After our previous study about methane (CH4) emissions from littoral marshes of the Three Gorges Reservoir (TGR), Chinese dams have raised a world-wide concern. Through measurements from the surface of the TGR, a CH4 emission rate was recorded as 0.26 ± 0.38 mg CH4 m−2 h−1 (Mean ± SD), relatively low compared with those from other hydropower reservoirs. We also recorded CH4 emission rate from the surface of downstream water, which was also relatively low (0.24 ± 0.37 mg CH4 m−2 h−1). Such result may indicate that TGR is not a great CH4 emitter (not “CH4 menace”). One possible reason for such a low emission rate is that measures to maintain water quality and protect environment and ecosystem decrease the input of organic materials (for methanogenesis), which in turn limits the CH4 production in the sediment of the TGR. We also found that CH4 emission from the flooding drawdown area (0.29 ± 0.37 mg CH4 m−2 h−1) was higher than other permanently flooded sites (0.23 ± 0.38 mg CH4 m−2 h−1). Because of annual vegetation re-growth, the drawdown zone is the especially important carbon source for methanogenesis in flooding season. Interestingly, we also observed that mean CH4 emission was significantly higher in winter than in spring and summer. This was partly due to seasonal dynamics of hydrology. In order to estimate the net CH4 emissions caused by the reservoir and reservoir operation, the best approach would be Life Cycle Analysis.
 Anthropogenic CH4 sources dominate present-day CH4 budgets, accounting for more than 60% of the global budget [Denman et al., 2007]. Compared with main anthropogenic sources of CH4, such as rice agriculture, waste, livestock and biomass burning, large hydropower reservoirs are suggested as one of the largest sources of human-caused CH4 emissions, annually emitting 10 and 70 Tg CH4, for low and high dissolved CH4 concentration scenarios[Lima et al., 2007a]. Much of the research on hydropower reservoir CH4 fluxes focuses on tropical and boreal regions with only a few studies in temperate and subtropical regions [Chen et al., 2009]. Therefore, more data from hydropower reservoirs will help us to better estimate the global dam CH4 emission.
 Among the dams in the world, the 2,335 m long and 185 m high Three Gorges Dam (TGD) on the Yangtze River of China is the largest and thus a good example. The Three Gorges Reservoir (TGR) is about 660 km long and 58,000 km2 in watershed area, greater than Switzerland [Stone, 2008; Wu et al., 2003]. When operating at full capacity, the total area of TGR is estimated to be about 1,080 km2 [Wu et al., 2004]. Moreover, after our previous study about CH4 emissions from littoral marshes of the TGR [Chen et al., 2009], Chinese dams have been misleadingly described as a “CH4 menace” in Nature News [Qiu, 2009], which exaggerated the fact and raised a world-wide concern about Chinese dams. In our previous paper, we focused on CH4 emission from marshes in the exposed drawdown area in summer. For a reasonable and comprehensive understanding of CH4 emission from TGR, this study aimed to measure CH4 emission from the surface of TGR.
2. Methods and Materials
 This study was conducted in the Three Gorges Reservoir, which is built on the Yangtze River in south-central China. This region has a humid subtropical monsoon climate, with a mean annual temperature of 15–19°C, mean annual precipitation of 1,250 mm, and relative humidity of 76%. For spatial variability, the sampling of CH4 fluxes was made at four sites along the reservoir in the river channel (Sandouping, Jiangjin, Fulin, River Delta in Kaixian), two in the drawdown zone (Wuxi and Kaixian; we measured CH4 emission from the surface at these sites during flooding, and from the surface of the small river channel during exposure) and one site after the dam (Figure 1). Considering the temporal fluctuation of water table at TGR, we tried to explore the effect of water table fluctuation on CH4 emission in this research. We took three campaigns of measurement when TGR was partially drained in spring (February 15, March 15 and April 15), two campaigns when the water table of TGR was 145 m in the summer (June 15 and September 15), and one campaign when the water table was 170 m in the winter (December 15).
