop agriculture in the state (Fong, 2000). During the period of our experimental studies, September 1996 through September 1998, NO3– concentrations at SHW 1 were seasonally uniform (2.2 mg L–1) and DO consistently high (7.2 mg L–1) indicating an oxidizing environment at the head of the hillslope flow path. Similar redox conditions at the upland-riparian boundary are described in several studies (Cooper, 1990; Jordan et al., 1993; Haycock and Pinay, 1993; Cey et al., 1999) including a forest riparian zone along the Boyne River in southern Ontario that received high NO3– (10 to 30 mg N L–1) from a sand aquifer (Hill et al., 2000).
Local ground water along the hillslope transect (Transect 1) decreased in Cl– and NO3– as it moved toward the channel. Decreases in Cl– probably resulted from rainwater dilution. Rainwater Cl– was an order of magnitude lower than local ground water. Deep ground water mixing (‘old’ ground water) with shallow ground water was unlikely even though Cl– concentrations of deep ground water were low enough to dilute the Cl– (Table 1). The surficial deposits along the Shingobee River at the study site are fine-grained sand and silts with a thin till unit underlying the deposits that passes beneath the Shingobee River (Winter et al., 2001, 2003). While the surficial sediments are permeable to rainwater, the low permeability till unit probably causes much of the older ground water to pass beneath the study area and discharge west of the river (Winter et al., 2001). The young ground water recharge dates in SHW 1, 11, and 13 (Triska et al., 2002) support this hypothesis.
Decreases in NO3– along Transect 1 were far greater than Cl– dilution suggesting that biotic processes accounted for a portion of the NO3– decrease. In these hillslope soils, the predominant biotic mechanisms for NO3– removal are plant uptake and denitrification. The depth to the water table is about 7 m at the ridge and decreases to 1.5 m near the base of the hillslope (SHW 12). Plant uptake is likely secondary to denitrification because of the depth to water (Peterjohn and Correll, 1984; Lowrance, 1992). Contact with tree roots would occur only where riparian ground water is forced close to the surface soils by the underlying till layer at the base of the hillslope, which would be similar to observations by Jordan et al. (1993) in a study on nutrient interception by a riparian forest receiving inputs from adjacent cropland in a mid-Atlantic coastal plain stream.
Dissolved oxygen is significantly lower (1.3 to 2.9 mg L–1) on the alluvial bench, suggesting denitrification may dominate NO3– removal in near-stream pore water. Hill et al. (2000) suggested that denitrification in the Boyne River riparian area occurred along a horizontal flow path where ground water DO was less 3 mg L–1. Cey et al. (1999) also found that NO3– concentrations dropped sharply when DO values fell below 2 mg L–1 in ground water upland of the Kintore Creek riparian area in Ontario, Canada, suggesting the NO3– decrease was due to denitrification. Ground water O2 measurements are made on relatively large volumes of water easily withdrawn by pumping so they reflect average O2 conditions in the aquifer. It is likely that O2 concentrations in isolated microsites (Usui et al., 2001) and "hotspots" in the riparian area (Hill et al., 2000; McClain et al., 2003) where denitrification occurs are lower than the average readings explaining denitrification in aquifers with 2 to 3 mg O2 L–1. If peaty soils or buried channel deposits were a source of C in the near-stream aquifer, we would expect lower DO levels. Presumably, organic matter derived from litter and root exudates transported to the water table supply the energy source for respiration and NO3– reduction by ground water bacteria on the alluvial bench (Pinay et al., 1993).
The wells in Transect 2 revealed a complex hydrogeologic environment along the stream bank. Based on Cl–, temperature, NO3–, and DO, it was possible to separate the bankside wells into two distinct ground water environments: alluvial wells upstream of the relict beaver dam influenced by stream water (SHW 10, 11, 12, 20, 21) and bankside wells downstream of the dam influenced by hillslope ground water (SHW 12, 16, 17, 18, 19).
Upstream alluvial wells were lower in Cl– and NO3– than the downstream bankside wells and temperature fluctuated seasonally from 0.8 to 18°C, similar to stream water. In addition, the piezometric surface of the water table generally indicated that alluvial water, consisting largely of stream water, flowed through the meander bend and mixed with local ground water below the hillslope gradient break. The temperature fluctuation in well SHW 12 at the gradient break appears to be an example of mixing between local ground water and stream water. In winter, it is cooler than ground water and warmer than stream water. Conversely, in summer it is warmer than ground water and cooler than stream water.
Bankside wells downstream of the relict beaver dam typically exceeded stream water Cl– levels indicating greater ground water than surface water influence. Chloride concentration in these wells generally exceeded 2.0 mg L–1, whereas stream water was approximately 0.85 mg L–1. One exception, SHW 16, had a mean Cl– concentration of 1.2 mg L–1. Additionally, the water temperatures of these bankside wells, including SHW 16, were relatively constant seasonally and similar to water in the hillslope wells, which suggests a strong ground water influence. Water temperatures for SHW 1, 13, 16, 17, 18, and 19 when averaged were 8.4°C in January and 9.0°C in September compared with the seasonal temperature fluctuation in stream water of 0.8 to 18°C. Finally, DO levels were high and comparable between these bankside wells and hillslope wells, and DO in both environments was substantially higher than in the alluvial wells.
