Does replacing chemical fertilizer with ryegrass (Lolium multiflorum Lam.) mitigate CH4 and N2O emissions and reduce global warming potential from paddy soil?

The incorporation of ryegrass (Lolium multiflorum Lam.; RG), a winter manure, could partly replace chemical N and reduce N loss during the succeeding rice seasons, but little is known about its impact on greenhouse gas emission. This study investigated the effect of different RG-urea substitution ratios (0% RG and 100% urea, 25% RG and 75% urea, 50% RG and 50% urea, 75% RG and 25% urea, 100% RG and 0% urea,) on net C and N mineralization, CH4 and N2O emissions in a paddy soil. Gas samples for CH4 and N2O fluxes were collected by using a closed chamber and determined by chromatograph method. Net C and N mineralization from the incorporated RG residue were tested by a mesh bag method. Net C and N mineralization from RG followed a single exponential decay model, with 95.5%-97.8% of the original C and 98.7%-99.3% of N released during 192 days. The RG-urea substitution ratio increased CH4 emission, but was negatively correlated with N2O emission. In comparison with 0% substitution, global warming potential (GWP) and greenhouse gas intensity (GHGI) were not significantly different for the 25% and 50% RG substitutions, but were significantly higher for the 75% and 100% substitutions (P < 0.05). Soil redox, C and N remaining in litter residue were key characteristics explaining CH4 emission, while NH4+-N and NO3−-N were correlated with N2O emission. The increased GWP by CH4 emission after RG incorporation could be offset by N2O reduction when RG-urea substitution ratio was 50% or less.


Introduction
Methane (CH 4 ) and nitrous oxide (N 2 O) are important greenhouse gases (GHGs) that contribute to global climate change (Stocker 2014). Rice paddy soil is one of the important sources for CH 4 and N 2 O production, and accounts for a significant proportion of global GHG emissions (Carter et al. 2014). Rice paddies are responsible for 48% of total anthropogenic CH 4 emission from agriculture (Carlson et al. 2016), while N 2 O emission from paddy soils contributes 19%-25% of the of the global N 2 O emissions (Van Groenigen et al. 2011). Therefore, it is crucial to mitigate GHG emissions from paddy soils in most rice production regions to maintain a sustainable grain food supply.
Green manure incorporation and chemical nitrogen (N) fertilizer applications are common soil management practices that may impact CH 4 and N 2 O emissions. First, the release of organic carbon (C) and inorganic N from these fertilizers determines the availability of substrates for CH 4 and N 2 O releasing microorganisms, such as methanogenic bacteria (Liu et al. 2022) and methanogenic Archaea (Zhang et al. 2018), ammonia oxidizers  and denitrifiers . Several studies have reported increased CH 4 emission after organic amendments such as green manure (Haque et al. 2012), crop residue, and animal manure incorporation (Bhattacharyya et al. 2012;Nguyen et al. 2020). N 2 O emission peaks are usually observed immediately after intensive chemical N fertilizer application (Shaukat et al. 2019). Meanwhile, soil conditions that drive CH 4 and N 2 O formation and consumption may be greatly changed following organic matter returning and fertilizer application. For example, during the early stage of organic matter decomposition, soil pH and dissolved oxygen concentration were dramatically reduced while dissolved organic C was increased in the paddy soil, which provide a favorable condition for methanogens and enhance CH 4 production during the rice growing seasons (Xu et al. 2017;Zhou et al. 2020).
The rapidly rising soil NH 4 + -N and NO 3 − -N concentrations after chemical N fertilizer application benefit nitrification and denitrification, and increase the potential of N loss in the form of N 2 O (Cecilio et al. 2022;Zhu et al. 2013). Replacing chemical N with green manures has potential to reduce N 2 O emission because inorganic N release from the incorporated organic matter is much slower than from chemical fertilizers (Mehnaz et al. 2019;Zhu et al. 2014). However, acetate released during anaerobic decomposition of organic matter acts as a terminal electron acceptor for respiration and provides substrate for methanogens' activities, resulting in increased CH 4 emissions (Amin et al. 2021;Zhu et al. 2012a, b). Nevertheless, most of the above observations focused on a single organic-chemical N substitution ratio, while the influence of different substitution ratios is poorly known.
The ratio of substitution of organic for chemical fertilizer is an important factor for determining C and N dynamics after fertilization (Lashermes et al. 2010) as well as the consequent changes in soil environment (Hou et al. 2022). In addition, the complex interaction between organic material and inorganic N under different organic-chemical N substitution ratios may greatly affect CH 4 and N 2 O emissions (Zou et al. 2009). We hypothesized that the trade-off between CH 4 and N 2 O emissions can be changed under different organic-chemical N substitution ratios.
Ryegrass (Lolium multiflorum Lam.; RG), a commonly used green manure in paddy soils, has potential to partly replace chemical N fertilizers in most Asian countries. Increased rice yield and N use efficiency have been observed after ryegrass incorporation alone or in combination with urea (Asagi and Ueno 2009;Zhu et al. 2014). However, the impacts of RG-urea substitution ratio on CH 4 and N 2 O emissions remains poorly understood. The objectives of this study were to: (1) test changes in overall emissions of CH 4 and N 2 O under different RG-urea substitution ratios and what is the optimal ratio; (2) explore the main soil factors as well as C and N mineralization from RG residue controlling CH 4 and N 2 O emission under different RG-urea substitution ratios.