 CH4 flux was measured with floating chambers. The chambers (25 cm in diameter, 40 cm in height) were made of cylindrical polyvinyl chloride (PVC) pipe with a floating ring. Four air samples from each chamber were taken at 10-min intervals over a 30 min period after enclosure, stored in 50 ml air-tight vacuumed vials. The CH4 concentration was determined by a gas chromatography (PE Clarus 500, PerkinElmer, Inc., USA), equipped with a FID (flame ionization detector) operating at 350°C and a 2 m Porapak 80–100 Q Column. The column oven temperature was 35°C and the carrier gas was N2 with a flow rate of 20 cm3 min−1. Detailed information has been described in our previous paper [Chen et al., 2009]. Total carbon (TC) concentration in the surface water was detected by a total organic carbon (TOC) analyzer (multiN/C 2100, Analytik Jena AG, Jena, Germany).
 Mean CH4 emissions were calculated by averaging the six replicates for each sampling day. One-way ANOVA was used to compare the fluxes among different sites. Data was analyzed with SPSS 11.5 statistical package.
3. Results and Discussions
3.1. CH4 Emissions From the Surface of TGR and Its Determinants
 Through measurements along the surface of TGR, we did not find significant variations in CH4 emission among sites (P = 0.355, Figure 2). However, we noticed that CH4 emission rate from the drawdown area (0.29 ± 0.37 mg CH4 m−2 h−1) was markedly higher than the permanently flooding sites (0.23 ± 0.38 mg CH4 m−2 h−1). Moreover, we observed that mean CH4 emission in winter was significantly higher than those in spring and summer (Figure 2). We also found a significant nonlinear relationship between total carbon (TC) and CH4 emission (Figure 3). Interestingly, there was a turning point when TC = 30 mg L−1, which will be useful for environmental modelers.
 Overall average CH4 emission (±SD) was 0.26 ± 0.38 mg CH4 m−2 h−1, which is much lower than our previous estimate (3.3 mg CH4 m−2 h−1) [Chen et al., 2009]. We also recorded CH4 emission rate from the surface of downstream water, which was also relatively low (0.24 ± 0.37 mg CH4 m−2 h−1, n = 36). Though TGR is located in Subtropical zone, such relative low value of CH4 emission is comparable to those from Boreal zone (Table 1). However, recent studies indicated that latitude shall not the best variable to indicate whether a dam will be a significant CH4 source or not [Barros et al., 2011; Lima et al., 2011]. For example, a Swiss study indicated that the total CH4 emission from Lake Wohlen was on average >150 mg CH4 m−2 d−1, which is the highest ever documented for a midlatitude reservoir [DelSontro et al., 2010]. Besides, the emission rate from TGR is lower than that from other hydroelectric reservoirs with flooded peat or forests [Fearnside, 2005].
Table 1. Methane Emissions From Different Aquatic Ecosystems From Boreal to Tropical Zonesa
 Chinese government have taken several measures to abase environmental problems after damming the Yangtze River just after Three Gorges [Ministry of Environmental Protection of PRC, 2001]. For example, since 2002, to conserve water quality, vegetation and solid waste is cleared out every year prior to impounding. Such clearance may have limited the C substrate supplies for methanogens by reducing the amount of dead plant matter and waste on the substrate surface. By 2010, 95% of wastewater from upstream and TGRR have been purified through 107 large wastewater treatment plants, which directly decreased the carbon input from cities and industry of upstream, and indirectly decreased carbon input through reducing the eutrophication rate of the reservoir. What's more, two of China's unprecedented conservation actions, the Natural Forest Conservation Program and the Grain to Green Program, were initiated with priority in TGRR and its upstream region. Such land-use changes and measures have abased carbon leaching from terrestrial natural and agricultural ecosystems in TGRR and its upstream.
