This is the accepted version of the article: Li, Xiaofei; Sardans i Galobart, Jordi; Hou, Lijun; [et al.]. «Climatic temperature controls the geographical patterns of coastal marshes greenhouse gases emissions over China». Journal of Hydrology, Vol. 590 (Nov. 2020), art. 125378. DOI 10.1016/j.jhydrol.2020.125378 This version is avaible at https://ddd.uab.cat/record/232211 under the terms of the license 1 Climatic temperature controls the geographical patterns of coastal marshes 2 greenhouse gases emissions over China 3 Xiaofei Li1,2,3, Jordi Sardans4,5, *, Lijun Hou2, *, Min Liu3, Chaobin Xu1 and Josep 4 Peñuelas4,5 1 5 Key Laboratory for Humid Subtropical Eco-geographical Processes of the Ministry of 6 Education, Fujian Normal University, Fuzhou, 350007, China, 2 7 State Key Laboratory of Estuarine and Coastal Research, East China Normal 8 University, Shanghai, 200241, China, 3 9 Key Laboratory of Geographic Information Science of the Ministry of Education, 10 School of Geographic Sciences, East China Normal University, Shanghai, 200241, 11 China, 4 12 CSIC, Global Ecology CREAF-CEAB-CSIC-UAB, Bellaterra 08193, Barcelona, 13 Catalonia, Spain, 5 14 CREAF, Cerdanyola del Valles 08193, Barcelona, Catalonia, Spain. 15  16 Corresponding Author: 17 E-mail address: j.sardans@creaf.uab.es (J. Sardans); ljhou@sklec.ecnu.edu.cn (L. Hou) 18 19 20 21 22 23 24 25 26 Abstract: Coastal marshes contribute greatly to atmospheric greenhouse gases (GHGs) 1 / 36 27 and have thus attracted considerable attention at regional and global scales. However, 28 the relative importance of climates and soil variables to the geographical variation of 29 GHG emissions at a large geographical scale remains poorly understood. We thus 30 investigated the spatial patterns of GHG (CO2, CH4 and N2O) emissions from coastal 31 marshes along a latitudinal gradient in China and identified the climatic and soil 32 variables controlling the emissions. Emissions of CO2, CH4 and N2O were estimated at 33 rates of 12.8–310 mg CO2 m–2 h–1, 23.6–986 µg CH4 m–2 h–1 and 1.58–110 µg N2O m– 2 34 h–1, respectively. The emissions of CH4, N2O and CO2 increased from high to low 35 latitudes and were strongly correlated with climatic temperature. These emissions also 36 varied significantly between temperate and subtropical climates and between winter 37 and summer. Soil Eh, total organic carbon (TOC), dissolved organic carbon (DOC), 38 total nitrogen, microbial biomass carbon (MBC) and nitrogen (MBN), bacterial 39 abundance were the key regulators of CH4 emissions. Soil Eh, TOC, DOC, MBC and 40 MBN were the crucial variables controlling N2O emissions. CO2 emissions across the 41 coastal marshes were controlled mainly by soil pH, TOC, DOC, MBC, MBN and 42 bacterial abundance. Although GHG emissions were strongly correlated with climatic 43 temperature, 28–67% of the variations in GHG emissions were attributed to soil 44 variables. The relative amplitudes (%) of the increases in GHG emissions, estimated on 45 the basis of projected annual increases of 2 or 4 °C by 2100, were larger in temperate 46 than subtropical coastal marshes, indicating that the coastal marshes with lower 47 temperature were more sensitive to global warming. The annual emissions of CO2, CH4 48 and N2O estimated from the Chinese coastal marshes accounted for 6.6, 0.62 and 0.70‰ 49 of the annual emissions from global terrestrial ecosystems, respectively. These results 50 suggest that coastal marshes will be important to future global warming and will play a 51 critical role in global GHG emissions. 2 / 36 52 Keywords: Greenhouse gases; climatic gradient; global warming; Chinese coastal 53 marshes 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 1. Introduction 3 / 36 77 Greenhouse gases (GHGs) have attracted much attention due to their environmental 78 consequences including global-warming potential (GWP) and ozone depletion (Lashof 79 and Ahuja, 1990; Meinshausen et al., 2009; IPCC, 2013). Methane (CH4) and nitrous 80 oxide (N2O) are important GHGs because they have approximately 30- and 300-fold 81 higher GWP over a 100 year time frame, respectively, than carbon dioxide (CO2) (IPCC, 82 2013). CO2 is the largest contributor to GHGs and is expected to greatly affect GWP. 83 Many recent studies have been devoted to identifying the natural sources of GHGs, 84 estimating their emissions and predicting the global climatic consequences of their 85 increasing concentrations at regional and global scales (Meinshausen et al., 2009; Yvon- 86 Durocher et al., 2014; Yuan et al., 2015; Murray et al., 2015; Abdul-Aziz et al., 2018). 87 Coastal wetlands store an estimated 10% of global soil carbon (C), so they are important 88 sources of CH4 and CO2 emissions (Murray et al., 2015; Yuan et al., 2015; Lu et al., 89 2017; Abdul-Aziz et al., 2018). Coastal wetlands can intercept 93% of the terrestrial 90 inputs of nitrogen (N) (Brin et al., 2010; Yang and Silver, 2016) and are also predicted 91 to increase N2O emissions under the increasing nitrate level (Turner et al., 2016). 92 Although the coastal wetlands account for a minor part of terrestrial area, they 93 contribute largely to atmospheric CH4, N2O and CO2 (Morrissey et al., 2014a; Murray 94 et al., 2015; Yuan et al., 2015; Yang and Silver, 2016; Jeffrey et al., 2018). Quantifying 95 the emissions of CH4, N2O and CO2 from coastal marshes and understanding the 96 variables controlling these emissions are therefore crucial for predicting future 97 emissions and their effects on climate change. 98 Coastal marshes have been reported to be responsible for the large global emissions 99 of GHGs because the microbial processes for GHGs production can be enhanced by 100 some changing environmental factors (Chen et al., 2015; Yuan et al., 2015; Abdul-Aziz 101 et al., 2018; Jeffrey et al., 2018; Emery et al., 2019). C and N availability, vegetation 4 / 36 102 invasion, tides, water level, pH and redox dynamics are the important factors affecting 103 local GHG emissions (Chen et al., 2015; Yuan et al., 2015; Yang and Silver, 2016; 104 Jeffrey et al., 2018; Emery et al., 2019). Soil organic matter characteristics (dissolved 105 organic carbon and total organic carbon) are suggested to be the important factors 106 controlling emissions of CO2 and CH4 because the enhanced microbial respiration and 107 anoxic organic matter decomposition can increase the yields of CO2 and CH4 108 (Morrissey et al., 2014a; Yuan et al., 2015; Abdul-Aziz et al., 2018). The microbial 109 communities and enzymatic activities for the CO2 and CH4 production are also greatly 110 affected by soil organic matter characteristics (Morrissey et al., 2013). Most studies 111 reported that N2O emissions are mainly controlled by NH4+ and NO3– (Borges et al., 112 2015; Murray et al., 2015; Wells et al., 2018) because N2O production is primarily 113 attributed to nitrification and denitrification (Hu et al., 2012; Beaulieu et al., 2015). Soil 114 microbial biomass carbon (MBC) and nitrogen (MBN) indicting the microbial activities 115 and assimilating C and N capabilities can also predict the GHG emissions (Morrissey 116 et al., 2014a; Li et al., 2019). Although functional gene abundances and microbial 117 activities are crucial for GHG production (Hu et al., 2012; Emery et al., 2019), they are 118 also significantly affected by soil substrates and conditions (Morrissey et al., 2014a; 119 von Sperber et al., 2017; Abdul-Aziz et al., 2018). Temperature, however, was identified 120 as an important factor accounting for the variations in GHG emissions in coastal 121 marshes (Abdul-Aziz et al., 2018). Soil substrates availabilities and microbial processes 122 for GHG production can be favored by higher temperature (Walter and Heimann, 2000; 123 Yvon-Duroche et al., 2014). Climatic temperatures can affect soil properties and alter 124 microbial communities (Kirwan et al., 2009; von Sperber et al., 2017), thereby affecting 125 GHG emissions (Nahlik and Mitsch, 2011). Therefore, the effects of climatic 126 temperature variability on GHG emissions may become stronger at a continental scale. 5 / 36 127 The geographical patterns of GHG emissions along a latitudinal gradient, however, 128 remain unclear at the continental scale, which may limit the predictions of climate 129 change models. This kind of information can highlight an understanding of how global 130 warming may influence GHG emissions. More studies of GHG emissions across 131 climatic gradient are therefore required to determine the importance of climatic 132 temperature (the 30-year average) and soil characteristics on the geographical variation 133 of GHG emissions. 