Factors Related with CH4 and N2O Emissions from a Paddy Field: Clues for Management implications

Paddy fields are major sources of global atmospheric greenhouse gases, including methane (CH4) and nitrous oxide (N2O). The different phases previous to emission (production, transport, diffusion, dissolution in pore water and ebullition) despite well-established have rarely been measured in field conditions. We examined them and their relationships with temperature, soil traits and plant biomass in a paddy field in Fujian, southeastern China. CH4 emission was positively correlated with CH4 production, plant-mediated transport, ebullition, diffusion, and concentration of dissolved CH4 in porewater and negatively correlated with sulfate concentration, suggesting the potential use of sulfate fertilizers to mitigate CH4 release. Air temperature and humidity, plant stem biomass, and concentrations of soil sulfate, available N, and DOC together accounted for 92% of the variance in CH4 emission, and Eh, pH, and the concentrations of available N and Fe3+, leaf biomass, and air temperature 95% of the N2O emission. Given the positive correlations between CH4 emission and DOC content and plant biomass, reduce the addition of a carbon substrate such as straw and the development of smaller but higher yielding rice genotypes could be viable options for reducing the release of greenhouse gases from paddy fields to the atmosphere.


Introduction
Climate change is a major environmental problem of the 21 st century caused mainly by increasing emissions of anthropogenic greenhouse gases (GHGs). Agriculture contributes about 20% of the present atmospheric GHG concentration [1]. Methane (CH 4 ) and nitrous oxide (N 2 O) are the two most important GHGs from agriculture, with global-warming potentials (GWP) of 28 and 265 CO 2 -equivalents, respectively, on a 100-year time horizon [2]. The atmospheric concentrations of CH 4 and N 2 O have increased rapidly from preindustrial levels of 722 and 270 ppb to present levels of 1830 and 324 ppb, respectively [2]. N 2 O is also the dominant gas that is catalytically a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 soil, and (3) investigate the relationships between soil physiochemical properties and CH 4 and N 2 O metabolism (production, oxidation, transport, and final emission).

Study area
The field experiments were carried out at the Wufeng Agronomy Field of the Fujian Academy of Agricultural Sciences (26.1˚N, 119.3˚E) in southeastern China during the early rice-growing season (April -July) in 2011 (Fig 1). The soil of the wetland paddy field was poorly drained, and the proportions of sand, silt, and clay particles in the top 15 cm were 28, 60, and 12%, respectively. Other soil properties (0-15 cm) at the onset of the experiment were: bulk density of 1.1 Mg m -3 , pH (1:5 with H 2 O) of 6.5, organic-carbon content of 18.1 g kg -1 , total nitrogen (N) content of 1.2 g kg -1 , total phosphorus (P) content of 1.1 g kg -1 , total potassium (K) content of 7.9 g kg -1 , and available-S of 1.26 mg kg -1 After rice transplant, the paddy field was flooded by 5-7 cm of water above the soil surface throughout the growing period (32 days) by using an automatic water-level controller. At the final tiller, water was drained and crop was non-flooded during about one week. Thereafter there was a period with alternating wet and drying treatments and when the rice was ripe, a dry period of about one week was stablished. After this, crop was harvested. The field was plowed to a depth of 15 cm with a moldboard plow and leveled two days before rice transplantation. Mineral fertilizers were applied in three splits as complete (NH 4 -P 2 O 5 -K 2 O, 16-16-16%; Keda Fertilizer Co., Ltd., Jingzhou, China) and urea (46% N) fertilizers. The ratios between N in complete NPK fertilizer and N from urea were 21:1, 7:3 and 9:4 in the first, second and third fertilization times, respectively. Fertilization management in the rice crop followed the typical practices of southern China. A basal fertilizer was applied one day before transplantation at a rate of 42 kg N ha -1 , 40 kg P 2 O 5 ha -1 , and 40 kg K 2 O ha -1 . Twenty-one-dayold seedlings (three seedlings per hill) of rice (cv. Hesheng 10, China) were transplanted manually at a spacing of 14 × 28 cm on 6 April to three replicate plots (10 × 10 m) at the study site. The second split of fertilizer was broadcasted during the tiller initiation stage (seven days after transplanting (DAT)) at a rate of 35 kg N ha -1 , 20 kg P 2 O 5 ha -1 , and 20 kg K 2 O ha -1 . The third split was broadcasted during the panicle initiation stage (56 DAT) at a rate of 18 kg N ha -1 , 10 kg P 2 O 5 ha -1 , and 10 kg K 2 O ha -1 . As is habitual in this area we used butachlor at 1.5 kg ha -1 two days before rice transplantation to remove small grasses. No specific permissions were required for these locations/activities, the location only have plant rice that no need protect, and our study experiment were also safety. Moreover, the field studies did not involve endangered or protected species.

