2.1. Site Description
 TGR is located within the watercourse of the Yangtze River, which has been flooded since 2003. The TGR is about 660 km in length, and has a surface area of 1084 km2, including the drawdown area of 450 km2. Water levels in TGR fluctuate seasonally by 30 m. The water level is held at 145 m during the flood season (June–August) to control flooding, which drains the drawdown area. The water level is gradually increased to 175 m after the flood season in order to increase the efficiency of electricity generation [Zhou et al., 2010]. The capacity to generate electricity at TGR was 18.2 million kW in 2010 (84.7 billion kWh of electricity were actually produced in 2010), making the TGR the largest hydroelectric producer in the world.
 The TGR is located in a subtropical monsoon climate zone with an annual mean temperature of 16.3–18.2°C and annual precipitation of 987–1326 mm [Guo et al., 2007]. Nearly 80% of the precipitation falls in the hot-wet season (April–September); only 20% falls in the cool-dry season (October–March) [Guo et al., 2007]. This study was carried out at three sites in the drawdown area near Zigui (30°51′N, 110°58′E), Wushan (31°03′N, 109°51′E), and Yunyang (30°56′N, 108°39′E), which are 2 km, 120 km and 240 km upstream from the Three Gorges Dam (Figure 1).
 Three kinds of plots (fallow land, cropland, and deforested land) at each site were selected in September 2009 when the plots were drained and the water level was below 146 m. A rice paddy (Oryza sativa), located at an elevation of 172 m in Yunyang, was added in June 2010 as a kind of agricultural wetland; thus making 10 experimental plots over 3 sites (Table 1). As lands below 160 m were frequently inundated by normal water level fluctuations or summer floods, only lands above 160 m were reclaimed for crop cultivation. Croplands were thus only located in the upper zone of the drawdown area at each of the three sites, while fallow and deforested lands were located in the lower zone, except for the deforested land in Wushan. Maize (Zea mays L.) was grown in the croplands of Zigui and Wushan, and sesame (Sesamum indicum L.) was grown in the cropland of Yunyang and the deforested land in Wushan. In the deforested lands, trees were cut down before the reservoir was impounded and many stumps were left on site.
Table 1. Elevation, Duration of Inundation, pH, and Soil Carbon and Nitrogen Levels at the Sampled Sitesa
|Site||Plot||Elevation (m)||Inundated Duration (d)||Soil pH||Organic C (g kg−1)||Total N (g kg−1)|
|Zigui||Fallow land||149||314||6.42 ± 0.16(a)||10.72 ± 2.67(a)||0.78 ± 0.17(a)|
| ||Cropland||160||175||6.64 ± 1.01(a)||11.18 ± 3.56(a)||0.90 ± 0.19(a)|
| ||Deforested land||148||326||6.21 ± 0.32(a)||11.44 ± 1.65(a)||0.85 ± 0.14(a)|
| ||Average||152||272||6.42 ± 0.22||11.11 ± 0.36||0.81 ± 0.12|
|Wushan||Fallow land||151||296||8.00 ± 0.20(a)||8.11 ± 1.13(a)||1.05 ± 0.10(a)|
| ||Cropland||157||229||8.28 ± 0.08(a)||12.39 ± 2.43(b)||1.26 ± 0.24(b)|
| ||Deforested land||161||167||8.12 ± 0.10(a)||11.64 ± 2.69(b)||1.19 ± 0.20(a, b)|
| ||Average||156||231||8.13 ± 0.14||10.71 ± 2.29||1.17 ± 0.11|
|Yunyang||Fallow land||149||314||8.29 ± 0.10(a)||4.68 ± 2.38(b)||0.57 ± 0.17(b)|
| ||Cropland||160||175||8.63 ± 0.50(a)||2.66 ± 0.57(a)||0.34 ± 0.04(a)|
| ||Deforested land||150||305||8.25 ± 0.05(a)||3.55 ± 0.43(b)||0.48 ± 0.18(b)|
| ||Average||153||265||8.39 ± 0.21||3.63 ± 1.01||0.46 ± 0.12|
| ||Rice paddies||172||214||8.45 ± 0.26(a)||16.49 ± 2.47(c)||2.69 ± 0.83(c)|
2.2. CH4 Emission Measurements
 CH4 emissions were measured using the static chamber method [Duchemin et al., 1995]. Two kinds of static chambers were deployed: 1) a floating static chamber for measuring CH4 emissions from the water surface, and 2) a closed static chamber was used for measuring CH4 emissions from dry land or rice paddies during the drained season. The same chambers were used for measuring CH4 emissions from dry lands both in the inundated and drained seasons. The floating static chamber (65 cm in length × 45 cm in width × 40 cm in height) consisted of a plastic box without a cover that was wrapped in light-reflecting and heatproof films to prevent temperature variation inside the chambers; in addition, plastic foam collars were fixed onto opposite sides of the chamber. The headspace height inside the chamber was about 30 cm. When the fallow