134 Coastal marshes are important ecosystems in China, with an area of 5.8×104 km2 135 (Guan, 2012), representing various climatic types and a long latitudinal gradient from 136 temperate to tropical monsoon climates. Mean month air temperature (2011-2015) 137 along the gradient is highly variable in winter (–20 to +20 °C) but less variable in 138 summer (28–35 °C). Mean annual temperature (MAT) (1985-2015) from north to south 139 along the gradient increases from 9.0 to 22.9 °C, which can drive the variation in GHG 140 emissions. In addition, the climatic temperature can shape microbial communities and 141 affect soil substrates availabilities, further affecting GHG emissions along the coastal 142 marshes. The scale of this climatic gradient is ideal for predicting changes in GHG 143 emissions under scenarios of global warming and for identifying associated 144 environmental consequences on the global scale. We thus characterized these coastal 145 marshes with divergent climatic types to answer three questions. (1) Do CH4, N2O and 146 CO2 emissions have a latitudinal gradient because the temperature increases from the 147 temperate to the subtropical coastal marshes? (2) What is the role of the main soil 148 variables individually and together with temperature in controlling the emissions of 149 CH4, N2O and CO2 from the coastal marshes across the climatic gradient? (3) Do coastal 150 marshes located in temperate climate have stronger feedbacks to global warming than 151 marshes in subtropical climate? We addressed these questions by investigating 11 6 / 36 152 coastal marshes along a latitudinal gradient. We measured the emissions of CH4, N2O 153 and CO2 and soil and climatic variables to identify the crucial factors controlling the 154 emissions. Our study can enable the understanding of GHG emissions mediated by 155 climate and the effect of global warming on emissions at a large geographical scale. 156 157 2. Materia and methods 158 2.1. Study area and sample collection 159 The coastal marshes in China span from temperate to tropical climates (approximately 160 20–40°N, 108–124°E) (Figure 1). These marshes across a broad geographical region 161 are characterized by the irregularal and regular semi-diurnal tides and divergent 162 environmental conditions (Guan, 2012). The climate, especially the temperature, varies 163 greatly along the latitudinal gradient. The sample sites were located in the cities of 164 Dandong (DD), Tangshan (TS), Weifang (WF), Qingdao (QD), Dafeng (DF), Shanghai 165 (SH), Wenzhou (WZ), Fuzhou (FZ), Shantou (ST), Zhuhai (ZH) and Beihai (BH). 166 These sites spanned two climatic zones: DD, TS, WF, QD and DF have a temperate 167 monsoon climate, and SH, WZ, FZ, ST, ZH and BH have a subtropical monsoon climate. 168 More detail information on the sites and climatic temperatures is provided in Table 1. 169 We collected gas and soil samples from the low-tide bare flat of each site on 4–20 170 July and 5–21 December 2016. Coastal marshes vegetation had a great influence on 171 GHGs emissions. There were more than five vegetation types at some sites, so it was 172 very difficult to collect gas and soil samples. The precipitation and monsoons between 173 the study sites were different due to the large climatic gradient. However, the bare flat 174 in the coastal marsh was permanently water saturated and strongly affected by the tides. 175 Therefore, we only investigate the bare flat after the ebb tide to compare the GHGs 176 emissions under the consistent settings and methodologies. Three independent plots of 7 / 36 177 the low-tide bare flat were investigated at each sampling site. Gas samples for GHG 178 analyses were collected using static chambers (Li et al., 2019). The chambers were 179 made of polypropylene with a diameter of 30 cm and a height of 50 cm. When the lid 180 was closed, it compressed a rubber gasket cemented to its underside against a horizontal 181 flange at the top of the base walls, thus providing a gas seal (Li et al., 2019). The 182 chambers were inserted in the coastal marshes, maintaining an airtight system. Gas 183 samples (50 mL) were collected at 0, 2 and 4 h at the bare flat after the ebb tide using a 184 syringe. The samples were then immediately transferred to pre-evacuated airtight gas 185 bags (Dalian Delin Gas Packing Co., Ltd., China). After collection of gas samples, soil 186 (10 cm deep) samples for determination of geochemical characteristics were also taken 187 with PVC tubes from the sampling plots of each site. Totally, 99 gas samples and 33 188 soil samples were collected for each seasonal survey, and they were stored in coolers 189 with ice and transported to the laboratory within two days. The materials for collecting 190 the pore water were not prepared, thus the soil pore water was not collected during the 191 sampling. Air temperature (AT) near the marshes was determined using a temperature 192 sensor (U333S, UTI-T Instrument Corp., Guangdong, China), soil redox potential (Eh) 193 at a depth of 10 cm was measured using an ORP meter (AZ8551, AZ Instrument Corp., 194 Taiwan, China), and surface water salinity was determined using a conductivity meter 195 (Model 30, YSI Inc., Yellow Springs Instrument, Yellow Springs, OH, USA). 196 197 2.2. Measurement of soil parameters 198 The soil of 0-10 cm was homogenized to measure the soil properties. Soil bulk density 199 was measured from core samples (Yuan et al., 2015). Water content was determined 200 from the weight lost from fresh soil after drying at 60 °C for 72 h. Soil pH was measured 201 using a pH meter (Mettler-Toledo, Zurich, Switzerland) after soil was mixed with CO2- 8 / 36 202 free water at a ratio (w/v) of 1:2.5 and. Soil NH4+ and NOx− (including NO3− and NO2−) 203 were extracted from 5 g of soil with 25 mL of 2 mol L–1 KCl, and the extraction solution 204 was filtered through a 0.45-µm membrane filter (Whatman, Clifton, USA). Their 205 concentrations were determined using a continuous-flow analyzer (SAN++, Skalar 206 Analytical B.V., Breda, Netherlands). Dissolved organic C (DOC) and sulfate (SO42–) 207 were extracted from 25 g of soil with 50 mL of deionized water, and the supernatant 208 was filtered through a 0.45-µm membrane filter (Whatman, Clifton, USA). DOC and 209 SO42– concentrations were measured using a Shimadzu C analyzer (TOC Vcph, 210 Shimadzu, Kyoto, Japan) and an ion chromatograph (ICS-2000, Dionex, Camberley, 211 UK), respectively. After 300 g of freeze-dried soil was acidized with 1 mol L–1 HCl, 212 total organic C (TOC) and total N (TN) contents were determined using a CN Elemental 213 Analyzer (Vario, Elementar, Hanau, Germany) and δ13C and δ15N enrichments were 214 measured using a Flash Elemental Analyzer connected to an Isotope Ratio Mass 215 Spectrometer (Thermo Fisher Scientific Inc., Bremen, Germany). MBC and MBN 216 contents were determined using chloroform fumigation (Vance et al., 1987; Brookes et 217 al., 1985). 218 219 2.3. Determination of GHG emissions and GWP 220 CH4, N2O and CO2 concentrations in gas samples were analyzed using a gas 221 chromatograph (GC-2014, Shimadzu, Kyoto, Japan) equipped with a flame ionization 222 detector (FID) and electron capture detector (ECD). The column, FID and ECD 223 temperatures were 50, 200 and 250 °C, respectively (Li et al., 2019). The concentrations 224 of gas standards were analyzed after every 10 samples to ensure the accuracy of 225 detection. The GHG emissions rates were estimated from 0 to 2 h and from 2 to 4 h, 226 respectively, on the basis of equation (1): : 9 / 36 𝐶𝑒𝑛𝑑− 𝐶𝑖𝑛𝑖𝑡 𝑉 ×𝑃 227 𝐸= ×𝐴× (1) (𝑡𝑒𝑛𝑑− 𝑡𝑖𝑛𝑖𝑡 )×𝑆 𝑅×𝑇𝑘 228 where E is the emission of CH4, N2O or CO2 (mg m–2 h−1); Cinit and Cend are the initial 229 and final concentrations of CH4, N2O or CO2 (µmol L−1), respectively; A is the molar 230 weight of CH4, N2O or CO2 (g mol−1); V is the chamber volume (L), P is the normal 231 atmospheric pressure (1 atm = 1013.15 hPa); R is the common gas constant (0.082056 232 L atm K-1 mol-1); Tk is the absolute temperature at the time of gas sampling (K); S is the 233 chamber area (m2), and tinit and tend are the start and end times of the samplings (h), 234 respectively; The molar weights of CH4, N2O and CO2 are the common gas constants 235 at normal atmospheric pressure (16, 44 and 44 g mol−1 for CH4, N2O and CO2, 236 respectively). The average of the two time-interval measurements were used as the 237 GHG emissions rates. 