Introduction

The nitrous oxide (N2O) molecule is a potent greenhouse gas, with a global warming potential 298 times that of carbon dioxide over a 100 year time period1. It is also a precursor to reactions involved in the depletion of stratospheric ozone2. A major source of anthropogenic N2O emissions is the intensive grazing of grasslands and the resulting ruminant urine deposition that occurs3,4. Thus, in order to achieve mitigation of N2O emissions from intensively managed pasture soils it is important to identify and understand the processes that lead to N2O formation and consumption within ruminant urine-affected soil.

Typically, ruminant urine-N deposited onto pasture soil is comprised of >70% urea-N. Upon contact with the soil the urea begins to hydrolyse, forming ammonium (NH4+) resulting in a rapid elevation of soil pH to 8.0 or higher5. The equilibrium between NH4+ and ammonia (NH3) is pH driven6,7. Soil pH >7.0 leads to elevated NH3 concentrations in the soil, that not only result in NH3 volatilization8 but which can also inhibit the microbial oxidation of nitrite (NO2−) by Nitrobacter sp.9,10. As the pH decreases to ca. <7.0, the equilibrium between NH4+ and NH3 shifts in favour of NH4+, which may undergo clay mineral fixation, plant uptake, immobilization or nitrification11.

Production of N2O may occur via the microbial pathways of nitrification, denitrification, and nitrifier-denitrification12. However, under ruminant urine-affected soil it is bacteria, not archaea, that respond to the high concentration of NH4+ substrate that forms in the soil following ruminant urine deposition13,14, since bacterial nitrifiers operate under conditions of high inorganic NH4+ inputs14,15,16. During the conventional nitrification process bacteria produce N2O as a by-product of NH2OH oxidation17 or during nitrifier-denitrification following nitric oxide (NO) reduction15. However, the major source of N2O emissions from ruminant urine-affected soil occurs as a result of the NO3− formed, as a consequence of nitrification. Under anaerobic conditions microbes denitrify NO3− to sequentially form NO2−, NO and N2O, which are all obligate intermediaries of the denitrification pathway12,18,19,20 to finally create dinitrogen (N2). In order to conserve both energy and oxygen, nitrifier-denitrification may occur in response to limited soil oxygen conditions21, whereupon nitrifiers convert NO2− to NO, N2O and N212 although the significance of this process may have been overestimated in some studies22. In addition to these N2O production pathways, N2O may also be produced as ‘hybrid’ N2O via codenitrification, a process involving two different N pools20,23. Spott et al.20 reviewed possible biotic and abiotic reactions that may be included under the term ‘codenitrification’. For example, abiotic reactions involving reduced iron (Fe2+) and NO2−, may occur at the interface between an aerobic zone overlying an anaerobic zone when NO2− diffusing downwards meets Fe2+24,25. However, this process is unlikely to contribute significantly to N2O emissions due to insufficient Fe2+ ion concentrations in most soils26,27. A more common abiotic reaction that occurs in acidic soil (pH < 5.0) is that of chemodenitrification (abiotic-nitrosation), whereby NO2− and H+ react to form nitrous acid (HNO2), which can then react with amino compounds, NH2OH, NH4+ or other organic N compounds resulting in the formation of N2O28,29. However, under alkaline conditions when oxygen is depleted codenitrification may occur via biologically mediated nitrosation20,30. Under such conditions the hydrogen atom in an organic compound is replaced with a nitroso group (−N=O). Enzymatic nitrosyl compounds attract nucleophile compounds (e.g. NH2OH, NH4+, hydrazine (N2H2), amino compounds and NH3) resulting in hybrid N2O or N2 species, containing one N atom derived from the nucleophile and one N atom derived from the nitrosyl compound20. Recent studies have revealed the significant contribution of codenitrification to gaseous N losses from grassland soils30,31,32. Using a 15N tracer approach, Laughlin and Stevens32 found evidence for fungal dominated 15NO3− depletion leading to hybrid N2 emissions where 92% of the N2 evolved was derived from codenitrification. Selbie et al.30 confirmed, in-situ, the dominance of codenitrification derived N2 under urine patch conditions when 56% of applied urine was codenitrified. Recently, studies have found further evidence for N2O production via codenitrification under simulated ruminant urine patch conditions31,33. However, knowledge about the nucleophile species that potentially partake in codenitrification under ruminant urine patch conditions is still lacking. Different N substrates (as potential nucleophiles) such as amino acids, NH4+ and NH2OH have previously been proven to be capable of generating hybrid N2O/N2 in vitro when utilized by one microbial species in combination with either NO3− or NO2−34,35,36,37. Amino acids have been reported to be freely available within the soil solution, for example, phenylalanine (8–50 µg N g−1 soil) and glycine (35–193 µg N g−1 soil) were measured in long-term agricultural land on a Stagni-Haplic Luvisol38 and in different cattle manure treated crop fields on a sandy Orthic Luvisol39. Reported concentrations of NH2OH are orders of magnitude lower, for example, Liu et al.40 reported concentrations of <0.0348 µg N g−1 in a forest soil, while NH4+ and NH3 are routinely reported following ruminant urine deposition events41. Therefore, we hypothesise that in a soil matrix under simulated ruminant urine deposition the N substrates applied in this study will be utilized for codenitrification reactions, with a microbial preference for NH2OH and that these reactions would be mainly fungi driven.

