Non-native Brachiaria humidicola with biological nitrification inhibition capacity stimulates in situ grassland N2O emissions

Introduction Brachiaria humidicola, a tropical grass, could release root exudates with biological nitrification inhibition (BNI) capacity and reduce soil nitrous oxide (N2O) emissions from grasslands. However, evidence of the reduction effect in situ in tropical grasslands in China is lacking. Methods To evaluate the potential effects of B. humidicola on soil N2O emissions, a 2-year (2015–2017) field experiment was established in a Latosol and included eight treatments, consisting of two pastures, non-native B. humidicola and a native grass, Eremochloa ophiuroide, with four nitrogen (N) application rates. The annual urea application rates were 0, 150, 300, and 450 kg N ha−1. Results The average 2-year E. ophiuroides biomass with and without N fertilization were 9.07–11.45 and 7.34 t ha−1, respectively, and corresponding values for B. humidicola increased to 31.97–39.07 and 29.54 t ha−1, respectively. The N-use efficiencies under E. ophiuroide and B. humidicola cultivation were 9.3–12.0 and 35.5–39.4%, respectively. Annual N2O emissions in the E. ophiuroides and B. humidicola fields were 1.37 and 2.83 kg N2O-N ha−1, respectively, under no N fertilization, and 1.54–3.46 and 4.30–7.19 kg N2O-N ha−1, respectively, under N fertilization. Discussions According to the results, B. humidicola cultivation increased soil N2O emissions, especially under N fertilization. This is because B. humidicola exhibited the more effective stimulation effect on N2O production via denitrification primarily due to increased soil organic carbon and exudates than the inhibition effect on N2O production via autotrophic nitrification. Annual yield-scaled N2O emissions in the B. humidicola treatment were 93.02–183.12 mg N2O-N kg−1 biomass, which were significantly lower than those in the E. ophiuroides treatment. Overall, our results suggest that cultivation of the non-native grass, B. humidicola with BNI capacity, increased soil N2O emissions, while decreasing yield-scaled N2O emissions, when compared with native grass cultivation.