238 Annual emissions of CH4, N2O and CO2 were calculated as: 1 239 T= × (∑11 11 𝑖=1 𝐸𝑠𝑢𝑚𝑚𝑒𝑟+∑𝑖=1 𝐸𝑤𝑖𝑛𝑡𝑒𝑟 )×t×S (2) 2×11 240 where T is the total annual emission of CH4, N2O or CO2 (Tg y–1); Esummer and Ewinter 241 are the emissions of CH4, N2O or CO2 (mg m–2 h−1) in summer and winter, respectively; 242 t is the time (365 d) and S is the total area of the marshes (5.8×104 km2). 243 Annual C and N exports from the CH4, N2O and CO2 emissions were calculated as: 12 12 244 C== × 𝑇𝐶𝐻4 + × 𝑇𝐶𝑂2 (3) 16 44 28 245 N= × 𝑇𝑁2𝑂 (4) 44 246 where C is the annual C export by CH4 and CO2 emissions (Gg y–1); N is the annual N 247 export by N2O emissions (Gg y–1); TCH4 is the total annual emission of CH4 of Chinese 248 coastal marshes (Tg y–1); TCO2 is the total annual emission of CO2 of Chinese coastal 249 marshes (Tg y–1); TN2O is the total annual emission of N2O of Chinese coastal marshes 250 (Tg y–1). 10 / 36 251 GWP (Mg CO2eq ha–1 y–1) of the marshes over a 100-year horizon for evaluating the 252 environmental implications of GHGs was calculated as (Neubauer and Megonigal, 253 2015): 254 GWP = (ECH4 ×25 + EN2O×298 + ECO2) ×t×104 (5) 255 where ECH4, EN2O and ECO2 are the mean emissions of CH4, N2O and CO2 (mg m–2 h−1), 256 respectively; t is the time (365 d) and ×104 is a factor for converting ha to m2. 257 258 2.4. Bacterial abundance 259 Genomic DNA was extracted from 0.25 g of soil using the Power Soil DNA Isolation 260 Kit (Mo Bio, USA). The extracted DNA and the products from polymerase chain 261 reactions (PCR) were then verified using 1.0% agarose gel electrophoresis. Bacterial 262 abundance was estimated using qPCR targeting the 16s rRNA gene amplified with 263 primers 341F (5'–CCT ACG GGA GGC AGC AG–3') and 534R (5'–ATT ACC GCG 264 GCT GCT GGC A–3') (Philippot et al., 2011). The reaction mixtures included 12.5 μL 265 of Maxima SYBR Green/RoxqPCR Master Mix (Fermentas, Lithuania), 1 μL of each 266 primer (10 μmol L–1, Sangon, China), 1 μL of template DNA and 9.5 μL of sterile 267 ddH2O. Thermal-cycling conditions for the qPCR were: 50 °C for 2 min, 95 °C for 15 268 min, followed by 35 cycles of 95 °C for 15 s, 60 °C for 30 s and 72 °C for 30 s. The 269 qPCR was run in triplicate using an ABI 7500 Sequence Detection System (Applied 270 Biosystems, Canada) with the SYBR Green qPCR method. Average amplification 271 efficiency was 99.3%, and gene copies (log10) were significantly correlated with the 272 threshold cycle number (CT) (r2 = 0.991). The bacterial abundances in the soils were 273 calculated using standard curves. 274 275 2.5. Statistical analyses 11 / 36 276 GHG emissions, bacterial abundance and environmental properties were compared 277 among the sampling sites using a one-way analysis of variance (ANOVA) followed by 278 Tukey's honestly significant difference test. Significant differences in GHG emissions 279 between the temperate and subtropical climates and between December and July were 280 identified by independent-samples t-tests at P < 0.05. The interactive effects of site and 281 seasonal variabilities in GHG emissions were identified using a two-way ANOVA. A 282 partial-correlation analysis controlling for climatic temperature was used to describe 283 the relationships between GHG emissions and soil characteristics. Stepwise multiple 284 regression analysis was used to identify the most important factors affecting the 285 variabilities in GHG emissions across the coastal marshes. The criteria for inclusion in 286 the stepwise multiple regression analysis was F to enter of >0.1 and F to remove of 287 <0.05 (Harrison and Matson, 2003). We performed a principal component analysis 288 (PCA) with all soil variables and gas emissions as the variables for detecting patterns 289 of the variable relationships ordering samples as functions of site and/or season. A 290 structural equation model (SEM) in the path analysis was constructed to identify the 291 direct and indirect effects of the soil and climatic variables on the GHG emissions. 292 293 3. Results 294 3.1. Site descriptions 295 Abiotic and biotic characteristics of the coastal marshes are provided in Table 2. Soil 296 density varied slightly across the study sites, ranging from 1.13 to 1.41 g cm–3. Water 297 contents ranged from 50.3 to 69.2 %. The soil had a neutral pH, varying between 7.72 298 and 8.07. Eh varied greatly among the study sites, ranging from -32 to -158 mV. The 299 salinity of the seawater ranged from 11.8 to 27.9 ‰. TOC and TN contents ranged 300 largely from 1.09 to 10.2 mg g–1 and from 0.14 to 1.15 mg g–1, respectively, with C/N 12 / 36 301 ratios of 6.4–25.8. DOC increased from high to low latitudes, with contents of 0.11– 302 0.93 mg g–1. MBC and MBN contents were generally higher at the subtropical than 303 temperate sites, varying between 11.7 and 113 µg g–1 and between 2.27 and 17.5 µg g– 1 304 , respectively. NH4+ and NOx– contents were higher at the subtropical than the 305 temperate sites. NH4+ contents ranged from 1.53 to 12.1 µg g–1, contributing 68–91% 306 of total dissolved inorganic N (DIN). NOx– contents ranged from 0.61 to 2.13 µg g–1, 307 accounting for 9–32% of total DIN. SO42– contents varied greatly, ranging from 0.21 to 308 0.96 mg g–1. δ13C and δ15N varied between -26.3 and -20.3‰ and between 4.35 and 309 7.15‰, respectively. Bacterial abundances varied greatly across the sites, ranging from 310 1.1× 109 to 11.3 × 109 copies g–1. 311 312 3.2. GHG emissions 313 Emissions of CH4 ranged from 23.6 to 211 µg m–2 h–1 and from 29 to 986 µg m–2 h–1 in 314 December and July, respectively (Table 3). Emissions of N2O ranged from 2.0 to 103 315 µg m–2 h–1 and from 3.9 to 110 µg m–2 h–1 in December and July, respectively. Emissions 316 of CO2 were 1000-fold higher than the emissions of CH4 and N2O, ranging from 9.9 to 317 310 mg m–2 h–1 and from 50.9 to 153 mg m–2 h–1 in December and July, respectively. 318 Emissions of CH4, N2O and CO2 varied significantly spatially among the study sites 319 (one-way ANOVA, P < 0.001, Table 4). Emissions of CH4 and N2O varied significantly 320 across seasons (one-way ANOVA, P < 0.001), but emissions of CO2 did not 321 significantly differ (one-way ANOVA, P = 0.942). The interaction between site and 322 season, however, influenced the emissions of the three gases (two-way ANOVA, P < 323 0.001). 324 Emissions of CH4, N2O and CO2 increased from high to low latitudes in both 325 December and July (Figure 2), indicating an important geographical pattern of GHG 13 / 36 326 emissions from the marshes. Emissions were significantly higher at the subtropical than 327 temperate sites in both December and July (t-test, P < 0.05) (Figure 3). Emissions of 328 CH4 at the subtropical sites varied significantly between December and July (t-test, P 329 < 0.05). Emissions of CO2 at the temperate sites were significantly higher in July than 330 in December (t-test, P < 0.05). N2O emissions, however, did not significantly differ 331 seasonally at the temperate or subtropical sites (t-test, P > 0.05). 332 333 3.3. Influences of climatic temperature and biogeochemical variables 334 AT was significantly correlated with the emissions of CH4 (r = 0.66, P < 0.01), N2O (r 335 = 0.59, P < 0.01) and CO2 (r = 0.57, P < 0.01) at all sites (Figure 4a-c). MAT was also 336 correlated with CH4 (r = 0.80, P < 0.01), N2O (r = 0.85, P < 0.01) and CO2 (r = 0.66, P 337 < 0.01) emissions (Figure 4d-f). CH4 emissions were negatively correlated with Eh (r 338 = -0.87, P < 0.001) and δ13C (r = -0.45, P = 0.036). CH4 emissions were strongly 339 positively correlated with C/N and TOC, DOC, TN, NH4+, NOx–, MBC and MBN 340 contents (r > 0.42, P < 0.05). Eh and TOC, TN, NOx– and MBC contents, however, 341 remained significantly correlated with CH4 emissions after controlling for climatic 342 temperature in a partial-correlation analysis (lrl > 0.51, P < 0.01) (Table 5). N2O 343 emissions were correlated negatively with Eh (r = -0.54, P = 0.009) and strongly 344 positively with TOC, DOC, MBC and MBN contents (r > 0.41, P < 0.05). Eh and DOC 345 contents affected N2O emissions the most along the latitudinal gradient after controlling 346 for climatic temperature. CO2 emissions were negatively correlated with Eh (r = -0.46, 347 P = 0.03) and pH (r = -0.58, P = 0.004) in the partial-correlation analysis. The C/N ratio, 348 TOC, DOC, MBC and MBN contents and bacterial abundance were strongly correlated 349 with CO2 emissions (r > 0.44, P < 0.05). Only pH and bacterial abundance, however, 350 remained significantly correlated with CO2 emissions (r = -0.60, P = 0.005; r = 0.49, P 14 / 36 351 = 0.029, respectively) after a partial-correlation analysis to account for climatic 352 temperature. 