Experimental description
Two open glasshouse pot experiments were conducted at the experimental station of Yangtze University, Jingzhou, Hubei Province, China (30°20′N, 112°12′E) in 2021. The regional climate was subtropical with annual mean air temperature 16.5 °C and precipitation 1200 mm. The paddy soil (0-15 cm) was collected from a nearby farmer's paddy field followed by manual removal of stems, roots, stones, etc. The soil was then air-dried and sieved to pass 2 cm mesh. Soil properties were: sand 265 g kg −1 , clay 134 g kg −1 , silt 601 g kg −1 , pH 6.2, total N 1.4 g kg −1 , organic C content of 14.2 g kg −1 , available phosphorus 15.6 mg kg −1 and available potassium 87.0 mg kg −1 . The fresh RG (cv. Muteli) aboveground litter had moisture content 79%, dry matter N content 4.94% and dry matter C content 38.1%. Experiment 1: effects of RG-urea substitution ratio on CH 4 and N 2 O emission The experiment comprised five RG-urea substitution ratios: 0%, 25%, 50%, 75% and 100% with 3 replicates for each treatment. All the treatments received a total amount N at a rate of 100 mg N kg −1 air-dried soil. The details for RG and urea application amount in each treatment were shown in Table 1. Plastic pots (20 cm in diameter and 15 cm in height), each containing 3 kg air-dried soil was used. The aboveground litter of RG was weighed, cut into 2-3 cm pieces and thoroughly mixed with soil in mid-April 2021. Superphosphate and potassium chloride were applied at 0.61 g kg −1 and 0.21 g kg −1 air-dried soil, respectively. On April 25, three rice seedlings (cv. Tianliangyou 616, 30 d old) were transplanted per pot. The seedlings were watered to 2 cm flooding depth from the beginning to the end of the experiment and received no topdressing fertilizer during the whole growing season. The plants were harvested at 40 cm cutting height for the first maturity, leaving the stubbles for ratooning to get a second harvest. Experiment 2: net C and N mineralization during RG decomposition In experiment 2, the mesh bag method (Zhu et al. 2014) was used to measure net C and N mineralization dynamics from the incorporated RG litter. Treatments were the same as in experiment 1. RG aboveground litter was cut and placed in 0.5 mm mesh size nylon bags (10 cm × 10 cm) and buried in the soil with 39 replications for each treatment.

