This is the accepted version of the article: Wang, Weiqi; Wang, Chun; Sardans i Galobart, Jordi; [et al.]. «Multiple trade- offs between maximizing yield and minimizing greenhouse gas production in Chinese rice croplands». Land degradation & development, Vol. 31, Issue 10 (June 2020), p. 1287-1299. DOI 10.1002/ldr.3507 This version is avaible at https://ddd.uab.cat/record/222641 under the terms of the license 1 MULTIPLE TRADEOFFS BETWEEN MAXIMIZING YIELD AND 2 MINIMIZING GREENHOUSE GAS PRODUCTION IN CHINESE RICE 3 CROPLANDS 4 5 Weiqi Wang1,2,3,4*#, Chun Wang1,2#, Jordi Sardans3,4*, Yunying Fang5, Bhupinder Pal 6 Singh5, Huaru Wang6, Xiaoting Huang1,2, Congsheng Zeng1,2, Chuan Tong1,2, Josep 7 Peñuelas3,4 1 8 Key Laboratory of Humid Subtropical Eco-geographical Process, Ministry of Education, 9 Fujian Normal University, Fuzhou 350007, China 2 10 Institute of Geography, Fujian Normal University, Fuzhou 350007, China 3 11 CSIC, Global Ecology Unit CREAF-CEAB-CSIC-UAB. 08913 Cerdanyola del Vallès. 12 Catalonia, Spain 4 13 CREAF. 08913 Cerdanyola del Vallès. Catalonia, Spain 5 14 NSW Department of Primary Industries, Elizabeth Macarthur Agricultural Institute, 15 Menangle, NSW 2568, Australia 6 16 Yellow River Institute of Hydraulic Research, Zhengzhou 450003, China 17 18 Corresponding author: W. Wang, Key Laboratory of Humid Subtropical Eco- 19 geographical Process, Ministry of Education, Fujian Normal University, Fuzhou 350007, 20 China, Institute of Geography, Fujian Normal University, Fuzhou 350007, China. Email: 21 wangweiqi15@163.com; 22 J. Sardans, CSIC, Global Ecology Unit CREAF-CSIC-UAB, 08913 Cerdanyola del 23 Vallès, Catalonia, Spain, CREAF, 08913 Cerdanyola del Vallès, Catalonia, Spain. Email: 24 j.sardans@creaf.uab.es 25 # Equal contribution 1 26 Running title: 27 Tradeoffs between yield and greenhouse gas production 28 2 29 ABSTRACT 30 Globally, paddy fields are a major anthropogenic source of greenhouse gas (GHG) 31 emissions from agriculture. There is, however, limited understanding of relationships 32 between GHG production with fertilizer management, rice varieties, and soil variables. 33 This information is crucial for minimizing the climatic impacts of rice agriculture. Here, 34 we examined the relationships between soil GHG production and management practices 35 throughout China. The current doses of N-fertilizer (73-272 kg ha-1) were negatively 36 correlated with rice yield and with CO2 or CH4 production, and positively correlated with 37 N2O production, thus suggesting N-overfertilization. Impacts on soil traits such as 38 decreasing pH or the availabilities of other nutrients could be underlying these 39 relationships. Rice yield was highest and GHG production was lowest at sites using 40 intermediate levels of P- and K-fertilization. CO2 and CH4 production and emissions were 41 positively related with soil water content. The yield was higher and N2O productions were 42 lower at the sites with japonica rice. Our results strongly suggest that current high doses 43 of N-fertilizers could be reduced to thus avoid the negative effects of excessive N input 44 on GHG production without any immediate risk of rice production loss. Current 45 intermediate doses of P- and K-fertilization should be adopted across China to further 46 improve rice production without the risk of GHG emissions. The use of different rice 47 varieties and strategies of water management should be re-examined in relation to crop 48 production and GHG mitigation. 49 KEYWORDS: paddy field, greenhouse gases, yields, nitrogen, phosphorus, soil 50 nutrients 3 51 1. INTRODUCTION 52 Rice currently feeds more than 50% of the global population (Haque, Kim, Ali & Kim, 53 2015), but production will need to increase by 40% by the end of 2030 worldwide to meet 54 the demand for food from the growing population (FAO, 2009). Sustaining soil fertility 55 and increasing rice yields are therefore of utmost importance. An increased nutrient 56 supply can stimulate the growth and grain yield of rice plants (Ali, Oh & Kim, 2008; 57 Wang et al., 2014) but can also influence the potential of paddy fields to produce and emit 58 greenhouse gases (GHG). Globally, paddy fields are a major anthropogenic source of 59 greenhouse gas (GHG) emissions from agriculture (Tan, 2011). Paddy fields are very 60 important sources of GHG, especially methane (CH4) and nitrous oxide (N2O) (Myhre et 61 al., 2013), so minimizing the release of these very potent GHG could contribute to 62 mitigating their adverse impacts on climate change (Li et al., 2006). Chinese GHG 63 emissions from agricultural systems account for ~ 40% of Chinese GHG emissions, hence 64 requiring detailed investigations. Furthermore, since sixty percent of the Chinese 65 population depends on rice-based food, so protecting China's rice production for food 66 security is important (Zhu, 2006). 67 There is, however, limited understanding of relationships between GHG production 68 with fertilizer management, rice varieties, and soil variables. This information is crucial 69 for minimizing the climatic impacts of rice agriculture, especially for the agricultural 70 sustainable development in China. Improving the status of soil nutrients in paddy fields 71 for improved rice yield while decreasing GHG emissions, or at least not increasing it, is 72 a challenging option. However, intense fertilization can also induce rises in GHG 73 emissions in paddy soils (Fan et al., 2016), and great soil nutrient concentration also can 74 be related to GHG emissions in paddy soils (Wang et al., 2017a). Rice crops in China are 75 frequently overfertilized (Cheng &Li, 2007) so a general analysis of the relationships of 4 76 fertilization and soil traits with GHG emissions and crop yield is necessary to detect the 77 level of over-fertilization, GHG emissions and yield, and moreover, to improve the future 78 management of the sustainable development of rice agriculture. Here, we examined the 79 relationships between soil GHG production and management practices throughout China. 80 Most current studies associating GHG emissions from paddy fields with soil 81 properties have been conducted at a single location by applying different treatments that 82 modify soil properties (Wassmann et al., 1998; Wang et al., 2014). Several strategies for 83 managing rice crops, such as from fertilizer and herbicide application to straw or water 84 management, are aiming to increase rice production. However, such management 85 strategies may also increase or decrease CO2, CH4 and/or N2O emissions (Li et al., 2005, 86 2013; Liu et al., 2016; Jiang, Chen, Sun, Song & Huang, 2015; Launio, Asis, Manalili & 87 Javier, 2016; Trinh et al., 2017; Wang et al., 2016c; Zhang et al., 2016a,b; Jiang, Chen, 88 Sun, Song & Huang, 2015; Launio, Asis, Manalili & Javier, 2016; Trinh et al., 2017; 89 Wang et al., 2016c; Zhang et al., 2016a,b). Many of these studies have reported links 90 between GHG emissions and various soil traits, such as pH (Wang et al., 2017a), redox 91 potential, (Fan et al., 2016; Wang et al., 2017b), salinity (Olsson et al., 2015), sulfate 92 concentration (Dong et al., 2011; Theint, Susuki, Ono & Bellingrath-Kimura, 2014; Wang 93 et al., 2017b), N content (Zhu, Zhang & Cai, 2011; Zheng, Zhang & Cai, 2013; Zhao et 94 al., 2015; Wang et al., 2017c) and soil P concentration (Adhya, Pattnaik, Satpathy, 95 Kumaraswamy & Sethunathan, 1998; Zheng, Zhang & Cai, 2013; Sheng et al., 2016). 96 However, only few studies have tested the differences in GHG productions and emissions 97 at sites with different soil traits and management strategies, including fertilization. 98 Studying soil conditions on a large scale, including coastal and inland paddy soils at large 99 regional scales, and their relationships with GHG productions and emissions are thus 100 warranted for increasing rice production while controlling GHG emissions. 