353 The stepwise analysis identified the important predictors of the variations in the GHG 354 emissions along the latitudinal gradient (Figure 5). MBC contents, Eh and MAT were 355 the most important predictors of the variations in CH4 emission at all sites, explaining 356 75.4%, 11% and 4.1% of the total variance, respectively. MAT, AT, Eh and NOx– 357 contents were the most important factors affecting N2O emissions, with MAT as the 358 best predictor, explaining 70.9% of the total variance along the gradient. MAT was the 359 most crucial factor controlling CO2 emissions, explaining 40.5% of the total variance, 360 and both pH and AT also affected CO2, explaining 14% and 11.3% of the total variance 361 across the sites, respectively. The PCA clearly indicated that the first axis was mainly 362 loaded by the C and N contents, temperature, Eh and GHG emissions (Figure 6). The 363 samples from the subtropical sites were loaded toward higher temperatures, C and N 364 contents and GHG emissions and toward lower Eh, salinity and SO42– content. The 365 samples collected in winter and summer were differently loaded, with no clear patterns 366 among the sites in the separation of the winter and summer samples. Climatic 367 temperature also had a strong indirect positive effect on the emissions of CH4 and CO2 368 by affecting Eh, bacterial abundance and TOC, DOC, MBC and MBN contents (Table 369 S1), and TOC contents had an indirect negative influence on N2O emissions (Figure 7). 370 371 3.4. Annual C and N exports and GHG emissions 372 The estimated GWP for the marshes ranged from 2.96 to 21.9 Mg CO2eq ha–1 y–1 across 373 the sites (Table S2), with a mean GWP of approximately 10.0 Mg CO2eq ha–1 y–1. 374 Annual emissions of N2O, CH4 and CO2 from the marshes (total area of 5.8×104 km2) 375 were estimated at 1.80–44.41 Gg y–1, 14.89–304 Gg y–1 and 1.56×104–1.17×105 Gg y– 15 / 36 1 376 , respectively. Annual C and N exports from the emissions were 1.64±0.32×104 Gg y– 1 377 and 9.14±2.89 Gg y–1, respectively. Based on the soil densities at a depth of 10 cm, 378 total area of the marshes and mean contents of TOC and DIN from the homogenized 379 soils (0–10 cm), the estimated C and N storages in Chinese coastal marshes were 380 approximately 3.89±0.43×104 Gg and 56.05±4.49 Gg, respectively. The annual losses 381 of C and N from the GHG emissions were therefore about 42% and 16% of the C and 382 N storages, respectively, regardless of the sequestration of C and N, inorganic carbon 383 and organic nitrogen. 384 Increases in the emissions of CH4, N2O and CO2 under global-warming scenarios 385 of 2 and 4 °C by 2100 (IPCC, 2007) were estimated using the fitting models at 2.20– 386 5.43 µg m–2 y–1, 1.86–4.78 µg m–2 y–1 and 4.32–6.10 mg m–2 y–1 (Table 6), respectively 387 (Figure 4d-f). These enhanced emissions thus represented approximately 4.62–11.4%, 388 27.5–70.6% and 9.13–12.9% of the current mean emissions of CH4 (47.6 µg m–2 y–1), 389 N2O (6.77 µg m–2 y–1) and CO2 (47.3 mg m–2 y–1), respectively, for the temperate coastal 390 marshes and approximately 0.67–1.66%, 4.03–10.3% and 2.98–4.21% of the current 391 mean emissions of CH4 (328 µg m–2 y–1), N2O (46.2 µg m–2 y–1) and CO2 (145 mg m–2 392 y–1), respectively, for the subtropical coastal marshes. 393 394 4. Discussion 395 4.1. Geographical variation along the latitudinal gradient 396 Many studies have reported GHG emissions from coastal marshes (Yuan et al., 2015; 397 Abdual-Aziz et al., 2018; Emery et al., 2019), but the geographical variations of the 398 emissions at a continental scale remain unclear. We examined the geographical 399 distributions and seasonal variations in emissions of CH4, N2O and CO2 along a 400 latitudinal gradient of Chinese coastal marshes. Although some sites were close to the 16 / 36 401 cities (Table 1), soil C and N pools and substrates availabilities may vary greatly and 402 may be affected by temperature along the coastal marshes. The soil respiration rates and 403 microbial communities for GHGs production are also the long-term results of sampling 404 site temperature (Wang et al., 2016; von Sperber et al., 2017). Therefore, the variations 405 in GHG emissions were not controlled by the urban sizes but by the climatic 406 temperature and soil variables of the coastal marshes. The geographical patterns of the 407 emissions increased from high to low latitudes in both summer and winter. AT and MAT 408 were significantly correlated with the emissions of CH4, N2O and CO2 in all coastal 409 marshes (Figure 4). A higher capacity of biological activity under higher temperatures 410 in wetland areas, where water is not limited, is expected to be associated with higher C 411 and N contents (both depending on biological fixation) in marshes (Emery et al., 2019). 412 The higher microbial biomass and higher levels of C and N transformation processes 413 could thus increase the CO2, CH4 and N2O emissions (Morrissey et al., 2014a; Abdual- 414 Aziz et al., 2018). Therefore, our statistical analyses suggest that the high temperature 415 and associated high MBC and MBN contents were the underlying mechanisms 416 contributing to high GHG emissions. Some of previous studies have consistently 417 reported that temperature was the most important factor affecting microbial respiration 418 and GHG emissions in coastal marshes (Xie et al., 2014; Abdual-Aziz et al., 2018). The 419 emissions of CH4, N2O and CO2 in our study were consistently higher in the subtropical 420 than temperate marshes (Figure 3), even though AT did not differ significantly between 421 them in July (P > 0.05), indicating that potential GHG emissions are the long-term 422 effects of temperature. The dependency of GHG emissions on temperature is also likely 423 attributed to that the soil microbial respiration and microbial communities for GHGs 424 production are the results of the long-term effects of site temperature (von Sperber et 425 al., 2017; Nottingham et al., 2019). The difference in temperature in latitude 24 and 21 17 / 36 426 was less variable between December and July. The MAT was also comparable with air 427 temperature at sites ZH and BH where TOC, DOC, MBN and C/N were higher in 428 December than in July, which could contribute to the higher emissions of CO2 and N2O 429 at these sites. Interestingly, the summer samples from all temperate sites were more 430 towards the right along the PC1 axis (thus toward conditions of the subtropical sites), 431 as expected, whereas a comparison between the summer and winter samples from the 432 subtropical sites did not indicate this general trend (Figure 6). These results therefore 433 suggest that the mean annual temperature in our study was the best predictor of the 434 geographical variation in GHG emissions across the coastal marshes. Soil C and N 435 availabilities, MBC and MBN could also explain the variation in GHG emissions across 436 the coastal marshes. 437 438 4.2. Influences of temperature and soil properties on GHG emissions 439 4.2.1. Factors affecting CH4 emissions 440 CH4 emissions were significantly higher in July than in December, because high 441 contents of labile organic C from the increased decomposition of organic matter can 442 further favor CH4 production at high temperatures (Gelesh et al., 2016). 443 Methanogenesis, responsible for CH4 production, is also substantially driven by 444 temperature (Martin and Moseman-Valtierra, 2017) and is highly sensitive to increasing 445 temperatures (Abdual-Aziz et al., 2018), leading to higher emissions of CH4 in 446 subtropical coastal marshes and warming seasons. Temperature had an indirect effect 447 on the CH4 emissions, because temperature strongly influences redox reactions and C 448 availability (Figure 7a). The anaerobic decomposition of organic matter is the main 449 pathway of CH4 production, but the content and quality of organic C strongly affects 450 CH4 emission by altering the abundance and community structure of methanogenic 18 / 36 451 archaea (Wagner et al., 2005). Yuan et al. (2015) found a strong link between CH4 452 emissions and DOC contents in coastal marshes, further suggesting that degradable C 453 was the main contributor to CH4 production. Our study indicated that both TOC and 454 DOC contents significantly influenced the CH4 emissions. A partial-correlation 455 analysis controlling for temperature suggested that DOC contents were not correlated 456 with CH4 emissions, likely because DOC contents covaried with temperature changes. 