Brachiaria humidicola Cultivation Enhances Soil Nitrous Oxide Emissions from Tropical Grassland by Promoting the Denitriﬁcation Potential: A 15 N Tracing Study

: Biological nitriﬁcation inhibition (BNI) in the tropical grass Brachiaria humidicola could reduce net nitriﬁcation rates and nitrous oxide (N 2 O) emissions in soil. To determine the effect on gross nitrogen (N) transformation processes and N 2 O emissions, an incubation experiment was carried out using 15 N tracing of soil samples collected following 2 years of cultivation with high-BNI Brachiaria and native non-BNI grass Eremochloa ophiuroide . Brachiaria enhanced the soil ammonium (NH 4+ ) supply by increasing gross mineralization of recalcitrant organic N and the net release of soil-adsorbed NH 4+ , while reducing the NH 4+ immobilization rate. Compared with Eremochloa , Brachiaria decreased soil gross nitriﬁcation by 37.5% and N 2 O production via autotrophic nitriﬁcation by 14.7%. In contrast, Brachiaria cultivation signiﬁcantly increased soil N 2 O emissions from 90.42 µ g N 2 O-N kg − 1 under Eremochloa cultivation to 144.31 µ g N 2 O-N kg − 1 during the 16-day incubation ( p < 0.05). This was primarily due to a 59.6% increase in N 2 O production during denitriﬁcation via enhanced soil organic C, notably labile organic C, which exceeded the mitigated N 2 O production rate during nitriﬁcation. The contribution of denitriﬁcation to emitted N 2 O also increased from 9.7% under Eremochloa cultivation to 47.1% in the Brachiaria soil. These ﬁndings conﬁrmed that Brachiaria reduces soil gross nitriﬁcation and N 2 O production via autotrophic nitriﬁcation while efﬁciently stimulating denitriﬁcation, thereby increasing soil N 2 O emissions.


Introduction
Nitrous oxide (N 2 O) concentrations in the atmosphere have increased by more than 20% since pre-industrial times and are responsible for 6% of current global warming [1]. N 2 O emissions are also an important factor in stratospheric ozone depletion [2], with agricultural soil accounting for approximately 66% of global anthropogenic N 2 O emissions, mainly due to the excessive input of synthetic N fertilizers [3,4]. The increasing use of synthetic fertilizers is also causing increased nitrifier activity, transforming modern agricultural systems into high-nitrifying environments [5,6].
Ammonia oxidation is the rate-limiting first step of nitrification, producing N 2 O as a by-product [7]. Biological nitrification inhibition (BNI) is a rhizospheric process whereby specific inhibitors exudated or released from the plant's roots suppress the activity of nitrifying bacteria [8]. This process is widely found in major crops, such as sorghum [9], rice [10], wheat [11,12] and maize [13], as well as in certain forage species [14] and trees [15]. Brachiaria humidicola, a tropical grass native to East and Southeast Africa, has a strong BNI capacity due to the release of the specific compound brachialactone in its root exudates [14,16]. Previous studies have shown that soil collected from established Brachiaria plots shows a remarkable decrease in the net nitrification rate during incubation compared with soil cultivated with non-BNI plants [16][17][18][19]. Meanwhile, Subbarao et al. [14] found that both the soil ammonia oxidation rates and cumulative N 2 O emissions were reduced by almost 90% after Brachiaria pasture planting compared with soybean or plant-free plots during a three-year field experiment in Colombia. However, in contrast, Vazquez et al. [20] found no apparent differences in the gross nitrification rates in the soil in which different Brachiaria genotypes with differing BNI capacities were grown. N 2 O is produced by a number of simultaneous N transformation processes [21]. Denitrification produces N 2 O as an intermediate product during the reduction in nitrate (NO 3 − ) to N 2 and is considered a much more potent source of N 2 O than nitrification in grassland soil [22]. However, it remains unclear whether cultivation of exotic Brachiaria in tropical pastures results in a reduction in soil denitrification potential and N 2 O emissions due to the decrease in supply of NO 3 − substrates for denitrifiers. In this study, we therefore established an incubation experiment using a 15 N tracing technique with soil samples collected from an experimental field cultivated with Brachiaria and the native grass Eremochloa ophiuroide, which has no BNI capacity. The objectives were to: (1) determine the effect of Brachiaria on soil N transformation rates in terms of gross nitrification and denitrification rates; and (2) understand how cultivation of Brachiaria affects soil N 2 O emissions. We hypothesized that Brachiaria cultivation would reduce nitrification by releasing biological nitrification inhibitors, thereby reducing the availability of NO 3 − for denitrification and together with nitrification, decreasing soil N 2 O emissions.

