Effects of Drainage on Greenhouse Gas Emissions and Yields of Lowland Rice-Wheat Rotation System in East China

The subtropical region of East China is characterized by abundant water and temperature resources conducive to crop cultivation, and large areas of lowland have been widely used for agricultural planting. To explore the feasible methods of greenhouse gases (GHGs) reduction for rice-wheat rotation system, shallow ditch (SD) and deep ditch (DD) treatments in the wheat season were set up for drainage to control the water content in soil, with the conventional non-ditching treatment as the control group (CG). Results showed that methane (CH 4 ) emissions from paddy soil were in the majority in global warming potential (GWP) in rice-wheat rotation system. In the three years, compared with that of CG, the CH 4 cumulative emissions of SD and DD were reduced by 65.80% and 63.42% (rice season), and 101.37% and 77.28% (wheat season), respectively; the nitrous oxide (N 2 O) cumulative emissions of SD and DD were reduced by 27.62% and 11.30% (rice season), and 1.53% and -37.40% (wheat season), respectively; the total GWP produced by SD and DD in the three years was reduced by 58.78% and 52.22%, respectively; GHG emission intensity (GHGI) of SD and DD declined by 60.67% and 53.85%, respectively; the CH 4 emission ux was signicantly positively correlated with atmospheric temperature and 5 cm ground temperature, but negatively correlated with soil Eh; when the soil Eh value was lower than -150 mV, the CH 4 emission ux increased signicantly, indicating that -150 mV was the key soil Eh value for CH 4 emissions in this area; in addition, both SD and DD led to markedly decrease in soil organic matter content and an increase in soil pH. The ndings indicate that SD and DD not only ensure stably increasing production, but also effectively reduce GHG emissions.

agricultural planting, whose distribution is related to rivers and lakesides (Nillesen and Kok 2015). In the context of increasing demand for grain production, it is of great global signi cance to mitigate and adapt to climate change for researching and promoting the reduction of GHGs in rice-wheat rotation system. CH 4 is an organic gas produced by methanogens decomposing organic matter in an extremely anaerobic environment. For the paddy elds, soil water control is a highly important factor affecting CH 4 and N 2 O emissions, and the in uences of water managements such as draining and alternate wetting and drying (AWD) on the CH 4 and N 2 O emissions in farmlands have been well documented (Liang et al. 2017;Liang et al. 2016;Schimel et al. 2011). Related research has shown that the higher soil water content in winter will make the soil remain an extremely anaerobic state, thus promoting the emissions of CH 4 (Wu et al. 2018;Xu et al. 2003;Zhao et al. 2020).
According to Kang et al. (2002) and Jain et al. (2014), water control in non-rice growing period is effective in controlling CH 4 emissions in China's rice growing period. Related scholars reported that drainage helps reduce CH 4 emissions by destroying the extreme anaerobic environment of methanogens, while ooded paddy soils are global sources of CH 4 due to the lack of oxygen (David et al. 2018;Liang et al. 2016). N 2 O emissions from continuously ooded paddy elds are very low, and they are often ignored (Johnson-Beebout et al. 2009).
However, water managements on farmlands usually produce large amount of N 2 O emissions. Linquist et al. (2012) showed that the drainage in the wheat season leads to an increase in N 2 O emissions from rice elds and a signi cant decrease in CH 4 emissions. Liu et al. (2019) showed through meta-analysis that draining in paddy elds is an effective way to mitigate GHG emissions in rice ecosystem, the global warming potential (GWP) of CH 4 and N 2 O under mid-season drainage decreases signi cantly compared with that under continuous ooding.
Furthermore, Hou et al. (2016) reported that water management has a signi cant impact on GHG emissions from rice-wheat rotation system, and long-term ooded rice elds have lower N 2 O emissions. Another report pointed out that if a water control method is used in the ooded farmland in southwest China in winter, the total CH 4 emissions during the growth period of the next rice season will be effectively reduced by approximately 63-72% (Cai et al. 1998). Water control measures can inhibit CH 4 emissions, but it may increase N 2 O emissions at the same time (Cai et al. 2000). However, it remains unclear whether the corresponding increase in N 2 O emissions can partially offset or overcompensate the positive impact on CH 4 mitigation, leading to an increase or decrease in GWP (Khalid et al. 2019). Moreover, most studies on the impact of water management on N 2 O and CH 4 emissions are restricted to the rice season. Relatively few studies have been devoted to the effects of water control during the wheat season on the GHG emissions of the entire annual rice-wheat rotation (Hou et al. 2016).
In order to explore a feasible plan for reducing GWP under the rice-wheat rotation pattern, a three-year experiment was conducted, with the rice-wheat two-cropping farmland of Chao Lake polder area as the research object. Two water control treatments easy to be implemented and a control treatment were designed. In addition, the in uencing factors for CH 4 emissions from paddy soils were analyzed, such as atmospheric temperature, 5 cm ground temperature and soil Eh. The results are applicable to the emission reduction and e ciency enhancement technology of subtropical rice-wheat rotation farmland cultivation in China.

