Preceding crops changed greenhouse gases emission and carbon neutrality under maize-rice and double rice cropping systems

ABSTRACT Conversion from double rice cropping (RR) to maize-rice cropping (MR) have been occurring in recent years in Asia. However, effects on the environment by introducing maize into paddy still need more examination. The objective of this study was to assess differences in greenhouse gases emissions, carbon footprint and carbon neutrality between MR and RR cropping systems from 2016 to 2018. Results showed that the global warming potential (GWP) under MR cropping was 13.0 t CO2 eq ha−1, which was 20.0% higher than that of RR (10.8 t CO2 eq ha−1) due to its higher N2O emission. Although the GWP of MR increased, it significantly decreased carbon footprint by 17.3% in comparison with RR due to its higher gain yield. Simultaneously, MR had a substantial increase in net ecosystem economic efficiency by 62.5%. Nevertheless, carbon uptake was significantly higher in MR system, its carbon neutrality was lower than that of RR by 69.3% average. In conclusion, MR could be an effective sustainable cropping system with high yield and low carbon footprint in the subtropical regions. Considering the higher GWP of MR, further research is needed on appropriate agronomic practice to reduce carbon loss to realize higher carbon neutral potential.


Introduction
Atmospheric greenhouse gas (GHG) emissions, such as carbon dioxide (CO 2 ), methane (CH 4 ) and nitrous oxide (N 2 O), are an important contributor to global climate change. Of these, food-system emissions amounted to 18 Gt CO 2 equivalent per year globally, representing 34% of total GHG emissions (Crippa et al. 2021). Meanwhile, agroecosystem is a great carbon sink due to the higher photosynthetic capacity in crops (Huang et al. 2018) and soil organic carbon (SOC) sequestration potential (Shang et al. 2021). So even though being a significant source of GHG, agricultural mitigation has become a major focus worldwide due to its great potential (Smith et al. 2008). As a major agricultural country in the world, China accounts for 17%-20% of its total GHG emissions from agriculture (Li et al. 2020). Therefore, assessing the effects of agricultural practices on soil carbon changes and developing low-carbon cropping patterns in China to achieve carbon neutrality are urgently needed (Yue et al. 2017;Chen et al. 2022).
Offsetting of GHG emissions can be achieved by emissions reduction and the enhancement of natural carbon sinks to remove carbon from the atmosphere (Piao et al. 2022). Soil carbon sequestration as a way for GHG mitigation has a huge reduction potential. As the most important terrestrial carbon sink, direct measurement of soil organic carbon (SOC) is the most convincing evidence to evaluate soil carbon sequestration (Follett 2001;Pan et al. 2004). However, this method has limitations in detecting small changes in short-term due to the substantial background SOC stock (Don et al. 2007). In contrast, aiming to determine short-term carbon changes, carbon neutrality, that is, the difference between carbon uptake and carbon loss was introduced (Jia et al. 2012;Liu et al. 2019). Via the carbon neutrality approach, we can assess ecosystem carbon sequestration potential under the current conditions. Carbon footprint (CF), as a valid indicator for assessing the environmental impact of agricultural activities, provides a powerful tool for exploring mitigation options of GHGs emissions relative to crop production (Lal 2004;Yue et al. 2017). The carbon footprint of agriculture is mainly analyzed by the direct or indirect GHGs emissions associated with a complete crop life cycle, including the yieldscaled or area-scaled carbon emissions of all inputs to the farming land (Mogensen et al. 2014). The life cycle assessment (LCA) method enables a comprehensive analysis of all GHG generated related to agricultural production processes (Brentrup et al. 2004;Masuda 2019). Zhang et al. (2017) estimated the carbon footprint of grain production in China, and showed that rice had the largest CF, followed by wheat and maize. Besides the crop type, the carbon footprint could vary with farm practices and cropping systems (Yan et al. 2015). Identifying the main contributors to the carbon footprint might provide an effective reference for GHG emission reductions. Besides the carbon footprint, another consideration is the improvement in economic profits of maize-rice cropping system, particularly the net ecosystem economic efficiency (NEEE) (Li et al. 2015a). Previous studies have shown that the introduction of maize can increase yields and improve resource efficiency (Li et al. 2015b). Nevertheless, the effect of cropping systems conversion to maize-rice on NEEE is poorly understood in terms of assessing the combined economic and environmental impacts.
China is the leading rice cultivation country, with the area harvested being 30.4 million ha in 2020 (FAO 2021). China has a large area of paddy fields in the middle reaches of Yangtze river, where different cropping systems are practiced, such as double-rice, paddy-upland rotation, etc. (Guo et al. 2017). Recently, maize has been introduced to rice rotation, and there was a switch from double paddy rice to maize-rice rotation due to the changes in rural economic and social conditions (Timsina et al. 2010). Introduction of maize into double-rice system may change the carbon cycle, and cause 'pollution-swapping' of GHGs emissions, that is, from dominant CH 4 emissions towards N 2 O and CO 2 . As the 100-year global warming potential of N 2 O is approximately 9 times that of CH 4 , suggesting that the introduction of maize into conventional rice cropping systems may increase or decrease total GHG emissions. Some researchers reported that introduction of maize into paddy could decrease global warming potentials (GWP) by resulting in a decrease in CH 4 emission due to upland maize cultivation (Weller et al. 2015(Weller et al. , 2016Janz et al. 2019). Maize introduction also leads to change in indirect GHGs emissions due to difference in agriculture practices with rice cultivation (Sun et al. 2019). However, little is known on changes in the carbon neutrality potential in paddy due to replacing early rice by spring maize.
Therefore, the objectives of this study were: (1) to assess the changes of GHGs emissions, carbon footprint at the yield scale, net ecosystem economic efficiency and carbon neutrality potential after introducing maize to double rice cropping systems; (2) to determine the feasibility of maize-rice cropping systems as a sustainable system for paddies by comprehensive evaluation on the four aspects above.

