HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (4) Citing Articles via Google Scholar Google Scholar Articles by Well, R. Articles by Myrold, D. D. Search for Related Content PubMed Articles by Well, R. Articles by Myrold, D. D. Agricola Articles by Well, R. Articles by Myrold, D. D. Related Collections Vadose Zone Processes and Chemical Transport Nutrient Cycling Soil Methods/Instrumentation

DIVISION S-3—SOIL BIOLOGY & BIOCHEMISTRY

A Proposed Method for Measuring Subsoil Denitrification In Situ

^a Institut für Bodenwissenschaft, Von Siebold–Str. 4, D-37075 Göttingen, Germany
^b Dep. of Crop and Soil Science, Oregon State Univ., Agric. Life Sci. Bldg. 3017, Corvallis, OR 97331-7306

* Corresponding author (David.Myrold{at}orst.edu)

	ABSTRACT

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
We present and test a new approach for measuring denitrification in unsaturated subsoils in situ. The procedure consists of applying a test solution containing ¹⁵NO^-₃ to the subsoil, measuring steady-state ¹⁵(N₂ + N₂O) concentration in the pore space of the ¹⁵N-labeled soil, and determining denitrification rates by fitting measured and simulated concentrations. We adapted and evaluated a numerical, spherical diffusion model to simulate production and diffusion of ¹⁵(N₂ + N₂O) evolved from a ¹⁵(N₂ + N₂O)-producing subsoil volume surrounded by an infinite nonproducing soil volume. The effects of gas diffusivity, gas production rate, and spatial distribution of ¹⁵NO^-₃ on steady-state ¹⁵(N₂ + N₂O) concentrations were investigated with computer simulations, which proved the methodological principle theoretically. Field experiments were done using two different probe designs. Initially, multipurpose probes were used for applying ¹⁵NO^-₃ solution to the subsoils and subsequent collection of gas samples, but denitrification rates were unrealistically high, probably because of depressed soil gas diffusivity caused by the probe installation procedure. Nondisturbing capillary probes were used for two subsequent in situ experiments. Here, in situ ¹⁵(N₂ + N₂O) concentrations were near the detection limit and resulted in denitrification rates between 0 and 7 g N ha^-1 d^-1 10 cm^-1, which were comparable to rates determined by laboratory reference methods. We concluded that the in situ ¹⁵N gas emission method is a promising tool for measuring subsoil denitrification because it overcomes some of the limitations of the alternative techniques.

Abbreviations: a_p, ¹⁵N atom fraction of the NO^-₃ pool • d, fraction of denitrified N₂

	INTRODUCTION

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
LEACHING OF NO^-₃ is well recognized to be a large problem for groundwater reservoirs throughout the world. In an extensive review by Spalding and Exner (1993), the magnitude of the problem was described at a global scale. Nitrate contamination of groundwater is an ongoing process and extensive research (Smith et al., 1991; Geyer et al., 1992) has investigated various sources and sinks for NO^-₃ in soil. Recent research has focused on possible N transformations that may take place when NO^-₃ leaches from surface soils to the groundwater. In particular, the denitrification process has been considered as a possible sink for NO^-₃ in the rooting zone and in the vadose zone (Rice and Rogers, 1993; Van Cleemput, 1998).

Surface soil denitrification has been well investigated in the past (Nieder et al., 1989; Aulakh et al., 1992). Various direct and indirect methods for measuring this process are available (Mosier and Klemedtson, 1995). Relatively few data exist for subsurface denitrification in the unsaturated zone (Rice and Rogers, 1993). This might be explained in part by the limitation of available methodology: First, high measuring sensitivities are required because denitrification rates are generally low in subsoils compared with surface soils; second, direct methods, such as the ¹⁵N gas emission (Siegel et al., 1982) or the acetylene inhibition methods (Mosier and Klemedtson, 1995), are either not applicable at all or require major modifications.

Various investigations have demonstrated the potential for denitrifying activity in subsoils, either by measuring potential denitrification (anoxic incubations with NO^-₃ and organic C added) or denitrification capacity (anoxic incubations with only NO^-₃ added) (Parkin and Meisinger, 1989; Yeomans et al., 1992; Sotomayor and Rice, 1996; Richards and Webster, 1999). All of these investigations resulted in the conclusion that the low or nearly undetectable rate of denitrification in the subsoil was governed by low concentrations of organic C.

The fate of subsurface NO^-₃ has been used in various studies as an indirect indicator of denitrification. Schulte-Kellinghaus (1988) investigated NO^-₃ dynamics in the vadose zone of loess soils and found no detectable NO^-₃ disappearance, which is indicative of denitrification except in layers that were high in fossil organic matter. Funk et al. (1996) measured repetitively NO^-₃ and Cl^- profiles in the vadose zone of loess soils and found denitrification rates up to 0.5 kg ha^-1 d^-1 N 33 cm^-1 that were attributed to fossil surface soil horizons containing organic matter and reduced sulfur compounds. The importance of organic C coming from the topsoil was supported by investigations of Rice and Rogers (1990). When investigating three different soils, a silty clay loam, a sandy loam, and a silty loam, they observed that the silty loam soil, which earlier had received applications of manure, exhibited a measurable decline in the NO^-₃:Cl^- ratio down through the soil profile. The selective removal of NO^-₃ indicated the existence of denitrifying activity in the subsoil, and laboratory experiments showed denitrifying potential to a 200-cm depth (Rice and Rogers, 1990).

Further studies used techniques to measure gaseous denitrification products in undisturbed soil samples. Weier et al. (1993a) incubated undisturbed ¹⁵N-labeled soil cores taken between 5- to 110-cm depth from two subtropical clay soils. Significant ¹⁵(N₂ + N₂O) production was detected in each depth and treatment. Production was enhanced by glucose addition, and the main denitrification product was ¹⁵N₂. ¹⁵Nitrogen gas emission and ¹⁵N balance methods also were used to quantify denitrification in undisturbed subsoil monoliths from a fertilized loess soil (Well, 1994). Denitrification was determined to be 5 to 9 kg N ha^-1 y^-1 in the 40- to 70-cm depth. Castle and Arah (1998) incubated ¹⁵N-labeled cores from glacial till subsoils at varying oxygen partial pressures and found denitrification rates varying from 0.2 to 2.5 mg N kg^-1 d^-1 under low O₂ and natural C concentrations. Clough et al. (1999) investigated 100-cm long, intact columns of a grassland soil in the laboratory. To detect subsoil denitrification activity, ¹⁵N-NO^-₃ solution with or without glucose was injected at the 80-cm depth and ¹⁵N was subsequently measured in pore space, in the column headspace, and in the solid phase. A quantitative estimate of denitrification rates was not obtained because the ¹⁵N balance deficit was approximately four times the ¹⁵N recovered in the gas phase.