Experiment site
Page 4/26 The experiment was conducted at the Chao Lake Agricultural Environment Experimental Station of Anhui Agricultural University from 2013 to 2015. The monitoring site was located in Tangzui Village, Qi Town, Chaohu City (117°41'6"E and 31°39'50"N with an elevation of 17 m), which belongs to the north subtropical monsoon climate zone. The annual average precipitation here is 1358.3 mm, the average temperature is 16.8°C, the annual frost-free period is 247 days, and the sunshine hours are 2106 h. According to the statistics in 2014, the grain crop planting area was around 49,800 ha in Chaohu City, with a total output of approximately 319,000 tons of crop, and rice production of about 256,000 tons.
This eld was managed according to the traditional practices in the area, wherein the rice was ooded during the growing season, and the uncultivated elds were left after harvest until the next cultivation. The rice and winter wheat varieties used for the experiment were Huiliangyou 996 and Yangmai 16.
The soil type of this monitoring site in the Chao Lake lowland area is submerged paddy soil with a pH value (H 2 O) of 6.19, organic carbon content of 23.71 g kg −1 , total nitrogen content of 1.29 g kg −1 , and physical clay content of 488 g kg −1 . The physical and chemical properties under different treatments from 0 cm to 20 cm soil are also shown in Table 1.  Fig. 1). Each treatment was performed with 3 replicates. The designed depth of the ditch was determined based on actual situations. A large ditch outside the farmland was approximately 60-80 cm deep. The interval between the eld drains was 3 m while ditches were interconnected.
The bund used to separate the experimental plots was made of cement. Each plot had water inlets and drains, and the drains were connected one another. Local routine management was adopted for the eld management of each treated plot. The wheat season land was plowed before planting, and its depth was around 15 cm. Rice would be irrigated 4-5 times during the growth period and irrigated for 1-2 days before the three fertilizations, and the depth was approximately 6-7 cm at the end of the roasting eld. Regular pest and disease control was performed during the growing period to ensure the normal growth state of the rice and wheat.
N fertilizer, P fertilizer, and K fertilizer were urea, superphosphate, and potassium chloride, respectively. The speci c fertilization scheme and fertilization amount are shown in Table 2. The managements of wheat and rice elds in the three years are shown in Tables 3 and 4.  of 10 minutes. 60 mL of gas was sampled each time. The data were recorded, and the Eh value of the soil was monitored.
The samples were taken in the wheat season once a week, and every 3 to 5 days in the rice season, which could be adjusted rationally according to weather conditions. In the periods of fertilization, topdressing, and roasting, the samples were taken every two days. The collected gas samples were measured for CH 4 and N 2 O concentrations by gas chromatography (Brooker 450-GC) within 24 hours. And FID detector was used to detect CH 4 . The detection conditions were as follows: column temperature: 50°C; detector temperature: 250°C; nitrogen ow, 10 mL/min; hydrogen: air ow 30 mL/min, 300 mL/min. N 2 O was detected with the Nl63ECD Detector under the following conditions: detector temperature of 300°C and nitrogen ow rate of 300 mL/min. CH 4 and N 2 O emission ux can be calculated by the following formula:

Calculation of yield, GWP and GHGI
To analyze the actual effect of the different water control methods on crop yield in the wheat season, production measurement was conducted after crop maturation and the average economic output of the three groups of repeated plots was taken.
GWP was calculated as CO 2 equivalents (CO 2 -eq) over a 100-year time horizon using the radiative forcing potential of 298 for N 2 O and 25 for CH 4 relative to CO 2 (IPCC, 2013): In the formula: GHGI is the GHG emission intensity (t/hm 2 , calculated as CO 2 ); Y represents the yield of rice and wheat (kg hm -2 ).

Soil properties
To clarify the mechanism of CH 4 production in rice elds, soil properties were analyzed during a complete wheat season. The HW type soil temperature automatic recorder was used to measure the 5 cm ground temperature. The soil Eh value was determined by in-situ measurement with an Eh meter. The model of the intelligent portable redox potential meter is QX6530, provided by the Nanjing Institute of Soil Science, Chinese Academy of Sciences. We used soil drills in the community to randomly sample fresh soil samples of 0-20 cm from 3 locations The Eh electrode was inserted 3-5 cm below the soil surface. After each removal, we quickly inserted the electrode into the soil sample to a depth of about 2 cm, the Eh value was recorded after the reading became stable.
In each study area, a soil drill was used to take eld soil samples from 5 points 0-20 cm soil layer using the "Z" method. The soil pH was measured by the potentiometric method (water-soil ratio: 2.5:1), and soil organic matter Page 7/26 was determined by the potassium dichromate volumetric method with oil bath heating.

CH 4 emissions
CH 4 emissions from rice-wheat rotation system were mainly in the rice season, while drainage during the wheat season effectively reduced the CH 4 emissions in the middle and late stages in the rice season ( Fig. 2 and Table 5).
CH 4 emission ux from rice elds of CG showed many peaks after applying fertilization (Table 4), while the CH 4 emission ux of SD and DD generally had single peaks during the rice season.
In the rice season, compared with that of CG, the cumulative CH 4 emissions of SD and DD were reduced by 56. 36% and 53.88% in 2013, 74.95% and 72.03% in 2014, and 73.19% and 70.97% in 2015, respectively. The results showed that the CH 4 emissions of paddy elds of SD and DD were signi cantly less than those of CG (P<0.05, Table 5), while there was no signi cant difference between SD and DD (P>0.05). In addition, CH 4 emissions from rice seasons in the last two years declined signi cantly, and the cumulative CH 4 emissions in 2014 and 2015 were reduced by 56.24% and 50.64% (average value for each treatment), respectively, compared with 2013. Relatively speaking, the CH 4 emission reduction effect of SD in the rice season was slightly better than that of DD. Although the CH 4 emissions in the wheat season were low, they still existed, while drainage also signi cantly reduced CH 4 emissions during the wheat season. The total CH 4 emissions of SD and DD were signi cantly reduced by 139.66% and 59.48% in 2013, 93.11% and 94.09% in 2014, and 82.59% and 74.34% in 2015, respectively (P<0.05). Note: Lowercase letters indicate signi cant differences between treatments (P<0.05); ± means the standard errors (n=3) of the replicates. The same as below.

