| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a USGS, 345 Middlefield Rd., Menlo Park, CA 94025
b USGS, 12201 Sunrise Valley Dr., Reston, VA 20192
c USGS, 6000 J St., Placer Hall, Sacramento, CA 95819
d USGS, 810 Bear Tavern Rd., West Trenton, NJ 08628
e J.M. Denver, USGS, 1289 McD Dr., Dover, Delaware 19901
f USGS, 10615 SE Cherry Blossom Dr., Portland, OR 97216
* Corresponding author (ctgreen{at}usgs.gov).
Received for publication September 29, 2006.
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
Abbreviations: AVS, acid volatile sulfide • CA, California • CFC, chlorofluorocarbon • CFIRMS, continuous-flow isotope-ratio mass spectrometry • DOC, dissolved organic carbon • IN, Indiana • MD, Maryland • NE, Nebraska • WA, Washington
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
Despite the growing interest in contamination and natural remediation of NO3– in ground water, important questions remain about the overall regional and global importance of ground water denitrification (microbial reduction of NO3– to N2) (Seitzinger et al., 2006) and the sources of electron donors contributing to this microbial reaction. These questions persist in part because of the difficulties of comparing results from single-site studies that use a wide variety of methods (Groffman et al., 2006). Studies are needed that apply robust methods for measuring denitrification among multiple sites. As a part of the National Water-Quality Assessment Program's investigation of agricultural contamination, this study uses analysis of in situ gradients of stable isotopes, gases, and major ions to evaluate the controlling processes and the rate of denitrification at four agriculturally important sites across the United States.
Although various research studies have documented removal of NO3– via denitrification in individual aquifers (e.g., Trudell et al., 1986; Postma et al., 1991; Smith et al., 1991; Hiscock et al., 1991; Korom, 1992; Böhlke and Denver, 1995; Puckett and Cowdery, 2002; Böhlke et al., 2002; Tesoriero et al., 2005; Korom et al., 2005), the quantitative effects of ground water denitrification within larger regions remain highly uncertain. A recent review of denitrification at the global scale (Seitzinger et al., 2006) identified ground water as a potentially important sink for natural and anthropogenic N in the hydrologic cycle, but the role of ground water denitrification was highly uncertain because of the scarcity of transferable data. Similarly, forward and inverse models of watershed nitrogen fluxes have relatively large uncertainties associated with transport and reaction in ground water (e.g., Bicknell et al., 1997; Schwarz et al., 2006). Many reported rates are for sites that were specifically selected for known or suspected denitrifying activity (e.g., Trudell et al., 1986; Korom et al., 2005). As a result, the potential exists for bias in published rate estimates because of interest in the presence, rather than the absence, of natural attenuation reactions (Commission on Geosciences, Environment, and Resources, 2000). To improve assessment of the range of denitrification rates in regional ground water, previous analyses need to be combined with newer studies that investigate denitrification in multiple aquifers chosen for a range of important agricultural settings and without regard for the expected rates.
Questions remain as to the sources of electron donors that react with NO3– in shallow aquifers. In denitrification, NO3– serves as the terminal electron acceptor for bacteria that derive energy from the oxidation of a reduced substance. For electron donors, microbes commonly rely on organic C or, in the case of autolithotophs, reduced forms of S or Fe (Korom, 1992). Several studies have identified solid phases in aquifers as sources of electron donors for denitrification (Kölle et al., 1985; van Beek and van Puffelen, 1987; Frind et al., 1990; Postma et al., 1991; Böhlke and Denver, 1995; Robertson et al., 1996; Tesoriero et al., 2000; Böhlke et al., 2002; Böhlke et al., 2007). Dissolved organic carbon (DOC) in recharging ground water can also serve as an electron donor. Few studies have provided direct measurement of relict and recharging electron donors, and the relative contribution of recharging electron donors remains an area of active research (van Beek and van Puffelen, 1987; Starr and Gillham, 1993; Puckett and Cowdery, 2002; Siemens et al., 2003). Additional work is needed to understand the relative contribution of relict and recharging electron donors in agricultural settings.
Among methods to estimate denitrification rates at the scale of aquifers and larger, analysis of in situ concentration gradients of environmental (as opposed to injected) tracers has several advantages. Although biological samples and assays can provide important information about potential rates and microbial communities involved in denitrification, the rates derived from such methods are typically not representative of in situ rates at the scale of aquifers (Groffman et al., 2006). Sampling of ground water from observation wells minimally perturbs the system as compared to installation of microcosms, injection of solutes, or analysis in the laboratory of sediment cores or slurries. Because NO3– concentration gradients can result from the spatial or temporal variability of source concentrations it is important to include analyses of N2 in assessments of denitrification progress and rates (Vogel et al., 1981; Böhlke and Denver, 1995; Böhlke, 2002; Böhlke et al., 2002; Puckett and Cowdery, 2002). In addition, in situ gradients of stable isotopes can provide substantiating evidence as to the extent of denitrification as well as historical changes in sources. Combining these data with measurements of environmental age tracer concentrations allows estimation of denitrification rates at the scale of aquifers. Few studies have conducted a complete, simultaneous analysis of the various N species, particularly N2, coupled with an analysis of the redox processes, stable isotopes, and the ground water flow system involved, leaving many questions unanswered with respect to the true fate of NO3–.
