In Situ Quantification of Biological N2 Production Using Naturally Occurring 15N15N.

We describe an approach for determining biological N2 production in soils based on the proportions of naturally occurring 15N15N in N2. Laboratory incubation experiments reveal that biological N2 production, whether by denitrification or anaerobic ammonia oxidation, yields proportions of 15N15N in N2 that are within 1‰ of that predicted for a random distribution of 15N and 14N atoms. This relatively invariant isotopic signature contrasts with that of the atmosphere, which has 15N15N proportions in excess of the random distribution by 19.1 ± 0.1‰. Depth profiles of gases in agricultural soils from the Kellogg Biological Station Long-Term Ecological Research site show biological N2 accumulation that accounts for up to 1.6% of the soil N2. One-dimensional reaction-diffusion modeling of these soil profiles suggests that subsurface N2 pulses leading to surface emission rates as low as 0.3 mmol N2 m-2 d-1 can be detected with current analytical precision, decoupled from N2O production.

Introduction

Biological N2 production constitutes the main mechanism through which fixed nitrogen is returned to the atmosphere. While many methods have been developed for measuring N2 production in the field, obtaining accurate estimates of ecosystem fixed-nitrogen loss remains a challenge.[1,2] Field-based techniques often require nutrient amendments (e.g.,[15]N-labeled nitrate), manipulation of biochemical pathways (e.g., C2H2 inhibition of nitrous oxide reductase),[3] or sampling and incubation of soil cores, all of which introduce poorly constrained uncertainties.[4,5] For example, in nutrient amendment studies, the fraction of extant nitrogen substrate utilized must be accounted for, but it is often difficult to constrain. Moreover, biological N2 production can be stimulated by substrate addition, biasing measurements based on this approach. Soil-core incubations to evaluate N2 production may not require nutrient amendments, but instead require that the extant gases be replaced with a gas mixture to reduce or replace the ambient N2 background.[6] Ultimately, this suite of methods for quantifying N2 production rates can only probe short-term and potential rates of denitrification and other nitrogen-loss processes. Importantly, they may not integrate variation in activity that occurs over longer time scales at a given sampling site. Passive in situ measurements are rare, and fraught with a different set of complications: a recent attempt to use N2/Ar ratios to probe excess N2 production in situ found that physical fractionation of gases, combined with insufficient sensitivity, would likely preclude its widespread application.[7]

Stable isotopes of nitrogen at natural abundance levels could in principle be used to determine the amount of biologically produced N2 in soil gases as well. Variations in the 15N/14N ratio of N2, reported as a δ-value in per mil (‰) relative to atmospheric N2,can be caused by variability in the chemistry of N2 cycling, substrate δ15N, and physical transport. Nevertheless, a large isotopic contrast may exist between biological and atmospheric N2: strong isotopic fractionation for N2-yielding processes[8−10] can result in local deviations in the δ15N value of N2 relative to their substrates and the atmospheric background. However, closed-system and rate-dependent effects on isotopic fractionation,[11] the broad distribution of substrate δ15N values,[12] and physical fractionation affecting the elemental and isotopic composition of soil gases[13] are rarely well-characterized, rendering the interpretation of bulk δ15N values in soil N2 nonunique; disentangling the variations in δ15N of soil-N2 may not be possible without additional constraints.

We recently developed methods to measure 15N15N in N2 with high precision at natural abundances, which offers a new approach to quantifying in N2 production on local and global scales.[14] Together with 14N15N/14N14N ratios, measurements of the 15N15N/14N14N ratio in N2 yield a “clumped” isotope tracer, Δ30, which is defined below and also reported in per mil:

Unlike the δ15N value, Δ30 represents the proportional (rather than absolute) enrichment in 15N15N, quantified relative to a random distribution of 15N and 14N atoms in N2 molecules. The δ15N value of the substrate does not affect the Δ30 signature of a N2-yielding process because the Δ30 value is normalized against the bulk 15N/14N ratio (eqs and 5). Moreover, the Δ30 tracer is insensitive to physical fractionation and nitrogen fixation;[14,15] these processes tend to preserve proportions of 15N15N relative to 14N15N and 14N14N. Furthermore, the Δ30 values of the biological N2 thus far identified cluster near zero, while the Δ30 value of atmospheric N2 is 19.1 ± 0.1‰—a signature of upper-atmospheric gas-phase reactions.[14] It results in a large isotopic contrast between biological and atmospheric N2. Local subatmospheric Δ30 values in soils thus may reflect the presence of biological N2, which can be quantified through a clumped-isotope mass balance if the Δ30 signatures of different N2-producing pathways are sufficiently similar. Δ30 values may trace biological N2 production in situ using the same principles first laid out by Hauck and co-workers,[16,17] but without the need for nutrient amendments or isotopic labels.