5 101 We hypothesized that different fertilization practices, strategies of rice-crop 102 management, rice varieties, concentrations of soil carbon (C) and other nutrients, salinity 103 or/and pH would explain a large part of the differences in GHG emissions and yield 104 among sites, especially, the fertilization amount increment will increase the GHG 105 emissions but not the yield. Our results will provide information for improving strategies 106 and managing soil conditions toward more favorable traits to avoid a possible increase in 107 GHG production without yield loss. We pursued this objective by determining (1) the 108 relationships among fertilization dose, soil GHG production and rice yield in China, (2) 109 the relationships among GHG production, rice variety and environmental traits, such as 110 region (sites), location (coastal vs. inland) and cropping system (single or double) on these 111 relationships. 112 2. MATERIALS AND METHODS 113 2.1 Study area 114 This study was conducted throughout China (Figure 1). China has 2.45×107 ha of 115 cultivated rice, and 90% of the paddies are in the subtropics. China has a large area of 116 paddy rice, diverse soil types and different tillage and fertilizer management practices, all 117 of which may affect the content and distribution of nutrients and GHG emissions. We 118 collected the soils in the whole China rice cultivation areas choosing sites with contrasting 119 fertilization management. The studied ranges of fertilization were 73 -272 kg ha-1 for N 120 fertilizer, 48-150 kg ha-1 for P2O5, and 45-270 kg ha-1 for K2O. The main characteristics 121 of the different studied sites are showed in Table S1. To analyze the role of sea proximity 122 we separated the studied sites between inland and coastal rice crops depending on the 123 distance to sea line; when less than 20 km was considered coastal paddy field and more 124 than 20 km was considered inland paddy field. To analyze the role of fertilization intensity 125 we classified the distinct sites as low, intermediate and high intensity fertilized. In the 6 126 case of N-fertilization: Low <100 kg N ha-1 y-1, intermediate between 100-150 kg N ha-1 127 y-1 and high intensity >150 kg N ha-1 y-1. In the case of P-fertilization: Low<60 kg P2O5 128 ha-1 y-1 , intermediate between 60-75 kg P2O5 ha-1 y-1and high intensity>75 kg P2O5 ha-1 129 y-1. In the case of K-fertilization: Low <60 kg K2O ha-1 y-1, intermediate between 60-70 130 kg K2O ha-1 y-1 and high intensity >70 kg K2O ha-1 y-1. 131 132 2.2 Collection and measurement of soil samples 133 Soil samples were collected from 34 randomly selected paddy fields and five replicated 134 plots throughout China using a small core sampler (length and diameter of 0.5 and 0.1 m, 135 respectively) from the plowed (0-15 cm) soil layer in October to November 2015 (Figure 136 1). A total of 170 samples (34 sites × 1 soil layer × 5 replicates) were thus collected. We 137 attributed China into six regions for this study: Northeast China, North China, East China, 138 Center and South China, Southwest China and Northwest China (Sun, Huang, Zhang & 139 Yu, 2010). Moreover, the sampling site characteristics and paddy field management were 140 also investigated. The collected soil samples were also used for all analyses. The samples 141 were air-dried, and roots and visible plant debris were removed. Total C and N contents 142 were measured using a Vario EL III Elemental Analyzer (Elementar Scientific 143 Instruments, Hanau, Germany, Wang et al., 2015a; Wang et al., 2016a,b). Labile organic- 144 carbon (LOC) content was determined by digestion with 333 mM KMnO4 (Wang, Lai, 145 Wang, Oan & Zheng, 2015b), while NH4+ and NO3- were determined by extracting the 146 soils with 2 M KCl (Lu, 1999). Available N content was the sum of NH4+ and NO3-. The 147 total P content was determined, first by perchloric-acid digestion and then using a 148 sequence flow analyzer (San++, SKALAR Corporation production, Breda, The 149 Netherlands). To determine soil P availability, we used the Mehlich extraction method 150 and then using a sequence flow analyzer (San++, SKALAR Corporation production, 7 151 Breda, The Netherlands). 152 The soil CO2, CH4 and N2O productions were determined using anaerobic 153 incubation consistent with soils under water saturation (Wang et al., 2015a). Thirty grams 154 of fresh soil of the five core samples for each site were placed in 120-mL incubation 155 bottles, and two volumes of distilled water were added. The bottles were purged with N2 156 for 2 min to replace the O2 and were then sealed with a rubber stopper and incubated at 157 25 °C for 3 d. Five milliliters of gas were extracted from the headspaces each day, about 158 24 h interval in four times during incubation: 0, 24, 48 and 72 hours during incubation 159 experiments. This method has been successfully used in several previous studies 160 (Wassmann et al., 1998; Wang et al., 2017b). We also use the method, of Xu et al., 2016 161 to calculate annual gas emissions from some days (in our case 3 days) of sampling (Xu et 162 al., 2016). 163 The soil CO2 and CH4 concentrations in the samples of headspace air were 164 determined by gas chromatography (Shimadzu GC-2010, Kyoto, Japan). The soil N2O 165 concentrations in these samples were determined by gas chromatography (Shimadzu GC- 166 2014, Kyoto, Japan) using a stainless-steel Porapak Q column (2 m length, 4 mm OD, 167 80/100 mesh). A methane conversion furnace, flame ionization detector (FID) and 168 electron capture detector (ECD) were used for determining the CO2, CH4 and N2O 169 concentrations, respectively. The operating temperatures of the column, injector and 170 detector were adjusted to 45, 100 and 280 °C, respectively, for determining the CO2 171 concentrations; to 70, 200 and 200 °C, respectively, for determining the CH4 172 concentrations; and to 70, 200 and 320 °C, respectively, for determining the N2O 173 concentrations. Helium (99.999% purity) was used as a carrier gas (30 mL min-1), and a 174 make-up gas (95% argon and 5% CH4) was used for the ECD. The gas chromatograph 175 was calibrated before and after each set of measurements using 503, 1030 and 2980 µL 8 176 CO2 L-1 in He; 1.01, 7.99 and 50.5 µL CH4 L-1 in He and 0.2, 0.6 and 1.0 µL N2O L-1 in 177 He (CRM/RM information center of China) as primary standards (Wang et al., 2015c, d). 178 We used linear equations for calculating CO2, CH4 and N2O productions. 179 Other soil variables were also analyzed. Bulk density was measured using three 15 × 180 3 cm cores (Wang et al., 2016b), and was estimated by core mass dry weight divided by 181 core volume, and represent the averaged bulk density of 0-15 cm. Soil water content was 182 measured by the drying method (Lu, 1999). pH was measured with a PHS-3C pH meter 183 (Orion Scientific Instruments, Minnesota, USA) and salinity was measured using a 184 2265FS EC Meter (Spectrum Technologies Inc., Paxinos, USA; Wang et al., 2016b). Soil 185 particle size (percent clay, silt and sand contents) was measured by a Mastersizer 2000 186 laser particle-size analyzer (Malvern Scientific Instruments, Malvern, UK; Wang et al., 187 2016b). 188 189 2.3 Determination of soil C and nutrient contents 190 191 The total C, N and P contents, and labile organic carbon (LOC), available N, available P, 192 NH4+-N and NO3--N contents in the 0-15 cm soil profile were estimated by following the 193 approach of Mishra, Ussiri & Lal (2010): 194 CS = ∑ Cm × ρb × D 195 196 where CS is the total C, N or P content or LOC, available N, available P, NH4+-N or NO3- 197 -N content (kg m-2), Cm is the total C, N or P content or LOC, available N, available P, 198 NH4+-N or NO3--N content (g kg-1), ρb is the bulk density (kg m-3), D is the thickness of 199 each soil layer (0.15 m). 