Measurement of CH 4 and N 2 O emissions from the paddy field
Static closed chambers were used to measure CH 4 and N 2 O emissions during the growing period as described by Datta et al. [23]. The chambers were made of PVC and consisted of two parts: an upper transparent compartment (100 cm height, 30 cm width, 30 cm length) was placed on a permanently installed bottom collar (10 cm height, 30 cm width, 30 cm length). Three replicate chambers were used. Each of these chambers was placed in each plot. Fluxes were measured in triplicate (three days) each week in each plot. Gas samples were collected twice daily. We used the average of the resulting 3x3x2 measurements. Each chamber was installed with two battery-operated fans to homogenize the air inside the chamber headspace, a thermometer to monitor temperature changes during the gas-sampling period, and a gassampling port with a neoprene rubber septum at the top of the chamber for collecting gas samples from the headspace. Each chamber covered two rice hills. A wooden boardwalk was built for accessing the plots in the study area to minimize soil disturbance during gas sampling.
Gas fluxes were measured in triplicate weekly at all chamber locations to study the seasonal variation. A 100-ml plastic syringe equipped with a 3-way stopcock was used to collect gas samples from the chamber headspace 0, 15, and 30 min after chamber deployment. Gas samples were collected twice a day. The collected gas samples were immediately transferred to 100-ml air-evacuated aluminum foil bags (Delin Gas Packaging Co., Ltd., Dalian, China) sealed with a butyl rubber septum and transported to the laboratory for analysis of CH 4 and N 2 O. Additional headspace gas samples were collected hourly from 9:00 to 6:00 the following day to study the diurnal variation of CH 4 and N 2 O emissions during the tillering (36 DAT) and maturity (85 DAT) stages of the rice crop.
Determination of CH 4 and N 2 O concentrations in the headspace air samples CH 4 and N 2 O concentrations in the headspace air samples were determined by a gas chromatograph (Shimadzu GC-2014, Kyoto, Japan) packed with a Porapak Q column (2 m length, 4 mm OD, 80/100 mesh, stainless steel column). A flame ionization detector (FID) and an electron capture detector (ECD) were used for the determination of CH 4 and N 2 O concentrations, respectively. Helium (99.999% purity) was used as a carrier gas (30 ml min -1 ), and a make-up gas (95% argon and 5% CH 4 ) was used for the ECD. Calibration was conducted with 1.01, 7.99, and 50.5 μl CH 4 l -1 in He and 0.2, 0.6, and 1.0 μl N 2 O l -1 in He (CRM/RM Information Center of China) as primary standards.

Measurement (in situ) of rates of CH 4 production and oxidation
The rates of CH 4 production and oxidation were measured once every two weeks using acetylene (C 2 H 2 ) inhibition, which has been successfully used in both laboratory incubations and field-based studies [26][27][28][29]. After gas sampling for determining overall CH 4 emission, C 2 H 2 was added to the chambers at a headspace concentration of 4% to inhibit CH 4 oxidation. The chambers were incubated overnight to allow the translocation of the C 2 H 2 through the plants into the rhizosphere, which would inhibit nearly all microbial CH 4 oxidation [30,31]. The chambers were removed from the plants for 5 min the next morning to re-establish the ambient atmospheric conditions. The chambers were then redeployed, and the headspace gas was sampled for determining the rate of CH 4 production as described above for measuring total emission. The rate of CH 4 oxidation was estimated by subtracting the total CH 4 emission rate from the CH 4 production rate.