This study uses combined methods including analyses of major-element chemistry, dissolved gases, and stable isotopes, as well as age dating and numerical flow modeling to evaluate the extent of mitigation of nitrate contamination by denitrification in contrasting aquifers that are representative of important agricultural settings across the United States. At four shallow, sandy aquifers overlain by a wide range of unsaturated zone, climatic, and agricultural conditions, a single study approach is used to identify the histories of NO3– transport and denitrification in areas of recharging ground water. This article identifies the spatial variations in NO3– distribution and denitrification and examines the role of physical and chemical processes in controlling these variations. Denitrification rates are estimated and are compared to previous studies to evaluate the broader importance of denitrification in the types of shallow agricultural aquifers represented by these four study sites.
Site Descriptions
Four study sites were chosen in agricultural watersheds in the San Joaquin River watershed, California (CA); Elkhorn River watershed, Nebraska (NE); the Yakima River watershed, Washington (WA); and the Chester River watershed on the Delmarva Peninsula, Maryland (MD) (Fig. 1
). All four study sites are underlain by sandy aquifers (63 to 89% sand and gravel) with low organic matter content (0.8 to 2.3%) and several meters of overlying unsaturated zone. More detailed descriptions of the surficial sedimentary geology are provided in Table 1
. The water table was generally within 10 m of the land surface at the MD, WA, and CA sites, but reached a maximum of 22 m at NE. Saturated thicknesses of the aquifers range from 16 m at MD to 31 m at CA and NE, to 100 m at WA. Climates, irrigation practices, and crop types contrast among the sites. Climates ranged from arid/semiarid at the western sites (WA and CA) to humid continental at the mid-continent (NE) and humid subtropical at the eastern coast (MD) sites. The arid sites receive the majority of water during the growing season from irrigation from surface water diversions, which supports a wide variety of crops such as fruit and nut orchards, corn [Zea mays L.], pasture, vineyards, and vegetables. The main crop types at the humid sites are corn and soybeans [Glycine max (L.) Merr.] planted in rotation. Additional details of the characteristics of the watersheds containing each of these sites can be found in Capel (2008).
|
|
|
Materials and Methods |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
Well Installation
At each study location, multiple well nests were installed along transects that extended from the edge of a stream up the topographic gradient to a distance of approximately 0.5 km at MD and NE, 1 km at CA, and 2.5 km at WA. Each transect included one well nest in or near the riparian zone adjacent to the ground water/surface water interaction study area (Puckett et al., 2008). Upgradient well nests were located in ground water recharge areas beneath agricultural fields and were collocated with unsaturated zone study locations (Green et al., 2008). The length and location of each transect was constrained in part by practical considerations such as land owner permissions. At each well nest, multiple well screens were installed to extend to approximately 10 m below the water table at MD and NE and approximately 25 m below the water table at CA and WA. Each drilling site was assigned a three digit code (Fig. 1). Individual well screens were assigned a four digit code by appending the well location code with a letter starting with "p" for the shallowest screen and subsequent letters for deeper screens. Wells were constructed of 5-cm-diameter polyvinyl chloride pipe with slotted screens and were typically installed using hollow-stem augers. The annulus around the screen was filled with silica sand, and the remaining annulus was grouted with bentonite, with a cement cap at land surface. Well installation followed standard procedures as described in Lapham et al. (1997).
Dissolved Gases and Denitrification Progress
Dissolved N2 and Ar concentrations were used to estimate the quantities in ground water of dissolved gases originating from atmospheric and biological sources (Heaton and Vogel, 1981; Vogel et al., 1981; Böhlke and Denver, 1995; McMahon and Böhlke, 1996; Puckett et al., 2002; Böhlke et al., 2002). Samples for analysis of dissolved N2 and Ar were collected in filled serum bottles with no headspace and analyzed by gas chromatography after creation of low-pressure headspace in the laboratory (http://water.usgs.gov/lab/dissolved-gas/; verified 24 July 2007). Contamination and degassing were minimized by sampling using positive displacement pumps. Results of the analyses were corrected for solubility in sample water at laboratory temperatures and have typical uncertainties of ±1 to 2%.
In ground water, dissolved gases may originate from equilibrium exchange with the atmosphere at the water table, dissolution of entrapped air bubbles, and production by reactions such as denitrification. Typically, dissolved gases include some fraction from bubbles of air that become trapped under recharging water and entrained in the saturated zone. As long as the hydrostatic pressure remains greater than the total pressure of gases in solution, degassing is unlikely (Blicher-Mathiesen et al., 1998). Similarly, denitrification produces N2 that remains in solution in recharging ground water. Recent literature has used the terms "excess air" to refer to atmospheric gases originating from entrained bubbles, and "excess N2" to refer to N2 originating from denitrification.
