Influence of soil moisture on codenitrification fluxes from a urea-affected pasture soil

Intensively managed agricultural pastures contribute to N2O and N2 fluxes resulting in detrimental environmental outcomes and poor N use efficiency, respectively. Besides nitrification, nitrifier-denitrification and heterotrophic denitrification, alternative pathways such as codenitrification also contribute to emissions under ruminant urine-affected soil. However, information on codenitrification is sparse. The objectives of this experiment were to assess the effects of soil moisture and soil inorganic-N dynamics on the relative contributions of codenitrification and denitrification (heterotrophic denitrification) to the N2O and N2 fluxes under a simulated ruminant urine event. Repacked soil cores were treated with 15N enriched urea and maintained at near saturation (−1 kPa) or field capacity (−10 kPa). Soil inorganic-N, pH, dissolved organic carbon, N2O and N2 fluxes were measured over 63 days. Fluxes of N2, attributable to codenitrification, were at a maximum when soil nitrite (NO2−) concentrations were elevated. Cumulative codenitrification was higher (P = 0.043) at −1 kPa. However, the ratio of codenitrification to denitrification did not differ significantly with soil moisture, 25.5 ± 15.8 and 12.9 ± 4.8% (stdev) at −1 and −10 kPa, respectively. Elevated soil NO2− concentrations are shown to contribute to codenitrification, particularly at −1 kPa.

denitrification. It has been suggested that codenitrification results from microbially mediated N-nitrosation reactions [14][15][16] . Codenitrification is one of the least studied N loss pathways and its contribution to agricultural N 2 O and N 2 emissions remains unclear 17 .
Codenitrification is a process that co-metabolises organic N compounds, such as amines, to produce N 2 O and/or N 2 , and is also referred to as biotic N-nitrosation 16 . Codenitrification involves the replacement of a hydrogen atom in an organic compound with a nitroso group (-N=O). Under near neutral to alkaline soil pH conditions, common to pasture soils, codenitrification may occur via enzymatic catalysis ( Fig. 1), with enzymatic nitrosyl compounds (E-NO + or E-NO) attracting nucleophilic compounds 16,18 . Nucleophiles involved in codenitrification include hydroxylamine, ammonium (NH 4 + ), hydrazine, amino compounds, and ammonia (NH 3 ). The resulting gas products formed, N 2 O or N 2 , contain one N atom originating from the inorganic-N (e.g. NO 2 − ), and a second atom from the co-metabolised organic compound 16,18 . Significant rates of both partial and complete codenitrification are only likely to occur if nucleophile concentrations are at least one or two orders of magnitude greater than that of NO 2 − and NO 16 . Heterotrophic denitrification results in the reduction of NO 3 − to N 2 with nitrite (NO 2 − ), nitric oxide (NO), and N 2 O obligate intermediaries 19 (Fig. 1). Formation of the N 2 O molecule is recognized as occurring via parallel or sequential pathways 16 and references therein. In the parallel pathway simultaneous bonding of two NO 2 − or two NO molecules to an enzyme, where both NO 2 − and NO are derived from the same NO 3 − source, creates a non-hybrid N-N bond, thus precluding the occurrence of codenitrification 16 . However, a two-step reaction, the sequential pathway, results in either NO 2 − or NO initially bonding with an enzyme, which in turn may react with either free NO 2 − or NO to form a non-hybrid N-N bond, or alternatively, this enzyme bound N can act as an electrophile and react with nucleophiles (e.g. amines) to form a hybrid N-N bond (Fig. 1). Consequently, hybrid N-N gas production, codenitrification, can occur simultaneously as a result of conventional denitrification ( Fig. 1) 16 . Formation of hybrid N 2 has also been reported to occur when NH 3 , hydrazine (N 2 H 4 ) or amines are co-metabolised during codenitrification 20 .