Measurement (in situ) of CH 4 transport in the paddy field
The CH 4 transport pathways were measured once every two weeks following the method of Wang and Shangguan [32]. The rate of plant-mediated transport was determined by covering the entire surface inside the collar with plastic sheeting and then taping the sheeting to the base of the rice plants to block CH 4 ebullition and diffusional transport. The rates of total plant-mediated and diffusional transport were determined by covering the surface inside the collar with 0.15-mm gauze and then taping the gauze to the base of the rice plants to block CH 4 ebullition. The rate of diffusional CH 4 transport was then determined by subtracting the plant-mediated transport rate from the total plant-mediated and diffusional-transport rates. The rate of CH 4 ebullition was calculated by subtracting the plant-mediated and diffusional CH 4 transport rates from the total CH 4 emission rate. To measure the diurnal variation of GHG emissions we chose the 36 DAT as representing the tiller period and the 85 DAT representing the ripening stage.

Sampling of porewater and soil samples
Porewater was sampled in situ once every two weeks from April to July 2011. Three specially designed stainless steel tubes (2.0 cm inner diameter) were installed to a depth of 30 cm in each plot. Porewater samples were collected immediately after the measurements of CH 4 emission using 50-ml syringes, injected into pre-evacuated vials (20 ml), and stored in a cooling box in the field, and another part was injected into the 100 ml sample bottle. After transporting to the laboratory, the samples in the vials were stored at -20˚C until the analysis. Three soil porewater samples were randomly collected from the 0-30 cm layer of each plot. The soil samples were collected with a soil sampler (length 0.3 m and diameter 0.1 m) taking a core from the first 15 cm of soil profile.

Measurement (in situ) of porewater and soil properties
Before analysis, the vials were first thawed at room temperature and were then vigorously shaken for 5 min to equilibrate the CH 4 concentrations between the porewater and the headspace. The gas samples were taken from the headspace of the vials and analyzed for CH 4 concentration with the above gas chromatograph [30]. The porewater concentrations of sulfate and dissolved organic carbon (DOC) were determined using a sequence flow analyzer (San ++ , SKALAR Corporation production, Breda, The Netherlands) and a TOC Analyzer (TOC-V CPH, Shimadzu Corporation, Kyoto, Japan), respectively. Fresh soil (0-15 cm) was digested with 1M HCl to determine the total Fe concentration in the soil, and soil Fe 2+ and Fe 3+ concentrations were determined using the 1,10-phenanthroline and spectrometric method (UV-2450, Shimadzu Corporation, Kyoto, Japan) [33]. The concentration of soil available N (0-15 cm) was determined by alkaline hydrolysis diffusion [33]. The growth characteristics of rice (e.g. leaf, stem, grain, above-and belowground, and total biomasses) were determined by harvesting three plants (one per plot) on each measurement day, and the biomasses were determined by oven-drying the samples.

Calculation of CH 4 and N 2 O fluxes and porewater dissolved CH 4 concentration
The rates of CH 4 and N 2 O flux from the paddy field were expressed as the increase/decrease in CH 4 and N 2 O mass per unit surface area per unit time. CH 4 and N 2 O fluxes were calculated by: where F is the CH 4 or N 2 O flux (mg CH 4 m -2 h -1 or μg N 2 O m -2 h -1 ), M is the molar mass of the respective gas (16 for CH 4 and 44 for N 2 O), V is the molar volume of air at a standard state (22.4 l mol -1 ), dc/dt is the change in headspace CH 4 and N 2 O concentration with time (μmol mol h -1 ), H is the height of the chamber above the water surface (m), and T is the air temperature inside the chamber (˚C). The concentration of CH 4 dissolved in the porewater was calculated by Ding et al. [34]: where Ch is the CH 4 concentration (μl l -1 ) in the air sample from the vials, Vh is the volume of air in the bottle (ml), and Vp is the volume of the porewater in the bottle (ml).