200 The total C, N and P contents, and LOC, available N, available P, NH4+-N and NO3- 9 201 -N contents were calculated for each region by multiplying area content and areas of each 202 paddy distribution. The contents for each measured variables across China were 203 calculated as the sum for all regions. 204 205 2.4 Determination of soil CO2, CH4 and N2O productions 206 We used linear regressions for calculating soil CO2, CH4 and N2O productions 207 (Wassmann et al., 1998): dc VH MW Tst 208 P= ⋅ ⋅ ⋅ dt WS MV Tst + T 209 where P is the rate of CO2, CH4 or N2O production (µg-1 g-1 d-1), dc/dt is the recorded 210 change in the mixing ratio of CO2, CH4 and N2O in the headspace over time (mmol mol- 1 211 d-1), VH is the volume of the headspace (L), Ws is the dry weight of the soil (g), MW is 212 the molecular weight of CO2, CH4 or N2O (g), MV is the molecular volume (L), T is the 213 temperature (K) and Tst is the standard temperature (K). 214 The potential productions of CO2, CH4 and N2O, calculated as the average 215 productions per unit area per year [productions rate × soil depth × bulk density × days of 216 the year (365 d in 2015)]. The productions of CO2, CH4 and N2O were calculated for each 217 region as average production per unit area per year × area of the paddy fields in the 218 production region. The GHG productions for all of China was calculated as the sum of 219 productions form all paddy regions. 220 221 2.5 Statistical analyses 222 One-way ANOVAs with N-, P- and K-fertilization doses categorical (low, intermediate 223 and high intensity), regions, inland-coastal environments and cropping systems as 224 independent variables and gas production variables, and yield as a dependent continuous 10 225 variable were conducted with Bonferroni post hoc tests. The data were checked for 226 normality and homogeneity of variance, and if necessary, were log-transformed. We used 227 the Benjamini–Hochberg procedure to control the rate of false discovery (Benjamini and 228 Hochberg 1995) to analyze the relationships between gas emissions and all the studied 229 environmental soil and climate variables. The different individual analyses were listed in 230 rank order according with their ascending P-value. Thereafter, for each single analyses 231 each P-value was divided by the total number of test and thereafter multiplied by the false 232 discovery rate (habitually 0.25) then the values below 0.05 were considered as significant. 233 We also used Tukey’s method (Tukey, 1977) to detect and remove outliers. 234 We also performed multivariate statistical analyses. We used principal component 235 analyses (PCAs) to determine the overall differences of the soil variables and CO2, CH4 236 and N2O productions rates, and annual accumulated emissions among fertilizer doses. We 237 used all variables given the scarce multicollinearity existing among variables (see Table 238 S1), with no R2>0.6 between any pairwise variables. We have also estimated VIF for each 239 independent fixed variables in the mixed models. We conducted one-way ANOVAs with 240 Bonferroni post hoc tests of the scores of the first PC axis to determine differences among 241 the treatments. We then used general discriminant analyses (GDAs) to determine the 242 overall differences of soil traits; CO2, CH4 and N2O productions rates and annual 243 accumulated emissions and yield among fertilizer doses, and productivities among sites 244 with different yields. Discriminant analyses consist of a supervised statistical algorithm 245 that derives an optimal separation between groups established a priori by maximizing 246 between-group variance while minimizing within-group variance (Raamsdonk et al., 247 2001). GDA is thus an appropriate tool for identifying the variables most responsible for 248 differences among groups. The GDAs and PCAs were performed using Statistica 8.0 249 (StatSoft, Inc., Tulsa, USA). Before conducting these multivariate analyses, we selected 11 250 the sampling adequacy of individuals and the set of variables by the Barlett’s test of 251 spherity (<0.05) and the Kaiser-Meyer-Olkin measure (>0.50). We removed the variables 252 with communality values < 0.5 and perform the PCA and DGA analyses with the 253 variables with communality > 0.5. To perform these sampling adequacy analyses we used 254 the package psych (Revelle, 2010). 255 Significant differences in CO2, CH4 and N2O productions among does of fertilizers, 256 number of crops, crop location (inland vs. coastal) and soil traits were tested by general 257 mixed models using location with topography, site and plot as random nested factors. We 258 used the “lme” function of the “nlme” R package (Pinheiro, Bates, DebRoy, Sakar & 259 Core, 2016). Non-normally distributed variables were log-transformed. We chose the best 260 model for each dependent variable using the Akaike information criterion (AIC). We used 261 the MuMIn (Barton, 2012) R package in the mixed models to estimate the percentage of 262 the variance explained by the model. We conducted Tukey’s post hoc tests to detect 263 significant differences in the analyses for more than two communities using the 264 “multcomp” (Hothorn, Pretz & Wersfall, 2013) R package with the “glht” function. The 265 relationships of each soil variable with CO2, CH4 and N2O productions were determined 266 by simple regressions using Statistica 6.0 (StatSoft, Inc. Tulsa, USA). We used the 267 Benjamini–Hochberg procedure to control for rates of false discovery (Benjamini & 268 Hochberg, 1995). 269 We used structural equation modeling (SEM) to study the total effects of fertilizer 270 doses on accumulated gas emissions by both direct and indirect effects of soil traits. We 271 fit the different models using the SEM R package (Fox, Nie & Byrnes, 2013) and 272 determined the minimum adequate model using AIC. Standard errors and significance 273 levels (P values) of the total, direct and indirect effects were calculated using 274 bootstrapping (1200 repetitions) (Davison, Hinkley & Schechtman, 1986; Mitchell-Olds, 12 275 1986). 276 277 3. RESULTS 278 3.1 Effects of fertilization doses, regions, rice variety, inland-coastal environments and 279 cropping systems on soil GHG productions and rice yield 280 One-way ANOVAs with N-, P- and K-fertilization doses as independent categorical (low, 281 intermediate and high intensity) variables and gas production variables, and yield as a 282 dependent continuous variable, indicated that the lowest levels of N fertilization (<100 283 kg N ha-1 y-1) were associated with the highest annual yields. However, the highest levels 284 of N fertilization (>150 kg N ha-1 y-1) were associated with the lowest annual yields 285 (Figures 2d and S1). The lowest doses of N fertilizer were associated with the highest soil 286 CO2 and CH4 productions, and the lowest N2O productions, with the opposite patterns at 287 the highest doses of N fertilizer. The intermediate doses of P (60-75 kg P2O5 ha-1 y-1) and 288 K (60-70 kg K2O ha-1 y-1) fertilizers were the best, because they were associated with the 289 highest yields and the lowest CO2 and CH4 productions (Figures 3a,b, S2, 4a,b and S3). 290 The correlations among all the studied variables were shown in Table S2. 291 The balance between yield and GHG production was worst in the East China rice 292 crops, with the highest productions of CO2, CH4 and N2O and the lowest annual yield 293 (Figures 5, and S4). Coastal rice crops have higher CO2 and CH4 production (Figure 294 6a,b, S5a,b) and lower yield than inland rice crops (Figure 6d). Total annual yield was 295 notably similar at the sites with one and two annual rice crops (Figure 7d). The CO2 296 production rates were higher at sites with one annual crop than at sites with two annual 297 crops (Figure S6a,c). 