Introduction

Atmospheric CH4 is recognized as one of the most important greenhouse gases (GHG) due to its high global warming potential (GWP) (34 times greater than CO2 at the per molecule level in a 100-year time horizon) (Myhre et al. 2013). CH4 contributes to approximately 20% of the total global warming forcing (Schulze et al. 2009), and it is projected that CH4 emissions may increase by as much as 60% by 2030 (Smith et al. 2007). Paddy rice cultivation is a major source of global CH4 emissions, contributing to approximately 11% of the total global CH4 emissions (Smith et al. 2014). To meet food demands for the world’s growing population, rice production has to increase by 60% in the next few decades (Cassman et al. 1998). As the largest rice producer in the world, China produces approximately 36% of the global rice with 23% of the world’s total rice planting area (Dong et al. 2011).

Double rice cropping systems are widely adopted in southern China, representing the dominant crop system in the region (Yang et al. 2010). In the past several decades, new rice fields have been widely developed in southern China due to rapidly increasing rice demand from population growth. Land-use legacies arising from land-use change could impact ecosystem carbon and nitrogen cycles in various ways, such as changing soil nutrient pools (McGrath et al. 2001; Zhang and He 2004), modifying soil microbial communities (Chu et al. 2009) and altering GHG emissions (Eusufzai et al. 2010; Nishimura et al. 2011). Several studies showed that CH4 emissions were much lower in new rice fields recently converted from uplands than old rice fields (Eusufzai et al. 2010; Nishimura et al. 2011). Multi-year studies on CH4 emissions in new rice fields are necessary to investigate the lagged effects of land-use change (Hatala et al. 2012) and are important for improving GHG inventories. However, long-term studies on CH4 emissions in new rice fields converted from upland cropping systems are still scarce, especially in subtropical regions.

CH4 emissions in new rice fields can also be affected by many other factors, such as water and fertilization management practices. As an effective way to increase rice production, intensive management practices with high nitrogen inputs through fertilization have been adopted widely in south China. Nitrogen fertilization generally leads to greater N2O emissions (Zou et al. 2005; Shang et al. 2011). However, N fertilization can also reduce CH4 emissions (Banger et al. 2012; Linquist et al. 2012) and thus decrease the overall GHG of CH4 and N2O expressed as carbon dioxide equivalents (Dong et al. 2011). However, there are no consistent conclusions on the net effects of N fertilization on CH4 emissions (Dong et al. 2011; Shang et al. 2011; Banger et al. 2012). Some studies showed that N fertilization could increase CH4 emissions in rice paddies (e.g., Schimel 2000; Yang et al. 2010), while others have reported a significant decrease (approximately 30–50%) (Dong et al. 2011; Xie et al. 2010) or no effects (Lindau et al. 1991; Wang et al. 2012; Pittelkow et al. 2014).

Given the importance and complexity of the N fertilization effect (NFE) on CH4 emissions in rice paddies, numerous studies have been conducted (e.g., Bodelier et al. 2000a; Bodelier and Laanbroek 2004; Banger et al. 2012; Pittelkow et al. 2014). However, most studies have focused on N fertilization per se, such as N form and application rates (e.g., Bodelier et al. 2000b; Cai and Mosier 2000; Dong et al. 2011). Few have examined how site-specific environmental factors affect the NFE on CH4 emissions, except for Banger et al. (2012), who reported that the NFE on CH4 emissions was closely related to water management activities in rice paddies. Perring et al. (2016) suggested that land-use legacies could alter system dynamics by modulating contemporary environmental changes in biogeochemical processes, such as the effects of N addition on the carbon cycle. Therefore, the objectives of the current study were to examine if (1) land use legacies affect CH4 emissions, if (2) N fertilization affects CH4 emissions, and if (3) land use legacies affect the NFE in rice fields in subtropical China.

Methods

Site description

The experiment began in the late rice season in 2012 at the Qianyanzhou Ecological Research Station (26°44′46″N, 115°04′05″E), which is located in Jiangxi Province in south China. The region experiences a monsoon climate, with a dry season typically extending from July to September and a rainy season from April to June. According to meteorological data, the mean annual temperature is 18.0 °C, and the mean annual precipitation is approximately 1500 mm. The double rice cropping system is the main cropping system in this region. The soil is a typical red soil and is classified as Ultisols in USDA soil taxonomy. The parent material of the soil consists of red sandstone and sandy conglomerate. The topsoil consists of 17% clay (<0.002 mm), 25% silt (0.002–0.02 mm) and 56% sand (0.02–2 mm). The soil pH was 4.97, and the bulk density was 1.29 g/cm3.