Results

Soil pH, and mineral N

Within 6 h of applying the urea solution to the soil surface pH values increased uniformly in all treatments from an average of 5.6 ± 0.2 on Day −2 to >7.6 on Day 0. The surface soil pH peaked 30 h after the urea application, at 7.9, followed by a steady decline to 4.8 ± 0.1 on Day 9 (Fig. 1) in the positive control and all treatments. The surface pH in the negative control ranged from 5.4 ± 0.05 to 5.6 ± 0.06 over the course of the experiment (Fig. 1).

Figure 1
figure 1

Soil response to urea and treatment application. The N2O fluxes over time (a) of the no inhibition treatments. Before Day 9 the N2O fluxes did not significantly differ between the positive control and the treatments. On Day 9, the N2O fluxes of all treatments and the positive control increased as listed in Table 1, for simplicity only the non-inhibition treatments are depicted in Fig. 1 to visualize the range of increase. Below the NO2− concentration in the soils as measured in the NO2− control. (b) These partially destructive analysis was not performed within the treatment soils and the positive controls, but depicts the assumed NO2− concentration development within these soils. The soil surface pH was measured in all jars, however, all treatment soil surface pH values did not differ from the depicted positive control, in contrast to the negative control. (c) Each symbol represents a mean of n = 3, all error bars are ± SD.

Full size image

Soil NO2− concentrations were significantly elevated within the first 4 days following urea application (p < 0.05). Soil NO2− concentrations peaked at 1.5 ± 0.2 µg NO2−-N g−1 soil on Day 9, subsequent to the physical mixing and then decreased to 0.6 ± 0.1 µg NO2−-N g−1 soil on Day 11 (Fig. 1b).

Both the soil NO3− and NH4+ concentrations were higher (p < 0.01) in the positive control at Day 12 compared with the negative control. The NO3− concentrations in the positive control were in the range of 366 ± 122 µg NO3−-N g−1 soil while NH4+ concentrations were 174 ± 7 µg NH4+-N g−1 soil. The soil NO3− and NH4+ concentrations in the negative control were 64 ± 23 µg NO3−-N g−1 soil and 22 ± 1 µg NH4+-N g−1 soil, respectively.

N2O fluxes

Initially N2O fluxes increased within the first 48 h following urea application, with treatments and positive controls emitting 100–200 µg N2O-N m−2 h−1. From Day 4 to Day 8, the N2O fluxes from the urea-treated soil were <100 µg N2O-N m−2 h−1 across all treatments. Following N2O flux measurement on Day 8, the process of mixing the soil and/or the addition of N substrates increased N2O fluxes at Day 9 (Fig. 1). In the absence of microbial inhibition, the addition of the NH2OH substrate resulted in higher N2O fluxes (4496 µg N2O-N m−2 h−1) when compared to the amino acid (1796 to 2130 µg N2O-N m−2 h−1) and NH4+ (1405 µg N2O-N m−2 h−1) treatments on Day 9 (p < 0.001), 24 h after N substrate addition.

The magnitude of the decrease in the N2O fluxes, following inhibition treatment, varied due to inhibitor type and N substrate applied (Table 1). The N2O emissions were lower under fungal inhibition by 46, 34 and 21% in the glycine, phenylalanine, and NH2OH treatments, respectively, while fungal inhibition did not affect fluxes from the NH4+ treatment. Bacterial inhibition decreased N2O fluxes by 14, and 26% in the glycine and NH2OH treatments, respectively, while fluxes from the phenylalanine and NH4+ treatments were unaffected by bacterial inhibition (Table 1). Applying both inhibitors simultaneously (combined inhibition) resulted in N2O fluxes decreasing by 29–41% in all N substrate treatments (Table 1). In the glycine treatment fungal inhibition decreased N2O fluxes more than bacterial inhibition, but this decrease was not enhanced when the two inhibitors were combined (Table 1). While bacterial inhibition did not significantly lower N2O fluxes in the phenylalanine treatment, the fungal inhibition either alone or within the combined inhibition did decrease N2O fluxes (Table 1). Sterilizing effectively eliminated N2O fluxes in both the amino acid treatments, and the NH4+ treatment (Table 1). However, this was not the case when NH2OH was applied, where emissions decreased by 72% (Table 1).

