Effects of Split Application of Urea on Greenhouse Gas and Ammonia Emissions From a Rainfed Maize Field in Northeast China

Split application of nitrogen (N) fertilizers during different crop growth stages to fulfill the crop N requirements reduces soil mineral N concentrations and improves the efficiency of crop N fertilizer use, and can decrease nitrous oxide (N2O) emission from the soil. However, inconsistent results regarding N2O emissions have been reported in rainfed areas. Furthermore, few long-term studies have explained the effects of split N application on soil methane (CH4) flux, thus limiting complete assessment of the effects of split N application on total greenhouse gas (GHG) emissions. Therefore, long-term monitoring is urgently required to understand the impacts of split N application on GHG emissions in rainfed areas. In this study, a 6-year field experiment was conducted in a rainfed maize (Zea mays L.) field in Northeast China. The experiment included three treatments: no N application representing control (CK), single application at the sowing stage of maize (SU), and split N at the sowing and jointing stages at a ratio of 1: 2 (SF). Between the sowing and jointing stages, N2O emissions were significantly higher in SU than in SF. However, high N2O emissions were observed in SF for 1 month after N application at the jointing stage possibly because the time of N application coincided with optimum precipitation and soil temperature conditions, which stimulated N2O emissions. Overall, the total N2O emissions showed no significant difference between SU and SF. During the study period, split application of N fertilizer did not significantly affect the cumulative CH4 flux. Compared to CK, the yield-scaled GWP in SF treatment increased by 18.7% (p < 0.05). Ammonia (NH3) volatilization in SF was 272% higher than that in SU. The findings indicated that split N application exhibited an environmental risk by increasing the yield-scaled GWP and NH3 emissions in the field. Thus, this study suggested that single N application applied in the sowing stage should be employed in rainfed fields to mitigate the yield-scaled GWP and NH3 emissions, and maintain efficient maize yields.


INTRODUCTION
Nitrous oxide (N 2 O) is a potent greenhouse gas (GHG) with a global warming potential (GWP) that is 265 times more than that of carbon dioxide (CO 2 ) in a 100-year timescale; moreover, it significantly destroys the stratospheric ozone layer (IPCC, 2013). Therefore, reducing N 2 O emissions is urgently required. Fertilized soil is a primary source of N 2 O (Smith et al., 1998;Thompson et al., 2019); however, using suitable agricultural management measures can effectively reduce N 2 O emissions (Shi et al., 2013).
In Northeast China, maize fields comprise 42.2% of the total cropped area (Yearbook, 2019). However, maize requires high inputs of N fertilizers and, thus, is an important source of atmospheric N 2 O (Liu et al., 2011). A meta-analysis indicated that maize exhibited high emission factors (EF) compared with other crop types (Cayuela et al., 2017). Maize requires low N amounts during its early growth stages but high N amounts for several weeks during the growing season (Abendroth et al., 2011). During the period of first few months from N application to rapid N uptake, N 2 O is majorly lost. Delaying the application of N fertilizers to maize during the growing season can effectively reduce N 2 O emissions in rainfed systems (Dell et al., 2014). Using the DNDC (DeNitrification-DeComposition) model, Li et al. (2012) suggested that N 2 O emissions can be mitigated with the increase in the number of N fertilizer applications from spring maize fields in Northeast China. Split N applications during the growing season could improve the balance between soil N availability and crop N demand, and, subsequently, reduce the soil N amount available for conversion to N 2 O (Venterea and Coulter, 2015;Schwenke and Haigh, 2019). However, previous studies on the effects of split application of N fertilizer on N 2 O emissions are still inconclusive, with reports of increase (Venterea and Coulter, 2015), decrease (Aita et al., 2015;Schwenke et al., 2016), and no change in the N 2 O emissions (Zebarth et al., 2012;Venterea et al., 2016). Differences in the amount and time of precipitation between the single and split N fertilizer applications under rainfed agricultural conditions could contribute to these disparate results (Drury et al., 2012;Yu et al., 2016). Furthermore, difficulty in predicting the precipitation amount and distribution in rainfed areas and its significant impacts on N 2 O emissions served as a challenge to draw consistent conclusions during shortterm monitoring of N 2 O emissions in rainfed areas (Su et al., 2021). In addition, increased N application rate (100 kg N ha −1 ) gradually increased methane (CH 4 ) absorption in upland soils (Aronson and Helliker, 2010). Contrary to previous studies, recent studies indicated that the upland soil should be a source of CH 4 emission (Jiang C. M. et al., 2017;Lan et al., 2020). Therefore, the authors of this study could not draw a conclusion on the response of CH 4 flux to conventional N application (180 kg N ha −1 ) in upland cropping systems in the study area. Thus, increased upland field studies were required to reduce the uncertainty of the effects of N application on soil CH 4 fluxes.
Fertilization of agricultural soil is an important anthropogenic source of ammonia (NH 3 ), which is a secondary source of N 2 O (Gao et al., 2013). Therefore, simultaneous measurements of N 2 O and NH 3 can provide valuable insights on the effect of split N application on N losses. However, the effects of split application of urea on both N 2 O and NH 3 emission have not been studied extensively in rainfed areas.
In this study, we aimed at examining the effects of split application of N fertilizer on N 2 O and CH 4 emissions, maize yield, and yield-scaled GWP in a continuously rainfed maize field in Shenyang City, Northeast China, through long-term field experiments. Additionally, in the sixth year of the study, NH 3 emissions were determined to evaluate the effects of split application of N fertilizer on NH 3 loss. We hypothesize that split application of N fertilizer could mitigate N 2 O emissions, but might increase NH 3 losses from this rainfed maize field.