N 2 O emissions
The seasonal variation of N 2 O ux of SD and DD were approximately the same as that of CG (Fig. 3). In the wheat season, SD and DD treatments led to no signi cant emission reduction compared with CG. In some periods, the N 2 O ux of SD and DD was signi cantly higher than that of CG. In contrast, during the rice season, both SD and DD effectively reduced N 2 O emissions, and the cumulative N 2 O emissions of SD and DD were reduced by 35.71% and 53.57% in 2013, 20.51% and 6.41% in 2014, and 43.64% and 3.64% in 2015, respectively, with the decrease of SD reaching a signi cant level in the three years (P<0.05).
Over the entire rice-wheat rotation system, the cumulative N 2 O emissions in the three years were DD > CG > SD (Table 6). Compared with CG, the cumulative N 2 O emissions of SD declined by 8.50%, while DD increased by 24.38%. Although water management in the wheat season effectively reduced the total CH 4 emissions in the rice season, the N 2 O emissions in the wheat season were not lower than those under the conventional treatment, and the cumulative N 2 O emissions even exceeded those under the conventional treatment in some stages.

GWP, yield, and GHGI
The cumulative CH 4 emissions during the rice season was the main contributor to GWP of rice-wheat rotation system, and the GWP produced by CH 4 in the rice season accounted for 86.01%, 71.35%, and 65.85%, respectively, among the total processed by CG, SD and DD (Tables 7 and 8). The GWP calculated with the CH 4 and N 2 O emissions after the SD and DD treatments in 2013 was about half that after the CG treatment, with more than 98% caused by CH 4 . In 2014, the GWP generated by CH 4 and N 2 O after the SD and DD treatments were about onethird that after the CG treatment, and the GWP of all the treatments was reduced to different extents compared with that in 2013. Compared with the other two years, the GWP generated by CH 4 in the rice season in 2014 accounted for a relatively small proportion of the total GWP, especially in the SD and DD treatments wherein the percentage was approximately 75%, and SD treatment achieved the best emission reduction effect.
The drainage experiment improved the economic yields of both rice and wheat elds. The yields increased by 3.28% (SD) and 2.88% (DD) in the rice season, and increased by 7.64% (SD) and 4.71% (DD) in the wheat season, respectively. In addition, the GHGI can re ect the comprehensive impact of different treatments on crop yields and GHG emissions. In the rice season, the GHGI of SD and DD declined by 65.46% and 62.49%, respectively (three- year total) (P<0.05). Moreover, the GHGI of SD declined by 18.75% and GHGI of DD rose by 25.00% in the wheat season (P<0.05).  Table 8 Average yield (kg hm −2 ) and yield-scaled GWP (kg CO 2 -Eq kg −1 ) in the rice-wheat seasons in each treatment In the early stage of rice growth (0-28 days before rice eld sunning), the trend of the 5 cm ground temperature and CH 4 ux in the paddy soil was approximately the same (Fig. 4). The soil drying period in the eld was between 29th and 34th days after transplanting. During this period, the paddy eld surface water was evaporated, wherein the soil permeability and the oxygen content in the soil void increased, and then the soil environment changed from anaerobic to aerobic state, inhibiting the activity of methanogens, and nally reducing the CH 4 emissions. Accordingly, no signi cant correlation existed between the CH 4 emissions and 5 cm ground temperature. Generally, during the whole growth period of rice, the CH 4 emissions of paddy soil were signi cantly correlated with the 5 cm ground temperature.
As shown in Fig. 5, the correlation between the change in CH 4 emission ux and the 5 cm ground temperature was signi cant (P<0.01). As the 5 cm ground temperature increased, CH 4 emissions also increased gradually.
According to the comparison of the R 2 value between the 5 cm ground temperature and the CH 4 ux in the watercontrolled farmland, the tting degree of each treatment was higher (0.7 or higher), indicating that the 5 cm ground temperature plays an important role in the change in CH 4 ux. Figure 5. Relationship between the CH 4 emission ux and 5 cm ground temperature in each treatment.