Measurement of environmental and biotic parameters
Ambient air temperature (˚C) and air humidity (%) and soil electrical conductivity (mS cm -1 ), Eh (mV), pH, and temperature in the 0-15 cm layer were measured in situ each sampling day. Each final measurement was the average of three consecutive measurements. Air temperature and air humidity were measured using a Kestrel 3500 pocket weather meter (N K Scientific Instruments, Carlsbad, USA). Soil redox potential (Eh), pH, and temperature were measured with an Eh/pH/temperature meter (IQ Scientific Instruments, Carlsbad, USA), and soil electrical conductivity was measured using a 2265FS EC Meter (Spectrum Technologies Inc., Paxinos, USA).

Statistical analysis
The data were checked for normality and homogeneity of variance, and if necessary, were logtransformed. Pearson correlation analysis was also used to determine the relationships of porewater CH 4 concentration, CH 4 production, CH 4 oxidation, CH 4 transport, CH 4 emissions, and N 2 O emissions among them and with soil properties, rice growth, and meteorological variables. Stepwise regression analysis was used to determine the relationships of CH 4 and N 2 O dynamics with soil properties, rice growth, and meteorological variables. All statistical analyses were performed using SPSS Statistics 17.0 (SPSS Inc., Chicago, USA).

CH 4 emissions from the paddy field
The pattern/intensity of CH 4 emissions from the paddy field varied with the stage of rice growth (Fig 2). The emission rate was relatively low (0.04-0.55 mg m -2 h -1 ) during the initial stage (1-22 DAT) and increased as the crop matured. The rate peaked on 71 DAT (7.99 mg m -2 h -1 ) and then decreased following drainage and ripening of the rice crop (0.28-0.75 mg m -2 h -1 ). The seasonal average CH 4 emission rate was 3.53 ± 3.37 mg m -2 h -1 . Diurnal variation of CH 4 emissions from the paddy field differed significantly between the tillering (36 DAT) and maturity (85 DAT) stages (Fig 3). On 36 DAT, emission was maximum at 15:00 (9.36 mg m -2 h -1 ) and minimum at 9:00 (7.06 mg m -2 h -1 ), with an average of 8.98 ± 3.20 mg m -2 h -1 . CH 4 emissions on 85 DAT decreased continuously throughout the day from a maximum (1.07 mg m -2 h -1 ) at 9:00 to a minimum (0.05 mg m -2 h -1 ) at 6:00 the next day, with an average of 0.53 ± 0.37 mg m -2 h -1 .

N 2 O emissions from the paddy field
The pattern/intensity of N 2 O emissions from the wetland paddy field also varied with the stage of rice growth (Fig 2). The emission rate was relatively high (-5.21 to 359 μg m -2 h -1 ) during the initial stage (1-22 DAT), but decreased rapidly following the development of anaerobic conditions in the soil. Emission peaked on 8 (359 μg m -2 h -1 ) and 22 (217 μg m -2 h -1 ) DAT. The emission rate increased slightly after drainage at the final growth stage (-1.00 to 57.4 μg m -2 h -1 ). The seasonal average N 2 O emission rate was 36.1 ± 114 μg m -2 h -1 .
Changes in soil, porewater, weather, and plant parameters during the experimental period Ambient air temperature increased steadily over the period of rice growth from 18.9 to 33.5˚C, with an average of 27.2 ± 4.7˚C. Air humidity varied considerably between 30.6 and 95.9% during the same period, with an average of 74.5 ± 16.8% (Fig 4).