Motivated by this potential application, we conducted a broader survey of Δ30 values from biological processes. Specifically, we expanded our earlier characterization of Δ30 values from denitrifying bacteria[14] with new measurements of Δ30 signatures from anaerobic ammonia-oxidizing (anammox) bacteria and incubations of natural soils. The narrow distribution of biological Δ30 signatures that we find suggests that Δ30 values can indeed be used to quantify biological N2 production in soils, and possibly also other restricted environments. As a proof-of-principle application, we present two soil-gas depth profiles that show evidence for biological N2 production, and evaluate the sensitivity of the approach.

Experimental Methods

Isotopic analyses were performed on the ultrahigh resolution Nu Instruments Panorama mass spectrometer at the University of California, Los Angeles according to methods described previously.[14,18] The uniquely high resolution of the instrument allows the simultaneous measurement of 14N15N+/14N14N+ and 15N15N+/14N14N+ ratios at m/z = 29 and 30, with near-baseline resolution of 15N15N+ from 14N16O+ and 12C18O at m/z = 30. N2 gas samples (20–50 μmol) were isolated from experimental headspace and soil-derived gases using cryogenic purification on a high-vacuum sample preparation line followed by gas chromatographic separation from O2 and Ar before isotopic analysis. Cryogenic purification removes condensable gases (e.g., CO2 and some hydrocarbons) and was accomplished by passing the gas through a stainless-steel U-trap submerged in liquid nitrogen (−196 °C). The gas was then condensed onto silica gel pellets at −196 °C within the sample-injection loop of the gas-chromatographic system. N2 gas was separated from O2 and Ar using a molecular sieve 5A column (3 m × 1/8” OD) followed by a HayeSep D column (2 m × 1/8 in. OD) inline, all with a 20 mL min–1 He flow rate at 25 °C. The sample gases, air, and high-temperature standards of N2 (which were heated at 800 °C for 24–48 h over strontium nitride) were purified the same way and analyzed during the same analytical sessions. Analytical precision for replicate air samples during these sessions was ±0.006‰ for δ15N and ±0.08‰ for Δ30.

To determine the Δ30 signatures of N2 produced by anammox bacteria, headspace outflows from several anammox bioreactors at Radboud University were sampled. Outflows from bioreactors containing enrichment cultures of the genera Candidatus Kuenenia,[19] and Ca. Brocadia(20) (both freshwater genera), as well as Ca. Scalindua(21) (a marine genus) were sampled using a 8 mL sampling loop made of a 1/4 in. OD stainless steel tube. The gas mixture was transferred cryogenically to a pre-evacuated sample finger filled with silica gel at −196 °C for 15 min before flame-sealing. All enrichment cultures at Radboud University were grown on the same NH4SO4 + NaNO2 substrates, which had δ15N values of −0.5 ± 0.3‰ and −26.2 ± 0.3‰, respectively. Atmospheric contamination was monitored using gas chromatography–mass spectrometry of the outflow, using O2 (m/z = 32) as a proxy. A correction for air-N2 contamination in the bioreactor headspace was calculated from the O2 signal and a proportionality coefficient determined through a series of volumetrically calibrated mixtures of air in the 95% Ar/5% CO2 mixture used to flush the bioreactors. Measured air contamination varied between bioreactors, ranging from 0.6% for Kuenenia to 12.3% for Scalindua outflows, as a result of variable anammox activity compared to the flushing flow rate. After correction for background contamination (0.12–2.40‰ for δ15N and 0.1–2.3‰ for Δ30), duplicate collections showed reproducibility in δ15N and Δ30 within ±0.01‰ and ±0.3‰, respectively.

