Soil Formate Regulates the Fungal Nitrous Oxide Emission Pathway

Abstract

Fungal activity is a major driver in the global nitrogen cycle, and mounting evidence suggests that fungal denitrification activity contributes significantly to soil emissions of the greenhouse gas nitrous oxide (N₂O). The metabolic pathway and oxygen requirement for fungal denitrification are different from those for bacterial denitrification. We hypothesized that the soil N₂O emission from fungi is formate and O₂ dependent and that land use and landforms could influence the proportion of N₂O coming from fungi. Using substrate-induced respiration inhibition under anaerobic and aerobic conditions in combination with ¹⁵N gas analysis, we found that formate and hypoxia (versus anaerobiosis) were essential for the fungal reduction of ¹⁵N-labeled nitrate to ¹⁵N₂O. As much as 65% of soil-emitted N₂O was attributable to fungi; however, this was found only in soils from water-accumulating landforms. From these results, we hypothesize that plant root exudates could affect N₂O production from fungi via the proposed formate-dependent pathway.

The importance of fungal denitrification to the emission of nitrous oxide (N₂O; an important greenhouse gas) from soils has been demonstrated for a number of systems. Up to 89% of soil N₂O emissions could be attributed to fungal activity (25). Given that fungal biomass dominates in many ecosystems, its potential activity may be the dominant soil N₂O-emitting process. Fungal denitrifiers are ecologically significant because most fungal isolates capable of denitrification appear to lack nitrous oxide reductase, the enzyme that reduces N₂O to N₂ (32, 58). Therefore, unlike the end product of bacterial denitrification, which is mostly N₂, the end product of fungal denitrification is N₂O.

Accumulated evidence from work with fungal isolates indicates that the fungal pathway for the respiratory reduction of nitrogen oxides to nitric oxide (NO) and N₂O is different from that of bacteria. Fungal nitric oxide reductase (P450nor) is a cytochrome p450-containing enzyme that receives electrons directly from NADH for the reduction of NO to N₂O (32, 58). Consequently, small amounts of O₂ (hypoxia) are required to generate NADH from the oxidation of citrate in the tricarboxylic acid cycle (Fig. 1). This is in contrast to bacterial denitrification, where successive enzymes in the pathway are increasingly sensitive to O₂ inhibition (44, 59). The O₂ requirement for fungal denitrification, however, has not been tested explicitly in soil.

FIG. 1.

Proposed O₂- and formate-dependent fungal denitrification pathway developed from the cited works. ¹, Kuwazaki et al., 2003 (24); ², Zhou et al., 2002 (57); ³, Uchimura et al., 2002 (47); ⁴, Nakahara et al., 1993 (32); ⁵, Zhou et al., 2001 (58). TCA, tricarboxylic acid cycle; FDH, formate dehydrogenase; NAR, nitrate reductase; NIR, nitrite reductase; Acetyl-CoA, acetyl coenzyme A.

An interesting feature in some fungal denitrification pathways is the coupling of nitrate or nitrite reduction with formate (HCOO⁻) oxidation (Fig. 1) (24, 47). Low-molecular-weight organic acids, such as formate, are important root exudates (22) and are intermediates and by-products of anaerobic carbon metabolism (10). Formate (together with acetate) is the end product of the fermentation of citrate, oxaloacetate, and pyruvate. It is also produced from H₂ and CO₂ by a variety of anaerobic microorganisms (e.g., acetagens, sulfate reducers, and methanogens) (21). The amount of formate in aerobic soils is reported to range from 6 to 26% of the total extractable low-molecular-weight organic acids (51). Reported rhizosphere formate concentrations range from below the detection limit for clover (Trifolium repens) (9) to 117 μM for Norway spruce (Picea abies) (49) and 563 μM for quack grass (Elytrigia repens) (5). In addition to having external sources, fungi can produce formate. Under O₂-limited conditions, formate is produced from the fermentation of pyruvate (57). Because of its various sources and relative ubiquity in soils, formate-dependent respiratory reduction of nitrogen oxides to N₂O by fungi may be an important contributor to net N₂O emissions from soils.