Weier et al. (1991)(1993b) used an in situ injection technique in order to substantiate subsoil denitrification in the field. Test solutions containing varying combinations of ¹⁵N-NO^-₃, C₂H₂, and glucose were injected through brass probes that were installed in two clay soils between 5 to 10 cm. The application rates varied between 2 and 12 mL of test solution per probe. Soil gas samples were collected from the probes at 6 h to 14 d after applications and were analyzed for gaseous denitrification products. Denitrification rates were calculated assuming constant accumulation of denitrification gases in the pore space surrounding the probes. The increase of gas concentrations following the injections without glucose was low or undetectable, although core incubations with the same soil material had shown significant denitrification activity without glucose addition (Weier et al., 1993a).

Reliable measurement of subsoil denitrification using the above mentioned procedures is difficult. The methods for measuring denitrification potential or capacity are not suitable for quantifying actual rates and indirect methods such as NO^-₃:Cl^- ratios or NO^-₃ disappearance are not sufficiently sensitive for low level rates. Laboratory experiments with undisturbed subsoils are limited by the difficulty of collecting undisturbed samples in deeper layers and maintaining them close to in situ conditions. The reliability of the acetylene inhibition techniques is limited by methodological problems: In soils with low gas diffusivity, the poor performance of the acetylene core techniques was attributed to slow diffusion of acetylene and N₂O (Arah et al., 1991; Bronson et al., 1997; Mahmood et al., 1999); in aerobic soils, the catalytic decomposition of NO in the presence of O₂ and C₂H₂ potentially leads to underestimating of denitrification rates (Bollmann and Conrad, 1997a, b).

The aim of our project was to develop a method for in situ measurement of subsoil denitrification based on the principle of determining ¹⁵(N₂ + N₂O) gas production from denitrification of added ¹⁵NO^-₃.

	MATERIALS AND METHODS

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Principle
The ¹⁵N emission method has been used to determine denitrification in surface soils (Hauck et al., 1958; Rolston et al., 1976; Mosier et al., 1986; Buresh et al., 1991). This method consists of labeling of the soil NO^-₃ pool with ¹⁵NO^-₃ and collecting evolved denitrification gases inside a chamber on top of the soil surface. Denitrification rates are calculated from ¹⁵N gas phase concentrations that accumulate in the headspace. This principle cannot be applied directly to subsurface soil because only a small fraction of dinitrogenous gases evolved from a ¹⁵N-labeled subsoil accumulates, whereas the remainder leaves the production zone by diffusion. A modified version of the ¹⁵N emission method was developed in order to enable the in situ measurement of denitrification in subsurface soils (Nielsen et al., 1997). This method is based on the following assumptions: (i) a NO^-₃ pool of a defined volume of the subsoil can be highly enriched with ¹⁵N, (ii) the denitrification activity in that volume is homogeneously distributed, and (iii) the denitrification activity is constant across time. If gas diffusion were zero, then the production of denitrification gases could be directly determined from the ¹⁵N concentration in the gas phase. This approach is not applicable in unsaturated soils where diffusion is an important transport mechanism for gases. Therefore, in order to determine denitrification activity from the isotopic composition of the gas phase, diffusion must be taken into account. Fick's second law describes the change in concentration of a gas species across time in relationship to its concentration gradient as well as to the apparent gas diffusion coefficient and the air-filled porosity. Using a numerical solution of Fick's second law and adding a production term, gas concentration across time in an unsaturated soil may be described as follows (Anlauf et al., 1990):

¹⁵Nitrogen Labeling
¹⁵N labeling of the NO^-₃ pool of a roughly spherical subsoil volume can be done by applying a ¹⁵NO^-₃ solution via a subsoil well. In order to prevent enhancing denitrification by increasing the NO^-₃ concentration of the soil, the NO^-₃ concentration of the solution has to be adjusted to the NO^-₃ concentration of the natural soil solution. Macropore flow may be minimized by using a low application rate. The geometry of the initial ¹⁵NO^-₃ distribution may be determined by destructive sampling in preliminary experiments or by simulation of the solute transport. Because miscible displacement occurs during solute transport in soils, dilution of injected ¹⁵NO^-₃ will increase with distance from the well. The extent of ¹⁵N labeling is therefore defined by the ¹⁵N atom fraction of the NO^-₃ pool with distance from the well. During long-term measurements, leaching of ¹⁵N-labeled NO^-₃ and dilution from nitrification have to be taken into account. Because soil water content affects both gas diffusivity and denitrification activity, gas sampling for determining denitrification activity may only start after soil water in the ¹⁵N-labeled zone is in equilibrium with the surrounding soil.

Determining Steady-state ¹⁵Nitrogen Concentrations in the Gas Phase
Gas phase samples from the center of the ¹⁵N-labeled soil volume can be taken with a subsoil probe. Because gas samples have to be drawn from a defined location, a probe design that minimizes mass flow of soil gas during the sampling process is necessary. Isotopic composition of the gas samples can be determined by mass spectrometry.

Determination of Soil Physical Parameters
Several methods for measuring the apparent gas diffusion coefficient in situ or in soil cores are available (Sallam et al., 1984; Jellick & Schnabel, 1986). Actual air-filled porosities may be calculated from bulk density and water content.

Determination of Denitrification Activity from Steady-state ¹⁵Nitrogen Concentration in the Gas Phase using Computer Simulation
The computer program CO2.BAS, which simulates CO₂ concentrations in soil across time at a given production rate, was adapted from Anlauf et al. (1990) and modified to simulate concentrations of one or two gas species across time in a spherical production volume surrounded by an infinite nonproducing volume. The option to simulate two gases is necessary because two molecular species, ²⁹N₂ and ³⁰N₂, are formed during denitrification of ¹⁵NO^-₃. Computer simulations were based on Eq. 1 and adapted to spherical geometry, as described previously. The model consisted of 50, 1-cm thick shells. The time step was varied from 1 to 18 s. Model parameters were: N₂ production rate, production volume, air-filled porosity, and apparent diffusion coefficient. The boundary conditions consisted of an outer boundary with ²⁹N₂ and ³⁰N₂ concentrations constantly at natural abundance and an inner boundary (i = 0) that was a spherical shell with radius = 1 cm at the center of the ¹⁵N producing soil volume, with diffusion only in the positive direction.