Introduction
Nitrous oxide (N 2 O) is a potent greenhouse gas with a significant 100-year global warming potential that is 265 times higher than that of carbon dioxide on a per-molecule basis (IPCC, 2013). In addition, N 2 O depletes stratospheric ozone, which protects the earth from biologically damaging ultraviolet radiation (Ravishankara et al., 2009). Notably, the concentration of atmospheric N 2 O has increased from 270 ppb during the pre-industrial era to 335.55 ppb in 2022, with an average annual increase rate of 0.90 ppb over the last 2 decades (Lan et al., 2022). Agriculture reportedly emitted approximately 4.1 Tg N 2 O-N year −1 , accounting for approximately 66% of total global anthropogenic N 2 O emissions (UNEP, 2013). Using the dynamic land ecosystem model, Dangal et al. (2019) estimated that the net N 2 O emission from the global grasslands was 2.2 Tg N 2 O-N year −1 , which was responsible for 54% of the total agricultural N 2 O emissions.
To meet the increasing food demands, nitrogen (N) fertilizer and agricultural land are growing substantially (Foley et al., 2011). The global synthetic N fertilizer consumption has increased from 12 to 112 Tg N while that has risen from 0.8 to 24 Tg N in China during the 1961-2020. 1 However, the N-use efficiency (NUE) in China was only 28-35%, which is much lower than the global average (Liu et al., 2013;Han et al., 2015;Zhang et al., 2015).The heavy reliance of N fertilizers in agriculture has contributed to the stimulation of nitrifier activity and the trend toward high-nitrifying soil environments (Poudel et al., 2002;Bellamy et al., 2005).
Nitrification is closely related to N utilization and loss, and has become a key process to improve NUE and reduce N pollution (Subbarao et al., 2006;Beeckman et al., 2018). Nitrification is a microbes-driven process of oxidizing ammonia (NH 3 ) to nitrite and further to nitrate (NO 3 − ) and producing N 2 O as a byproduct (Stein, 2020). The NO 3 − produced during nitrification serves as a substrate and denitrification further reduces NO 3 − to dinitrogen and produces N 2 O as an intermediate product (Coskun et al., 2017a). Nitrification inhibitors can depress the activities of nitrifiers in soil, thereby delaying NH 3 oxidation and reducing N 2 O emissions and NO 3 − production (Rodgers, 1986;Coskun et al., 2017b). To date, a few synthetic nitrification-inhibiting compounds have been efficiently adopted in the field, such as nitrapyrin, dicyandiamide, and 3,4-dimethyl pyrazole phosphate (Weiske et al., 2001;Zerulla et al., 2001;Niu et al., 2018). Meta-analysis revealed that the combination application of nitrification inhibitors and urea reduced NO 3 − leaching by 48% and N 2 O emissions by 44% (Burzaco et al., 2014), and increased crop yields by 7.5% and NUE by 12.9% (Abalos et al., 2014a). However, synthetic nitrification inhibitors have certain limitations such as low cost-effectiveness, application challenges, poor biological stability, and environmental pollution risks (Subbarao et al., 2012;Coskun et al., 2017b;Wang et al., 2021).
Natural compounds with biological nitrification inhibition (BNI) have been found from litters, root exudates, tissue extracts, and rhizosphere of plants such as grasses, trees, and crops (Wang et al., 2021), including methyl 3-(4 hydroxyphenyl) propionate (MHPP), sorgoleone and sakuranetin from sorghum (Subbarao et al., 2013), 1,9-decanediol from rice (Sun et al., 2016) and brachialactone from Brachiaria humidicola grass . Some rootsecreted biological nitrification inhibitors (BNIs) like sorgoleone, sakuranetin, and brachialactone as well as linolenic acid and linoleic acid found in B. humidicola can inhibit both ammonia monooxygenase and hydroxylamine oxidoreductase activities (Coskun et al., 2017b), while 1,9-decanediol and MHPP only inhibits activity of ammonia mono-oxygenase (Zakir et al., 2008;Nardi et al., 2013, 1 https://www.ifastat.org/databases/graph/1 2020; Sun et al., 2016;Lu et al., 2019). Up to date, the functional validation of the BNIs is mainly performed in the pure culture system of a single strain Nitrosomonas europaea, and the effect in the complicated soil system remains to be tested (Subbarao et al., 2015). For example, sakuranetin isolated from sorghum shows a strong inhibitory activity in vitro-cultural bioassay but losses the inhibitory effect in soil-assay (Subbarao et al., 2013). Gopalakrishnan et al. (2009) found that the inhibition effect of BNIs is affected by soil type, and the BNIs derived from B. humidicola in Cambisol can inhibit 90% nitrification with comparable effects to dicyandiamide (50 mg kg −1 soil), but are less effective in Andosol during the 60-day incubation.
Forage grasses with biological nitrification inhibition (BNI) capacity exhibit approximately 2-fold greater productivity than those lacking such capacity in nutrient-limited ecosystems, based on an estimate of a newly developed model (Lata et al., 1999;Boudsocq et al., 2009). The B. humidicola, reportedly exhibits the strongest BNI function among tropical grasses reduces the NH 3 oxidation rate and N 2 O emissions significantly during a 3-year field experiment, when compared with soybean-planted or plant-free soils . During a short-term (29 days) monitoring period in Colombia, cumulative N 2 O emissions from a B. humidicola cv. Tully field was decreased by approximately 60% when compared with that in a Brachiaria hybrid cv. Mulato field under bovine urine amendment (Byrnes et al., 2017). In contrast, no significant effect on N 2 O emissions of the two forage genotypes was observed under cattle dung amendment in the same experimental site (Lombardi et al., 2022).
Latosol is a most widely distributed soil and covers 51.26% of the total area in Hainan Province, China. In the present study, a 2-year field experiment was conducted in a Latosol cultivated with B. humidicola and a native grass species, Eremochloa ophiuroides. We hypothesized that in situ N 2 O emissions from grasslands under cultivation with Brachiaria with higher BNI capacity are lower than in those cultivated with Eremochloa. The objectives of the present study were: (1) to determine whether the N 2 O emissions from a B. humidicola field are lower than those from an E. ophiuroides field in tropical Hainan Province, China; and (2) to evaluate the mitigation effects of B. humidicola on yield-scaled N 2 O emissions under the different N application rates. We also established an incubation with soils from the field experiment using a 15 N tracing technique to evaluate the influence of B. humidicola on the N transformation process rates and N 2 O production rates via nitrification and denitrification (Xie et al., 2022).