Table 1 Emission rates of total N2O (µg N2O-N m−2 h−1) of the inhibitor × N substrate treatments on Day 9.
Full size table

N2O-15N enrichment

The positive control (urea only at natural abundance) had a N2O-15N enrichment of 0.363 ± 0.004 (SD) on Day 9. At the same time, the addition of an N substrate resulted in small increases in the N2O-15N enrichments in all treatments with the following exceptions (Table 2): the phenylalanine treatment with either no inhibition or bacterial inhibition, and the NH4+ treatment with bacterial inhibition (Table 2). Within a given N substrate treatment, when comparing the N2O-15N enrichment of the no inhibition treatment and a specific inhibitor treatment, few treatment differences occurred. Under glycine only the sterilized soil treatment varied, with a higher N2O-15N enrichment relative to the no inhibition treatment (Table 2). Applying phenylalanine also resulted in enhanced N2O-15N enrichment, mostly when applied to the sterilized soil but this was not statistically different from the no inhibition treatment (Table 2). With NH4+ as the N substrate the N2O-15N enrichment was again highest in the sterilized soil treatment, but none of the inhibitor treatments caused N2O-15N enrichment to differ from the no inhibitor treatment (Table 2). The biggest shifts in N2O-15N enrichment with inhibition treatments occurred in the NH2OH treatment where applying bacterial inhibition, either alone or within the combined inhibition treatment, caused significant decreases in N2O-15N enrichment relative to the no inhibition treatment (Table 2).

Table 2 N2O 15N enrichment (atm%) of the inhibitor × N substrate treatments on Day 9, 24 h after the treatment application.
Full size table

N2O codenitrification

Increased 15N enrichment of the N2O fluxes revealed the formation of hybrid N2O (codenitrified N2O (N2Oco)). Amino acid and NH4+ treatments emitted 13–17 µg N2Oco-N m−2 h−1 in the case of no inhibition, while bacterial inhibition and/or fungal inhibition lowered these fluxes by >30% (Table 3). With sterilized soil under these N substrate treatments codenitrification fluxes ceased (Table 3). The N2Oco fluxes from the NH2OH treatment decreased significantly in the presence of the combined inhibition (>46%, Table 3) but not when applied individually. Under NH2OH, hybrid N2O fluxes equalled 3851 µg N2Oco-N m−2 h−1 with no inhibition present. Sterilizing the soil significantly lowered NH2OH derived codenitrification fluxes to 617 µg N2Oco-N m−2 h−1. This corresponded to a decrease of >83%, compared to the no inhibition treatment; or a decrease of >71%, compared to the combined inhibitor treatment (Table 3).

Table 3 Codenitrification fluxes (N2Oco, µg N2O-N m−2 h−1) of the inhibitor × N substrate treatments on Day 9, 24 h after the treatment application.
Full size table

Discussion

The hydrolysis of urea and its resulting products increases NH4+ and OH− concentrations in the soil5 with the latter responsible for the elevated soil surface pH observed in treatments containing urea. Urea application elevated soil NH4+-N concentrations, as evidenced by the higher concentrations in the positive control when compared with the negative control. Elevated soil pH will have resulted in the NH4+/NH3 equilibrium shifting towards NH35. However, by Day 8 the concentration of NH3 will have been relatively low based on soil pH values at this time5. While NH3 can inhibit NO2− oxidisers under urea-affected soil9,10 the elevated soil NO3−-N concentrations at the end of the experiment and the decline in NO2− from Day 1 to 7 demonstrates NO2− oxidisers were functioning. The soil NO3−-N concentration on Day 9 was higher when compared to a previous study by Rex et al.33, at a similar time following urea application. This higher soil NO3−-N concentration is likely to have occurred due to the reduced potential for nitrifier inhibition9,10, a consequence of the lower urea-N rate used in the current study. Considering the soil pH and inorganic-N dynamics it can be concluded that the application of urea was representative of conditions under a typical urine patch41,42, and that the N substrate treatments were applied during a period of relatively rapid inorganic-N transformation.