Production rates were assumed to be uniform in the producing volume. We assumed N₂ was the only denitrification product. This simplification was expected to have minor effects on the results because the N₂O fraction in denitrification gases from inorganic subsoils is generally low (Weier et al., 1993a; Well, 1994; Castle and Arah, 1998). Simulations of ²⁹N₂ and ³⁰N₂ emission included the option to vary production rates for the two N₂ species with distance from the center of the producing volume. This was necessary to simulate the decrease in ¹⁵N atom fraction of the NO^-₃ pool that results from dilution of ¹⁵NO^-₃ with natural abundance soil NO^-₃ during the process of ¹⁵N labeling of subsoils as described above. Simulated isotopic compositions of the gas phase are expressed in terms of d (fraction of denitrified N₂) and a_p (¹⁵N atom fraction of the NO^-₃ pool), calculated according to Mulvaney (1984). Denitrification activity is determined by fitting simulated steady-state concentrations to measured data.

Model Evaluation
Gas Diffusion Experiments
Laboratory experiments were performed in order to evaluate the above described computer simulations. The process of denitrification was simulated experimentally by applying test gases via capillary probes to an unconfined soil volume which allowed unrestricted gas diffusional exchange with the surrounding laboratory air. Within the volume of the soils, 13 probes were installed. One central probe was surrounded by peripheral probes arranged in octahedral geometry of two different distance levels with six probes per level. Two different materials were tested. In one case, an undisturbed, cylindrical soil monolith (28-cm diameter), which was removed from the subsoil (45- to 70-cm depth) of a Woodburn silt loam (fine-silty, mixed, superactive, mesic, Aquultic Argixerolls) was used. Probes consisted of hypodermic needles (15-cm length, 24 gauge) with a rubber septum at the upper end. Evaporation during the experiments was minimized by covering the monolith with a plastic bucket. In the second case, a cubic metal basket of 48.5-cm length was filled with dry medium sand. A cotton sheet was used as a gas permeable barrier holding the sand in the basket. Probes were made of brass tubing (1-mm i.d., 40-cm length) with a plastic Luer stop cock at the upper end. Peripheral probes were installed at 8.1 and 16.2 cm from the central probe. Gas flow rates of the various experiments were chosen to represent denitrification rates that might be expected in subsoils. Steady-state gas concentrations were then measured by sampling the soil atmosphere at varying distances from the application probes. Two different arrangements were tested: (i) gas application through the central probe and gas sampling with surrounding probes, which simulates single-point gas production; (ii) gas application through the six peripheral probes next to the central sampling probe, which simulates homogenous gas production distributed over a spherical volume, similar to the process of soil denitrification.

Nitrous oxide was used as the test gas in most of the experiments because this was the gas that could be detected at highest precision with the available equipment. In the soil monolith experiments, N₂O was applied from a compressed N₂ and N₂O mixture with 152 µL N₂O L^-1. Flow rates were controlled using a needle valve and throttle capillaries, and were relatively high. In the sand basket experiments, pure N₂O was applied at relatively low rates using a syringe pump. Nitrous oxide is not an ideal test gas for evaluating the simulation of ¹⁵N-enriched N₂ diffusion in soil because a single gas cannot be used to directly evaluate the version of the simulation program that simulates the concurrent diffusion of two gas species. Therefore, one experiment was conducted with ¹⁵N-enriched N₂ as a test gas, which was produced as follows: (¹⁵NH₄)₂SO₄ solution (63 atom % ¹⁵N) was added to a glass bottle. The ¹⁵N-labeled NH⁺₄ subsequently oxidized to N₂ by adding NaOBr solution (Warembourg, 1993). The concentration of NH⁺₄-derived N₂ in the final mixture was 546 µL L^-1. The gas was applied to the soil from a 4-L glass bottle by displacing the gas with a slow influx of a barrier solution (saturated Na₂SO₄ solution plus 10% (v/v) concentrated H₂SO₄). The conditions of the gas diffusion experiments are summarized in Table 1. The values of the apparent diffusion coefficient, of the air-filled porosity, and of the adjusted gas flow rate were used in the gas diffusion model to simulate steady-state concentrations. In the variant with gas application to the central probe, the simulations assumed that production occurs only in the center cell of the model. In the case of gas application to the peripheral probes, the simulations assumed that production is restricted to a spherical shell with mean radius equal to the distance between central sampling and the peripheral injection probes. We assumed that this was the best approximation to simulate the six peripheral point sources with our spherical model. The model parameters were measured values with one exception. Because diffusivity in the sand basket was too high to be measured with the method described below, D_a was calculated from water content and air-filled porosity using the empirical approach of Millington and Quirk (1961).

Homogenous ¹⁵N labeling is not achievable in field experiments because miscible displacement during application of the ¹⁵N solution results in dilution of applied ¹⁵NO^-₃ with soil NO^-₃. Therefore, the final distribution of ¹⁵NO^-₃ is not only defined by the amount of ¹⁵N solution applied to the soil, but also depends on the actual conditions of solute transport and on the concentration of NO^-₃ in the ¹⁵N solution compared with the initial soil solution. The response of steady-state d to varying patterns of ¹⁵NO^-₃ distribution was evaluated by assuming the following processes of ¹⁵NO^-₃ solution transport: (i) piston flow, that is, no dilution; or (ii) miscible displacement, that is, ¹⁵NO^-₃ solution added to the soil is diluted by the native soil solution. The fraction of ¹⁵NO^-₃ solution in the final soil solution was assumed to be (i) constant, or to (ii) decrease linearly between a 0- and 20-cm distance from the application point. The ratio N_ia:N_ib was set at 0.2, 0.5, 1, 2, or 5, where N_ia is the NO^-₃ concentration of the ¹⁵NO^-₃ solution and N_ib is the NO^-₃ concentration in the initial soil solution. The calculated ¹⁵NO^-₃ profiles were used to simulate steady-state d.

