| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Illinois State Geological Survey, 615 E. Peabody Dr., Champaign, IL 61820
b Dep. of Geology, Univ. of Illinois at Urbana-Champaign, 1301 West Green St., Urbana, IL 61801
c Illinois State Water Survey, 2204 Griffith Dr., Champaign, IL 61820
* Corresponding author (mehnert{at}isgs.uiuc.edu)
Received for publication March 8, 2006.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: DEA, denitrification enzyme activity • MPN, most probable number • org-N, organic N • TKN, total Kjeldahl nitrogen • TOC, total organic carbon • DO, dissolved oxygen
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Nutrients such as N can enter streams by soil erosion, storm runoff, discharge from drainage tiles, and ground water discharge. The N load contributed to streams by ground water is variable. Ground water discharge to streams (base flow) contributed 66% of the mean annual nitrate exported to the Raccoon River in Iowa (Schilling and Zhang, 2004), but contributed negligible N in Iowa's Walnut Creek watershed (Schilling and Wolter, 2001).
In some watersheds, N leaches through the root zone and below tile drains (Cambardella et al., 1999; Mehnert et al., 2005). Böhlke (2002) noted that 10 to 50% of applied N commonly leaches through well-drained agricultural fields into shallow ground water. Puckett et al. (1999) found similar N flux to shallow ground water while conducting a mass balance of N in a Minnesota watershed with coarse-grained soils and geologic materials. They determined that crop harvest (67.6% of total N source), net flux to ground water (11.1%), and denitrification (8.8%) were the three largest sinks for all N sources in this watershed. Puckett et al. (2002) confirmed that the N transported to ground water discharged to the Otter Tail River, bypassing riparian buffer strips. The hydrogeologic factors that negate the effectiveness of riparian zone denitrification include (i) denitrification in the upgradient aquifer due to the presence of organic carbon or other electron donors, (ii) long residence times (>50 yr) along the ground water flow paths allowing slow reactions to completely remove nitrate, (iii) dilution of nitrate-rich water with older, nitrate-poor water, (iv) bypassing the riparian zones due to extensive use of drains and ditches, and (v) movement of ground water along deep flow paths below reducing zones (Puckett, 2004). These factors also describe the factors that control denitrification in the shallow ground water of a watershed.
Besides shallow ground water, denitrification occurs in other portions of a watershed. Denitrification has been documented in surficial soils (Mosier et al., 2002), hyporheic zones, which includes in-stream plants and benthic sediments (Peterson et al., 2001), and riparian zones (Hill et al., 2004; Kellogg et al., 2005). While other researchers considered denitrification in the hyporheic zone, our purpose was to estimate denitrification in the shallow ground water of a tile-drained, Corn Belt watershed having fine-grained soils and geologic materials. We characterized the ground water geochemistry and conducted microcosm and field push-pull experiments to estimate the potential for denitrification in the subsurface. We measured stable isotope ratios to determine whether denitrification was occurring in the system. We developed a ground water model of the watershed, which we combined with mass-balance calculations to estimate the extent of denitrification in the shallow ground water of the watershed.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
Eleven monitoring wells were installed throughout the watershed to monitor the position of the water table and to collect shallow ground water samples (Fig. 1). The wells were positioned to provide water level data along the length of the stream (Wells 1, 4, 7, 9, and 11) and two transects perpendicular to the stream (Wells 4, 3, and 2 and Wells 7, 6, and 5). The wells were constructed with 5.0-cm diam. PVC, screened at approximately 3.8 to 2.3 m below ground surface, and sand packed from 3.8 to 0.76 m below ground surface. Above 0.76 m, each well annulus was sealed with bentonite and concrete. The long sand packs were intended to allow us to sample water from the water table throughout the year. Water levels in the monitoring wells and surface water levels at selected locations were monitored on a weekly or biweekly schedule. Water levels were determined manually using a water level meter (Heron Instruments, Burlington, ON). These measurements were recorded to the nearest 0.30 cm. The elevations of all water-level measuring points were determined by level surveying, based on the known elevations of five benchmarks throughout the watershed. Level surveying was conducted with an automatic level and a micrometer (Wild NA2 & GPM3 µm, Leica Geosystems AG, Heerbrugg, Switzerland). The micrometer allowed elevations to be measured to the fourth decimal place (0.0001 m).