Incubations of natural soils were performed to determine the Δ30 signatures of N2 produced by natural biological communities. Soils from three agricultural treatments at the Kellogg Biological Station (KBS) Long-Term Ecological Research site were used for these experiments. Soils at the site belong to the Kalamazoo series, which are fine-loamy, mixed mesic Typic Hapludafs.[22] Soils T1 and T2 are agricultural soils that have been under an annual corn–soybean–winter wheat rotation since 1989, with T1 conventionally tilled with a chisel plow and T2 being no-till. Soil T7 comes from a native early successional old field community (containing grasses, shrubs, and trees) that was established in 1989 and has been maintained by an annual spring burn since 1997. Incubations of 25-g soil samples were conducted in 125 mL glass serum bottles that were crimp-sealed using butyl rubber stoppers (Geomicrobial Technologies, Inc., Ochelata, OK, U.S.A.). Initially, after saturating the dried soils, an anaerobic headspace was created by sparging with He. The soils were then allowed to denitrify for 7–10 d to remove any initial oxidized N. At that point, the headspace was sparged again with He and then inoculated with glucose (0.3 mL, 1 M) and NaNO3 substrate (1 mL, 0.3 M; δ15N = 5.4‰). Production of N2 was allowed to proceed for 96 h to ensure collection of sufficient N2 gas for isotopic analysis. Gases were transferred cryogenically to a pre-evacuated silica-gel finger and flame-sealed prior to analysis at UCLA.

For the in situ study, soil gas samples from the KBS Interactions Experiment site were obtained from a monolith soil lysimeter. The lysimeter is located 5 m from the edge of Plot 13 (27 × 40 m total width), which had followed an annual corn–soybean–winter wheat rotation (conventional tillage, no fertilizer) until spring 2016, when planting was changed to Cave-in-rock switchgrass (Panicum virgatum L.). Constructed of stainless steel, the 2.29 × 1.22 × 2.03 m (L × W × D) monolith lysimeter was installed with a minimum of disturbance to the soil column approximately 5 cm above the soil surface in 1986 as described in Brown et al.[23] Gas sampling lines (stainless steel, 1.6 mm OD, 0.5 mm ID) were previously installed through the walls of the lysimeter and extend 30 cm outward. Each line was purged by removing 3 mL of soil gas (∼50 times the line volume) by gastight syringe and discarding the gas. Subsequently, 5 mL of gas for each sample was collected by gastight syringe and pushed through a 3 mL stainless-steel sampling bottle that had been previously purged with He gas. Gas samples were collected on 10/11/17 and 7/18/18 at depths of 24, 34, 50, 59, 77, 86, and 170 cm from the soil surface. On return to the laboratory, gases were cryogenically purified and transferred to a pre-evacuated silica-gel finger and flame-sealed.

Results and Discussion

Δ30 Values from Biological N2 Production Are Near Zero

Anammox enrichment cultures produced N2 with Δ30 values close to, but slightly different from the stochastic distribution of isotopes (Table ). Nitrogen gas produced by the two freshwater genera are characterized by Δ30 < 0 (i.e., N2 was “anticlumped”), while N2 produced by the marine Ca. Scalindua enrichment had Δ30 = 1.0 ± 0.3‰, indistinguishable from an equilibrium distribution of 15N isotopes at culturing temperatures (i.e., 1.0‰ at 35 °C). A positive correlation between Δ30 and δ15N values was observed when all anammox culture data are considered together (R = 0.86, p = 0.0009). The origins of this correlation were not investigated, but deserve further scrutiny: the apparent difference in Δ30 value between freshwater and marine species may point to a different biochemistry related to the gene organization and subsequent expression of hydrazine synthase enzyme.[24,25] In any case, the Δ30 values for N2 produced by freshwater anammox genera are close to that expected from combinatorial isotope effects:[26] the contrast in isotopic compositions between the NaNO2 (δ15N = −26.2‰) and NH4SO4 (δ15N = −0.5‰) substrates, by itself, would yield Δ30 = −0.2‰, close to the mean measured values of −0.2 ± 0.1‰ and −0.5 ± 0.3‰ (1σ) for Ca. Kuenenia and Ca. Brocadia, respectively. Isotopic fractionation during biological uptake[10] may cause additional variability in the δ15N value of the assimilated substrates, but the Δ30 value of the N2 produced is not expected to deviate more than ∼1‰ from zero because the combinatorial effect is a relatively weak function of the substrate δ15N contrast.[26]