Abiotic nitrosation is also a well-recognized phenomena 21,22 . In abiotic reactions, free NO 2 − derived from nitrification or denitrification processes is chemically transformed to produce the nitrosonium cation (NO + ) under acidic conditions. The NO + cation reacts with a nucleophile (e.g. amine) to produce a hybrid N-N linkage ( Fig. 1) 16 and references therein. This process differs from codenitrification since the formation of the NO + electrophile is chemically dependent on the soil pH and involves free NO 2 − in the soil solution as the precursor. Nucleophiles involved in abiotic reactions include hydroxylamine, NH 4 + , hydrazine, amines, and NH 3 . However, relatively high soil pH values under grazed pasture conditions mean that the equilibrium concentrations of free nitrosating agents are generally inadequate for abiotic nitrosation to be significant 16 .  16 , Weeg-Aerssens et al. 18 , Schmidt et al. 55 ) showing abiotic denitrification, parallel denitrification, sequential denitrification and codenitrification pathways. During abiotic production an electrophile (e.g. the nitrosonium cation NO + which is formed under acidic soil conditions) replaces the hydrogen atom of a nucleophile with a hybrid N-N bond formed following deprotonation. The parallel pathway results in a non-hybrid N-N bond as the result of two NO 2 − or two NO molecules being bound, simultaneously to one enzyme (E), which theoretically excludes the possibility of a nitrosation reaction occurring and the formation of a hybrid N-N bond 55,56 . However, a two-step process occurs in the sequential pathway when NO 2 − or NO molecules initially bind to an enzyme (E) followed by a free NO 2 − , or NO molecule, (originating from the original NO 3 − pool) reacting with the enzyme complexed N species to form a non-hybrid N-N bond. The two-step sequence also permits the enzyme complexed N species to function as an electrophile which is able to be to be attacked by nucleophiles producing a hybrid N-N bond. Nucleophiles able to partake in codenitrification reactions include amines, ammonium, hydrazine, and ammonia.
In grazed pastures ruminant urine deposition onto pasture soil temporarily elevates soil pH following urea hydrolysis, creating a urinary-N cascade that produces potential nucleophiles (e.g. NH 4 + and NH 3 ) at high concentrations. Simultaneously, enzyme bound nitrosating agents (E-NO + or E-NO), may be formed during denitrification of nitrate (NO 3 − ) or as supplied by NO 2 − or NO during processes such as nitrification of nitrifier-denitrification 19 . Thus urine patches are potentially conducive to codenitrification occurring. In the only in vivo study to date to focus on codenitrification, Selbie et al. 23 confirmed the occurrence of codenitrification within ruminant urine-affected pasture soil with 95% of the N 2 emitted over 123 days resulting from codenitrification, with N 2 the dominant product, and where the codenitrified N 2 was equivalent to 56% of the N applied. This experiment by Selbie et al. 23 received regular rainfall and it may be that the dominance of codenitrified N 2 over codenitrified N 2 O may have been the result of, as the authors suggest, hybrid N 2 O being converted to hybrid N 2 via heterotrophic denitrification (Fig. 1). Conceptually, the recognized environmental constraints on denitrification should also apply to codenitrification 16 , since codenitrification depends on enzyme bound nitrosyl compounds, formed during denitrification, being present ( Fig. 1). A key driver of denitrification is the soil's oxygen status, and wetter soils result in higher levels of anaerobiosis since oxygen diffuses 1 × 10 4 times slower through water when compared to air 24 . Thus wetter soils should have higher rates of codenitrification. In order to test this hypothesis, and better understand the constraints and importance of codenitrification in pasture soils, we performed an experiment using either saturated soil or soil at field capacity to determine relative rates of codenitrification. The objective of the study was to investigate the effect of soil moisture on the rate of codenitrification from simulated urine applied to a free draining permanent grassland soil.

Results
Soil moisture, pH, DOC and inorganic-N. The −1 kPa and −10 kPa moisture treatments imposed resulted in average WFPS values (%±s.e.m) of 88.9 ± 1.1 and 48.5 ± 0.4, respectively. The relative gas diffusivity values at −1 and −10 kPa were 0.0028 and 0.2079, respectively. There was a significant interaction of soil moisture and sampling date (p < 0.001) on soil pH, DOC and inorganic N contents (Figs 2-4). Soil pH in the non-urea treatment was generally constant over time (Fig. 2) regardless of soil moisture treatment, averaging 5.49 ± 0.11 (Stdev). However, soil pH (p < 0.001) increased within 6 hours of urea application, and increased further, peaking at 8.57 ± 0.29 and 8.78 ± 0.09 in the −1 kPa and −10 kPa treatments, respectively, on day 3 before declining over time (Fig. 2). On days 21 and 35 the soil pH was lower in the −1 kPa treatment than in the −10 kPa treatment (p < 0.001) with the reverse occurring on day 63 (p < 0.05).