Site Description and Experimental Design
The experiments were performed from May 2010 to April 2016 in a rainfed maize field at the Shenyang Agro-ecological Experimental Station (41°31′N, 123°22′E), Chinese Academy of Sciences, Liaoning Province, China. The mean annual precipitation of the study site is approximately 680 mm, with more than 80% of the annual precipitation concentrated from May to September. The mean annual temperature is 7.5°C. The soil is classified as Luvisol (World Reference Base) or Alfisol (Soil Taxonomy). The soil texture is silt loam (clay, 20.4%; silt, 50.1%; sand, 28.9%). The soil physical and chemical properties in the top 20 cm were as follows: soil bulk density, 1.25 g cm −3 ; soil pH, 5.8; total N content, 0.95 g kg −1 ; organic carbon content, 8.52 g kg −1 ; Olsen-P, 21.9 mg kg −1 ; and NH 4 OAc-K, 111.0 mg kg −1 .
The experimental design consisted of three replicates of a randomized complete block, with an area of 20 m 2 (4 m × 5 m) for each plot. The experimentation included three treatments: no N fertilizer application (CK), single application of 180 kg N ha −1 of urea at the sowing stage (SU), and N fertilizer application with the same rate and type as SU in two splits (SF). The SF treatment included the application of 60 kg N ha −1 and 120 kg N ha −1 at the sowing and jointing stages of maize, respectively. Calcium superphosphate (90 kg P 2 O 5 ha −1 ) and potassium chloride (90 kg K 2 O ha −1 ) were applied during sowing. Furthermore, maize (Zea mays L., variety of fuyou #9) were planted in early May with a density of 45,000 plants ha −1 , were seeded at 37 cm intervals, were not irrigated, and were harvested by the end of September. At harvest, maize yield and aboveground biomass yield were measured by harvesting all plants (20 m 2 ) in each plot. After harvest of aboveground biomass, they were manually removed.

N 2 O and CH 4 Emissions
Soil-to-atmosphere N 2 O/CH 4 fluxes were measured using static chambers (Dong et al., 2021), which consisted of a base chamber (56 cm × 28 cm × 10 cm) and a removable top chamber (56 cm × 28 cm × 20 cm), both made of stainless steel. Prior to sowing, the base chambers were inserted 10 cm into the soil in the center of a Frontiers in Environmental Science | www.frontiersin.org January 2022 | Volume 9 | Article 798383 row, and were maintained in the same place throughout the study period except during tillage, which was performed once a year in April. Gas samples were collected with a 50-ml syringe, equipped with a three-way stopcock, at 20 min intervals (t 0 , t 20, and t 40 ) between 9:00 and 11:00 a.m. after the chambers were closed. Subsequently, the collected samples were injected into 12 ml vacuum vials fitted with butyl rubber stoppers, and were analyzed using a gas chromatograph (Agilent 7890A, Shanghai, China) equipped with an electron capture detector and a hydrogen flame ionization detector. Gas samples were collected every 2-6 days during the growing seasons and every 7-15 days during the nongrowing seasons. The growing and nongrowing seasons were from May to September and from October to April of the following year, respectively.