Sampling and analysis
In Experiment 1, gas samples for CH 4 and N 2 O fluxes were collected using a static chamber ) at 3-7 d intervals after RG incorporation until the harvest of the ratoon crop. The pot with plant was placed in a closed PVP pipe chamber (height 110 cm, diameter 25 cm). An electric fan at the top of the chamber was used to mix the air in the chamber. The gas samples were collected from the chamber at 0, 10 and 20 min between 9:00 and 11:00 am and transferred to 0.5 L sample bags. CH 4 and N 2 O concentrations were determined by gas chromatography (Agilent 7890 B, USA) with a flame ionization detector (FID) at 200 °C and an electron capture detector (ECD) at 330 °C, respectively. Soil pH and redox state (Eh) were simultaneously measured by pH meter (Leici PHS-25, China) and ORP meter (Leici TR-901, China), respectively. Rice grain was separated by hand after harvest and was weight after oven drying at 75 °C to constant weight. The two seasons of grain yield were summed to make a total annual rice grain yield for each pot.

Calculations
The CH 4 and N 2 O fluxes were calculated in detail as follows: where F is CH 4 flux (mg m −2 h −1 ) or N 2 O flux (μg m −2 h −1 ); p is density of CH 4 or N 2 O under the where F is the CH 4 (mg m −2 h −1 ) or N 2 O (μg m −2 h −1 ) flux, i is the sampling time, D i+1 -D i is the number of days between two adjacent sampling time (h), and n is the total number of sampling intervals.
Global warming potential (GWP) was calculated as the sum of N 2 O (GWP N2O ) and CH 4 (GWP CH4 ) as IPCC (2021): The greenhouse gas intensity (GHGI) was calculated as follows: Percentage of C and N remaining in RG residues were calculated as described by Zhu et al. (2014): where Yc is the C or N remaining percent of RG residue, Yt is the C or N content of RG residues at each sampling time, Yi is the original C or N content in RG.
An exponential decay model was used to describe net C and N mineralization from RG litter over time as follows (Zhu et al. 2014): where YR is C or N remaining in RG residue at time t and k is the C (k C ) or N (k N ) release rate.

Statistical analysis
Before the analysis of variance (ANOVA), data was tested for normality by using exploration in IBM SPSS Statistics 19.0 (IBM Co., NY, USA). The data Seasonal CH 4 was log-transformed when it was not in normality (rice grain yield and GHGI). Analysis of variance (ANOVA) for RG substitution ratio on CH 4 and N 2 O emission fluxes, GWP, GHGI, grain yield and soil properties was conducted with SPSS Statistics 19.0 (IBM Co., NY, USA). The multiple comparisons significant differences were set at P < 0.05. Pearson correlation analysis was used to examine the correlation between greenhouse gas emissions and soil properties. Canoco 5 (Beijing Huanzhong Ruichi Technology Co., LTD, China) was used to make the redundancy analysis (RDA) between GHG emissions and soil properties.

CH 4 and N 2 O emissions
A major peak was observed at 38 d after urea and RG application for CH 4 emissions (Fig. 1a), and at 17 d for N 2 O emissions (Fig. 2a). The highest CH 4 peaks were found in the 100% RG treatment (41.3 mg m −2 h −1 ), followed by 75% RG, 50% RG, 25% RG and 0% RG (Fig. 1a). By contrast, N 2 O peaks were observed to be highest in 0% RG (784.9 μg m −2 h −1 ) and decreased with the rising RGurea substitution ratio (Fig. 2a).
Most of the CH 4 (54.3%-63.7%) was emitted before the tillering stage and 73.7%-82.1% of the CH 4 emission took place before the first season maturity. Regression analysis showed that there is a significant quadratic relationship between RG-urea substitution ratio and total CH 4 emission (y = 13.53x 2 -2.79x + 9.36, R 2 = 0.99, p < 0.01; Fig. 1b). As for N 2 O, 65.4%-79.6% of the seasonal emissions occurred during the tillering stage and 91.5%-98.3% of the total emissions were contributed before the first season. A significant quadratic relationship was identified between RG-urea substitution ratio and total N 2 O emission (y = 0.48x 2 -0.8x + 0.39, R 2 = 0.99, p < 0.01; Fig. 2b).