Study site
The field site was located in Danzhou, Hainan Province, China (109°29′ E, 19°30′ N). The region is characterized by a tropical monsoon climate, with a rainy season from May to October, and a dry season from November to April. The mean annual temperature and precipitation are 23.1°C and 1,823 mm, respectively. The soil is derived from granite and classified as a Latosol, with a sandy loam texture. Latosol is a most widely distributed soil in Hainan Province. The properties of surface soil (0-20 cm) prior to the field experiment are shown in Table 1

Experimental design
A field experiment was established in August 2015 and included eight treatments, consisting of two pastures, Brachiaria humidicola CIAT679 and Eremochloa ophiuroides, with four N application rates. The annual urea application rates were 0, 150, 300, and 450 kg N ha −1 , which were designated as BCK, BN1, BN2, and BN3, respectively, for B. humidicola, and ECK, EN1, EN2, and EN3, respectively, for E. ophiuroides. The plots measured 3 m × 4 m. The treatments, which had three replicates, were set up based on a randomized complete block design. During the first season from August 2015 to August 2016, 60 and 40% of urea was applied as basal fertilizer and top-dressing fertilizer, respectively, in the fertilized treatments. In the second season from August 2016 to August 2017, urea was added with three splits: 40% as basal fertilizer, and 30% as top-dressing fertilizer on 13 March and 9 June 2017, respectively. Calcium superphosphate (150 kg P 2 O 5 ha −1 ) and potassium chloride (105 kg K 2 O ha −1 ) were applied as basal fertilizer. All the fertilizers were dissolved in water and uniformly spread into the soil. Harvested plant samples were ovendried at 60°C to a constant weight, and then ground to less than 0.2 mm for analysis. Field management practices were similar to local practices and standardized at all plots. Specific dates are listed in Table 2.

Nitrous oxide flux measurement
Nitrous oxide fluxes were measured using the static chamber method. Before grass planting, a stainless steel chamber with a rectangular base (50 cm × 50 cm × 15 cm) and a 5-cm groove around the upper edge was permanently fit 10 cm into the soil. During gas sampling, a stainless chamber (50 cm × 50 cm × 50 cm) was inserted into the groove, which was filled with water to ensure airtightness. The chamber was covered with reflective film and foam to minimize air temperature change inside the chamber. A rubber plug with a mercury thermometer was fit tightly into the hole on the top of the chamber for use in measuring the chamber temperature while gas sampling. Two vents welded with stainless tubes were punched on top of the chamber, one connected to a rubber tube with a three-way stopcock for gas collection, and another one for ensuring air pressure equilibrium inside and outside the chamber. Gas samples were obtained once every other day during 1 week after each fertilization and once a week during the other period. Sampling was conducted between 7:00 am and 12:00 pm to minimize diurnal variation. Four gas samples were extracted from the chamber at 0, 10, 20, and 30 min after chamber closure using airtight plastic syringes and instantly injected into 20-ml pre-evacuated vials fitted with butyl rubber stoppers. The N 2 O concentrations were analyzed using a gas chromatograph (GC; Agilent 7890, Agilent Technologies, Santa Clara, CA, United States) equipped with a 63 Ni electron capture detector and a thermal conductivity detector. The N 2 O fluxes were calculated using the following equation (Niu et al., 2018): ; ρ is the density of N 2 O at 0°C and 760 mm Hg (kg m −3 ); V is the effective volume of the chamber (m 3 ); S is the soil area covered by the chamber (m 2 ); ΔC/Δt is the rate of N 2 O concentration increase in the chamber (ppbv N 2 O-N h −1 ); and, T is mean air temperature inside the chamber during sampling (°C).