Laboratory Experiment to Evaluate the Effect of ¹⁵Nitrogen Labeling on Gas Diffusivity
The process of ¹⁵N labeling by applying a solution produces a temporary increase of the water content. If the equilibration of soil water between labeled soil and surrounding soil was incomplete, this would cause reduced apparent gas diffusion coefficients and air-filled porosities in the vicinity of the probes thereby affecting denitrification activity and transport of gaseous denitrification products. A laboratory experiment was conducted in order to determine if incomplete equilibrium of water content after ¹⁵N labeling of the subsoil could result in a lasting reduction of gas diffusion coefficients. The same undisturbed soil monolith that was used earlier during the model evaluation experiments was placed on a filter plate. The soil was maintained at a constant matric potential of -1 kPa at the bottom of the monolith to simulate poor drainage because the suspected reduction of diffusivity was expected to be maximum under these conditions. A needle probe was installed in the center of the monolith. In situ gas diffusion coefficients were measured as described below. Subsequently, 2.5 L of water were injected to the needle probe in order to simulate the ¹⁵N labeling process in the field. After allowing the injected water to drain for a week, gas diffusion coefficients were measured again.

Field Evaluation Experiments
Multipurpose Probe for ¹⁵Nitrogen Labeling, Taking Liquid Samples, and Measuring Soil Matric Potential
We used a multipurpose subsurface probe that had been designed and tested earlier under laboratory conditions (Nielsen et al., 1997). The probe consisted of a glass fiber wick for applying ¹⁵N solution, a coil of silicone tubing for collecting gas samples without causing mass flow in the soil matrix, and a Tensionic (Moutonnet et al., 1993) for measuring matric potential and for sampling dissolved ions.

Experimental setup.
Multipurpose probes were used for ¹⁵N-labeling, sampling the soil atmosphere and the soil solution, and for measuring the soil water potential in the field. Experiments were done at the North Willamette site in a Woodburn silt loam starting in March and June 1995 and at the Lake Creek site in a Holcomb silt loam (fine, smectitic, mesic Typic Argialbolls) starting also in June 1995. Holes 5 cm in diameter and 75 cm deep were drilled with a bucket auger. Probes were inserted to the bottom of the holes. Dry sieved (<1 mm) subsoil previously collected from the same sites was poured into the 0.5-cm gap between the bottom part of the 4-cm OD probes and surrounding soil until the hole was filled to a level 2 cm above the glass fiber wick. After adding 2-cm-thick sealing layer of granular bentonite clay, the remainder of the hole was filled with fresh soil material. Three probes per site were installed. Two days after probe installation, 2.5 L of ¹⁵NO^-₃ solution (K¹⁵NO₃, 99 atom % ¹⁵N, 20 mg L^-1 N) was applied through the glass fiber wick at a rate of 1 mL min^-1. Subsequently, probes were sampled weekly to determine steady-state concentrations of ¹⁵N in the gas phase.

Random locations for in situ measurement of the apparent gas diffusion coefficients were selected between probes but outside the ¹⁵N-labeled soil. Every 2 wk, three measurements were made at a 75-cm depth.

In the spring experiment, the distribution of the ¹⁵NO^-₃ around the probes was determined 5 wk after ¹⁵N labeling by destructive sampling of the soil at discrete distances from the probes (3.5, 5, 7, 10, 12, and 17 cm; eight replicates for each distance).

Evaluation of a Modified Probe Design and Comparison with Reference Methods
Conclusions from multipurpose probe experiments suggested a modified experimental arrangement for further investigations. Consequently, additional experiments included testing a modified, capillary probe design, using reference methods to determine denitrification rates independently from in situ measurements, and comparing concentration profiles measured within the ¹⁵N-labeled soil with simulated values.

Experimental setup.
In February 1996, five multipurpose probes were installed at a 50-cm depth at the Brooks site in a Woodburn silt loam, as described in the previous section. Three capillary probes and three tensiometers were installed at the same depth. The capillary probes consisted of 100-cm-long hypodermic stainless steel tubing (24 gauge) reinforced by 10-mm diameter stainless steel tubing (1-mm wall thickness) with a sharp tip at one end. Guitar wire was inserted to the hypodermic tubing during installation to prevent clogging of the tip with soil. ¹⁵Nitrogen-labeled NO^-₃ solution (2.5 L at 18.2 mg L^-1 NO^-₃–N, 99 atom % ¹⁵N) was applied at a rate of 1 mL min^-1 to the multipurpose and capillary probes. Data were collected approximately every 2 wk until 2 May. Sampling the multipurpose probes was done as described previously. Capillary probes were initially sampled in the same way. After sampling the capillary probes on 2 May, they were pushed to a final depth of 85 cm in 5-cm intervals. A 4-mL gas sample was taken at each depth interval. This was done because precipitation during the experimental period caused downward transport of ¹⁵N-labeled NO^-₃ to an unknown extent. The center of the ¹⁵N-labeled soil was therefore below the 50-cm depth. Probes were lowered in order to determine maximum steady-state concentrations in the profile, which were assumed to occur at the center of the labeled soil.

On 11 May, capillary probes were installed on new locations of the same sites and were subsequently treated with ¹⁵N solution, as described before. On 16 May and 19 June, gas samples were collected from the capillary probes. In addition, gas samples were withdrawn from the soil surrounding the capillary probes in order to determine concentration profiles. This was done in the following manner: an additional capillary probe was installed at 5- and 10-cm distances from the center capillary probes that had been used for ¹⁵N labeling. After collecting a gas sample, the additional probe was removed. The remaining hole was immediately sealed with Bentonite clay in order to minimize disturbance of gas diffusion properties within the ¹⁵N-labeled soil. Four samples were collected horizontally around and one sample vertically beneath the center capillary probes.

On 16 May, peripheral samples were taken from Capillary Probe 1 only. On 19 June, the sample volume was reduced to 0.5 mL in order to minimize mass flow during the sampling process, and peripheral samples were taken from all capillary probes. Gas diffusion coefficients were determined by injecting test gases directly to the capillary probes after gas samples for ¹⁵N analysis were collected and also in additional capillary probes at random locations between the multipurpose and capillary probes.