457 Methanogenic archaea are favored in strongly reducing and anaerobic conditions, 458 CH4 emissions are thus generally higher in the environments with lower Eh. The strong 459 correlation between Eh and CH4 emission thus likely indicated that soil Eh was an 460 important factor for CH4 emissions in the coastal marshes. Previous studies also 461 reported that salinity had a negative effect on CH4 emissions in marshes with moderate 462 to high salinities (Vivanco et al., 2015; Abdul-Aziz, et al., 2018), the CH4 emissions 463 from the coastal marshes were strongly inhibited by the high salinity, indicating a 464 negative relationship of salinity with CH4 emission. Sulfate-reducing bacteria can 465 outcompete methanogens, and sulfate reduction can oxidize CH4 in anaerobic 466 environments, limiting CH4 emissions (Weston et al., 2014). Coastal marshes are 467 enriched in sulfates, so SO42– content is likely a strong factor leading to low CH4 468 emissions. It has been reported that organic matter inputs promote the microbial growth 469 rate and further increase the amount of microbial C anabolism (Chen et al., 2020). In 470 addition, higher microbial carbon anabolism implies more C converted to biomass, 471 leading to higher MBC (Spohn et al., 2016; Chen et al., 2020). The more microbes can 472 enhance the anaerobic decomposition of organic matter and further increase CH4 473 production (Morrissey et al., 2014a). Therefore, MBC content was an important factor 474 accounting for the variability of CH4 emissions across the coastal marshes because 475 MBC is affected by microbial C anabolism and soil C pools (Chen et al., 2020). 19 / 36 476 477 4.2.2. Factors affecting N2O emissions 478 Warming is predicted to intensify the metabolic biokinetics of enzymatic activity for 479 nitrification and denitrification (Greaver et al., 2016). Both nitrification and 480 denitrification are sensitive to the increase in temperature, with an optimal temperature 481 of about 20–25 °C (Martens-Habbena et al., 2009; Canion et al., 2014). AT in December 482 and MAT in subtropical zones are about 20 °C, but are <10 °C in temperate zones. The 483 modified nitrification and denitrification driven by the changes in temperature, however, 484 may create environmental conditions for shifting microbial activity between NH4+ and 485 NO3– (Martens-Habbena et al., 2009), further increasing N2O emissions. Increases in 486 temperature are also expected to enhance the amount of dissolved organic C (Laudon 487 et al., 2012), promoting a C-cycle feedback to N transformation, including 488 denitrification as a heterotrophic process. N2O emissions are therefore higher in 489 subtropical coastal marshes and warm seasons, because denitrification can increase 490 with the amount of dissolved organic C. Xiong et al. (2017) reported that organic C was 491 the primary factor affecting denitrification because of its yields of respiration energy. 492 In our study, however, TOC content had an indirect negative effect on N2O emissions 493 (Figure 7b), likely because C transformations enhanced by temperature and C 494 availability could limit N cycling processes (Craine et al., 2007; Nottingham et al., 495 2019). 496 Denitrification is active in reducing environments with low Eh (Stelzer and Scott, 497 2018), thereby leading to higher N2O emissions. Eh in our study significantly 498 influenced N2O emissions across the coastal marshes, suggesting that low Eh was 499 favorable for N2O emission. NOx– also had an important effect on the N2O emissions, 500 because NOx– is the electron acceptor for denitrification. Denitrifiers, however, can use 20 / 36 501 the shift in substrates from NOx– to N2O when environmental NOx– content is below 502 the threshold of ~1 µg g–1 (Ryden, 1983; Yuan et al., 2015). The NOx– contents in most 503 of our sampling sites were similar to or below the threshold for denitrifiers, likely due 504 to leaching by the tidal water, so N2O may be preferentially used in denitrification, 505 which can lower N2O emissions. Nitrification has been suggested to both directly and 506 indirectly affect N2O emission (Yuan et al., 2015). Nitrification can produce N2O by 507 reducing NH2OH to NO and then to N2O (Hu et al., 2012). The community of ammonia- 508 oxidizing archaea can use NH4+ to produce NOx– even at extremely low NH4+ contents 509 (Martens-Habbena et al., 2009). NH4+ content could be a potential factor affecting N2O 510 emissions, although the relationship between N2O emissions and NH4+ contents was 511 not significant (Table 5). 512 513 4.2.3. Factors affecting CO2 emissions 514 Temperature is an important factor affecting CO2 release, because the increase in 515 temperature can enhance the yield of CO2 from microbial respiration and organic-C 516 decomposition (Bond-Lamberty and Thomson, 2010; Lu et al., 2013). More 517 importantly, microbial respiration can be enhanced exponentially from 10 to 20 oC and 518 further yield more CO2 (Wang et al., 2016; Nottingham et al., 2019). The dependence 519 of bacterial and fungal growth on temperature has been described in the tropical forests 520 soils that is a very different system from the marshes, but the microbial communities 521 were consistent with long-term adaptation to temperature (Nottingham et al., 2019). 522 The range of MAT between the temperate and subtropical climates in our study was 523 appropriately 10 °C, which strongly determined the distinct geographical patterns of 524 CO2 emissions between the temperate and subtropical coastal marshes. The air 525 temperature along the gradient is highly variable in winter but less variable in summer. 21 / 36 526 The MAT increases significantly from north to south sites along the climatic gradient. 527 The emissions of GHGs also showed increasing gradients from north to south sites. 528 Therefore, the MAT could better explain the variations of GHGs emissions than air 529 temperature along the climatic gradient (Figures 4 and 5). The availability of organic 530 C also covaried with increasing temperature, and the TOC and DOC were not 531 significantly correlated to CO2 emissions after controlling for temperature (Table 5), so 532 temperature indirectly influenced CO2 emissions by affecting soil organic carbon 533 (Figure 7c). The CO2 emissions thus had a strong latitudinal gradient, increasing from 534 high to low latitudes, mainly because microbial respiration can be enhanced by 535 temperature. 536 The quality and quantity of organic C can strongly affect CO2 emissions (Voigt et 537 al., 2017; Emery et al., 2019). The TOC and DOC contents in our study, however, did 538 not strongly affect the CO2 emissions when a partial-correlation analysis controlled for 539 climatic temperature. This relationship suggests a crucial dependence of organic-C 540 availability on temperature across the marshes, thereby obscuring the important 541 influences of TOC and DOC on CO2 emission. The leaching and decomposition of 542 organic C are more intensified by higher temperatures in subtropical than temperate 543 climates, delivering more DOC for microbial respiration (Wang et al., 2016). Soil pH 544 was negatively correlated with CO2 emission (Table 5, Figure S1), but soil Eh may not 545 be the important factor affecting CO2 emission. Soil Eh is strongly affected by soil 546 water and tidal flooding (Li et al., 2015). The low soil Eh means low oxygen content, 547 which may inhibit the aerobic microbial respiration and further decrease the CO2 548 emissions. The C/N ratio was also an important variable affecting the CO2 emissions, 549 further suggesting that microbial respiration can increase in the environments enriched 550 in organic C. Bacterial abundance and MBC and MBN contents were positively 22 / 36 551 correlated with CO2 emissions (Table 5, Figure S1), suggesting that microbial 552 abundance and respiration play important roles in CO2 emission. 553 554 4.3. Implications for global warming and emissions 555 Increasing GHG emissions are expected to have many environmental consequences, of 556 which global warming is the most important for global climate change (Lashof and 557 Ahuja, 1990; Meinshausen et al., 2009). Global warming would in turn increase GHG 558 emissions by stimulating biogeochemical cycling due to higher temperatures (Greaver 559 et al., 2016), thus positively feeding back on GHG emission. Coastal marshes are highly 560 sensitive to global climate change and are predicted to play an important role in GHG 561 emissions (Martin and Moseman-Valtierra, 2017; Abdul-Aziz, et al., 2018). Global 562 warming shows a great challenge to predict GHG emissions from coastal marshes. 