298 We excluded hybrid and glutinous rice varieties from the analysis of the effects of 299 rice variety, because both were only at one site each. We focused on japonica and Hsien 13 300 rice varieties, which were the main varieties planted at the sites. On average, the yield 301 was higher and N2O productions were lower at the sites with japonica rice (Figure S7). 302 303 3.2 Gas productions and environmental traits 304 We observed scarce relationships with studied gas emissions and production with the 305 studied soil variables, the most strong was the positive relationships between total soil N 306 concentrations and CO2 emission and production and between soil water content and 307 methane emission and production (Table S2). The variables that mainly loaded on PC1 308 (correlation >0.4) were Soil N concentration, LOC soil extractable NH4+ and NO3- 309 concentrations, soil pH, soil salinity, soil bulk density, sand %, MAP, soil NO3- content, 310 CO2 emission rates and yield. The variables that mainly loaded PC1 were soil total C, N 311 and P concentrations, soil P availability, soil NO3- concentrations and contents, and soil 312 salinity. A PCA of all available data also indicated that the rice crops under the low to 313 intermediate doses of N fertilizer, and intermediate doses of P and K fertilizers had the 314 highest yields and lower CO2 and CH4 productions (Figure 8a). Whereas, crops receiving 315 high doses of N and K fertilizers had the lowest annual yields and the highest CO2 and 316 CH4 productions, as shown along the PC1 axis (Figure 8b). The PCAs indicated that the 317 North China and Northwest China rice crops were plotted by the two first PC axes toward 318 higher yield and soil P availability, total P content, C:N ratio and sand content, and toward 319 lower soil N:P ratio, labile C:N ratio and labile C:P ratio and lower CO2 and CH4 320 productions (Figure 8b). Fields in the East China, Center and South China, Southwest 321 China regions had opposite patterns. 322 Coastal rice crops have higher CO2 and CH4 emissions and lower yield than inland 323 rice crops. This was associated with higher soil N, water and clay contents in coastal rice 324 crops (Figure 8c). Figure 8d shows the overall differences in the environmental variables 14 325 between the sites with one and two annual crops. These two groups of sites were clearly 326 separated along the PC1 axis, with single-crop sites plotted toward higher annual yields 327 and soil P contents, salinities, C:N ratios and pHs, whereas the double-crop sites were 328 plotted toward higher soil N, water and LOC contents and N:P ratios, and higher CO2 329 productions. 330 The best mixed models (based on a low AIC and the highest R2 and 331 parsimoniousity), with rice variety and environmental traits as fixed independent 332 variables; location, topography, site and plot as random factors; and GHG productions as 333 dependent variables, indicated that 26% of the total variance of annual accumulated CO2 334 productions was explained by the length of the growth period, soil water content and bulk 335 density (Table S3). Thirty-nine percent of the total variance of CO2 productions rates was 336 explained by the growth-period length, soil water content and P-fertilizer dose. Forty- 337 seven and 50% of the total variance of annual accumulated CH4 productions and CH4 338 productions rate, respectively, were explained by the growth-period length, soil water 339 content, water source (river or groundwater), rice variety, bulk density, total soil N content 340 and P-fertilizer dose. Nineteen and 21% of the total variance of annual accumulated N2O 341 productions and N2O productions rate, respectively, were explained by the cropping 342 systems, soil P availability and soil NO3--N content. All six mixed models (Table S3), 343 corresponding to each of the production variables, explained >99% of the corresponding 344 total variance, while taking into account the random variables. 345 346 3.3 Relationships between annual yield and gas productions; fertilization effects 347 The GDA indicated that the crop sites with high yields generally had the highest GHG 348 productions, whereas crop sites with moderate yields generally had the lowest GHG 349 productions (Figure S8). Lower doses of N fertilizer were surprisingly associated with 15 350 higher yields but also with higher CH4 productions (Figure S9), whereas intermediate 351 doses of N fertilizer were associate with higher N2O productions, consistent with the 352 above results. Intermediate doses of P fertilizer were associated with the highest annual 353 yields and the lowest CO2 and CH4 productions (Figures S10), whereas intermediate 354 doses of K fertilizer were associated with the highest annual yields and the lowest CH4 355 productions (Figures S11) also consistent with the results of the one-way ANOVAs. 356 When the productivities at the various doses of N, P and K fertilizers (yield kg-1 357 fertilizer) were used as grouping dependent factors in the GDA, the sites with high N 358 productivity were associated with the highest N2O productions, sites with intermediate N 359 productivity were associated with the highest CH4 productions, and sites with low N 360 productivity were associated with higher CO2 productions (Figure S12). The lowest CO2, 361 CH4 and N2O productions were associated with sites with intermediate P fertilization 362 (Figure S13), and sites with intermediate K fertilization were associated with the lowest 363 CO2 and CH4 but not N2O productions (Figure S14). 364 365 3.4.SEM models 366 The SEM provided further evidence of the complex relationships among fertilizer doses, 367 GHG productions and annual yield. The N fertilization had a positive direct effect on 368 GHG productions and a negative effect on annual yield, and P fertilization had the 369 opposite pattern (Figure 9), and K fertilization did not significantly affect GHG 370 productions but generally negatively affected yield (Figure 9c). The N fertilization also 371 had indirect negative effects on annual yield by increasing the amount of labile soil carbon 372 content and by lowering soil pH. We also detected a negative direct and indirect effect, 373 by decreasing soil water content, effect of bulk density on soil CO2 productions (Figure 374 9b). 16 375 376 377 4. DISCUSSION 378 We studied potential emissions of carbon dioxide (CO2), CH4 and N2O from paddy fields 379 across China, because they are the most important GHG, contributing about 80% of the 380 current global radiative forcing (Myhre et al., 2013). The statistical analyses of our 381 database of 34 field sites throughout China clearly identified a general trend to over 382 fertilization with N. The current dose of N fertilizer was negatively correlated with rice 383 yield. This correlation was exacerbated by a positive correlation between N-fertilization 384 and N2O productions and negative correlations between N-fertilization and CO2 and CH4 385 productions. These results are generally consistent with previous findings, that have also 386 observed that higher N application doses had detrimential effects on yield and/or 387 increased gas emissions. Kim et al. (2016a) reported that rice yield increased with N- 388 fertilizer dose to levels of 110-130 kg N ha-1 y-1 and thereafter decreased at higher levels. 389 A meta-analysis of 24 field studies of rice crops in China demonstrated that N fertilization 390 increased CH4 and N2O emissions and decreased the ratio of yield-to-GHG emission, 391 which was highest for 150-200 kg N ha-1 y-1 (Feng et al., 2013) very similar results than 392 the observed in our study. Zhong, Wang, Yan & Zhao (2017) observed that the percentage 393 of N lost from leaching, ammonia volatilization and denitrification increased with the N- 394 fertilizer dose in a field experiment in China. Our results also provide evidence that both 395 increased soil N and P availability increased rice yield (Figure S4), even though the N- 396 fertilizer dose was negatively correlated with yield. 