Auxiliary variables measurement
Soil temperature was measured at 5 cm depth using a geothermometer. Soil water content was measured at three different positions in each plot with time domain reflectometry (TDR) probes and expressed as water-filled pore space (WFPS, %) as follows (Niu et al., 2018): WFPS volumetric water content total soil porosity = / where total soil porosity = 1 -(soil bulk density/2.65), 2.65 being the soil particle density (g cm −3 ).
Surface (0-20 cm) soil samples were collected at five different positions in each plot fortnightly using a stainless steel soil sampler and thoroughly mixed to form a composite sample. The samples were taken to the laboratory immediately and stored at −20°C before analysis. Soil exchangeable ammonium-N (NH 4 + -N) and nitrate-N (NO 3 − -N) were extracted with 2 M KCl (soil/KCl solution ratio of 1:5) by agitating for 1 h on a rotary shaker, and the concentrations were measured using a colorimetric method on a segmented flow analyzer (Skalar, The Netherlands; Chen et al., 2014). Dissolved organic C (DOC) was extracted with deionized water at a soil water ratio of 1:5, which was shaken for 0.5 h, followed by centrifugation for 15 min at   2,300 × g and filtration (<0.45 μm). Subsequently, the DOC was analyzed using the combustion oxidation nondispersive infrared absorption method on a TOC analyzer (vario TOC Cube, Elementar, Hanau, Germany). Soil samples were collected after the end of the field experiments. Soil pH was determined from soil-water suspensions (1:2.5 v/v) using a pH meter (SevenCompact, Mettler Toledo, Swiss). Soil organic C (SOC) was measured using the wet oxidation-redox titration method (Walkley and Black, 1934). Total N content in soil and plant was determined using an elemental analyzer (Vario MAX, Elementar, Germany). Soil available P was extracted with 0.05 M HCl and 0.025 M H 2 SO 4 , and determined using the molybdenum blue colorimetric method (Ye et al., 2019). Available K was extracted with ammonium acetate and analyzed using a flame photometer (Lu, 2000).

Data calculation and statistical analysis
Annual cumulative N 2 O emissions (E N2O , kg N 2 O-N ha −1 ) were calculated using the following equation (Chen et al., 2014): is the interval between the i th and the (i + 1) th measurement time (d); n is the total number of measurements; and 24 × 10 −5 was used for unit conversion.
The N 2 O emission factor of applied fertilizer N (EF, %) was calculated using the following equation: where E fertilizer and E control are the cumulative N 2 O emissions from the fertilized and control treatments, respectively; and N applied is the amount of fertilizer N applied to the corresponding treatment.
The yield-scaled N 2 O emission (mg N 2 O-N kg −1 biomass) was computed using the following equation (Venterea et al., 2011): where E N2O is the annual cumulative N 2 O emissions (kg N 2 O-N ha −1 ); and yield is the amount of grass biomass harvested annually (kg ha −1 ).
Fertilizer N-use efficiency (NUE, %) was calculated as follows: where N fertilizer and N control are the amount of N uptake in aboveground biomass (kg N ha −1 ) in the fertilized and control treatments, respectively; and N applied is the amount of the N applied to the corresponding treatment (kg N ha −1 ).
Significant differences among treatments were evaluated using one-way ANOVA followed by the Duncan test at p < 0.05. Spearman's correlation analysis was used to determine the relationships between N 2 O flux and soil WFPS, soil inorganic N, soil dissolved organic C, and air temperature. All statistical analyses were performed using IBM SPSS Statistics 26 for Windows (IBM corp., Armonk, NY, United States).

Soil characteristics
After 2 years of grass cultivation, soil pH in all the treatments increased when compared with that in the pre-treatment soil ( Table 3). The maximum soil pH was observed in the ECK treatment without N fertilization, and soil pH decreased with an increase in N application rate in both grasslands. SOC increased by 17.5-22.8% under B. humidicola cultivation and only by 5.8-15.1% under E. ophiuroides, when compared with the pre-treatment soil. Cultivation of both pastures promoted soil N accumulation significantly (p < 0.05); however, there were no significant differences in soil N accumulation among treatments under different N application rates (Table 3).

Grass yield
The biomass of B. humidicola ranged from 29.54 to 31.37 t ha −1 in the BCK treatment, which was 3.1-6.0-fold that of E. ophiuroide during the two seasons (Table 4). The N application increased biomass yield of B. humidicola by 11.3-25.8% (p < 0.05) but did not increase the biomass yield of E. ophiuroides, during the first season. During the 2016-2017 season, however, the biomass of both grasses was enhanced with N fertilization, and increased with an increase in the N application rate (p < 0.05).
The amounts of N uptake by B. humidicola under no N fertilization were 220.08 and 188.74 kg N ha −1 during the 2015-2016 and 2016-2017 seasons, respectively, and decreased to 44.22 and 69.82 kg N ha −1 for E. ophiuroide, respectively (Table 4). The mean NUE of the N applied under B. humidicola was 19.5-29.5% during the 2015-2016 season and increased to 47.9-59.2% during the 2016-2017 season, which was significantly higher than that under E. ophiuroides during both seasons.

