Methane uptake responses to heavy rainfalls co-regulated by seasonal timing and plant composition in a semiarid grassland

Heavy rainfalls caused by global warming are increasing widespread in the future. As the second greenhouse gas, the biological processes of methane (CH4) uptake would be strongly affected by heavy rainfalls. However, how seasonal timing and plant composition affect CH4 uptake in response to heavy rainfalls is largely unknown. Here, we conducted a manipulative experiment to explore the effects of heavy rainfall imposed on middle and late growing season stage on CH4 uptake of constructed steppe communities including graminoid, shrub and their mixture in Inner Mongolia, China. The results of mixed effect model showed that both heavy rainfalls decreased CH4 uptake. Nevertheless, the effect magnitude and the pathways were varied with seasonal timing. Relatively, the late heavy rainfall had larger negative effects. Structural equation model suggested that late heavy rainfall decreased CH4 uptake through decreased diffusivity, pmoA abundance, and NH4+-N content, as products of high soil water content (SWC). However, middle heavy rainfall decreased CH4 uptake only by increasing SWC. Additionally, aboveground biomass (AGB) had negative effects on CH4 uptake under both heavy rainfalls. Additionally, plant composition not only affected CH4 uptake but also regulated CH4 uptake in response to heavy rainfalls. Late heavy rainfall had less negative effect on CH4 uptake in graminoid community than in other two communities, in coincidence with less reduction in NH4+-N content and less increase in SWC and AGB. In contrast, we did not observe obvious difference in effects of middle heavy rainfall on CH4 uptake across three communities. Our findings demonstrated that magnitude and pathways of heavy rainfall effects on CH4 uptake were strongly co-regulated by seasonal timing and plant composition.