Global warming potential and greenhouse gas intensity
The GWP for seasonal CH 4 and N 2 O emissions showed a significant quadratic relationship with RG-urea substitution ratio (y = 496.58x 2 -263.31x + 387.85, R 2 = 0.99, p < 0.05; Fig. 3a). The values for GWP were lowest in the 25% and 50% RG treatments (6.3% and 6.5% lower than 0% RG, respectively), which were significantly lower than those in 75% and 100% RG. As shown in Table 2, contribution of CH 4 emission to the GWP (71.7%-96.7%) was much higher than that from N 2 O emission (3.3%-28.3%). CH 4 and N 2 O emissions from the first 45 d of the experiment accounted for 47.5%-56.4% of the total GWP while comparable RG-urea substitution ratios: 0% RG, 25% RG, 50% RG, 75% RG, 100% RG. Tillering stage represents cumulative CH 4 emissions from the start of experiment to rice tillering stage, filling stage represents cumulative CH 4 emissions from tillering stage to rice filling stage, first maturity represents cumulative CH 4 emissions from filling stage to first sea-son rice maturity, second maturity represents cumulative CH 4 emissions from first season harvest to the second season maturity. Bars indicate the SE (standard error). The quadratic linear regression relationship between CH 4 cumulative emission and RG-urea substitution ratio was significant (P < 0.01). Different letters indicate significant difference level (P < 0.05) Fig. 2 N 2 O emission flux (a) and cumulative N 2 O emissions (b) of different RG-urea substitution ratios: 0% RG, 25% RG, 50% RG, 75% RG, 100% RG. Tillering stage represents cumulative N 2 O emissions from the start of experiment to rice tillering stage, filling stage represents cumulative N 2 O emissions from tillering stage to rice filling stage, first maturity represents cumulative N 2 O emissions from filling stage to first sea-son rice maturity, second maturity represents cumulative N 2 O emissions from first season harvest to the second season maturity. Bars indicate the SE (standard error). The quadratic linear regression relationship between N 2 O cumulative emission and RG-urea substitution ratio was significant (P < 0.01). Different letters indicate significant difference level (P < 0.05) total GWP (43.6%-52.5%) were detected in the relatively longer following 46-125 d (Fig. 3b).
Rice grain yield in the 25% and 50% RG treatments was not significantly different from that in the 0% RG treatment, but was significantly higher than that in the 75% and 100% RG treatments (Table 2; p < 0.05). Similar as GWP, GHGI was higher in the 75% and 100% RG treatments as compared to the 0%, 25% and 50% RG treatments (Table 2).

Net C and N mineralization from RG residue
The majority of C (83.2%-86.9%) and N (97.4%-97.5%) from the incorporated RG residue was released during the first 29 d after start of the experiment (Figs. 4a and b). At the end of the experiment, less than 4.5% of C and 1.3% of N remaining remained in RG residue. The decay exponential model successfully described C (R 2 = 0.92-0.94, p < 0.05) and N (R 2 = 0.98-0.99, p < 0.05) release

Fig. 3
Global warming potential (GWP, a) of different RGurea substitution ratios: 0% RG, 25% RG, 50% RG, 75% RG and 100% RG. GWP CH4 represents GWP caused by CH 4 emission, GWP N2O represents GWP caused by N 2 O emission. Bars indicate the SE (standard error). The quadratic linear regression relationship between GWP and RG-urea substitution ratio was significant (P < 0.01). Different letters indicate significant difference level (P < 0.05). Percentage of GWP (b) before and after major CH 4 peaks. 0-45d represents GWP calculated by cumulated CH 4 and N 2 O emissions from the start of fertilization to 45 days, 45-192d represents GWP calculated by cumulated CH 4 and N 2 O emissions from 45 to 192 days  (Table 3). The values for k N were generally higher than for k C (0.0877-0.1368 vs 0.0689-0.0800), indicating that N release from RG residue was faster than C release. The values for k N increased with higher RG-urea substitution ratio while the opposite was true for k C ( Table 3).