Soil DOC was higher (P < 0.001) under the urea treatment throughout the experiment (Fig. 3) and within the urea treatment soil DOC concentrations were significantly lower at −1 kPa than at −10 kPa from day 3 to day 62 (Fig. 3). In the urea treatment soil DOC correlated strongly with soil pH at both −1 kPa (r = 0.79; p < 0.001) and −10 kPa (r = 0.89; p < 0.001).
Soil NH 4 + -N concentrations increased following urea application ( Fig. 4), peaking at day 3 and then declining over time with a faster rate of decline in the −1 kPa treatment from day 14 (p < 0.05) such that soil NH 4 + -N concentrations were lower at −1 kPa on days 35 and 63 (Fig. 4). The 15 N enrichment of the NH 4 + -N in the urea treatment declined from 44 to 37 atom% over the experiment with higher 15 N enrichment on days 14, 21 and 35 in the −10 kPa treatment (Fig. 5). Concentrations of NO 2 − -N increased from day 7 under the urea treatment and peaked at day 21, with more NO 2 − -N present in the −1 kPa treatment, prior to returning to background levels at day 35 (Fig. 4). Concentrations of NO 2 − -N, extracted from the urea treatment, were only sufficient for 15 N enrichment determinations on days 14 and 21, where the 15 N enrichment was higher (p < 0.05) at −1 kPa than at −10 kPa on day 14, with no differences on day 21 (Fig. 5). Soil NO 3 − -N concentrations also began to increase at day 7 under the urea treatment and were consistently higher (p < 0.001) in the −1 kPa treatment on days 14 and 21. Soil NO 3 − -N concentrations peaked on day 35, before they declined to be less than those observed in the    (Fig. 6). Adding urea at −10 kPa caused N 2 O-N fluxes to increase steadily from day 12 until they peaked at day 30 (449 μg m −2 h −1 ) where after they steadily declined to <10 μg m −2 h −1 by day 51 (Fig. 6). The highest N 2 O-N fluxes were observed at −1 kPa with urea addition, where a rapid increase in the flux occurred peaking at 11,603 μg m −2 h −1 on day 2, followed by a rapid decrease to 163 μg m −2 h −1 by day 7. Then the flux gradually increased until day 35 (9220 μg m −2 h −1 ) whereupon it too decreased to be 476 μg m −2 h −1 by day 61 (Fig. 6).
Soil moisture treatment influenced cumulative N 2 O-N fluxes (p < 0.001) with total emissions of 0.08 and 2.26 g N 2 O-N m −2 at −10 and −1 kPa, respectively, when averaged over plus and minus urea treatments. Similarly, application of urea increased cumulative N 2 O-N fluxes (p < 0.001) from 0.10 to 2.25 g N 2 O-N m −2 when averaged over soil moisture treatments. An interaction between soil moisture and N application (p < 0.002) resulted in higher cumulative N 2 O-N fluxes at −1 kPa when urea was applied equal to 3.99 g m −2 ( Table 1). The N 2 O-N emission factors for the urea-N applied, allowing for non-N fluxes equated to 4.14% and 0.18% of N applied at −1 kPa and −10 kPa, respectively.  Upon urea application, the atom % 15 N enrichment of the N 2 O emitted at −1 kPa increased steadily to reach a maximum value of 43.9 atom % 15 N on day 25 before declining at a relatively slow rate to a value of 36.3 atom % 15 N by day 59 (Fig. 7). With the exception of day 2, the atom % 15 N enrichment of the N 2 O emitted at −1 kPa was higher than that emitted at −10 kPa (P < 0.05) on any given day. At −10 kPa the atom % 15 N enrichment of the N 2 O flux was observed to increase abruptly at day 12, reaching a maximum of 32.8 on day 30 and thereafter declining relatively abruptly to remain at ca 10 atom % 15 N (Fig. 7). − concentration after day 35 coincided with higher denitrification fluxes. At −10 kPa denitrification fluxes were highest after the initial wetting up following treatment application where after they generally declined (Fig. 8). Consequently, cumulative denitrification as N 2 was higher (p = 0.055) at −1 kPa, totaling 8.61 g N m −2 , than at −10 kPa where observed fluxes were 1.98 g N m −2 (Fig. 8).