To determine hydraulic conductivity of the geologic materials, slug tests were conducted in all monitoring wells following the procedures described by Butler (1998). The slug test data were analyzed to estimate hydraulic conductivity using AquiferWin32, version 2 (Environmental Simulations International, 2000).
Other researchers provided data on surface water flow and quality (Keefer and Bauer, 2005) and tile water quality (Mirek, 2001). Surface water was sampled only at the gaging station. Eleven tile drains in the watersheds of Big Ditch and Wildcat Slough were sampled. These tiles drained a single field and were not regional drainage lines. In addition, the sampled tile drain locations were confidential per an agreement with landowners.
Water Sample Collection and Analyses
For each well, ground water was pumped through a flow cell (Hydrolab, Austin, TX) while monitoring dissolved oxygen (DO), pH, oxidation–reduction potential, conductivity, and temperature (Barcelona et al., 1985). After the readings stabilized, the ground water was pumped through a 0.45-µm filter (Groundwater Capsule, Gelman, Ann Arbor, MI). Separate samples were collected for nitrate (NO3–), ammonium (NH4+–N), total Kjeldahl nitrogen (TKN), Fe, and total organic carbon (TOC). Samples for TKN/NH4+–N, Fe, and TOC were preserved with 0.2% H2SO4. Nitrate samples were not acidified because the sulfate would have overwhelmed the ion chromatograph detector. All samples were stored on ice for transportation to the laboratory. The NO3––N samples were either analyzed the same day they were collected or stored overnight at 4°C and analyzed the next day. All other samples were stored at 4°C until they were analyzed.
Water samples were collected from May 2000 through June 2003. The wells were sampled about 20 times during that period, approximately once every 3 mo and more often around planting season (May and June). Some wells were sampled as few as 10 times during the project (low water levels prevented sampling), but other wells were sampled up to 26 times.
Nitrate was determined by anion chromatography using USEPA Method 300.0, rev. 2.11 (USEPA, 1993). Ammonium was determined by automated colorimetry using USEPA Method 350.1, rev. 2.0 (USEPA, 1993). Total Kjeldahl nitrogen was determined by NaOH digestion and automated colorimetry using USEPA Method 351.2, rev. 2.0 (USEPA, 1993). Total N concentrations were defined as the sum of the NO3––N and TKN concentrations. Total organic carbon was determined using persulfate-ultraviolet oxidation and infrared spectrometry using USEPA Method 415.2 (USEPA, 1983). Ferrous iron was determined colorimetrically (Clesceri et al., 1998). All samples were analyzed within the holding times specified by the methods.
Natural variations in the isotope ratios of nitrogen (15N/14N) and oxygen (18O/16O) in the nitrate ion were used as indicators of the extent of denitrification. Isotopic fractionation associated with microbial denitrification provides a method to quantify NO3– losses due to denitrification independently of dilution and advection effects on NO3– concentrations. Denitrification of 14N-bearing NO3– proceeds at a rate slightly greater than that of 15N-bearing NO3–, and this leads to the enrichment of 15N-bearing NO3– in the residual unreacted fraction (Mariotti et al., 1981). Similarly, the residual NO3– is enriched in 18O relative to 16O as denitrification proceeds.
All isotope ratio analyses are reported in delta notation to facilitate analysis and display of the very small variations in the ratios observed and the very high precision typical for these analyses. The 
18O value represents variation of measured 18O/16O ratios in parts per thousand (
or per mil), with the zero point on the 18O scale set equal to the 18O/16O ratio of the seawater reference standard, standard mean ocean water (SMOW). The mathematical definition is:
 | [1] |
15N is the same equation as 18O. The reference standard for 15N is air nitrogen. Isotopic analyses of N and O were performed using published methods (Silva et al., 1994, 2000; Wassenaar, 1995) with some modification (Hwang et al., 1999; Panno et al., 2001).