Soil properties
The corresponding Eh values were mostly between -150 and -300 mV when the CH 4 emission ux was higher; when the soil Eh value was lower than -150 mV, the CH 4 emission ux increased signi cantly with the decrease of Eh; whereas the corresponding CH 4 emission ux was close to zero when the Eh value was higher than -100 mV ( Fig. 6-a, b, c).
In the wheat season, the changes in soil organic matter (28.20-26.59 g/Kg) and soil pH (5.79-6.12) of CG were relatively more stable. In contrast, the soil organic carbon content of both SD and DD showed declining trends, among which the soil organic matter of SD dropped from 29.6 g/Kg to 14.75 g/Kg while that of DD dropped from 28.62 g/Kg to 16.68 g/Kg. The lowland soil in Chao Lake area was an acidic soil type, and it was observed that the drainage would raise the soil pH: The soil pH of SD increased from 5.29 to 6.67 while that of DD increased from 5.15 to 6.41 ( Fig. 7-a, b).

CH 4 emissions
In this study, the annul emissions of CH 4 were lower than those in the study from the subtropical permanently The cumulative CH 4 emissions in the rst year were notably too high in this study (Table 5), which could be attributed to the fact that the drainage released the CH 4 originally stored in the soil and improved the aerobic state of the soil, not only inhibiting the production of CH 4 , but also increasing the activity of methane oxidizing bacteria.
The existence of a large number of anaerobic zones in the soil structure during the initial drainage stage is suitable for the production of CH 4 (Yuan et al. 2016). Furthermore, through the comparison of the CH 4 emissions in the three years, it was found that multiple peaks often existed in the early and late stages of rice growth (Fig. 3), and they all appeared after applying fertilization (Tables 2 and 3 and 4), suggesting that the application of fertilizer provided an excessive carbon source to the soil in a short time, thus promoting the emissions of CH 4 , consistent with the nding of previous research (Fan et al. 2016;Masuda et al. 2018). Over the entire rice-wheat rotation system, it can be seen that the peaks of N 2 O ux in the rice season roughly coincided with the low value of CH 4 ux, all in the period of rice eld sunning. CH 4 tends to be generated under anaerobic conditions (Cai et al. 1997). During the period of rice eld sunning, the CH 4 emissions are very low due to the soil aeration enhanced by rice eld sunning, which destroy the extreme anaerobic environment of methanogens (Nayak et al. 2015).
Therefore, rice eld sunning can be regarded as an effective measure for CH 4 emission reduction in paddy elds.
Temperature mainly controls CH 4 emissions and absorption by affecting the activity against methanogens and methane oxidizing bacteria (Green et al. 2018). The soil Eh is a characterization of the soil moisture status of the soil oxygen availability. Gas exchange is reduced between the atmosphere and the soil when the soil is ooded (Nobrega et al. 2016). Accordingly, the soil is in an extremely anaerobic environment, wherein the soil Eh decreases rapidly and the production activity of CH 4 bacteria is enhanced. Under ooding conditions, polysaccharides in the paddy elds produce a large amount of acetic acid, H 2 , and CO 2 during the decomposition process, which offer abundant basic substances to the production of CH 4 bacteria, thereby signi cantly increasing the CH 4 emission ux of the rice elds (Nobrega et al. 2016). The corresponding Eh values were mostly between -150 and -300 mV when the CH 4 emission ux was high in this study, indicating that -150 mV may be the critical value affecting the CH 4 emission ux in lowland paddy soils. However, due to the complicated eld conditions, the relationship between CH 4 emissions and soil Eh has not been de nitely clari ed. Szafranek-Nakonieczna and Stępniewska (2015) reported that when the soil Eh value is less than 240 mV, the CH 4 production and emission will be concentrated, inconsistent with the results of this study. The discrepancy among these results may be due to differences in soil texture. These simple compounds prone to mineralization are the carbon source and energy source for CH 4 production.
CH 4 in rice elds is mainly produced by the decomposition of soil organic matter in those soils by methanogens in the soil under anaerobic conditions (Takakai et al. 2019;Yuan et al. 2018). Both SD and DD reduce soil organic matter content, and low organic matter content may inhibit the activities of methanogens. Related research has revealed a signi cant positive correlation between CH 4 emission ux and soil organic matter content (Pramanik et al. 2014). The results of this study indicated that the reduction of organic matter caused by drainage may be the dominant reason for CH 4 emissions. The possible reason is that after soil ooding, anaerobic decomposition of organic matter is dominated, and microorganisms deprive the organic matter of oxygen in the soil to form various reducing substances, resulting in a rapid decline of soil Eh. The faster the Eh value declines, the more the reducing substances will be produced (Verdi et al. 2018).
Moreover, soil pH signi cantly in uences the activities and growth of methanogens and methane oxidizing bacteria that control CH 4 production and oxidation, respectively. Some studies have demonstrated that methanogens have the ability to adapt to acidic environments (Lin et al. 2017;Wu et al. 2018). In this study, drainage in the wheat season led to an increase in soil pH, and lower CH 4 emissions were observed in the rice season, consistent with the ndings by Jeffery et al. (2016) that the increase in soil pH is conducive to the survival of methane oxidizing bacteria, thus reducing the CH 4 emissions. However, another study reported that an increase in soil pH will promote the activity of methanogens (Liu et al. 2011). There are also reports that the production of CH 4 is affected by a combination of various soil properties (Shen et al. 2021). Therefore, effects of other soil properties should be considered when the effects of organic matters on CH 4 emissions are assessed.