563 Some studies, however, have reported that increasing temperatures can tend to enhance 564 GHG emissions in coastal marshes (Xie et al., 2014; Yang and Silver, 2016; Abdul-Aziz, 565 et al., 2018). 566 Assuming that the latitudinal gradient was a main consequence of temperature 567 changes and that local environmental variables remained constant across the marshes, 568 we can conservatively estimate the response of GHG emissions to future increases in 569 global temperature. The increased emissions of CH4, N2O and CO2 in response to 570 proposed increases in annual temperature of 2–4 °C by 2100 (IPCC, 2007) represent 571 4.62–11.4%, 27.5–70.6% and 9.13–12.9% of the current mean emissions for temperate 572 coastal marshes, respectively. The increases in the emissions of CH4, N2O and CO2 573 would account for 0.67–1.66%, 4.03–10.3% and 2.98–4.21% of the current mean 574 emissions for subtropical coastal marshes, respectively. These results are consistent 575 with the report that climatic warming has already altered the extent of biogeochemical 23 / 36 576 processes (Lu et al., 2013; Greaver et al., 2016; Voigt et al., 2017), possibly contributing 577 to increased GHG emissions. The increasing amplitudes of emissions induced by global 578 warming in the marshes were larger for N2O than CH4 and CO2. The increases in the 579 amplitudes of GHG emissions were larger in the temperate than subtropical marshes 580 (Table 6), indicating that the temperate marshes had stronger feedbacks to the 581 increasing temperatures. It has been reported that the soil microbial respiration rates 582 incubated under the increasing temperature increased logarithmically with increasing 583 latitude and the sensitivity of soil microbial respiration to the increasing temperatures 584 is higher in the soils from locations with lower temperature relative to high temperature 585 (Wang et al., 2016). The GHG emissions from low temperature soils showed high 586 temperature sensitivity relative to high temperature soils (Zou et al., 2018). We 587 therefore propose that coastal marshes at high latitudes with lower climatic 588 temperatures are anticipated to emit more GHGs under future climatic warming. 589 GHG emissions in coastal marshes are currently increasing globally and contribute 590 greatly to global emissions (Yang et al., 2016; Abdul-Aziz et al., 2018; Emery et al., 591 2019). The total area of Chinese coastal marshes is 5.8×104 km2, equivalent to 0.39‰ 592 of the global terrestrial area (1.49 ×108 km2). The CH4 emissions from these marshes 593 were estimated at approximately 0.119 Tg CH4 y–1, which represent approximately 0.62‰ 594 of annual global CH4 emissions from terrestrial environments (192 Tg CH4 y–1, Tian et 595 al., 2015). N2O emissions from the coastal marshes were lower than CH4 emissions, 596 estimated at a rate of 0.014 Tg N2O y–1, equivalent to approximately 0.70‰ of annual 597 global N2O emissions from terrestrial environments (20 Tg N2O y–1, Zaehle et al., 2011). 598 CO2 emissions were the largest of the three gases, about 500- and 3600-fold larger than 599 for CH4 and N2O, respectively. The total emissions of CO2 (0.051 Pg CO2 y–1) from the 600 marshes accounted for approximately 6.60‰ of annual global CO2 emissions from 24 / 36 601 inland waters (7.7 Pg CO2 y–1, Raymond et al., 2013). These GHG emissions from the 602 marshes are large, considering their small surface area, reinforcing the understanding 603 that coastal marshes are foci of GHGs. These results suggest that Chinese coastal 604 marshes play a critically important role in global GHG emissions and are anticipated to 605 be important to future global climate change. 606 607 4.4. Study uncertainty and future outlook 608 This study highlights the importance of climatic temperature on GHG emissions and 609 the relationship between global climate change and GHG emissions in coastal marshes 610 along the latitudinal gradient. Uncertainty remains, however, about the relative 611 importance of climatic temperature and local conditions on GHG emissions. Salinity is 612 a crucial factor affecting biogeochemical processes at moderate ranges (Morrissey et 613 al., 2014b), which can decrease the role of temperature in GHG emissions (Abdul-Aziz, 614 et al., 2018). Differences in temperature and nutrient availability are highly variable 615 both spatially and temporally, likely contributing to the local variability of annual GHG 616 emissions across the coastal marshes. The GHG emissions and soil C and N contents 617 missed the vertical variation within the changing soil depth, making it uncertain for 618 projecting the contribution of GHG emissions from Chinese coastal marshes to global 619 terrestrial environments. In addition, GHG emissions in our study were determined 620 from the bare flat after the ebb tide, actually the vegetation exhibits a great influence 621 on GHG emissions (Yuan et al., 2015), leading to great uncertainty of the estimates of 622 GHG emissions from coastal wetlands. Multiple factors should thus be considered for 623 predicting GHG emissions in coastal wetlands under a future changing climate and 624 environment. More importantly, the variation in GHG emissions within a study site may 625 be equivalent to the total variability in the emissions along a latitudinal gradient 25 / 36 626 (Bridgham et al., 2013). This variability emphasizes the importance of local 627 environmental parameters other than climatic temperature in determining the 628 geographical variation in GHG emissions from coastal marshes (Abdul-Aziz et al., 629 2018). Our relative large number of samples sites, however, enabled us to clarify a large 630 and consistent latitudinal gradient in GHG emissions that was correlated strongly with 631 climatic temperature. 632 The responses of biogeochemical processes to environmental conditions are 633 inherently complex and nonlinear, so predicting the ecological consequences of global 634 warming is difficult. C and N transformations may compete under changing 635 environments and temperature, facilitating or inhibiting other processes (Nottingham 636 et al., 2019). We used a linear model that was an abstraction of reality and simplified 637 many complex processes for predicting GHG emissions in response to increases in 638 annual temperature. Statistical analyses of experimental data for determining ecological 639 and biogeochemical dynamics in in situ field environments are also difficult. The fitting 640 model in our study suggested that mean annual temperature was crucial for predicting 641 GHG emissions, but 28–67% of the variations in the GHG emissions were attributed to 642 soil variables and not to climatic temperature. Our results for the tropical and 643 subtropical marshes, however, should not be extrapolated, because the optimum 644 temperatures for C and N transformations (e.g. enzymatic dynamics) are about 20–25°C 645 (Canion et al., 2014), equivalent to our MAT. This equivalence suggests that the GHG 646 emissions in the subtropical marshes were less sensitive to the increase in annual 647 temperature relative to temperate marshes. Global warming is also expected to have a 648 more important influence on productivity and biogeochemical processes of northern 649 coastal marshes (Kirwan et al., 2009). These results further indicate divergent responses 650 to global warming of coastal marshes with different climates, potentially affecting the 26 / 36 651 predictions of GHG emissions at the large geographical scales. 652 653 5. Conclusions 654 Emissions of CH4, N2O and CO2 in the Chinese coastal marshes increased along the 655 latitudinal gradient from high to low latitudes. These emissions were highly dependent 656 on the increase in climatic temperature. Soil substrate availabilities and bacterial 657 abundance affected by the temperature were also the key drivers of the geographical 658 variations of GHG emissions. The increase in GHG emission was larger in the 659 temperate than subtropical marshes under the proposed increase in temperatures of 2– 660 4 °C by 2100, suggesting that the feedbacks to global warming were stronger in the 661 temperate than subtropical marshes. Chinese coastal marshes play a crucial role in GHG 662 emissions, representing 0.62–6.60‰ of the global annual emissions from terrestrial 663 environments and inland waters. Our results refine our understanding of the 664 environmental consequences of global climate change to coastal marshes and suggest 665 that these marshes are important for predicting future GHG emissions. 