Field Experiment and Soil Sampling
The field experiment was established in Danzhou, Hainan Province, China (109 • 29 E, 19 • 30 N), in August 2015. The area has a tropical monsoon climate, with an annual mean air temperature of 23.1 • C and annual precipitation of 1823 mm. The soil was developed from granite and classified as Latosol according to the US soil taxonomy. The field experiment involved eight treatments consisting of two forage grasses and four N application rates, with three replicates each. The two forage species were the introduced exotic grass Brachiaria humidicola CIAT679, which has a high-BNI capacity [14], and the native tropical grass Eremochloa ophiuroide, which has no BNI capacity. The four N application rates were 0, 150, 300 and 450 kg N ha −1 year −1 . Plot size was 4 × 3 m and all plots were arranged according to a randomized block design. N fertilizer urea was surface applied prior to irrigation. In the first growing season, 60% urea was applied in August 2015 as a basal fertilizer, with the remaining 40% applied in April 2016 as a top-dressing. In the second growing season, 40% urea was applied in August 2016 as a basal fertilizer, while 30% was applied in March and 30% in June 2017 as a top-dressing. The grasses were harvested using a lawnmower one day before each top-dressing. Phosphorus and potassium application rates were 120 kg P 2 O 5 ha −1 year −1 (calcium superphosphate) and 120 kg K 2 O ha −1 year −1 (potassium sulfate), respectively, with both applied annually as a basal fertilizer in August.
In March 2017, approximately 2 years after establishment of the field experiment and 6 months after the last fertilization, surface soil (0-20 cm) was collected from 10 different positions in each Brachiaria and Eremochloa plot treated with 150 kg N ha −1 year −1 . The samples were then pooled to form a composite sample for each treatment. After removal of visible roots and litter, the fresh soil was sieved through a 2 mm mesh then divided into two subsamples, one of which was stored at 4 • C for incubation and the other which was air-dried for further analysis. Soil pH was measured in a 1:2.5 soil:water sample ratio using a DMP-2 mV/pH detector (Quark Ltd., Nanjing, China). Soil organic C (SOC) and total N (TN) were determined by wet-digestion with H 2 SO 4 -K 2 Cr 2 O 7 and on a CN analyzer (Vario Max CN, Elementar, Hanau, Germany), respectively, while NH 4 + and NO 3 − were extracted using 2 M potassium chloride (KCl) at a 1:5 soil:solution ratio then analyzed using a continuous-flow autoanalyzer (Skalar, Breda, The Netherlands). Dissolved organic carbon (DOC) was extracted using deionized water at a soil:water ratio of 1:5 with shaking for 0.5 h then analyzed using a TOC analyzer (Vario TOC cube, Elementar, Hanau, Germany). Soil available K + was extracted with ammonium acetate and analyzed using a flame photometer (FP640, INASA, China). Soil properties are presented in Table 1.