 Therefore, we understand that the comparatively low CH4 emission rate at TGR is mainly due to the limited substrate availability for methanogens, not only due to the warm climate. Similarly, CH4 emissions from the downstream of TGR were also low due to substrates limited (Figure 2). The significant relation between TC and CH4 emission in this study partially evidenced that carbon input from upstream and surrounding terrestrial ecosystems was an important substrate for CH4 production in TGR. Though some studies reported that during the first few years of impounding CH4 emissions are much higher because of more CH4 generated from carbon stock in soil and the remained biomass [Fearnside, 2005]. In our study, we did not find high CH4 emission from the surface of TGR. This partially indicated that carbon input from upstream and surrounding terrestrial ecosystems were relatively low even in the first years of TGR.
 In this study, we found that CH4 emission from the flooded drawdown area was higher than other permanently flooded sites. We also observed that mean CH4 emission in winter was significantly higher than spring and summer (Figure 2). This was partly due to seasonal dynamics of hydrology. Because hydrology is a master variable mediating many of the biogeochemical processes (such as O2 concentration) in aquatic systems or wetlands, dramatic changes in hydrological conditions (such as water depth and water retention time) should result in measurable differences in processes such as GHG fluxes [Altor and Mitsch, 2008]. Altor and Mitsch  reported that CH4 emission were twice as high in continuously inundated zones during the steady-flow year compared to the flood pulsed year. Hydrology of TGR in spring and summer was highly dynamical due to fast drainage in spring and frequently flood pulses in summer (Figure 1), thus soluble carbon has only short retention time for methogenesis. In winter, the water table depth was higher than in and it fluctuated less possibly allowing more phytoplankton production and thus more substrates to support CH4 production [Yang et al., 2010].
3.2. Limitations and Rough Estimate
 As one of pilot studies about CH4 emissions from dams in China [Chen et al., 2009; Yang et al., 2009], this brief report only presented limited data about CH4 diffusive emissions from the surface of TGR. Static chambers used in this study can only monitor the diffusive emission. For more accurate measurement of GHG fluxes from dams, a high-proficiency and long-term monitoring system should be established to monitor different emission pathways using novel techniques [Lima et al., 2007b; Ramos et al., 2006]. Therefore, we did not and still cannot have data from bubbles, turbines and spill ways due to limited techniques and conditions [Qiu, 2009]. Though we cannot present a whole picture about CH4 emission from TGD at present, calculating with the surface area (1080 km2), we roughly estimated the annual diffusive CH4 flux from TGR was 2.46 Gg (1 Gg = 109 g), which was less than 10% of our previous estimate [Chen et al., 2009]. Moreover, the estimate (2689 ± 201 Gg CH4 yr−1) through modeling by Lima et al. , not carefully validated by data from China, thus was probably an overestimate of CH4 emissions from hydropower reservoirs in China. Furthermore, Lima et al.  overestimated dam methane emissions due partly to misuse of spillway ICOLD data (I. B. T. Lima, personal communication, 2011). For our continuous study to get a preliminary budget of CH4 and carbon dioxide emissions from TGD or even Chinese dams, the best approach would be Life Cycle Analysis, which includes several models, such as Biome Carbon Loss Model [Abril et al., 2005], Downstream Emission Model [Lima et al., 2007b], etc.
 This study was financially supported by Natural Science Foundation Project of CQ CSTC (2009BB7182), Chinese Postdoctoral Foundation (20090460058) and Key Laboratory of Mountainous Ecological Restoration and Biological Resources Utilization, Chinese Academy of Sciences (KXYSWS0902). This study was also funded by National Natural Science Foundation of China (31100348). We must give personnel thanks to Rong Sun and Qiang Wang for their assistance in our field sampling. Wan Xiong, an expert for ESP, was thanked for her great and patient help in our writing and reasoning. Two anonymous reviewers and I. B. T. Lima are thanked for detailed evaluation and constructive suggestion on our manuscript.