666 667 Acknowledgments 668 This study was supported by the National Natural Science Foundation of China (Grants 669 41725002, 41701548, 41761144062, 41671463 and 41601008), Natural Science 670 Foundation of Fujian Province (Grant Number 2018J01737). We thank two anonymous 671 reviewers for their valuable comments on the early version of this manuscript. We also 672 thank our research team for their assistance with collecting the field samples. J.S. and 673 J. P. were funded by the European Research Council Synergy grant SyG-2013-610028 674 IMBALANCE-P. 675 27 / 36 676 References 677 Abdul-Aziz, O.I., Ishtiaq, K.S., Tang, J., Moseman-Valtierra, S., Kroeger, K.D., 678 Gonneea, M.E., Mora, J., Morkeski, K., 2018. Environmental controls, emergent 679 scaling, and predictions of greenhouse gas (GHG) fluxes in coastal salt marshes. J. 680 Geophys. Res.-Biogeo. 123(7), 2234–2256. 681 Beaulieu, J. J., Nietch, C. T., Young, J. L., 2015. Controls on nitrous oxide production 682 and consumption in reservoirs of the Ohio River Basin. J. Geophys. Res.-Biogeosci. 683 120, 1995–2010. 684 Bond-Lamberty, B., Thomson, A., 2010. Temperature-associated increases in the global 685 soil respiration record. Nature 464, 579–582. 686 Borges, A. V., Darchambeau, F., Teodoru, C. R., Marwick, T. R., Tamooh, F., Geeraert, 687 N., Omengo, F.O., Guérin, F., Lambert, T., Morana, C., Okuku, E., Bouillon, S., 2015. 688 Globally significant greenhouse-gas emissions from African inland waters. Nature 689 Geosci. 8(8), 637–642. 690 Bridgham, S.D., Cadillo-Quiroz, H., Keller, J.K., Zhuang, Q., 2013. Methane emissions 691 from wetlands: biogeochemical, microbial, and modeling perspectives from local to 692 global scales. Global Change Biol. 19(5), 1325–1346. 693 Brin, L.D., Valiela, I., Goehringer, D., Howes, B., 2010. Nitrogen interception and 694 export by experimental salt marsh plots exposed to chronic nutrient addition. Mar. 695 Ecol. Prog. Ser. 400, 3–17. 696 Brookes, P.C., Landman, A., Pruden, G., Jenkinson, D.S., 1985. Chloroform fumigation 697 and the release of soil nitrogen: a rapid direct extraction method to measure microbial 698 biomass nitrogen in soil. Soil Biol. Biochem. 17, 837–842. 699 Canion, A., Overholt, W.A., Kostka, J.E., Huettel, M., Lavik, G., Kuypers, M.M., 2014. 700 Temperature response of denitrification and anaerobic ammonium oxidation rates 28 / 36 701 and microbial community structure in Arctic fjord sediments. Environmen. 702 Microbiol. 16(10), 3331–3344. 703 Chen, X.B., Xia, Y.H., Rui, Y.C., Ning, Z., Hu, Y.J., Tang, H.M., He, H.B., Li, H.X., 704 Kuzyakov, Y., Ge, T.D., Wu, J.S., Su, Y.R., 2020. Microbial carbon use efficiency, 705 biomass turnover, and necromass accumulation in paddy soil depending on 706 fertilization. Agric. Ecosyst. Environ. 292, 106816. 707 Chen, Y.P., Chen, G.C, Ye, Y., 2015. Coastal vegetation invasion increases greenhouse 708 gas emission from wetland soils but also increases soil carbon accumulation. Sci. 709 Total Environ. 526, 19–28. 710 Craine, J.M., Morrow, C., Fierer, N., 2007. Microbial nitrogen limitation increases 711 decomposition. Ecology 88(8), 2105–2113. 712 Emery, H.E., Angell, J.H., Fulweiler, R.W., 2019. Salt marsh greenhouse gas fluxes and 713 microbial communities are not sensitive to the first year of precipitation change. J. 714 Geophys. Res.-Biogeo. 124, 1071–1087. 715 Gelesh, L., Marshall, K., Boicourt, W., Lapham. L., 2016. Methane concentrations 716 increase in bottom waters during summertime anoxia in the highly eutrophic estuary, 717 Chesapeake Bay, U.S.A. Limnol. Oceanogr. 61(S1), S253–S266. 718 Greaver, T.L., Clark, C.M., Compton, J. E., Vallano, D., Talhelm, A.F., Weaver, C.P., 719 Band, L.E., Baron, J.S., Davidson, E.A., Tague, C.L., Felker-Quinn, E., Lynch, J.A., 720 Herrick, J.D., Liu, L., Goodale, C.L., Novak, K.J., Felker-Quinn, E., 2016. Key 721 ecological responses to nitrogen are altered by climate change. Nat. Climate Change 722 6(9), 836–843. 723 Guan, D.M., 2012. Coastal wetlands of China. Marine Press, Beijing, China. 724 Harrison, J., Matson, P., 2003. Patterns and controls of nitrous oxide emissions from 725 waters draining a subtropical agricultural valley. Global Biogeochem. Cy. 17, 1080. 29 / 36 726 Hu, Z., Lee, J.W., Chandran, K., Kim, S., Khanal, S.K. 2012. Nitrous oxide (N2O) 727 emission from aquaculture: a review. Environ. Sci. Technol. 46(12), 6470–6480. 728 Intergovernmental Panel on Climate Change (IPCC), 2007. Fourth Assessment Report, 729 Climate Change 2007: Synthesis Report. Cambridge University Press, Cambridge, 730 UK. 731 Intergovernmental Panel on Climate Change (IPCC), 2013. Climate Change 2013: The 732 Physical Science Basis. Contribution of Working Group I to the Fifth Assessment 733 Report of the Intergovernmental Panel on Climate Change. In T. F. Stocker, and 734 others [eds.], Cambridge Univ. Press. 735 Jeffrey, L.C., Santos, I.R., Tait, D. R., Makings, U., Maher, D.T., 2018. Seasonal drivers 736 of carbon dioxide dynamics in a hydrologically modified subtropical tidal river and 737 estuary (Caboolture River, Australia). J. Geophys. Res.-Biogeo. 123(6), 1827–1849. 738 Kirwan, M.L., Guntenspergen, G.R., d'Alpaos, A., Morris, J.T., Mudd, S.M., 739 Temmerman, S., 2010. Limits on the adaptability of coastal marshes to rising sea 740 level. Geophys. Res. Lett. 37(23). 741 Kirwan, M.L., Guntenspergen, G.R., Morris, J.T., 2009. Latitudinal trends in Spartina 742 alterniflora productivity and the response of coastal marshes to global 743 change. Global Change Biol. 15(8), 1982–1989. 744 Lashof, D.A., Ahuja, D.R., 1990. Relative contributions of greenhouse gas emissions 745 to global warming. Nature 344(6266), 529. 746 Laudon, H., Buttle, J., Carey, S.K., McDonnell, J., McGuire, K., Seibert, J., Shanley, J., 747 Soulsby, C., Tetzlaff, D., 2012. Cross-regional prediction of long-term trajectory of 748 stream water DOC response to climate change. Geophys. Res. Lett. 39(18). 749 Li, X.F., Gao, D.Z., Hou, L.J., Liu, M., 2019. Salinity stress changed the 750 biogeochemical controls on CH4 and N2O emissions of estuarine and intertidal 30 / 36 751 sediments. Sci. Total Environ. 652, 593–601. 752 Li, X.F., Hou, L.J., Liu, M., Zheng, Y.L., Yin, G.Y., Lin, X.B., Cheng, L., Li, Y., Hu, X., 753 2015. Evidence of nitrogen loss from anaerobic ammonium oxidation coupled with 754 ferric iron reduction in an intertidal wetland, Environ. Sci. Technol. 49, 11560–11568. 755 Lu, M., Zhou, X.H., Yang, Q., Li, H., Luo, Y.Q., Fang, C.M., Chen, J.K., Li, B., 2013. 756 Responses of ecosystem carbon cycle to experimental warming: a meta- 757 analysis. Ecology 94(3), 726–738. 758 Lu, W., Xiao, J., Liu, F., Zhang, Y., Liu, C.A., Lin, G., 2017. Contrasting ecosystem 759 CO2 fluxes of inland and coastal wetlands: a meta-analysis of eddy covariance 760 data. Global Change Biol. 23(3), 1180–1198. 761 Martens-Habbena, W., Berube, P.M., Urakawa, H., José, R., Stahl, D.A., 2009. 762 Ammonia oxidation kinetics determine niche separation of nitrifying Archaea and 763 Bacteria. Nature 461(7266), 976–979. 764 Martin, R.M., Moseman-Valtierra, S., 2017. Different short-term responses of 765 greenhouse gas fluxes from salt marsh mesocosms to simulated global change drivers. 766 Hydrobiologia 802(1), 71–83. 767 Meinshausen, M., Meinshausen, N., Hare, W., Raper, S.C., Frieler, K., Knutti, R., Allen, 768 M.R., 2009. Greenhouse-gas emission targets for limiting global warming to 2 oC. 769 Nature 458(7242), 1158. 770 Morrissey, E.M., Berrier, D.J., Neubauer, S.C., Franklin, R.B., 2014a. Using microbial 771 communities and extracellular enzymes to link soil organic matter characteristics to 772 greenhouse gas production in a tidal freshwater wetland. Biogeochemistry 117(2-3), 773 473–490. 774 Morrissey, E.M., Gillespie, J.L., Morina, J.C., Franklin, R.B., 2014b. Salinity affects 775 microbial activity and soil organic matter content in tidal wetlands. Global Change 31 / 36 776 Biol. 20, 1351–1362. 777 Murray, R.H., Erler, D.V., Eyre, B.D., 2015. Nitrous oxide fluxes in estuarine 778 environments: response to global change. Global Change Biol. 21(9), 3219–3245. 779 Nahlik, A. M., Mitsch, W. J., 2011. Methane emissions from tropical freshwater 780 wetlands located in different climatic zones of Costa Rica. Global Change Biol. 17(3), 781 1321–1334. 782 Neubauer, S.C., Megonigal, J.P., 2015. Moving beyond global warming potentials to 783 quantify the climatic role of ecosystems. Ecosystems 18, 1000–1013. 784 Nottingham, A.T., Bååth, E.R.L.A.N.D., Reischke, S., Salinas, N., Meir, P., 2019. 