Excess air and excess N2 concentrations in ground water were estimated using the concentrations of N2 and Ar, their solubility in water (Weiss, 1970), the atmospheric pressure, and the recharge temperature. This calculation required several assumptions, including (i) the only source of Ar was the atmosphere, (ii) the only sources of N2 were the atmosphere and denitrification, and (iii) excess air was not fractionated. In addition, it was assumed that all samples at each study site were recharged at the same temperature, but with varying amounts of excess air. The latter assumption is considered to be the largest source of uncertainty because of possible variations in recharge temperatures, but in the absence of other noble gas measurements, this procedure minimizes errors associated with minor air contamination or partial degassing during sampling (Böhlke et al., 2002). The recharge temperature was based on the annual average ground water temperature in monitoring wells near the water table (13.0°C at MD, 13.3°C at WA, 19.2°C at CA, and 10.8°C at NE). The excess air concentration in each sample was calculated with
 | [1] |
 | [2] |
 | [3] |
The cumulative progress of denitrification that has occurred in a ground water sample is given by the concentration of N2,denit and it can also be expressed as the fraction of the original NO3– that has been denitrified (
denit):
 | [4] |
 | [5] |
denit would indicate that half of the original NO3– had been denitrified somewhere along the ground water flow path after the water ceased to exchange with unsaturated-zone air in the recharge area. Uncertainties in the reaction progress estimate were caused mainly by uncertainties in the assumed recharge conditions affecting the calculation of [N2,denit].
Stable Isotopes
The concentrations and isotopic compositions of NO3– and N2 in selected samples were used to assess NO3– sources and effects of denitrification (Vogel et al., 1981; Böhlke et al., 2002). The N and O isotopic composition of NO3– was determined by bacterial reduction to N2O and continuous-flow isotope-ratio mass spectrometry (CFIRMS) (Sigman et al., 2001; Casciotti et al., 2002; Coplen et al., 2004). The isotopic composition of dissolved N2 was determined by gas chromatograph separation and CFIRMS on headspace gas leftover after gas chromatograph analysis of dissolved gas concentrations. The isotope data were normalized according to recommendations in Böhlke and Coplen (1995) and Böhlke et al. (2003). The isotopic composition of the [N2,denit] component, 
15N[N2,denit], was calculated using a mass balance equation:
 | [6] |
15N[N2,equil] = +0.7 
and 15N[N2,bub] = 0.0 . The 15N value of the initial NO3–, (15N[NO3–]0), then was calculated from mass balance:
 | [7] |
15N and 13C and water samples for 18O (http://isotopes.usgs.gov; verified 24 July 2007). Sediments were collected from the upper 0.5 m near selected well nests and from drilled cores below the water table. Sediments were dried, ground, homogenized, and analyzed for 15N and 13C by CFIRMS following decomposition in an elemental analyzer. Data were normalized as described in Qi et al. (2003), with typical reproducibilities of ± 0.5 (2-sigma). For 13C of the organic carbon fraction, samples were acid treated to remove carbonates before decomposition. Water samples were analyzed by CO2 equilibration and dual-inlet mass spectrometry, normalized to Vienna Standard Mean Ocean Water and Standard Light Antarctic Precipitation with typical reproducibilities of ± 0.2 (2-sigma).
Ground Water Dating by Chlorofluorocarbons, Sulfur Hexafluoride, and Tritium
Samples for analyses of the chlorofluorocarbons CFCl3 (CFC-11), CF2Cl2 (CFC-12), and C2F3Cl3 (CFC-113) were collected in glass bottles with aluminum-lined caps and analyzed by electron-capture gas chromatography with a detection limit of 0.3 pg kg–1 for CFC-11 and CFC-12, and 1.0 pg kg–1 for CFC-113 (Busenberg and Plummer, 1992; http://water.usgs.gov/lab/cfc/; verified 24 July 2007). Samples for analysis of sulfur hexafluoride (SF6) were collected in glass bottles with polyseal plastic-lined caps and analyzed by electron-capture gas chromatography with a detection limit of 0.01 fmol L–1 and a precision of about 20% near the detection limit and about 3% at concentrations above 0.1 fmol L–1 (Busenberg and Plummer, 2000).
For this study, two methods were used to estimate recharge dates based on measured concentration of tracers. First, piston flow estimates of "apparent" ages were assigned on the basis of a comparison of CFC or SF6 equilibrium partial pressures, corrected for recharge temperature and excess air with chronologies of atmospheric concentrations (Busenberg and Plummer, 1992, 2000). This has been a common approach in previous studies using ground water ages to calculate denitrification rates. It relies on the assumption that all water in a sample is recharged at the same time and that the concentration of age-dating tracers in the sample is identical to the concentration at the water table at the time of recharge. In a few cases, tritium was used for quality assurance to screen for possible gross errors in CFC or SF6 age dates. Tritium samples were collected in safety-coated glass bottles sealed with polyseal plastic-lined caps. Tritium was enriched by electrolysis and analyzed by liquid scintillation counting (Thatcher et al., 1976) with an analytical uncertainty of approximately ±0.7 tritium unit. Permissible age ranges were assigned by comparing chronologies of tritium in rainwater decayed to the present.
The second method for estimating ages was a "mixed-age" analysis. Ground water dispersion and sampling across a range of depths can result in samples that contain a mixture of waters of different ages. In such cases, the average age can differ substantially from the piston flow age estimate from a single tracer (Zuber, 1986; Cook and Böhlke, 2000; Weissmann et al., 2002). For CFC dating, the discrepancy is small for waters younger than 20 to 30 yr, such as at the MD site. The discrepancy increases with age, however, and is potentially significant for the older waters at the CA, NE, and WA sites. To account for possible effects of mixing of different waters on apparent ground water ages, age distributions were assigned to individual samples. For NE and WA, multiple age-dating tracers were available for most samples. For these samples, several dozen unimodal and bimodal discrete-age probability density functions were manually constructed by assigning ad hoc proportions across the range of ages to represent interpretations of possible mixing scenarios. To calculate modeled tracer concentrations, the atmospheric concentration record was used to assign concentrations to each age category. Concentration was multiplied by the proportion of water in that category, and the overall concentration was determined by summing these products for all age categories. The age distribution that reproduced the measured tracer concentrations was chosen as the best fit scenario.