Experimental design

We designed an experiment with four treatments in a split-block design with four replicates: old rice paddy with fertilization (OF), old rice paddy without fertilization (OUF), new rice paddy with fertilization (NF), and new rice paddy without fertilization (NUF). The old rice paddy was less than 100 m away from the new rice paddy. The old rice paddy had been cultivated continuously for 10 years before the experiment started. The new rice paddy was converted from a 5-year plantation of oil tea camellia (Camellia oleifera). The rice paddies had been cultivated following a conventional double-rice cropping system with an early rice season from April to July, a late rice season from July to November, and a fallow season normally from November to the following April. Both the late and early rice were transplanted at a density of 235,000 hills/ha with a spacing of 25 cm by 17 cm. Each plot had an area of 10 m × 10 m. The fertilization rate was 180 kg N/ha in each rice growing season, which is a typical rate for the study area. As a common practice for double rice systems, fertilizers were applied twice during each growing season. Compound fertilizers (N:P2O4:K2O = 15%:15%:15%) were applied first as a basal fertilizer at a rate of 108 kg N/ha at the time of rice transplanting, which was May for the early rice season and July for the late rice season. In the early-tillering stage of rice growth, we applied urea at a rate of 72 kg N/ha to the field. All plots were under typical water management, with an intermittent irrigation regime (flooding-drainage-moist during the tillering, panicle and ripening stages, respectively).

CH4 flux measurements

Static chambers were used to collect air samples, which were then analyzed with a gas chromatograph (GC system, 7890A, Agilent Technologies, Santa Clara, CA, USA). The chamber system is composed of two parts: a movable cylindrical steel chamber body and a fixed steel base with a diameter of 49 cm, covering an area of 0.188 m2. The base was inserted into the soil to a depth of 15 cm for the entire rice growing season. The chamber body has a diameter of 49 cm and a height of 69 cm. At the beginning of each round of measurements, we positioned the chamber body on top of the base and used a silicon tube to seal the joint to form a closed system. Two mixing fans were installed inside each chamber for mixing the air in the chamber space. The fans were powered by batteries and positioned at the top of the chamber to minimize disturbing the boundary layer. We turned on the fans before deploying the chamber body to avoid perturbation to the chamber pressure when turning on the fans. We also installed a pressure balance tube at the top of the chamber. Five rice plants were included in each base frame. Gas sampling was usually carried out twice per week. After fertilizer was applied, we collected gas samples every day for approximately a week.

We calculated CH4 fluxes by measuring CH4 concentration increases inside each chamber when it was closed. Gas samples from the closed chamber were collected with a custom-built manifold, which consisted of a 12 mL Exetainer vial (Labco Limited, Lampeter, Ceredigion, UK), an air pump (KNF Neuberger, Inc., Trenton, NJ, USA), a pressure sensor and a solenoid valve. The pump stopped when the pressure inside the vial reached 3 bars. The vials were evacuated with a vacuum pump prior to the field measurements. Five gas samples were collected in each chamber at 10 min intervals. The first sample was collected 30 s after the chamber was closed. The measurements for each chamber were completed in approximately 40 min. The gas samples were then transported to the laboratory and analyzed using a gas chromatograph (GC System, 7890A, Agilent Technologies) equipped with an electron capture detector (ECD) and a flame ionization detector (FID). CH4 fluxes were determined by fitting a linear regression model where CH4 concentration was the dependent variable and sampling time was the independent variable. The following equation was used to calculate CH4 fluxes:

$$F = \rho \cdot V/A \cdot P/{P_0} \cdot {T_0}/\left( {T + 273} \right) \cdot d{C_t}/dt$$
(1)

where F is the CH4 flux rate (mg C/m/h), ρ is the air density (1.29 kg/m3), V is the chamber volume (m3), A is the base area (m2), P is the site air pressure, P 0 is standard air pressure at sea level (101.3 kpa), T is air temperature in the chamber (°C), T 0 is a reference air temperature of 273 K, and dC t /dt is growth rate of the CH4 concentration in the chamber (mg C/h).