Introduction
Methane (CH 4 ) is a powerful greenhouse gas and strongly contributes to global warming and resultant changes in precipitation, as the global warming potential is 28-36 times than that of carbon dioxide (CO 2 ) at 100-y timescale (Jiang et al., 2012;Fischer and Knutti, 2016;Otto et al., 2018;IPCC, 2021). Aerobic soils are important CH 4 sink, in which 9-47 Tg CH 4 year −1 from the atmosphere was oxidated by methanotroph through methane monooxygenase (MMO) (Elango et al., 1997;Fest et al., 2015;Yue et al., 2019Yue et al., , 2022. The subunit genes of MMOs, specifically pmoA, are used as biomarker genes for the presence and abundance of bacterial methanotrophs (Fest et al., 2015;Tentori and Richardson, 2020). Therefore, understanding effects of changes in precipitation on CH 4 uptake in drylands and underlying microbial mechanisms have great implications for prediction of future carbon cycling and its feedback to climate changes.
It has been confirmed that precipitation changes are expected to significantly influence the intensity of CH 4 sinks (Aronson et al., 2019;Martins et al., 2021). For example, increased precipitation by 30% significantly increased CH 4 uptake in temperate deserts (Yue et al., 2019). In contrast, CH 4 uptake was decreased and unchanged by increased precipitation in alpine meadows and in degraded steppe grasslands, respectively (Chen W. W. et al., 2013;Wu et al., 2020). Meta-analysis studies suggested that increased precipitation can decrease CH 4 uptake in terrestrial ecosystems at the global scale Yan et al., 2018). Although these results highlighted the important role of increased precipitation in regulating CH 4 uptake in aerobic soils, to date, there is great uncertainty about the effects of extreme precipitation with several days (e.g., heavy rainfall events), rather than chronic increases in precipitation at seasonal timescale, on CH 4 uptake.
Effects of chronic increases in precipitation and heavy rainfall on CH 4 uptake may be largely different. Soil moisture controlled CH 4 uptake through affecting the methanotroph community and altering air-soil diffusion (Wei et al., 2015). A bell-shaped relationship was observed between soil moisture and CH 4 uptake with CH 4 uptake reached the peak at intermediate soil moisture (Dijkstra et al., 2013;Li et al., 2016;Zhang et al., 2021). Above the optimum soil moisture, soil moisture would limit oxygen (O 2 ) diffusion in soils and depress the activity of methanotroph communities, inhibiting CH 4 uptake (Curry, 2007;Liptzin et al., 2011;Zhuang et al., 2013). Hence, slight increases in precipitation may promote methanotroph community and thereby increase CH 4 uptake while heavy rainfall caused saturation soil moisture and thereby would reduce CH 4 uptake. Additionally, CH 4 uptake is sensitive to soil ammonium (NH 4 + -N) and nitrate (NO 3 − -N). NH 4 + -N had an inhibiting effect on CH 4 uptake mainly through replacing CH 4 to be oxidized by methanotroph (Schnell and King, 1994;Yue et al., 2022), while NO 3 − -N had an inhibiting effect on CH 4 uptake through changing methanotroph activity and composition or enhancing soil oxidation potential and environment (Le Mer and Roger, 2001;Yue et al., 2022). Increased precipitation is likely to enhance soil inorganic nitrogen by accelerating mineralization (Cabrera and Kissel, 1988;Bai et al., 2012). In contrast, soil inorganic nitrogen may decline through leaching and runoff under heavy rainfalls (Borken and Matzner, 2009;Cregger et al., 2014). Thus, slight increases in precipitation and heavy rainfall are likely to induce opposite impacts on CH 4 uptake through the pathway of NH 4 + -N and NO 3 − -N content. Furthermore, seasonal timing and plant composition potentially modulate CH 4 uptake in response to heavy rainfalls. Previous studies suggested that seasonal timing strongly regulates effects of heavy rainfall on multiple ecosystem attributes such as soil water, carbon, and nitrogen availability, as well as plant biomass and phenology Post and Knapp, 2020;Li et al., 2022). Therefore, we except that impacts of heavy rainfall on CH 4 uptake may be also regulated by seasonal timing. Indeed, Zhao et al. (2017) found that CH 4 uptake was reduced by 62% and 45% during the period of middle and late heavy rainfall, respectively. Besides, there were significant differences in the composition and abundance of methanogens in soil with different plant species, resulting in different potential of CH 4 uptake (Dai et al., 2015). For example, CH 4 uptake capacity was stronger in soil of oat than that in native vegetation (Hüppi et al., 2022). Moreover, plant communities with higher-diversity were less negatively affected by floods and mature plants can withstand flooding better than seedlings (Gattringer et al., 2017;Wright et al., 2017). Although several studies had reported that plant community composition and seasonal timing could moderate heavy rainfall effects on CH 4 uptake capacity (Liebner et al., 2015;Tong et al., 2017;Zhao et al., 2017;Yue et al., 2022), it is unknown the interactions on CH 4 uptake in the face of heavy rainfall.
To explore the individual and especially interactive effects of heavy rainfall timing and species composition on CH 4 uptake in response to heavy rainfall, we conducted a field experiment in which heavy rainfall occurring in middle and late growing season were imposed on plots with three experimental plant communities of graminoids, shrubs and their combination, respectively. We hypothesized that: (1) Heavy rainfalls would suppress CH 4 uptake due to reduced diffusivity and methanotrophs activity, regardless of seasonal timing, (2) Heavy rainfall occurring in middle growing season with high air temperatures may cause less saturated soil conditions, thus CH 4 uptake is likely to be less decreased by middle heavy rainfall than late heavy rainfall, and (3) Plant community composition would adjust CH 4 uptake in response to heavy rainfalls though soil moisture, inorganic content, aboveground biomass and methanotrophs activities.

Study site
We carried out the study at the Research Station of Animal Ecology (44°18′ N, 116°45′ E 1079 m.a.s.l) in a semiarid grassland of Inner Mongolia Autonomous Region, China. The study site has a temperate continental semi-arid climate, of which the mean annual precipitation (1953 to 2017) is 281 mm and the mean annual temperature is 2.5°C. The plant species in the study region is mainly dominated by xeric rhizomatous grasses, needle grasses and perennial forbs such as Leymus chinensis, Stipa grandis and Medicago falcata. The soil type in this experimental region is classified as chestnut soil consisting of 60% sand, 18% clay and 17% silt.