We hypothesized that soil N₂O production by fungi is formate and O₂ dependent and that land use and landform influence the proportion of N₂O attributable to fungi. Land use factors, such as tillage, have an influence on fungi by physically disturbing the soil (17), while fertilizer applications can have an inhibitory effect on fungi (4, 16). Land use and landform also affect water distribution (18, 30) and the quantity and quality of soil organic matter (48), which, in turn, can affect soil fungi. Using substrate-induced respiration inhibition (SIRIN) under anaerobic and aerobic conditions, in combination with ¹⁵N gas analysis, we evaluated the importance of formate and O₂ to N₂O production by fungi in cultivated and uncultivated soils at the St. Denis National Wildlife Area in Saskatchewan, Canada.

MATERIALS AND METHODS

Study site.

The St. Denis National Wildlife Area in central Saskatchewan, Canada (52°12′N, 106°5′W), is typical of the North American prairie pothole region. It contains 216 wetlands within an area of 3.84 km² (19) in the Dark Brown soil zone. Soil types range from thin Typic Calciborolls and thick Typic Haploborolls for water-shedding landform elements to Albic Argiborolls and Argic Cryaquolls in water-accumulating elements (54). Soils (Weyburn Association) developed on loamy unsorted glacial till parent materials in hummocky terrain with slope classes ranging from 10 to 15% (31). Six ephemeral wetlands (three cultivated and three uncultivated) were selected for study. Ephemeral wetlands are those depressions in hummocky landscapes that contain standing water in the spring but typically dry out during the growing season (18).

A digital elevation model was used to subdivide the cultivated wetlands (CW) into convex (CX; topographically high positions with a positive profile curvature that sheds water) and cultivated depression center (CD; level positions, roughly circular in shape, which temporarily collect rain or snowmelt water) landform elements (55). These two landforms represent the extremes in terms of N₂O emission, soil moisture conditions, and biological productivity within the cultivated landscape (19, 54, 55). Uncultivated wetlands (UW) were found in nonagricultural portions of the site. These were divided into two landform elements roughly equivalent to those in the cultivated wetlands. Basin centers (BC) are level areas covered by 99 nongrass plant species, collect rain and snowmelt water, and are analogous to the CD elements. Uncultivated wetlands also included nonlevel fringe areas covered with Bromus inermis, termed riparian grass (RG) areas. The RG elements represent the driest areas within uncultivated wetlands and, in this sense, are analogous to CX elements in the cultivated wetlands.

Soil sampling and soil carbon and nitrogen determination.

Three cultivated and three uncultivated wetlands (each constituting a land use replicate [6]) were sampled on 12 September 2006. Five soil cores (0 to 15 cm; 15-cm internal diameter [i.d.]) were collected from the individual landform elements in each of the six wetlands. The cores from each individual landform element were bulked together to form a composite sample for that location, yielding a total of 12 composite samples (2 land uses × 2 landforms × 3 replicates). Samples were transported on ice, and subsamples were collected for soil formate extraction and lyophilization (for phospholipid fatty acid [PLFA] extraction). The remaining soil was air dried just enough (<24 h) to pass through a 2-mm sieve without smearing and stored at −20°C. Inorganic nitrogen (2 M KCl extracts) (28), soil organic carbon (dry combustion) (37), and total nitrogen (Dumas method) (34) were determined by standard methods, and results are listed in Table 1.

TABLE 1.

Organic carbon and mineral and total nitrogen in soils of the St. Denis National Wildlife Area, Saskatchewan, Canada^a

Land use	Landform	SOC (%)	NH4+	NO3−	Total N (%)	C/N
Cultivated	CX	2.3 (0.1)	1.8 (0.1)	1.7 (0.2)	0.2 (0.0)	9 (0)
	CD	3.2 (0.2)	2.7 (0.1)	2.9 (0.4)	0.4 (0.0)	9 (0)
Uncultivated	RG	2.4 (0.1)	3.4 (0.4)	1.6 (0.2)	0.3(0.0)	9 (0)
	BC	3.5 (0.1)	4.7 (0.8)	6.0 (1.6)	0.4 (0.0)	9 (0)

^aReported values are means (μg N g⁻¹ soil) with standard errors in parentheses.

SIRIN incubations to determine the contribution of fungi to N₂O emissions.