15 N Tracing Experiment
The soil incubation experiments consisted of two NH 4 NO 3 treatments with three repetitions each, with labelling of either ammonium ( 15 NH 4 NO 3 , 10.23 atom % excess) or nitrate (NH 4 15 NO 3 , 10.28 atom % excess) with 15 N. Six sets of 250 mL incubation bottles (six bottles per set) were prepared with 30 g fresh soil (on oven-dried basis). After 24 h pre-incubation, 2 mL of 15 NH 4 NO 3 solution or NH 4 15 NO 3 solution was then added at a rate of 50 mg NH 4 + -N kg −1 soil and 50 mg NO 3 − -N kg −1 soil, respectively. The bottles were sealed with cling film punctured with seven pin holes to allow gas exchange then incubated for 16 d at a water holding capacity (WHC) of 60% and a temperature of 25 • C in the dark. Water lost during incubation was compensated for by adding deionized water using a mini pipette to maintain a constant weight. Prior to incubation, a pre-experiment was conducted to confirm the optimal incubation time and gas sampling time interval for identifying the N 2 O flux peaks and meeting the requirement of data-input for the 15 N tracing model.
Gas sampling and destructive soil sampling were carried out 2, 98, 194, 290, and 386 h after NH 4 NO 3 application, respectively. At each sampling point, gas samples were collected using a 50 mL syringe from a specific set of bottles at 0 and 6 h after sealing with an air-tight lid. The samples were then immediately injected into two pre-evacuated gas vials with a butyl-rubber stopper for analysis of N 2 O concentrations and the isotopic composition of 15 N 2 O. In advance of the first gas collection, the bottles were injected with 50 mL of fresh gas to maintain air pressure then after the second collection, the lids were replaced with the punctured cling film. At the same time as gas sampling, NH 4 + and NO 3 − were extracted from another set of soil samples using 100 mL 2 M KCl. After extraction, the soil was rinsed repeatedly with deionized water to remove any residual inorganic N then oven-dried at 50 • C for soil organic N testing. The soil and solution samples were both stored at −20 • C until use. N 2 O concentrations in the sampled gas samples were measured using a gas chromatograph (Agilent 7890, Agilent Technologies, Santa Clara, CA, USA) equipped with a 63 Ni electron capture detector. For isotopic analysis, extracted NH 4 + was separated by distillation with MgO, thereafter NO 3 − was converted to NH 4 + with Devarda's alloy in another distillation [23]. Released ammonia was absorbed in boric acid solution, and NH 4 + concentration was measured using 0.02 M sulfuric acid. After acidification, the solution was dried in an oven at 50 • C and 15 N enrichment of NH 4 + was determined using an isotope ratio mass spectrometry (IRMS 20-22, SerCon, Crewe, UK). While 15 N enrichment of N 2 O and organic N were measured using a MAT 253 mass spectrometer (Thermo Finnigan, Bremen, Germany).

15 N Tracing Model
A full process-based 15 N tracing model ( Figure 1) was used to simultaneously quantify the gross N transformation rates in each soil sample [24]. Average NH 4 + and NO 3 − concentrations and 15 N excess values (average ± standard deviations) from the two 15 N-labeled treatments were included in the model. The model calculated the gross N transformation rates by simultaneously optimizing the kinetic parameters for the various N transformation processes to minimize misfit between the modeled and observed NH 4 + and NO 3 − concentrations and respective 15 N enrichments. A Markov chain Monte Carlo metropolis algorithm (MCMC-MA) was used for parameter optimization, since it is known to be efficient to simultaneously estimate a large number of parameters [25,26]. This algorithm performed a random walk in model parameter space in order to find the global minimum and was shown to be robust against local minima [24]. The optimization procedure produced a probability density function for each model parameter, from which the mean and standard deviation of three parallel sequences were then calculated [25]. To obtain the best parameter set for 15 N tracing analysis that was able to simulate the observed data, various combinations of kinetic settings of individual processes were evaluated ( Table 2 shows the final version of the parameter set). The most appropriate model to describe the measured N dynamics was then selected according to the Akaike information criterion for each model version [25].  [24] and N2O production pathways from specific N pools (b). Norg: soil organic N (including soil labile organic N and recalcitrant organic N), NH4 + : ammonium, NO3 − : nitrate, NH4 + ads: ammonium adsorbed to soil, NO3 − sto: stored nitrate, SOM: soil organic matter. Abbreviations for the transformations are as in Table 2.
The initial pool sizes for soil NH4 + -N and NO3 − -N were estimated by extrapolating the first two sets of data back to the time point zero [27]. Based on the kinetic settings and the final parameters set, average gross transformation rates were then calculated over the whole incubation period and presented in units of μg N g −1 soil d −1 (Table 2).

