The study site was located at the Jilin Songnen Grassland Ecosystem National Observation and Research Station at the Changling Horse Breeding Farm in western Jilin province in northeastern China (44°34′25″N, 123°31′6″E; 138-176 m above sea level). The study area has a semi-arid temperate continental monsoon climate. In the past 65 years (1953-2017), average annual air temperatures ranged from 3.40°C to 7.58°C, with an average value of 5.6°C (National Meteorological Information Center). The average annual precipitation is 445 mm, with more than 80% occurring during the growing season (1953-2017, National Meteorological Information Center). The percentage distributions of precipitation pulses of different sizes and their contributions to total precipitation during the growing season over the past 65 years are shown in Table S1 . Although precipitation pulses smaller than 5 mm were frequent (63%), they only accounted for 13.01% of the total precipitation during the growing season. In contrast, extreme pulses above 50 mm only had a frequency of 1.38%, but their contribution to the total precipitation (13.40%) was comparable to pulses below 5 mm in size.

The vegetation of the studied meadow steppe is dominated by Leymus chinensis. The zonal soil at the study site is classified as Salic Solonetz (World Reference Base for Soil Resources) or an Aqui-Alkalic Halosol (Chinese soil classification) (Ren et al, 2019; Shi et al., 2019). The soil is saline-alkaline with a pH value of 8.0-10.0 (Cui et al., 2021). Table S2 shows the soil’s chemical and physical properties at 0-10 cm depth as measured in the control and long-term N addition plots before the precipitation pulse treatments were applied.

In 2010, an area of 100 m × 100 m was fenced off in the L. chinensis meadow steppe, prohibiting grazing and mowing. In 2011, a long-term N addition experimental platform was established in the fenced area. Five blocks were set up in the experimental region, each with an area of 20 m × 10 m. Each block was then divided into two plots, each with an area of 10 m × 10 m. One plot was randomly assigned to N addition (10 g N m^-2 yr^-1) in each block, and the other unfertilized plot served as the control. In the long-term N addition plots, urea was applied every year in May and July at 5 g N m^-2. In early May 2020, the intact soil columns were collected by the soil column collector (external diameter = 30.3 cm, height = 50 cm). The collector rotated and cut into 40 cm of the soil, and then the intact soil columns were collected and placed in the pots (internal diameter = 30.3 cm, height = 45 cm). Ten intact soil mesocosms (height = 40 cm) were collected from each plot (n = 100). Half of the soil mesocosms (n = 50) were used for CH₄ flux measurement, and the other half (n = 50) were used for soil sampling. The collected soil mesocosms were placed in five blocks in the rainout shelter at the research station, with each block containing 20 mesocosms (ten control (CK) and ten N addition (NA) mesocosms). The mesocosms were buried 40 cm underground to reduce environmental interference and were watered to restore plant growth. From the middle of June to July 10^th, the mesocosms were watered once a week with 1.44 L water (equal to a 20 mm precipitation pulse) to keep consistent soil water status. At the beginning of both May and July, urea was added at 5 g N m^-2 to the surface of the N addition mesocosms. The precipitation pulses were conducted on July 31^st. Based on the 65 years of historical precipitation in the study site ( Table S1 ), five precipitation pulse sizes (0 mm, 5 mm, 10 mm, 20 mm, and 50 mm) were designed and applied randomly to a pair of the control (P0, P5, P10, P20, and P50) and N addition mesocosms (NP0, NP5, NP10, NP20, and NP50) in each block. The water was evenly sprayed into the P0 (0 L), P5 (0.36 L), P10 (0.72 L), P20 (1.44 L), and P50 (3.6 L) treatments using the sprayer in both control and N addition mesocosms. The CH₄ fluxes and soil samples were collected immediately after watering (0 h), as well as at 2, 4, 8, 12, 24, 72, 144, 288, and 432 h after the precipitation pulse treatments.

In June 2020, the open base collar (20 cm × 20 cm × 10 cm high), with a U-shaped groove (2.5 cm in width) around the upper edge, was permanently inserted into the soil of each mesocosm before the precipitation pulse experiment. The open-bottom chamber was tightly fitted to the collar during the gas sampling and sealed with water. The gas samples were collected from inside the chambers using a 100 mL plastic syringe fitted with three-way stopcocks at 0 min, 30 min, and 60 min after the chamber closure. The collected gas samples were immediately transferred to vacuumed gas sampling bags (LB-301, Dalian Delin Gas Packing Co., Ltd, Dalian, China). The concentrations of CH₄ were analyzed within one week using the N₂O/CH₄ analyzer (Model 913-1054, Los Gatos Research Inc., Mountain View, CA, USA).