Importance of soil properties for CH 4 and N 2 O emissions
Soil NH 4 + -N concentrations were much higher than NO 3 − -N concentrations during the experiment (Figs. 5a and b). Both NH 4 + -N and NO 3 − -N increased steadily and peaked 14-29 d after RG and urea application. NH 4 + -N and NO 3 − -N appeared to be highest in 0% RG and lowest in 100% RG treatment. Redox state (Eh value) decreased gradually within 30-40 d after the start of the experiment and increased steadily thereafter. Lower Eh values tended to be observed in higher RG-urea substitution ratios (Fig. 5c). Soil pH decreased during the first two months especially for the 25%, 50% and 100% RG treatments (Fig. 5d).
Distance-based RDA showed that the first two components accounted for 63.1% and 64.8% of the variation in CH 4 and N 2 O emissions, respectively (Figs. 6a and c). C remaining (F = 49.8, p = 0.002), N remaining (F = 7.8, p = 0.014) and soil Eh (F = 6.9, p = 0.032) had a significant impact on CH 4 emission when NH 4 + -N (F = 15.4, p = 0.004) and NO 3 − -N (F = 6.6, p = 0.034) showed a significant impact on N 2 O emission (Table 4). Variance decomposition showed that 90.3% of the variance of CH 4 emission could be explained by C and N remaining in RG residue, soil Eh and other indicators (Fig. 6b), while 84% of the variance of N 2 O emission was explained by soil NH 4 + -N, NO 3 − -N and C remaining in RG residue (Fig. 6d).

Effects of RG-urea substitution ratio on CH 4 emission
Incorporation of organic amendments such as green manure and crop straw is an important soil practice to increase soil organic C in agricultural fields, but the enhanced active C input may stimulate CH 4 production in paddy soils (Fu et al. 2018). This was confirmed by the rapidly increasing CH 4 emission Fig. 4 Percentage of C (a) and N (b) remaining in ryegrass residue during 192 days after fertilization under different RG-urea substitution ratios: 25% RG, 50% RG, 75% RG and 100% RG Table 3 The rate of C (k C ) and N releases (k N ) from ryegrass (RG) residue under different RG-urea substitution ratios, estimated using a single exponential model after RG incorporation, especially in the treatments with high RG-urea substitution ratios (Fig. 1). Dramatically increased CH 4 emissions have also been reported in flooded paddy soils with fresh green manure incorporation (Lee et al. 2010;Liu et al. 2019). In this study, C remaining in the incorporated RG litter had the strongest impact on CH 4 emissions (F = 49.8, p < 0.01; Fig. 6a), suggesting that C input through RG incorporation was an important source for CH 4 production ( Table 4). RG-urea substitution ratio affected the rate of green manure decomposition, which consequently determined the availability of soil C substrates for methanogens (Dalal et al. 2008;Xu et al. 2017). Soil organic C pools such as dissolved organic C has been shown to be correlated with CH 4 emissions after green manure application in long term field experiments (Raheem et al. 2019;Xu et al. 2017). Soil condition is another factor affecting the activity of methanogens and methane-oxidizing bacteria which are responsible for CH 4 production and consumption in the soil. In the present study, effects of soil condition changes by RG-urea substitution ratio on GHG emission could be better measured in a pot study than in field experiments, because some undesirable interferences such as rainfall and leaching could be well prevented. In the present study, rapid decomposition of RG litter may have reduced dissolved oxygen concentration in the flooded paddy soil, resulting in a decrease of soil Eh (Fig. 5c). A low Eh provides a suitable condition for methanogens and reduces CH 4 oxidation in the soil (Muhammad Significantly higher CH 4 emissions in low redox soils have also been reported by Fan et al. (2020) and Xu et al. (2017). About 5% of the variance in CH 4 emission was explained by changes in soil Eh while 0.6% was explained by pH (Fig. 6b). Decreased soil pH by RG incorporation (Fig. 5d) was attributed to the formation of simple organic acids such as acetic acid and butanoic acid during the process of RG fermentation. This lower pH and increased acetic acid availability stimulates the activity of methanogens (Baumann et al. 2009;Qualls 2005).  Effects of RG-urea substitution ratio on N 2 O emission Green manure applied alone or in conjunction with chemical fertilizers has been reported to improve N use efficiency and decrease N losses into the environment Liang et al. 2022;Zhu et al. 2012a, b Cowan et al. 2021). In the current study, a major N 2 O peak was detected following urea application while several minor peaks were measured afterwards (Fig. 2a). Similar with Zhang et al. (2020), both the major peak and seasonal N 2 O emissions were significantly decreased with the increase of RG-urea substitution ratio, in spite that the soils received a same amount of total N input (Fig. 2). Soil mineral N concentrations, especially NH 4 + -N content, were significantly decreased with RG-urea substitution ratio in the peaking stage (Fig. 5a). This may result in a reduced N available for nitrification or coupled nitrification-denitrification which are the main mechanisms for N 2 O production in agricultural soils Nie et al. 2021). In addition, RG planting has the potential to take up native soil mineral N which may be lost in the form of N 2 O during winter seasons. Another possible reason for N 2 O reduction could be attributed to the lower redox condition after RG incorporation which reduced the activity of ammonia oxidizers (Tao et al. 2018). These results supported the beneficial effect of replacing chemical N with green manure on N 2 O mitigation. To trace N 2 O production from different N sources in a paddy ecosystem, 15 N labeled green manure or chemical fertilizer should be used in further studies.