The rapid increase in N2O fluxes following inhibitor application was partially the result of physically mixing the soil in order to distribute the inhibitors, which resulted in entrapped N2O, in the soil, being released43. Furthermore, soil, not previously exposed to oxygen, would have become exposed and thus there is also the possibility that inhibition of N2O reductase occurred, preventing complete denitrification44. However, the application of substrate-N also contributed to the N2O flux as demonstrated by the increased N2O-15N enrichments, particularly in the case of the NH2OH treatment (Fig. 1a).

Soil N2O emissions are strongly driven by the presence and turn-over of NO2− which is the ‘gate-way molecule’ for N2O production9,45. In the current study soil NO2− concentrations were elevated on Day 9 but at concentrations lower than previously observed (e.g. Clough et al.31) due to the lower urea application rate in the current study preventing NH3 inhibition of NO2− oxidation45. Hence, the ensuing N2O emissions most likely result from the net effects of microbial processes utilising NO2− and/or the N substrate added.

The effects of the microbial inhibitors, cycloheximide, streptomycin and heat sterilization on N2O production were assessed 12 h after inhibitor application since maximum efficacy is reported within 24 h of application46. The decline in the N2O fluxes following fungal inhibition within the amino acid and NH2OH treatments demonstrates fungal mechanisms were responsible for a portion of the N2O produced (21–46%). Previous studies have shown fungi are able to produce N2O32,33,47,48. Nitric oxide reductase (P450nor), is a key feature of fungal denitrification and has been observed to require hypoxia and either NO3− or NO2− substrate to generate N2O47,49: these conditions occurred within the current study. Biotic N2O emissions from non-autoclaved soil suspensions can be stimulated by the presence of both NH2OH and NO3−, as was the case in the NH2OH treatment of the current study. Thus, the decline in N2O emissions in the NH2OH treatment, with fungal inhibition, implies a fungal mechanism was partially responsible for the N2O flux, via NH2OH utilisation.

With bacterial inhibition, the decline in the N2O flux under the NH2OH treatment likely occurred due to the bacterial inhibitor preventing the function of the ammonia oxidising bacteria (AOB), which utilise NH2OH to gain energy50. Increased mRNA transcription levels of the functional genes present in AOB that encode for NH2OH oxidoreductase (haoA), and the reductases for NO2− and NO, which are nirK and norB, respectively, become elevated following NH2OH application50. A similar result and explanation might have been expected following bacterial inhibition in the NH4+ treatment, given that NH2OH is an intermediate in the nitrification pathway, however the result was not statistically significant (Table 1). Lower N2O fluxes from the glycine treatment under bacterial inhibition may have also resulted from a diminished nitrification rate of the NH4+ derived from the mineralized glycine-N, and thus delivering less NO2− to the soil pool. However, this did not occur under the phenylalanine treatment possibly because it is a more complex molecule and potentially slower to be mineralized, and thus potentially bacteria played less of a role in the N2O fluxes derived from phenylalanine. Again, with glycine the combined inhibition treatment demonstrated the role of fungi in generating N2O. This was also the case with phenylalanine where the combined inhibition cut N2O emissions to a level comparable to fungal inhibition alone.

The near complete suppression of N2O emissions in the amino acid and NH4+ treatments, under the combined inhibition treatment, demonstrates that the observed N2O fluxes were almost entirely from biologically driven processes. As previously shown, from the δ13C signatures of respired amino acid–CO2-C, amino acids are readily mineralized, forming NH4+51. Consequently, amino acids will contribute to N2O fluxes if this NH4+ is nitrified, or via the denitrification of the nitrification products51. The residence time of amino acids in soils is generally reported in hours and depends on soil type51,52,53. However, the lack of a significant N2O flux response to amino acid and NH4+ substrate additions at Day 9, relative to the positive control (Fig. 1), is most likely due to the large background NH4+ pool present at the time of N substrate addition, derived from the urea addition. Hence, the NH4+ formed from either amino acid mineralization or direct NH4+ addition will have been diluted by at least 10-fold, assuming all substrate-N was immediately available. Furthermore, it is likely other amino acids were also present to further dilute the amino acid additions. For example, after extracting three soils McLain and Martens51 found the sum of 18 amino acids to range from 9 to 20 g kg−1 of soil, when examining an arid grassland (Well-drained Typic Torrifluvents of the Pima series). In contrast to the soil used in this study, these amino acid concentrations referred to a non-irrigated soil with an expected lower microbial abundance.