In general, increases in 18O and 15N are indicative of denitrification. The Rayleigh fractionation model relates the increase in 18O or 15N to the amount of denitrification. A close approximation to the Rayleigh equation is given by
 | [2] |
gives the isotopic composition measured for a ground water sample, 0 gives the nitrate before it has undergone any denitrification, f gives the fraction of nitrate remaining as denitrification proceeds, and 
expresses the strength of the isotopic fractionation and is given by Eq. [3]:
 | [3] |
Microbiological Studies
Samples of geologic materials were collected from various depths at each monitoring well site to characterize the spatial and vertical variability of populations of denitrifying bacteria. An initial set of sediment samples was collected in April 2000 as boreholes for the monitoring wells were drilled using a hollow stem auger drill rig. In September 2000, a second set of samples was collected using a direct push rig at six sites to seek a more detailed vertical profile. To prevent contamination, sediment samples were collected from the center of the core using aseptic technique. Most probable number (MPN) analyses for denitrifying bacteria were started within 1 wk of sample collection. The sediment samples were stored at 4°C for up to 3 mo until all analyses were completed although the more time sensitive assays were completed as quickly as possible.
The sediment and water samples were enumerated using a modified MPN method for denitrifying microorganisms (Tiedje, 1994). For three-tube MPN determinations, R2A broth medium was used with 0.5 mM nitrate. The samples were incubated for 2 wk, then the number of positive and negative tubes were counted for each sample. Positive tubes were determined by using a diphenylamine indicator test, which detects residual nitrate or nitrite (Tiedje, 1994). Plate counts on R2A agar plates with 0.5 mM nitrate were also used for selected samples from Wells 1, 7, and 9 to enumerate the number of nitrate-reducing microorganisms. The plates were inoculated with the same dilutions as the MPN tests, incubated for 2 wk in an anaerobic chamber (T 
22°C), and the number of colony forming units were then counted.
Samples of geologic materials for denitrification experiments were collected at the same time as samples for enumeration. Denitrification rates in the sediments were measured using a modification of the standard soil denitrification enzyme activity (DEA) assay (Tiedje, 1994). In this assay, 160-mL serum bottles were filled with 10 g of sediment and phosphate buffer for a total volume of 100 mL. The 60-mL headspace in these bottles was filled with 90% nitrogen gas and 10% acetylene (added as an inhibitor of N2O reduction). Samples were incubated at 25°C. The starting nitrate concentration was 2 mM (28 mg L–1). To determine the effect of available carbon, the DEA was determined by measuring the N2O production rate from sediment samples with four different carbon amendments: no carbon, 0.2 mM (2.4 mg L–1), 2 mM, 8 mM, and 20 mM. The carbon sources, acetate and glucose, were added in equimolar amounts. Chloramphenical, an antibiotic typically used to inhibit new growth in DEA analysis, was not used because it has been shown to underestimate the actual denitrification rates (Pell et al., 1996) and because N2O production was undetectable when it was used in this study. Instead of the usual 2-h DEA test used for surface soils, headspace gas samples were taken at different intervals depending on the rate of N2O production and the carbon amendments applied: 3 to 4 d (2, 8, and 20 mM C), 1 wk (0.2 mM C), and 8 wk (no carbon). This was necessary because the number of bacteria per gram of sediment is much lower than found in surface soils. The rate was determined using the maximum slope of a linear fit line on a graph of N2O produced vs. time for each sample.