Calculations
The N2O flux (F, μg N2O-N kg −1 h −1 ) was calculated as follows:  Table 2.  4 Oxidation of NH 4 NO 3 Dissimilatory NO 3 − reduction to NH 4 4 Adsorption of NH 4 4 Release of adsorbed NH 4 Values followed by different lowercase letters within the same row indicate a significant difference between treatments (no overlap of 85% confidence intervals). N lab : soil labile organic N, N rec : soil recalcitrant organic N. Kinetic types: 0 = zero order, 1 = first order.
The initial pool sizes for soil NH 4 + -N and NO 3 − -N were estimated by extrapolating the first two sets of data back to the time point zero [27]. Based on the kinetic settings and the final parameters set, average gross transformation rates were then calculated over the whole incubation period and presented in units of µg N g −1 soil d −1 (Table 2).

Calculations
The N 2 O flux (F, µg N 2 O-N kg −1 h −1 ) was calculated as follows: where ρ is the density of gas under standard conditions (1.25 kg N 2 O-N m −3 ); ∆C is the variation in gas concentrations during the 6 h gas sampling period (ppbv); V is the volume of the flask (m −3 ); T is the incubation temperature; ∆t is the incubation time (h); and W is the dry weight of the soil (kg). Cumulative N 2 O emissions (E, µg N 2 O-N kg −1 ) were calculated as follows: ; i is the ith measurement; and t i+1 -t i represents the time interval between the two adjacent measurements. N 2 O is thought to be derived from three N transformation process: autotrophic nitrification, heterotrophic nitrification, and denitrification. The relative contributions of each process to the N 2 O emissions were therefore calculated as follows [28]: where AN, HN and DN represent autotrophic nitrification, heterotrophic nitrification and denitrification, respectively; a N 2 O , a a , a h and a d represent the 15 where N 2 O T is the total N 2 O production rate during the entire incubation time.

Statistical Analyses
Statistical analysis was not applied to the parameter results since the 15 N tracing model contained plenty of iterations [24]. Accordingly, differences between treatments were considered significant at an alpha level of 0.05 if the 85% confidence intervals did not overlap. Differences in soil properties and N 2 O emissions between treatments were determined using an independent t-test. All statistical analyses were carried out using SPSS Statistics (version 26.0, IBM corp., Armonk, NY, USA) for Windows.   15 N enrichment of the NH4 + pool decreased, while that of the NO3 − pool increased following the addition of 15 NH4 + , suggesting that mineralization of soil organic N and NH4 + oxidation occurred simultaneously (Figure 2c,d). Meanwhile, 15 N enrichment of NO3 − decreased after the application of NH4 15 NO3, suggesting that natural or a low abundance of NO3 − entered this pool. In contrast, 15 N enrichment of NH4 + increased slightly after the application of NH4 15 NO3, suggesting that the direct conversion from 15 NO3 − to 15 NH4 + was negligible.    15 N enrichment of the NH 4 + pool decreased, while that of the NO 3 − pool increased following the addition of 15 NH 4 + , suggesting that mineralization of soil organic N and NH 4 + oxidation occurred simultaneously (Figure 2c,d)

