Responses of CH4 and N2O fluxes to land-use conversion and fertilization in a typical red soil region of southern China

Land-use conversion and fertilization have been widely reported as important management practices affecting CH4 and N2O fluxes; however, few long-term in situ measurements are available after land-use conversion from rice paddies to upland cultivation, especially those including the initial stages after conversion. A 3-year field experiment was conducted in rice paddies and a newly converted citrus orchard to measure CH4 and N2O fluxes in response to land-use conversion and fertilization in a red soil region of southern China. Annual CH4 and N2O emissions averaged 303.9 kg C ha−1 and 3.8 kg N ha−1, respectively, for the rice paddies over three cultivation years. Although annual N2O emissions increased two- to threefold after the conversion of rice paddies to citrus orchard, the substantial reduction in CH4 emissions and even shift into a sink for atmospheric CH4 led to significantly lower CO2-eq emissions of CH4 and N2O in the citrus orchard compared to the rice paddies. Moreover, distinct CH4 emissions were observed during the initial stages and sustained for several weeks after conversion. Our results indicated that the conversion of rice paddies to citrus orchards in this region for higher economic benefits may also lead to lower aggregate CH4 and N2O emissions.

to a sink for atmospheric CH 4 6, 9 . However, when and to what extent can the soil act as an atmospheric CH 4 sink after these land-use conversions still remain unclear. Moreover, most of the existing studies have focused on comparisons of different types of land uses that have been converted for many years [18][19][20][21] . Thus, little information is available for understanding the dynamics of CH 4 and N 2 O fluxes and the underlying mechanisms involved during the initial stages after land-use conversion.
China is one of the most important rice-producing countries in the world, accounting for 20% of the global rice production area 6,22 . The annual totals for CH 4 and N 2 O emissions from Chinese rice paddies were approximately 4.5-7.5 Tg C yr −1 and 32-51 Gg N yr −1 , respectively, based on long-term field measurements and model simulations [22][23][24][25] . Red soil, one of the typical agricultural soils in subtropical China, covers approximately 11.8% of the country's land surface, producing 80% of the rice, and supporting 22.5% of the population of China 26 . During the past decades, the red soil regions, which are the most densely populated, have experienced significant changes in land use due to increased socio-economic development and demand for livestock products. In particular, conversion of rice paddies to upland cultivation for growing vegetables, fruits and economic forest has been locally advocated to meet increasing market demands and gain higher economic returns in these regions 6,18 . Such conversions not only can alter the physical, chemical and biological properties of the soil, but also can impact on the soil C and N turnover and GHG emissions. However, detailed long-term measurements of combined CH 4 and N 2 O fluxes during such conversions are still limited, and the impact of environmental factors and management practices on CH 4 and N 2 O fluxes during such conversions are not fully understood, especially for the red soil regions in China.
Therefore, over a 3-year period, we conducted in situ measurements of CH 4 and N 2 O fluxes from conventional paddy fields and a citrus plantation recently converted from paddy field in a typical hilly, red soil region of southern China. The main objectives of this study were to investigate the response characteristics and temporal changes of CH 4 and N 2 O fluxes during land-use conversion from paddy fields to a citrus plantation and to assess the impact of environmental factors and fertilization on CH 4 and N 2 O fluxes. Eventually, this study also attempted to examine whether the conversion of rice paddies to citrus plantation shows potential for mitigating CH 4 and N 2 O emissions.

Results
Weather conditions and soil properties. The  compared to the long-term site average (1509.0 mm), and more than 70% of the rainfall occurred between March and August (Fig. 1a). Soil temperature showed a temporal pattern similar to that of air temperature for the three consecutive cultivation years (Fig. 1c). Land-use conversion from rice paddy field to a citrus orchard slightly increased the soil temperature but significantly reduced the soil moisture (Fig. 1b). However, fertilization did not significantly affect the soil temperature and moisture for either land-use type during the entire measurement period.