N 2 O emissions
Soil moisture regimes are important for the activities of nitrifying and denitrifying bacterial enzymes, major drivers of N 2 O emissions (Cai et al. 1997;Lan et al. 2013). Notably, since ooding conditions will stimulate the denitri cation process of soil nitrate nitrogen, the N 2 O emissions of each treatment increase to varying degrees.
Drainage in the wheat season has a certain inhibitory effect on CH 4 emissions, but the content of N 2 O is increased compared with that under the conventional treatment, because water management affects the content of O 2 in the soil, weakening anaerobic condition of soil gradually and promoting nitri cation .
When the soil is in an alternating aerobic and anaerobic condition, its moderate O 2 concentration is conducive to the production of N 2 O in the nitri cation and denitri cation processes, which leads to massive N 2 O production and emissions (Song et al. 2020;Vilarrasa-Nogue et al. 2020;Wang et al. 2017). Therefore, the dry and wet alternation of the soil caused by drainage satis es both the above conditions and greatly promotes the emission of N 2 O, and wheat eld soils have the largest amount of emissions (Liao et al. 2020;Xu et al. 2013). In the wheat season, the peak of N 2 O emissions of each treatment emerged after fertilization. Abalos et al. (2016) and Wang et al. (2021) reported that applying fertilization will produce a large amount of nitrogen to the soil, which provide a reaction substrate for nitrifying and denitrifying bacteria. Affected by the typhoon bringing frequent rainfall in 2015, the soil was in aerobic and anaerobic alternation, which promoted the process of nitri cation and denitri cation, so the N 2 O emissions greatly increased and uctuated (Gao et al. 2018;Hou et al. 2016), directly leading to a signi cant increase in N 2 O emissions of rice-wheat rotation system in 2015 (Table 7). In order to further clarify the mechanism of N 2 O emissions, it is necessary to study the impact of rainfall pattern on N 2 O emissions.
Dryland soil has great porosity, making the oxygen easily get in, so the extreme anaerobic environment required for the survival of methanogens is blocked, while the activity of methanotrophic bacteria is preserved (Hofmann et al. 2016). Liu et al. (2015) reported that winter-wheat elds are often regarded as the sink of CH 4 . In this experiment, CH 4 emissions in the wheat season were very low but should not be ignored (Table 5). It is worth noting that the CH 4 emissions of rice-wheat rotation farmland in this study were too high, while the N 2 O emissions were considerably low. Wen et al. (2017) reported that the substantial reduction of N 2 O to N 2 under anaerobic condition may lead to lower N 2 O emissions. The lowland polder areas along the middle and lower reaches of the Yangtze River are mostly derived from the development of water, and the groundwater level is strongly affected by the nearby main waters ). It has been reported that groundwater level uctuation is a key predictor for CH 4 and N 2 O emissions from cultivated lowland (Wen et al. 2020). Evans et al.
(2021) observed that the high groundwater level in lowland areas usually leads to CH 4 increase and N 2 O decrease, which is a scienti c issue that cannot be ignored in global climate change. Therefore, in arable lands with different groundwater levels, the effect of water control treatments on the total greenhouse effect caused by CH 4 and N 2 O needs to be further assessed. For the further exploration of the GHG emission reduction technologies applicable to lowland polder areas, it is suggested that long-term observations be conducted on soil water content and groundwater levels in the future to study the speci c mechanisms of water control and emissions reduction.