The measurement of CH4 fluxes in the new rice paddy continued for 3 years, while that in the old rice paddy was carried out only in the first year of the 3-year study. Annual cumulative emissions of CH4 for all treatments were directly computed from the measured fluxes by linear interpolation for the days without measurements. We quantified the NFE on CH4 fluxes using the difference (∆F) of CH4 fluxes between the plots with and without fertilization. A negative NFE meant that N fertilization reduced CH4 emissions, while a positive NFE meant that N fertilization stimulated CH4 emissions. N fertilization-induced CH4 emissions were defined as CH4 emissions per kg N fertilizer and were calculated by dividing the difference of CH4 emissions between the plots with and without fertilization by the amount of N fertilizer (Bouwman 1996; Banger et al. 2012).

Measurements of climatic factors

Air temperature inside the chamber headspace was measured with a thermocouple simultaneously with gas sampling. Soil temperature and soil moisture at a 5 cm depth were automatically recorded by an automatic data-logging system via wireless technology. Water depth in the rice fields was monitored using a steel ruler when the fields were flooded. Air temperature and precipitation were collected from an on-site automatic meteorological station adjacent to the experimental plots.

Plant biomass

Aboveground rice biomass was determined just before harvesting. The rice plants were sampled at four random subplots at an area of 1 m x 1 m in each plot. The rice plants were then separated into grain and straw, which were oven dried at 65 °C for 48 h for dry matter weight determination. Rice roots were sampled at three 1 m × 1 m subplots immediately after harvesting in each plot. The roots were washed thoroughly, oven dried and weighed.

Soil sampling and laboratory analysis

Soil samples in each plot were collected every 10–15 days during the study period to analyze the concentrations of soil nitrate (NO3 −–N) and soil ammonia (NH4 +–N) from both treatments. In each plot, soils were collected using a 5-cm diameter auger in the plough layer (0–10 cm) at five randomly selected locations and then mixed as one sample. All fresh soil samples were sieved through a 2.0 mm sieve and stored in a refrigerator for nutrient analysis.

Soil pH was measured with a glass electrode in a 1:2.5 soil:water solution at the end of each rice growing season. Soil NO3 −–N and NH4 +–N were extracted with 1 M KCl, and the filtrate was analyzed by an AA3 HR AutoAnalyzer (SEAL Analytical GmbH, Norderstedt, Germany). Soil samples for total N and soil organic C were collected at the end of each season and air-dried. Soil total N and total C contents were analyzed by dry combustion using vario MACRO cube (Elementar Analysensaysteme GmbH, Langenselbold, Germany), while soil inorganic C was analyzed using a 08.53 calcimeter (Eijkelkamp Soil & Water, Giesbeek, The Netherlands).

Statistical analysis

All data were tested for normal distribution before statistical analysis, and for data that not normally distributed, a natural logarithm transformation with certain constants were adopted. Duncan’s multiple range test was used to analyze the differences of soil temperature, moisture, soil pH, soil organic carbon content and rice biomass among the treatments in different studying years. Two-way ANOVA was used to determine the main and interactive effects of land-use legacies and N fertilization on CH4 fluxes. One-way ANOVA with Duncan’s multiple range test was used to examine the differences of the NFE on CH4 fluxes and annual CH4 emissions among the rice plots under different land-use years. A paired-sample T test was used to analyze the effects of N fertilization on annual mean fluxes and annual cumulative emissions of CH4 in rice plots under different land-use years. All statistical analyses were conducted with IBM SPSS 22.0 (IBM, New York, USA).

Results

Environmental factors

The seasonal patterns and variations of soil temperature, precipitation and soil moisture in the rice plots are shown in Fig. 1. There were no significant differences of soil temperature and soil moisture among the different treatments during the study. Annual precipitation did not vary much among the years, with values of 1639, 1587 and 1510 mm in the three years, respectively. The soil organic carbon (SOC) content in the new rice fields was significantly lower than that in the old rice fields (Table 1). There was no significant difference of SOC in the new rice paddy among the different study years (Table 1). The soil pH in the new rice paddy increased from 4.93 to 5.11 in the NF plots and from 4.85 to 5.09 in the NUF plots during the three-year study, which was markedly lower than that in OF and OUF (Table 1).