With the exception of NH2OH, the near-zero N2O emissions after applying the N substrates to the sterilized soils indicated that the N2O fluxes were dominated by biotic processes. This was not the case for NH2OH where the N2O flux from the sterilized soil was ∼28% that of the no inhibition treatment. It has previously been shown that the NH2OH molecule may decompose abiotically to produce N2O50,54,55,56.

The lack of any corresponding shifts in the relatively low 15N enrichments of the N2O evolved from the amino acid treatments, under the various inhibition treatments, suggests fungi were not directly utilising the amino acids for N2O production. The codenitrification product depends on the redox state of the N-donor, and prior studies have shown amines (-R-NH2) to be codenitrified to N247. Thus, the lack of any corresponding shifts in the relatively low 15N enrichments of the N2O evolved from the amino acid treatments may have also been the result of N2 being produced. Despite this, fungal inhibition lowered amino acid derived codenitrified N2O (Table 3), indicating that products derived from the amino acid mineralization are involved in fungal codenitrification. The lack of any bacterial inhibition effect on the codenitrification flux demonstrates the dominant role of fungi in codenitrification33.

Increasing 15N enrichment of the N2O molecule demonstrates that the N2O-N partially derives from a 15N enriched source. In the case of the NH2OH, applied with an enrichment of 98 atom% 15N, the highly 15N enriched N2O emissions demonstrate the applied NH2OH contributed strongly to the evolved N2O flux.

Using soil suspensions Spott and Stange57 concluded N2O production from NH2OH in soil was complex due to the interaction of production pathways involving both abiotic formation and biogenic formation, resulting from both codenitrification and denitrification. Adding the NH2OH substrate to the sterilized soil (abiotic conditions) the 15N enrichment of the N2O (∼44 atom%) aligned closely with the calculated 15N enrichment of 49 atom% that indicates hybrid N2O production via abiotic N-nitrosation. The formation of N2O via NH2OH reacting with NO2− occurs due to abiotic nitrosation processes58, and has been previously observed in sterilized soils56. The NH2OH compound has also been reported to decay abiotically to form N2O with the process slowed down when NO2− is preesent58. However, had this been the main process for N2O formation the 15N enrichment of the N2O evolved would have aligned more with the applied NH2OH-15N enrichment. The combined inhibition treatment significantly decreased the N2O codenitrification flux by 50% (Table 3) compared to the no inhibition treatment (Table 2) indicating abiotic reactions were also contributing substantially to the observed N2O flux.

Fungi contributed to N2O production when NH2OH was applied, as indicated by the flux decrease under the fungal inhibition treatment, however, the lack of any change in the N2O-15N enrichment indicates fungal inhibition was not affecting the process generating 15N enriched N2O. Conversely, the further decrease in both the N2O flux and N2O-15N enrichment in the bacterial inhibition and the combined inhibition treatments, showed that the N2O production process was inhibited, and that less 15N enriched NH2OH contributed to the N2O flux produced. Therefore, the codenitrification flux also tended to decline in the presence of the bacterial inhibitor. Bacterial inhibition diminishes, amongst others, the activity of AOB and thus (i) lowers the consumption of NH2OH via bacterial nitrification, (ii) lowers the enrichment of the nitrification products derived from 15N enriched NH2OH, and thus (iii) the formation of 15N enriched nitrification intermediaries NO2− and NO declines. Since NO2− and NO have been shown be involved in codenitrification, decreases in the concentration of these molecules would lead to lower N2O fluxes with lower 15N enrichment. Furthermore, had 15N enriched NH2OH progressed to NO2− then any denitrification of this NO2− that contributed to the 15N enriched N2O pool, would also have occurred at a slower rate or been prevented with inhibition of bacterial denitrifiers.