Table 1

Clumped-Isotope Composition of N2 (±1σ) Derived from Experimental Cultures of Denitrifying or Anammox Bacteria

	substrate	Δ30 (‰)	n	Reference
natural soils
KBS T1 (conventional agricultural)	KNO3	–0.1 ± 0.1	3	this work
KBS T2 (no-till agricultural)	KNO3	0.1 ± 0.3	3	this work
KBS T7 (early successional)	KNO3	0.2 ± 0.2	4	this work
anammox enrichment cultures
Kuenenia spp.	NH4SO4 + NaNO2	–0.2 ± 0.1	3	this work
Brocadia spp.	NH4SO4 + NaNO2	–0.5 ± 0.3	2	this work
Scalindua spp.	NH4SO4 + NaNO2	1.0 ± 0.3	3	this work
denitrifying bacteria
Pseudomonas stutzeri	KNO3	0.9 ± 0.4	4	(14)
Paracoccus denitrificans	KNO3	0.6 ± 0.2	5	(14)

Anaerobic incubation of KBS soils yielded N2 with Δ30 values indistinguishable from the stochastic distribution of isotopes (i.e., all within 0.2‰; see Table ). Unlike in previous axenic laboratory cultures of denitrifying bacteria,[14] no statistically significant dependence on reaction extent or δ15N values was observed (p = 0.2–0.4 for a slope of zero, depending on the soil; see Table S1 of the Supporting Information, SI).

Compiling these results with those from earlier experiments on bacterial denitrifiers[14] shows that biological N2 production yields Δ30 values between −0.7‰ and +1.4‰, with a weak dependence, if any, on bulk δ15N values (Table S1). Moreover, the lack of Δ30 fractionation during biological nitrogen fixation[14] suggests that it preserves Δ30 values in the N2 residue. Atmospheric N2, in contrast, is characterized by Δ30, atm = 19.1 ± 0.1‰ (Figure ).[14]

Figure 1

Clumped-isotope composition of N2 derived from experimental cultures of denitrifying or anaerobic ammonia-oxidizing bacteria reported here and in ref (14). Substrates for experiments were as follows: USGS34 (KNO3, δ15N = −1.8‰) for denitrifying bacteria, NaNO2 (δ15N = −26.2‰) and NH4SO4 (δ15N = −0.5‰) in the anammox bioreactors, and bulk NaNO3 (δ15N = 5.4‰) for soil incubations.

Using Δ30 Values to Detect Biological N2 Fraction in Soil Gas

Due to the large and relatively invariant Δ30 contrast between atmospheric and biologically produced N2, we suggest here that Δ30 values in N2 can be used to quantify biologically produced N2 in soils via mass balance. To illustrate this concept, we first write the two-component mixing equations for the N2 isotopologue ratios in soil, 29Rsoil and 30Rsoil, in terms of the biological N2 fraction (fbio) and the N2 isotopologue ratios of atmospheric and biological N2 (subscripts “atm” and “bio,” respectively):While the soil-gas 29Rsoil and 30Rsoil values can be measured (as δ15Nsoil and Δ30,soil values) and 29Ratm and 30Ratm are known, this system of equations remains under-constrained. However, the proportionality between 29Rbio and 30Rbio coming from a relatively invariant biological clumped-isotope signature (Δ30,bio) provides a way forward.

Two-component mixing is linear in Δ30 values if the biologically produced N2 has the same 15N/14N ratio as that of the atmosphere, i.e., δ15Nbio = δ15Natm, yielding eq :[14,27]In that case, the soil-gas Δ30 value (Δ30,soil) would be simply related to fbio and the atmospheric (Δ30,atm) and biological clumped-isotope signatures. Measurements of Δ30,soil would allow one to solve for fbio:

Unknown and variable δ15Nbio values lead to deviations from this relationship, and uncertainty in fbio. However, for Δ30,soil values close to Δ30,atm (i.e., mixtures dominated by atmospheric N2), eqs and 9 retain much of their accuracy over a wide range of δ15Nbio values (Figure ). For example, when δ15Nbio is 20‰ different from δ15Natm, the fbio value derived from eq is within 6% of the true fbio value (e.g., a calculated fbio of 0.094 when the true fbio is 0.1). The expected range of Δ30,bio values coming from natural communities of ±1‰—i.e., the range observed in laboratory experiments—results in an additional ±6% relative uncertainty in fbio (e.g., an error of ±0.006 for fbio = 0.1). Both errors are similar to that contributed by analytical uncertainty for fbio = 0.1 (resulting in a cumulative uncertainty of ±10% if added in quadrature), but they quickly decrease in importance as fbio decreases: for fbio = 0.01, analytical uncertainty of ±0.08‰ in Δ30 results in an asymmetrical uncertainty of +36% and −56% fbio, i.e., fbio = 0.010–0.006+0.004. Therefore, analytical uncertainty dominates Δ30-based estimates of fbio for fbio < 0.1. Current analytical uncertainties suggest that soil gas containing ≥1% biological N2 will be detectable in Δ30,soil values.

Figure 2

Effects of bulk isotopic composition of biologically produced N2 on clumped-isotope based mass balance of biological and atmospheric N2. Inset shows mixing nonlinearity over the entire range of mixing fractions, which is most pronounced near a biological fraction of 0.5.

To test this concept, we obtained two depth profiles of δ15N and Δ30 values in N2, along with N2O concentrations, from a monolith lysimeter installed in the KBS Interactions site. We found that many Δ30,soil values were less than or equal to Δ30,atm (Figure and Table S2), ranging from 18.8‰ to 19.1‰. One sample analysis (34 cm depth on 10/11/17) was rejected based on apparent contamination that resulted in an abnormally elevated Δ30 value (4σ above the mean atmospheric value measured during the analytical session). The largest Δ30,soil depletions (−0.3 ± 0.1‰ relative to Δ30,atm), observed in both profiles, correspond to 1.6–0.5+0.4% of soil N2 at those depths being derived from biological processes. Soil-N2 δ15N values were equal to or slightly lower than the atmospheric value, although they differed between profiles: the profile obtained in July 2018 had δ15N values close to the atmospheric value, while the profile obtained in October 2017 had subatmospheric δ15N values ranging from −0.4 to −0.6‰. N2O concentrations increased nearly monotonically with increasing depth, with values exceeding 1000 parts per billion (ppb) at 170 cm depth (Figure ). Taken together, these data imply an active nitrogen cycle and the presence of biological N2 in these soils.

Figure 3

Depth profiles of Δ30 and δ15N values in N2 drawn from the same monolith lysimeter at the KBS LTER Interactions Experiment site on 7/18/18 (A) and 10/11/17 (B). Mean measured atmospheric Δ30 values were 19.04 ± 0.03‰ (1 s.e.m., n = 5) during the analysis period (dashed lines). Solid lines show depth profiles calculated using a 1-D diffusion-reaction model for each sampling date that are consistent with the Δ30 data (10/11/17 profile offset by −0.4‰). Note that these best-fit profiles for Δ30 may not be unique solutions due to the number of adjustable parameters in the model, e.g., the duration, width, and depth distribution of the assumed biological N2 pulse. Shaded areas therefore represent the range of N2 production rates that describes the analytical 1σ of Δ30 values (i.e., + 25% and −30% relative to the solid lines). Dashed lines denote isotopic compositions in the free atmosphere.

Figure 4

Depth profiles of N2O concentrations from the same samples as shown in Figure , along with illustrative steady-state profiles for uniform N2O production rates and 5% gas-filled porosities (D = 0.0026 cm2 s–1).[31,32]

Gas Diffusion and Denitrification Hot-Spots Can Explain Observed Soil Δ30 Profiles

A further understanding of the chemical and isotopic signatures measured in the soil gas can be obtained using a one-dimensional diffusion-reaction model based on Fick’s second law:where D is the effective gas diffusivity and J(z,t) is the production rate of a gas, which may be depth-(z) and time-(t) dependent. We treat the soil-gas system as a diffusive column ventilated to the atmosphere at the top (z = 0) and with zero permeability at the bottom (z = 180 cm). At steady state (∂C/∂t = 0) the depth profile is described by ∂2C/∂z2 = −J/D; because J and D are positive as defined, concentration depth profiles at steady state should monotonically decrease toward the atmospheric value. Isotopic tracers may increase or decrease toward the top depending how they are defined, but the change with depth should be monotonic toward the atmospheric value.