At CA, multiple age tracers were generally not available. For each CA sample, a discrete-age probability density function was estimated using the backward random-walk particle-tracking approach outlined in Weissmann et al. (2002) with a dispersivity of 4 cm. Modeled tracer concentrations were calculated in the same manner as for NE and WA. For each well screen, models were fitted to data by shifting all age categories in the probability density function up or down by a uniform offset until the mean tracer concentration matched the observed value. For the CA, NE, and WA data, the arithmetic averages of the best-fit age probability density functions were used as the mixed-age estimates.
Solid Phase Electron Donors and Mineralogy
To estimate the concentration of electron donors in sediments, cores were analyzed for mineralogy, Fe(II), Fe(III), and sulfides. Mineralogy was determined by X-ray diffraction. Iron species were characterized using hydrochloric acid and hydroxylamine extractions. Sulfides in the form of acid-volatile sulfide (AVS) and pyrite-S were characterized by a variety of methods (see Puckett et al., 2008).
Numerical Modeling
The MODFLOW code (Harbaugh et al., 2000) was used to develop three-dimensional numerical models of ground water flow in the vicinity of the well transects at all four sites. Flow models were manually calibrated using MODFLOW's parameter estimation routines and with manual adjustments to match hydraulic heads measured in observation wells, surface water elevations, and observations of hydraulic gradients in the stream beds. All models were steady-state and approximated the average base flow conditions. To constrain the flow and boundary conditions of local-scale site models, the model domains were nested within existing regional-scale flow models. Additional details about model construction and calibration are available in Capel (2008).
The flow vectors from the local-scale models were used to run backward advective particle tracking in MODPATH (Pollock, 1994), which allows estimation of the recharge source areas contributing to each well screen (e.g., Modica et al., 1998; Rayne et al., 2001; Paradis et al., 2007). For all study locations, digitized maps of crop types and advective particle tracking results were used to characterize the spatial scale and current land use in the source area of each well screen. Source areas for well transects were estimated by drawing a polygon around the particle trajectories associated with all wells within each transect. The length of the source area was measured along a line intersecting the approximate center of the source area for each well in the array. In addition, average path lengths were computed for each well based on all particles contributing to that well.
|
Results and Discussion |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
|
Inter-site Comparison of NO3– Concentrations and Denitrification at the Aquifer Scale
To provide a general overview and comparison of NO3– and redox chemistry among sites, the medians and interquartiles of travel times, O2, denitrification progress, and NO3– were calculated for each site (Fig. 3
). The median values for each site are based on the full set of data from all monitoring wells at that site. The results reveal broad-scale geochemical trends indicating that, despite differences in hydrology and land use, these four aquifers can be seen to represent different stages along a gradual and consistent progression of redox conditions. Study locations with old median ground water ages tend to have low median dissolved O2 concentrations. From MD to WA to CA to NE, a trend of increasing age (in both piston flow and mixing based estimates) corresponds to a trend of decreasing O2 concentrations (Fig. 3). Along with the trend in O2 concentration, the prevalent electron acceptors (as determined using the hierarchical scheme of Paschke et al. (2007) shift from O2 at MD to SO42–, Mn(IV), Fe(III), NO3–, and O2 at NE. Among all four study locations, the correlations between aquifer-scale median O2, redox status, and age are similar to patterns commonly observed along individual flow paths in aquifers receiving oxic recharge (Champ et al., 1979; Chapelle et al., 1995). The occurrence of such patterns at the aquifer scale implies that the reactive characteristics of these aquifers may be similar, whereas deviations from the correlation between age and redox status would imply more variability in the reaction rates among the sites. To the extent that reactive characteristics are similar, age may serve as a useful predictor of redox status for groups of aquifers.
|
N-species concentration data indicate that N2 is the primary end product of NO3– reduction in these aquifers. Possible alternative products such as NH4+ and NO2– are seldom detected, and concentrations are below 0.02 mmol L–1 in all ground water samples from all sites. Nitrous oxide (N2O) can occur as an intermediate product of denitrification, but in the low O2 zones where denitrification occurs in aquifers, a large fraction of produced N2O is reduced to N2 (Well et al., 2005). This is corroborated by N2O concentrations at N22, where the zone of active denitrification contains only 0.1 µmol L–1 N2O (Green et al., 2008) as compared to 351 µmol L–1 excess N2.
Stable isotope ratios and excess N2 gas concentrations indicate that some denitrification occurs at all study locations (Fig. 3). Denitrification even occurs locally at MD, where measured O2 concentrations always exceed 3.0 mg L–1 (0.094 mmol L–1) (Fig. 3d), indicating that individual samples may contain water from both aerobic and anaerobic zones. In each study unit, values of 15N[NO3–] and 18O[NO3–] of ground water samples tend to be positively correlated (Fig. 4a
). The apparent fractionation slopes (
15N/18O) for MD (0.8), CA (2.0), and NE (1.3) are within or close to the reported range of 1 to 2 for NO3– reduction processes including denitrification (Böttcher et al., 1990; Granger et al., 2004). The apparent slope at WA is 0.5, which indicates that factors other than denitrification, such as variability of source isotopic composition, may affect the trends in stable N and O isotopes more strongly at WA than at the other study locations. In a plot of denitrification progress (the fraction of N occurring as excess N2) versus increase in 15N[NO3–] since recharge (Fig. 4b), most samples plot near a region bounded by Rayleigh fractionation curves with apparent fractionation factors (
) of –5 to –20, which are similar to values reported for denitrification in ground water elsewhere (Mariotti et al., 1988; Böttcher et al., 1990; Böhlke et al., 2002). Samples with relatively small apparent fractionation effects may contain mixtures of waters with varying amounts of denitrification as a result of heterogeneous redox conditions encountered during transport, whereas some of the larger fractionation effects are closer to those expected in homogeneous systems.