NH 3 Emissions
The NH 3 volatilization was measured using a modified vented chamber method as described by Li et al. (2019). The chamber is made up of PVC pipes with a height of 15 cm and a diameter of 15 cm. Two round sponges with 16 cm in diameter and 2 cm in thickness were placed in each chamber. Before use, the sponges were moistened with 15 ml of phosphate/glycerol solution (5% phosphate and 4% glycerol). One sponge was placed 5 cm away from the soil surface to absorb NH 3 volatilized from the soil. Another sponge was placed on the top of the chamber to absorb the NH 3 volatilized from the air entering the chamber through the vent. NH 3 in the phosphate solution of each sponge in the chamber was determined by shaking 300 ml of 1 M KCl for 60 min. Ammonium ions were qualitified with a continuous flow analyzer (Alliance, Futura, France). The sponge in the chamber at the sowing stage of maize was replaced and sampled every day during 4-7 days after sowing, at 2-4 days intervals during the second and third week, and thereafter every 7 days until NH 3 became undetectable. During the topdressing period, NH 3 was observed continuously for 9 days. Next, NH 3 was monitored once a week.

Soil Measurements
During measurements of gas fluxes, the soil temperature at 5 cm of soil depth was simultaneously recorded using a bent stem thermometer. Soil moisture was monitored at a depth of 0-5 cm using time domain reflectometry and was expressed as waterfilled pore space (WFPS). Daily precipitation and air temperature data were acquired from the weather station of the Shenyang experimental station of ecology, Shenyang City.

Statistical Analyses
The N 2 O/CH 4 fluxes were calculated using a linear regression of concentration versus time during the chamber closure period. Cumulative N 2 O/CH 4 emissions were calculated from May to April of the following year by adding the average values of the two adjacent fluxes, and multiplying the sum by the interval time (Ding et al., 2011). The GWP was measured with 265 kg of CO 2 and 28 kg of CO 2 for 1 kg of both N 2 O and CH 4 , respectively, while the yield-scaled GWP (kg CO 2 t −1 yield) was calculated by dividing the GWP by the maize yield (Mosier et al., 2006;Van Groenigen et al., 2010). NH 3 loss from the soil was calculated by the following formula: Here, G represents NH 3 captured by the sponge during each sampling (mg N), a is the cross-sectional area of the chamber (m 2 ), d is the sampling duration (d), and 10 -2 is the conversion coefficient. The loss rate (%) of NH 3 was determined by dividing the cumulative NH 3 emissions (kg N ha −1 ) with the amount of N applied (kg N ha −1 ).
Differences in N 2 O, CH 4 , and NH 3 emissions were tested using one-way analysis of variance (ANOVA) followed by Tukey's test with p 0.05 as the significance level. Furthermore, two-way ANOVA was used to examine whether there were significant differences of the N 2 O flux, CH 4 flux, GWP, maize yields, and yield-scaled GWP among the three treatments in the experimental years. Analyses were performed using the Statistical Package for the Social Sciences (SPSS) 18.0 for windows (SPSS, Chicago, IL, United States).

Environment and Soil Conditions
The air temperature ranged from −22.8°C to 28.3°C (Figure 1), and the annual mean air temperatures were 7.9, 8.3, 7.7, 8.1, 9.5, and 9.3°C during 2010-2015, respectively. The soil temperature trends were consistent with the air temperature changes. The soil temperature at 5 cm depth ranged between −14.0°C and 35.0°C ( Figure 1). In May 2012, a dry period persisted from May 3 to 29, during which 7.6 mm of precipitation was recorded, and the soil WFPS at 0-5 cm depth decreased to 24% on May 28. In May 2015, a heavy precipitation event (61.5 mm) occurred on May 12, and the total precipitation was 71.9 mm. During July 5-24 2015, precipitation was not recorded, and the soil WFPS gradually declined to 10% on July 24. Furthermore, heavy precipitation events occurred in August 2010, August 2012, and July 2013 ( Figure 2). During 2010-2015, the total annual precipitation for each year was 1,075, 645, 910, 690, 409, and 627 mm, respectively, with a mean annual precipitation of 680 mm, indicating that 2010 and 2012 witnessed heavy precipitation, while 2014 witnessed extreme droughts. During the nongrowing period in 2012, 230.3 mm of precipitation was recorded, which was the highest during the study period.