During each gas sampling occasion, soil temperature was measured and soil samples were collected. At the end of the precipitation pulses, aboveground vegetation in the pots was clipped to estimate aboveground biomass. Soil temperature was measured at a 5 cm layer of each soil mesocosm using a thermocouple penetration probe (Li 6400-09 TC, Li-cor Biosciences, Lincoln, NE, USA). Soil samples were taken at 0-10 cm depth of soil mesocosm using a stainless-steel corer (inner diameter of 3.3 cm). The soil samples were placed in sterile bags, transported to the laboratory with a cooler box, and stored at 4°C for subsequent analysis. The fresh soil samples were sieved using 2 mm mesh and divided into two subsamples in the laboratory. Fresh soil subsamples were analyzed for ammonium nitrogen (NH₄ ⁺-N), nitrate nitrogen (NO₃ ^–-N), dissolved organic carbon (DOC), and microbial biomass carbon (MBC). The other subsamples were air-dried to determine pH and total carbon content (TC). The soil temperature, soil moisture, NH₄ ⁺-N content, NO₃ ^–-N content, and pH value were measured on each sampling campaign, and the DOC, MBC, and TC were measured at the end (432 h) of precipitation pulse treatments.

The soil moisture was determined by the oven-drying method. The pH values of the soils were measured in a 1:2.5 (soil: water ratio) suspension with a PHS-3E glass pH electrode (PHS-3E, Shanghai Precision & Scientific Instrument Co., Ltd, Shanghai, China). Soil NH₄ ⁺-N and NO₃ ^–-N content were extracted using 2 mol L^-1 KCl solution by shaking for 1 h before being analyzed using a Lachat flow-injection auto-analyzer (Futura Flow Analyser, Alliance Instruments, Frepillon, France). TC was analyzed with an elemental analyzer (Vario Max CN, Elementar, Hanau, Germany). MBC was determined by the chloroform fumigation-extraction method (Vance et al., 1987). Extractable organic C in the fumigated and unfumigated samples was measured using an elemental analyzer. The MBC was calculated as the differences in DOC in the soil extracts between the fumigated and unfumigated samples. The amount of organic carbon in the un-fumigated soil extracts was used as DOC (Zhao et al., 2016). The AGB was determined by oven-drying, and harvested biomass was oven-dried at 65°C to a constant mass when weighed.

The CH₄ fluxes were calculated from the change in CH₄ concentrations with time (Eq. S1). Cumulative CH₄ fluxes were linearly and sequentially accumulated from the fluxes between every two adjacent measurement intervals (Eq. S2). The impact-treatment is the relative effects of precipitation pulses of different sizes on CH₄ fluxes (average or cumulative) from 0 mm pulse (Eq. S3).

Two-way ANOVA was conducted to examine the effects of precipitation pulses, N addition, and their interactions on CH₄ fluxes as well as biotic and abiotic factors. Multiple comparisons were determined using Tukey’s HSD test at a probability level of 95% (P < 0.05). The correlations between CH₄ fluxes (average and cumulative) and average soil moisture were tested using a linear model. Binary linear functions were used to test the dependence of CH₄ fluxes on soil moisture and soil temperature in each treatment. A quadratic equation was developed to describe the relationship between CH₄ fluxes (average and cumulative) and precipitation pulse sizes. The structural equation model (SEM) was performed to analyze the direct and indirect effects of precipitation pulse and N addition on cumulative CH₄ fluxes using the lavaan package (Rosseel, 2012). The chi-square (χ²) test (P > 0.05), comparative fit index (CFI) > 0.9, and standardized root mean-square-residual (SRMR) value < 0.08 were used to indicate if the SEM models fit well. Statistical analyses were conducted using R Statistical Software (Version 4.1.2, R Corporation, Vienna, Austria) and IBM SPSS Statistics (IBM SPSS Statistics 25.0, IBM Corporation, Chicago, IL, USA). Results were presented as mean ± 1 standard error (SE). The graphics were drawn using OriginPro 2018 software (OriginPro 2018, OriginLab Corporation, Northampton, MA, USA).