|
Variability of NO3– Concentrations within Aquifers
Correlations among solutes indicate that non-reactive processes such as N loading rates and physical mixing control the distribution of N concentrations in these aquifers. Table 2
shows Pearson product moment correlations between NO3– and other solutes. Each correlation value is based on all available samples from an individual study site. In contrast to sites where rapid denitrification produces low NO3– in zones with high excess N2 and low O2 (e.g., Tesoriero et al., 2000), NO3– does not correlate with excess N2 at these study sites, and only correlates with O2 at the NE site (Table 2). Correlations of NO3– with [NO3–]0 are 0.87 or greater at all sites, indicating that NO3– concentrations have not changed greatly since the time of recharge, and that substantial denitrification progress is limited to low [NO3–]0 samples. Nitrate frequently correlates with specific conductance, Ca2+, Mg2+, K+, Cl–, SO42–, and DOC, which are typically associated with agricultural applications such as manure, N fertilizers, dolomitic lime, potash, and gypsum. At these sites, the lack of correlations that would indicate a strong influence of denitrification on NO3– concentrations and the abundant correlations of NO3– with other agricultural chemicals indicate that the history and distribution of application rates are the primary determinant of the observed concentrations of NO3–.
|
|
15N[NO3–] and 18O[NO3–] values of relatively young, undenitrified samples (Fig. 4a). For aerobic samples with more than 0.5 mg L–1 O2 (0.016 mmol L–1) (the lower threshold of Chapelle et al. (1995) is used to allow comparison among sites with no "strictly aerobic" samples) and water age <25 yr, 15N[NO3–] values are 6.4 to 10.7 for NE, 5.4 to 8.5 for CA, 5.2 to 16.4 for WA, and 2.3 to 6.1 for MD. Values of 18O[NO3–] in these aerobic samples are 4.1 to 5.3 for NE, –0.54 to 1.2 for CA, –4.2 to 0.6 for WA, and 0.46 to 5.4 for MD. Sources of soil N or mineral fertilizer typically produce 15N[NO3–] of approximately –2 to 8 (Heaton, 1986; Fogg et al., 1998), which fits the range of most samples from these study locations. Higher values, especially greater than 10, may reflect modern manure sources, such as at well N21p, completed in shallow ground water below an actively grazed pasture (15N = 10.7), and well W20p (15N = 16.4), completed in shallow ground water at the edge of a field with manure applications. For samples with measurements of 18O of water and of NO3– (Fig. 4c), these measurements define a linear trend that is similar to one proposed for nitrification (microbial oxidation of NH4+ to NO3–) in soils (Amberger and Schmidt, 1987). The negative correlations at all sites between NO3– and pH (Table 2) are consistent with production of H+ ions with NO3– during nitrification and transport into aquifers with limited acid neutralizing capacity, as noted previously in the vicinity of the MD site (Böhlke and Denver, 1995).
The reconstructed 15N[NO3–]0 values in denitrified samples generally are similar to local measured values of 15N in bulk organic matter in surface soils (Fig. 5
). The ranges of 15N for recharging NO3– in young, aerobic ground water and for surface soils at WA, CA, and NE were slightly higher than those at MD, and were also higher than background values observed in aquifer sediments. The elevated 15N values of surface soils and reconstructed NO3– sources at WA, CA, and NE may reflect natural differences in the soil N pools or may indicate a partial replacement of the reactive soil N pool by 15N-enriched sources such as manure. Based on site average concentrations of 0.0010 to 0.0017 g g–1 N, approximate bulk density of 1.8 g cm–3, and sampling depth of 30 cm, the shallow soils at these sites contain 5400 to 9000 kg ha–1 N, which amounts to 38 to 92 times the corresponding annual N application rates at each site (Green et al., 2008). Therefore, it is feasible that agricultural applications of N have replaced a substantial portion of the shallow soil N pool within a few decades. The 15N values of aquifer sediments were relatively low and uniform among all study locations (Fig. 5) (median 15N ranging from 3.1 to 4.9) irrespective of redox status or agricultural practices, indicating that the N in these aquifer sediments is relict and has not been altered by interaction with the modern,15N-enriched N species.
|
15N[NO3–]0 values at some of the sites may reflect changes in the dominant NO3– sources over time. At CA, the reconstructed 15N[NO3–]0 values increase with age (Fig. 6a
), consistent with a shift from manure to chemical fertilizers over the last half century. This trend matches the history inferred from aerial photographs that show that, starting 40 yr ago, the predominant agriculture in the source area began to switch from row crops, which typically receive manure, to orchards, which typically do not (Ruddy et al., 2006). Similarly, at MD, 15N[NO3–]0 was higher for older samples (5.4 to 5.7; water age = 17 yr) compared to younger samples (2.1 to 2.3; water age = 2 to 5 yr), reflecting the increasing use of mineral fertilizers following the cessation of chicken manure applications in the mid-1990s, as reported by the land owner.
|
Trends in Denitrification within Aquifers
The time required for denitrification to deplete NO3– in ground water varies widely among sample points at the four study locations, as shown by the scatter of data in a plot of denitrification progress (denit, Eq. [4]) versus age (Fig. 6c). The depletion of O2 with age is more systematic (Fig. 6d). Concentrations of O2 typically decline during the first 20 yr after recharge and then remain below 0.15 mmol L–1.