As expected, land-use conversion and fertilization significantly changed the soil properties (Table 1 and Fig. 2). Before land-use conversion (May 2012), all the measured soil properties were not significantly different among different treatments (Table 1). However, the soil bulk density, TN and inorganic N (NH 4 + and NO 3 − ) content substantially increased three years after land-use conversion from rice paddy fields to a citrus orchard (May 2015), whereas the soil pH and DOC values significantly decreased (P < 0.05). Seasonal variations in the soil inorganic N content over the entire observation period were basically regulated by N application (Fig. 2a,b).  Table 1. Main soil properties (0-10 cm) in study sites before and after land conversion. Data are shown as the means with standard errors for four spatial replicates. Different letters in the same column indicate significant differences (P < 0.05) between corresponding treatments. OF = orchard with fertilization, ONF = orchard without fertilization, PF = paddy with fertilization, and PNF = paddy without fertilization. The relatively high NH 4 + and NO 3 − contents were primarily observed within 10 days after fertilization. Overall, fertilization significantly increased (P < 0.05) the NH 4 + and NO 3 − contents for both land-use types. The dynamics of soil DOC concentration were mainly affected by irrigation and fertilization activities, especially for the rice paddy (Fig. 2c). Soil DOC concentrations in the fertilized rice paddy plots were significantly higher compared to the control (P < 0.05); however, the highest soil DOC values in the rice paddy gradually decreased with successive cultivation years (Fig. 2c).  (Fig. 3). In paddy fields, unperceivable N 2 O fluxes were observed during the flooding periods, whereas substantial emissions occurred after fertilization and at the end of the cropping period, when fields dried off and/or were rewetted by rainfall. Fertilizer application significantly and consistently increased the N 2 O emissions from paddy fields during the three cultivation years (P < 0.05), and this stimulating effect of fertilization on N 2 O emissions was enhanced with successive years (Fig. 3a). Over the three cultivation years, the cumulative N 2 O emissions ranged from 3.18 to 6.18 kg N ha −1 in the PF, significantly higher than those in the PNF ( Fig. 5 and Table 2 Land-use conversion from rice paddy to citrus orchard significantly increased the N 2 O emissions during the entire observation period (P < 0.05, Fig. 3b and Table 2). The high emission peaks in the citrus orchard were mainly linked to fertilization and sharp increases in the soil moisture following irrigation or rainfall events. However, the cumulative N 2 O emissions from the citrus orchard gradually decreased with consecutive cultivation years. Over the entire measurement period, the annual cumulative N 2 O emissions ranged from 7.0 to 10.33 kg N ha −1 in the ONF and from 9.91 to 16.25 kg N ha −1 in the OF (Fig. 5 and Table 2). Compared to the control, the application of fertilizer in the citrus orchard significantly increased N 2 O emissions (P < 0.05, Table 2), and the EF d values were variable for the citrus orchard, ranging between 1.29 and 1.99%. However, in contrast to the rice paddy, the stimulating effect of fertilization on N 2 O emissions in the citrus orchard gradually decreased with successive cultivation years ( Table 2). General linear model analysis indicated that the N 2 O emissions were significantly affected by land-use conversion, fertilization and year, as well as by their interactions (P < 0.01, Table 2). During the entire observation period, the N 2 O emissions were significantly positively correlated with the soil NO 3 − content for both land-use types (P < 0.05, Fig. 6a).

CH 4 fluxes.