Reference Methods for Measuring Subsoil Denitrification
Acetylene inhibition soil core method.
We followed the procedure described by Mosier and Klemedtson (1995). Soil cores 4.6-mm i.d. by 150-mm length were collected in perforated PVC pipes using a drop-hammer probe. Access holes were drilled with a 10-cm bucket auger in order to enable core collection from subsurface horizons. Cores were placed in 500-mL PVC incubators that were closed with rubber stoppers. On the same day, samples were transported to the laboratory, where 30 mL of C₂H₂ were injected into the incubators through a rubber septum. Cores where then incubated for a maximum duration of 48 h at temperatures measured in the field at the 50-cm depth. Up to four gas samples were collected at 4- to 12-h intervals and analyzed for N₂O. Cores were collected from the Lake Creek and Brooks sites between January and July 1996.

¹⁵Nitrogen emission monolith method.
On 14 February 1996, undisturbed soil monoliths were collected at the Brooks site. An excavation procedure was used to fit monoliths from the 45- to 55-cm depth inside PVC rings of 205-mm i.d. by 105-mm height. In the laboratory the monoliths were installed in the experimental arrangement that enabled ¹⁵N labeling of the soil, controlling soil matric potential and temperature, and collecting evolved denitrification gases, as described by Well (1994). Monoliths were placed on suction plates consisting of a confined layer of silica flour (30 to 125 µm). ¹⁵Nitrogen solution (18 mg L^-1 NO^-₃–N, 63 atom % ¹⁵N, 1 mg CaSO₄ L^-1) was applied to monolith surfaces at a rate of 1 mL min^-1. The solution was distributed through a manifold of six hypodermic needles that dripped on a layer of glass fiber wick on top of the surface. Plastic covers were placed on top of the PVC rings and were sealed with rubber sleeves to collect denitrification gases. Covers were kept closed long enough to accumulate ¹⁵N gas concentration sufficient for isotopic analysis. Matric potential was controlled by the level of the drainage reservoir and was kept at -6.4 or -1 kPa in two subsequent experiments.

Analytical Methods
Apparent Diffusion Coefficient
An in situ method was used to determine D_a (Jellick and Schnabel, 1986). Capillary probes, similar to those used in the modified in situ dentrification experiments, were pushed into the subsoil during the field experiments. Either O₂ or He (50 mL) was then injected instantly to the subsoil with a plastic syringe. At 3, 4, and 5 min after injection, 4-mL gas samples were collected from the probe, transferred to Vacutainers (Becton Dickinson, Rutherford, NJ), and subsequently analyzed for O₂ or He. Oxygen was measured using a gas chromatograph (Carle Series 100, AGC, Loveland, CO) equipped with a thermal conductivity detector and a molecular sieve 5A column kept at 85°C. The carrier gas was He. For He analysis, a gas chromatograph (model 3700, Varian Assoc., Inc., Walnut Creek, CA) equipped with a ⁶³Ni electron capture detector and a Porapaq Q column kept at 35°C was used. The carrier gas was 95% Ar:5% CH₄ (v/v). Apparent diffusion coefficients were then determined using the gas diffusion model described above. In situ gas diffusion coefficients were evaluated with a chamber method for measuring gas diffusion of undisturbed soil cores (Frede, 1986). Soil cores (50-mm i.d. by 40-mm height) were removed from the North Willamette site at the 60- to 65-cm depth. Core values for D_a were 0.0014 ± 0.0008 cm² s^-1. In situ measurements at the same site and same date resulted in D_a = 0.0017 ± 0.0004 cm² s^-1. Thus, data obtained with the in situ method are reliable. All of the diffusion coefficients determined during the laboratory and field experiments were measured with the in situ method.

Air-Filled Porosity
Bulk density was measured with the core method (Blake and Hartge, 1986). Air-filled porosity was then calculated using water content measurements.

Analysis of Soil NO^-₃ and ¹⁵Nitrogen Atom Fraction of the NO^-₃ Pool
Soil NO^-₃ was determined by extracting 20 g wet soil in 50 mL 0.01 M CaCl₂ for 1 h and subsequent NO^-₃ analysis of the extract using an ion chromatograph (model 2000i, Dionex Corp., Sunnyvale, CA). The isotopic composition of NO^-₃ in the soil was determined by adding 0.25 mL of a denitrifying culture (Pseudomonas aeruginosa) to 2 mL of soil extract in a 4-mL Vacutainer, flushing the Vacutainer with He, incubating for 3 to 10 d, and analyzing the headspace for the isotopic composition of N₂ plus N₂O (Risgaard-Petersen et al., 1991).

¹⁵Nitrogen Gas Analysis
Gas samples were collected using 1-mL or 5-mL glass syringes. Usually a 4-mL sample was injected into a 3-mL Vacutainer for storage. When smaller sample volumes were desired, 0.5-mL samples were injected into He-flushed glass vials (11 mL).

The isotopic composition of the N₂ plus N₂O pool was determined with two different mass spectrometers. Until December 1995 we used a Roboprep C/N solid sample analyzer in line with a Tracermass isotope ratio spectrometer (Europa Scientific, Crewe, UK). Gas samples were injected into the carrier gas [He, purity >99.999% (v/v)] through a gas injection port mounted between the combustion unit and the reduction column of the Roboprep unit. After December 1995 we used an ANCA-TG trace gas sampler in line with a 20/20 isotope ratio mass spectrometer (Europa Scientific, Crewe, UK), as described by Stevens et al. (1993). Sensitivities for m/z 29 and m/z 30 were 4.62 x 10^-6 and 1.21 x 10^-6 for the Tracermass and 6.99 x 10^-7 and 2.97 x 10^-6 for the 20/20 mass spectrometers.

	RESULTS AND DISCUSSION

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Model Evaluation
Gas Diffusion Experiments
Relative differences between measured and simulated steady-state concentrations (Table 1) were highest (30–39%) for peripheral probes that were most distant to the injection probe, whereas the relative differences of the proximate and center probes were smaller (7–23%). Some difference between measured and simulated values might be expected because the geometry of the diffusion experiments (cylindrical or cubical boundaries) was different from the computer simulation (spherical boundary), and any discrepancy due to differences in geometry would be most important for the most distant probes, which were closest to the boundaries. Nevertheless, the differences of the comparisons with replicate measurements were not significant (P > 0.95), except for Experiment 2b at 16.2-cm distance. The variability of the measured values was generally high, however, possibly because of anisotropic diffusion or imperfect positioning of the probes.