Substrate-induced respiration inhibition studies involved the following treatments: control soil (no added microbial inhibitor), cycloheximide-amended soil, and streptomycin-amended soil. Whereas cycloheximide was chosen for targeted suppression of fungal activities, streptomycin was chosen as a broad-spectrum bacterial activity inhibitor (3). Soils were removed from cold storage, thawed at room temperature, and packed into a 10-ml volume at the bottom of a 55-ml glass culture tube (22-mm i.d.) to yield bulk densities similar to those observed in the field. Soil water content was determined using standard procedures with an assumed particle density of 2.65 g cm⁻³ (46). Preliminary experiments determined that when repacked cores are initially wetted to 50% or 70% water-filled pore space (WFPS), a burst of N₂O was observed at 24 h after wetting, declined to background levels at 48 h after wetting, and remained at background levels for several days thereafter. Thus, soil water content was first adjusted to 50% WFPS, and the culture tube was capped with Parafilm and stored at 4°C for 48 h prior to treatment.

Preliminary experiments also were conducted to determine the optimal cycloheximide and streptomycin concentrations for use with the CD and BC soils. Optimum concentrations of the inhibitors were determined using a modification of the method described by Laughlin and Stevens (25). The modifications included bringing the soils to 70% WFPS and adding potassium formate (2 mg formate-C g⁻¹ soil; equivalent to about 2% of the maximum soil organic carbon found in the St. Denis soils) and potassium nitrate (80 μg NO₃-N g⁻¹ soil; this gave a ratio of soil organic carbon to total N [C/N ratio] of 24:1). The optimum concentration of both cycloheximide and streptomycin was found to be 4 mg inhibitor g⁻¹ soil for the CD soils and 8 mg inhibitor g⁻¹ soil for the BC soils. These levels of inhibitor also were applied to the respective upland soils. Because cycloheximide is soluble only in methanol, the cycloheximide stock solution consisted of 4 g of cycloheximide dissolved in 10 ml of methanol and dispersed in deionized water to a final volume of 50 ml. Streptomycin and control (no-biocide) solutions were made with a similar volume and methanol/water ratio. The addition of the inhibitors resulted in an increase in soil moisture content to approximately 60% WFPS. Thus, the tubes were recapped with Parafilm and incubated in the dark at 23°C for an additional 24 h prior to the addition of the formate and nitrate (11). After adjustment of the soils to a final moisture content of 70% WFPS, the tubes were sealed with butyl-rubber caps and incubated in the dark at 23°C for a 24-h incubation. After 24 h, a 20-ml gas sample was withdrawn from the headspace using a 20-ml disposable syringe equipped with a 25-gauge needle and injected into a preevacuated 12-ml Exetainer vial (Labco Ltd., United Kingdom). Headspace N₂O concentrations were determined using a gas chromatograph equipped with an electron capture detector (55).

¹⁵N stable isotope incubation with N₂O, ammonium, and nitrate analyses.

Modified SIRIN incubations were repeated for the CD and BC soils in which the N substrate was enriched with ¹⁵N. That is, the added KNO₃ consisted of 78 μg KNO₃-N g⁻¹ soil and 2 μg K¹⁵NO₃-¹⁵N g⁻¹ soil (at 98% ¹⁵N enrichment). The ¹⁵N studies were conducted under both aerobic and anaerobic conditions. Anaerobic systems were prepared by replacing the headspace air in the culture tubes with ultrahigh-purity N₂ (i.e., working in an anaerobic chamber under a N₂ atmosphere). Gas samples were collected as described for SIRIN incubations, after which the soils were destructively sampled for ammonium and nitrate by extraction with 2 M KCl.

¹⁵N-labeled N₂O was analyzed by isotope ratio mass spectrometry at the University of California—Davis Stable Isotope Facility (Davis, CA). The fractionation of ¹⁵N-labeled N₂O was calculated as described by Arah (1) and Stevens et al. (41). The diffusion disk technique described by Stark and Hart (40) and modified by Bedard-Haughn et al. (7) was used to concentrate soil ammonium and nitrate for ¹⁵N analysis.

Soil N₂O emissions in response to increasing concentrations of formate.

Soils were removed from cold storage, thawed at room temperature, and packed to a 10-ml volume at the bottom of a 55-ml glass culture tube to yield bulk densities similar to those observed in the field. The soil water content was adjusted to 50% WFPS, and the tubes were capped with Parafilm and stored at 4°C for 48 h. The soils were then amended with formate (at concentrations of 0.0, 0.1, 0.2, 0.4, 0.8, and 1.6 nmol formate-C g⁻¹ soil) and adjusted to 70% WFPS; the tubes were then sealed with butyl-rubber caps and incubated in the dark at 23°C for 24 h. Gas sampling and analysis were carried out as described for the SIRIN incubations. The effect of formate additions on N₂O emissions was modeled using the Enzyme Kinetics Module (single-substrate model) in SigmaPlot version 9.0 (Systat Software, Inc., San Jose, CA).