Gross N Transformation Rates
The 15 N tracing model described the measured data in the test soil with a correlation coefficient (R 2 ) of 0.99. The estimated gross rates of the 12 N transformation processes are shown in Table 2. The dynamic rates of labile organic N (labile organic N mineralization into NH 4 + and immobilization of NH 4 + into labile organic N) and adsorbed NO 3 − (adsorption of NO 3 − and release of adsorbed NO 3 − ) were negligible in both test soils. Meanwhile, the gross mineralization rate of recalcitrant organic N in the Brachiaria soil was 2.02 µg N g −1 soil d −1 , which was significantly higher than that in the Eremochloa soil. In contrast, the gross rate of mineral NH 4 + immobilization in the Brachiaria soil decreased to 2.94 µg N g −1 soil d −1 from 3.57 µg N g −1 soil d −1 in the Eremochloa soil (Figure 3), and as a result, Brachiaria planting increased the mineralization-immobilization turnover of NH 4 + (M/INH 4 + ) from 44.6% in the Eremochloa soil to 68.7%. Both the adsorption rate of NH 4 + and the release rate of adsorbed NH 4 + were significantly lower in the Brachiaria compared to Eremochloa soil. Meanwhile, the net exchange (release minus adsorption) of mineral NH 4 + between these two pools was 0.64 µg N g −1 soil d −1 in the Brachiaria soil, almost double that in the Eremochloa soil (0.33 µg N g −1 soil d −1 ), suggesting an increase in NH 4 + supply. Autotrophic nitrification was a dominant NO3 − production process in both sets of soils, while heterotrophic nitrification was negligible. Brachiaria planting reduced the autotrophic nitrification rate to 1.44 μg N g −1 soil d −1 from 1.98 μg N g −1 soil d −1 in the Eremochloa soil. The Brachiaria soil also showed a lower nitrification capacity (ONH4 + /M) than the Eremochloa soil (71.3 vs. 126.0%, respectively). Immobilization of NO3 − overwhelmingly surpassed the rate of dissimilatory nitrate reduction to ammonium

Autotrophic nitrification was a dominant NO 3
− production process in both sets of soils, while heterotrophic nitrification was negligible. Brachiaria planting reduced the autotrophic nitrification rate to 1.44 µg N g −1 soil d −1 from 1.98 µg N g −1 soil d −1 in the Eremochloa soil. The Brachiaria soil also showed a lower nitrification capacity (ONH 4 + /M) than the Eremochloa soil (71.3 vs. 126.0%, respectively). Immobilization of NO 3 − overwhelmingly surpassed the rate of dissimilatory nitrate reduction to ammonium (DNRA) in both sets of soil, representing the primary NO 3 − consumption process under our experimental conditions. The NO 3 − immobilization rate in the Eremochloa soil was 0.88 µg N g −1 soil d −1 , nearly four times greater than that in the Brachiaria soil. Meanwhile, the NO 3 − retention capacity and availability in the Eremochloa soil, expressed as the ratio of NO 3 − consumption to production, was significantly higher than in the Brachiaria soil (44.4 vs. 13.9%, respectively).

N 2 O Production Pathways and Emissions
The N 2 O flux peak occurred on day 4 of the incubation in the Eremochloa soil and on day 12 in the Brachiaria soil (Figure 4), although there was no apparent difference in the N 2 O production rates from heterotrophic nitrification between the Eremochloa and Brachiaria soil ( Table 3) (DNRA) in both sets of soil, representing the primary NO3 − consumption process under our experimental conditions. The NO3 − immobilization rate in the Eremochloa soil was 0.88 μg N g −1 soil d −1 , nearly four times greater than that in the Brachiaria soil. Meanwhile, the NO3 − retention capacity and availability in the Eremochloa soil, expressed as the ratio of NO3 − consumption to production, was significantly higher than in the Brachiaria soil (44.4 vs. 13.9%, respectively).