Two potential concerns about the adequacy of the gas diffusion experiments are (i) the high water solubility of N₂O, the gas used in most of the experiments; and (ii) the potential confounding effect of convection at the flow rates we used. Because N₂O is highly soluble in water, soil water content affects the N₂O concentration in the air-filled pore space. However, when a state of equilibrium exists between gas and liquid phase, soil water content only affects the storage capacity of the total pore space. Simulations indicated that storage capacity did not affect steady-state concentrations (data not shown); consequently, N₂O dissolution in soil water should not affect the interpretation of these diffusion experiments. This was also confirmed by the good agreement between measured and simulated values when we used a relatively insoluble gas in Experiment 1c.

The flow rates in the soil monolith experiments were relatively high, and thus convective flow might have been a confounding factor. We estimated the relative significance of convection compared with molecular diffusion by computing the Peclet numbers for the center-injection experiments. The Peclet number is a dimensionless parameter that expresses the relative importance of convection and diffusion: Peclet numbers much <1 are indicative of diffusion-dominated transport. For example, the Peclet number for Experiment 1c was 0.01, which suggested that convection was negligible and diffusion was the dominant transport process in the soil monolith experiments. This was further confirmed in the sand basket experiments that were done at much lower flow rates (Peclet numbers for Experiments 2a and 2b were 0.00002 and 0.0002, respectively).

Because most of the measured N₂O concentrations of experiments fitted the simulated data satisfactorily (Table 1), we concluded that the gas diffusion model was suitable for determining gas production rates from steady-state concentrations. The accuracy of this approach is limited, however, by heterogeneity of gas diffusivity and the difficulty in measuring representative values for this parameter.

Effect of Model Parameters on Steady-State Concentrations
Apparent diffusion coefficient and denitrification activity. Denitrification rates were simulated as a gas volume production per soil volume. At room temperature, 1 µL L^-1 h^-1 N₂ is equal to 28 g ha^-1 d^-1 N to 10 cm. Figure 1 shows the simulated time course of gas concentrations following the beginning of the gas production at D_a = 0.001 cm² s^-1. Concentrations became approximately constant within 2 d, indicating that steady-state concentrations in field experiments can be expected to be attained rapidly. This also clearly demonstrates that, because of diffusional gas transport out of the production zone, the initial period of almost linearly increasing concentrations is relatively short. Consequently, it would be difficult to determine production rates from initial accumulation rates.

View larger version (20K): [in this window] [in a new window]   Fig. 1. Simulation of the fraction of denitrified N2 (d) across time at various distances from the center of the labeled soil volume assuming the following model parameters: Radius of 15N-labeled soil = 14 cm; homogeneous 15N distribution in labeled soil volume with 15N atom fraction of the NO-3 pool = 0.5; Da = 0.001 cm2 s-1; and denitrification rate = 1 µL N2 L-1 soil.

View larger version (17K): [in this window] [in a new window]   Fig. 2. Simulated steady-state fraction of denitrified N2 (d) as a function of denitrification rate and Da. Radius of 15N-labeled soil = 14 cm; homogeneous 15N distribution in labeled soil volume with 15N atom fraction of the NO-3 pool = 0.5.

View larger version (13K): [in this window] [in a new window]   Fig. 3. Simulated steady-state of denitrified N2 (d) as a function of the radius of 15N-labeled soil. Homogeneous 15N distribution in labeled soil volume with 15N atom fraction of the NO-3 pool = 0.5; Da = 0.001 cm2 s-1; and denitrification rate = 1 µL N2 L-1 soil.

View larger version (30K): [in this window] [in a new window]   Fig. 4. Calculated profiles of the 15N atom fraction of the NO-3 pool (ap).

Conclusions from the Model Evaluations
The laboratory experiments confirmed that subsoil gas production can be determined by simulations based on measured values of steady-state concentrations and gas diffusivity. Steady-state ¹⁵N-gas concentrations in pore space can be expected to build up rapidly after the equilibration of soil water following the injection of a ¹⁵N test solution, which was shown to be complete within 1 wk. The computer simulations of diffusive gas transport have demonstrated that in the pore space of an unconfined, roughly spherical ¹⁵N labeled subsoil volume of 5 L, a measurable steady-state concentration of ¹⁵N gases is to be expected if the denitrifying activity is at a level >5 g ha^-1 d^-1 N to 10 cm, which is often the case in subsoils. These findings also indicate that earlier ¹⁵N and C₂H₂-injection techniques using application rates of only a few milliliters of test solution and assuming accumulation of denitrification gases within the amended soil volume (Weier et al., 1991, 1993b) were not adequate for quantifying denitrifying activity because (i) diffusive transport was not taken into account, and (ii) because denitrifying activity in the relatively small amended soil volumes was probably not high enough to produce measurable steady-state concentrations.

Multipurpose Probe Evaluation
Measured ¹⁵N atom fractions of the a_p around the subsurface probes at the end of the spring experiment are given in Fig. 5 . Data were fitted to a first-order exponential function:

were R is the distance from the probe center in centimeters. Results of apparent diffusion coefficients, air-filled porosities, steady-state fractions of denitrified N₂ in the gas phase, as well as denitrification rates are given in Table 4. Diffusion coefficients were constant during spring but increased slightly during the summer experiments.

View larger version (15K): [in this window] [in a new window]   Fig. 5. Measured and fitted values of the 15N atom fraction of the NO-3 pool (ap) at the North Willamette site. The regression equation, ap = 0.695e-0.145xR, was significant (R = 0.60, P < 0.0001).

View this table: [in this window] [in a new window]   Table 4. Denitrification rates at the North Willamette and Lake Creek sites as determined with the 15N in situ method. Depth of multipurpose probes = 75 cm.

We suspected that the probe installation procedure was a potential source of error because the filling of gaps around multipurpose probes with dry sieved soil during installation might disturb various soil processes. Lack of aggregation in this material might result in a deficit of continuous air-filled pores at low matric potentials observed during the experiments. Low gas diffusivity would then result in elevated denitrification activity and inhibit transport of gaseous denitrification products. Further experiments with a modified probe design and installation procedure are necessary to obtain more reliable measurements. The accuracy of a modified procedure may be evaluated by using reference methods of denitrification measurement. An additional verification of simulated denitrification rates may be obtained by measuring concentration profiles within the ¹⁵N-labeled soil. If the simulated rates were correct, then simulated concentration profiles should be in agreement with measured data.