The physical and chemical disparity between the alluvial and downstream bankside wells and the similarity between the hillslope and downstream bankside wells lead us to conclude that the downstream bankside wells consisted of hillslope ground water flowing toward the channel. As termini of hillslope ground water flow paths, it is possible to determine the fates of NO3– in local ground water originating near or beyond SHW 1. Shifts in NO3/Cl– ratios between the top and bottom of the potential flow paths between SHW1 and SHW 12, 16, 17, 18, and 19 can be used to identify dilution by rain water and biotic NO3– retention. Well SHW 12 had a NO3/Cl– ratio of 104, approximately 15% of the NO3/Cl– ratio in SHW 1. Well SHW 16 had a NO3/Cl– ratio of 504, approximately 75% of the NO3/Cl– ratio in SHW 1. Well SHW 17 had a NO3/Cl– ratio of 640, similar to the ratio in SHW 1. Well SHW 18 had a NO3/Cl– ratio of 3, approximately 1% of SHW 1, and well SHW 19 had a NO3/Cl– ratio of 269, approximately 40% of SHW 1. Based on these ratios, biotic NO3– retention accounted for 25 to 95% of the NO3– loss in four of the potential flow paths, with dilution being solely responsible for NO3– loss in the flow path to SHW 17.
Numerous ground water studies of agricultural runoff have concluded that NO3– processing by denitrification is dependent on (i) routing of ground water through riparian areas to the discharge zones (Devito et al., 2000), (ii) interception with organic-rich deposits and soils (Hill, 1990), and (iii) passage through areas of steep redox changes defined by electron donors and acceptors (Hedin et al., 1998; Böhlke et al., 2002). If the denitrification capacity is fully utilized, large amounts of NO3– can be removed as shown in many studies by decreased NO3– concentrations between upland and riparian ground water. While NO3– losses were observed along the ground water flow paths in this study, the NO3– levels were low relative to ground water impacted by fertilized row-crop agriculture. In addition, the presence of DO in the wells and the variability in denitrification potentials along the flow paths suggest that DOC was low and patchy. If the rate of NO3– loading in ground water recharge were to increase dramatically along these flow paths, assuming the denitrifying capacity of the riparian soils is at an optimum, then NO3––rich ground water would likely pass through the soils unaltered. This result is contrary to NO3– retention studies where NO3– buffer strips are viewed as NO3– retention zones irrespective of loading rates if loading is routed through shallow riparian flow paths (Haycock and Burt, 1993; Haycock and Pinay, 1993; Jacobs and Gilliam, 1985).
Nitrate Transformations in the Experimental Well Array
Dissolved oxygen and BDOC were principal factors controlling NO3– retention in ground water of the experimental well array. Nitrate and Br– were the only solutes added to Injection 1 in January 1998, and DO levels remained high throughout the experiment (>7 mg L–1). With high DO and low winter pore water temperatures, the NO3/Br ratio was constant during the injection indicating minimal NO3– retention. When glucose was added with NO3– and Cl– to Injection 2 in January, both DO and the NO3/Cl ratio decreased during passage of the tracer. Nitrate retention in Well 1 began after DO reached its minimum of 2.2 mg L–1 (20% saturation), suggesting DO concentrations above 3 mg L–1 (25% saturation) were the primary control on NO3– retention in January. Glucose stimulated NO3– retention first by stimulating heterotrophic metabolism causing DO to decrease and second by providing C and energy for NO3––reducing bacteria (Groffman, 1994; Hill et al., 2000).
Background DOC measured in October 1997 ranged from 1.8 to 2.9 mg L–1 in wells SHW 13, 17, 19, and 21 compared with 5.2 mg L–1 in the stream. In September 1998, DOC was 1.5 and 2.5 mg L–1 in experimental Wells 1 and 3. Thus it is reasonable to assume background DOC was between 1.5 and 2.9 mg C L–1 in the experimental well array in January. Since ground water velocities were slower in winter than summer maximizing reaction time, there is reason to believe that the DOC was not sufficient in quantity or quality to support denitrification at the high NO3– levels of Injection 1.
In September, average background DO was 2.5 mg L–1 (26% saturated) in the well array compared with 6.6 mg L–1 in January (54% saturated), suggesting late summer BDOC was sufficient to lower background DO in the wells. Nonetheless, natural NO3– levels began to decrease in Injection 3 only after added glucose caused a drop in DO from 2.5 to 0.65 mg L–1 (7% saturation). As the glucose plume passed, O2 and NO3– gradually returned to pre-injection levels. Even though DO in ground water was near or below a threshold of 2 to 3 mg L–1 (25% saturation) where NO3– reduction has been shown to occur, reduction of natural NO3– levels did not begin without adding glucose. These results suggest that NO3– reduction was limited by BDOC in September. Therefore, we conclude that BDOC at this site would be inadequate to completely reduce elevated NO3– concentrations present in agriculturally impacted ground water.