The precipitation pulse (PP) treatments caused an immediate increase in soil moisture that was related to the PP sizes ( Figure 1A ). Soil moisture peaked at around 2-4 h and then decreased until the end of precipitation pulse treatments. The PP treatments of 5 mm, 10 mm, 20 mm, and 50 mm significantly enhanced the average soil moisture (P = 0.000, df = 4; Figure 1B ), which quadratically increased with PP size in both control and N addition treatments (all P = 0.000, df = 24; Figure S1 ). N addition (NA) as well as the interaction of PP and NA, however, did not substantially alter the average soil moisture ( Figure 1B ). Soil temperatures had significant temporal dynamics in the P0 and NP0 treatments, showing a unimodal trend ( Figure 1C ). PP, NA, and their interaction, however, had no significant effect on the temporal dynamics and average soil temperature ( Figure 1C, D ).

The PP treatments significantly affected the temporal dynamics of NH₄ ⁺-N content ( Figure 2A ), NO₃ ^–-N content ( Figure 2C ), and pH value ( Figure 2E ). The peaks of NH₄ ⁺-N content, NO₃ ^–-N content, and pH mainly occurred at 0-8 h, 2-12 h, and 24-144 h after the PP treatments, respectively. All sizes of PP significantly increased average NO₃ ^–-N content (P = 0.000, df = 4; Figure 2D ) but not average NH₄ ⁺-N content ( Figure 2B ). The 20 mm and 50 mm PP treatments significantly increased the pH, while the 5 mm PP treatment significantly decreased the pH (P = 0.000, df = 4; Figure 2F ). NA significantly increased NH₄ ⁺-N content (P = 0.000, df = 1; Figure 2B ) and NO₃ ^–-N content (P = 0.000, df = 1; Figure 2D ), but significantly decreased the pH (P = 0.000, df = 1; Figure 2F ). The interaction of PP and NA significantly affected the pH (P = 0.000, df = 4; Figure 2F ) but had no significant effects on NH₄ ⁺-N content ( Figure 2B ) or NO₃ ^–-N content ( Figure 2D ).

At the end of precipitation treatment, PP treatments significantly affected the DOC content (P = 0.005, df = 4; Figure 3A ), MBC content (P = 0.004, df = 4; Figure 3B ), and AGB (P = 0.026, df = 4; Figure 3D ), but not the TC ( Figure 3C ). Compared with the 0 mm PP treatment, 50 mm PP treatment significantly decreased DOC content ( Figure 3A ), while the 10 mm, 20 mm, and 50 mm PP treatments significantly increased MBC content and AGB ( Figure 3B, D ). NA significantly increased AGB (P = 0.000, df = 1; Figure 3D ), but did not substantially alter DOC content (marginal effect, P = 0.085, df = 1; Figure 3A ), MBC ( Figure 3B ), and TC ( Figure 3C ). The interaction of PP and NA had no significant effect on the DOC, MBC, TC, or AGB ( Figure 3 ).

The studied grassland acted as a sink for CH₄ in the P0 and NP0 treatments, with fluxes ranging from -8.22 to -3.78 μg m^-2 h^-1 and -6.91 to -2.57 μg m^-2 h^-1, respectively ( Figure 4A ). The 5 mm PP treatment slightly suppressed CH₄ uptake and average CH₄ fluxes, but not at statistically significant levels ( Figures 4A, B ). The 10 mm, 20 mm, and 50 mm PP treatments, however, substantially changed the temporal dynamics and the source-sink relationship of CH₄ fluxes ( Figure 4A ). The 10 mm PP treatment shifted CH₄ fluxes from sinks to sources within 2 to 4 h, while 20 mm and 50 mm PP treatments immediately triggered the release of CH₄ ( Figure 4A ). The CH₄ fluxes primarily peaked at 12 h and relatively decreased the CH₄ uptake by 203-362% and 243-333% compared with P0 and NP0 treatments, respectively ( Figure 4A ). After that, CH₄ releases decreased and absorption resumed at 144 h until the end of PP treatments ( Figure 4A ).