lands, croplands, and deforested lands were drained, the above-described chambers were placed on permanently positioned aluminum bases (65 cm in length × 45 cm in width × 10 cm in height) with water grooves on top and inserted into the soil in order to ensure a tight fit at the air-soil interface. To avoid disturbing plant respiration and photosynthesis, any new growth grasses around the base were cleared before the gas samples were collected. A closed, static, steel frame chamber (50 cm in length × 50 cm in width × 75 cm in height for measuring emissions from rice paddies) was covered in polyethylene plastic film (85% transparent) and used to collect gas samples from rice paddy [Duan et al., 2005; Zheng et al., 2011a]. A silicone tube (0.6 cm and 0.4 cm outer and inner diameters, respectively) was inserted into the upper side of the chamber to collect gas samples and another silicone tube was inserted into the chamber to keep air pressure balanced between the inside and outside of the chamber. All measurements were performed in triplicate.
 In either flooded or drained seasons, the gases in the headspace of the chamber were collected into air-sampling bags (0.5 L; Hedetech, Dalian, China) five times every 10 min over a 40 min period using a hand-driven pump (ICQS-1; Beijing Municipal Institute of Labour Protection, Beijing, China). The gas samples were transported within 2–3 days after sampling to the State Key Laboratory of Urban and Regional Ecology (Beijing, China) for analysis using a gas chromatograph (Agilent 6820; Agilent Technologies, Santa Clara, USA) equipped with a flame ionization detector (FID) and separated with a Teflon column (2 m × 3 mm) packed with TDX-01(60/80 mesh). The oven, injector, and detector temperatures were 80°C, 150°C, and 300°C, respectively. The flow rate of the carrier gas (N2) was 30 mL min−1, and the flow rate of H2 and compressed air were set to 20 and 30 mL min−1, respectively. Standard CH4 gas (10.2 ppm in air; provided by China CH4 National Research Center for Certified Reference Materials, Beijing) was used to quantify the CH4 concentration in one of every 10 samples, which kept the coefficient of variation of the CH4 concentration in the replicated samples below 1%.
 We separated the diffusive and bubble emissions based on the change of CH4 concentration in the chambers. The CH4 emission was considered diffusive if the linear correlation between the CH4 concentration in the chamber and the elapsed time had r2 greater than 0.90 [Marani and Alvalá, 2007]. If the CH4 concentration was punctuated by one or more abrupt increases and the initial concentration in the chambers (at time t = 0) was close to the ambient air concentration, the abrupt increases were most easily explained by interception of rising gas bubbles by the chamber [Keller and Stallard, 1994]. The diffusive CH4 emission (Fd; mg CH4 m−2 h−1) was determined using
where ρ is the density of CH4 under standard conditions (0.714 kg m−3), dc/dt is the slope of the linear regression of the CH4 concentration in the chamber versus time, H is the height of the chamber above the water or soil surface (0.3 m for dry lands and 0.75 m for rice paddies), and T is the air temperature (°C).
 The bubble CH4 emission (Fb; mg CH4 m−2 h−1) was determined using
where ΔC is the gas concentration difference between the beginning and the end of the enclosure time in the chamber (ppm) and ΔT is the total time of emplacement (h).
 Cropland, deforested land, and fallow land were selected along one side of the Yangtze River at each site (Figure 1 and Table 1). The rice paddy plot was located at 4 km upstream of Yunyang (Figure 1). In each sampling plot, all measurements were conducted at the same places during all seasons, regardless of inundation or drainage. CH4 emissions were measured once or twice per month in the morning, weather depending, for 15 months (November 2009 to January 2011) in the fallow lands, croplands, deforested lands, and rice paddies at each growing stage and twice per month after the harvest. 17–21 times measurement was carried out at each land use of dry lands and 11 times at the rice paddy plot. CH4 emissions were measured from 8:00–18:00 at 2 h intervals about once every two months in the fallow land, cropland and deforested land in Zigui in order to assess the variability in CH4 emission during the daytime.