397 These results supported the hypothesis that increasing the doses of N fertilizer can 398 have more detrimental than positive effects on yield and GHG productions, even though 399 the availability of N to plants is important and positive for rice production. The results 17 400 thus strongly suggest a possible N saturation and/or ineffective N fertilization beyond a 401 certain level, with no extra positive effects but mostly negative effects of the higher N 402 input on rice production. Current reports, however, demonstrate an interest to optimize 403 N-fertilization doses in rice crops. Zhu, Zhang, Zhang, Deng & Zhang (2016) reported a 404 decrease in GHG emissions and an increase in rice yield as N-fertilizer doses decreased 405 and plant density increased (~50% increase) in northeastern China. Zhang et al. (2016b) 406 similarly demonstrated that the doses of N fertilizer typical of Nanjing paddy fields 407 (China) could be decreased without decreasing rice yield but could decrease GHG 408 emission. 409 Different studies have reported contradictory effects of N fertilization on GHG 410 emissions from rice crops (see the review by Cai, Shan & Xu, 2007). Several studies, 411 however, have observed an increase in N2O emissions with increases in N-fertilizer dose 412 in rice croplands (Zhang et al., 2014; Kim, Jeong, Kim, Kim & Kim, 2016b). 413 Our data analyses indicated that P- and K-fertilizer doses generally had positive 414 effects on rice yield and decreased GHG productions. The relationships among P- and 415 mainly K-fertilization, yield and GHG productions in rice croplands have not been 416 studied much, as compared to the corresponding relationships with N-fertilization. Our 417 results are nonetheless consistent with other experimental studies (Li et al., 2013; Datta, 418 Santra & Adhya, 2013). Datta, Santra & Adhya (2013) found that N fertilization 419 explained more of the changes in CH4 emissions than P or K fertilization. In contrast, Li 420 et al. (2013) observed that increasing the dose of P fertilizer decreased CH4 and N2O 421 emissions but increased yield. 422 However, our study has some potential limitations. The incubation period of three 423 days could not be sufficient to capture the overall patterns of GHG production and 424 emission, and thus has limited reflection of the field condition, but it is still very useful 18 425 to compare the potential emission among studied sites. In the paddy fields, the first peak 426 of methane emission generally occurs within a month after transplanting, just according 427 with our study. But a second peak would occur at approximately two months and this is 428 mainly governed by the stable low soil redox potential and neutral soil pH, and the 429 increased release of plant-borne carbon sources. (e.g. Ly et al., 2013; Vu et al., 2015). 430 This can be partially corrected by using Xu et al. 2016 method to estimate all year gas 431 emissions. But all in all this can explain why the estimated methane production in this 432 paper is lower than in other reports in Asian countries (Yan et al., 2003; Vo et al., 2018). 433 434 4.1.Regions, locations and cropping systems 435 Our results suggest that rice yield can be increased without increasing or even decreasing 436 GHG productions. For example, our results indicated that yields were higher for the 437 japonica than the Hsien varieties. The N2O productions were several times higher for the 438 Hsien varieties, leading to a better balance between yield and GHG productions in fields 439 with japonica varieties, even though CH4 productions were clearly higher for the japonica 440 varieties. Adequate irrigation is fundamental for assuring high rice yield (Sun et al., 2016), 441 but our data also suggest that GHG productions can be reduced by regulating doses of N-, 442 P-, and/or K-fertilization. We have also observed that CH4 and N2O productions were 443 much higher at sites irrigated and flooded with river water than at sites irrigated and 444 flooded with groundwater or water from superficial reservoirs, whereas average yield did 445 not differ significantly between these sites. This result is difficult to interpret in the 446 context of this study; further studies should aim to find out the cause of these differences. 447 The balance between yield and GHG production was worst in the East China rice crops, 448 with the highest productions of CO2, CH4 and N2O because in this area the temperature 449 is relatively higher, and there are more active substrates (Wang et al., 2015). The North 19 450 China and Northwest China rice crops had the opposite patterns because in this area the 451 lower temperature limits the substrates decomposition, such as soil organic carbon, and 452 then the microbes act to convert plant residues into humus in the soil (Cui et al., 2008). 453 The lower temperature can decrease decomposition and the CO2 and CH4 release from 454 soil by mediating the microbe growth (Tang et al., 2017). 455 Moreover, in these areas paddy soils had relative higher pH and water comes from 456 rivers that can provide more substrate to the paddy and also have longer growth period 457 and more illumination than in other areas of china. Furthermore, the North China and 458 Northwest China rice crops lower greenhouse gases production was related with the 459 higher C:N ratio. The C:N ratio controls the CO2 and CH4 release. There is more limited 460 carbon decomposition with relatively higher C:N ratios (Windham, 2001). The soil GHG 461 productions tended to be lower (significantly for CO2) at inland than coastal sites because 462 he coastal paddy fields had higher carbon and nitrogen concentration, and therefore more 463 substrates for soil GHG productions (Delaune et al., 2018). 464 Moreover, the CO2 production rates were higher at sites with one annual crop than 465 at sites with two annual crops because the two annual areas, mostly in the south of China, 466 had soils rich in ferric oxide, which, in turn, favored C fixation in soil, and more stable 467 soil carbon, thus decreasing C release in form of methane (Wang et al., 2014). 468 Furthermore, CH4 production was higher under single crop. Single crop areas had lower 469 temperature during the whole year, thus lowering decomposition, increasing carbon 470 storage in the soil and providing more substrates for CH4 production when the 471 temperature increases during rice growth period. However, N2O is higher under double 472 crops because of more applications of N fertilizer. 473 CH4 production was higher under single crop. Single crop areas had lower 474 temperature during the whole year, thus lower decomposition, and more carbon can be 20 475 storage in the soil and more substrates is able for CH4 production when the temperature 476 increase during rice growth period. However, N2O is higher under double crops, because 477 along the year the N fertilizer applied is in higher amount in the doubles crop areas. 478 Our multivariate analysis of all data from all sites indicated that the highest average 479 yields were accompanied by the highest GHG productions. Optimizing rice production 480 and GHG production is thus interesting and challenging. Our analyses provide some clues 481 for optimization, suggesting that some advances could be achieved by adjusting N- 482 fertilization to a level that improve rice production but avoid the negative effects of 483 excessive N input. Further, GHG productions could be reduced by maintaining adequate 484 levels of P and K fertilizers, mostly at intermediate doses of the current range of P and K 485 fertilization across China, and by re-examining the use of different rice varieties and 486 strategies of water management. 