The PP significantly affected the average CH₄ fluxes (P = 0.000, df = 4; Figure 4B ). More specifically, the 10 mm, 20 mm, and 50 mm PP treatments significantly enhanced the average CH₄ fluxes (109-171%), but the 5 mm PP treatment had no significant effect ( Figure 4B ). The relative effect of PP treatment on CH₄ fluxes increased significantly with increasing PP sizes (P = 0.000, df = 3; Figure 4C ). The relationship between the average CH₄ fluxes and PP sizes could be fitted by a quadratic equation, explaining the 87% (P = 0.000, df = 24) and 81% (P = 0.000, df = 24) variation in the average CH₄ fluxes in the control and N addition treatments, respectively ( Figure 4D ). N addition significantly enhanced the average CH₄ fluxes (P = 0.048, df = 1; Figure 4B ) and the impact-treatment of average CH₄ fluxes (P = 0.001, df = 1; Figure 4C ). The interaction of PP and NA, however, had no significant influence on average CH₄ fluxes ( Figure 4B ) or the impact-treatment of average CH₄ fluxes ( Figure 4C ).

Cumulative CH₄ fluxes showed a decreasing trend following the 0 mm and 5 mm PP treatments, as CH₄ continued to be absorbed from the atmosphere by the soil ( Figures 4A , 5A ). The 10 mm, 20 mm, and 50 mm PP treatments significantly increased the cumulative CH₄ fluxes, reaching the highest cumulative fluxes at 144 h ( Figure 5A ). After that, cumulative CH₄ fluxes were reduced and the areas were converted into CH₄ sinks. During the experimental period, cumulative CH₄ fluxes ranged from -0.17 to -0.03 mg CH₄ pot^-1 and -0.16 to -0.02 mg CH₄ pot^-1 in the control and N addition treatments, respectively ( Figure 5A ).

The PP treatments significantly affected the cumulative CH₄ fluxes (P = 0.000, df = 4; Figure 5B ) and the impact-treatment of cumulative CH₄ fluxes (P = 0.000, df = 3; Figure 5C ). The 10 mm, 20 mm, and 50 mm PP treatments significantly reduced the absorption of CH₄ (68-88%), with a higher relative effect on cumulative CH₄ fluxes; in contrast, the 5 mm PP treatment did not have any effect ( Figure 5B, 5C ). There was a significant quadratic relationship between the cumulative CH₄ fluxes and PP sizes, which could explain the 64% (P = 0.000, df = 24) and 60% (P = 0.000, df = 24) variation in the cumulative CH₄ fluxes in the control and N addition treatments, respectively ( Figure 5D ). NA significantly increased the cumulative CH₄ fluxes (P = 0.015, df = 1; Figure 5B ), but had a marginal effect on the impact-treatment of cumulative CH₄ fluxes (P = 0.065, df = 1; Figure 5C ). The interaction of PP and NA, however, did not significantly affect the cumulative CH₄ fluxes ( Figure 5B ) and the impact-treatment of cumulative CH₄ fluxes ( Figure 5C ).

The binary linear model showed that the temporal dynamics of CH₄ fluxes were mainly driven by soil moisture and soil temperature after the PP treatments ( Table 1 ). Only the P0 and NP0 treatments had no significant relationship between CH₄ fluxes and soil moisture and soil temperature ( Table 1 ). The CH₄ fluxes were significantly positively correlated with soil moisture and soil temperature in the NP5 treatment (P < 0.01, df = 49), but were significantly negatively correlated with soil moisture in the P5 treatment (P < 0.05, df = 42). The CH₄ fluxes from control and N addition treatments were both significantly positively correlated with soil moisture and soil temperature in the 10 mm (both P < 0.01), 20 mm (both P < 0.01), and 50 mm (both P < 0.001) PP treatments. The degree of fitting increased with increasing PP sizes, explaining the 10-34% and 18-28% change in the temporal dynamics of CH₄ fluxes in the control and N addition treatments, respectively ( Table 1 ).

Structural equation model (SEM) analysis indicated that N addition significantly increased the NO₃ ^–-N content (P < 0.001), thereby decreasing the soil pH (P < 0.01; Figure 6 ). The pH change directly increased the cumulative CH₄ fluxes (P < 0.05) and indirectly enhanced the MBC (P < 0.01) and DOC (P < 0.001) through decreased TC (P < 0.01). The MBC and DOC significantly increased the cumulative CH₄ fluxes (both P < 0.05), while the TC had a marginal effect on cumulative CH₄ fluxes (P < 0.1). The precipitation pulses significantly increased the soil moisture (P < 0.001), which directly increased the cumulative CH₄ fluxes (P < 0.01). Additionally, soil moisture positively affected NO₃ ^–-N content (P < 0.001), pH (P < 0.05), and MBC (P < 0.001), but a negative effect on DOC (P < 0.01) and TC (P > 0.05); this, in turn altered the cumulative CH₄ fluxes. All factors jointly explained the 41% variation in cumulative CH₄ fluxes that were observed ( Figure 6 ).