At the CA site, progression of denitrification is relatively uniform across the transect, and denit correlates positively with apparent age (P = 0.005) (Fig. 6c). On the basis of a least squares linear regression of the data, denitrification begins approximately 8 yr after recharge and depletes 50% of NO3– in the ground water after 31 yr of reaction time. The timing of onset of denitrification is consistent with trends observed at other sites (Böhlke and Denver, 1995; Tesoriero et al., 2000; Puckett and Cowdery, 2002; Tesoriero et al., 2005). The sample locations with active denitrification (0.0 < denit < 1.0) define a broad zone that spans 20 m in thickness and 1 km in width (Fig. 7a
).
|
15N[NO3–] and 18O[NO3–] in the unsaturated zone above N22p are comparable to source isotopic compositions (Green et al., 2008), suggesting that the onset of denitrification is below the top of the capillary fringe. Complete denitrification of source water appears to occur within a 3.5-m-thick zone between the water table and N22p (Fig. 7b). Similarly, the denitrification progress and water age at N20p imply denitrification of 84% of source NO3– within 11 yr of travel time. In contrast, at N23 and N21, low concentrations of excess N2 relative to NO3– are consistent with more gradual denitrification. Across all depths at N23, the denitrification progress increases with age such that 50% denitrification occurs after 13 yr of reaction time (Fig. 6c). At N21, apparent ages and N species concentrations imply rates of 50% denitrification in less than 24 yr. Overall, the data from NE indicate 50% denitrification in <8 to 25 yr, although, as previously noted, high denitrification progress is limited to low [NO3–]0 samples, and it is not evident that the same progress will occur for higher [NO3–]0 waters. The zone of active denitrification ranges in thickness from <3.5 m at N22 to 6.2 to 10.2 m at N23 (Fig. 7b). The variation in rates along the NE transect may result in part from the unusual orientation of the transect relative to the regional ground water gradient (Fig. 1). NE is the only study location with modeled source areas in riparian zones, which often have relatively low [NO3–]0 and high rates of ground water denitrification (e.g., Vidon and Hill, 2004). At WA and MD, denitrification appears to be limited in extent due to aerobic conditions. In samples from both study sites, denitrification progress was less than 0.5 (Fig. 6c) and O2 was frequently greater than 0.05 mmol L–1 (Fig. 6d). At WA, zones of active denitrification were disconnected and irregular. For example, at W21 and W24 there was no evidence of denitrification at any depth. At W31 and W32, excess N2 was detected in shallow wells but not in deeper ground water. Denitrification rates comparable to those at CA may occur locally, for example in the vicinity of W32p and W23q, but the overall rate of denitrification at this site is low. Three-quarters of the samples have denitrification progress less than 0.13, and most of the oldest ground water samples (apparent age >50 yr) have O2 greater than 0.1 mmol L–1 and show no evidence of denitrification. The variability of denitrification at WA likely reflects the large variability of agricultural land use within the source area, and intensive application of manure at some locations. At MD also, aerobic conditions likely limit the zone of active denitrification, which was encountered only at M20. At this location, denitrification progress was 0.10 at the shallowest well M20p (no age estimates available), 0.16 over 6 yr travel time to the middle well M20q, and 0.14 over 17 yr travel time to the deepest well, M20r.
The combined data from this multi-site study and previous studies using similar methods indicate that many of the denitrification rates reported in the literature may be skewed high relative to the range of rates in surficial aquifers like the ones considered here. At these study locations, zero-order rates as calculated by the excess N2 divided by the water age were 0 to 0.03 µmol N L–1 d–1 for WA, 0 to 0.06 µmol N L–1 d–1 for MD, 0 to 0.14 µmol N L–1 d–1 for CA, and 0 to 0.12 µmol N L–1 d–1 for NE. Figure 8 shows maximum zero-order in situ denitrification rates from previous studies. The references were obtained primarily from tables in Korom (1992) and Groffman et al. (2006) along with several more recent studies. The reported maximum denitrification rates summarized in Fig. 8 range from less than 10–2 to greater than 104 µmol N L–1 d–1. Even lower rates (10–5 to 10–4 µmol N L–1 d–1) have been reported from aquifers with ground water residence times that are thousands of years long as determined by 14C dating (e.g., Vogel et al., 1981; McMahon et al., 2004). Rates tend to be highest for studies using injection of solutes into aquifers, and rates are intermediate for estimates based on naturally occurring gradients of NO3– concentrations. The lowest rates are reported consistently by studies, such as this one, that analyze naturally occurring gradients of excess N2 concentrations. Moderate rates for N2 gradient methods are for studies in unusual redox settings, such as below a dairy waste lagoon (Singleton et al., 2007) and where the method is applied to a selected portion of the zone of active denitrification (Tesoriero et al., 2000). For injection-based studies, the high rates commonly reflect selection of sites or portions of aquifers with known or suspected denitrification (e.g., Trudell et al., 1986; van Beek and van Puffelen, 1987; Starr and Gillham, 1989; Korom, 1991; Korom et al., 2005). In the studies of NO3– gradients, estimated rates may be elevated due to historical trends that interfere with interpretation of NO3– gradients or due to selection of sites with known denitrification. In a study employing all three methods, Tesoriero et al. (2000) reported three rates that correspond to the pattern observed in the literature: the highest rates were reported for a reactive zone tested by injection of NO3–, intermediate rates were based on NO3– gradient, and the lowest rate was based on N2 gradients at the same location (Fig. 8).