Conclusions

Codenitrification occurs when N-donors, such as those studied here (NH4+, glycine, phenylalanine and NH2OH) react with a nitrosyl compound, to form hybrid N2O. Using selective microbial inhibition treatments, and simulating a ruminant urine patch environment, we demonstrated that all the used 15N-labelled N substrates contributed to codenitrification in a soil matrix. Hydroxylamine was the most important N substrate with respect to increasing the N2O flux and contributing to codenitrification (85.7% of total flux), likely because of its more reactive character compared to the other N substrates. The codenitrification N2O fluxes following amino acid-15N addition were orders of magnitude lower (0.7–1.2% of total flux), potentially due to dilution from antecedent amino acids or their break down products, which in turn means that a contribution of these natural amino acids could be assumed under the experimental conditions. Fungal inhibition resulted in a significant decline in the formation of amino acid derived codenitrification fluxes, underlining once more the importance of fungal codenitrification vs. bacterial codenitrification. The relatively lower codenitrification N2O fluxes with amino acids may also be a result of the microbial community structure that is present20. Alternatively, codenitrification of NH2OH to form N2O has been reported in the absence of organic electron donors59 hence, given that codenitrification is in principle dependent on organic carbon respiration a lack of organic substrate or variations in its form may have favoured codenitrification of NH2OH20. The results of this study, demonstrated that codenitrification occurs via multiple pathways in a pasture soil following a simulated bovine urine event. Codenitrification resulting from the presence of NH2OH is likely to be the dominant process, in the short-term following the deposition of ruminant urine with its relatively high urea-N loading. The results warrant further in situ investigation of the dynamics of potential N-donors, in conjunction with N2O fluxes, under ruminant urine patches.

Materials and Methods

Experimental design

A bulked soil sample was taken from a sandy loam pasture soil on the Lincoln University dairy farm (0–10 cm), New Zealand (43°38′25.23″S, 172°27′24.71″E, Typic Immature Pallic Soil, (USDA: Udic Haplustept)). The pasture consisted of perennial rye grass (Lolium perenne L.) and white clover (Trifolium repens L.). Field moist soil was sieved (4 mm) to remove stones and plants and then placed into jars (250 mL, Ø 8.1 cm), corresponding to 100 g dry weight (ca. 82 cm3), and moistened to 50% of water-holding capacity33 (ca. 83% water-filled pore space).

Initially the jars, with soil, were placed in an incubator, in the dark, at 23 °C and wetted-up daily to preincubation weight. After four days, any germinated weed seedlings were removed and the experimental period of 14 days commenced (Day −2 to Day 11). An aqueous urea solution (500 µg urea-N g dry soil−1) was applied on Day 0 in order to simulate a bovine urine deposition event31,60. On Day 8, microbial inhibition treatments were applied with the N substrate treatments applied immediately after this in an aqueous solution (4 mL) as noted below.

Treatments consisted of 15N enriched N substrate species (glycine (98), L-phenylalanine (98), NH4+ (99) and NH2OH (98); atom% 15N enrichment in bracket) with each N substrate treatment further split into five microbial inhibition treatments (no inhibition, fungal inhibition, bacterial inhibition, fungal and bacterial inhibition (‘combined inhibition’) and soil total microbial inhibition (heat sterilised soil)). Treatments were replicated thrice. The amino acid-N concentrations were based on the findings of Scheller and Raupp39, and in order to apply a realistic concentration, these were applied at a rate of 20 µg N g−1 dry soil. Hydroxylamine and NH4+ were applied at equal N rates for comparative purposes.

According to Anderson and Domsch61 cycloheximide, a fungal inhibitor, was applied at a rate of 8 mg g−1 soil and streptomycin, a bacterial inhibitor, at a rate of 5 mg g−1 soil. Both chemicals were applied as a dry powder on to the soil surface and subsequently mixed into the soil with a spatula for 1 min. The combined inhibition included the simultaneous application of cycloheximide and streptomycin and was designed to inhibit both bacteria and fungi. Sterilizing (as complete microbial inhibition) was performed by heating the soil. This was achieved by microwaving the soil in the jars for 4 minutes, remoistening the dry soil, and then microwaving the jars for another 3 minutes, as microwave heating is a proven method to stop microbial activities62,63. Thereafter, the microwaved soils were readjusted to 50% water-holding-capacity and also mixed for 1 minute. The control treatment contained urea, but no inhibitors were applied, and the soil was mixed to replicate the physical disturbance of the other treatments. Immediately after application of the inhibitor treatments the N substrate treatments were applied according to treatment at a rate of 20 µg N g−1 dry soil, without subsequent mixing.

In addition, three further control treatments were set up; a positive control (soil with urea but no N substrate or inhibitor addition (n = 3), also physically mixed on Day 8; a negative control (n = 3) consisting of soil without urea, inhibitors, or N substrates, also physically mixed on Day 8; and a separate NO2− control (soil with urea but no N substrate addition, physically mixed on Day 8) for soil NO2−-N sampling at 4 different times over the duration of the experiment.