Discussion
Inorganic-N pools and 15 N enrichment. Following urea application to the soil the ensuing hydrolysis produces NH 4 + and bicarbonate (HCO 3 − ) ions. The HCO 3 − ions are further hydrolysed to produce hydroxide ions (OH − ) and carbon dioxide 25 and it is this second hydrolysis reaction that generated the observed increase in soil pH under the urea treatments (Fig. 2). Elevated soil pH also influences the equilibrium between NH 4 + and ammonia (NH 3 ): as soil pH becomes elevated (>7.0) concentrations of NH 3 26,27 . Thus the slower decline in the NH 4 + concentration observed in the −10 kPa treatment, under urea, may have been due to NH 3 inhibition of nitrification. In favour of this were both the relative gas diffusivity of the soil being 2 orders of magnitude higher at −10 kPa, which would have facilitated NH 3 diffusion through the soil, and the soil pH remaining higher for longer (Fig. 2). The latter would have promoted the presence of NH 3 for longer. A slower rate of decline in soil pH at −10 kPa also demonstrates nitrification was slower, since nitrification results in the net release of H + ions 19 . Further evidence to support a slower rate of NH 4 + oxidation can be found in the slower rate of increase in ammonium oxidizing bacteria (AOB) gene and transcript abundance 28 .
Elevated soil NO 2 − concentrations resulted from nitrification of NH 4 + and their increase, from day 5 until day 20, occurred over a period when soil pH was sufficiently high to result in NH 3 generation. Ammonia toxicity acts more strongly on nitrifier NO 2 − oxidation than nitrifier NH 4 + oxidation 29 . It has been shown that solution-phase NH 3 (slNH 3 ) inhibits NO 2 − oxidation, as evidenced by strong relationships between cumulative slNH 3 and cumulative NO 2 − and static copy numbers of the nxrA gene, which is associated with nitrite oxidoreductase, and as a consequence soil NO 2 − is strongly correlated with N 2 O production 29 . The high N 2 O fluxes that occurred, between ca. days 7 to 35, at both −1 and −10 kPa under urea, where the soil NO 2 − concentrations were elevated strongly demonstrates this, and it can be assumed slNH 3 induced NO 2 − toxicity lead to the ensuing N 2 O emissions. The higher NO 3 − concentrations observed under urea on days 14 and 21 at −1 kPa were a consequence of the more rapid nitrification rates in this treatment, while the lower NO 3 − concentration in this treatment observed at day 63 resulted from higher denitrification induced losses of NO 3 − , which is further supported by the increase in soil pH under this treatment, since denitrification results in a net release of OH − ions 19 .
The 15 N enrichment of the NH 4 + pool, under urea, shows that it was predominantly derived from the urea applied, regardless of soil moisture treatment. The fact the NH 4 + pool 15 N enrichment was initially ca. 5 atom% lower than the urea solution applied was likely due to the release of NH 4 + as a consequence of the high soil pH solubilising soil organic matter, as demonstrated by the elevated DOC concentrations under the urea treatment. Solubilisation of soil organic matter is routinely observed following urine or urea application to soil 30 15 N enrichment at −10 kPa, when compared to −1 kPa, further supports the fact there was a slower rate of nitrification at −10 kPa. The increase in the NO 3 − pool 15 N enrichment over time, in both the −1 and −10 kPa treatments, demonstrates the NO 3 − pool was initially dominated by antecedent soil NO 3 − as in fact occurred (Fig. 4c).  31 , are due to the chemically induced anoxia that results from the hydrolysis reactions generating both NH 3 and CO 2 , as demonstrated in situ 32 . Such high fluxes were not observed at −10 kPa during this period because the higher relative gas diffusivity of the soil at −10 kPa ensured the soil was not as anaerobic.  15 N enrichment of the NO 2 − pool may possibly have arisen due to the method of treatment application where, in the −10 kPa treatment the urea solution infiltrated further and contacted a greater soil volume than at −1 kPa, as evidenced by the greater release of DOC at −10 kPa (Fig. 3), and which would have resulted in a more uniform NO 2 − pool. It is likely that, at −1 kPa, denitrification of antecedent NO 3 − occurred and that this generated sufficient NO 2 − to isotopically dilute the relatively small 15  − pool, and given the compatible conditions for denitrification, it can be assumed that denitrification of the NO 3 − pool dominated N 2 O production after day 30, and this assumption is supported by the elevated denitrification flux occurring after this time (Fig. 8).