N 2 O Emissions
The N 2 O emission peaks during 2010-2015 were 84.8, 27.7, 99.6, 23.8, 96.5, and 50.6 μg N m −2 h −1 for the SU treatment, respectively, and 249.8, 62.6, 93.6, 33.9, 172.2, and 160.5 μg N m −2 h −1 for the SF treatment, respectively, with the maximum daily N 2 O fluxes observed after top-dressing in the SF treatment. The highest peaks in the SF treatment appeared on July 28, 2010 (249.8 μg N m −2 h −1 ) after precipitation (76.7 mm on July 20 and 21, 2010) following N fertilizer top-dressing at the jointing stage. The maximal peak of N 2 O emissions in 2010 was 7 times higher than that in 2013 (33.9 μg N m −2 h −1 ). Moreover, N 2 O emissions associated with freezing and thawing were observed in 2012-2013 ( Figure 3).
Among the three treatments, cumulative N 2 O emissions were significantly higher in 2010 than that in other years between the sowing and jointing stages in the CK and SF treatments (p < 0.05). Compared with the SF treatment, significantly higher N 2 O emissions were observed in the SU treatment between the sowing and jointing stages during 2011, 2012, 2014, and 2015 (p < 0.05, Table 1). Because of higher N application in the SU treatment than in the SF treatment during maize sowing, N 2 O emissions were 107.6% greater in the SU treatment vs SF treatment between the sowing and jointing stages. Except for 2012, the N 2 O emissions did not differ between the CK and SF treatments during 2010-2015 between the sowing and jointing stages (Table 1). Furthermore, during the 6-year study period, a significantly linear relationship between the soil N 2 O emissions between the sowing and jointing stages and N input was observed ( Figure 4). The N 2 O emissions in 1 month after top-dressing were 29.8-53.1% of the annual cumulative N 2 O emissions in the SF treatment. Compared with the SU treatment, the N 2 O emissions increased by 169.4% in the SF treatment in the month after top-dressing (Table 1). However, no significant difference was observed during 2010-2015 between the CK and SU treatments in 1 month after top-dressing ( Table 1).
The CK treatment exhibited the lowest N 2 O emissions, with an average flux of 0.466 kg N ha −1 during the 6 years ( Table 1). The annual cumulative N 2 O emissions from the SU treatment were 0.343-2.539 kg N ha −1 , with extremely high N 2 O emissions observed in 2010 (Table 1). No significant differences were observed between SU and SF in average N 2 O emissions across the 6-year observation.

CH 4 flux
The CH 4 flux did not exhibit seasonal or annual trends, and each treatment represented either a small sink or a small source, with the CH 4 magnitude ranging from −60.5 to 46.0 μg m −2 h −1 ( Figure 5). Negative flux days accounted for 70.7% of the total sampling days across all treatments. The CH 4 flux was positively correlated with soil WFPS (p < 0.01). Furthermore, soil CH 4 uptake between the sowing and jointing stages was significantly in linear relation with the N input ( Figure 6).
Compared with CK, CH 4 absorption increased by 40.3% in SU in the maize growing season. Total CH 4 uptakes in the CK, SU, and SF treatments were −0.33, −0.38, and −0.35 kg CH 4 -C ha −1 year −1 , respectively. Two-way ANOVA results indicated that treatments had no significant effect on the cumulative CH 4 flux at the annual scale ( Table 2, p > 0.05).

Maize Yield, GWP, and Yield-Scaled GWP
Maize yield in the SU and SF treatments was not significant ( Table 3, p > 0.05). The GWP of SU and SF were significantly higher than that of CK ( Table 3, p < 0.05). The yield-scaled GWP  Frontiers in Environmental Science | www.frontiersin.org January 2022 | Volume 9 | Article 798383 were 21.4, 22.9, and 25.3 kg CO 2 eq t −1 grain for CK, SU, and SF, respectively. Compared to CK, the yield-scaled GWP in SF treatment increased by 18.7% (Table 3, p < 0.05).