|
denit divided by travel time) are highest at C22r in the most upgradient well cluster (Fig. 6c), where the highest zero-order rates also occur. At NE, the rates of denitrification progress appear similar for N20, adjacent to the river, and N22, distant from the river (Fig. 6c), and the highest zero-order rates occur at N22. The highest zero-order rates at WA occurred at W23q, over a km from the riparian zone. MD was the only site in which the well closest to the stream (M20) had the highest rate of denitrification progress and zero-order production of excess N2.
Electron Donor Controls on Reduction of NO3– and O2
Dissolved and solid phase chemical data indicate that electron donors originating in recharge water are minor contributors to denitrification and other redox reactions at these four study locations. Based on commonly assumed electron-balance reaction stoichiometries during denitrification, the molar ratios of NO3– reduced to electron donors oxidized are 0.8 for organic C(0), 2.8 for sulfide in FeS2, and 0.2 for Fe(II) (Böhlke, 2002). Aqueous phase Fe(II) and sulfide are not present in sufficient quantities in recharging water to account for the observed losses of NO3– or, further, the observed loss of O2. In shallow ground water, Fe(II) was never greater than 0.0005 mmol L–1, and sulfide was never greater than 0.042 mmol L–1 in any sample, as compared to typical concentrations of 0.5 to 1 mmol L–1 NO3– and 0.2 mmol L–1 O2 in young, shallow samples. At the maximum detected concentrations, the dissolved Fe(II) and sulfide could account for denitrification of 0.0001 and 0.07 mmol L–1 NO3–, respectively. Dissolved organic carbon is more abundant, and data at the NE site are consistent with DOC reactions in recharging water. At the NE site, DOC decreases linearly with water age (P = 0.04) and the specific ultraviolet absorbance (a potential indicator of the reactivity of DOC (Weishaar et al., 2003)) also tends to decrease linearly with water age (P = 0.02). The decrease in dissolved O2 is associated with a decrease in DOC (P = 0.002) with a least squares regression slope of 1.5 moles O2 per 1 mole DOC. These trends are consistent with oxidation of the more labile fraction of DOC during transport, or could possibly result from changes in quantity and quality of DOC relating to historical changes in the types of DOC sources. The change in DOC concentration from the highest to lowest concentrations at NE is 0.47 mmol L–1. Based on a simple electron balance (O2/C(0) = 1), this amount of DOC could reduce all O2 at its highest concentration (0.05 mmol L–1) in addition to 0.34 mmol L–1 NO3–. The 0.17 mmol L–1 excess N2 produced by this reaction, however, could not fully account for the 0.3 to 0.4 mmol L–1 excess N2 at many wells, indicating that other electron donors contribute to denitrification. At the other sites, recharging DOC does not contribute substantially to reduction of O2 or NO3–. Median DOC concentrations in recharging ground water are low (0.04 mmol L–1 at MD, 0.16 mmol L–1 at WA, and 0.10 mmol L–1 at CA) and are nearly identical to the median concentrations in O2 depleted (<1.0 mmol L–1 O2) ground water at those sites. In general, recharging electron donors are insufficient to account for losses of O2 and NO3– at the aquifer scale, and relict, solid phase materials in the aquifer are likely the primary sources of electron donors for O2 reduction and denitrification.
Concentrations of solid phase electron donors are low at the four study locations. Aquifer sediments at WA contained the highest median concentrations of Fe(II) (1400 µg g–1) and acid volatile sulfides (AVS) + pyrite-S (median = 3.9 µg g–1) (Fig. 9 ). Elsewhere, median Fe(II) ranged from 46 to 670 µg g–1 and median AVS + pyrite-S ranged from 0.1 to 1.6 µg g–1. The highest concentrations of organic C occurred at MD (median = 900 µg g–1) with medians of approximately 500 µg g–1 at the other study sites. Interestingly, the aquifers with more denitrification (CA and NE) tended to have lower organic C content. Nevertheless, concentrations of solid phase electron donors are sufficient to account for the losses of O2 and NO3– at these sites. For example, assuming a porosity of 0.33 and bulk density of 1.8 g cm–3, the aquifer material containing 500 µg g–1 Fe(II) could reduce 0.16 mmol L–1 of ambient O2 and 9.7 mmol L–1 NO3– (see Trudell et al., 1986). Likewise the AVS + pyrite-S concentrations of 50 µg g–1 could reduce ambient O2 plus 6.7 mmol L–1 NO3–. For organic carbon, 1000 µg g–1 could reduce ambient O2 and 364 mmol L–1 NO3–, which corresponds to tens to hundreds of pore volumes with the maximum observed NO3– concentrations at these sites. Although the total concentrations of solid phase electron donors clearly indicate the potential for reduction of O2 and NO3–, the quantities of reactive electron donors are uncertain because a substantial portion may be inert or physically inaccessible to denitrifying microbes (Korom, 1992; Puckett and Cowdery, 2002). The electron donor concentrations are relatively low compared to published values for other shallow aquifers with comparable rates of denitrification. For example, Böhlke et al. (2002) reported 100 to 200 µg g–1 of pyrite-S and 1000 to 2300 µg g–1 organic C in the recharge area of a glacial outwash sand aquifer in Minnesota where denitrification with rates on the order of 0.04 µmol N L–1 d–1 were related in part to pyrite oxidation. Larger reservoirs of FeS2 have been reported in denitrifying aquifers elsewhere (e.g., Postma et al., 1991; Robertson et al., 1996; Böhlke et al., 2007).