Experiment design
The experiment was began in 2012. In this study, we reported the data measured in 2021. According to the statistical analysis of Frontiers in Ecology and Evolution 03 frontiersin.org ∼60-year (1953-2012) historical meteorological data provided by The Xilin Gol League Meteorological Administration, the longest continuous rainfall period of daily precipitation (≥ 3 mm) was 20 days during the growing season. The total effective precipitation was calculated over all 20 days periods, which was 250 mm. Thus, heavy rainfall was defined as 250 mm rainfall over 20 d in this study (12.5 mm d −1 ) . We used a two-way split-plot experiment design to study the effect of heavy rainfall on CH 4 uptake joint control of seasonal timing and plant composition. The main treatment had 9 plots and each main plot was made up of 3 sub-plots, thus there were total 27 sub-plots in heavy rainfall treatments experiment (Supplementary Figure S1). Specifically, three heavy rainfall treatments were set up in the main plots with three replicates: ambient control, mid-stage heavy rainfall (HR-mid, 15 July-5 August) and latestage heavy rainfall (HR-late, 15 August-5 September), respectively. Three plant community compositions were set up in the sub-plots: graminoid (Leymus chinensis and Stipa grandis), shrub (Caragana microphylla and Artemisia frigida) and their mixture. Plant seeds of dominant local species were cultivated at the start of the study in early May 2012. The total coverage of graminoids, shrub and mixture community were 70, 80, and 75%, respectively. Twenty-seven 2 m × 2 m sub-plots were established with 1-m intervals between sub-plots. The ambient control sub-plots remained uncovered year-round. Heavy rainfall treatment sub-plots were covered with 27 m 2 rainout shelters (4.5 m × 6 m, height 3 m) to prevent natural rainfall and greenhouse effect during the treatment periods. The rainout shelters were made of transparent polyester fiber material to ensure no significant shading. Non-target plant seedlings in each subplot were weekly removed to maintain fixed plant community composition during the entire growing season.

CH 4 flux and soil water content measurements
Soil water content (SWC) and CH 4 fluxes were measured three times monthly during the growing season in 2021. Time domain reflectometry (TDR 300 Soil Moisture Meter) with 20 cm probe was used to record SWC. CH 4 fluxes were measured by laser-based fast greenhouse gas analyzer with an in-house closed chamber. The data collection frequency of 1 Hz was utilized to measure CH 4 fluxes (Kang et al., 2018). The volume of cube chamber is 1.25 × 10 5 cm 3 , which was equipped with two electric fans in the center of the chamber ceiling to mix air concentration. Laser-based fast greenhouse gas analyzer has two 20-m rubber internal pipes, which were used to connect with the closed chamber through two 2-cm diameter holes on top of the chamber. Two pipes were used to transport gas from the greenhouse gas analyzer to the chamber and return from the chamber to the analyzer. Each subplot had a stainless frame (length × width × height = 50 cm × 50 cm × 10 cm) with 2 cm wide water groove. Each frame was installed and inserted 7 cm deep in soil and retained 3 cm above ground. Enough water should be put into the grooves of frames to guarantee gas tightness before mounting the chamber on the frame. Gas sampling area in each sub-plot was measured between 9:00 am and 10:00 am local time. In each sampling area, the gas in chamber was measured for 10 min, and the chamber should be opened for 2 min before the next measurement. We calculated CH 4 flux from the linear slope.
where F is the CH 4 flux rate [mg/(m 2 ·h)]; dc dt / is the cumulative growth rate of CH 4 ; M and P represent the molar mass of CH 4 (g/mol) and the air pressure (Pa), respectively; V 0 and P 0 represent the standard molar volume (22.41 m 3 /mol) and standard air pressure (101,325 Pa), respectively; T and T 0 represent the absolute temperature inside the chamber (oK) and absolute temperature (oK), respectively; and H is the effective height of the chamber (m).

Above-ground biomass measurement
Harvest method was used to measure above-ground biomass (AGB). We harvested all aboveground living plant tissues in a 50 cm × 50 cm quadrat of each sub-plot on September 21st, 2021. All plant tissues of each quadrat were put in the oven and dried at 65°C until they had constant weight.

Soil property measurement
Three soil cores were taken in 20 cm deep in each plot using an auger (2.5 cm in diameter) on September 21st, 2021. Roots and stones of soil were removed from three core sets and homogenized by 2 mm sieves. The extract was a mixture solution of 10 g fresh soil sample and 40 ml 0.5 M K 2 SO 4 solution, which were shaken for 30 min in shaker. After mixing well, NH 4 + -N and NO 3 − -N concentration of soil sample were detected by a continuous flow automatic ion analyzer (SEAL Analytical GmbH, Norderstedt, Germany) (Wachendorf et al., 2008).