Fig. 1
figure 1

Daily mean soil temperature at a 5 cm depth, daily precipitation (upper panel), and seasonal dynamics of soil moisture (bottom panel) during the study period

Full size image
Table 1 Climatic factors, soil properties and rice biomass (mean ± SE) under various treatments during the study period
Full size table

Seasonal dynamics of CH4 fluxes

CH4 fluxes fluctuated considerably in all treatments during the study period (Fig. 2) with similar seasonal patterns. During the rice growing seasons, CH4 fluxes gradually increased until seasonal peak fluxes were reached at approximately 10–25 days after rice transplanting when the plots were waterlogged. After midseason drainage, CH4 fluxes dropped sharply in all plots. Thereafter, CH4 fluxes remained at a low level due to the alternating dry and wet cycles until rice harvest. During the fallow seasons, CH4 fluxes were negligible, fluctuating around zero.

Fig. 2
figure 2

Seasonal variations of CH4 fluxes under the different treatments during the study years of 2012–2013 (upper panel), 2013–2014 (middle panel) and 2014–2015 (lower panel)

Full size image

The effects of land-use legacies on CH4 emissions

Land-use legacies significantly affected CH4 fluxes (Table 2). CH4 fluxes from the old rice plots were significantly higher than that from the new rice plots. CH4 fluxes increased with land-use years irrespective of N fertilization, following the order of 1-year < 2-year < 3-year < 10-year (Table 3). As a result, the cumulative emissions of CH4 were significantly lower in the new rice paddy than the old rice paddy, regardless of N fertilization and land-use years in the new rice plots (Fig. 3). Annual CH4 emissions also increased with the land-use years of the rice plots (Fig. 3). Annual CH4 emissions increased by 40.7 and 49.5%, respectively, in the second and third year of cultivation of the NF plots compared to the first year. In parallel, the NUF plots observed an increase of 0.8 and 39.8%, respectively, in the second and third year compared to the first year after the new rice fields were developed.

Table 2 Results of a two-way ANOVA for the main and interactive effects of land-use legacies and N fertilization on CH4 fluxes
Full size table
Table 3 Annual mean CH4 fluxes (mg C/m2/h, mean ± SE) and the NFE on CH4 fluxes in rice plots under different land-use years
Full size table
Fig. 3
figure 3

Annual cumulative emissions of CH4 in the different rice paddy land-use years. Vertical bars indicate standard errors of four replicates, and different letters indicate significant differences at the p < 0.05 level

Full size image

The effects of N fertilization on CH4 emissions

N fertilization had no significant effect on CH4 fluxes during the study period (Table 2). However, the NFE on CH4 fluxes varied significantly with the land-use years of the rice plots (Table 3). In the first year, N fertilization significantly decreased CH4 fluxes from the new rice plots, whereas N fertilization had no significant effects on CH4 fluxes from the 10-year-old rice plots (Table 3). N fertilization had no significant effects on CH4 fluxes during the second and third year of cultivation of the new rice plots (Table 3). N fertilization significantly reduced CH4 emissions by 36.9% from the new rice plots during the first year of cultivation, whereas it had little effect on the annual CH4 emissions from the 10-year-old rice paddy (Fig. 3). N fertilization-induced CH4 emissions increased with the land-use years of rice plots, following the order of 1-year < 2-year < 3-year < 10-year with values of −0.21, −0.07, −0.05 and 0.13 kg CH4/ha/kg N, respectively.

Discussion

Land-use legacies affect CH4 emissions

The seasonal pattern of CH4 fluxes in the present study was in line with many previous studies (Sass et al. 1992; Yagi et al. 1996; Jia et al. 2001; Xu and Inubushi 2004). The results showed that the seasonal pattern of CH4 fluxes was mainly determined by water regime and rice growing stages during the rice-growing season, irrespective of the former land use history and nitrogen fertilization. Annual CH4 emissions from the old rice plots in this study were within the range of previous studies (Feng et al. 2013). However, annual CH4 emissions from the new rice plots were much lower than the old rice plots at the site and those under similar climatic conditions and management practices in southern China (e.g., Tang et al. 2014).

Lower CH4 emissions from the new rice fields compared to the old rice fields were also supported by the literature. For example, CH4 emissions from a rice paddy recently converted from upland soybean were significantly lower than that from a long-term rice paddy (19-year-old) (Eusufzai et al. 2010), and the introduction of upland crops into the double rice cultivation significantly decreased CH4 emissions in subsequent paddy rice cultivation (Nishimura et al. 2011; Weller et al. 2016). According to previous studies, lower CH4 emissions in the new rice fields converted from upland soils can be ascribed to land-use legacy effects, which included higher soil redox potential (Eh), lower soil organic matter content (Eusufzai et al. 2010; Nishimura et al. 2011), and the lack of labile organic substrate in the upland soils (Hatala et al. 2012).