487 488 5. CONCLUSION 489 Rice production has historically been improved using N fertilizers, but we found that the 490 current paddy field sites in China with relatively low N fertilization had high rice 491 production and low soil CO2 and CH4 potential productions, even though rice yield was 492 positively correlated with soil N availability. In contrast to N fertilization, sites using 493 intermediate doses of P and K fertilizers had the highest rice yields and the lowest soil 494 GHG potential productions. The large capacity of soil to accumulate P in non-available 495 forms and the large capacity of K leaching by the high use of water in rice crops together 496 with a less direct role of P and K in CH4 and N2O productions could explain these results. 497 The analysis of all our data strongly suggests that increasing rice yield and 498 minimizing GHG productions can be further optimized in China. The use of different rice 499 varieties and strategies of water management and fertilization with implications for 21 500 minimizing GHG productions should thus be re-examined. 501 502 ACKNOWLEDGMENTS 503 The authors would like to thank Zhu Defeng, Fu Rongfu, Yang Binjuan, Xu Hui, Zhou 504 Jianxia, Liu Jianhui, Huang Jinlong, Chen Chuanbing, Chen Youyang and Chen 505 Xiaoxuan for their assistance with field sampling and laboratory analysis. Funding was 506 provided by the National Science Foundation of China (41571287; 31000209), Research 507 Project of Public Institute of Fujian Province (2018R1034-1), Outstanding Young 508 Research Talents in Higher Education of Fujian Province (2017), Spanish Government 509 grant CGL2016-79835, Catalan Government grant SGR 2017-1005 and European 510 Research Council Synergy grant ERC-SyG-2013-610028, IMBALANCE-P. 511 512 CONFLICS OF INTEREST 513 The authors declare no conflicts of interest. 514 515 RERERENCES 516 517 Adhya, T. K., Pattnaik, P., Satpathy, S. N., Kumaraswamy, S., & Sethunathan, N. (1998). 518 Influence of phosphorus application on methane emission and production in flooded 519 paddy soils. Soil Biology and Biochemistry, 30, 177–181. https://doi.org/ 520 10.1016/S0038-0717(97)00104-1 521 Ali, M. A., Oh, J. H., & Kim, P. J. (2008). Evaluation of silicate iron slag amendment on 522 reducing methane emission from flood water rice farming. Agriculture, Ecosystems 523 and Environment, 128, 21–26. https://doi.org/10.1016/j.agee.2008.04.014 524 Barton, K. (2012). MuM. In: Multi-model inference. R package version 1.7.2. 525 http://CRAN. R-283 project.org/package=MuMIn. at http://cran.r-project.org/ 526 package=MuMIn> 527 Benjamini, Y., & Hochberg, Y. (1995). Controlling the false discovery rate: a practical 528 and powerful approach to multiple testing. Journal of the Royal Statistical Society B, 22 529 57, 289–300. http://www.jstor.org/stable/2346101 530 Cai, Z., Shan, Y., & Xu, H. (2007). Effect of nitrogen fertilization on CH4 emissions from 531 rice fields. Soil Science and Plant Nutrition, 53, 353–361. 532 https://doi.org/10.1111/j.1747-0765.2007.00153.x 533 Cheng, S.H., & Li, J. (2007). Modern Chinese rice. Beijing: Jindun Press. 534 Cui M, Zhang X, Cai Q, Wang Y, Fan H, & Zhou J. (2008). Relationship between black 535 soil development and climate change and geomorphological evolution in northeast 536 china. Geographical Research, 27(3), 527-535. DOI:10.1360/jos172601 537 Datta, A., Santra, S. C., & Adhya, T. K. (2013). Effect of inorganic fertilizers (N, P, K) 538 on methane emission from tropical rice field of India. Atmospheric Environment, 66, 539 123–130. https://doi.org/10.1016/j.atmosenv.2012.09.001 540 Davison, A. C., Hinkley, D. V., & Schechtman, E. (1986). Efficient Bootstrap Simulation. 541 Biometrika, 73, 555–566. https://www.jstor.org/stable/2336519 542 Delaune, R. D., White, J. R., Elsey-Quirk, T., Roberts, H. H., Wang, D. Q. (2018). 543 Differences in long-term vs short-term carbon and nitrogen sequestration in a 544 coastal river delta wetland: implications for global budgets. Organic Geochemistry, 545 123, 67-73. DOI:10.1016/j.orggeochem.2018.06.007 546 Dong, H. B., Yao, Z. S., Zheng, X. H., Mei, B. L., Xie, B. H., Wang, R., Deng, J., Cui, 547 F., & Zhu, J. G. (2011). Effect of ammonium-based, non-sulfate on CH4 emissions 548 from a paddy field with a typical Chinese water management regime. Atmospheric 549 Environment, 45, 1095–1101. https://doi.org/10.1016/j.atmosenv.2010.11.039 550 Fan, X. F., Yu, H. Y., Wu, Q. Y., Ma, J., Xu, H., Yang, J. H., & Zhuang, Y. Q. (2016). 551 Effects of fertilization on microbial abundance and emissions of greenhouse gases 552 (CH4 and N2O) in rice paddy fields. Ecology and Evolution, 6, 1054–1063. doi: 553 10.1002/ece3.1879 554 FAO [Food and Agricultural Organization of the United Nations]. (2009). OECD-FAO 555 Agricultural Outlook 2011–2030. 556 Feng, J., Chen, C., Zhang, Y., Song, Z., Deng, A., Zheng, C., & Zhang, W. (2013). Impacts 557 of cropping practices on yield-scaled greenhouse gas emissions from rice fields in 558 China: a meta-analysis. Agriculture Ecosystems and Environment, 164, 220–228. 559 https://doi.org/10.1016/j.agee.2012.10.009 560 Fox, J., Nie, Z., & Byrnes, J. (2013). sem: Structural equation models (R. package version 561 3.1-1). Retrieved from http://CRAN.R-project.org/package=sem 562 Haque, M. M., Kim, S. Y., Ali, M. A., & Kim, P. J. (2015). Contribution of greenhouse 23 563 gas emissions during cropping and fallow seasons on total global warming potential 564 in mono-rice paddy soils. Plant and Soil, 387, 251–264. doi:10.1007/s11104-014- 565 2287-2 566 Hothorn, T., Bretz, F., & Wesrfall, P. (2013). Package “mulcomp” (WWW document). 567 U.R.L. http://cran.stat.sfu.ca/web/packages/mulcomp/mulcomp.pdf. (accessed 568 19.12.17) 569 Jiang, J. Y., Chen, L. M., Sun, Q., Sang, M. M., & Huang, Y. (2015). Application of 570 herbicides is likely to reduce greenhouse gas (N2O and CH4) emissions from rice- 571 wheat cropping systems. Atmospheric Environment, 107, 62–69. 572 https://doi.org/10.1016/j.atmosenv.2015.02.029 573 Kim, G. W., Jeong, S. T., Kim, G. Y., Kim, P. J., & Kim, S. Y. (2016b). Evaluation of 574 carbon dioxide emission factor from urea during rice cropping season: a case study 575 in Korean paddy soil. Atmospheric Environment, 139, 139–146. 576 https://doi.org/10.1016/j.atmosenv.2016.05.033 577 Kim, G. W., Gwon, H. S., Jeong, S. T., Hwang, H. Y., & Kim, P. J. (2016a). Different 578 responses of nitrogen fertilization on methane emission in rice plant included and 579 excluded soils during cropping season. Agriculture, Ecosystems and Environment, 580 230, 162–168. https://doi.org/10.1016/j.agee.2016.06.005 581 Launio, C. C., Asis, C. A., Manalili, R. G., & Javier, E. F. (2016). Cost-effectiveness 582 analysis of farmers’ rice Straw management practices considering CH4 and N2O 583 emissions. Journal of Environmental Management, 183, 245–252. https://doi. 584 org/10.1016/j.jenvman.2016.08.015 585 Li, C., Salas, W., Deangelo, B., Rose, S. (2006). Assessing alternatives for mitigating net 586 greenhouse gas emissions and increasing yields from rice production in china over 587 the next twenty years. Journal of Environmental Quality, 35, 1554– 588 1565.doi:10.2134/jeq2005.0208 589 Li, C. S., Frolking, S., Xiao, X. M., Moore, B., Boles, S., Qiu, J. J., Huang, Y., Salas, W., 590 & Sass, R. (2005). Modeling impacts of farming management alternatives on CO2, 591 CH4 and N2O emissions: A case study for water management of agriculture of China. 592 Global Biogeochemical Cycles, 19, GB3010. https://doi.org/10.1029/ 593 2004GB002341 594 Li, C. F., Zhang, Z. S., Guo, L. J., Cai, M. L., & Cao, C. G. (2013). Emissions of CH4 and 595 CO2 from double rice cropping systems under varying tillage and seedling methods. 