Global warming potential and greenhouse gas intensity
The assessment of GWP is important because of the dramatically different effects of RG substitution on CH 4 and N 2 O emissions in this study. The majority of GWP (71.7%-96.7%) were caused by CH 4 emission, confirming that CH 4 is the dominant GHG responsible for radiative forcing in the studied paddy soil (Peyron et al. 2016;Wang et al. 2016). Nevertheless, GWP in 25% and 50% RG-urea substitution ratios were slightly decreased when compared with 0% RG (361.5-362.1 g CO 2 -eq m −2 vs 386.5 g CO 2 -eq m −2 ) while CH 4 -induced GWP increased with RG-urea substitution ratio (Table 2). This confirms a positive trade-off between CH 4 emission and N 2 O emission for the 25% and 50% RG-urea substitution treatments (Shang et al. 2011). However, RG-urea substitution ratios higher than 50% tended to increase GWP due to the rising CH 4 emission. Nearly half of the GWP was observed during the first 45 d after fertilization (Fig. 3b), suggesting that GHG mitigation strategies in paddy soil with organic matter incorporation should be focused on the early stages of organic material decomposition and rice growth. Shang et al. (2011) suggested incorporating organic matter when the soil is drained and before rice is planted in order to minimize CH 4 and N 2 O emissions. The high C/N ratio of RG litter which enhanced N competition between crop roots and soil microorganisms may be the reason for the decreased rice yield, especially in high RG-urea substitution ratios (Manzoni et al. 2008;Cao et al. 2020). However, rice grain yield was not significantly decreased by 25% and 50% RG-urea substitution ratios, making their GHGI values comparable with 0% RG ( Table 2). The increased GHGI under 75% and 100% RG-urea substitution could be attributed to the promoted GWP and the reduced grain yield. It should be mentioned that in farmer's practice, plant root with higher C/N ratio than aboveground litter is incorporated in the soil simultaneously which may decompose more slowly. The optimal RG-urea substitution ratio for rice growing and GHG mitigation should be overestimated in the current study while only the aboveground part was tested.

Conclusions
This study demonstrated that RG-urea substitution ratio played an important role in CH 4 and N 2 O emissions. CH 4 emissions were positively related with RG-urea substitution ratio, while the opposite was true for N 2 O emissions. Compared with 0% RG, the increased CH 4 emissions in the 25% RG and 50% RG substitution ratios were offset by decreased N 2 O emissions, leading to comparable GWP and GHGI values. The RG-urea substitution ratio affected C and N mineralization from RG residues, which further affected CH 4 and N 2 O emissions in the paddy soil. Soil Eh, C and N remaining were key characteristics correlated with CH 4 emission while NH 4 + -N, NO 3 − -N and C remaining were main factors on N 2 O emissions. As CH 4 was the main contributor to GWP in paddy soils, further studies should be taken to reduce CH 4 fluxes, especially in the early stage of organic matter returning.