Relationships of CH 4 production, oxidation, and transport with emissions
The rate of CH 4 production was low early in the season (Fig 8) but had become significantly higher by 71 DAT. The rate of CH 4 oxidation was low throughout the growth period except during 71-78 DAT. The rate of plant-mediated transport of CH 4 was low during the initial growth and ripening stages but then rose to a peak on 71 DAT (7.02 mg m -2 h -1 ). The rate of CH 4 ebullition was also low during the initial growth and ripening stages but had increased by 43 DAT (1.36 mg m -2 h -1 ). The rate of CH 4 diffusional transport was low throughout the growth period. Compared to other sampling days, porewater CH 4 concentrations were significantly higher to a depth of 30 cm on 71 DAT (69.52 μmol l -1 ) and significantly lower on 15 DAT (1.40 μmol l -1 ), with an average of 17.4 μmol l -1 . CH 4 emission was positively correlated with CH 4 production (R = 0.994), plant-mediated transport (R = 0.997), ebullition (R = 0.940), CH 4 diffusion (R = 0.530), and dissolved CH 4 concentration in the porewater (R = 0.620) ( Table 1). About 94% of the CH 4 produced (calculated as the percentage of production to total emission) from the anaerobic soil environment was released into the atmosphere, and less than 6% was oxidized (Fig 8). Plant-mediated transport was the most important pathway, contributing about 98% of the CH 4 emission, and ebullition and diffusion together contributed only 2% of the CH 4 emission.    We used stepwise regression analysis to identify the most important variable(s) controlling CH 4 metabolism. CH 4 emission was largely governed by sulfate, available N, and DOC concentrations, stem biomass, air temperature, and relative humidity, which together explained Factors Related with CH 4 and N 2 O Emissions from a Paddy Field 92% of the variance (Table 3). Sulfate and available N concentrations, stem and leaf biomasses, air temperature, and relative humidity together explained 96% of the variance in CH 4 production. In contrast, about 88% of the variance in CH 4 oxidation could be accounted for by changes in aboveground and stem biomasses, air temperature, relative humidity, and soil Eh and available N concentration. Plant-mediated transport of CH 4 was controlled by above-and belowground biomasses, air temperature, and soil Fe 3+ concentration and pH, which together explained 91% of the variance. Stem biomass, air and soil temperatures, relative humidity, and soil available N concentration accounted for 69% of the variation in CH 4 ebullition. Soil salinity and Eh, belowground and leaf biomasses, and air humidity together explained 73% of the variance in CH 4 diffusional transport. Over 82% of the variance of the dissolved CH 4 concentration was explained by a combination of stem biomass, relative humidity, and soil temperature, salinity, and available N concentration.  The stepwise regression analysis indicated that 95% of the variation in N 2 O emission could be explained by changes in air temperature, leaf biomass, soil Eh, pH, and available N and Fe 3+ concentrations (Table 3).

Crop yield and GHG per Mg of crop yield
The crop yield in this paddy field was 8.1 Mg ha -1 , so CH 4 and N 2 O emissions per Mg of crop (grain) yield were 9.62 and 0.1 kg, respectively.

Temporal patterns of CH 4 and N 2 O emissions
The diurnal variation of CH 4 emissions from the paddy field differed significantly between the tillering (36 DAT) and ripening (85 DAT) stages of rice growth (Fig 2). Diurnal emission was maximum during the tillering stage at 15:00, which might be attributed to a higher soil temperature in the afternoon that enhanced microbial CH 4 production. Luo et al. [14] reported that soil temperature had a significant effect on the diurnal variation of CH 4  Factors Related with CH 4 and N 2 O Emissions from a Paddy Field spruce forest in Germany, a tropical rain forest in Australia, and an ungrazed semi-arid steppe in China. The diurnal variation of CH 4 emission may also be partly explained by the photosynthetic activity of the rice plants. Up to 52% of the CH 4 emissions from paddy soils comes the exudation of labile organic carbon from roots to the rhizosphere for methanogenesis i.e. comes from photosynthesis. The other 48% is emitted from old soil carbon [35]. The photosynthetic activity of rice plants during the day probably causes a rapid translocation of photosynthates to belowground tissues, hence promoting CH 4 production and emissions from soils [35,36]. Lai et al.[36] reported that the duration of CH 4 production was much shorter than the lag in CH 4 flux of 9-12 hours after maximum photosynthetic activity in a sedge-dominated community in a northern peatland. Microbial respiration in soils may also increase CO 2 concentrations in soil water due to the absence of photosynthesis at night, because the CO 2 cannot be used for photosynthesis. Higher CO 2 concentrations could reduce the pH of soil solutions, which in turn could lower CH 4 production and emissions from soils at night. The diurnal CH 4 emission during ripening was maximum at 9:00. Young rice plants contributed a substantial proportion of their photosynthates to soils compared to mature plants [37]. The diurnal variability of CH 4 flux during ripening was likely not strongly limited by the supply of labile carbon substrates, but by other environmental factors. The CH 4 peak in the early morning could be attributed to winddriven ventilation immediately after sunrise that promoted the mass transport of CH 4 produced and stored in soil pores during the relatively calm night. CH 4 emission, production, and plant-mediated transport had similar seasonal patterns, with higher rates on 71 DAT (young panicle differentiation stage) and lower rates during the initial growth and ripening stages (Fig 8). Such seasonal patterns were typical for CH 4 metabolism in paddy fields, especially for the overall CH 4  The diurnal variation of N 2 O emissions from the paddy field differed significantly between the tillering (36 DAT) and ripening (85 DAT) stages (Fig 2). Emissions during the tillering stage were maximum and minimum at 9:00 and 18:00, respectively, similar to emissions in a natural wetland [41]. Emission during ripening was maximum at 9:00. A combination of low nighttime temperature and N 2 O diffusion rate from the paddy soil to the atmosphere likely led to the accumulation of N 2 O in the soil. The accumulated N 2 O would be released to the atmosphere the following morning in a burst due to turbulence-driven mass flow. Higher daytime N 2 O emission has also been reported for mangroves during the early growth stage [20,42], which could be due to changes in N chemistry, O 2 availability, or soil temperature [43,44].