Soil formate extraction and analysis.

Soil formate was extracted using the centrifugation drainage technique described by van Hees et al. (50). Extracts were stored at −20°C prior to analysis and were analyzed using the capillary electrophoresis method for anion detection described by Swallow and Low (43). The analyses were carried out using a Waters Quanta 4000 capillary electrophoresis system (Waters Corporation, Millford, MA) equipped with a 60-cm by 75-μm (i.d.) fused silica column. The electrolyte buffer was 5 mM sodium chromate-0.4 mM OFM-BT (Waters Corporation) adjusted to pH 8 with lactic acid and filtered through a 0.45-μm Millipore membrane filter (Millipore, Billerica, MA). Anions were detected at 254 nm using indirect UV detection. Samples and standards were analyzed according to the following sequence: (i) 30-s hydrostatic injection, (ii) 5-min run time at 25 kV, (iii) 1.5-min rinse with 0.1 M NaOH, (iv) 4-min rinse with Nanopure deionized water, and (v) 5-min purge with electrolyte buffer.

PLFA extraction and analysis.

PLFAs were extracted from lyophilized soil samples (4.0 g) using the method described by White et al. (53). Briefly, lipids were extracted from soils in a mixture of methanol, chloroform, and phosphate buffer (2:1:0.8 [vol/vol/vol]). Lipids were separated on silica gel columns (Bond Elut, Varian, Inc., Mississauga, Canada) by sequential applications of chloroform, acetone, and methanol (with the methanol fraction containing the phospholipids) (20). Phospholipids were methylated and separated by using a Hewlett Packard 5890 Series II gas chromatograph equipped with a 25-m by 0.2-mm (i.d.) Ultra 2 column (J&W Scientific; MIDI, Inc., Newark, DE) and a flame ionization detector. Lipid peaks were identified by comparisons of retention times to those in the TSBA version 4.1 and CLIN version 4.0 lipid libraries (MIDI, Inc.). Methylnonadecanoate (19:0) fatty acid (1 μg) was added to each sample as the internal standard before the methylation step (20). The PLFA 18:2ω6,9 was used as the fungal biomass marker (2).

Statistical analysis.

The SIRIN and ¹⁵N gas data were imported into Minitab (v. 11.21; State College, PA) and analyzed using two-way analysis of variance (ANOVA) with a general linear model (α = 0.1) after verification that the data met the ANOVA assumptions (using the Anderson-Darling test for normality and Bartlett's and Levene's tests for homogeneity of variance). Classification variables were landform and treatment. Because there were significant differences in landform (P = 0.001) and landform versus treatment (P < 0.001), Tukey's pairwise comparison was used to assess the significance of treatment differences within a landform. PLFA data were compared using one-way ANOVA and Tukey's pairwise comparisons with landform as the classification variable.

RESULTS

Nitrous oxide emissions attributable to fungal activity were found (P < 0.1) in both CD and BC soils (Fig. 2). Indeed, CD and BC soils incubated with cycloheximide produced 46% and 65% less N₂O, respectively, than the control soil. The wide-spectrum bacterial inhibitor streptomycin also decreased N₂O emissions in the CD (−51%) and BC (−47%) soils. The inhibitors had no significant effect on N₂O emissions from the RG soils, but they produced increased emissions in soils from the CX landform elements. Either these water-shedding landforms, RG and CX, do not have significant fungal N₂O emission, or the inhibitors were ineffective in these landforms. Thus, because our focus was on fungal denitrification, subsequent incubation studies designed to probe the regulation of fungal denitrification with ¹⁵N-labeled nitrate and inhibitors was restricted to the CD and BC soils.

FIG. 2.

Nitrous oxide emission from a SIRIN assay incubated at 70% WFPS for 24 h. Reported values are means with standard error bars (n = 3). An asterisk above the error bar denotes N₂O emission from a treatment incubation with results that were significantly different from those of the control incubation (P < 0.1).