Push-pull tests (Istok et al., 1997) were conducted to determine in situ nitrate reduction rates in shallow ground water (McDonald, 2003). The push-pull test involves the injection and extraction from the same well of a test solution containing both a reactive and a conservative tracer. The test solution used in these experiments consisted of 15 L of local ground water spiked with approximately 10 mg L–1 NO3––N, 100 mg L–1 Br–, and 8 mg L–1 (0.135 mM) acetate (Na-salt) as a carbon source. This acetate amendment is equivalent to an organic carbon concentration of 3.2 mg L–1, which is near the average concentration found in the shallow aquifer (0.5–1 mg L–1). Therefore, the rates estimated using this method were likely to reflect in situ conditions. A chaser solution (5 L of distilled, deionized water) was used to push the test solution out of the well casing and into the subsurface. A resting phase of 0.57 h between the injection and extraction phases was found to be optimal, although resting phases of up to 22 h were attempted. No pumping occurred in the resting phase to allow time for denitrification to occur.
Estimation of Denitrification via Mass Balance
Estimation of denitrification within the shallow ground water of the watershed requires data on ground water flow and N concentrations. We modeled the steady-state ground water flow. Transient flow was not included in our model because accurately modeling transient flow through tiles was too complex to model. Thus, any denitrification in transient flow was ignored in this preliminary estimation. A steady-state model of shallow ground water and surface water flow was developed to define the hydraulic characteristics of the watershed (e.g., hydraulic conductivity and recharge) and to quantify ground water flow in the watershed during base-flow conditions (typically during late summer and early fall). During base-flow conditions, only ground water is discharging to the stream. Tiles are not discharging to the stream during base-flow conditions because the water table is below the tiles. Surface water flow within the Big Ditch watershed was measured at 156 L s–1 on 17 Nov. 2001 (Mehnert et al., 2005). No tile flow was observed on this date. This value of stream flow and the corresponding water levels in 11 monitoring wells and surface water sites were used to manually calibrate a two-dimensional ground water model. Calibration with both surface water flow and ground water levels allows one to uniquely estimate recharge (Sanford, 2002). This model was developed using the analytical element modeling software (Haitjema, 1995), GFLOW (Haitjema Software, 2001). The GFLOW software allows one to model surface water flow and ground water flow without assigning boundary conditions.
Denitrification in the shallow ground water was estimated with a simple approach. We assumed that high nitrate N concentrations that enter shallow ground water are reduced to zero by denitrification, before ground water discharges to the stream. This conceptual model is based on two widely known facts. First, N concentrations in soil water and tile drainage are generally high (e.g, Cambardella et al., 1999). Second, during base-flow conditions (generally late summer and early fall), ground water is the sole source of water in the stream, and N concentrations in the stream are very low. During base-flow conditions, nitrate N concentrations in Big Ditch were <0.06 mg L–1 even though the stream base flow varied from year to year (6.2 to 3560 L s–1). Similar seasonal variation in N concentrations has been observed within the Mississippi River Basin (Fenelon and Moore, 1998; Goolsby et al., 2001). For the case of steady-state ground water flow, the amount of nitrate N entering shallow ground water and the amount denitrified were assumed to be equal and were estimated using the following relationship:
 | [4] |
N is the difference in N concentration at the beginning and end of the flow paths. Ground water flow in this watershed can be conceptualized as a set of nonoverlapping flow paths with different lengths and depths that arc beneath the ground surface. The beginning of the flow paths is the water table, just below the tiles. The end of the flow paths is the stream. We used the average concentration of N in water from tile drains to compute N. The N concentration was assumed to be zero at the end of the flow paths. Ground water recharge was determined by steady-state flow modeling using uniform geologic and hydrogeologic conditions to describe the watershed. |
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
|
Geochemistry
The NO3––N concentration of the stream water at the outlet was quite variable throughout the year (Fig. 3), with high concentrations (up to 20 mg L–1) in late winter through spring and low concentrations in late summer and fall. High NO3––N concentrations were observed during high flows, while low concentrations were observed during low flows. During high-flow periods, organic N (org-N) and NO3––N concentrations were both high due to discharge from tile drains. During low-flow periods, NO3– was the predominant N species, accounting for >90% of total N in more than 80% of the samples (Keefer and Bauer, 2005). The potential for in-stream and benthic denitrification to remove N from the stream was evaluated by other researchers. Based on field data, the hyporheic zone of the Big Ditch watershed was not an effective sink for N in 2001 (Royer et al., 2004) or in 2002 (Schaller et al., 2004).