24 596 Atmospheric Environment, 80, 438–444. https://doi.org/10.1016/ 597 j.atmosenv.2013.08.027 598 Li, X., Wang, H., Gan, S. H., Jiang, D. Q., Tian, G. M., & Zhang, Z. J. (2013). Eco- 599 stoichiometry alterations in paddy soil ecosystem driven by phosphorus application. 600 Plos One, 8, e61141. doi: 10.1371/journal.pone.0061141 601 Liu, Y., Hu, C., Mohamed, I., Wang, J., Zhang, G. S., Li, Z. G., & Chen, F. (2016). Soil 602 CO2 emissions and drivers in rice-wheat rotation fields subjected to different long- 603 term fertilization practices. Clean - Soil, Air, Water, 44, 867–876. 604 https://doi.org/10.1002/clen.201400478 605 Ly, P., Jensen, L.S., Bruun, T.B., de Neergaard, A. (2013). Methane (CH4) and nitrous 606 oxide (N2O) emissions from the system of rice intensification (SRI) under a rain-fed 607 lowland rice ecosystem in Cambodia. Nutrient Cycling in Agroecosystems, 97, 13– 608 27. DOI 10.1007/s10705-013-9588-3. 609 Lu, R. K. (1999). Analysis methods of soil science and agricultural chemistry. Beijing: 610 Agriculture Science and Technology Press. 611 Mishra, U., Ussiri, D. A. N., & Lal, R. (2010). Tillage effects on soil organic carbon 612 storage and dynamics in Corn Belt of Ohio USA. Soil and Tillage Research, 107, 88– 613 96. https://doi.org/10.1016/j.still.2010.02.005 614 Mitchell-Olds, T. (1986). Jackknife, bootstrap and other resampling methods in 615 regression analysis. Annals of Statistics, 14, 1316–1318. https://www.jstor. 616 org/stable/2241454 617 Myhre, G., Shindell, D., Bréon, F. M., Collins, W., Fuglestvedt, J., Huang, J., Koch, D., 618 Lamarque, J. F., Lee, D., Mendoza, B., Nakajima, T., Robock, A., Stephens, G., 619 Takemura, T., & Zhang, H. (2013). Anthropogenic and Natural Radiative Forcing. In: 620 Climate Change 2013: The Physical Science Basis. Contribution of Working Group 621 I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change 622 [Stocker, T. F., Qin, D., Plattner, G. K., Tignor, M., Allen, S. K., Boschung, J., Nauels, 623 A., Xia, Y., Bex, V., & Midgley, P. M. (eds.)]. Cambridge University Press, 624 Cambridge, United Kingdom and New York, NY, USA. pp 714. 625 Olsson, L., Ye, S., Yu, X., Wei, M., Krauss, K. W., & Brix, H. (2015). Factors influencing 626 CO2 and CH4 emissions from coastal wetlands in the Liaohe Delta, Northeast China. 627 Biogeosciences, 12, 4965–4977. doi:10.5194/bg-12-4965-2015 628 Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., & Core, T. R. (2016). nlme: Linear and 629 Nonlinear Mixed Effects Models. R package version 3.1-126, http://CRAN. 25 630 Raamsdonk, L. M., Teusink, B., Broadhurst, D., Zhang, N. S., Hayes, A., Walsh, M. C., 631 Berden, J. A., Brudle, K. M., Kell, D. K., Rowland, J. J., Westerhoff, H. V., van Dam, 632 K., & Oliver, S. G. (2001). A functional genomics strategy that uses metabolome data 633 to reveal the phenotype of silent mutations. Nature Biotechnology, 19, 45–50. 634 http://dx.doi.org/10.1038/83496 635 Revelle, W. (2010). Package “psych.” Retrieved August 24, 2010, from 636 http://personality‐project.org/r/psych_manual.pdf 637 Sheng, R., Chen, A. L., Zhang, M. M., Whiteley, A. S., Kumaresan, D., & Wei, W. X. 638 (2016). Transcriptonal activities of methanogens and methanotrophs vary with 639 methane emission flux in rice soils under chronic nutrient constraints of phosphorus 640 and potassium. Biogeosciences, 13, 6507–6518. doi: 10.5194/bg-13-6507-2016 641 Sun, W., Huang, Y., Zhang, W., & Yu, Y. (2010). Carbon sequestration and its potential 642 in agricultural soils of China. Global Biogeochemical Cycles, 24, 1154–1157. 643 https://doi.org/10.1029/2009GB003484 644 Sun, H., Zhou, S., Fu, Z., Chen, G., Zou, G., & Song, X. (2016). A two-year field 645 measurement of methane and nitrous oxide fluxes from rice paddies under 646 contrasting climate conditions. Scientific Reports, 6, 28255. doi: 647 10.1038/srep28255 648 Tang Q C. (2011). Greenhouse gas emission in China's agriculture: situation and 649 challenge. China Population, Resources and Environment, 21(10), 69-75. DOI: 650 10.3969/j. issn.1002 －2104.2011.10.011. 651 Tang J, Cheng H, & Fang, C. (2017). The temperature sensitivity of soil organic carbon 652 decomposition is not related to labile and recalcitrant carbon. Plos One, 12(11), 653 e0186675. DOI:10.1371/journal.pone.0186675 654 Theint, E. E., Suzuki, S., Ono, E., & Bellingrath-Kimura. S. D. (2014). Influence of 655 different rates of gypsum application on methane emission from saline soil related 656 with rice growth and rhizosphere exudation. Catena, 133, 467–473. https://doi. org/ 657 10.1016/j.catena.2014.12.003 658 Trinh, M. V., Tesfai, M., Borrell, A., Nagothu, U. S., Bui, T. P. L., Quynh, V. D., & 659 Thanh, L. Q. (2017). Effects of organic, inorganic and slow-release urea fertilizers 660 on CH4 and N2O emissions from rice paddy fields. Paddy and Water Environment, 661 15, 317–330. doi: 10.1007/s10333-016-0551-1 662 Tukey, John W (1977). Exploratory Data Analysis. Addison-Wesley. 26 663 Vo, T.B.T., Wassmann, R., Tirol-Padre, A., Cao, V.P., MacDonald, B., Espaldon, 664 M.V.O., Sander, B.O. (2018). Methane emission from rice cultivation in different 665 agro-ecological zones of the Mekong river delta: seasonal patterns and emission 666 factors for baseline water management. Soil Science and Plant Nutrition, 64, 47–58 667 . https://doi.org/10.1080/00380768.2017.1413926 668 Vu, Q.D., de Neergaard, A. Tran, T.D., Hoang, Q.Q., Ly, P., Tran, T.M., Jensen, L.S. 669 (2015). Manure, biogas digestate and crop residue management affects methane gas 670 emissions from rice paddy fields on Vietnamese smallholder livestock farms. 671 Nutrient Cycling in Agroecosystems, 103, 329–346 . DOI 10.1007/s10705-015- 672 9746-x 673 Wang, W., Lai, D. Y. F., Li, S. C., Kim, P. J., Zeng, C., Li, P., & Liang, Y. (2014). Steel 674 slag amendment reduces methane emission and increases rice productivity in 675 subtropical paddy fields in China. Wetlands Ecology and Management, 22, 683–691. 676 doi: 10.1007/s11273-014-9364-4 677 Wang, W., Wang, C., Sardans, J., Min, Q., Zeng, C., Tong, C., & Peñuelas, J. (2015a). 678 Agricultural land use decouples soil nutrient cycles in a subtropical riparian wetland 679 in China. Catena, 133, 171–178. https://doi.org/10.1016/j.catena.2015.05.003 680 Wang, W., Lai, D. Y. F., Wang, C., Pan, T., & Zeng, C. (2015b). Effects of rice 681 incorporation on active organic carbon pools in a subtropical paddy field. Soil and 682 Tillage Research, 152, 8–16. https://doi.org/10.1016/j.still.2015.03.011 683 Wang, W., Lai, D. Y. F., Sardans, J., Wang, C., Datta, A., Pan, T., Zeng, C., Bartrons, M., 684 & Peñuelas, J. (2015c). Rice straw incorporation affects global warming potential 685 differently in early vs. Late cropping seasons in Sourtheastern China. Filed Crops 686 Research, 181, 42–51. https://doi.org/10.1016/j.fcr.2015.07.007 687 Wang, W., Sardans, J., Lai, D.Y.F., Wang, C., Zeng, C., Tong, C., Liang, Y., & Peñuelas, 688 J. (2015d). Effects of steel slag application on greenhouse gas emissions and crop 689 yield over multiple growing seasons in a subtropical paddy field in China. Filed 690 Crops Research, 171, 146–156. https://doi.org/10.1016/j.fcr.2014.10.014 691 Wang, W., Min, Q., Sardans, J., Wang, C., Asensio, D., Bartrons, M., & Peñuelas, J. 692 (2016a). Organic cultivation of jasmine and tea increases carbon sequestration by 693 changing plant and soil stoichiometry. Agronomy Journal, 108, 1–13. 694 doi:10.2134/agronj2015.0559 695 Wang, W., Sardans, J., Zeng, C., Tong, C., Wang, C., & Peñuelas, J. (2016b). Impact of 696 plant invasion and increasing floods on total soil phosphorus and its fractions in the 27 697 Minjiang River esturine wetlands, China. Wetlands, 36, 21–36. doi: 10.1007/s13157- 698 015-0712-9 699 Wang, W., Wu, X. H., Chen, A. L., Xie, X. L., Wang, Y. Q., & Yin, C. M. (2016c). 