Experimental site
The field experiments were conducted from March 2016 to October 2018 at the Qujialing county (112°46′E, 30°53′N), Hubei Province, China. The region has a humid subtropical monsoon climate with abundant water and heat resources. The daily and seasonal temperature and rainfall during the experimental years were shown in Figure S1. The basic soil at 0-20 cm depth featured as pH 6.53, organic C 14.6 g kg −1 , total N 1.44 g kg −1 , available P 9.83 mg kg −1 and available K 98.2 mg kg −1 .

Experimental design and agronomic management
The experiment covered two cropping system, including maize-rice cropping system (MR) and double rice cropping system (RR). A randomized block design with three replications was adopted and each plot area was 85.5 m 2 . The plots were surrounded by ridges 30 cm in height. The ridges were wrapped with black plastic film which was inserted down 30 cm depth in the soil to prevent lateral water movement.
All agronomic management practices in the plots were identical to those of local farmer's practice. The information on planting and inputs for the two cropping systems are shown in Table  S1. The local cultivars were chosen for the experiment, including Fengken 139 for the spring maize, Ganzaoxian 11 for early rice and Tianliangyou 953 for late rice. The land was prepared using a small rotary tiller and basal fertilizer were applied before spring maize sowing or rice transplanting. Spring maize was manually sown with row spacing of 60 cm and plant spacing of 22 cm. Herbicide was immediately sprayed after maize sowing. Before rice transplanting, the plot land was soaked for 2-3 days. Both early and late rice were manually transplanted with 25 cm in row spacing and 13 cm in hill spacing. Fertilizer of urea (46.0% N), calcium super phosphate (12.0% P 2 O 5 ) and potassium chloride (60.0% K 2 O) were used for all crops. All of P, 40% of N and 50% of K were applied as basal fertilizer for maize and rice. During maize growing period, 20% of N and 50% of K were applied at the 6-leaf stage and the remainder 40% of N was top dressed at 12-leaf stage. During rice growth period, 20% of N fertilizer was top dressed at the tillering stage of rice, meanwhile, totals of 40% of N and half of K fertilizer were top dressed at the booting stage. No irrigation occurred during spring maize growth period. An alternating wetting and drying irrigation regime was practiced during rice growth stage. The crop residues were removed from the field after harvest. Inputs of fertilizer, seeds, herbicides, pesticides, electricity for irrigation, and diesel consumption were collected during the experiment.

Plant and soil sampling and measurements
Plant samplings were conducted at crop maturity. Five representative hills of rice plants or maize plants were taken at diagonal five points at the respective plot. The samples were oven-dried to a constant weight at 85°C to determine the biomass. Carbon concentrations in the root, stalk, leaf and grain were determined using an elemental analyzer (VarioMax CNS, Elementar, Germany). Rice plants of three 3-m 2 replicates in each plot were harvested to calculate the grain yield. Three sampling sites were chosen in the middle rows of each plot and the adjacent 30 maize plants at each sampling site were harvested to determine the maize grain yield. The final grain yield for rice and maize was adjusted to the standard moisture content (14%).
Soil samples (0-30 cm depth) at five diagonal positions were collected at crop harvest in each plot in 2017. Soil NH 4 + -N and NO 3 − -N were measured by the indophenol blue spectrophotometry with KCl (2 M) extraction. Nmin content was expressed in the sum of NH 4 + -N and NO 3 − -N contents. Soil active reducing substances and ferrous iron (Fe 2+ ) were determined according to Chen et al. (1997).