Soil and environmental variables
Air temperature (AT) ranged from 5.4 to 32.6°C, with an average of 24.8°C over the 2-year measurement period, and there was no apparent difference between two growth seasons (Figure 1). Soil temperature (ST) at 5 cm depth ranged from 13.7 to 34.8°C, a trend similar to that of AT (ST = 0.758AT + 7.062, R 2 = 0.42, p < 0.01). Mean rainfall was 2,341 and 2,373 mm during the 2015-2016 and 2016-2017 seasons, respectively. Precipitation mainly occurred in the rainy season, from May to October, accounting for 87% of the total annual precipitation. Soil moisture fluctuated from 5.1 to 58.3% WFPS, and the mean WFPS in all the treatments was 33.5-38.3%, with no significant differences among treatments.
Soil NH 4 + -N concentration peaks occurred approximately 1 week after each fertilization, and decreased to a constant level 40 days later Frontiers in Microbiology 05 frontiersin.org  Yield, grass aboveground biomass; N uptake, the amount of N uptake in aboveground biomass; NUE, nitrogen use efficiency. Different letters in the same column indicate significant differences between treatments (p < 0.05).

FIGURE 1
Temporal variation in precipitation, air temperature, and soil temperature at 5 cm depth, and water-filled pore space (WFPS). Vertical bars denote the standard errors of the mean (n = 3). ( Figure 2A). The mean soil NH 4 + -N concentrations under the BCK and ECK treatments were 4.60 and 4.05 mg N kg −1 , respectively and increased to 10.46-14.93 and 10.09-15.91 mg N kg −1 in the BN and EN treatments, respectively, increasing with increases in the N application rate. Mean soil NH 4 + -N concentrations were not significantly different between the B. humidicola and E. ophiuroides fields under similar N application rates. Soil NO 3 − -N concentrations in the BCK and ECK treatments were on average 3.21 and 2.59 mg N kg −1 , respectively ( Figure 2B). Under N fertilization, mean soil NO 3 − -N concentrations increased to 5.61, 6.02, and 8.42 mg N kg −1 in the BN1, BN2, and BN3 treatments, respectively, which were higher than the corresponding values under E. ophiuroides cultivation, excluding BN2 (p < 0.05).

Nitrous oxide emissions
Nitrous oxide flux peaks emerged after each fertilization, and increased with an increase in the N application rates. The highest flux was 544.60 μg N 2 O-N m −2 h −1 in the BN3 treatment on 31 August 2016, which was 3-fold that in the EN3 treatment (Figure 3). During  (Table 5). Annual N 2 O emissions in the B. humidicola fields were higher than those in the E. ophiuroides fields, regardless of N fertilization rate (p < 0.05; Figure 4A). They were also greater during the first respectively, and increased to 1.43-3.28 and 1.65-3.64 kg N 2 O-N ha −1 under N fertilization, respectively. The annual N 2 O emissions increased linearly with an increase in the N application rate in the B. humidicola fields (E N2O = 0.0092 N applied + 2.76, R 2 = 0.97); conversely, they exhibited exponential correlations with the N application rate in the E. ophiuroides fields (E N2O = 1.24e 0.021Napplied , R 2 = 0.95).
The N 2 O emission factor (EF) of the N applied ranged from 0.74 to 0.98% for B. humidicola, and decreased to 0.11-0.47% for E. ophiuroides over the 2 years ( Figure 4B). The EF increased with an increase in the N application rate only under E. ophiuroides cultivation.

Yield-scaled nitrous oxide emissions
The mean yield-scaled N 2 O emissions in the BCK and ECK treatments were 95 and 206 mg N 2 O-N kg −1 biomass, respectively, over the 2 years ( Figure 5). Under N fertilization, they increased to 128, 132, and 183 mg N 2 O-N kg −1 biomass in the BN1, BN2, and BN3 treatments, respectively, which were significantly lower than the corresponding values in the EN treatments by 26.76-46.04%. The reduction increased with an increase in the N application rate.