In the current study, the soil organic carbon (SOC) content in the new rice plots was significantly lower than in the old rice plots regardless of N fertilization management (Table 1). The long-term continuous cultivation of paddy rice favors SOC sequestration (Huang et al. 2012). The cropping duration of paddy rice could be the main reason for the significant difference of SOC between the new and old rice plots. Since SOC is an important substrate for CH4 production (Watanabe et al. 1999), the significant differences of CH4 emissions between the new and old rice plots were probably caused by lower SOC contents in the new rice plots inherited from upland plantations compared with the old rice plots.

Furthermore, CH4 emissions significantly increased with time after establishment of the new rice fields (Fig. 3). The increasing trend of CH4 emissions also reflected the effects of land-use legacies on CH4 emissions and indicated that the effects of land-use legacies weakened with increasing time of new rice paddy cultivation (Eusufzai et al. 2010; Hatala et al. 2012; Nishimura et al. 2011). However, significant increases of SOC content in rice paddies usually takes decades (Zhang and He 2004), and during the study period, we found no significant increase of SOC content with land-use years of the new rice fields (Table 1). Given that climatic conditions were consistent and rice biomass did not significantly increase during the study period (Table 1), the increasing trend of annual CH4 emissions in the new rice plots might be caused by other factors of land-use legacy, such as soil Eh.

Soil pH is an index that reflects proton concentration in the soil, which can influence redox reactions. The increase of soil pH in wetlands indicates a significant reduction of soil Eh, especially when strong oxidants (e.g., Fe3+, Mn4+) are present (Reddy and Delaune 2008). In the current study, soil pH significantly increased with the land-use years of the rice fields, which was consistent with CH4 emissions (Table 1). The increasing trend of soil pH with increasing time of the new rice plots indicates that soil Eh had decreased. A lower soil Eh could favor CH4 production (Le Mer and Roger 2001), and thus, CH4 emissions increased with the land-use years of the rice fields, which was consistent with that we found in the current study. CH4 production in flooded soils is also very sensitive to pH (Yan et al. 2005), which has been reported to have an optimum range of 6.7–7.1 (Aulakh et al. 2001). However, Yan et al. (2005) reported that CH4 emissions were much higher in soils with a pH of 5.0–5.5 compared to other soils. These findings suggest that soil pH per se may have little effect on CH4 emissions in the current study.

Unfortunately, we did not measure the variation of soil Eh, CH4 production and oxidation potential directly. To elucidate the effects of land-use legacies on CH4 emissions in the new rice fields, further studies need to be conducted even though several studies have revealed the effects of land-use legacies on soil C and N cycles (McGrath et al. 2001; Zhang and He 2004; Eusufzai et al. 2010; Nishimura et al. 2011). However, our study may provide new clues for understanding CH4 emissions from new double-rice paddies in subtropical regions (Yang et al. 2010; Dong et al. 2011; Shang et al. 2011; Banger et al. 2012). Our findings also suggest that regional GHG inventories should consider not only land use types but also land use history because of the effects of land-use legacies.

The N fertilization effect on CH4 emissions

In the study, N fertilization significantly reduced CH4 emissions in the first year after the establishment of the new rice plots (Table 3). Generally, CH4 emissions from rice fields can be determined by the balance between CH4 production and oxidation (Bodelier and Laanbroek 2004). Krüger et al. (2001) reported that 10–90% of produced CH4 was oxidized in the rhizosphere of rice plants, owing to oxygen diffusion via rice aerenchyma in the rhizosphere (Wassmann and Aulakh 2000). Fertilization-enhanced plant growth can increase the oxygen supply to the rhizosphere through a better developed aerenchyma system, thus stimulating CH4 oxidation (Aulakh et al. 2000). In the present study, N fertilization significantly increased rice biomass by approximately 66–110% in the new rice plots (Table 1), and previous studies have shown that greater rice biomass is highly associated with aerenchyma tissues (Aulakh et al. 2000). This suggests that fertilization could have increased CH4 oxidation in the current study. Furthermore, better plant growth, as indicated by greater biomass, may also provide more organic substrates for methanogens through litter decomposition and root exudation (Jia et al. 2001), resulting in greater CH4 production (Aulakh et al. 2001; Watanabe et al. 1999). When N fertilization-induced CH4 oxidation exceeds its production (Xu and Inubushi 2004), N fertilization decreases the overall CH4 emissions, as measured by CH4 efflux in the current study. Otherwise, N fertilization may increase or have no effect on CH4 emissions.