Relative to the appropriate controls, N₂O emissions from the CD and BC soils exhibited a 30-fold increase when incubated with formate (Fig. 3A and B). Furthermore, the formate effect was most pronounced under hypoxic conditions (i.e., soils incubated at 70% WFPS with an aerobic headspace). In the presence of cycloheximide, N₂O emissions from CD and BC soils decreased by about 30% when they were incubated in a low-oxygen environment. Conversely, there was no difference between incubations with or without the inhibitor when soils were incubated under anaerobic conditions (i.e., soils at 70% WFPS with a nitrogen atmosphere), suggesting a bacterial source for the N₂O. In the presence of formate, nearly all of the N₂O emitted was derived from the labeled ¹⁵NO₃⁻ pool (Fig. 3C and D). In the absence of formate, the majority of the N₂O was derived from the unlabeled soil-N pool. Nitrogen-15 enrichment of the ammonium pool was not observed during the incubation period (data not shown).

FIG. 3.

SIRIN assay with ¹⁵N-labeled nitrate incubated at 70% WFPS for 24 h. Reported values are means with standard error bars (n = 3). (A and B) Total N₂O emissions. An asterisk above the error bar denotes N₂O emission from a treatment incubation with results that were significantly different from those of the control incubation (P < 0.1). (C and D) Percentage of emitted N₂O attributable to denitrification or processes that oxidized ¹⁵NO₃⁻ to ¹⁵N₂O.

In general, N₂O emissions increased as the amount of formate added to the soils increased (Fig. 4). Moreover, soil N₂O emissions demonstrated a hyperbolic dependence on formate concentration. As such, Michaelis-Menten kinetics was used to describe the formate-dependent N₂O emissions from all soils (r² values ranged from 0.94 to 0.99). Soils from water-accumulating (CD and BC) landforms had greater affinity (lower K_m values) for formate than soils from drier landforms (CX and RG). Not surprisingly, then, there was less extractable formate in soils from water-accumulating landforms than in soils from water-shedding or drier landforms.

FIG. 4.

N₂O emitted as a function of added formate concentration. Closed circles indicate experimental means of N₂O emitted with standard error bars (n = 3); the solid line is the modeled response. Units for V_max are nanomoles N₂O-N nanomoles formate-C⁻¹ day⁻¹. K_m has the same units as the x axis. The r² is the goodness of fit for the nonlinear regression line as determined by the Enzyme Kinetics Module for SigmaPlot 9.0 using the single-substrate (formate) model. Reported soil formate concentrations are means with the standard error in parentheses (nanomoles formate-C grams⁻¹ soil). ND, not detected.

Nitrous oxide emissions attributable to fungal activity were greatest in the water-accumulating landforms (Fig. 2). However, the water-accumulating landforms yielded the least extractible fungal PLFA (Fig. 5A) and exhibited the lowest fungus/bacterium PLFA ratios (Fig. 5B).

FIG. 5.

Fungal biomass in soil as determined by PLFA extraction. Shown are the mol% of fungal PFLA biomass marker (A) and fungi/bacteria PLFA biomass marker (B). Reported values are means with standard error bars (n = 3). Different letters below error bars denote significant differences (α = 0.1); lowercase letters are used for comparisons between soils of cultivated wetlands, and uppercase letters are used for comparisons between soils of uncultivated wetlands.

DISCUSSION

The O₂ and formate dependence of N₂O emissions associated with fungi can be explained by the proposed fungal denitrification pathway (Fig. 1). To date, fungal nitric oxide reductase activity and the P450nor genotype have been reported for only a few fungi isolated from soil and wastewater reactors (52, 56, 58). In all cases, the expression of the P450nor gene was essential for the proposed fungal denitrification pathway. Thus, it was not surprising to find that fungal denitrification was not directly linked to fungal biomass. Rather, it is likely linked to the presence and expression of a specific functional genotype, P450nor.

The hyperbolic dependence of N₂O emission on formate concentration, as modeled by Michaelis-Menten kinetics, supports our contention, outlined in Fig. 1, that a formate dehydrogenase/nitrate reductase couple and/or a formate-dependent nitrite reductase is directly linked to N₂O emissions. The enzymatic characteristics of the formate-dependent N₂O emissions differed between soils of water-accumulating landforms (CD and BC) and the soils from the corresponding water-shedding or drier landforms (CX and RG). For example, the higher affinity (lower K_m) for formate in CD and BC soils would impart a competitive advantage to those fungi in soils with low soil formate concentrations. Also, the potential maximum formate-dependent N₂O emission per unit formate-C (V_max) was 2.5 to 6 times greater in soils from water-accumulating than from upland landforms. The reported enzymatic parameters and magnitude of the formate-dependent N₂O emissions may be underestimated, because experimental additions of formate can encourage nitrogen immobilization by increasing the C/N ratio (29).