Main CH 4 metabolic process controlling CH 4 emission
Frenzel and Karofeld [10] reported that 80-99% of the CH 4 produced in paddy soil was oxidized in the rhizosphere, and Jia et al. [45] reported that an average of 36.3-54.7% of the CH 4 produced was oxidized in rice paddy soils. The fraction of the CH 4 oxidized in our study was much smaller (9.6%), perhaps due to a smaller methanotrophic community, less efficient transport of oxygen to the rhizosphere, or both. Further studies of microbial diversity and anaerobic CH 4 oxidation are required to elucidate the causes of the limited role of methanotrophs in the paddy soil.
Plant-mediated transport was the most important pathway of CH 4 transport (about 98% of total CH 4 emission). Ebullition and diffusion contributed only 2% to the overall CH 4 emission. Frenzel and Karofeld [10] reported that ebullition accounted for only <1% of the total CH 4 emitted to the atmosphere from a raised bog, which was similar in magnitude to our findings. Boose and Frenzel[46] found a strong influence of ebullition on CH 4 emissions from rice microcosms during the early vegetative and late senescence phases, but such variations among growth stages were not seen in the present study. The high CH 4 emissions at the young panicle differentiation stage (71 DAT) might be attributed to the rapid development of plant aerenchyma, which is supported by the higher rate of plant-mediated CH 4 transport on 71 DAT (Fig 6). Jia et al. [45], however, reported a higher CH 4 emission from paddy soils during the tillering stage compared to the panicle initiation stage, which could be explained by less oxidation of the produced CH 4 . CH 4 oxidation at our study site, though, did not have a significant influence on overall CH 4 emissions. The relationships among CH 4 emissions production and transport, but not with CH 4 oxidation, suggest that CH 4 emissions are mainly related with CH 4 production and CH 4 transport and less or not with CH4 oxidation.