Measurement of direct CH 4 , N 2 O and CO 2 emissions
CH 4 and N 2 O emissions were determined in situ using a static chamber-gas chromatography method. The chamber was approximately 0.5 m or 1.2 m high (depending on crop height) and 0.35 m in diameter. At the top of the chamber, a sampling hole was made to connect a three-way valve, and a thermometer installed to determine the temperature inside the chamber. Two electric fans were installed inside the chamber to mix the gas. Each chamber is equipped with a 25 cm high steel base, which has a groove around it in 8 cm depth and 5 cm width. The steel bases were inserted into the soil about 18 cm depth at three sites along diagonal line in each plot 1 d before the gas sampling to alleviate interference on soil gas producing. Once the chambers were inserted into the groove of the bases, the water was poured into the groove to seal the chamber to prevent gas leakage. Then, gas was extracted at 0 min, 10 min, 20 min and 30 min using a 20 mL plastic syringe and then immediately injected into the 20 mL vacuumed glass vial sealed. Samplings were conducted between 8:00 and 11:00 am, because the soil temperature during this period was close to the mean daily soil temperature (Zou et al. 2005). The height and internal temperature of the chamber were recorded at each time of gas collection. The steel bases were removed from the field after gas collection and reinserted in the same sites 1 d before the next gas sampling event. The gas was collected every 7-10 days during the crop growing season and every 10-15 days during the fallow period, with adjustment according to the fertilizer application and weather conditions. The CH 4 and N 2 O concentrations were analyzed simultaneously by a gas chromatograph meter (Shimadzu GC-14B). The working condition was similar to that as described by Sun et al. (2019). The gas concentration was calculated by linear regression of the gas concentrations collected four times at each sample point. The gas emission flux was calculated according to the equation given by Zheng et al. (1998).
The CO 2 emissions were measured using a LI-8100A soil CO 2 flux system (Li-Cor Inc., Lincoln, NE, USA). The rate of CO 2 emission from the soil was inferred by measuring the rate of increase in CO 2 concentration in the respiratory chamber. Three measurements were made in each sample plot on each sampling day to get the average CO 2 flux (mg m −2 h −1 ) measurement for each plot. Measurements of CO 2 were taken at the same time as the CH 4 and N 2 O sampling.
The cumulative seasonal CH 4 , N 2 O and CO 2 emissions were calculated according to the equation, as described by : Where CE is the seasonal total emissions (kg ha −1 ), F i and F i+1 are the gas emission fluxes (mg m −2 h −1 ) from two consecutive sampling days, and d is the number of days between two consecutive sampling days.

Calculation of carbon footprint
The global warming potential (GWP) of the agroecosystem equals the total CO 2 emission equivalents of greenhouse gases based on the LCA method (Ma et al. 2013), including direct emissions of CH 4 and N 2 O from soil and indirect emissions from energy consumption of farm operations and agrochemicals inputs. Carbon footprint can be calculated by dividing GWP by the grain yield as follows (Ma et al. 2013): Where the unit of GWP was kg CO 2 eq ha −1 and the unit of CF is kg CO 2 eq kg −1 . AI i is the amount of input for each item in this study, as shown in Table S1. EF i is the emission factor corresponding to each item. According to the China Life Cycle Database (CLCD v0.7, IKE Environmental Technology CO., Ltd., China), the EF i for urea, calcium superphosphate, potassium chloride, diesel for machinery and electricity for irrigation are 2.39, 0.30, 0.15, 0.89 kg CO 2 eq kg −1 and 1.23 kg CO 2 eq MJ −1 , respectively. According to Ecoinvent v2.2 (Swiss Life Cycle Inventory Center, Switzerland), the EF i for seeds, herbicides and pesticides are 0.58, 10.15 and 16.61 kg CO 2 eq kg −1 , respectively. E(N 2 O) and E(CH 4 ) are the cumulative emissions of N 2 O and CH 4 from the field, respectively. fN 2 O and fCH 4 are the GWP coefficient at a 100-year time horizon and, are 298 and 34 kg CO 2 eq kg −1 (IPCC 2013), respectively.