DNA extraction and qPCR
DNA were extracted from fresh soil of 0.5 g by Power Soil DNA Isolation Kit (MOBIO Laboratories, United States) according to the specification information. DNA quality of soil was assessed by NanoDrop 2000 UV-Vis spectrophotometer (Thermo Scientific, Wilmington, Delaware, USA). The methanotrophic pmoA gene abundance was determined by quantitative polymerase chain reaction (qPCR) using Eppendorf Masterpiece realplex sequence detection system (Applied Biosystems 7500/7600). Standard curves were created with plasmid DNA in ten-fold serial dilutions. The primer sets were used for pmoA: 5′-GGNGACTGGGACTTCTGG-3′ and 5′-CCGGMGCAACGTCYTTACC-3′. The 20 μL qPCR reaction consisted of 1 μL DNA template, 0.2 μL of front and back primer, 10.4 μL mixture solution of ROX and Takara SYBR ® Premix Ex Taq ™ (Perfect RealTime) and 8.4 μL sterile water. After the reaction solution has been thoroughly mixed, the hole in the 96-well plate was filled with 20 μL qPCR reaction solution. Additionally, the contamination was detected by adding 19 μL qPCR reaction solution into the hole of 96-well plate without DNA template during the experiment. The sequential reaction conditions for the pmoA gene were set as: an initial denaturation at 95°C for 30 s, followed by 40 cycles at 95°C for 30 s, 60°C for 45 s, and 68°C for 45 s, with a final extension at 80°C for 30 s.
Frontiers in Ecology and Evolution 04 frontiersin.org

Statistical analyses
We conducted Duncan's multiple comparison to test differences of heavy rainfall with seasonal timing, plant composition and their interaction effects on variables including CH 4 uptake, soil water content, aboveground biomass, pmoA abundance, NH 4 + -N and NO 3 − -N content. Mixed-effects models were conducted using the NLME package in R v.3.4.4 (R Core Team, 2018) to compare the effects of middle and late growing season heavy rainfall, plant composition, and the interaction effects of plant composition and heavy rainfall on the above variables, respectively. Structural equation model (SEM) analyzes were performed using the piecewise SEM package to explore direct and indirect impacts of heavy rainfall on CH 4 uptake (Domeignoz-Horta et al., 2020). The most variation can explain by this model, including low Akaike Information Criterion (AIC), a nonsignificant Chi-squared test (p > 0.05) and high Comparative Fit Index (CFI > 0.9).

Result
Seasonal dynamics and response of soil moisture content to heavy rainfalls Total growing season precipitation (GSP) of control, HR-mid and HR-late were 351 mm, 513.16 mm, and 559.02 mm, respectively. Regardless of seasonal timing, SWC was significantly increased by heavy rainfalls in all three communities (p = 0.01 and <0.0001 for HR-mid and HR-late, respectively; Figure 1 and Table 1). Overall, SWC in HR-late seems slightly higher than that in HR-mid ( Figures 1E,F; Supplementary Figure S2). Plant community composition had no significant effects on SWC (p = 0.54 for composition, Table 1). However, the SWC in graminoid community was slightly less increased by heavy rainfalls compared with shrub and mixture communities (Supplementary Figure S2).