Generally, CH 4 fluxes from rice paddies were pronounced primarily during the waterlogging stages. Substantial CH 4 emissions were observed during rice-growing seasons, while no pronounced CH 4 emissions or a minor sink were observed during the fallow periods (Fig. 4a). During the rice-growing seasons, CH 4 fluxes increased steadily until the emission peak was attained several weeks after rice transplanting under waterlogging conditions. Thereafter, CH 4 fluxes decreased dramatically with the drying of fields and decreasing soil water content due to mid-season drainage and then remained at low rates until rice harvest. Although fertilization did not substantially alter the seasonal pattern of CH 4 fluxes in rice paddies, it significantly decreased the magnitude of CH 4 emissions (P < 0.05, Fig. 4a and Table 2). Over the three cultivation years, the mean annual cumulative CH 4 emission was 286. 65 kg C ha −1 in the PF, which was an average of 12% lower than that in the PNF ( Fig. 5 and Table 2). During the entire observation period, the CH 4 fluxes from rice paddies were significantly positively correlated with the soil DOC concentrations (P < 0.05, Fig. 6b). Land-use conversion from rice paddy to citrus orchard significantly decreased the CH 4 fluxes during the entire observation period (P < 0.05, Fig. 4b and Table 2). There were no regular and consistent seasonal patterns of CH 4 flux for the citrus orchard during our measurement period. In the newly converted citrus orchard, the soils remained a source of atmospheric CH 4 , which was sustained for several weeks after land-use conversion. Thereafter, CH 4 emissions gradually decreased, and the soils became a weak sink for atmospheric CH 4 in the beginning of October 2012, i.e., approximately 2 months after land-use conversion (Fig. 4b). After that, the citrus orchard was generally a sink for CH 4 , with some sporadic CH 4 emissions during phases with high soil moisture. As a result, cumulative CH 4 fluxes from the citrus orchard gradually increased with consecutive cultivation years. Over the entire observation period, the annual cumulative CH 4 fluxes increased from −0.41 to −1.84 kg C ha −1 in the OF and from −0.29 to −1.7 kg C ha −1 in the ONF ( Fig. 5 and Table 2). Compared to the control, there was no significant influence of fertilization on the mean annual CH 4 fluxes in the citrus orchard. The cumulative CH 4 fluxes over the entire measurement period varied significantly with land-use conversion, fertilization and year, as well as by their interactions (P < 0.01, Table 2). Table 2, the mean annual CO 2 -eq emissions of CH 4 and N 2 O were 15.31 and 15.55 t CO 2 -eq ha −1 for PF and PNF over the three cultivation years, respectively. These values were 2.5 and 3.8 times higher than those from OF and ONF, respectively, indicating that the land-use conversion of rice paddy field to a citrus orchard significantly decreased the net GHG emissions (P < 0.001). The significantly higher CO 2 -eq emissions in rice paddies were mainly due to the substantial CH 4 emissions during rice-growing seasons. The highest CO 2 -eq emissions in the citrus orchard were observed in the first year after land-use conversion, irrespective of fertilization, after which the annual CO 2 -eq emissions gradually decreased with consecutive cultivation years ( Table 2). The application of N fertilizer significantly increased CO 2 -eq emissions in the citrus orchard, mainly due to the substantial increase in N 2 O emissions from the fertilized treatment. However, there was no significant influence of fertilization on the CO 2 -eq emissions in rice paddies, which was largely ascribed to the counteractive effects of stimulated N 2 O emissions and depressed CH 4 emissions due to fertilization in rice paddies. General linear model analysis indicated that the CO 2 -eq emissions were also significantly affected by land-use conversion, fertilization and year, as well as by their interactions (P < 0.01, Table 2).

Discussion
Over the past decades, numerous measurements of CH 4 and N 2 O fluxes have been conducted in rice paddies, documenting paddy fields as significant sources of atmospheric CH 4 and N 2 O 3, 27-29 . However, due to increased demand for livestock products and crop diversification, paddy field-converted upland cultivation systems (e.g., vegetables and orchard) have become increasingly adopted as agricultural systems, especially in the red soil regions of southern China 18,30 . Although significantly lower CH 4 but higher N 2 O emissions were generally observed in upland cultivation compared to rice paddies 9, 31 , most previous studies have focused on the comparison of CH 4 and N 2 O fluxes in different land-use types that have been converted for many years and seldom consider the early stages after land-use conversion [16][17][18][19][31][32][33] . Recently, several studies have suggested that in addition to accounting for GHG fluxes from specific land-use types, GHG dynamics during actual land-use changes should be also considered 34 , and these studies have recommended that further studies should be designed to monitor the entire conversion process 35,36 . Although this is an important issue, only a few studies have carried out     The mean annual CH 4 and N 2 O fluxes from rice paddies in this study were 303.9 kg C ha −1 and 3.8 kg N ha −1 , respectively, over the three cultivation years. These values were within the ranges identified by previous studies in double rice-cropping system with similar