Gas sampling and analysis

On Day −2, −1, 0, 1, 2, 4, 6, 7, 8 (before inhibitor application), 9, 10 and 11, the jars were sealed with lids equipped with rubber septa. Jar headspace gas samples were taken with a plastic syringe, fitted with a three-way-stop cock and a 25G hypodermic needle, and injected into a previously evacuated Exetainer® vials (Labco Ltd., High Wycombe, UK). The first gas sample (12 mL) was taken immediately after sealing the jar headspace. The second gas sample was taken after 1 h, only from the positive control to verify the linearity of the increase in the headspace gas concentration, and the third gas sample was taken after a 2 h incubation time (12 mL, all jars). On Days 8, 9, 10 and 11, the third gas sample (30 mL), was split between a 6 mL Exetainer® that received 12 mL, and an evacuated and helium flushed 12 mL Exetainer® that received 18 mL for 15N-N2O determination.

Nitrous oxide concentrations were determined using a gas chromatograph (SRI-8610, SRI Instruments, Torrance, CA) coupled to an autosampler (Gilson 222XL; Gilson, Middleton, WI) equipped with a 63Ni electron capture detector64. PeakSimple 4.44 software (SRI Instruments, Torrance, CA) and several N2O standards (range 0–100 µL L−1,BOC, New Zealand) were used to determine the N2O concentrations. The N2O fluxes (µg N2O-N m−2 h−1) were determined using the following equation:

$${N}_{2}O\,flux=(\frac{V\times \Delta {N}_{2}O\times P}{R\times T})\times {m}_{N}\times {t}^{-1}\times {A}^{-1}$$

V = headspace volume (L). ΔN2O = change in headspace N2O concentration during sampling (µL L−1). P = pressure (atm). R = gas constant (0.08206 L atm K−1 mol−1). T = temperature (K). mN = mass of N per mole of N2O (g mol−1). t = time (h). A = soil surface area (m2).

The 15N enrichment of the N2O evolved was determined by analysing the gas samples with a continuous-flow-isotope ratio mass spectrometry CFIRMS (Sercon 20/20; Sercon, Chesire, UK) inter-faced with a TGII cryofocusing unit (Sercon, Chesire, UK). If required, gas samples were diluted by injecting 4 mL of sample gas into a helium-filled 12 mL Exetainer® (1:4 dilution).

The measured 15N concentration of the headspace N2 was close to natural abundance thus a determination of the N2 flux was not possible, hence, the N2 emissions were not considered further.

Codenitrification calculations

As previously reported20 conventional denitrification produces N2O (non-hybrid N2O) while N2O produced via codenitrification results in an N atom from NO2− and an N atom from a co-metabolised compound producing a hybrid N-N species, such as N2O. The following calculations determine the codenitrification flux, assuming that hybrid N2O only arises from codenitrification. We do not distinguish between the roles of biotic and abiotic reactions in this process. However, the use of biological inhibitors and soil sterilization indicate the relative roles of abiotic and biotic processes in producing hybrid N2O.

For the N2O evolved it was assumed that this was generated from one 15N enriched pool-fraction (d′D) with 15N enriched N (15N atom fraction q′D), and a fraction (d′N, equal to 1 − d′D) derived from a pool or pools at natural abundance (15N atom fraction q′N).

The ratios r’1 and r’2, were determined from the N2O m/z ion currents at m/z 44, 45 and 4665:

$${r^{\prime} }_{1}={}^{45}i/{}^{44}i$$
(1)
$${r^{\prime} }_{2}={}^{46}i/{}^{44}i$$
(2)

where, 44i, 45i and 46i represent the ion-currents of the N2O mass fractions 44, 45 and 46.

Then, following Arah65 (equations 22 and 23), the values of the 15N atom fraction of the sample (a′s) and the 46N2O component of the molecular fraction, of the N2O molecule, in the sample (x′s) were calculated using r′1 and r′2, while allowing for the presence of oxygen isotopes.

In Arah65 a′s and x′s are defined as follows:

$${{\rm{a}}^{\prime} }_{{\rm{s}}}=(1-{{\rm{d}}^{\prime} }_{{\rm{D}}}-{{\rm{d}}^{\prime} }_{{\rm{N}}})\ast {{\rm{a}}^{\prime} }_{{\rm{A}}}+{{\rm{d}}^{\prime} }_{{\rm{D}}}\ast {{\rm{a}}^{\prime} }_{{\rm{D}}}+{{\rm{d}}^{\prime} }_{{\rm{N}}}\ast {{\rm{a}}^{\prime} }_{{\rm{N}}}$$
(3)
$${{\rm{x}}^{\prime} }_{s}=(1-{{\rm{d}}^{\prime} }_{{\rm{D}}}-{{\rm{d}}^{\prime} }_{{\rm{N}}})\ast {{{\rm{a}}}^{2^{\prime} }}_{{\rm{A}}}+{{\rm{d}}^{\prime} }_{{\rm{D}}}\ast {{{\rm{a}}}^{2^{\prime} }}_{{\rm{D}}}+{{\rm{d}}^{\prime} }_{{\rm{N}}}\ast {{{\rm{a}}}^{2^{\prime} }}_{{\rm{N}}}$$
(4)