Calculation of net ecosystem economic efficiency (NEEE)
The NEEE was calculated with reference to the method of Zhang et al. (2015): Where yield gains are calculated based on prevailing crop prices and grain yield. Agricultural activity costs incorporate fertilizers, pesticides, machinery operations, irrigation and seedling trays based on current prices. GWP costs are the product of carbon-trade price (103.7 CYN t −1 CO 2 -eq) and GWP (Li et al. 2015a).

Calculation of carbon neutrality
The annual carbon neutrality of the cropping system was estimated by the difference between the total C uptake and total C loss from the boundary of the system. Considering the remove of straws and grain from the field in this study, the totally net C uptake was simplified to be the fixed CO 2 in roots, litters and rhizodeposition which entered the soil. Carbon loss included direct soil CO 2 emissions (C Re ) and all equivalent CO 2 (GWP) from soil CH 4 , N 2 O, agricultural inputs and management. The annual carbon neutrality was calculated using the following equations: Carbon neutrality ¼C root þC litter þC rhizodeposition À C Re À C GWP (5) where the unit of all items was t CO 2 eq per ha for calculation. Root biomass was estimated from aboveground dry biomass (Table S2) with root/shoot ratios of 0.1 for rice and 0.09 for maize (Huang et al. 2007). Litter accounts for 5% of the total biomass (Kimura et al. 2004). Rhizodeposition C was estimated from exudates, root hairs and fine roots sloughed off with rice and maize accounting for 11% and 8.5% of total biomass carbon, respectively (Jones et al. 2009). The fixed C in roots, litters and rhizodeposition were shown in Table S2. Then, the uptake of in roots, litters and rhizodeposition were estimated from the fixed C (NPP) via the NPP/GPP ratio of 0.52 (Zhang et al. 2009).

Data analysis
Analyses of variance were performed using the Statistix 8.0 statistical package. The least significant difference (LSD) was computed to evaluate the differences between treatments at p < 0.05.

Direct CH 4 , N 2 O and CO 2 emission fluxes
Soil N 2 O flux in MR cropping system was higher than that of RR cropping system at most of sampling dates during the crop growing season (Figure 1(a)). Furthermore, MR had higher emission peaks than RR, which occurred in April, June and late July, reaching 1981 μg m −2 h −1 -2329 μg m −2 h −1 over the sampling period. While the highest peak of N 2 O flux in RR cropping was only 315 μg m −2 h −1 -454 μg m −2 h −1 among the sampling events. Both MR and RR kept lower N 2 O fluxes at most of sampling dates during the late rice period and the fallow period. There was no significant difference in averaged seasonal N 2 O flux in fallow period between RR and MR (Figure 1(a)). But the averaged seasonal N 2 O flux in the first season of MR was 243.2% and 344.8% greater than that of RR in 2017 and 2018, respectively. However, the averaged seasonal N 2 O flux during late rice season of MR was not consistent among both years, presenting as a higher value than that of RR in 2017, and comparable to that of RR in 2018.
Higher CH 4 fluxes were clearly observed in 2016 and 2017, while relatively low in 2018 (Figure 1  (b)). In particular, the highest CH 4 flux occurred in August in RR cropping was 29.5%, 13.8% and 31.4% greater than that in MR cropping in 2016, 2017 and 2018, respectively. In contrast, CH 4 emission flux was generally low during the fallow period and no significant difference among both cropping systems. There were only two small peaks in May and June in early rice season. Nevertheless, the averaged seasonal CH 4 emission fluxes of RR were significantly higher than those of MR in first and second crop seasons, with 32.9% and 187.1% higher averaged across the years of 2017 and 2018, respectively (Figure 1(b)).
There were clear seasonal fluctuations in soil CO 2 flux during the crop growing season from 2016 to 2018 (Figure 1(c)). The soil CO 2 flux varied little and were low during the fallow period, but mainly concentrated during crop growth and was distinctively higher in the first crop season each year. The CO 2 flux of MR remained relatively high and ranged from 282 mg m −2 h −1 to 1059 mg m −2 h −1 over the maize growing season, which was higher than that in early rice season of RR on most of sampling dates. There was little distinction in averaged seasonal CO 2 flux between MR and RR during the fallow period and late rice season. However, the averaged seasonal CO 2 flux over the first crop season was significantly higher in MR than that in RR by 106.2% in 2017 and 115.3% in 2018.