Discussion
Annual N 2 O emissions from this tropical grassland varied from 1.35 to 9.54 kg N 2 O-N ha −1 , which was within the 0-29.1 kg N 2 O-N ha −1 range in grasslands as reported previously (Mosier et al., 1996;Wolf et al., 2010;Merbold et al., 2014;Luo et al., 2017). Out of expectation, N 2 O emissions from the B. humidicola field were Annual soil nitrous oxide (N 2 O) emission (A) and emission factor of the fertilizer nitrogen applied (B) in the Brachiaria humidicola and Eremochloa ophiuroides fields. Vertical bars denote the standard errors of the mean (n = 3). Different letters indicate significant differences between treatments for the same measurement year and mean at p < 0.05. *indicates the significant difference between 2 years for the same treatment at p < 0.05.  Lata et al. (2022). Previous study suggested that Brachiaria genotype with high BNI capacity reduced almost 50% of N 2 O emission when compared with soybean or plant-free soils . Byrnes et al. (2017) reported that B. humidicola cv. Tully with high BNI capacity reduced approximately 60% of N 2 O emissions in the field when compared with the Brachiaria hybrid cv. Mulato having low BNI capacity during the 29-day monitoring period. Planting B. humidicola with high BNI capacity reduced soil N 2 O emissions by 18.3% when compared with B. humidicola with low BNI capacity in a 21-day pot experiment (Teutscherová et al., 2022). The active substances with BNI capacity, such as methyl-p-coumarate, methyl ferulate, and brachialactone, have been identified from exudates of B. humidicola (Gopalakrishnan et al., 2009;Subbarao et al., 2009). Brachialactone can simultaneously block the enzymatic pathways of ammonia monooxygenase and hydroxylamino oxidoreductase . The inhibitory potential reportedly increases with an increase in grass root density (Subbarao et al., 2007;Boudsocq et al., 2009). Subbarao et al. (2009) estimated that B. humidicola roots can potentially release 2.6 × 10 6 -7.5 × 10 6 ATU (allylthiourea units) ha −1 day −1 BNI activity in the South American savannas, which is equivalent to the application of 6.2-18 kg ha −1 nitrapyrin based on 1 ATU being equal to 0.6 μg of nitrapyrin. Karwat et al. (2017) demonstrated that B. humidicola, like dicyandiamide, significantly suppresses soil nitrification potential. In a previous study, using the 15 N tracing incubation with soils collected from the B. humidicola and E. ophiuroides plots at the field experiment end, we found that B. humidicola decreased the autotrophic nitrification rate and N 2 O production rate via nitrification by 27.3 and 14.7%, respectively when compared with E. ophiuroides (Xie et al., 2022). This indicated that in the test soil, B. humidicola efficiently inhibited nitrification and N 2 O production via nitrification. Subbarao et al. (2009) observed that cultivation of B. humidicola reduced the abundance of both ammonia-oxidizing archaea (AOA) and bacteria (AOB) in a Vertisol with pH 7.40 when compared with soil cultivated with soybean. Hink et al. (2018) further reported that although both AOA and AOB were capable of N 2 O production under high NH 4 + -N concentrations, the contribution of AOB was greater in a soil with pH 6.50. In the test acid soil with pH 5.42, it is likely that both AOA and AOB participated in NH 3 oxidation and N 2 O production. Further investigations are required to determine the relative importance of AOA and AOB in N 2 O production, and the suppression effects of B. humidicola on AOA and AOB activity (Nuñez et al., 2018). Byrnes et al. (2017) suggested that by increasing N uptake, B. humidicola with high BNI capacity more efficiently decreased soil NO 3 − -N availability and potential denitrification than B. humidicola with low BNI capacity, thereby reducing N 2 O emissions. Abalos et al. (2014b) reported that mixed cultivation of Folium perennial L. and Poi trivialize L. decreased soil NO 3 − -N concentrations and consequent N 2 O emissions when compared with either monoculture at an N application rate of 60 kg N ha −1 . They suggested that the trends were attributed to L. perennial taking up N using the "scale strategy" by increasing root biomass, and P. trivialize absorbing N via the "precision strategy" by providing access to N hotspots that were not emptied by L. perennial. In the present study, although B. humid cola cultivation increased N uptake, N 2 O emissions were positively correlated with pasture yield and N uptake, indicating that increased N uptake by B. humid cola did not reduce N 2 O emissions. In the present study, the mean soil NO 3 − -N concentration under N fertilization ranged from 5.60 mg N kg −1 in the BN1 treatment to 8.45 mg N kg −1 in the BN3 treatment, which was higher than the 5 mg N kg −1 threshold for occurrence of denitrification (Dobbie and Smith, 2003), and indicated that although B. humid cola efficiently increased N uptake and partially inhibited nitrification, soil NO 3 − -N under N fertilization was still higher than the threshold value for denitrification in the test field.