The effects of precipitation pulses on CH₄ fluxes are very complex. Previous studies have shown that precipitation pulses could stimulate (Harms and Grimm, 2012; Petrakis et al., 2017), suppress (Zhao et al., 2017; Ren et al., 2019), or not significantly affect CH₄ fluxes (Mariko et al., 2007; Ni et al., 2019). The effects mainly depended on the precipitation pulse size, ecosystem type, and soil moisture status (Harms and Grimm, 2012; Kim et al., 2012; Petrakis et al., 2017; Zhao et al., 2017; Ni et al., 2019). Without manipulated precipitation, the studied grassland acted as a net sink of CH₄ ( Figures 4A , 5A ), which is consistent with the findings of previous studies (Dalal and Allen, 2008; Wang et al., 2014b; Yu et al., 2017; Zhao et al., 2017). The average rate of CH₄ absorption of the studied grassland (-5.23 ± 0.36 μg m^-2 h^-1, Figure 4B ) was, however, much lower than that of the semi-arid grassland from northeast China (-74.31 ± 62.51 μg m^-2 h^-1) as well as the Qinghai-Tibetan Plateau (-31.29 ± 21.78 μg m^-2 h^-1) (Wang et al., 2014b). As expected, the large (10 mm, 20 mm, and 50 mm) precipitation pulses had a negative pulsing effect on CH₄ fluxes and significantly suppressed CH₄ absorption, but the small (5 mm) precipitation pulse did not significantly alter it ( Figure 5B ).

Without water supplementation (0 mm PP treatment), the soil was arid and soil moisture was relatively stable ( Figure 1A ). Low soil moisture limited the activity of methanotrophs due to water stress, and the absorption rate of CH₄ was, therefore, very low (Le Mer and Roger, 2001; Borken et al., 2006; Aronson et al., 2019). Although the 5 mm PP treatment substantially increased average soil moisture ( Figure 1B ), it did not significantly alter the temporal dynamics or cumulative CH₄ fluxes ( Figures 4A , 5B ). Ni et al. (2019) also confirmed that short-term precipitation pulses did not alter CH₄ fluxes in a forest ecosystem, suggesting that small precipitation changes do not alter O₂ in soil pore spaces enough to affect CH₄ fluxes. As expected, the large precipitation pulses (10 mm, 20 mm, and 50 mm) altered the source-sink relationship of CH₄, converting CH₄ fluxes of the studied grassland from a sink to a source ( Figures 4A , 5A ), which were consistent with an in situ field study (Ren et al., 2019). The temporal dynamics of CH₄ fluxes were controlled by the soil moisture and soil temperature following the precipitation pulses ( Table 1 ). While the precipitation pulses significantly increased soil moisture, they did not alter soil temperature ( Figure 1 ). The changes in soil moisture dynamics caused by precipitation pulses, therefore, may be responsible for the changes in CH₄ fluxes. In this study, there were several potential mechanisms by which precipitation pulses could have triggered the CH₄ source-sink conversion. Firstly, the infiltration of soil water caused by the precipitation pulses could displace CH₄ trapped in the soil pore space and release it to the atmosphere, especially in the case of the 50 mm PP treatment. Secondly, a large precipitation pulse could increase the soil water availability and alleviate water limitation, decreasing the redox potential and availability of O₂ in favor of anaerobic processes, thereby promoting methanogenic activity and suppressing CH₄ oxidation (Harms and Grimm, 2012; Kim et al., 2012). Thirdly, the increased availability of water could stimulate microbial, specifically methanogen, biomass (Huang et al., 2015; Venturini et al., 2022). Fourthly, the availability of substrates before PP treatments was accumulated through microbial metabolism, soil aggregates shattering, and organisms death, which would be rapidly utilized by methanogen under these increased soil moisture conditions (Unger et al., 2010; Harms and Grimm, 2012; Kim et al., 2012; Lado-Monserrat et al., 2014). Though precipitation pulses break the balance between the CH₄ production and consumption and change the source-sink relationship of CH₄ fluxes through physical and biological processes ( Figure 4A ), they cannot permanently change the nature of grassland as a CH₄ sink ( Figure 5B ). In the context of global change, however, it appears that the sink strength of grassland ecosystems will decrease with the increase of heavy precipitation pulses in the future (Zhao et al., 2017; Ren et al., 2019).