Response of CH 4 uptake to heavy rainfalls
Over the growing season, averaged CH 4 uptake significantly decreased by HR-mid (p = 0.03) and HR-late (p < 0.0001) in all three communities ( Figures 2E,F). The reductions were mainly occurred during the period of the heavy rainfalls. Relatively, HR-late had larger negatively effects on CH 4 uptake than HR-mid. There were significant differences of CH 4 uptake among three communities with the least CH 4 uptake in shrub community (p < 0.0001, Table 1; Supplementary Figure S3). HR-late effects on CH 4 uptake depended on plant composition (p = 0.03 for HR-late × Composition), with the least decreased CH 4 uptake in graminoid community than that in other two communities. There was also a marginally significant interaction between HR-mid × Composition (p = 0.08) on CH 4 uptake, but the effects of HR-mid on CH 4 uptake were similar across three communities. Collectively, negative effects of heavy rainfalls on CH 4 uptake modulated by seasonal timing and plant community composition.
Response of pmoA abundance, AGB, NH 4 + -N and NO 3 − -N to heavy rainfalls Similarly, heavy rainfalls effects on soil inorganic N content and pmoA abundance were changed with seasonal timing and plant community composition. Both two heavy rainfalls significantly declined pmoA abundance for three communities but the effects were larger in mixture community than in other two communities (p = 0.01 and 0.0001 for HR-mid × Composition and HR-late × Composition) ( Figure 3A; Table 1). Regardless of plant composition, HR-mid and HR-late unchanged and significantly increased AGB, respectively ( Figure 3B; Table 1), NH 4 + -N was significantly increased by HR-mid but significantly declined by HR-late in three communities ( Figure 3C). Similarly, HR-mid significantly increased NO 3 − -N content ( Figure 3D; Table 1), mainly in graminoid and mixture communities. In contrast, HR-late had little effects on NO 3 − -N content.

The influence of abiotic and biotic factors on CH 4 uptake
Structural equation model showed that SWC had directly negative impacts on CH 4 uptake and pmoA abundance under two heavy rainfalls. Additionally, CH 4 uptake negatively correlated with AGB under two heavy rainfalls. However, CH 4 uptake positively correlated with pmoA abundance in HR-late but not in HR-mid. SWC had significantly positive impact on NH 4 + under HR-mid, while opposite impact between SWC and NH 4 + was found in HR-late. Moreover, NH 4 + positively correlated with pmoA abundance only in HR-late. NO 3 − had no significant relationships with CH 4 uptake ( Figure 4). In short, heavy rainfalls with different seasonal timing decreased CH 4 uptake through different pathways.

Discussion
Understanding extreme precipitation scenario on CH 4 uptake has important implications for predicting future global climate changes and terrestrial C cycling. To explore how seasonal timing and plant composition affected CH 4 uptake in response to heavy rainfalls, we conducted a manipulative experiment in a semiarid grassland of Inner Mongolia, China. In this study, we identified CH 4 uptake in response to heavy rainfall are regulated by independent and especially interactive effects of heavy rainfall timing and plant composition. Our results demonstrate that seasonal timing strongly controls size and pathway of negative effects of heavy rainfall on CH 4 uptake and importantly the regulating effects of plant composition on CH 4 uptake response to heavy rainfall via soil water content, pmoA abundance, NH 4 + -N content and AGB.
Heavy rainfalls decrease CH 4 uptake CH 4 uptake was significantly decreased by heavy rainfalls, regardless of seasonal timing and plant community composition in our study (Figure 2). Previous study showed that soil moisture and CH 4 uptake had a hump-shaped relationship, where the optimum Frontiers in Ecology and Evolution 05 frontiersin.org moisture is 10 % -12 % for highest CH 4 uptake in a semiarid and arid soils (Dijkstra et al., 2011;Li et al., 2016;Yue et al., 2022). Moderate soil moisture could significantly promote CH 4 uptake, which could be significantly inhibited by too-low or too-high soil moisture (Van den Pol-van Dasselaar et al., 1998;Dijkstra et al., 2011). In our study, SWC were above 12% in all treatments throughout the growing season. High soil moisture induced by heavy rainfalls would cause anaerobic soil conditions, low soil oxygen (O 2 ) concentrations and CH 4 diffusion ( Figure 1). Additionally, pmoA abundance decreased in both two heavy rainfalls ( Figure 3A). Taken together, these results  suggested that experimental heavy rainfalls continuously decreased CH 4 diffusivity and O 2 availability and thus inhibited the activity of methanotrophs, supporting our first hypothesis. As a result, SWC showed negatively relationship with CH 4 uptake under two heavy rainfalls in our study (Figure 4).