|
The results from the four sites indicate that solid phase electron donors strongly contribute to O2 reduction and denitrification, as has been observed elsewhere. If current N fluxes to shallow ground water persist, the continued consumption of a finite pool of relict solid phase electron donors by recharging O2 and NO3– will eventually deplete electron donors and reduce the rates of denitrification. The time required to produce a substantial effect is difficult to predict. Although the concentrations of electron donors in sediments are sufficient to reduce many pore volumes of recharge water, much of the electron donor pool may be inert over time scales of interest for contaminant transport. To address questions about the long-term prospects for denitrification, additional work is needed to quantify the reactivity of the relict electron donors.
|
Conclusions |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
Results show that denitrification does not strongly affect the distribution of NO3– throughout these four study transects. At all sites, NO3– exhibits strong correlations with other agricultural chemicals, indicating that the NO3– originates from agricultural sources, and that the application history mainly controls the distribution of NO3– mass in the aquifers. Numerical models of transect source areas show that the variability in observed concentrations for each study site relates to the scale of agricultural land use. More uniform land use within a source area corresponds to more uniform concentrations across the sample transect, further indicating that agricultural practices influence the spatial and temporal trends of NO3–.
Reconstructed isotopic source signatures and analyses of solid phase isotopes revealed information about the sources of NO3– contamination at these sites. Decreased values of reconstructed 15N[NO3–]0 in relatively young waters reflect an overall shift away from 15N-enriched sources, such as manure, and toward 15N-depleted sources, such as chemical fertilizers. Similarly, higher values of 15N of surface soils at WA, CA, and NE, as compared to MD, may reflect higher applications of 15N-enriched sources such as manure.
Trends in excess N2, O2, and isotopic fractionation indicate that some degree of denitrification occurs at all sites. All of the sites appear to be electron donor limited, however, and denitrification proceeds at the expense of relict electron donors in aquifer material. High values of fractional denitrification progress are limited to samples with initially low concentrations of NO3–, and it remains unclear whether the current reaction rates will persist as recently recharged waters with high NO3– encounter and deplete the labile relict electron donors. Although all these areas of ground water recharge below agricultural fields show some denitrification, rates tend to vary within each site, and are often too slow to substantially mitigate NO3– contamination before transport to discharge zones or deeper ground water receptors.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMaterials and MethodsResults and DiscussionConclusionsREFERENCES |
|---|
15N determinations of nitrates containing mass-independent 17O. Rapid Commun. Mass Spectrom.
18
:245–250.[CrossRef][Web of Science][Medline]This article has been cited by other articles:
|
|
 |
|
  L. M. Debrewer, S. W. Ator, and J. M. Denver Temporal Trends in Nitrate and Selected Pesticides in Mid-Atlantic Ground Water J. Environ. Qual., September 2, 2008; 37(5_Supplement): S-296 - S-308. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. D. Capel, K. A. McCarthy, and J. E. Barbash National, Holistic, Watershed-Scale Approach to Understand the Sources, Transport, and Fate of Agricultural Chemicals J. Environ. Qual., May 1, 2008; 37(3): 983 - 993. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. J. Puckett, C. Zamora, H. Essaid, J. T. Wilson, H. M. Johnson, M. J. Brayton, and J. R. Vogel Transport and Fate of Nitrate at the Ground-Water/Surface-Water Interface J. Environ. Qual., May 1, 2008; 37(3): 1034 - 1050. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. T. Green, L. H. Fisher, and B. A. Bekins Nitrogen Fluxes through Unsaturated Zones in Five Agricultural Settings across the United States J. Environ. Qual., May 1, 2008; 37(3): 1073 - 1085. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. V. Steele, H. M. Johnson, M. W. Sandstrom, P. D. Capel, and J. E. Barbash Occurrence and Fate of Pesticides in Four Contrasting Agricultural Settings in the United States J. Environ. Qual., May 1, 2008; 37(3): 1116 - 1132. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. H. Duff, A. J. Tesoriero, W. B. Richardson, E. A. Strauss, and M. D. Munn Whole-Stream Response to Nitrate Loading in Three Streams Draining Agricultural Landscapes J. Environ. Qual., May 1, 2008; 37(3): 1133 - 1144. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. L. Domagalski, S. Ator, R. Coupe, K. McCarthy, D. Lampe, M. Sandstrom, and N. Baker Comparative Study of Transport Processes of Nitrogen, Phosphorus, and Herbicides to Streams in Five Agricultural Basins, USA J. Environ. Qual., May 1, 2008; 37(3): 1158 - 1169. [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 | |||