Cumulative emissions and GWP
The seasonal cumulative emission of CH 4 , N 2 O and CO 2 under two cropping systems were quite different ( Table 1). The seasonal cumulative CH 4 , N 2 O and CO 2 emissions remained similar between MR and RR during the fallow period. However, the highest CH 4 emissions occurred during the rice growing season, while the highest N 2 O and CO 2 emissions were clearly observed during the maize growing season.
The introduction of MR significantly decreased cumulative CH 4 emissions in first crop season by 48.6% and 72.3% in 2017 and 2018 (Table 1), respectively. Meanwhile, there were also notable reductions in CH 4 by 9.7% and 19.9% in second season under MR cropping in comparison with RR in 2017 and 2018, respectively. Totally annual emission of CH 4 under MR cropping was 54.0 kg ha −1 , which was lower than that of RR cropping by 24.2% averaged across two experimental years. Conversely, N 2 O cumulative emissions were higher during the maize season, averaging 13.8 kg ha −1 over two years, and lower during the late rice season, averaging 2.45 kg ha −1 over two years (Table 1). The cumulative N 2 O emissions of MR were significantly higher by 308.7% and 362.2% in first and second season, respectively, than that of RR over two years. The introduction of maize in 2016-2017 was followed by higher N 2 O emissions in the subsequent rice season, approximately twice as high as in the double rice cropping system, whereas this was not observed in 2017-2018. Overall averaged across two experimental years, the annually cumulative N 2 O emission under MR cropping reached to 20.4 kg ha −1 , which was 2.41 times of that under RR cropping (8.5 kg ha −1 ).
The seasonal variation in cumulative CO 2 emissions in MR cropping was similar to that of N 2 O, remarkably higher in the spring maize season and obviously lower in all rice cropping season (Table 1). Cumulative CO 2 emissions from the two cropping systems differed significantly only in the first season and remained consistent for the rest of the season. It was increased by 137.1% and 114.2% in spring maize season than that in early rice season in 2017 and 2018, respectively. Thus, the introduction of maize significantly increased annually cumulative CO 2 emissions to 33,012.5 kg ha −1 averaged across both years, which was higher by 45.2% than that under RR cropping.
The annually GWP of MR and RR cropping were 12,993.5 kg CO 2 eq ha −1 and 10,823.5 kg CO 2 eq ha −1 averaged across both experimental years, respectively ( Table 2). The difference in GWP during the fallow period is consistent with the cumulative emissions, as the GWP was derived from direct emissions. The GWP in first crop season of MR was greater than that of RR by 51.8% in 2017 and 50.4% in 2018, but varied little in late rice season. During maize growing season, the GWP mostly originated from direct N 2 O emissions, which contributed 66.7% in 2017 and 62.9% in 2018 (Table 2). In contrast, the GWP in rice season was mainly caused by CH 4 and electricity consumption for irrigation. Among the indirect emissions, emissions from farm operations were obviously greater than those from agricultural inputs in rice cropping system, while it was reversed in maize due to high nitrogen fertilizer. Totally, higher indirect emissions were observed in RR than in MR cropping. Values are mean ± standard errors. Different small letters in the same column indicate significant differences among both cropping systems at p < 0.05.
The annual GWP differed significantly among two cropping system, mainly from the discrepancy of direct CH 4 and N 2 O emissions in first season. In terms of annual GWP, MR was 20.3% and 19.7% higher than RR in 2017 and 2018, respectively (Table 2).

Crop yield, carbon footprint and net ecosystem economic efficiency
Grain yields varied considerably between MR and RR, with similar trends in both experimental years (Table 3). Compared to RR cropping, MR cropping significantly increased grain yield by 74.7% in the first crop season and by 18.6% in the late rice season averaged across both years. Consequently, the annual grain yield of MR increased by 39.9% in 2017 and 50.7% in 2018 compared with that of RR. The MR cropping exhibited a higher annual yield potential than RR cropping. This was mainly due to the high yields of maize, which contributed 78.3 to 80.0% of the increase. Due to higher grain yield, the carbon footprint (CF) of MR was significantly lower than that of RR, with the exception that CF differed little in first season of 2017 (Table 3). Compared with early rice, maize in MR had lower CF by 4.6% and 21.1% in 2017 and 2018, respectively. Notably, the carbon footprint was significantly reduced by 16.8% and 16.9% in late rice season after the introduction of maize in 2017 and 2018, respectively. In short, MR cropping greatly reduced annual CF by 17.3% compared with RR averaged across both experimental years.
The highest net ecosystem economic efficiency (NEEE) was obtained in maize cultivation season, which was 1.2 times that of the early rice season averaged across both years (Table 3). Furthermore, MR cropping also had the higher NEEE in the late rice season, increasing by 28.4% averaged across both years. As a result, the annual NEEE of MR was distinctly higher, approximately 1.5 and 1.7 times than that of RR in 2017 and 2018, respectively.