When letting d′N equal (1 − d′D) and a′A equal the 15N enrichment at natural abundance (0.003663) Eqs 3 and 4, when set to equal zero, become:

$$0={{\rm{d}}^{\prime} }_{{\rm{D}}}\ast {{\rm{a}}^{\prime} }_{{\rm{D}}}+(1-{{\rm{d}}^{\prime} }_{{\rm{D}}})\ast 0.003663-{{\rm{a}}^{\prime} }_{{\rm{s}}}$$
(5)
$$0={{\rm{d}}^{\prime} }_{{\rm{D}}}\ast {{{\rm{a}}}^{{\rm{2}}^{\prime} }}_{{\rm{D}}}+(1-{{\rm{d}}^{\prime} }_{{\rm{D}}})\ast {0.003663}^{2}-{{\rm{x}}^{\prime} }_{s}$$
(6)

Since a′s and x′s are known the values of d′D and a′D can be determined using the Solver function in Microsoft ExcelTM, while setting the target value at zero, with the result accepted when the target value is <1 × 10−5.

Then the codenitrification flux was calculated according to Clough et al. (2001) as:

$${{\rm{d}}}_{{\rm{CD}}}=-\,\Delta {}^{45}{\rm{R}}{p}_{1}^{2}/(\,-\,{\Delta }^{45}{\rm{R}}{p}_{1}^{2}+\Delta {}^{45}{\rm{R}}{p}_{1}{p}_{2}+{q}_{1}{p}_{2}-{q}_{2}{p}_{1})$$
(7)

were dCD is the fraction of N2O within the headspace derived from codenitrification and Δ45R is the 45N2O/44N2O ratio, while p1 (0.9963) and q1 (0.0037) are fractions of 14N and 15N in the natural abundance pool, and where q2 equals a’D, derived above, with p2 equal to 1 − q2.

Finally the codenitrification flux was determined as:

$${{\rm{N}}}_{2}{{\rm{O}}}_{{\rm{CD}}}={{\rm{d}}}_{{\rm{CD}}}\times ({\rm{total}}\,{{\rm{N}}}_{2}{\rm{O}}\,{\rm{flux}})$$
(8)

Surface pH and inorganic-N measurement

Surface pH was measured on Days −2, 0, 1, 3, 5, 7, 9 and 11, by adding one drop of deionised water to the soil surface and then placing a flat surface pH probe (Broadley James Corp., Irvine, California) onto the soil surface.

The NO2− concentration in the unmixed NO2− control (soil + urea solution) was determined by subsampling soil with a corer (diameter 1.6 cm, depth 1.5 cm). The soil was then blended with 2 M potassium chloride (KCl), adjusted to pH 8 with potassium hydroxide66 at a 1:6 ratio. This procedure was performed on Days 1, 4, 6 and 10.

Subsamples of moist soil (4 g dry weight) were taken after Day 11, from the positive and negative controls, and extracted with 2 M KCl in order to determine the NH4+ and NO3− concentrations at the end of the experiment67,68. Inorganic-N concentrations in the extracts were determined using Flow Injection Analysis67.

Statistics

The single jars were defined as experimental units by the independent applications of treatments. The experiment focussed on achieving the most sensitive test of treatment differences and inference is not claimed for a population wider than the paddock, used for sampling. All statistical analyses were performed using SigmaPlot 13.0 (Systat Software Inc., Chicago). For each variable of interest a general linear model (ANOVA equivalent) was fitted with N substrate treatment or a factorial combination of N substrate treatment and inhibition method as explanatory variables. Using this method, the different inhibition treatments within each N substrate treatment were compared. Tests for normality (Shapiro-Wilk test) and variance (Brown-Forsythe test) were used to evaluate the residuals and define the most powerful test for each comparison of means. Hence, means comparisons were adjusted for multiplicity using Tukey, Holm-Sidak, Dunn’s or Student’s t-test adjustments to p values.