N2O Production Pathways and Emissions
The N2O flux peak occurred on day 4 of the incubation in the Eremochloa soil and on day 12 in the Brachiaria soil (Figure 4), although there was no apparent difference in the N2O production rates from heterotrophic nitrification between the Eremochloa and Brachiaria soil ( Table 3). The average N2O production rate of autotrophic nitrification was 3.19 μg N2O-N kg −1 d −1 in the Eremochloa soil, while it was significantly lower at 2.78 μg N2O-N kg −1 d −1 in the Brachiaria soil. In contrast, the average N2O production rate during denitrification increased sharply from 0.55 μg N2O-N kg −1 d −1 in the Eremochloa soil to 4.25 μg N2O-N kg −1 d −1 in the Brachiaria soil, representing a 7.7-fold increase.   Cumulative N2O emissions during incubation were significantly higher in the Brachiaria soil (144.31 μg N2O-N kg −1 ) than that estimated as 90.42 μg N2O-N kg −1 in the Eremochloa soil. Meanwhile, denitrification contributed to 47.1% of the emitted N2O, exceeding the contributions of autotrophic and heterotrophic nitrification in the  Values represent means ± standard deviation (n = 3). Different lowercase letters within the same column denote a significant difference between treatments at p < 0.05. f AN , f HN and f DN denote the contribution of autotrophic nitrification, heterotrophic nitrification and denitrification to N 2 O production, respectively.
Cumulative N 2 O emissions during incubation were significantly higher in the Brachiaria soil (144.31 µg N 2 O-N kg −1 ) than that estimated as 90.42 µg N 2 O-N kg −1 in the Eremochloa soil. Meanwhile, denitrification contributed to 47.1% of the emitted N 2 O, exceeding the contributions of autotrophic and heterotrophic nitrification in the Brachiaria soil (Table 3). In contrast, only 9.7% of the produced N 2 O was derived from denitrification in the Eremochloa soil.

Brachiaria Humidicola Cultivation Enhanced the Soil NH 4 + Supply
This study revealed that the gross mineralization rate of soil recalcitrant organic N was significantly enhanced by 28.7% under cultivation of Brachiaria compared with Eremochloa, accelerating the renewal of soil organic N due to its slight increase. This is consistent with the findings of Teutscherová et al. [29,30] who revealed a positive priming effect of high-BNI Brachiaria on native soil organic N decomposition in Colombian pastures compared with a low-BNI genotype. It is suggested that grasses with a dense root system stimulate organic N mineralization by enhancing microbial biomass and activity through the release of large amounts of dead roots and exudates into the soil [31,32]. It is thought that the increase in organic C accelerates the formation of aggregates and reduces the effective diffusion coefficient of oxygen in the soil, in turn inducing a shift in dominant microbes from aerobes (Gram-negative bacteria) to facultative and/or anaerobic microbes (Gram-positive bacteria and fungi) [33][34][35][36]. In general, Gram-positive bacteria and fungi preferentially utilize soil recalcitrant organic matter [37,38]. It is therefore likely that the increase in soil organic C observed here was mainly due to the increase in organic C input from the high biomass of dead roots and exudates under Brachiaria cultivation, resulting in more efficient growth of Gram-positive bacteria and fungi and an increase in the turnover of recalcitrant organic C compared with Eremochloa cultivation.
In contrast, the immobilization rate of NH 4 + in the Brachiaria soil decreased compared with the Eremochloa soil. This differs from the results of Vazquez et al. [20] who showed that gross NH 4 + immobilization was enhanced in the soil cultivated with high-BNI Brachiaria genotypes, while the NO 3 − concentration and N losses remained low. It has been reported that a high C/N ratio in high-BNI plant soil reduces N availability for microbial N immobilization when no N fertilizers are added or when only limited (animal) urine deposition occurs [14,29,39,40]. In contrast, cultivation of high-BNI Brachiaria genotypes in the same field experiment results in a significant reduction in microbial NH 4 + immobilization rates at 7 and 21 days after application of N fertilizers at a rate 50 kg N ha −1 [41]. These results suggest that microbial NH 4 + immobilization is dependent on soil NH 4 + availability. Compared with Eremochloa, Brachiaria more efficiently reduced the gross rate of soil NH 4 + adsorption than the release rate of adsorbed NH 4 + , causing an increase in the net release rate of adsorbed NH 4 + . This may have been due to two possible reasons. Firstly, roots of Brachiaria can distribute within the 20-40 cm soil layer, allowing effective absorption of non-exchangeable K + from deeper soil layers [42], dramatically increasing available K + in the surface soil. The higher availability of K + also outcompetes NH 4 + for soil adsorption sites, thereby reducing soil adsorption of NH 4 + since both have a similar ionic radius and physical properties [43,44]. To date, however, the influence of Brachiaria planting on the soil available K + is less studied. Further study is required to evaluate how Brachiaria cultivation affects the soil available K at the different K application rates. Secondly, the adsorption capacity of NH 4 + in soil is also affected by the content of clay and organic matter [45,46]. Organic matter with plenty of polar atom groups, such as carboxyl and phenolic hydroxyl, contribute to the negative charge and is the main source of variable soil charge. It is therefore likely that NH 4 + as a cation is not as efficiently adsorbed, while NH 4 + from decomposed organic N is released during the mineralization of native recalcitrant organic C under cultivation of Brachiaria.
Overall, cultivation of Brachiaria therefore reduced the rates of NH 4 + immobilization and adsorption and enhanced the rates of organic N mineralization and adsorbed NH 4 + release, in turn increasing soil NH 4 + availability and supply compared with Eremochloa cultivation. These results suggest that in the Brachiaria soil, reduced application rate of K fertilizer may increase the adsorption of NH 4 + from the test soil.