Carbon neutrality
Over two cropping system, carbon neutrality was negative (Figure 2), implying that either maizerice cropping system or double rice cropping system was acting as a carbon source. Different discrepancies in carbon uptake can be found in different treatments, with a variation from 20.7 to 25.7 t CO 2 eq ha −1 . MR showed a greatly increase of carbon uptake by 15.5% to 24.1% than that of RR over two years. In contrast, carbon loss ranged from 31.8 to 46.9 t CO 2 eq ha −1 , and MR exhibits a higher amount of carbon loss as well, with a significant increase of 32.8% to 41.9%. In this case, the increase of carbon loss outweighed the increase of carbon uptake in MR. Consequently, the higher negative carbon neutrality was found in MR, with a noticeable decrease of 93.7% in 2017 and 45.0% in 2018 in comparison to RR cropping ( Figure 2). Different small letters in a same column in the same crop season indicate significant differences between both cropping systems at p < 0.05.

Effect of cropping systems on the GHGs emissions
In the current study, significant CH 4 fluxes were only observed during rice season and were negligible in fallow and maize season (Figure 1(b)), which is in line with previous study (Sun et al. 2019). Considerable studies proved that the anaerobic conditions in irrigated rice field promoted microbial methanogenesis (Wang et al. 2012;Tariq et al. 2018) and further found lower CH 4 emissions in the dry season (Shang et al. 2011). When switching to maize instead of early rice, the activity of methanogenic bacteria was suppressed due to the reduced water content and improved soil aeration. Accordingly, soil inundation resulted in higher CH 4 emissions during rice cultivation in this experiment (Figure 1(b), Table 1), as reported in previous studies (Weller et al. 2016;Jiang et al. 2020). Similarly, the higher CH 4 fluxes in late rice season were observed at many sampling dates than those in the early rice season (Figure 1(b)). This result was partially due to the higher air temperature during the late rice season than that in first season ( Figure S1). Studies have shown that the higher temperature can accelerate the decomposition rate of soil organic matter (Khalil et al. 2008) and enhance the activities of dominant methanogens , thereby increasing CH 4 emissions in second season (Figure 1(b)). In addition, late rice could access more fresh soil organic matter than early rice due to its immediately transplanting right after the first season crop harvest and which will also account for its higher CH 4 emissions (Figure 1(b)). Noticeably, when maize was the crop instead of early rice, significant CH 4 reductions occurred in the in late rice season (Figure 1(a), Table 1). The lower CH 4 emissions in later rice season of MR could be explained by the aerobic condition during maize season, which adversely affects activity of the methanogenic bacteria or led to substantial mortality (Breidenbach et al. 2016). Meanwhile, previous studies have found that the number of methanogens in fields continued to decrease after maize cultivation and were more difficult to recover in a short time, thus affecting the communities of methanogenic bacteria during the later rice season. Breidenbach et al. 2016). Another reason for reducing CH 4 emissions was the production of oxidants following improved soil aeration during the maize season, in particular the re-oxidation of Fe (II) (Van Bodegom et al. 2000), which could extend into the latter crop season and accelerate the oxidation of methane. In our study, the Fe (II) content and total active reducing substances of MR were significantly lower than that of RR (Table S3), implying that more oxidants in the maize and late rice season. Therefore, the lower CH 4 emissions in late rice of MR were probably ascribed to production of less methane and oxidation. This result was consistent with those reported by Sun et al. (2019). In addition, higher soil inorganic N was observed in late rice season of MR cropping than RR cropping (Table S3), which could be due to more nitrogen fertilizer application in maize season which left more residue N in late rice season. This effect benefited late rice growth (Table 3, Table S2) and might enhance methanotrophs growth due to higher rhizosphere C inputs and better ventilation by aerenchyma (Liang et al. 2013) and result in increase of CH 4 oxidation in late rice season of MR cropping. Greater interannual variation in CH 4 emissions were observed in this study, especially during the late rice season (Figure 1(b)), which could be influenced by the disparity in rainfall ( Figure S1). Less precipitation dropped during late rice season in 2018 required more irrigation events (more electricity consumed as shown in Table S1), which could improve soil ventilatory condition and thus suppressed CH 4 emissions (Figure 1(b)). In addition, the carbon loss was far more than carbon input into the soil under MR and RR cropping each year (Figure 2, Table S2), and this indicated that SOC declined with the year under our experimental conditions and partially cut down the C source for CH 4 producing. Unlike the reduced CH 4 emissions, the incorporation of maize led to a substantial increase in N 2 O emissions from MR compared to RR (Figure 1(a), Table 1). This trade-off may eliminate the impact of introducing maize on greenhouse gas emissions and mitigation in light of the diverse growing condition and agricultural activities. During the experiment, N 2 O emissions in maize growing season were approximately 4 to 5 times higher than in rice growing season, as shown in previous studies (Weller et al. 2015;Sun et al. 2019). In general, N 2 O is produced as an intermediate product of nitrification and denitrification, and its production is influenced by changes in the soil air condition and increases in the substrate (Zheng et al. 2000). The introduction of maize into paddy rice reduced the soil water content and changed the conditions from anaerobic to aerobic, which is preferable to the nitrification process (Butterbach-Bahl et al. 2013). Thus, the higher N 2 O emission was obtained during maize cultivation than rice cultivation (Figure 1(a)). Moreover, the higher nitrogen fertilizer was used in maize production than in rice production (Table S1), increasing the availability of substrates (Table S3) and facilitating the emission of N 2 O during nitrification process. Thereby, the N 2 O peak emissions of MR occurred after fertilization during the maize growing season, a situation also founded in early rice cultivation (Figure 1(a)). Similar effects of more nitrogen fertilizer were found by previous research (Sun et al. 2019). On the other hand, although the application of nitrogen fertilizer in the paddy field increased N 2 O emissions, the flooded condition of the paddy field suppressed both nitrification and denitrification (Abao et al. 2000), ultimately exhibiting lower emissions. Accordingly, N 2 O emissions during the rice season were generally low (Figure 1(a), Table 1). However, more residual N from maize season resulted in higher soil inorganic N content (Table S3) possibly resulted in higher N 2 O emissions in later rice season of MR cropping than that of RR cropping (Figure 1(a)). More N 2 O emissions were observed in 2017 than that in 2018 (Table 1), which had close relations to the sharp decline in soil inorganic N supply for denitrification in 2018 (Table S3). This phenomenon might be a reason that soil organic matter possibly declined with the year in view of more C loss ( Figure 2) and somewhat slowed down soil organic N mineralization.
The CO 2 emission flux and the cumulative CO 2 emission were substantially greater in the maize season than in the rice season ( Figure 1c, Table 1), which were consistent with previous study which found that conversion of paddy rice fields to upland crops enhanced soil CO 2 emissions (Breidenbach et al. 2016). Normally, 85% of CO 2 production in agricultural ecosystems originates from soil microbial activity and 15% from crop root respiration (Davidson et al. 1998). Thus, soil aeration during the spring maize season accelerated soil microbial activity and enhanced soil respiration (Figure 1(c), Table 1). In addition, some studies reported that higher nitrogen fertilizer can boost CO 2 emissions due to plant growth, root activity and respiration (Xu and Wan 2008). This can be verified by having higher applied nitrogen fertilizer (Table S1) and higher root and litter carbon in the maize season (Table S2). However, opposite interannual variation in CO 2 emissions were observed in comparison with those of CH 4 and N 2 O emissions. More CO 2 emissions occurred in 2018, especially showed tremendous increase in late rice season (Table 1). Less rainfall and higher temperature during the late rice season in 2018 than in 2017 ( Figure S1) provided strong reasons for this observation.