Influence of environmental factors on CH 4 and N 2 O emission: clues for mitigation
Both CH 4 production and emission in this subtropical paddy field were predominantly controlled by the soil concentrations of sulfate and DOC and the biomass of the rice plants. Sulfate concentration was the most important factor controlling CH 4 emission due to the inhibitory effects of sulfate on the activities of soil methanogens during rice cultivation [47]. Our study demonstrated significant and negative correlations between soil sulfate concentration and CH 4 production and emission, in accordance with the results from other studies simulating the effects of sulfate deposition on paddy fields [47][48][49]. Sulfate reduction is thermodynamically more favorable than CH 4 production in the anaerobic degradation of soil organic matter [50], so an increase in the availability of sulfate ions would suppress the activity of methanogens and CH 4 production during rice cultivation [51]. Acetate and molecular hydrogen (H 2 ) are important methanogenic substrates but may also be consumed by oxidation with electron acceptors such as sulfate [50]. An increase in soil sulfate concentration in an Italian paddy field reduced the H 2 partial pressure in the soil below the threshold concentration for methanogens, thus inhibiting H 2 -dependent methanogenesis [51]. The drainage of paddy fields in Texas, USA, reduced acetate concentrations and concomitantly increased the soil sulfate concentration and decreased CH 4 production, suggesting that sulfate reducers were able to outcompete methanogens for the available acetate, a labile carbon substrate [51]. But we must be careful with the direct link between sulfate concentration and CH 4 release because of the max value of CH 4 emission occurred during flood period, and at the as time, sulfate reduction is also very high due to the redox state in anaerobic environment. Therefore, this relationship might be driven by the water condition in the paddy field, and does not mean a causal relationship.
Sulfate concentration in our study, however, was significantly and positively correlated with soil N 2 O emissions and was governed by mechanisms similar to the effects of Fe 3+ on  [57] reported that intermittent irrigation of rice paddies significantly decreased CH 4 emission but stimulated N 2 O emission. Kudo et al. [56] also reported that mid-season drainage successfully mitigated CH 4 emissions from paddy fields but led to a sharp increase in N 2 O emissions. The results from these studies suggest a trade-off between mitigating CH 4 and N 2 O fluxes from paddy fields. N 2 O is a more potent GHG than CH 4 , especially on a long-term basis, so the overall global-warming potential of the various management strategies should be carefully considered [58].
Our study showed that CH 4 emission from a subtropical paddy field was negatively correlated with N 2 O emission, which may mainly have been due to the differential responses of various microbial groups (e.g. methanogens and nitrifying and denitrifying bacteria) to critical environmental factors [59]. For example, sufficiently low Eh is required for CH 4 formation, because methanogenic bacteria metabolize organic substances under strictly anaerobic conditions [60]. Soil denitrification, however, could become dominant under such conditions, and most of the intermediate products (i.e. NO and N 2 O) could be reduced to the final product N 2 , especially in N-limited areas such as our study site [61] where N 2 O emission can be negative (Fig 2). Van Rijn [62], however, reported that denitrification in soils with a high Eh could become a dominant microbial process that promotes N 2 O production but inhibits methanogenic activity.
DOC concentration is an indicator of the availability of substrates for CH 4 production that in turn influences CH 4 emission [63]. A large amount of DOC could be derived from the exudation of organic acids from roots, and about 61-83% of the carbon in these exudates could serve as methanogenic substrates and eventually be converted into CH 4 [36]. The concentration of soil organic carbon was lower at our study site (18.1 g kg −1 ) than in other paddy fields [64,65]. CH 4 production in our paddy field was thus probably limited by the availability of organic substrates, which was supported by the positive response of CH 4 production to carbon addition reported for wetland soils in the same study area [66]. The observed increasing concentration of methane dissolved in pore water after 43 DAT (Fig 8) coincided with the reflooded period after the drainage period when much carbon from roots and plants has been deposited in the soil, and therefore was available in pore water.