fertilization rates 6, 9, 28, 37 , but were much higher than those from other studies in single rice-cropping systems 3,15,38 . The mean annual CH 4 and N 2 O fluxes from the paddy field-converted citrus orchard were −1.19 kg C ha −1 and 10.88 kg N ha −1 , respectively, over the entire study period, which were generally close to previous observations 18,39,40 . However, the annual CH 4 uptake values in the newly converted orchard were significantly lower than those found by Liu et al. 20 , who reported an average annual CH 4 uptake of 2.61 kg C ha −1 y −1 in a pine plantation in a subtropical region of southern China. This discrepancy might be partly attributed to the difference in the length of time since establishment of the orchard plantation, i.e., newly converted versus 12 years old. Many publications have indicated that soil-atmosphere CH 4 exchange can be strongly affected by soil disturbances, such as land-use change and agricultural practices, and that these effects may persist for years to decades 41,42 ; this possibility was confirmed by the gradually increasing CH 4 oxidation capacity after land-use conversion in our study (Fig. 5). In addition, the annual N 2 O emissions from the citrus orchard in this study, especially during the first year after conversion, were generally greater than those from some earlier estimates 20, 39 , probably due to persistent anaerobic conditions during the initial stages after conversion from rice paddy and due to the high amount of basal fertilizer (370 kg N ha −1 y −1 ). However, the annual N 2 O emissions from the citrus orchard gradually decreased with consecutive cultivation years due to changes in the soil environmental conditions and management practices. Therefore, our results suggested that long-term, continuous measurements over several years after land-use conversion are needed to provide reliable estimates of the changes in annual CH 4 and N 2 O fluxes due to land-use change and highlighted the importance of measurements during the initial stages after conversion.
The conversion of rice paddy to citrus orchard significantly reduced CH 4 emissions and changed the soil from an emission source to sink for atmospheric CH 4 over the entire measurement period. Although notable quantities of CH 4 emissions were observed during the initial stages after conversion, the emission rates were significantly lower than those from rice paddies, which were consistent with previous studies showing that CH 4 emissions can occur during non-flood conditions due to anaerobic microsites 9,43 . This reduction in CH 4 emissions after the conversion of rice paddy to citrus orchard can be primarily explained by the shift from anaerobic to aerobic conditions due to improved soil aeration and the regeneration of oxidants, particularly the re-oxidation of Fe(II) 17,31 . Another driving factor for the reduction in CH 4 emissions after conversion is the inhibiting effect of SCIEntIfIC REPORts | 7: 10571 | DOI:10.1038/s41598-017-10806-z aerated conditions on the methanogenic archaeal community. While investigating the abundance of methanogenic archaea at our study site in the years 2013-2014, Liu et al. found significantly lower methanogenic archaea abundance in the citrus orchard compared to rice paddies 44 . This finding is in good agreement with results from previous studies that also reported decreasing numbers of methanogens and reductions in resident and active archaea in drained rice paddies 45,46 . Moreover, significantly higher soil DOC content due to the large quantities of retained crop residues in rice paddies compared to citrus orchards might also contribute to higher CH 4 emissions in paddy fields because available soil organic C is the predominant source of methanogenic substrates 15,37 .
In contrast to decreasing CH 4 emissions, the conversion of rice paddy to citrus orchard resulted in a significant increase in N 2 O emissions in this study. These results are in line with earlier observations showing that the formation of aerobic conditions caused by land-use change can result in reduced CH 4 fluxes at the expense of increasing N 2 O emissions 9,47 . Soil moisture is one of the key factors driving N 2 O emissions from many ecosystems due to its role in the stimulation of microbial activity and in the delivery of electron donors and acceptors, as well as in the diffusion of gases in soil [3][4][5][14][15][16] . Although land-use conversion from rice paddy to citrus orchard significantly reduced soil moisture in this study, the strict anaerobic conditions during flooding periods in paddy fields might favour reduction of N 2 O to N 2 through denitrification processes, thus leading to lower N 2 O emissions 4,5,30 . In our study, the N 2 O emissions from rice paddies during the flooding periods were generally low. Substantial emissions only occurred only during periods within several weeks following fertilization and the drying of fields. In addition, the significantly lower soil NO 3 − content and limited nitrate availability under anaerobic conditions might contribute to the lower N 2 O emissions in rice paddies compared to those in citrus orchards, since strong positive correlations between N 2 O emissions and soil NO 3 − content were observed for both land-use types in this study (Fig. 6a). Furthermore, the increased abundance of ammonia-oxidizing archaea in upland cultivations compared to rice paddies could further explain the increasing N 2 O emissions due to enhancement of nitrification processes 46 .