Magnitude and pathways of heavy rainfall effects on CH 4 uptake depend on seasonal timing
Although two heavy rainfalls had negative effects on CH 4 uptake, the effect magnitude varied with seasonal timing. Consistent with the second hypothesis, CH 4 uptake is less decreased by HR-mid than HR-late in all three communities (Figure 2). This may be because HR-mid received less precipitation than HR-late (513.16 mm vs. 559.02 mm, Figure 1). In addition, higher air temperatures during the period of HR-mid would induce larger evapotranspiration. As a result, HR-mid caused less saturated soil conditions than HR-mid, which was reflected by slightly lower SWC in HR-mid than in HR-late ( Figure 1F). As discussed above, SWC had negative impacts on CH 4 uptake in our study. Thus, lower SWC and corresponding less saturated soil conditions under HR-mid induced less reduction in CH 4 uptake.
Structural equation model showed that SWC and resultant anaerobic conditions were main controller of CH 4 uptake (Wei et al., 2015;Zhou et al., 2021). AGB also had direct negative effects on CH 4 uptake under both heavy rainfalls ( Figure 4B). Previous studies showed similar trends that increased AGB may contribute to increasing soil water-holding capacity, maintaining high soil moisture and inhibiting soil substrate availability. As a result, methanotrophs activities and CH 4 oxidation in soil were inhibited (Robson et al., 2007;Zhang et al., 2012;Tang et al., 2018). Besides, high SWC directly and indirectly inhibited pmoA abundance through decreasing NH 4 + -N content, ultimately, suppressing CH 4 uptake in HR-late. CH 4 was oxidated by methanotroph, thus, it is not surprising that low pmoA Seasonal dynamic and mean of CH 4 uptake under heavy rainfall treatments in graminoid (A,D), shrub (B,E), and graminoid +shrub (C,F) plots. Different letters above bars in d, e and f indicate significant difference among treatments at p ≤ 0.05. The orange and blue shaded regions in a-c indicate the periods of the HR-mid (heavy rain imposed in middle of the growing season, orange line) and HR-late (heavy rainfall imposed late in the growing season, blue line) treatments, respectively. Error bars show one standard error of the mean.
Frontiers in Ecology and Evolution 07 frontiersin.org abundance would limit CH 4 uptake (Degelmann et al., 2010;Zhang et al., 2019;Kaupper et al., 2021;). Previous studies have found that the process of methanotrophs using CH 4 as both an energy and carbon source generally requires soil NH 4 + -N as N source (Rigler and Zechmeister-Boltenstern, 1999;Schimel and Weintraub, 2003;Bürgmann, 2011), resulting in decreased soil NH 4 + -N content can inhibit methanotrophs activities and pmoA abundance (Le Mer and Roger, 2001;Xu and Inubushi, 2007;Yue et al., 2016Yue et al., , 2022. However, some findings of other studies suggest that decreased NH 4 + -N availability in soils can promote CH 4 oxidation as higher NH 4 + -N can replace CH 4 to be oxidized by methanotroph (Song et al., 2020;Yue et al., 2022). Therefore, the net effects of NH 4 + -N on CH 4 uptake depend on the relative size of the two processes. Nevertheless, the mechanism was not suitable for HR-mid. Similar to HR-late, HR-mid declined pmoA abundance, however, it had no significant correlation with CH 4 uptake. This may be because increased NH 4 + -N content under HR-mid leaded to the oxidation of NH 4 + -N instead of CH 4 by methanotrophs, thus resulting in the most decreased CH 4 uptake in mixture community, although the largest reduction of pmoA abundance were found in graminoid community. As NO 3 − -N content was little impacted by heavy rainfalls, it had no effects on CH 4 uptake in this study. Our results are consistent with the finding that soil NO 3 − -N concentrations and CH 4 uptake had no correlation in a subtropical plantation forest ecosystem (Wang et al., 2014). Taken together, our study proved that the pathways underlying CH 4 uptake in response to heavy rainfall depend on seasonal timing.