785 Adaptation of soil microbial growth to temperature: using a tropical elevation 786 gradient to predict future changes. Global Change Biol. 25, 827–838. 787 Philippot, L., Tscherko, D., Bru, D., Kandeler, E., 2011. Distribution of high bacterial 788 taxa across the chronosequence of two alpine glacier forelands. Microbial 789 Ecol. 61(2), 303–312. 790 Raymond, P.A., Hartmann, J., Lauerwald, R., Sobek, S., McDonald, C., Hoover, M., 791 Butman, D., Striegl, R., Mayorga, E., Humborg, C., Kortelainen, P., Dürr, H., 792 Meybeck, M., Ciais, P., Kortelainen, P., 2013. Global carbon dioxide emissions from 793 inland waters. Nature 503(7476), 355. 794 Ryden, J., 1983. Denitrification loss from a grassland soil in the field receiving different 795 rates of nitrogen as ammonium nitrate. J. Soil Sci. 34, 355–365. 796 Spohn, M., Klaus, K., Wanek, W., Richte, A., 2016. Microbial carbon use efficiency 797 and biomass turnover times depending on soil depth–Implications for carbon cycling 798 Soil Biol. Biochem. 96, 74–81. 799 Stelzer, R.S., Scott, J.T., 2018. Predicting nitrate retention at the groundwater-surface 800 water interface in Sandplain Streams. J. Geophys. Res.-Biogeo. 123, 2824–2838. 32 / 36 801 Tian, H.Q., Chen, G.S, Lu, C.Q, Xu, X.F, Ren, W., Zhang, B., Banger, K., Tao, B., Pan, 802 S.F., Liu, M.L., Zhang, C., Bruhwiler, L., Wofsy, S., 2015. Global methane and 803 nitrous oxide emissions from terrestrial ecosystems due to multiple environmental 804 changes. Ecosyst. Health Sustain. 1(1), 1–20. 805 Turner, P.A., Griffis, T.J., Baker, J.M., Lee, X., Crawford, J.T., Loken, L.C., Venterea, 806 R.T., 2016. Regional-scale controls on dissolved nitrous oxide in the Upper 807 Mississippi River. Geophys. Res. Lett. 43(9), 4400–4407. 808 Vance, E.D., Brookes, P.C., Jenkinson, D.S., 1987. An extraction method for measuring 809 soil microbial biomass C. Soil Biol. Biochem. 19, 703–707. 810 Vivanco, L., Irvine, I.C., Martiny, J.B., 2015. Nonlinear responses in salt marsh 811 functioning to increased nitrogen addition. Ecology 96(4), 936–947 812 Voigt, C., Lamprecht, R.E., Marushchak, M.E., Lind, S.E., Novakovskiy, A., Aurela, 813 M., Martikainen, P.J., Biasi, C., 2017. Warming of subarctic tundra increases 814 emissions of all three important greenhouse gases–carbon dioxide, methane, and 815 nitrous oxide. Global Change Biol. 23(8), 3121–3138. 816 von Sperber, C.; Chadwick, O. A.; Casciotti, K. L.; Peay, K. G.; Francis, C. A.; Kim, A. 817 E.; Vitousek, P. M., 2017. Controls of nitrogen cycling evaluated along a well– 818 characterized climate gradient. Ecology 98(4), 1117–1129. 819 Wagner, D., Lipski, A., Embacher, A., Gattinger, A., 2005. Methane fluxes in 820 permafrost habitats of the Lena Delta: effects of microbial community structure and 821 organic matter quality. Environmen. Microbiol. 7(10), 1582–1592. 822 Walter, B.P., Heimann, M., 2000. A process-based, climate-sensitive model to derive 823 methane emissions from natural wetlands: Application to five wetland sites, 824 sensitivity to model parameters, and climate. Global Biogeochem. Cy. 14, 745–765. 825 Wang, Q., He, N.P., Yu, G.R., Gao, Y., Wen, X.F., Wang, R.F., Koerner, S.E., Yu, Q., 33 / 36 826 2016. Soil microbial respiration rate and temperature sensitivity along a north-south 827 forest transect in eastern China: Patterns and influencing factors. J. Geophys. Res.- 828 Biogeo. 121(2), 399–410. 829 Wells, N. S., Maher, D. T., Erler, D. V., Hipsey, M., Rosentreter, J. A., Eyre, B. D., 2018. 830 Estuaries as sources and sinks of N2O across a land use gradient in subtropical 831 Australia. Global Biogeochem. Cy. 32, 877–894. 832 Weston, N.B., Neubauer, S.C., Velinsky, D.J., Vile, M. A., 2014. Net ecosystem carbon 833 exchange and the greenhouse gas balance of tidal marshes along an estuarine salinity 834 gradient. Biogeochemistry 120(1–3), 163–189. 835 Xie, X., Zhang, M. Q., Zhao, B., Guo, H.Q., 2014. Dependence of coastal wetland 836 ecosystem respiration on temperature and tides: A temporal perspective. 837 Biogeosciences 11(3), 539–545. 838 Xiong, Z.Q., Guo, L.D., Zhang, Q.F., Liu, G.H., Liu, W.Z., 2017. Edaphic conditions 839 regulate denitrification directly and indirectly by altering denitrifier abundance in 840 wetlands along the Han River, China. Environ. Sci. Technol. 51, 5483−5491. 841 Yang, W.H., Silver, W.L., 2016. Gross nitrous oxide production drives net nitrous oxide 842 fluxes across a salt marsh landscape. Global Change Biology, 22(6), 2228–2237. 843 Yuan, J.J., Ding, W.X., Liu, D., Kang, H., Freeman, C., Xiang, J., Lin, Y.X., 2015. 844 Exotic Spartina alterniflora invasion alters ecosystem–atmosphere exchange of CH4 845 and N2O and carbon sequestration in a coastal salt marsh in China. Global Change 846 Biol. 21(4), 1567–1580. 847 Yvon-Durocher, G., Allen, A. P., Bastviken, D., Conrad, R., Gudasz, C., St-Pierre, A., 848 Thanh-Duc, N., del Giorgio, P.A., 2014. Methane fluxes show consistent temperature 849 dependence across microbial to ecosystem scales. Nature 507(7493), 488–491. 850 Zaehle, S., Ciais, P., Friend, A.D., Prieur, V., 2011. Carbon benefits of anthropogenic 34 / 36 851 reactive nitrogen offset by nitrous oxide emissions. Nat. Geosci. 4(9), 601. 852 Zou, J.L., Tobin, B., Luo, Y.Q., Osborne, B., 2018. Differential responses of soil CO2 853 and N2O fluxes to experimental warming. Agr. Forest Meteorol. 259, 11–22. 854 855 856 857 858 859 860 Figure captions 861 Figure 1. Sites of Chinese coastal marshes along the latitudinal gradient. Sites DD, TS, 862 WF, QD and DF have a temperate climate, and SH, WZ, FZ, ST, ZH and BH have a 863 subtropical climate. DD, TS, WF, QD, DF, SH, WZ, FZ, ST, ZH and BH are 864 abbreviations for the cities of Dandong, Tangshan, Weifang, Qingdao, Dafeng, 865 Shanghai, Wenzhou, Fuzhou, Shantou, Zhuhai and Beihai, respectively. 866 Figure 2. Emissions of CH4 (a), N2O (b) and CO2 (c) from the coastal marshes along 867 the latitudinal gradient. The bars denote standard errors (n = 3). 868 Figure 3. Emissions of CH4 (a), N2O (b) and CO2 (c) in the temperate and subtropical 869 coastal marshes. DD, TS, WF, QD and DF are grouped into temperate sites, and SH, 870 WZ, FZ, ST, ZH and BH are grouped into subtropical sites. The horizontal lines indicate 871 medians, the small squares indicate means, the asterisks indicate outliers, the boxes 872 indicate the 25th and 75th percentiles and the bars indicate the range from the 5th to the 873 95th percentiles. Different letters above the plots with the same color denote significant 874 differences between the temperate and subtropical marshes (P < 0.05). The asterisks 875 above and below the plots denote significant differences between December and July 35 / 36 876 (P < 0.05). 877 Figure 4. Variations in CH4, N2O and CO2 emissions with air temperature (a-c) and 878 mean annual temperature (d-f). Emissions of CH4, N2O and CO2 were log(10) 879 transformed prior to testing. Correlations (r2) and significance level (P) are shown. 880 Figure 5. Stepwise multiple regression analyses of the greenhouse-gas fluxes and 881 environmental traits. The data were log(10) transformed normally before entry into the 882 models. For this model, the final set of predictors are shown in sequence. (+) and (–) 883 denote positive and negative correlations between the greenhouse-gas fluxes and the 884 predictors, respectively. MBC, microbial biomass carbon; Eh, redox potential; MAT, 885 mean annual temperature; AT, air temperature; NOx–, NO3– plus NO2–. 886 Figure 6. 2D plots of the variables (A) and cases (B) of the principal component 887 analysis(PCA) TN, total soil nitrogen; NH4+, soil extractable ammonium; NOx–, soil 15 888 nitrate/nitrite extractable; N, δ15N; Eh, redox potential; pH, soil pH; SO42–, soil 889 extractable sulfate; CN, soil total C/N ratio; 13C, δ13C; MAT, mean annual temperature; 890 SD, soil density; AT, air temperature; Microbes, bacterial abundance; MBC, microbial 891 biomass carbon; MBN, microbial biomass nitrogen; TOC, total organic carbon; DOC, 892 dissolved organic carbon; SWC, soil-water content; CO2, CO2 emission; CH4, CH4 893 emission; N2O, N2O emission. The scores in panel B are the ±95% confidence intervals 894 of the samples for each site and season. 895 Figure 7. Standardized direct and indirect effects derived from structural equation 896 model (SEM) for CH4 (a), N2O (b) and CO2 (c). MAT, mean annual temperature; Eh, 897 redox potential; MBC, microbial biomass carbon; NOx–, NO3– plus NO2–; TOC, total 898 organic carbon; MBN, microbial biomass nitrogen; DOC, dissolved organic carbon; 899 Microbes, bacterial abundance. 36 / 36