Eleven tile drains in the Big Ditch and an adjacent watershed were monitored during water years (October through September) 1998, 1999, and 2000. For these 3 yr, the flow-averaged total N concentration of water discharging from tile drains was 16.6 mg L–1 for fields planted to corn and 12.3 mg L–1 for fields planted to soybeans (Mirek, 2001). Approximately 60% of these tile drain samples had total N concentrations that exceeded 10 mg L–1 (Mirek, 2001). For water years 2001 and 2002, a subset of these tile drains was monitored. The flow-averaged total N concentrations were 8.7 and 17.9 mg L–1 for water years 2001 and 2002 (G. Collins and M. David, personal communication, 2004).
One or more N species were generally detected in ground water from all wells except from Well 2. Total N concentrations in ground water were generally <5 mg L–1, except for two wells (3 and 6). Nitrate was the most common N species, comprising over 90% of the total N in over 70% of the samples with at least 1 mg N L–1. Organic N and NH4+–N made up a significant fraction of total N in many samples with total N values <4 mg L–1 (Fig. 4). The ground water samples showed considerable temporal variability in total N (Fig. 5) and spatial variability in the N species concentrations. Nitrate was the predominant N species in Well 11 (Fig. 5) and five other wells (Wells 3, 5, 6, 7, and 8). Organic N and NH4–N were usually the predominant species in Well 9 and four other wells (Wells 1, 2, 4, and 10). The spatial and temporal variability in N speciation and concentration were probably caused by local differences in biodegradable organic matter, recharge, and hydraulic conductivity.
|
|
 | [5] |
Redox conditions varied across the watershed but were favorable for denitrification at some sites. Denitrification occurs only under anoxic conditions or with restricted oxygen availability when electron donors are available (Korom, 1992; Starr and Gillham, 1993). The median DO value was less than 2 mg L–1 in five wells. In three wells, some DO values were less than 2 mg L–1 (Fig. 6). Several sets of ground water samples were analyzed for Fe, which was usually detected when the DO value was less than 
1.5 mg L–1. Because O2 rapidly oxidizes soluble Fe2+ to insoluble iron oxide (Stumm and Morgan, 1996), DO and Fe are expected to be mutually exclusive in ground water. The apparent co-occurrence of O2 and Fe may have been the result of pumping a mixture of shallow oxic and deeper anoxic water. Denitrification is known to occur in anoxic microenvironments in the subsurface (Pedersen et al., 1991). Limited soil sampling from this watershed (Shiffer, 2001) indicated that anoxic microenvironments are, indeed, present.
|
Isotopic Geochemistry
The results of the isotopic analyses are summarized in Table 1 and Fig. 7 and presented in detail in Mehnert et al. (2005). The tile drain 15N and 18O values cluster closely around mean values of 5.2 and 6.1, respectively. The few exceptions (e.g., there are only three 15N values greater than 7.4) are elevated values probably caused by soil-zone denitrification. We assume the tile drain 15N and 18O values provide a good estimate of the nitrate input to the ground water system.
|
|
18O and 15N relative to the tile drains, indicating little or no denitrification (Table 1). 18O and 15N values in Wells 3, 5, and 11 give strong evidence for the loss of large fractions of initially present nitrate via denitrification depending on the well and sampling time. The median DO for Well 11 was <2 mg L–1 but the median DO for Wells 3 and 5 was approximately 6 mg L–1(Fig. 6). These higher DO values for Wells 3 and 5 would not favor denitrification, but may indicate that denitrification is occurring in microenvironments or that DO varies temporally because the isotopic and geochemical samples were collected at different times.