NH 3 Volatilization
During sowing, the maximum NH 3 fluxes were observed after 3 d of basal fertilizer application (Figure 7). After 1 month of the basal fertilizer application, NH 3 flux reached the background level. However, during the jointing stage, NH 3 flux was the highest (approximately 2.51 kg N ha −1 d −1 ) immediately after fertilization. The NH 3 emission fluxes persisted for approximately 1 week at the maize jointing stage. Furthermore, NH 3 volatilization in the SF treatment accounted for 16.7% of the total N applied during this stage. Although NH 3 volatilization during top-dressing persisted for a short period, the intensity of NH 3 volatilization was high. Thus, the cumulative NH 3 loss from SF was 272% more than that from SU.

Effects of Amount and Distribution of Precipitation on N 2 O Emissions
The inter-annual N 2 O emissions significantly fluctuated during the study period. Similar results were reported by previous studies (Du et al., 2006;Scheer et al., 2008;Song et al., 2009;Zhang et al., 2014). Since the rates and methods of N application were identical each year, the variations in the inter-annual N 2 O emissions could be strongly related to fluctuations in the amount and distribution of precipitation.  In 2015, after a prolonged dry period (20 d), the soil WFPS at 0-5 cm depth was 10% (July 24); additionally, the precipitation (24.2 mm) on July 25 induced a large N 2 O flux in the SF treatment (160 μg N m −2 h −1 ). Pulse of N 2 O fluxes following rewetting were commonly observed in croplands, tropical forests, and grazed pastures (Davidson, 1992;Van Haren et al., 2005;Barton et al., 2008;Kim et al., 2010). The mechanisms underlying this phenomenon may be due to the   Frontiers in Environmental Science | www.frontiersin.org January 2022 | Volume 9 | Article 798383 enhanced metabolic activities of microorganisms, gas displacement, and reduced diffusivity (Kim et al., 2010).

Annual N 2 O Emissions
Cumulative N 2 O emissions for the SU treatment in 2010 were approximately 7 times more than those in 2013 because 2010 received the highest precipitation (998 mm, 47% more than the mean annual precipitation). However, heavy precipitation did not always induce higher N 2 O emissions. This was particularly evident in the last month of the growing season (August 2012), which received 315.2 mm of precipitation. These results were consistent with the results of Liu et al. (2011). In this study, low N 2 O emissions in the presence of high precipitation could be explained by low soil nitrate concentrations (2.32 mg N kg −1 ), which were lower than the threshold required for denitrification (Dobbie and Smith, 2003). Therefore, this means that soil nitrate content could be a limiting factor of N 2 O emissions. Despite lower annual precipitation in 2014 than in 2013, cumulative N 2 O emissions in SU in 2014 were approximately 2.6 times more than those in 2013 possibly because of the time of the precipitation events. The precipitation in May 2013 (15.4 mm) contributed to the low N 2 O emissions in 2013 between the sowing and jointing stages. Although the precipitation in July 2013 after top-dressing (228 mm) was the highest during the study period, only a relatively small peak of emissions occurred after top-dressing in the SF treatment ( Figure 3). Thus, the timing of the precipitation events and its precipitation distribution strongly affected N 2 O emissions (Burton et al., 2008).
In the nongrowing season, N 2 O emissions played an important role in annual N 2 O budgets and were increased by significantly high precipitation. For example, during significantly high precipitation in the nongrowing season of 2012-2013, N 2 O emissions from the freeze-thaw cycles contributed 43% to the annual N 2 O budgets in the SU treatment during 2012-2013. Therefore, during development and designing of N 2 O mitigation strategies, emissions associated with the nongrowing season CK: control; SU: single application of N fertilizer; SF: split application of N fertilizer. C CH4 b, C CH4 j and T CH4 denotes cumulative CH 4 emission between the sowing and jointing stages, 1 month after N applied at the jointing stage and the whole observation year, respectively. Lowercase letters within one column indicate differences among fertilizer treatments within the same year; capital letters in the row indicate differences among years within the same treatment.
Different lowercase letters indicate significant differences (p < 0.05)." with" the same letters were not significantly different (p > 0.05). Frontiers in Environmental Science | www.frontiersin.org January 2022 | Volume 9 | Article 798383 should be considered. Moreover, we speculated that the heavy precipitation and soil temperature along with split application would cause NH 3 , nitrogen oxide (NO), and NO 3 − -N losses. Pulse emission of NO after fertilization and precipitation/irrigation contributed to 88% of the annual NO emissions (Cui et al., 2012). Thus, studies are required to simultaneously quantify N 2 O, NH 3 , NO, and NO 3 − -N losses that can improve the estimation of complete impacts of split application of fertilizers on nitrogen losses.