Effect of cropping systems on the GWP, carbon footprint and NEEE
In our study, maize-growing season was of much greater importance for GWP when compared with the rice-growing season, accounting for almost 50% of GWP under MR cropping ( Table 2). The tradeoff between CH 4 and N 2 O during the maize-growing season could not counteract its GWP associated with tremendous N 2 O emissions. The introduction of MR increased GWP by 19.7-20.3% compared to RR, which is inconsistent with the results of previous studies (Weller et al. 2015;Sun et al. 2019;Jiang et al. 2020). This result can be attributed to higher N fertilization (up to 300 kg N ha −1 ) resulted in higher N 2 O emissions during the maize season, amounting to 25% higher than in previous reports (Sun et al. 2019;Jiang et al. 2020). Thus, the contribution of N 2 O emissions in GWP accounting outweighed that of CH 4 emissions in MR cropping (Table 3), while the equivalent CO 2 from CH 4 emissions were the main part of GWP in previous reports (Sun et al. 2019;Jiang et al. 2020). Furthermore, the basic SOC of our experimental site was remarkably lower than that in previous reports (Sun et al. 2019;Jiang et al. 2020) and evidently had lower substrates for CH 4 emission and further reduced its importance in GWP accounting. This result suggests a considerable potential for reducing N 2 O production and indirect emissions through less nitrogen fertilizer use during maize cultivation to lower its GWP.
The carbon footprint, as a composite indicator of ecological and yield benefits, is a guide to exploring greenhouse gas emission reductions under different cropping systems (Lal 2004). During the trial, differences in yields and CF under different cropping systems were observed, with MR having higher yields and a lower carbon footprint compared to RR cropping (Table 3). Our findings were consistent with the previous reports (Sun et al. 2019;Jiang et al. 2020). In our study, the maize season of MR was the most productive as a C 4 plant and maize generally had a higher yield potential than other cereal crops. It is also remarkable that the late rice season yields in the MR were improved (Table 3). This observation was most probably caused by the aerobic conditions of spring maize cultivation which accelerated the mineralization of soil organic matter and improved the nutrient utilization during the late rice season. Moreover, more residual fertilizer N was passed to the late rice season and thus increased the yield of late rice. This means that having a higher GWP does not contribute to a larger CF and that yield plays a huge role. Furthermore, with successive maize-rice cropping systems, performance in yield and CF became progressively better.
Recently, the net ecosystem economic efficiency (NEEE) of different cropping systems is of major consideration in crop production Li et al. 2015a), demonstrating the relationship between the economic profitability and environmental sustainability of different cropping systems. To our knowledge, there are few reports on NEEE of introducing maize into double rice cropping system. Our study showed that each crops season and annual NEEE of MR were remarkably higher than those of RR (Table 3). This result can be explained by the fact that the yield benefits from the introduction of maize have the potential to outweigh the environmental costs of high GWP. Accordingly, higher grain yield gains can compensate for the costs of agricultural activities and GWP. Our study further showed that the conversion of double rice cropping system to maize-rice cropping system provides an effective system with higher yields, lower CF and higher economic efficiency.