The depth profiles are not in steady state with respect to N2. At steady state, deeper soil-gas would have accumulated low-Δ30 biological signals over time, resulting in Δ30,soil values increasing from depth to the surface. The N2O depth profiles show accumulation at depth, but the N2 profiles do not (Figures and 4). Instead, Δ30,soil values are close to atmospheric values at depth, decrease at mid-depths, and return to atmospheric values at the surface. Pulsed biological N2 production over a limited depth range is required to reproduce these mid-depth minima in Δ30,soil values. Specifically, a quiescent period with respect to N2 production, which ventilates the soil down to 170 cm, must precede the pulse. Quantitative ventilation is not necessary, however; the quiescent period need only be long enough to dilute remnant Δ30,soil signals from earlier events beyond the limits of detection (∼5 days for the expected diffusivities; see below). Denitrification “hot moments” related to heterogeneities in soil moisture and organic carbon availability[28,29] have the appropriate temporal and spatial variability. The contrast between N2 and N2O depth profiles suggest that their production during these hot moments can be temporally decoupled. Moreover, the accumulation of N2O at depth argues against ventilation via gas exchange at the lysimeter–soil interface as the origin of the nonsteady-state Δ30,soil depth profile.

The shapes of the Δ30,soil depth profiles can be reproduced by solving eq using a 0.1-day N2 production pulse of Gaussian shape (1 cm full width at half-maximum), followed by a small time lag between the N2 pulse and sampling (see Table ). Here, we assume an air-filled porosity, ε, of 0.05—resulting in a calculated[30,31] soil diffusivity of D = 0.0039 cm2 s–1 for N2—and Δ30,bio = 0. The assumed ε value is within the plausible range for these soils[32] so it is appropriate for illustrative purposes. Using these parameters, the modeled Δ30,soil depth profile for July 2018 (the best-fit curve using a least-squares algorithm) reflects a depth-integrated gross production of 10.4 mmol N2 m–2 remaining in the soil after a 12.9 mmol N2 m–2 pulse (Figure A). The δ15N values of N2 in that profile can be reproduced if the biological N2 has δ15N = −11‰, on average. Note that the particular pulse shape, duration, and sampling lag used here (Table ) is likely one of many that can explain the data and therefore not meant to be diagnostic; consequently, the profile is considered a local (rather than global) best fit. The depth-integrated gross production, however, should be robust for a given air-filled porosity. For example, the model can also yield a satisfactory fit of the data using a 10-fold longer initial N2 pulse length of 1 day with a correspondingly weaker peak pulse peak of 0.3 nmol N2 cm–3 s–1 (instead of 2.6 nmol N2 cm–3 s–1). Both scenarios yield scaled-up N2 pulse magnitudes (3–4 kg N ha–1) that are consistent with peak N2 fluxes observed in previous in lab[33] and field[34] experiments.

Table 2

Model Parameters Used to Derive Profiles in Figure a

sampling date	center depth (cm)	pulse peak (nmol N2 cm–3 s–1)	sampling lag (h)	N2 production (mmol N2 m–2)
10/11/17	59	1.2	1.9	5.7
7/18/18	37	2.6	26.4	12.9

Pulses are Gaussian (1 cm full width at half maximum), occurring for 0.1 days.

The modeled Δ30 depth profile for October 2017 shown in Figure implies a depth-integrated gross production of 5.7 mmol N2 m–2 using a pulse centered at 59 cm (Figure B, Table ). Unlike for the July 2018 profile, the δ15N values of N2 in that profile cannot be explained by biological N2 production alone. Gravitational fractionation over this depth range would increase δ15N values by <0.01‰, so other physical mechanisms such as diffusive fractionation and/or water vapor flux fractionation[13] may be especially important for this profile. Sampling took place the morning after a heavy overnight precipitation event (>40 mm), implicating a physical isotope effect such as a hydrologically driven diffusive influx of atmospheric N2. These physical mechanisms will not affect Δ30,soil values significantly because they fractionate proportionately over a small δ15N range.[14,15] In addition, solubility fractionation does not seem to affect clumped-isotope compositions of sparingly soluble gases,[14,35] despite its effects on both elemental[36] and bulk-isotope composition.[37] Consequently, the Δ30 tracer shows a clearer measure of biological N2 production than the δ15N value of N2.