Frontiers in
Using 15 N paired incubation ( 15 NH 4 NO 3 and NH 4 15 NO 3 ), we found that the N 2 O production rate during denitrification in the B. humid cola soil increased by 7.7-fold when compared with the E. ophiuroides soil and the contribution of denitrification to N 2 O emissions sharply enhanced from 9.7% in the E. ophiuroides soil to 47.1% (Xie et al., 2022). In the present study, B. humidicola biomass was 3-6-fold greater than that of E. ophiuroides and SOC was more efficiently increased under B. humidicola. Horrocks et al. (2019) also observed that 1-year cultivation of B. humidicola increases SOC content and improves aggregate stability in Colombia. Fisher et al. (1994) and Amézquita et al. (2004) attributed the SOC enhancement to rapid accumulation of B. humidicola roots and exudates. Plant reportedly release as much as 40% of net photosynthetic C into the rhizosphere (Marschner, 2011), which in turn provides more labile C substrates for denitrifiers (Wu et al., 2017). Enhanced SOC at least exhibits two stimulation effects on denitrification. Firstly, enhanced SOC promotes the formation of anaerobic microsites for denitrification by stimulating Yield-scaled N 2 O emissions from the Brachiaria humidicola and Eremochloa ophiuroides field. Vertical bars denote the standard errors of the mean (n = 3). Different letters indicate the significant differences between treatments for the same measurement year and for mean at p < 0.05. *indicates the significant difference between two measurement years for the same treatment at p < 0.05.
Frontiers in Microbiology 09 frontiersin.org aggregation (Bollmann and Conrad, 2004). Secondly, increased organic C availability reduces the soil moisture threshold for the occurrence of denitrification (Rochette et al., 2000;Van Groenigen et al., 2004;Chantigny et al., 2013) resulting in increased denitrification potentials. Our results indicate that cultivation of exotic B. humidicola with a much higher biomass compared with E. ophiuroides stimulated N 2 O production during denitrification by providing more organic C, which in turn masked N 2 O reduction by inhibiting nitrification, thereby enhancing N 2 O emissions. Comparing yield-scaled N 2 O emissions has been suggested to be an effective way of evaluating the tradeoff between production and environmental impacts and determining the economic feasibility of N 2 O emission mitigation methods (van Groenigen et al., 2010;Grassini and Cassman, 2012). In the present study, yield-scaled N 2 O emissions from B. humidicola field with and without N fertilization during two seasons were 128.80-183.02 and 93.02 g N kg −1 biomass, respectively, which were significantly lower than the corresponding values under E. ophiuroides cultivation (171.07-221.62 and 186.93 g N kg −1 biomass, respectively). In addition, we observed interannual shifts in yield-scaled N 2 O emissions in both grasslands, which was primarily driven by changes in annual N 2 O emission in B. humidicola fields, whereas they were driven by changes in biomass yield in E. ophiuroides fields. The lower yield-scaled N 2 O emissions under B. humidicola cultivation compared with under E. ophiuroides indicated that although B. humidicola increased annual N 2 O emissions, it was more environmentally friendly based on its higher forage productivity and NUE.

Conclusion
In the present study, B. humidicola exhibited higher yields and NUE, and in contrast and unexpectedly, induced higher soil N 2 O emissions when compared with E. ophiuroides. Although cultivation of B. humidicola with high BNI capacity reduced N 2 O production rate via nitrification, however, it more efficiently enhanced N 2 O production rate than E. ophiuroides via denitrification due to increased SOC and exudate concentrations, thereby increasing N 2 O emissions ( Figure 6). When compared with under E. ophiuroides, however, the lower yieldscaled N 2 O emissions under B. humidicola cultivation indicated that although B. humidicola increased annual N 2 O emissions, it was more environmentally friendly based on its higher forage productivity and NUE.

Data availability statement
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

Author contributions
WD and DL: conceptualization. LX, YN, and DL: field experiment. LX and ZC: data analysis. LX, WD, and LM: writing. WD: funding acquisition. All authors contributed to the article and approved the submitted version.

Funding
This work was financially supported from the National Natural Science Foundation of China (41730753, 41977049 and 42077029), the International Partnership Program of Chinese Academy of Sciences (151432KYSB20200001), and International Atomic Energy Agency coordinated research project (D15020).
Frontiers in Microbiology 10 frontiersin.org Publisher's note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.