The utilization of synthetic N fertilizers is usually considered an important regulator of CH 4 and N 2 O fluxes in agriculture fields [5][6][7][8]28 . In this study, the application of N fertilizer generally resulted in a suppression of annual CH 4 emissions from rice paddies over the 3-year measurement period, which was in accordance with some earlier observations 6,8,43 . However, previous studies on the effect of synthetic fertilizers on CH 4 emissions from rice paddies are inconsistent. Either increased CH 4 emissions or no significant change in emissions due to fertilization from paddy fields has also been reported in some other studies 28,30,48 . In paddy fields, the application of N fertilizer, especially ammonium-based fertilizers, has been found to promote the growth and activity of methane-oxidizing bacteria, especially in soil around rice roots 49 , thus resulting in increased consumption of CH 4 . Moreover, for red soils with sandy loam texture, as in the present study, CH 4 oxidation under urea-based fertilization is likely further simulated by partially aerobic soil conditions due to the porous and percolating nature of soil 8 . In contrast to rice paddies, no significant effect of fertilization on CH 4 fluxes was observed in the citrus orchard, which was in agreement with previous studies of upland cultivation areas 30,50,51 , probably because both CH 4 production and oxidation are simultaneously affected by N fertilization 17,50 . Consistent with numerous previous studies conducted in paddy fields and upland orchard 8,28,39 , N 2 O emissions were significantly enhanced by fertilization in both land-use types in this study. These were mainly due to the fact that fertilizer application can markedly increase the soil inorganic content, as also shown in our study (Table 1), thereby providing sufficient substrate for microbial nitrification and denitrification for the production of N 2 O 7-9, 18 . The emission factors for N 2 O were estimated to be 0.47-0.96% and 1.29-1.99% for the rice paddy and citrus orchard, respectively. These results are comparable to previous estimates from paddy fields 5,8,25 and upland orchard 6,18,39 . However, large discrepancies in observed EF d for N 2 O in upland orchard, ranging from 0.2-2.2%, have been observed in earlier publications 18,20,39,52 likely due to the relatively short-term measurements and course sampling intervals, as well as the differences in the usage of the "baseline 39 ", i.e., background emission.
The average annual CO 2 -eq emissions of CH 4 and N 2 O was 15.43 t CO 2 -eq ha −1 for rice paddies over the three cultivation years, similar to values reported in previous studies conducted in the same regions 3,8,28,30 . These values are also within the range of 75-22,237 kg CO 2 -eq ha −1 for rice paddies reported by Linquist et al. 53 , who estimated aggregate emissions of CH 4 and N 2 O by collecting 328 measurements globally. However, the annual CO 2 -eq emissions were significantly reduced following the conversion of rice paddies to a citrus orchard ( Table 2), indicating that the effect of significantly reduced CH 4 emissions was only marginally offset by the simultaneously increased N 2 O emissions after land-use conversion. In general, the economic benefits from upland orchard and vegetables were higher than those from rice paddies in our study region 6,30 . Therefore, lower climate impacts but higher economic incomes can be achieved synchronously by the conversion of rice paddies to citrus orchards in this region. Moreover, the application of fertilizer had no significant effect on the CO 2 -eq emissions in rice paddies, which is consistent with the hypothesis of Zou et al. 25 , who surmised that fertilization generally depresses or dose not influence aggregate emissions of CH 4 and N 2 O from rice paddies, depending on the fertilizer application rate. However, fertilization resulted in significantly higher CO 2 -eq emissions in the citrus orchard, which can be explained by the substantial increase in N 2 O emissions due to higher input of mineral N. It is noteworthy that our analysis of CO 2 -eq emissions did not include the net exchange of CO 2 between agroecosystems and the atmosphere or changes in soil organic carbon. Results from the literature indicated that rice paddies are usually found to be a weak atmospheric CO 2 sink 47 and that carbon sink strength is typically lower than that of mature orchards 54 . Meanwhile, numerous studies have reported that changes in soil organic C are difficult to detect because the magnitude of change is small during several years and because there is a high degree of spatial variation 9,11 . Therefore, more long-term studies including measurements of the climatically important C-and N-trace gas fluxes (CO 2 , N 2 O and CH 4 ) and estimates of changes in soil organic C are needed to provide a complete evaluation of the overall GHG balance during land-use conversion.