Modified Capillary Probe Design
Values of D_a shown in Table 5 are averages for the main capillary probes of the ¹⁵N experiments. We used additional capillary probes along with the main capillary probes to measure D_a, and found that the main capillary probe averages were not lower than total averages (data not shown). This proved that the ¹⁵N labeling procedure did not affect gas diffusion properties during the sampling period and was consistent with the findings in the monolith laboratory experiment.

Simulated denitrification rates in the capillary probes were generally low compared with multipurpose probe values. In Exp. 1, two of the capillary probes exhibited elevated rates on the first date after ¹⁵N labeling was complete. Otherwise, rates ranged between 0 (below detection) and 7 g ha^-1 d^-1 N to 10 cm. On 2 May, detectable steady-state concentrations were observed at the 60- to 75- and 50- to 65-cm depths in probes C1 and C2, respectively. Maximum concentrations were at the 65-cm depth in both probes.

During Exp. 2, all ¹⁵N values of the samples collected at the 5- and 10-cm distance from the capillary probes were below detection. Sampling of the capillary probes was replicated 2 h after the first sampling of each date. Data from the second sampling were always below detection. On 16 May, when the gas sample volume was 4 mL, only Probe C1 showed a measurable signal in the first replicate. The D_a in this probe was the lowest among the six measured values on that date. There was also a considerable physical resistance during the injection of the test gas which was not experienced in the other probes. This might indicate that Probe C1 was located within a zone of low gas diffusivity compared with the average, perhaps within an aggregate.

Denitrification rates were higher on 19 June, when 0.5-mL gas samples were collected. This might be attributed to a lower dilution of samples from mass flow during the sampling process.

Reference Methods
Denitrification rates measured with the acetylene method decreased with depth when more than one depth interval was sampled (Table 6). Saturated conditions resulted in highest rates. Results from the Brooks site during unsaturated conditions ranged between 0.2 and 8.1 g ha^-1 d^-1 N to 10 cm, and were thus in the same order of magnitude as simulated rates based on ¹⁵N values at the capillary probes from the same site.

A verification of simulated steady-state concentration profiles was not possible because the accuracy of isotope analysis was insufficient at this level of denitrification activity. The high denitrification value on 19 June and the information about gas diffusivity at Probe C1 demonstrate that soil heterogeneity with respect to gas diffusion might be a problem of the in situ method. An experimental design that enables sampling of a larger soil volume might overcome this issue. Furthermore, it would be desirable to evaluate potential effects from heterogeneous gas diffusion properties. However, this was not feasible with the one-dimensional, spherical model used in this study because it assumes constancy within the spherical shells. It would be necessary to adapt a three-dimensional diffusion model in order to simulate arbitrary patterns of heterogeneity.

The effects observed from reducing the gas sampling volume as well as collecting replicate samples in the capillary probes demonstrates the importance of a sampling procedure that avoids mass flow.

The relative agreement with the reference methods only demonstrates that the in situ method is within the range of expected values, but it is certainly not sufficient to prove its accuracy. Given the high variation of the results from all methods, more replicates would be necessary to prove accuracy at a meaningful level of precision. Furthermore, most of the in situ rates were below detection. Finally, the acetylene inhibition and ¹⁵N emission monolith are also subject to various sources of error. To obtain a conclusive proof of accuracy for the in situ method, experimental conditions would have to warrant sufficient accuracy for both the in situ and the reference methods. This might be achieved with a large volume of homogenous substrate with denitrification activity high enough to allow the use of indirect reference methods such as the ¹⁵N balance and the NO^-₃:Br^- or NO^-₃:Cl^- methods.

	CONCLUSIONS

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
These experiments indicate that the in situ ¹⁵N emission method is, in principle, suitable for measuring subsoil denitrification. The results of the field experiments were not sufficient to prove the accuracy of this method, however. Further evaluation is desirable under conditions providing greater reliability of results. This could be achieved by increasing the volume of ¹⁵N-labeled soil or the sensitivity of isotope analysis, or by selecting sites with higher denitrification activity. The problem of ¹⁵NO^-₃ leaching during the experiments could be overcome by choosing a cylindrical instead of a spherical geometry of the ¹⁵N-labeled soil. This would require a three-dimensional gas diffusion model for simulating denitrification activity, however, as well as the use of multiple probes for ¹⁵N labeling and gas sampling.

Because the presented procedure requires isotope analysis of gas samples and simulating of gas diffusion in soil, it is more demanding than the existing alternative methods; however, several advantages make it worthwhile to pursue its further development. To our knowledge, an alternative technique allowing quantitative in situ denitrification measurement in unsaturated subsoils without excavating any soil material does not exist. Because it does not require collecting soil material, it is advantageous for measuring denitrification in the deeper vadose zone where difficulty of soil sampling is limiting the application of core or monolith methods. Therefore, it is also convenient for small experimental plots that would be destroyed by repetitive subsoil sampling. Finally, the use of the ¹⁵N technique avoids the potential errors of the acetylene inhibition techniques.

	ACKNOWLEDGMENTS

 
We thank the following individuals and organizations for access to field sites: Steve Griffith and the USDA-ARS National Forage Seed Production Research Lab for the Lake Creek site; the OSU North Willamette Research and Extension Center for the North Willamette site; and Mark Steele and Norpac Foods, Inc. for the Brooks site. We thank Nancy Ritchie, James Tang, and Stephanie Korschun for assistance with laboratory and field work, and Maria Dragila for help with Peclet number calculations. This research was supported by USDA Special Water Quality grant 93-34214-8900 to D. Myrold and by a postdoctoral fellowship to R. Well from the Alexander von Humboldt Foundation.

	NOTES

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Contribution from the Oregon Agric. Expt. Stn., Technical paper no. 11823.

Received for publication March 20, 2000.