The rice plants were an important factor controlling CH 4 emissions from the soil. The development of an internal gas-space ventilation system (aerenchyma) in some vascular plants provides improved aeration for submerged organs in anoxic soils below the water table [67]. These aerenchymatous tissues can also serve as gas conduits for the transport of CH 4 from the rhizosphere to the atmosphere, while bypassing the aerobic, methane-oxidizing layers [68]. Molecular diffusion and bulk flow are two other mechanisms involved in the transport of CH 4 from soil to the atmosphere [67]. The respiratory uptake of oxygen by plants creates a diffusion gradient that draws O 2 from the atmosphere to the rhizosphere [17]. This gradient is accompanied by an upward diffusion of CH 4 from the rhizosphere to the atmosphere via the aerenchyma down the concentration gradient. The bulk transport of CH 4 by plants, though, is driven by pressure differences. Differences in temperature or water-vapor pressure between the internal air spaces in plants and the surrounding atmosphere generate a pressure gradient that drives gases to vent from leaves to the rhizosphere in bulk and then vent back to the atmosphere through old leaves or rhizomes connected to other shoots. During this convective flow, CH 4 produced in the rhizosphere can also be rapidly flushed to the atmosphere [67]. This plant-mediated pathway enables an efficient transport of CH 4 with minimal resistance [17].
The N 2 O emissions from the paddy field varied with the stage of rice growth. Emissions were higher during initial growth compared to other stages due to the increased availability of substrates for N 2 O production subsequent to the N fertilization, in agreement with previous studies [69]. Our stepwise regression analysis also indicated that the fluctuation of soil Eh between the alternating wet and dry periods was an important factor influencing N 2 O emission, consistent with the findings of similar studies [70]. Furthermore, the soil Fe 3+ concentration was significantly and positively correlated with N 2 O emission ( Table 2). Previous studies have also found that variations in N 2 O emissions from subtropical paddy fields were dependent on Fe 3+ concentrations during rice growth, which may be due to two mechanisms [71,72]. Firstly, higher Fe 3+ concentrations could increase the biological oxidation of ammonium that is known to generate N 2 O by nitrification [73]. Secondly, higher concentrations of Fe 3+ in the soil could also lead to an increase in Fe 2+ concentrations, which in turn would promote the reduction of nitrite to N 2 O [54, 73].
The results of our study also indicated a significant, positive correlation between soil Eh and N 2 O emission. A decrease in Eh in a subtropical swamp was accompanied by a reduction in soil N 2 O concentrations that could directly reduce the emissions of N 2 O to the atmosphere [74]. Previous studies have also reported higher emissions of N 2 O during the day than at night as a result of higher dissolved O 2 concentrations, which indirectly supported the findings of a positive correlation between soil Eh and N 2 O emission [43,44]. Based on our results and those from previous studies, the positive relationship between N 2 O emission and Eh in our paddy field was likely because nitrification was the major mechanism governing N 2 O production, as supported by the very low concentrations of dissolved NO 3 measured in the porewater samples that were below the detection limit of our ion chromatograph (<0.01 mg l -1 , ICS2100, Dionex Corporation, Sunnyvale, USA). This study provides results indicating that both models of CH 4 emissions and management strategies to reduce CH 4 emissions should take into account the trade-off between N 2 O and CH 4 emissions. For instance, some previous studies claim that the use of sulphate N-fertilization reduced CH 4 emissions [11], however, as commented this could increase N 2 O emissions. The limited number of available observations of CH 4 N 2 O emissions in relation to environmental variables under field conditions have constrained the parameterization and validation of process-based biogeochemistry models [15,16]; this study can thus contribute to improve models of CH 4 and N 2 O emissions. Moreover, the results of this study indicate the interest of management strategies that can reduce methane emission by decreasing the production or transportation or increasing the oxidation. Furthermore, application of electron acceptors, such as Fe 2+ , SO 4 2-, or NO 3 could also be considered.

Conclusions
Methanogenesis and plant-related CH 4 transport are the two main processes governing the overall CH 4 emission from paddy fields. The reduction of CH 4 production in soil is thus critical for any attempt to mitigate CH 4 emissions from paddy fields. The sulfate concentrations were negatively correlated with CH 4 emissions, so the amendment with sulfate fertilizers may be a viable option to reduce CH 4 emission from the soil. Acid and sulfate deposition (by rainwater) are increasing in this part of China, so CH 4 production and emissions would likely be suppressed to some extent due to an increase in sulfate availability. Our results, however, suggest that the use of sulfate fertilizer may increase N 2 O emissions from paddy fields. The stepwise regression analysis showed that plant-related parameters, such as stem and leaf biomasses, were the most important factors controlling CH 4 and N 2 O emissions with significant, positive correlations. Rice cultivars with low biomasses should thus be selected for reducing the plant-mediated transport of CH 4 and N 2 O through the aerenchymatous tissue. CH 4 emission was positively correlated with DOC concentration, so our results suggest that the addition of carbon substrates such as straw could be an option for mitigating the GHG emissions from paddy fields.
Supporting Information S1 Appendix. Field data used in this study.