Temperature also controlled NO3– retention in ground water. Injections 3 and 4 were performed in September 1998, when alluvial ground water temperatures were approximately 18.0°C compared with 5.5°C in January. In September, the peak NO3– concentration at Well 1 was 1.7 times higher than in January (27 vs. 16 mg N L–1) and passed through >48 h faster. But the NO3– mass as a percentage of the Cl– mass was 20% lower in September than January indicating a higher proportion of biological processing in September. Furthermore, in September, virtually all added NO3– was depleted before reaching Well 2 whereas in January greater than 75% of the added NO3– mass persisted into Wells 2 and 3. In numerical modeling of seasonal denitrification rates in sediment perfusion cores from the streambed adjacent to this site, Sheibley et al. (2003) reported low winter denitrification rates relative to summer. Because denitrification consists of sequential temperature-dependent enzyme reactions, rates would be faster in September.
Ground water velocity was not a significant factor controlling NO3– retention in the experimental well array. Nitrate retention occurred in both seasons, but the rates were higher in September in spite of faster travel times to the wells reducing reaction time between NO3– and attached biota. Presumably, physical-chemical controls such as DO and temperature have a larger impact on NO3– retention than velocity at the observed rates. However, it is important to note that in ground water environments, subsurface microbial communities largely depend on water exchange to supply nutrients and carry away metabolic products.
When glucose was added with the tracer in Injections 2, 3, and 4, DO decreased shortly after the tracer arrived, demonstrating the microbial community's ability to rapidly increase metabolism and alter redox following the appearance of BDOC. As added NO3– decreased in January and September, NO2– and N2O appeared in a classic denitrification sequence. In September, NO3– was completely reduced to N2O in the presence of C2H2. Duff and Triska (1990) also found NO3– was reduced to N2O in a NO3–/C2H2 injection along a shallow ground water flow path adjacent to a northern California stream. Hill et al. (2000) used in situ C2H2 injections to locate sites of denitrifying activity in their study of the Boyne River subsurface riparian denitrification. Their injections to piezometers were different from our flow path studies in that samples of fresh ground water were withdrawn over 7-d intervals from the same piezometers that were used for the injections. Their results revealed that significant denitrification was restricted to a narrow zone of steep NO3– and N2O decline in the agricultural plume margin located near interfaces of sand and either peat or buried river-channel deposits. Though denitrification completely reduced a large quantity of added NO3– along our relatively short flow path (196, 296, and 161 mg of NO3––N in Wells 1, 2, and 3, respectively), the steep zone of NO3– reduction was the result of adding glucose and not naturally occurring.
The long-term trends in physical and chemical properties of ground water near the Shingobee River and results from the ground water injections in the experimental well array reveal a potential scenario whereby alluvial ground water originating in the stream channel interacts with local ground water to transform NO3– moving toward the channel. Along the meander of the river at this alluvial deposit, stream water BDOC and O2 are gradually depleted during transport through the alluvial flow path adjacent to the channel. Where it mixes with local ground water high in DO and NO3–, the BDOC eventually reduces O2 of the hillslope ground water and NO3– becomes the terminal oxidant.
|
CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Nonetheless, hydrologic and biotic processes occurring in the riparian zone decreased background NO3– concentrations along hillslope flow paths by dilution and denitrification. The presence of DO in the wells and variability in denitrification potentials along the flow paths suggest that DOC was low and patchy. Presumably, organic matter derived from litter and root exudates transported to the water table supplied the electron donor for respiration and NO3– reduction by ground water bacteria.
Dissolved oxygen, BDOC, and temperature were the factors controlling nitrate retention along the experimental flow path. Dissolved oxygen levels, regulated by BDOC, appeared to be the principal controlling factor. Nitrate was completely reduced to N2O in the presence of glucose, C2H2, and elevated NO3– concentrations simulating agricultural NO3– loading. This result indicated the importance of organic C for establishing a redox state suitable for denitrification. Though denitrification completely reduced a large quantity of added NO3– along the relatively short flow path in September, the steep zone of NO3– reduction was controlled by the added organic C and was not naturally occurring.
Under current land use along this reach of the Shingobee River, most NO3– associated with local ground water is retained (denitrified) or diluted before reaching the channel. However, the combined monitoring and experimental approaches suggest that under more intensive fertilized row-crop agriculture, biotic processing and physical dilution would be overwhelmed without added organic C or a change in the natural redox state of the riparian zone and excess NO3– would be transported to the channel degrading river water quality.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | |||