If these biological N2 pulses are isolated in time, then equivalent surface N2 fluxes F can be derived from the reaction-diffusion models, and the results compared to previous measurements of KBS soils. For one-dimensional diffusion, the equation F = [N2,bio] × D/z describes the instantaneous surface gas flux, where [N2,bio] is the concentration of biological N2 and z is the depth from the surface. The results for z = 5 cm, the biological N2 flux from the top 5 cm of soil, are shown in Figure . The flux F for the two profiles ranges from 0.1–2.9 mmol N2 m–2 d–1 (3–81 mg N m–2 d–1) during the first 10 days after the pulse events, with a prolonged period of low, but nonzero flux lasting several times longer (e.g., F = 0.1–0.2 mmol N2 m–2 d–1 for the 7/18/18 profile between 10 and 20 days after the pulse). These estimates are comparable to previous amendment-stimulated N2 production rates from these soils.[38,39] In particular, Bergsma et al. (2001) reported surface N2 fluxes of 0.2–2.0 mmol N2 m–2 d–1 (6–55 mg N m–2 d–1) during a four-day experiment utilizing a surface flux chamber and an amendment of 15N-labeled KNO3.[38] The model-derived fluxes strongly depend on the assumed air-filled porosity ε—which was not measured directly and can vary in time and space—so this agreement may be coincidental. Nevertheless, the two methods appear to yield results on the same order of magnitude. More well constrained in situ soil-atmosphere fluxes can be obtained with concurrent measurements of soil physical properties.

Figure 5

Calculated surface-flux time series for biological N2 derived from the Δ30 depth profiles in Figure (solid lines). Shaded areas represent the range of N2 production rates that describes the analytical 1σ of Δ30 values (i.e., + 25% and −30% in production rate). Dotted lines represent total biological N2 corresponding to each profile and show incomplete soil degassing after 20 days.

The only comparable in situ method for quantifying biological N2 production in soils is the N2/Ar method. Yang and Silver (2012) reported a relatively high detection limit of 3.9 mmol N2 m–2 d–1 for surface-flux measurements,[7] larger than the calculated peak surface fluxes shown in Figure . While the method can analytically resolve N2 excesses of less than 0.1%,[40] physical fractionation of N2 and Ar in soils presents substantial systematic uncertainties in these environments. We hypothesize that measurements of N2/Ar soil profiles may yield limited improvements in uncertainty because the physical mechanisms complicating the interpretation of δ15N values of N2 (e.g., the water vapor flux fractionation)[13] fractionate N2/Ar ratios to a greater degree, offsetting any analytical sensitivity advantages. Soil Δ30,soil depth profiles, in contrast, are insensitive to physical fractionation, revealing evidence for biological N2 production in soil profiles despite the lower analytical sensitivity of the method.

N2 fluxes into the atmosphere can be derived from Δ30,soil profiles if soil physical properties (i.e., air-filled porosity and diffusivity) are determined independently. The method could be used to compare in situ production rates to incubation- and amendment-based methods in field studies, or to obtain independent estimates using an array of spatially dispersed observations across soil types and conditions. Time series of soil-gas profiles similar to those shown here, sampled through lysimeters or air-permeable tubing, would provide a long-term perspective on soil N2 production dynamics, which is presently difficult to access without perturbing soil biogeochemistry and is useful for models.[41] Analytical throughput (2–3 samples/day) and availability of instrumentation are currently limiting factors for the Δ30 approach, but the relatively long ventilation time scales of certain soils may still allow weekly to-monthly sampling to capture the impacts of hot moments.

The initial results reported here suggest that Δ30,soil signals are sufficiently large that the approach can be used in future assessments of site- and ecosystem-scale loss of fixed nitrogen. Furthermore, the approach can also be applied to marine environments to investigate both the magnitude and mechanisms of fixed-nitrogen loss in low-oxygen zones.[42] Finally, constraining biological N2 production globally using Δ30,atm appears possible in principle if the terms related to upper-atmospheric chemistry in the global Δ30 budget—both the isotopic reordering rates and Δ30 endmembers—can be refined.