Formate stimulated nonfungal N₂O emissions. In the bacterial pathway, formate mediates the reduction of nitrate to ammonia (nitrate ammonification) (8), with N₂O being released as a result of the nonspecific action of dissimilatory nitrate reductase. Although energetically less favorable than denitrification (42), nitrate ammonification is reported to be the more important process in respiratory nitrate reduction under anaerobic conditions and when nitrate is limiting (44, 45). However, this did not appear to be the case; i.e., we did not observe any ¹⁵N enrichment of the ammonium pool as would have been expected. Alternatively, bacteria capable of mixotrophy (e.g., Paracoccus denitrificans) can denitrify nitrate using formate/H₂ as an electron donor (38, 39).

Under hypoxic conditions, heterotrophic nitrification may also contribute to N₂O release. Evidence suggests this is an important N transformation process for organic N and NH₄⁺ in a variety of soil systems (11, 26, 35). Heterotrophic nitrification derives energy from organic C oxidation rather than NH₃ oxidation (23), with N₂O being a product of the incomplete oxidation of hydroxylamine. Castaldi and Smith (11) demonstrated that fungal heterotrophic nitrification can be the dominant N₂O-emitting process in woodland and arable soils. In addition, heterotrophic nitrifiers produce N₂O from denitrification of nitrification products (i.e., fungal nitrifier denitrification) (12). Fungal heterotrophic nitrification would produce primarily ¹⁴N₂O. Under hypoxic conditions with formate, ¹⁵N₂O dominated, whereas without formate, only 10% of the N₂O was ¹⁵N₂O. Thus, under hypoxic conditions, there may be two fungal pathways contributing to N₂O release: fungal denitrification via P450nor and fungal heterotrophic nitrification. Our results suggest that the relative importance of these two processes is linked to soil formate concentrations.

Nitrous oxide emissions associated with fungi were landform, but not land use, dependent. It is often reported that fungal biomass is greater in native (uncultivated) soils than in adjacent cultivated soils and that fungal biomass increases with decreasing land use intensity (e.g., in no-tillage versus conventional-tillage soils) (13, 14, 17). Such was not the case in the present study; instead, differences in the fungal community appeared to be linked to landscape position (i.e., landform) through its influence on water redistribution. For example, fungal biomass (mycorrhizal and saprotrophic) has been shown to decline in soils that experience periods of flooding or high soil water content (30). This presumably reflects decreased mycorrhizal associations (33) or decreased saprotrophic activity under conditions of decreased O₂ availability imposed during these periods.

We did not observe fungus-linked N₂O emissions in the CX and RG landforms on the basis of our cycloheximide inhibition experiment. The cycloheximide concentrations were optimized for the CD and BC landforms and then applied to the CX and RG landforms. We hypothesized that since soils of CD and BC landforms have the highest levels of organic matter (Table 1), the levels of cycloheximide that inhibit fungi in these soils would be effective in the soils of water-shedding landforms, CX and RG. If this was incorrect, we may have underestimated the importance of fungus-linked N₂O emissions in the water-shedding landform elements.

Technical limitations and questions associated with the specificity of the inhibitors and substrate (formate) dictate that future investigations will require novel ideas to validate the findings of this work. To validate the specific activity of the inhibitors, quantification of 18S and 16S rRNA gene fragments or transcripts (36) could shed light on how the 18S or 16S pool is affected by the different inhibitor treatments. Isolation and characterization of the P450nor and formate dehydrogenase enzymes by mass spectrometry (32, 47) from soil would provide credence to our hypothesis as shown in Fig. 1. Inoculation with and activity monitoring of fungi with wild-type and mutant genotypes of these two enzymes into soil may also support our arguments.

For the first time, soil N₂O emissions related to fungi are linked to soil formate concentration and O₂ availability. This finding poses interesting management considerations. First, the selection of crop varieties that exude smaller amounts of formate (9) could mitigate N₂O production from fungi in areas where fungal denitrification is prevalent. Second, the benefits of improved soil fertility and plant growth through H₂ release from nodules of legumes (15, 27) could be countered by increased N₂O production from soil fungi because of formate production from H₂ and CO₂ (21).