	REFERENCES

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 

• Anlauf, R., K.C. Kersebaum, L.Y. Ping, A. Nuske-Schüler, J. Richter, G. Springob, K.M. Syring, and J. Utermann. 1990. Models for processes in the soil. Catena, Cremlingen, Germany.
• Arah, J.R.M., K.A. Smith, I.J. Crichton, and H.S. Li. 1991. Nitrous oxide production and denitrification in Scottish arable soils. Soil Sci. 42:351–367.
• Aulakh, M.S., J.W. Doran, and A.R. Mosier. 1992. Soil denitrification—significance, measurement and effects of management. Adv. Soil Sci. 18:1–57.
• Blake, G.R., and K.H. Hartge. 1986. Bulk density. p. 363–375. In A. Klute (ed.) Methods of soil analysis. Part 1—Physical and mineralogical methods. Agron. Monogr. 9. 2nd ed. ASA and SSSA, Madison, WI.
• Bollmann, A., and R. Conrad. 1997a. Acetylene blockage technique leads to underestimation of denitrification rates in oxic soils due to scavenging of intermediate nitric oxide. Soil Biol. Biochem. 29:1067–1077.
• Bollmann, A., and R. Conrad. 1997b. Enhancement by acetylene of the decomposition of nitric oxide in soil. Soil Biol. Biochem. 29:1057–1066.
• Bronson, K.F., H.-U. Neue, U. Singh, and E.B. Abao. 1997. Automated chamber measurements of methane and nitrous oxide flux in a flooded rice soil: I. Residue, nitrogen, and water management. Soil Sci. Soc. Am. J. 61:981–987.[Abstract/Free Full Text]
• Buresh, R.J., S.K. De Datta, M.I. Samson, S. Phongpan, P. Snitwongse, A.M. Fagi, and R. Tejasarwana. 1991. Dinitrogen and nitrous oxide flux from urea basally applied to puddled rice soils. Soil Sci. Soc. Am J. 55:268–273.[Abstract/Free Full Text]
• Castle, K., and J.R.M. Arah. 1998. Denitrification in intact subsoil cores. Biol. Fertil. Soils 28:12–18.
• Clough, T.J., S.C. Jarvis, E.R. Dixon, R.J. Stevens, R.J. Laughlin, and D.J. Hatch. 1999. Carbon induced subsoil denitrification of ¹⁵N labeled nitrate in 1 m deep soil columns. Soil Biol. Biochem. 31:31–41.
• Frede, G. 1986. Der Gasaustausch des Bodens. (In German.) Göttinger Bodenkundl. Ber. 87:1–130.
• Funk, R., F.X. Maidl, and F. Fischbeck. 1996. Denitrification in deeper soil layers of arable land on loess in southern Germany. Z. Pflanzenernaehr. Bodenkd. 159:207–213.
• Geyer, D.J., C.K. Keller, J.L. Smith, and D.L. Johnstone. 1992. Subsurface fate of nitrate as a function of depth and landscape position in Missouri Flat Creek watershed, U.S.A. J. Contam. Hydrol. 11:127–147.
• Hauck, R.D., S.W. Melsted, and P.E. Yankwich. 1958. Use of N-isotope distribution in nitrogen gas in the study of denitrification. Soil Sci. 86:287–291.
• Jellick, G.J., and R.R. Schnabel. 1986. Evaluation of a field method for determining the gas diffusion coefficient in soils. Soil Sci. Soc. Am. J. 50:18–23.[Abstract/Free Full Text]
• Mahmood, T., R. Ali, F. Azam, and K.A. Malik. 1999. Comparison of two versions of the acetylene inhibition/soil core method for measuring denitrification loss from an irrigated wheat field. Biol. Fertil. Soils 29:328–331.
• Millington, R.J., and J.M. Quirk. 1961. Permeability of porous solids. Trans. Faraday Soc. 57:1200–1207.
• Mosier, A.R., W.D. Guenzi, and E.E. Schweizer. 1986. Soil losses of dinitrogen and nitrous oxide from irrigated crops in northeastern Colorado. Soil Sci. Soc. Am. J. 50:344–348.[Abstract/Free Full Text]
• Mosier, A.R., and L. Klemedtson. 1995. Measuring denitrification in the field. p. 1047–1062. In R.W. Weaver, J.S. Angle, and P.J. Bottomley (ed.) Methods of soil analysis. Part 2—Microbiological and biochemical properties. SSSA Book Ser. 5. SSSA, Madison, WI.
• Moutonnet, P., J.F. Pagenel, and J.C. Fardeau. 1993. Simultaneous field measurement of nitrate-nitrogen and matric pressure head. Soil Sci. Soc. Am. J. 57:1458–1462.[Abstract/Free Full Text]
• Mulvaney, R.L. 1984. Determination of ¹⁵N-labeled dinitrogen and nitrous oxide with a triple collector mass spectrometer. Soil Sci. Soc. Am. J. 48:690–692.[Abstract/Free Full Text]
• Nieder, R., G. Schollmeyer, and J. Richter. 1989. Denitrification in the rooting zone of cropped soil with regard to methodology and climate, a review. Biol. Fertil. Soils 8:219–226.
• Nielsen, T.H., D.D. Myrold, and R. Well. 1997. Combination probe for ¹⁵N labeling and measurement of ¹⁵N atom fraction, water potential, and denitrification activity in subsoil. Soil Sci. Soc. Am. J. 61:802–811.[Abstract/Free Full Text]
• Parkin, T.B., and J.J. Meisinger. 1989. Denitrification below the crop rooting zone as influenced by surface tillage. J. Environ. Qual. 18:12–16.
• Rice, C.W., and K.L. Rogers. 1990. Denitrification in the vadose zone: A mechanism for nitrate removal No. 281. Kansas Water Resources Research Inst., Manhattan, KS.
• Rice, C.W., and K.L. Rogers. 1993. Denitrification in subsurface environments: Potential source for atmospheric nitrous oxide. p. 121–132. In D.E. Rolston et al. (ed.) Agricultural ecosystem effects on trace gases and global climate change. ASA Spec. Publ. no. 55. ASA, CSSA, and SSSA, Madison, WI.
• Richards, J.E., and C.P. Webster. 1999. Denitrification in the subsoil of the Broadbalk Continuous Wheat Experiment. Soil Biol. Biochem. 31:747–755.
• Risgaard-Petersen, N., S. Rysgaard, and N.P. Revsbech. 1991. A sensitive assay for determination of