N 2 denitrification and codenitrification of N 2 and N 2 O. As expected denitrification occurred at higher
rates under the more anaerobic moisture treatment as a result of the lower Dp/Do conditions promoting denitrification in the presence of NO 3 − substrate. The N transformations that ensued following urea hydrolysis, and hydrolysis itself, generated previously recognized codenitrification nucleophiles that include NH 4 + , NH 3 , and possibly organic-N compounds such as amines 16 . The latter might occur as a result of the dissolution of soil organic matter. While the enzymatically utilized NO 2 − and NO compounds, that form electrophiles, are generated during nitrification and denitrification 19 . Codenitrification N 2 O fluxes were generally low for both treatments, with measurable values mainly associated with the initial soil wetting. Conversely, codenitrification to N 2 was observed to peak on day 12, regardless of soil moisture, when NH 3 , NH 4 + and NO 2 − were all present at an elevated soil pH (≥7.70), and at relatively high concentrations. Thus it is possible that either NH 3 or NH 4 + were undertaking the role of the nucleophile at this time, since the elevated pH (>5.5) would have prevented any significant abiotic nitrosation occurring via NO + formation 16 . Recently, however, the formation of both N 2 O and N 2 , under both oxic and anoxic conditions, was reported in an in vitro experiment maintained at pH 6.2-6.9 where either live fungi or fungal necromass were incubated with glutamine and NO 2 − 33 . A subsequent isotope experiment with glutamine and 15 NO 2 − demonstrated the hybrid formation of N 2 after an incubation period of >7 days, again under either oxic or anoxic conditions 33 . Hence, based on this recent study, even though the soil in the current study was at a pH (≥7.70) sufficient to prevent acidic pathways of abiotic hybrid N-N bonds forming, we cannot rule out the possibility that abiotic reactions, under alkaline conditions, contributed to the codenitrification flux measured in the current experiment.
Production of N 2 O or N 2 via biotic codenitrification may result from the actions of archaea, bacteria or fungi. While archaea have been found to generate N 2 O through N-nitrosating hybrid formation 34 they are unlikely to have been the dominant mechanism in the current study since archaea are thought to prefer low N conditions 35,36 and urea addition resulted in lower ammonia oxidizing archaea gene copy numbers 28 . The codenitrification observed is most likely to be the result of fungi or bacterial activity. Delineation of the relative contributions made by fungi or bacteria to codenitrification is beyond the scope of the present study, however, future studies should aim to examine relative fungal and bacterial contributions.
Spott et al. 16 conceptualized that the recognized constraints on denitrification might also apply to codenitrification, and thus higher codenitrification fluxes might be expected under more anaerobic conditions. The current results support this concept: after day 30 the higher daily codenitrification fluxes under the more anaerobic (−1 kPa) soil moisture conditions, when at the same time denitrification fluxes were higher, resulted in higher cumulative codenitrification fluxes. This reinforces the fact that NO 2 − and or NO play a key role in the codenitrification process. The NO molecule has been observed to readily diffuse within the soil profile 37 , at relatively high concentrations, during denitrification and this would result in reactions with nucleophiles.
Unlike the results of Selbie et al. 23 codenitrification did not dominate the N 2 fluxes observed in the current study. This could be the result of the experimental system used in the current study differing to that used by Selbie et al. 23 . Differences include the lack of a pasture turf and associated microbiology and root exudation, the use of sieved repacked soil that may also have altered the fungal-bacterial community structure or activity as a result of sieving, constant soil moisture contents as opposed to wetting and drying events, and the lack of other climatic variables such as wind and rainfall.
SCIENTIfIC RepoRts | 7: 2185 | DOI:10.1038/s41598-017-02278-y In particular, fungal populations may have been reduced on sieving, and given that fungal P450 NOR is implicated in supplying enzyme bound nitrosating agents this could have had a significant influence on the results 38 . Given that enzyme bound nitrosating agents produced during denitrification may also consist of metal-nitrosyl complexes 16 any differences in soil Fe and Cu levels between studies may also explain the observed differences in codenitrification. Likewise, differences in the kinetic properties of different nucleophiles, combined with the ratio of NO or NO 2 − availability to nucleophile concentration, have also been shown to significantly impact on codenitrification/denitrification: lower K m and high nitrosyl donor/nucleophile ratios have been shown to reduce the level of codenitrification 15,20 .