For the first time it was demonstrated that cumulative CH₄ fluxes increased quadratically with precipitation pulse size in both the control and N addition treatments ( Figure 5D ). These results further suggest that precipitation pulses suppress CH₄ uptake by controlling soil moisture ( Figures 5B , 7B , S1 ). Average soil moisture in the 0-10 cm layer quadratically increased with precipitation pulse size after precipitation treatments (all P = 0.000, df = 24; Figure S1 ). Contrary to expectations, the optimal relationship between soil moisture and precipitation pulse size was not linear. Extreme precipitation pulses (20 mm and 50 mm) caused soil moisture to penetrate deeper and significantly increased average soil moisture in the 10-30 cm soil layer (P = 0.000, df = 4; Figure S2 ). The 10 mm, 20 mm, and 50 mm precipitation pulses significantly increased aboveground biomass ( Figure 3D ), which then also consumed a large amount of soil water. The increased aboveground biomass can also increase soil water by reducing soil evaporation, but the effect is largely unknown. Soil moisture infiltration and plant growth together decreased soil moisture in the 0-10 cm soil layer when larger precipitation pulse treatments were applied, resulting in a quadratic increase in average soil moisture with precipitation pulse size. Cumulative CH₄ fluxes increased linearly with average soil moisture due to the different responses of gas diffusion and activity of microbe to the increased soil moisture ( Figure 7 ). The small precipitation pulses did not alter soil water contents and O₂ concentrations enough to affect the CH₄ oxidation environment (Ni et al., 2019). The large precipitation pulses significantly inhibited the soil gases diffusion participating in CH₄ oxidation and suppressed CH₄ oxidation (Liu et al., 2008; Zhao et al., 2017). At the same time, the large precipitation pulses decreased the redox potential, created a saturated soil condition and lasted for a few days, which were conducive to methanogenesis (Harms and Grimm, 2012; Decock and Six, 2013; Petrakis et al., 2017). This study saw the release of CH₄ from the soil to the atmosphere under these conditions ( Figure 4A ), confirming that large precipitation pulses (10 mm, 20 mm, and 50 mm) stimulated the activity of methanogens (Ren et al., 2019). The decreased soil gases diffusion participating in CH₄ oxidation and increased activity of methanogens, therefore, led to a decrease in CH₄ uptake with increasing soil moisture. Ecologists have confirmed that CH₄ uptake decrease with increasing soil moisture in grassland ecosystems (Jiang et al., 2010; Shrestha et al., 2012; Liu et al., 2021a; Wu et al., 2021). In conclusion, precipitation pulses suppressed CH₄ uptake by increasing soil moisture and exhibited a quadratic relationship with the cumulative CH₄ fluxes.

Effects of N addition on CH₄ fluxes have been extensively studied in many ecosystems (Aronson and Helliker, 2010; Deng et al., 2020; Wu et al., 2022). As expected, N addition significantly suppressed the absorption of CH₄ in the studied grassland ecosystem ( Figures 4B , 5B ), consistent with the results of the meta-analyses (Aronson and Helliker, 2010). In the present study, N addition significantly increased the NH₄ ⁺-N content ( Figure 2B ) and NO₃ ^–N content ( Figure 2D ), along with aboveground biomass ( Figure 3D ), while significantly decreasing the pH value ( Figure 2E ). N addition, therefore, could decrease the absorption of CH₄ or increase the production of CH₄ through changing amounts of NH₄ ⁺-N, NO₃ ^–-N, and AGB, as well as altering pH value, suppressing CH₄ uptake (Fang et al., 2014; Yang et al., 2017; Chen et al., 2019; Kong et al., 2019; Ren et al., 2019).