Plant composition regulates responses of CH 4 uptake to heavy rainfalls
Multiple lines of evidence proved that plant composition is a controlling factor in regulation of soil CH 4 oxidation. CH 4 uptake would increase with enhanced plant diversity as high plant biodiversity promoted microbial activities (Altor and Mitsch, 2006;Bouchard et al., 2007;Schultz et al., 2011;Hassan et al., 2019). Niklaus et al. (2016) showed that the presence of legume plants inhibited soil CH 4 oxidation capacity due to decline in plant N acquisition. Likewise, CH 4 A B

FIGURE 3
Responses of pmoA abundance (A), above-ground biomass (AGB) (B), NH 4 + -N (C) and NO 3 − -N (D) to heavy rainfall treatments in three plant communities. Different letters above bars indicate significant difference among treatments at p ≤ 0.05.
Frontiers in Ecology and Evolution 08 frontiersin.org uptake had significant differences among three communities, where CH 4 uptake was less in shrub community than in graminoid and mixture communities in our study (Supplementary Figure S3; Table 1). It may be because shrubs had harmful effects on methanotrophs activities and CH 4 uptake through the chemistry of root exudates and N competition among plants and microbes (Zak et al., 2003;Hassan et al., 2019). Importantly, plant composition regulated CH 4 uptake to heavy rainfalls, reflected by significant interactions between plant composition and heavy rainfalls (Table 1). Negative effects of HR-late on CH 4 uptake was the least in graminoid community than in other two communities. The potential explanation may be that HR-late had larger positive effects on SWC and AGB and larger negative effects on pmoA abundance and NH 4 + -N content in shrub and mixture communities ( Figures 1F, 3A), although the interactions were only significant on SWC and pmoA abundance. This finding supports the third hypothesis that plant composition would regulate CH 4 uptake in response to heavy rainfalls though soil moisture, inorganic content, aboveground biomass and methanotrophs activities. Although previous studies proved that plant communities with higher-diversity are less negatively affected by floods (Gattringer et al., 2017;Wright et al., 2017), the least increase in SWC, AGB and the least decrease NH 4 + -N content were observed in graminoid community under HR-late. Previous studies have reported that precipitation infiltration and evaporation rate vary with plant species composition. Evaporation and transpiration can remove water from shallow soil layers after rainfall and thus decrease soil moisture (Coughenour, 1984;Weltzin et al., 2003;Springer et al., 2006;MacIvor and Lundholm, 2011;Moore et al., 2022). Graminoids are shallower rooting with deeper and faster infiltration and faster evaporation rate, leading to lower soil moisture and the duration of soil saturation in graminoid community (Springer et al., 2006). Therefore, HR-late had the least negative effect on CH 4 uptake in graminoid community under HR-late. Overall, SWC, pmoA abundance and NH 4 + -N content had the least reduction in graminoid, leading to the least decreased CH 4 uptake in HR-late. However, HR-mid had similar negative effects on CH 4 uptake across three communities although the interaction between HR-mid and plant composition on CH 4 uptake was statistically significant (p = 0.08). Therefore, we concluded that CH 4 uptake in response to climate extremes jointly controlled by interaction of seasonal timing and plant composition.

Conclusion
Our results highlight the vital role of seasonal timing and plant composition in regulating heavy rainfall effects on CH 4 uptake. Specifically, although both heavy rainfalls reduced CH 4 sink, late heavy rainfall had larger negative effects than middle heavy rainfall. This is because decreased NH 4 + -N induced by late heavy rainfall had negative effects on pomA abundance and further suppressing CH 4 sink, in addition to directly negative effects of high soil moisture induced by heavy rainfall. Besides, shrub community had lower CH 4 uptake than graminoid and mixture communities. Moreover, late heavy rainfall had the least negative effects on CH 4 uptake in graminoid communities than in other two communities, indicating that climate extremes-driven shifts in dominant species would in turn alter ecosystem feedbacks. Therefore, to improve prediction accuracy of terrestrial ecosystems feedbacks to climate changes, we encourage future studies to further quantify the interactive effects between seasonal timing and plant on regulating carbon cycling in response to climate extremes.

FIGURE 4
Structural equation models analysis of the direct and indirect effects of soil, microbe and plant variables on CH 4 uptakes under HR-mid (A) and HR-late (B) treatment. SWC: soil water content; AGB: aboveground, biomass; NH 4 + and NO 3 − : soil ammonium and nitrate content. Solid and dashed lines indicate significant (p ≤ 0.05) and nonsignificant (p > 0.05) relationships, respectively. Width of the line is proportional to the strength of path coefficients expressed by the numbers adjacent on lines. r 2 values denote the proportion of variance explained for each variable.
Frontiers in Ecology and Evolution 09 frontiersin.org

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.