In conclusion, this study provided insight into the integrated evaluation of CH 4 and N 2 O fluxes and their relationships with management practices following the conversion of rice paddies to a citrus orchard over three consecutive cultivation years in a typical red soil region of southern China. Our results not only confirmed that land-use conversion from rice paddy to citrus orchard significantly decreased CH 4 emissions and increased N 2 O emissions, but also demonstrated that the citrus orchard could persist as a source for atmospheric CH 4 for several weeks after conversion from paddy fields and then gradually shift to a CH 4 sink with increasing oxidation capacity over the three cultivation years. Thus, our results highlighted the importance of measurements during the initial stages following land-use conversion and suggested additional long-term continuous observations over several years after conversion. The substantial changes in CH 4 and N 2 O fluxes following land-use conversion were mainly due to significant alterations in soil environmental conditions (i.e., shifting from anaerobic to aerobic) and soil properties originating from the remarkably different management practices between the two land-use types. Moreover, fertilization significantly increased N 2 O emissions from both land-use types but substantially reduced CH 4 emissions from the rice paddies and had no significant effect on CH 4 fluxes in the citrus orchard. As a result, the CO 2 -eq emissions of CH 4 and N 2 O were significantly reduced following the conversion of rice paddies to citrus orchard, irrespective of fertilization. Overall, reduced aggregate CH 4 and N 2 O emissions and higher economic benefits can be achieved simultaneously by the conversion of rice paddies to citrus orchards in this red soil region.

Materials and Methods
Study site. The experimental site is located at the Qianyanzhou Ecological Research Station (26°44′N, 115°04′E) of the Chinese Academy of Science in Jiangxi Province, southern China. This site has a subtropical monsoon climate with a mean annual air temperature of 18.0 °C and a mean annual precipitation of approximately 1509.0 mm during 1989-2010 6 . The soil type is typical red soil found in middle-subtropical China, classified as Ultisols and some of the Alfisols and Oxisols based on soil taxonomy of the USA 18 . The soil texture is sandy loam, with 58% sand, 31% silt, and 11% clay. Double cropping of rice paddy is the main cropping system in this region, with late rice (late July to late November), a fallow period (late November to next late April), and early rice (late April to late July) in rotation. Other soil properties both before and 3 years after land-use conversion are shown in Table 1.
Experimental design. The two most prevalent agricultural land uses in our study area were selected, namely, rice paddy (Oryza sativa L.) and citrus orchard (Citrus reticulata). The experimental site was previously paddy fields for more than 10 years and had been partly converted to a citrus orchard in June 2012. Under each land-use type, two fertilizer treatments (i.e., conventional fertilization and no fertilization) were established. The conventional fertilization treatment followed the local cropping regimes and farmer fertilization practices. The fertilizers used were compound fertilizer (15% N) and urea (46% N). The other treatment was a control without fertilization, with additional management practices being the same as for the fertilization treatment. Therefore, four treatments were established: citrus orchard with fertilization (OF) and without fertilization (ONF) and rice paddy with fertilization (PF) and without fertilization (PNF). All treatments were arranged in a randomized block design with four replicates, for a total of 16 experimental plots (12 × 14 m) that were separated by buffer strips. In the PF, compound fertilizer was applied before rice transplanting, and urea was applied in the form of top dressing at the tillering stage, whereas in the OF, compound fertilizer combined with urea was uniformly spread on the soil surface. To ensure survival and yield, a floodwater layer of 5-7 cm in depth was maintained in the paddy fields until mid-season drainage, and basal fertilizer was amended to a depth of 50 cm before the citrus orchard was established. Details of the cultivation and fertilization practices during the study period are shown in Table S1. CH 4 and N 2 O flux measurement. Fluxes of CH 4 and N 2 O were simultaneously measured from June 2012 until July 2015 using a static opaque chamber-gas chromatograph (GC) method as described in Yao et al. 8 . and Zheng et al. 55 . A stainless steel collar (diameter = 40 cm) was pre-installed in the centre of each plot before rice transplanting or orchard planting. The top edge of the collar contains a groove (5 cm in depth) filled with water to seal the rim of a chamber during gas collection. Cylindrical sampling chambers with a diameter of 40 cm and height of 0.39 or 0.69 m (according to the plant height) were covered with a layer of thermal insulation to minimize air temperature changes inside the chamber and equipped with a circulating fan to ensure complete gas mixing during the gas sampling period. The base frames were kept in the same location throughout the entire measurement period in the orchard plots, whereas those in the paddy fields were removed before tillage and placed (24 h before the measurement) in the location marked for subsequent measurements.