This study confirms the role of anaerobic soil conditions in enhancing codenitrification fluxes under ruminant urine/urea deposition. It also demonstrates for the first time that high levels of NO 2 − , or other transitional N compounds ensuing from NO 2 − , that may occur during nitrification, are also able to contribute to codenitrification processes. To progress knowledge of codenitrification in grazed pastures more detailed studies are now required to both identify the microbial pathways operating and the relative importance of the possible nucleophiles and nitrosating agents that occur in grazed pastures.

Materials and Methods
Soil collection and experimental design. Soil was collected in early spring (March) from a permanently grazed dairy pasture at the Teagasc Moorepark Research Centre, County Cork, Ireland (8°15′W, 52°9′N). The top 5 cm of soil was removed and the A-horizon was sampled, 5-20 cm depth. Soil physical and textural characteristics are shown in Table 2. Cows had not grazed the pasture for over one month so recent urine deposition sites were avoided. The soil is classified as a Typical Brown earth from the Clashmore Series 39 , or as a Haplic Cambisol in the World Reference Database 40 . Field moist soil was then bagged and shipped to Lincoln University, New Zealand, following appropriate biosecurity protocols. It was then sieved (≤2 mm) to remove any stones, plant roots or earthworms. Sieved soil, with a gravimetric water content (θg) of 0.24 g water g −1 soil, was then packed into stainless steel rings (7.3 cm internal diameter, 7.4 cm deep) to a depth of 4.1 cm at a bulk density of 1.1 Mg m −3 , the latter simulating the in situ soil bulk density. This resulted in a total porosity of 0.58 cm 3 pores cm −3 soil. Packed soil cores were then arranged in a factorial experiment replicated four times.
Treatments consisted of two levels of soil moisture, −1 kPa and −10 kPa simulating 'near-saturation' and 'field-capacity' , respectively, and two levels of urea, (0 and 1000 kg N ha −1 ), replicated 4 times, with 7 destructive sampling times (112 cores in total). Preliminary tests showed that −1 and −10 kPa corresponded to 53% and 30% volumetric water content, or 91% and 52% water-filled pore space (WFPS). Soil cores were maintained at these water contents using tension tables 41 . Soil relative gas diffusivity values were calculated using the values for air-filled pore space and total porosity and the generalized-density corrected equation of Chamindu Deepagoda et al. 42 ; Equation 9b. It is recognized that artificial urine simulation does not generate identical effects to ruminant urine 43 , that urea contributes >70% of the total urine-N pool 6,44 , and that this N source is predominately responsible for the subsequent dynamics and transformations of organic and inorganic N in the soil under ruminant urine patches. Thus, in order to apply the N treatments, soil cores were wetted up on the tension tables to a point where there remained the capacity to add a further 10 mL of liquid, without inducing drainage. Subsequently, in the plus N treatment, 10 mL of a urea solution (42 g urea-N L −1 ; 50 atom%, Cambridge Isotope Laboratories Inc., USA) was slowly applied to the soil surface, to avoid drainage, to mimic an extreme bovine urine deposition event with a potentially high N 2 flux. Real urine could not be used since there was a need to have the urea-N highly enriched with 15 N to detect N 2 fluxes. In the nil N treatment 10 mL of deionized water was applied instead of a urea solution. Tension tables were maintained in a room with a mean temperature of 20 °C.

Soil chemical analyses.