However, N fertilization had no significant effects on CH4 emissions in the old rice plots and the new rice plots during the second and third year of cultivation even though N fertilization obviously increased rice biomass (Table 1). Moreover, N fertilization-induced CH4 emissions increased with the land-use years of the rice paddy. Theoretically, the effects of N fertilization on CH4 emissions could be influenced by site-specific environmental factors (Dong et al. 2011; Banger et al. 2012), which can lead to large interannual variability. In the study, N fertilization and water management practices were consistent between the old and new rice plots and among different years. Therefore, we examined soil temperature and pH, which could influence CH4 emissions, as suggested by former studies (Lu et al. 2000; Krüger et al. 2001; Conrad 2002; Yan et al. 2005). During the study period, the patterns and variations of soil temperature were similar among the three years (Fig. 1), suggesting that temperature was unlikely to cause the changing effects of N fertilization on annual CH4 emissions. Meanwhile, the changing trend of soil pH with land-use years was synchronous between the rice plots with and without fertilization (Table 1), suggesting that soil pH was also not the cause for the weakening negative effects of N fertilization. The results show that soil temperature and pH had little contribution to the variation of NFE in rice paddies, which are in line with a previous study (Banger et al. 2012).

We found that N fertilization-induced increase of soil NO3 −–N decreased with land-use year in the new rice plots (Fig. 4). The increase of soil NO3 −–N content induced by N fertilization was much larger in the 1-year rice plots than in the older ones, which might be the results of varying soil microbial communities with the land-use years of the new rice plots. Chu et al. (2009) reported that soil ammonia-oxidizing bacteria (AOB) communities had larger population sizes and were more diverse in rice paddies recently converted from upland cultivation compared to consecutive rice paddies. The results suggested that relatively more NH4 +–N from N fertilization was converted to NO3 −–N in the first year of cultivation of the new rice paddy compared to the following two years and the old rice paddy. According to the literature, NO3 −–N can reduce CH4 emissions through competing for H2 with denitrifying bacteria (Klüber and Conrad, 1998) or by inhibiting methanogenesis with toxic denitrification intermediates (Roy and Conrad, 1999). Therefore, the relatively stronger reduction effects of N fertilization on CH4 emissions during the first year of cultivation of the new rice plots could be induced by larger differences in the soil NO3 −–N concentration between the NF and NUF plots due to land-use legacies.

Fig. 4
figure 4

N fertilization-induced soil NO3 −–N increase in rice paddy soils with different land-use years. Vertical bars indicate the standard errors of four replicates

Full size image

The results indicate that the effects of N fertilization on CH4 emissions have been coupled with the effects of land-use legacies, as suggested by Perring et al. (2016), and that land-use legacies can interact with environmental changes such as N addition as well as modulate the carbon cycle. Considering that most studies were carried out in long-term rice paddies, our study may provide new clues for understanding the N fertilization effect on CH4 emissions. However, the coupled effects of land-use legacies and environmental change are complicated (Perring et al. 2016). More studies need to be completed to understand the transforming states of the soil biotic and abiotic processes in the new rice paddies and their interaction with N fertilization.

Conclusions

Field results showed that CH4 emissions from the new rice plots were significantly lower than the old rice plots. There was an increasing trend of annual CH4 emissions with land-use years of cultivation in the new rice paddies. Annual CH4 emissions increased with land-use year of the rice paddies, following the order of 1-year < 2-year < 3-year < 10-year. N fertilization significantly reduced CH4 emissions by 36.9% in the 1-year-old new rice plots, whereas it had no significant effect on CH4 emissions in the older rice plots. These results suggest that land-use legacies can significantly affect CH4 emissions in subtropical regions and may influence the N fertilization effect on CH4 emissions. Land-use legacies and their possible coupled effect with N fertilization on CH4 emissions should be considered in regional GHG inventories.