N 2 O Flux
Our results showed that split fertilization significantly reduced the cumulative N 2 O emissions by 51.8% compared with single fertilization between the sowing and jointing stages. The acquired significant linear relationship between the soil N 2 O emissions (between the sowing and jointing stages) and N input, which is consistent with previous studies (Bouwman, 1996;Kim et al., 2013), was adopted by IPCC (2019) to estimate the N 2 O emissions in managed agricultural soils. However, nonlinear responses of N 2 O emissions to N fertilizer could be observed when high amount of N fertilizer was applied (Takeda et al., 2021).
During the study period, N 2 O emission pulse persisted for 1 month after top-dressing ( Figure 3). The combined effect of N fertilization and heavy precipitation were major drivers of N 2 O emissions (Zhang et al., 2014). Thus, the increase in N 2 O emissions after top-dressing offsets the benefits of split application during the early growth stages of maize (Aita et al., 2015).

Annual N 2 O Emissions
No significant difference in the annual N 2 O emissions was observed between the SF and SU treatments (p > 0.05). In contrast to our hypothesis, split fertilizer application did not mitigate N 2 O emissions. Similar results were observed by Wang et al. (2016), who reported no signficant differences in N 2 O emissions between the split N application and single N application in a rainfed maize-wheat rotation system in the semiarid Loess Plateau. Additionally, Venterea et al. (2016) observed no significant differences between single urea application and split application on the N 2 O emission in a rainfed maize field during a 2-year study period. Thus, factors, such as soil moisture content, temperature, and C availability, that affect N 2 O production may be comparatively more important in regulating N 2 O flux than the crop stage (Venterea et al., 2016).
Based on the findings, we concluded that the single N fertilizer application was a better alternative than the two split applications in mitigating the N 2 O emissions in the rainfed maize fields in Northeast China. In the same region,  reported 1.42 kg N 2 O-N ha −1 emissions in three split applications with N applied at a rate of 150 kg N ha −1 during the maize growing season. Thus, this type of N application emitted more N 2 O (1.42 vs. 1.02 kg N 2 O-N ha −1 ) despite having low rates of N application (150 vs. 180 kg N ha −1 ). In addition, single fertilizations could reduce the fuel expenses and mitigate CO 2 emissions from tractors that are generally used for fertilizer application (Huérfano et al., 2015). Therefore, compared with split fertilization, single N application can reduce the farming costs, conserve the environment, and maintain high maize yields.

N 2 O Direct Emission Factors
In this study, the average EF in the SU (0.308%) and SF (0.376%) were much lower than 1% as suggested by IPCC (2006) probably because of low soil organic carbon (SOC) content (Wang et al., 2016). Stehfest and Bouwman (2006) concluded that N 2 O emissions were significantly lower from soils with SOC <3%. A review by Cayuela et al. (2017) indicated that soil under dry Mediterranean conditions, with an average annual precipitation >450 mm, exhibited an EF of 0.32%. Australia uses a countryspecific EF of 0.2% for N fertilizer application in rainfed agriculture (Schwenke et al., 2016). The present study conducted in a rainfed area exhibited an EF that was lower than that in irrigated regions, such as North China Plain (0.61-0.77%), probably because of the absence of irrigation and low precipitation (Jiang J. et al., 2017). Previous studies conducted in Northeast China and Loess Plateau, which were under rainfed conditions, indicated that the EF in these regions was similar to that observed in our study site (Chen et al., 2014;Wang et al., 2016), and exhibited a substantial impact on estimating the total national N 2 O emissions inventories from the cropping systems (Cayuela et al., 2017). The EF estimated in the present study could assist in quantifying the contribution of maize fields in rainfed systems to the national N 2 O emission inventories in China.