There are multiple underlying mechanisms for N addition suppressing CH₄ uptake. First, the increased NH₄ ⁺-N competes with CH₄ for methane monooxygenase (MMO), which decreased the combination point of MMO to CH₄, thereby reducing the oxidation of CH₄ (Schnell and King, 1994). Second, NH₄ ⁺-N is oxidized to hydroxylamine (NH₂OH) and nitrite (NO2--N) by CH₄ monooxygenase or ammonia-oxidizing microorganisms, which has a toxic effect on methanotrophs (Bodelier, 2011). Third, the NH₄ ⁺-N and NO₃ ^–-N content in the N addition treatment were 1.96 and 2.77 times greater than that of the unfertilized treatment, respectively ( Figure 2D ), resulting in osmotic stress and suppressing the activity of methanotrophs (Bodelier and Laanbroek, 2004; Saari et al., 2004; Yang et al., 2017). Four, the increased AGB could allocate more C to promote root exudates, which would improve substrate availability for methanogens (Waldo et al., 2019). Additionally, N addition could enhance litter mass input to soil and nutrient return from litter decomposition by increasing AGB, thereby alleviating the C limitation on methanogens (Gong et al., 2015; Ren et al., 2019). As a result, methanogens enhanced the CH₄ production, which in turn offset the absorption of CH₄ and suppressed CH₄ uptake (Ren et al., 2019). It was found that cumulative CH₄ fluxes were marginally significant positively correlated with the changes in average NH₄ ⁺-N (P = 0.114, df = 49; Figure S3A ) and NO₃ ^–-N content (P= 0.077, df = 49; Figure S3B ), as well as AGB (P= 0.074, df = 49; Figure S3C ). This implies that N addition could decrease the absorption of CH₄ by increasing NH₄ ⁺-N content and NO₃ ^–-N content, and/or increase the production of CH₄ by increasing AGB, ultimately suppressing CH₄ uptake.

N addition significantly reduced soil pH value by 0.13 units ( Figure 2F ), half of the average global level for terrestrial ecosystems (Tian and Niu, 2015). A significant positive correlation between cumulative CH₄ fluxes and pH value was detected (P= 0.042, df=49; Figure S3D ), consistent with the findings of Ren et al. (2019). Contrary to expectations, the decreased pH enhanced CH₄ uptake in saline-alkaline soils, possibly due to reduced pH alleviating the physiological stress of saline-alkaline conditions on methanotrophs. In summary, N addition suppressed CH₄ uptake not by reducing pH value, but by increasing NH₄ ⁺-N content, NO₃ ^–-N content, and AGB.

Both the precipitation pulses and N addition significantly suppressed CH₄ uptake ( Figures 4B , 5B ). In contrast, the precipitation pulses and N addition together had no interactive effect on CH₄ uptake ( Figures 4B , 5B ). Several potential mechanisms could explain this result. First, cumulative CH₄ emissions after precipitation pulses were significantly affected by soil moisture (P < 0.001, df = 49), pH (P < 0.001, df = 49), and DOC (P < 0.05, df = 49; Figure S4 ). There were no significant interactive effects between precipitation pulses and N addition on soil moisture ( Figure 1B ) and DOC ( Figure 3A ). The precipitation pulses and N addition had an interactive effect on the pH ( Figure 2F ). The pH value, however, had opposite responses to precipitation pulses and N addition ( Figure 2F ), as precipitation pulses significantly increased the pH, whereas N addition significantly decreased it ( Figure 2F ). Precipitation pulses and N addition, therefore, could not interactively affect CH₄ uptake by interactively affecting soil moisture, pH, and DOC. Second, precipitation pulses suppressed CH₄ uptake by increasing soil moisture, whereas N addition suppressed CH₄ uptake by increasing NH₄ ⁺-N, NO₃ ^–-N, and AGB. Precipitation pulses and N addition inhibited the absorption of CH₄ through different pathways. Third, the structural equation model showed that pH value was a key factor in precipitation pulses, and N addition interactively affected CH₄ fluxes. Precipitation pulses significantly increased pH by increasing soil moisture, while N addition decreased pH through increasing NO₃ ^–-N content ( Figure 6 ). The precipitation pulses and N addition together, therefore, had opposite effects on pH. Additionally, N addition decreased cumulative CH₄ fluxes by decreasing pH (total correlation coefficient: 0.100), whereas the precipitation pulses increased cumulative CH₄ fluxes by increasing soil moisture ( Figure 6 ). Precipitation pulses and N addition together, therefore, had an offset effect on cumulative CH₄ fluxes, rather than a synergistic suppressing effect. In summary, there were no interactive effects of precipitation pulses and N addition on cumulative CH₄ fluxes.