2.3. Environmental Variables
 During the inundated season, the following parameters were measured in situ: (1) velocity, using LS1206B, Midwest Group, Beijing, China; (2) pH, using HI 8424, Microcomputer HANNA, Rome, Italy; and (3) turbidity, using HI93703, Microcomputer HANNA. Air temperature, water temperature, and water depth were also measured in the field using alcohol thermometers and sounding ropes, respectively. In addition, water samples were collected once every month using plastic bottles (0.5 L) at a depth of 0.5 m and kept in a refrigerator at 4°C until laboratory analyses could be performed. Water quality parameters, including nitrate nitrogen (NO3−-N), ammonium nitrogen (NH4+-N), total nitrogen (TN), and total phosphorus (TP), were measured according to the methods of the State Environmental Protection Administration of China [Qi et al., 2002]. NO3−-N was determined by the spectrophotometric method with phenol disulfonic acid, and NH4+-N was determined by Nessler's reagent spectrophotometric method [Qi et al., 2002]. TN and TP were analyzed by peroxodisulfate oxidation of the original water samples. Total organic carbon (TOC) and total inorganic carbon (TIC) were detected using a total organic carbon analyzer (Liquic TOC; Elementar Co., Hanau, Germany). In the drained season, soil temperature was measured in the field using a waterproof thermometer (AD-5604; A&D Co., Tokyo, Japan) at a depth of 5 cm. Soil samples were collected once every month from a depth of 0–20 cm for laboratory analysis. Soil pH and organic carbon and total nitrogen levels were measured using a pH meter (Delta 320 pH meter; Mettler-Toledo Instruments Ltd., Shanghai, China) and elemental analyzer (Vario EL III; Elementar Co.), respectively.
2.4. Estimation of Total CH4 Emissions From the Drawdown Area
 To estimate the total CH4 emissions from the TGR drawdown area, two kinds of land (rice paddy and dryland, which included fallow land, cropland, and deforested land) and two periods (inundated and drained seasons) were studied. For each type of land use, the annual CH4 emission was calculated as the summation of the products of the CH4 emission rate, areas, and lengths of the inundated and drained season (equation (3)). Most of rice paddies were located in upland above the elevation of 170 m and inundated for 214 days with less variation (about 137 days as rice paddies in anaerobic condition, and 77 days as submerged drawdown areas). So the CH4 emission from all rice paddies was estimated with a constant inundation period. The total CH4 emission (TE) in the drawdown area was estimated by summing dryland and rice paddy emissions (equation (3)):
where i is the elevation; Pi is the inundated duration (days) for dryland at each elevation; FEin and FEdr are the CH4 emission rates of dryland during the inundated and drained seasons, respectively; Ai is the area of dryland at each elevation (range: 145–175 m) that was calculated from digital elevation model (DEM) data of the TGR drawdown area using ArcGIS 9.3 (ESRI Co., Redlands, USA); P′ is the inundated duration of the rice paddy; FEin′ and FEdr′ are the CH4 emission rates of the rice paddy during the inundated and drained seasons, respectively; and A0 is the total area of rice paddy, which accounted for 11% of the drawdown area [Ye et al., 2006].
2.5. Statistical Analysis
 One-way analysis of variance (ANOVA), in combination with the Tukey test, was used to analyze differences in soil pH, organic carbon, and total nitrogen levels between different land uses at the same site. The normal distribution of the CH4 emissions during the inundated and the drained seasons were tested using Kolmogorov-Smirnov Test. The skewed CH4 emissions distributions required the data set to be normalized by natural logarithmic transformation, and then differences between sites and land uses were tested using ANOVA. Two-way ANOVA was used to test interactions between sites and land uses in terms of CH4 emission during the inundated season. CH4 emissions were linearly regressed in terms of soil temperature at a depth of 5 cm, air-water temperature difference, air temperature, NH4+-N content and water depth. A multiple regression model was used to determine the key environmental variables that influenced CH4 emission during the inundated season. Data were analyzed with the SPSS 16.0 statistical package.