After treatment application and throughout the experiment, on days 0, 3, 7, 14, 21, 35, and 63, soil inorganic N concentrations were determined by destructively sampling 16 soil cores (2 levels of urea × two levels of soil moisture × 4 replicates). Soil cores were fully extracted, homogenized, and a subsample was taken to determine θg: by drying the soil at 105 °C for 24 hours. A flat surface pH electrode was used to determine soil pH (Broadley James Corp., Irvine, California). Then further soil subsamples were extracted (equivalent of 10 g dry soil: 100 mL 2 M KCl shaken for 1 hour) and filtered (Whatman 42) to determine soil inorganic-N. The Gas flux determinations. Nitrous oxide and N 2 fluxes were regularly determined, from two days before until 63 days after treatment application using only the last batch of soil cores to be destructively analysed. This was performed by placing a soil core into a 1-L stainless steel tin fitted with a gas-tight lid and rubber septa.   Table 2. Physical and textural characteristics of soil sampled.
was taken for N 2 O-15 N enrichment and N 2 flux determination after 3 hours, after which cores were returned to the tension tables. Gas samples were taken using a 20-mL glass syringe fitted with a 3-way tap and a 0.5 mm by 16 mm needle and placed in either 6 mL vials for the N 2 O flux determinations or 12 mL vials for the N 2 O-15 N enrichment and N 2 flux samples (Exetainer; Labco Ltd., Lampeter, UK). An automated gas chromatograph (8610; SRI Instruments, Torrance, CA), coupled to an autosampler (Gilson 222XL; Gilson, Middleton, WI), was used to determine N 2 O gas concentrations in the samples, as previously described 49 . A continuous-flow-isotope mass spectrometer (Sercon 20/20; Sercon, Chesire, UK) inter-faced with a TGII cryofocusing unit (Sercon, Chesire, UK), was used to determine the 15 N enrichment of the N 2 O-N and N 2 -N gas samples 50 . The ion currents (I) at mass to charge ratios (m/z) of 44, 45, and 46 facilitated the calculation of the N 2 O molecular mass ratios 45 R ( 45 I/ 44 I) and 46 R ( 46 I/ 44 I). The N 2 O sources were subsequently allocated to either the fraction derived from the denitrifying pool (d' D ) of enrichment aD or the fraction derived from the pool or pools at natural abundance d' N = (1-d' D ) using the method of Arah (1997). The ion currents at m/z 28, 29, and 30 permitted the N 2 molecular ratios 29 R ( 29 I/ 28 I) and 30 R ( 30 I/ 28 I) to be quantified. Differences between the N 2 molecular ratios of the enriched and ambient atmospheres were expressed as Δ 29 R and Δ 30 R The N 2 flux was subsequently calculated using three methods: (i) The enrichment of the denitrifying pool ( 15 X N ) was calculated using Δ 29 R and Δ 30 R, and then the N 2 flux 51 , (ii) Using only the Δ 30 R data with the assumption that the enrichment of the denitrifying pool was aD 52 and the equation of Mulvaney 53 (iii) Using Δ 29 R and Δ 30 R to calculate the relative contributions of denitrification (N 2DN ), according to method (ii), and codenitrification (N 2CO ).
Increases in Δ 29 R and Δ 30 R may occur from denitrification but codenitrification contributes most to Δ 29 R where the ratio of Δ 29 R to Δ 30 R is always 272 54 . By assuming all Δ 30 R was the result of denitrification, method (ii), N 2DN was calculated. Then using the 'backsolver' facility in Microsoft Excel TM , the contribution of Δ 29 R to N 2DN was determined. The difference between the total measured value of Δ 29 R and Δ 29 R determined for N 2DN was assigned to codenitrification. Thus the fraction of the total number of moles of N 2 in the headspace, resulting from codenitrification (d CD ) were calculated as: where p 1 (0.9963) and q 1 (0.0037) represent the atom fractions of 14 N and 15 N in the natural abundance pool, respectively, and p 2 and q 2 are the atom fractions of 14 N and 15 N in the enriched NO 3 − pool, respectively, from which codenitrification is assumed to occur. Using the headspace volume of the sample chamber, corrected for standard temperature and pressure, the mass of N 2 -N in the headspace was determined with the amount derived from denitrification or codenitrification ascertained by multiplying by d D or d CD , respectively. Data analyses. Data were analysed using the Glimmix procedure within the SAS ® software version 9.4 (SAS, 2014). Cumulative results were analysed for the +N treatment only. For all other variables, analyses was as N treatment × moisture × day or moisture × day factorials. Any repeated measurements over time were modelled using correlation structures and spatial covariance was used to model the unequally-spaced time measurements. Residual checks were made and, where required, log transformation was used to correct for skew and non-constant variance. Multiplicity adjustments were made for simple effects within interactions, as interest was primarily in comparisons within time points.