Effect of Brachiaria humidicola Cultivation on Soil N 2 O Emissions
As expected, cultivation of Brachiaria significantly reduced autotrophic nitrification and related N 2 O production, compared with Eremochloa. This is consistent with the findings of Subbarao et al. [14] who reported that Brachiaria soil reduced the ammonia oxidation rate by 90% during a 3-year field experiment compared with soybean and plant-free soil. Subbarao et al. [47] revealed that the release of brachialactone exudate by Brachiaria blocked the activities of both ammonia monooxygenase and hydroxylamino oxidoreductase, thereby reducing ammonia oxidation in pure culture with the ammonia oxidizer Nitrosomonas europaea. In line with this, a reduction in the abundance of ammonia oxidizers was observed in a previous field study of Brachiaria cultivation [18,40]. It was also revealed that BNI compounds were able to persist for a long time and tended to accumulate over time, remaining effective even after Brachiaria pasture was subsequently replaced with maize [48][49][50]. However, Vazquez et al. [20] and Teutscherová et al. [41] found no comparable differences in gross nitrification rates between Brachiaria genotypes with differing BNI capacities in soil with a high organic C content in the tropical savanna in Colombia. Moreover, they attributed the reduced inhibition of potential net nitrification rates to strong microbial immobilization of NH 4 + , and a subsequent reduction in soil NH 4 + availability for ammonia oxidizers.
In the present study, the reduction in N 2 O production in the Brachiaria soil via autotrophic nitrification was at the lower end of the range of 16.8-90.0% reported by previous studies [51]. This suggests that less NH 4 + was converted into N 2 O during autotrophic nitrification in test acidic soil. In general, competition for available NH 4 + exists between autotrophic nitrification and microbial immobilization or adsorption of NH 4 + [25,52]. As discussed above, reduced rates of NH 4 + immobilization and NH 4 + adsorption together with higher mineralization rates of organic N provided more NH 4 + substrates for ammonia oxidizers in the Brachiaria compared to Eremochloa soil. This suggests that the suppression effect of high-BNI Brachiaria on N 2 O emissions may also depend on soil NH 4 + availability, with greater suppression in soil with relatively low available NH 4 + [20,29,53,54]. In contrast to autotrophic nitrification, Brachiaria did not alter the N 2 O production rate via heterotrophic nitrification. In general, heterotrophic nitrification is carried out by a large variety of bacteria and fungi [55], with heterotrophic nitrifiers using both organic and inorganic N as a substrate, and possibly producing more N 2 O than autotrophic nitrifiers [56,57]. In the present study, we supposed that heterotrophic nitrifiers would use organic N as a unique substrate since the model only can select one substrate pool for running. Thus, it is very likely that the gross rate of heterotrophic nitrification in the Brachiaria and Eremochloa soil was underestimated. In general, fungal nitrification in acidic soil is not inhibited by popular synthetic nitrification inhibitors such as acetylene (C 2 H 2 ), dicyandiamide (DCD) and nitrapyrin [58]. Therefore, our results suggest that the 2-year cultivation with Brachiaria had no effect on N 2 O production via heterotrophic nitrification.