700 Mitigating effects of ex situ application of rice straw on CH4 and N2O emissions from 701 paddy-upland coexisting system. Scientific Reports, 6, 37402. http://dx.doi. org/ 702 10.1038/srep37402 703 Wang, C., Lai, D. Y. F., Sardans, J., Wang, W. Q., Zeng, C. S., & Peñuelas, J. (2017a). 704 Factors related with CH4 and N2O emissions from a paddy field: Clues for 705 management implications. Plos One, 12, e0169254. doi: 10.1371/journal. 706 pone.0169254 707 Wang, W., Sardans, J., Wang, C., Zeng, C., Tong, C., Asensio, D., & Peñuelas, J. (2017b). 708 Relationships between the potential production of the greenhouse gases CO2, CH4 709 and N2O and soil concentrations of C, N and P across 26 paddy fields in southeastern 710 China. Atmospheric Environment, 164, 458–467. 711 http://dx.doi.org/10.1016/j.atmosenv.2017.06. 023 712 Wang, W., Neogi, S. , Lai, D. Y. F. , Zeng, C. , Wang, C. , Zeng, D. (2017c). Effects of 713 industrial and agricultural waste amendment on soil greenhouse gas production in a 714 paddy field in southeastern china. Atmospheric Environment, 164, 239–249. 715 https://doi.org/10.1016/j.atmosenv.2017.05.052. 716 Wassmann, R., Neue, H. U., Bueno, C., Lantin, R. S., Alberto, M. C. R., Buendia, L. V., 717 Bronson, K., Papen, H., & Rennenberg, H. (1998). Methane production capacities of 718 different rice soil derived from inherent and exogenous substrates. Plant and Soil, 719 203, 227–237. doi: 10.1023/A:1004357411814 720 Windham, L. (2001). Comparison of biomass production and decomposition between 721 Phragmites australis (common reed) and Spartina patens (salt hay grass) in brackish 722 tidal marshes of New Jersey, USA. Wetlands, 21:179–188. 723 Xu, X., Chen, C., & Xiong, Z. (2016). Effects of biochar and nitrogen fertilizer 724 amendment on abundance and potential activity of methanotrophs and methanogens 725 in paddy field. Acta Pedologica Sinica, 53, 1517–1527. 726 DOI:10.11766/trxb201604210087 727 Yan, X., Ohara, T., Akimoto, H. (2003). Development of region-specific emission factors 728 and estimation of methane emission from rice fields in the East, Southeast and South 729 Asian countries. Global Change Biology, 9, 237 – 254. 730 Zhang, X., Yin, S., Li, Y., Zhuang, H., Li, C., & Liu, C. (2014). Comparison of greenhouse 28 731 gas emissions from rice paddy fields under different nitrogen fertilization loads in 732 Chongming Island, Eastern China. Science of the Total Environment, 472, 381–388. 733 https://doi.org/10.1016/j.scitotenv.2013.11.014 734 Zhang, G. B., Yu, H. Y., Fan, X. F., Yang, Y. T., Ma, J., & Xu, H. (2016a). Drainage and 735 tillage practices in the winter fallow season mitigate CH4 and N2O emissions from a 736 double-rice field in China. Atmospheric Chemistry & Physics, 16, 11853–11866. 737 doi:10.5194/acp-16-11853-2016 738 Zhang, Z. S., Chen, J., Liu, T. Q., Cao, C. G., & Li, C. F. (2016b). Effects of nitrogen 739 fertilizer sources and tillage practices on greenhouse gas emissions in paddy fields of 740 central China. Atmospheric Environment, 144, 274–281. https://doi.org/ 741 10.1016/j.atmosenv.2016.09.003 742 Zhao, Z., Yue, Y. B., Sha, Z. M., Li, C. S., Deng, J., Zhang, H. L., Gao, M. F., & Cao, L. 743 K. (2015). Assessing impacts of alternative fertilizer management practices on both 744 nitrogen loading and greenhouse gas emissions in rice cultivation. Atmospheric 745 Environment, 119, 393–401. https://doi.org/10.1016/j.atmosenv. 2015.08.060 746 Zheng, Y., Zhang, L. M., & He, J. Z. (2013). Immediate effects of nitrogen, phosphorus, 747 and potassium amendments on the methanotrophic activity and abundance in a 748 Chinese paddy soil under short-term incubation experiment. Journal of Soils and 749 Sediments, 13, 189–196. doi: 10.1007/s11368-012-0601-2 750 Zhong, Y., Wang, X., Yang, J., & Zhao, X. (2017). Tracing the fate of nitrogen with 15N 751 isotope considering suitable fertilizer rate related to yield and environmental impacts 752 in paddy field. Paddy and Water Environment, 15, 943–949. doi: 10.1007/s10333- 753 017-0606-y 754 Zhu, X., Zhang, J., Zhang, Z., Deng, A., & Zhang, W. (2016). Dense planting with less 755 basal nitrogen fertilization might benefit rice cropping for high yield with less 756 environmental impacts. European Journal of Agronomy, 75, 50–59. https://doi. 757 org/10.1016/j.eja.2016.01.003 758 Zhu, T. B., Zhang, J. B., & Cai, Z. C. (2011). The contribution of nitrogen transformation 759 processes to total N2O emissions from soils used for intensive vegetable cultivation. 760 Plant and Soil, 343, 313–327. doi: 10.1007/s11104-011-0720-3 761 Zhu. (2006). System of rice intensification. Beijing: China Agricultural Science and 762 Technology Press. 763 29 764 765 766 Figures 767 Figure 1. Locations of the field sites for soil sampling across China. 768 Figure 2. Rice yield and total accumulated productions of CO2, CH4 and N2O for the N- 769 fertilizer annual doses (low < 96 kg ha-1, intermediate 96-140 kg ha-1, high more than 140 770 kg ha-1). Different letters indicate significant statistical differences (P<0.05). 771 Figure 3. Rice yield and total accumulated productions of CO2, CH4 and N2O for the P- 772 fertilization annual doses (low < 62 kg ha-1, intermediate 72-92 kg ha-1, high more than 773 72 kg ha-1). Different letters indicate significant statistical differences (P<0.05). 774 Figure 4. Rice yield and total accumulated productions of CO2, CH4 and N2O for the K- 775 fertilizer annual doses (low < 70 kg ha-1, intermediate 71-81 kg ha-1, high more than 81 776 kg ha-1). Different letters indicate significant statistical differences (P<0.05). 777 Figure 5. Rice yield and total accumulated productions of CO2, CH4 and N2O in the 778 various regions. Different letters indicate significant statistical differences (P<0.05). 779 Figure 6. Rice yield and total accumulated productions of CO2, CH4 and N2O at inland 780 vs coastal sites. Different letters indicate significant statistical differences (P<0.05). 781 Figure 7. Rice yield and total accumulated productions of CO2, CH4 and N2O at sites 782 with one vs two rice crops. Different letters indicate significant statistical differences 783 (P<0.05). Different letters between brackets indicate marginal significant statistical 784 differences (P<0.1). 785 Figure 8. First two PC axes of the PCA of all soil variables, GHG productions and yield 786 showing the areas (95% confidence intervals) occupied by sites with the N-, P- and K- 787 fertilizer doses (A), the areas (95% confidence intervals) occupied by sites in the various 788 regions (B), showing the areas (95% confidence intervals) occupied by sites for the inland 30 789 vs coastal locations (C) and showing the areas (95% confidence intervals) occupied by 790 sites with one vs two rice crops (D). The Acronyms mean: CO2 rate= CO2 productions 791 rate, CH4 rate= CH4 productions rate, N2O rate= N2O productions rate, [C]=Soil C 792 concentration, [N] = Soil N concentration, [P] = Soil P concentration, NH4= Soil NH4+ 793 concentration, NO3= Soil NO3- concentration, NO3cont= Soil NO3- content, LOC= Soil 794 labile organic carbon concentration, P avai= Total soil available P concentration, 795 Sand=Percentage in weight of sand in dry soil, Clay= Percentage in weight of clay in dry 796 soil, pH=Soil pH, Water=Soil water content, salinity=Soil salinity, Bulkd=Soil bulk 797 density, Yield=Rice crop yield, MAP=Mean Annual Precipitation, MAT=Mean Annual 798 Temperature. 799 Figure 9. SEM models with soil CO2 production rate (A), soil accumulate CO2 production 800 (B) and rice crop yield (C) as end endogen variables. Numbers on the arrows are the 801 estimates of the effects of one variable (near the beginning of the arrow) over another 802 (near the end of the arrow) and the corresponding P values (in parentheses). Red arrows 803 indicate negative relationships, and black arrows indicate positive relationships. 31