In Fig. 7, a line representing a hypothetical trend for denitrification is given for the case where = –12.0 for N and = –5.7 for O2 (Mariotti et al., 1988). If these are the correct values of for this system, the isotope data imply roughly 30 to 70% loss of nitrate for most of the well samples. However, these estimates are semi-quantitative because is known to vary with geochemical conditions (Hübner, 1986; Mariotti et al., 1988; Böttcher et al., 1990), and the initial isotopic values, before denitrification, are not well constrained.
As an alternative, the elevated 15N values in ground water nitrate could be input from an isotopically heavy source of nitrate such as manure (Kendall, 1998). The latter scenario is untenable in this watershed, as N application is done almost exclusively with anhydrous ammonia and not with manure. No manure-intensive land use exists in the Big Ditch or neighboring watersheds. Additionally, this explanation does not account for the elevated 18O values, which are fully consistent with earlier work on denitrification (Böttcher et al., 1990).
Samples from seven monitoring wells (Wells 1, 2, 4, 7, 8, 9, and 10) consistently had low NO3––N (<1 mg L–1) and five of these wells (Wells 2, 4, 7, 9, and 10) often had low DO (generally <2 mg L–1). In light of the strong denitrification inferred from the isotopic data in three of the four wells with greater NO3––N concentrations, we assert that these seven wells had nearly complete denitrification. It is highly unlikely that the ground water in these wells was recharged with low NO3––N concentrations, as high NO3––N concentrations were found throughout the tile drains in the watershed (Mirek, 2001). It is more likely that the water at these sites was recharged at upland areas and denitrified as it flowed to the stream.
Microbial Studies
Denitrifiers were enumerated from cores collected at Wells 1, 3, 6, 7, 9, and 10. Most probable numbers for geologic samples, collected to depths of 10.7 m, ranged from 105 to 106 cells g–1 soil. Higher MPNs were observed in coarse-grained geologic materials than fine-grained geologic materials.
Denitrification rates were measured in the lab by modified DEA, without and with carbon amendments (Table 2). Assays done with carbon amendments were completed within 3 or 4 d after sample collection. However, the rates measured in samples without carbon required 8 wk to give significant N2O production. No increase in rates was observed over the course of any of the assays, which indicates little or no new growth of new denitrifying bacteria had occurred. The rates were variable for a given carbon amendment (a range is provided in Table 2) and increased as the carbon concentration increased. This result was expected because the availability of carbon or other electron donors commonly limits the activity and growth of denitrifying bacteria. The DEA results indicated that denitrifiers are present and respond quickly when appropriate nutrients are available. In the shallow geologic materials, these nutrients may be available on a cyclic basis depending on precipitation and subsequent recharge.
|
The potential for denitrification in the shallow ground water of the watershed was estimated for a time frame of 1 yr, assuming denitrification in the upper 4.6 m of sediments. A denitrification rate of 0.45 mg N kg–1 soil-day was selected based on adjusting DEA results to the lower average carbon content of the aquifer. Denitrification enzyme activity data were used instead of push-pull data since they were collected from all wells. However, as noted the push-pull data was in agreement with the DEA results. Assuming this constant denitrification rate, the equivalent of 84% of the N fertilizer applied in a given year (168 kg N ha–1 on half of the watershed) could be denitrified. The total nitrate removed may be less because the amount of nitrate in the ground water varies seasonally.
Estimated Denitrification in Watershed Ground Water
Low N concentrations (<4 mg L–1) were found in most wells. The microbiological studies indicate that the potential denitrification in the shallow geologic materials of this watershed is significant. The isotopic data indicate that up to 70% denitrification has occurred in three of four locations where it can be assessed in the shallow subsurface of the watershed (Table 1, Fig. 7). We inferred that at least partial denitrification occurred at other locations with low N concentrations in shallow ground water, as ground water flows slowly (years to decades) toward the first-order streams or Big Ditch. Although we did not study individual flow paths, it is reasonable to expect complete denitrification in the subsurface, as described by Puckett (2004).