CH 4 Fluxes
The absence of distinct seasonal trends of CH 4 flux estimated in this study was also observed in other studies (Liu et al., 2015;Wang et al., 2015;Fan et al., 2021). Soil WFPS only explained 3.3% of the variations in CH 4 fluxes in the CK treatment. Additionally, emission or uptake of CH 4 was controlled by several factors, such as soil texture, soil nutrients, and microbial activity (Carter et al., 2011;Dijkstra et al., 2013;Ran et al., 2017;Fan et al., 2021).
Some studies have indicated that N fertilizer application exceeding 100 kg ha −1 year −1 can inhibit CH 4 absorptions (Aronson and Helliker, 2010). In this study, the cumulative CH 4 absorptions between the sowing and jointing stages increased with increasing N fertilization. Furthermore, low or no fertilization inhibited the growth of methanotrophs, whereas high fertilization promoted the growth and activity of methanotrophs and improved the methane oxidation capacity (Wang et al., 2015;Ran et al., 2017).

Yield-Scaled GWP
There was no significant difference in maize grain yields between single application of urea at the sowing stage and two split applications at the sowing and jointing stages (p > 0.05), which was consistent with the previous study (Venterea et al., 2016). Therefore, in the local maize planting mode, the single application at the seeding stage can be used instead of the traditional application at the seeding stage and jointing stage, which can not only save labor costs but also effectively reduce greenhouse gas emissions. Single fertilization can be adopted and used by farmers in maize cultivation through extensive field trials and demonstrations. However, splitting N application has also been reported to increased crop yield by previous studies (Wang et al., 2016;Du et al., 2019;Lu et al., 2021). The differences between these studies may be related to the different amount and times of fertilization, as well as the differences in soil properties, climate, and other factors. With the increased times of split N fertilizer, there generally may be an increase in maize yields, but it may also increase greenhouse gas emissions, inevitably increasing labor costs. Therefore, it is necessary to further study the effects of fertilization frequency and the application rate on GHG and NH 3 emissions, as well as their relationship with yields. In the present study, the yield-scaled GWP ranged from 4.42 to 31.58 kg CO 2 eq Mg −1 in the N-fertilized treatment throughout the study period except 2010 (significantly high N 2 O emissions were observed in 2010 because of high precipitation). Jiang C. M. et al. (2017) reported 430 kg CO 2 eq −1 ha −1 and 43.9 kg CO 2 eq Mg −1 of GWP and yield-scaled GWP of maize, respectively. The relatively small values of GWP in our study could be because of low fertilizer splits [twice in our study vs. thrice in the study by ] and high maize yields. Moreover, GWP was significantly lower than that reported by Ma et al. (2013), who observed 610-620 kg CO 2 -eq Mg −1 of grain in a rice-wheat rotation field in eastern China probably due to high N application (480 kg −1 N ha −1 ) and high CH 4 emissions in rice fields.

Effects of Split Application of Nitrogen Fertilizer on NH 3 Emissions
The NH 3 volatilization loss was significantly higher in the basal stage than in the top-dressed urea application. The difference in NH 3 loss between the basal and top-dressed fertilization may be associated with high air temperature in the top-dressed fertilization (17°C vs. 25°C) and precipitation (4.9 mm on May 18 and 2 mm on May 19 vs. 2.5 mm on June 27 and 11.8 mm on June 29), which accelerated urea hydrolysis, thereby increasing the soil NH 4 + -N content (21 mg N kg −1 dry soil). The NH 3 EF of the SF treatment was 16.7%, which was higher than that observed by Li et al. (2019), who used ammonium sulfate as a N fertilizer. Thus, the increased NH 3 and N 2 O losses during top-dressing may affect the maize yield.

CONCLUSION
Split application of urea initially decreased N 2 O emissions between the sowing and jointing stages, but an increase in N 2 O emissions was observed after top-dressing in the SF treatment, that offset the benefits of early mitigation of N 2 O. During the study period, no significant differences were observed between SU and SF in N 2 O emission. However, split application of urea significantly increased NH 3 losses compared with single application. Moreover, no significant difference in the maize yield was observed between the SU and SF treatments. From the perspective of environment, single fertilization is more suitable than split fertilization in study area.

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 authors.

AUTHOR CONTRIBUTIONS
WY and HX contributed to conceptualization; HS and SK contributed to methodology; DD and SK contributed to investigation; DD contributed to writing-original draft preparation; WY and HS contributed to writing-review and editing; DD and SK contributed to visualization; WY contributed to supervision; HX and DD contributed to project administration; HX and DD contributed to funding acquisition. All authors have read and agreed to the published version of the manuscript.