As unexpected, cumulative N 2 O emissions were significantly higher in the Brachiaria soil compared to the Eremochloa soil as measured in the field (4.30 and 1.54 kg N 2 O-N ha −1 under Brachiaria and Eremochloa cultivation, respectively), although N 2 O emissions via autotrophic nitrification decreased. This is in contrast with previous results in which Brachiaria establishment was found to reduce soil N 2 O emissions by 20-90% compared with plants without BNI capacity [14,47]. In the present study, the N 2 O production rate in the Brachiaria soil during denitrification increased 6.7-fold, while the contribution ratio of denitrification to emitted N 2 O dramatically increased compared with Eremochloa. These results clearly suggest that the enhanced N 2 O emissions in the Brachiaria soil were primarily due to an increased denitrification potential. There were two suggested possibilities. Firstly, compared with Eremochloa, Brachiaria cultivation sharply reduced the NO 3 − immobilization rate, which outnumbered the rate of DNRA as the primary consumption process of NO 3 − . This suggests that cultivation of Brachiaria increased soil NO 3 − availability by reducing the ratio of total NO 3 − consumption through microbial immobilization of NO 3 − and dissimilatory NO 3 − reduction to NH 4 + (I NO3 + D NO3 ) to total NO 3 − production (O NH4 + O Nrec ), thereby providing more NO 3 − for denitrification. It was previously suggested that higher NO 3 − concentrations in the soil tend to result in a higher N 2 O/N 2 ratio during denitrification, thereby favoring N 2 O emissions [59]. Secondly, in this study, Brachiaria cultivation significantly enhanced soil organic C, notably dissolved organic C, due to the increase in plant biomass and especially biomass of roots and exudates. Increases in soil organic C were also found to promote the formation of anaerobic microsites for denitrification by stimulating aggregation and soil respiration [60,61]. Meanwhile, an increase in organic C was also found to reduce the minimum soil moisture threshold for the occurrence of denitrification [62,63]. For example, Wan et al. [64] found that the addition of starch to sandy loam soil treated with nitrate-based fertilizers stimulated N 2 O production through denitrification. The results of this study therefore suggest that although cultivation of Brachiaria suppressed autotrophic nitrification, it significantly increased the soil denitrification potential and subsequent N 2 O production by increasing soil organic C, notably labile organic C, through an increase in plant biomass, thereby stimulating soil N 2 O emissions.

Conclusions
This study examined the effect of the exotic tropical grass Brachiaria, which has a high-BNI capacity, on soil N transformation processes and N 2 O emissions. Cultivation of Brachiaria significantly decreased the gross rate of autotrophic nitrification and N 2 O production during nitrification. In contrast, Brachiaria also increased the gross mineralization rate of soil recalcitrant organic N and reduced microbial NH 4 + immobilization and NH 4 + adsorption, thereby increasing the NH 4 + supply for nitrification compared with native Eremochloa. Brachiaria planting caused a significant increase in soil N 2 O emissions, primarily due to an increased denitrification potential as a result of reductions in NO 3 − immobilization and an increase in soil labile organic C. Further studies are now required to determine the effects of K fertilizers on the adsorption and availability of NH 4 + . The effect of synthetic nitrification inhibitors together with the biological nitrification inhibitors released from Brachiaria on the mitigation of N 2 O emissions in tropical pastures also requires further clarification.

Conflicts of Interest:
The authors declare no conflict of interest.