Gas samples were taken daily for 5 days after fertilization and once or twice per week for the remaining period and were collected between 09:00-11:00 AM local time. Five air samples were taken from the headspace of each chamber at an interval of 10 min after chamber closure using plastic syringes attached to a three-way stopcock and were stored at room temperature for analysis within a few hours. The chamber headspace temperature was recorded for gas density correction in the flux calculation using a thermometer. The concentrations of CH 4 and N 2 O were determined by a gas chromatography (Agilent 7890 A, California, USA) equipped with a flame ionization detector (FID) and an electron capture detector (ECD), respectively, as detailed in previous studies 6,55,56 . The fluxes were calculated based on the slope of linear or nonlinear regression between concentration and time and were determined as the mean of the four fluxes from the four spatial replications. Annual cumulative CH 4 and N 2 O fluxes were sequentially computed from the fluxes between every two adjacent intervals of measurements and were estimated by linear interpolation 6,8 . The fertilizer-induced direct emission factors (EF d ) for N 2 O were calculated by subtracting the total cumulative emissions of N 2 O in the control treatments from the corresponding cumulative emissions in the fertilized treatments and dividing the result by the fertilizer application rate 5 . For calculating the GHG balance, annual CH 4 and N 2 O fluxes were converted into CO 2 equivalents, taking into account the specific radiative forcing potential of 298 for N 2 O and 34 for CH 4 relative to CO 2 for a 100-year time horizon 1 .
SCIEntIfIC REPORts | 7: 10571 | DOI:10.1038/s41598-017-10806-z Auxiliary measurements. Daily precipitation and air temperature were obtained from the Qianyanzhou meteorological station. The soil temperature and moisture (0-10 cm) for each plot were measured using a potable digital thermometer (JM624, Tianjin, China) and a moisture probe meter (TDR100, Spectrum, USA), respectively. Soil water-filled pore space (WFPS) was calculated from the bulk density (BD) and volumetric soil water content using a particle density of 2.65 g cm −3 . Floodwater depths in the paddy were measured daily during the flooding period using a ruler. Soil samples (0-10 cm) were collected prior to land-use conversion to determine background information and once per month or every two months between June 2012 and July 2015 for physiochemical property measurements. Soil pH was measured at a soil:water ratio of 1:2.5 using a pH meter. Soil total nitrogen (TN) and dissolved organic carbon (DOC) were determined using an automated C and N analyzer (Elementar, Hanau, Germany). Soil ammonium (NH 4 + −N) and nitrate (NO 3 -−N) were extracted from 20 g of fresh soil with 1 M KCl (soil:water = 1:5 w/v) and quantified colourimetrically using a flow injection analyzer (Seal AA3, Norderstedt, Germany).

Statistical analyses.
All data are presented as the mean and standard error of mean unless otherwise stated.
Analysis of variance (ANOVA) was used to examine differences in soil properties among the four treatments. The impacts of land-use conversion, fertilization and year on CH 4 and N 2 O fluxes were conducted using general linear models for analysis of variance together with the least significant difference test. The relationships between trace gas fluxes and soil properties were evaluated using a nonlinear regression model. SPSS 20.0 statistical software (IBM Co., New York, USA) was used to conduct statistical analyses. The figures were prepared using Origin 8.5 software (Origin Lab Corporation, USA).