Effect of cropping systems on the carbon neutrality
The carbon neutrality can provide an effective tool to assess the carbon fixation capacity of agroecosystems in the short term through carbon uptake and carbon loss (Smith et al. 2010). In our study, carbon uptake in both cropping systems were significantly lower than the carbon losses ( Figure 2). It demonstrates that carbon uptake was not sufficient to offset carbon loss, resulting in MR and RR being carbon sources with carbon neutral values of −10.7 to −21.3 t CO 2 eq ha −1 . This result is partially supported by previous research suggesting MR as a source of carbon (Sun et al. 2019). During the trial, the maize-rice cropping system dramatically increased carbon uptake by 15.5 to 24.1% compared to the RR (Figure 2), due to the higher carbon photosynthetic fixation in spring maize than in early rice and more root-C and litter-C entered the soil (Table S2). Concurrently, an increased carbon loss in MR of 32.8 to 42.0% was caused by massive greenhouse gas emissions (Figure 2), especially in the maize season when CO 2 emissions were more than twice as high as in the rice season (Table 1). In particular, direct CO 2 emissions from ecosystems account for 60.8-76.5% of carbon loss for MR and RR. This finding is supported by previous studies (Wu et al. 2018;Jiang et al. 2020), showing that CO 2 is the dominant contributor to carbon loss. Previous reports have shown that the conversion of rice paddy to upland cultivation tends to increase soil organic matter mineralization (Ma et al. 2013;Wu et al. 2018), which leads to soil organic carbon loss. Thus, greater negative carbon neutral values were observed in MR than in RR cropping ( Figure 2). However, it was possible to balance or increase SOC by increasing C inputs into the soil from crops production. One potential approach is to further raise maize and rice productivity to increase root rhizodeposition. Another possible way is to return the residue to the soil by optimization of nitrogen fertilizer management. Some previous studies have shown that reasonable straw return combing with N application method could enhance SOC, reduce reactive N loss and decrease C footprint without extra input Bai et al. 2021). Further studies are required to access the impact of maize-rice cropping on carbon neutrality in paddy.

Conclusions
Introduction of maize into double rice cropping system led to great influence on GHG emissions, CF and carbon neutrality. The GWP was increased by 20.0% under MR caused by higher N 2 O emissions during spring maize season averaged across two experimental years. However, the higher GWP did not result in a higher CF in MR, on the contrary, which was significantly lower by 14.0 to 20.5% compared to RR cropping. Simultaneously, conversion of RR to MR gained a substantial increase in yield and NEEE. These results suggest that the introduction of maize-rice cropping system can be an alternative cropping system under the requirements of crop production demand, increased economic efficiency and reduced carbon footprint. However, both MR and RR had a negative carbon neutrality due to higher carbon loss than carbon fixation, and carbon neutrality of MR was significantly lower than RR cropping. Further research is needed on appropriate agronomic practice to enhance the carbon neutrality potential under MR cropping.