Our observed water quality data could also be explained by the long residence time hypothesis. However, this long residence time hypothesis seems to be less credible than the complete denitrification hypothesis that we discuss below. For the long residence time hypothesis, one assumes that denitrification does not occur in the subsurface, that modern, N-rich water is recharging shallow ground water, and that only older, N-poor ground water is discharging to the stream during base flow. This ground water is old and was recharged before widespread N fertilization (likely the mid 1960s in this watershed), and thus would be N-poor ground water. Essentially, no N discharges to the stream because the older ground water dilutes the younger ground water. Mehnert et al. (2005) estimated that ground water discharging to the stream has a wide range of residence times, from months to many decades. Any ground water with residence times of months to 30 or 40 yr would have measurable N concentrations and thus would deliver N to the stream during base flow. Given the N concentrations in shallow ground water (8.7 mg L–1) and the stream (<0.06 mg L–1), the volume of older ground water would have to exceed the volume of younger ground water by a factor of 147 to sufficiently dilute N in the stream.
To estimate denitrification for the watershed using Eq. [4], we assumed that the N concentration at the beginning of the ground water flow paths was the flow-averaged mean N concentration from tile drains for water years 2000 through 2002. To allow for uncertainty in the input N concentration, the mean N concentration was altered by ± 50%. The resulting values are close to the median of the minimum and maximum monthly averages of 6.4 and 22.9 mg L–1 reported for the 11 tile drains for water years 1998 through 2000 (Mirek, 2001).
The mass-balance model has several limitations. First, as explained above, some NO3–N (<2.5 mg L–1) may be produced below the tiles by oxidation of soil org-N. Second, corn and soybean roots grow as deep as 2 m, which is deeper than the average tile depth of 1 m. Therefore, some NO3–N may be taken up by plant roots below the tiles. However, these two effects partially offset. The size of any of these effects may be less than the ± 50% range of assumed N concentrations in recharge water. Nevertheless, this first approximation of denitrification in the subsurface should be useful to examine the scope of denitrification in this environment.
Three values of recharge were used to develop a range of N flux from the watershed. Stream flow during base-flow conditions was evaluated to determine the first quartile, median, and third quartile of stream flow for three water years. These values are reported in Tables 3
through 5. For the median values of stream flow, steady-state ground water flow contributed 6.9% of the total stream flow for water year 2001 and 5.0% for water year 2002. Ground water recharge and ground water discharge to the stream are equivalent in a steady-state model. Thus, our model results indicate that steady-state ground water recharge is the source of 5.0 to 6.9% of stream flow.
|
|
|
through 5) varied with ground water recharge and the change in N. The best estimate for the mass of N denitrified in shallow ground water was equivalent to 2.3% of N applied for water year 2000, 1.2% of N applied for water year 2001, and 3.1% of N applied for water year 2002. Water years 2000 and 2001 were dry years, whereas water year 2002 was a normal, but relatively wet year. Higher N flux during a wetter year that follows a dry year has been noted by others because N is thought to be stored in the soil during the dry year, then flushed from the soil during the wet year (e.g., Goolsby et al., 2001). Over the course of the 3 yr monitored, the estimated mass of N denitrified in shallow ground water ranged from 0.3 to 6.4% of applied N, so these estimates vary considerably. The N exported from the watershed via surface water was estimated to be equivalent to 12.5% of applied N in 2000, 13.1% in 2001, and 11.3% in 2002 (Keefer and Bauer, 2005). Our estimates of N denitrified in shallow ground water represent 9 to 27% of the estimated N exported from the watershed via surface water.
These estimates should be considered minimum values because our simple model only accounts for steady-state ground water flow and ignores transient ground water flow. Thus, any additional N transported by transient ground water flow, such as flow from precipitation events (e.g., Cey et al., 1998), is not included in the estimates in Tables 3 through 5. However, some of our isotopic data show that transient ground water remains in the subsurface long enough to allow for partial and possibly complete denitrification.