Yield, GWP, and GHGI
The previous studies mainly focused on the cumulative emission and integrative GWP of CH 4 and N 2 O from the paddy elds. Although there are related reports on the assessment of GHG in the entire rice-wheat rotation system, that in the lowland polder area remains unclear ). China's subtropical regions are widely planted with rice, but it has always been a di cult point in related research to ensure stable crop production or even production increase when reducing GHG emissions from rice paddy soils.
Among the total GWP treated by CG, SD, and DD, the contribution rate of CH 4 was 86.01%, 71.35%, and 65.85%, respectively, in this study. Therefore, the most effective way to reduce the total amount of GHG in rice elds is usually to reduce CH 4 emissions (Lagomarsino et al. 2016). Jiang et al. (2019) pointed out that compared with continuous submerged irrigation, non-continuous submerged irrigation effectively reduces the GWP of rice elds by 44%. In south China, after adopting mid-term drainage in the double-cropping rice planting area, the GWP in the early and late rice seasons are reduced by 22% and 41%, respectively (Liang et al. 2017). Drainage in the wheat season signi cantly reduces the GWP of the entire rice-wheat rotation system, even with increased N 2 O emissions in some years, and the net GWP of the entire system has dramatically declined because of signi cantly reduced CH 4 emissions. In this study, it was found that the yield of rice increased, because drainage can improve the soil permeability of the paddy eld, and the oxygen supply is su cient, creating good growth conditions for the rice, so the GHGI of the entire rice-wheat rotation system was signi cantly reduced. In general, based on the current situation that a large amount of fertilizers are used to increase production and environmental protection is neglected, the drainage is a planting technique worth promoting.

Conclusion
The CH 4 emissions in the rice season were the main contributor to GWP of the rice-wheat rotation cropping system. The CH 4 and N 2 O emissions of the rice-wheat rotation system were obviously affected by drainage.
Compared with those of CG, the CH 4 cumulative emissions of SD and DD were reduced by 65.80% and 63.42% (rice season), and 101.37% and 77.28% (wheat season), respectively, while the N 2 O emissions increased in some years. Drainage led to an increase in the yields of both rice and wheat. In the rice season, the GHGI of SD and DD reduced by 65.46% and 62.49%, respectively (three-year total) (P<0.05). The GHGI of SD declined by 18.75% and GHGI of DD rose by 25.00% in the wheat season. It can be inferred that SD and DD ensure stably increasing production and effectively reduce GHG emissions. The CH 4 emission ux was signi cantly positively correlated with air temperature and 5 cm ground temperature, but negatively correlated with soil Eh. When the soil Eh value was lower than -150 mV, the CH 4 emission ux increased markedly with the decrease of Eh. In addition, both SD and DD led to marked decrease in soil organic matter content and an increase in soil pH, which may be the reason for decreased CH 4 emissions.