Previous studies support the expectation that our values underestimate the amount of N denitrified in the shallow ground water. For common grain-production systems, nitrate leaching below the root zones was found to range from 10 to 30% of applied N (Meisinger and Delgado, 2002). For a coarse-grained, outwash aquifer in Minnesota, 49% of the nitrogen fertilizer applied was transported past the root zone and 22% of the nitrogen fertilizer applied was denitrified (Puckett et al., 1999).
Data from this study indicate that denitrification occurring in the shallow ground water of watersheds with fine-grained soils and geologic materials will vary with precipitation and may be a significant N sink. Drainage management has been found to significantly reduce N losses in tile drainage (Gilliam et al., 1979; Deal et al., 1986; Drury et al., 1996) and may offer a potential remedy to reduce the high N loads transported from some Midwestern watersheds. By reducing tile discharge, more N-laden ground water may flow through the shallow subsurface where denitrification can occur as demonstrated by our results.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Denitrifying bacteria were found at depths to 10 m throughout the study area. Microcosm experiments were performed to estimate denitrification rates. The maximum denitrification rates were 0.2 mg N (g soil)–1 d–1 for aquifer sediments only and 4.2 mg N (g soil)–1 d–1 for sediments plus 2.4 mg L–1 organic carbon. The rates increased for higher organic carbon amendments. The results of push-pull tests were consistent with first-order denitrification in the subsurface with initial rates of 4.8 to 6.7 mg N L–1 d–1, which was in agreement with the microcosm results. The microcosm and denitrification experiments indicated that a significant fraction of nitrate in ground water could be consumed by denitrification within the typical residence time of shallow ground water.
Total N concentrations in ground water were similar to those of tile drainage (>8 mg L–1) in two wells. In the other wells, N concentrations were variable but usually <2 mg L–1. These results are consistent with high-N water moving below the tile drains to recharge shallow ground water and undergoing subsequent denitrification. Nitrate was the predominant N species in all samples with over 2 mg L–1 total N. Organic N and NH4–N accounted for a significant fraction of total N (10 to 40%) in some samples. Most samples had both detectable Fe and DO, probably because ground water from different strata was being withdrawn from all wells. Therefore, redox conditions were favorable for denitrification at some levels at all sites. Total N concentrations in stream water were very low (<0.4 mg L–1) during periods of low flow (base flow). This is consistent with denitrification before discharge to the stream.
At the site with the highest total N concentrations, the isotopic composition of nitrate in ground water was the same as in tile drainage, which is consistent with recharge of shallow ground water with high-N water from the surface. At other sites, the nitrate in ground water was isotopically heavier in both N and O than in tile drainage, which is consistent with denitrification in the subsurface. At these sites, the degree of denitrification estimated from the changes in isotopic composition ranged from 30 to 70%.
A mass balance of denitrification was developed. The recharge rate was obtained from a steady-state, ground water model and the NO3–N concentration in recharge water was assumed to be the average concentration in tile drainage. We assumed that, as ground water flowed from the water table to the stream, denitrification was complete. The amount of N denitrified per year was equivalent to 0.3 to 6.4% of the applied fertilizer N. The best estimates varied by water year—2.3% in 2000, 1.2% in 2001, and 3.1% in 2002, corresponding to differences in precipitation. More denitrification occurred in 2002, a year of normal precipitation, than in 2000 and 2001, 2 yr of below average precipitation. The estimates of N denitrified in shallow ground water represent 9 to 27% of the estimated N exported from the watershed via surface water. Thus, denitrification in the shallow subsurface of watersheds with fine-grained soils and geologic materials may be a significant N sink compared with N exported via surface water.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  L. E. Gentry, M. B. David, F. E. Below, T. V. Royer, and G. F. McIsaac Nitrogen Mass Balance of a Tile-drained Agricultural Watershed in East-Central Illinois J. Environ. Qual., July 23, 2009; 38(5): 1841 - 1847. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. D. Robertson and L. C. Merkley In-Stream Bioreactor for Agricultural Nitrate Treatment J. Environ. Qual., January 13, 2009; 38(1): 230 - 237. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | |||
