Soil–air greenhouse gas fluxes influenced by farming practices in reservoir drawdown area: A case at the Three Gorges Reservoir in China

Journal of Environmental Management 181 (2016) 64e73

Contents lists available at ScienceDirect

Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman

Research article

Soileair greenhouse gas fluxes influenced by farming practices in reservoir drawdown area: A case at the Three Gorges Reservoir in China Zhe Li a, b, *, Zengyu Zhang d, Chuxue Lin b, Yongbo Chen b, Anbang Wen c, Fang Fang d a

CAS Key Lab on Reservoir Environment, Chongqing Institute of Green and Intelligent Technology, Chinese Academy of Sciences, Chongqing, 400714, China China Three Gorges Corporation, Beijing, 100038, China Institute of Mountain Hazards and Environment, Chinese Academy of Sciences, Chengdu, 610041, China d College of Urban Construction and Environmental Engineering, Chongqing University, Chongqing, 400045, China b c

a r t i c l e i n f o

a b s t r a c t

Article history: Received 15 January 2016 Received in revised form 23 April 2016 Accepted 30 May 2016

The Three Gorges Reservoir (TGR) in China has large water level variations, creating about 393 km2 of drawdown area seasonally. Farming practices in drawdown area during the low water level period is common in the TGR. Field experiments on soil-air greenhouse gas (GHG) emissions in fallow grassland, peanut field and corn field in reservoir drawdown area at Lijiaba Bay of the Pengxi River, a tributary of the Yangtze River in the TGR were carried out from March through September 2011. Experimental fields in drawdown area had the same land use history. They were adjacent to each other horizontally at a narrow range of elevation i.e. 167e169 m, which assured that they had the same duration of reservoir inundation. Unflooded grassland with the same land-use history was selected as control for study. Results showed that mean value of soil CO2 emissions in drawdown area was 10.38 ± 0.97 mmol m
2 h
1. The corresponding CH4 fluxes and N2O fluxes were
8.61 ± 2.15 mmol m
2 h
1 and 3.42 ± 0.80 mmol m
2 h
1. Significant differences and monthly variations among land uses in treatments of drawdown area and unflooded grassland were evident. These were impacted by the change in soil physiochemical properties which were alerted by reservoir operation and farming. Particularly, Nfertilization in corn field stimulated N2O emissions from March to May. In terms of global warming potentials (GWP), corn field in drawdown area had the maximum GWP mainly due to N-fertilization. Gross GWP in peanut field in drawdown area was about 7% lower than that in fallow grassland. Compared to unflooded grassland, reservoir operation created positive net effect on GHG emissions and GWPs in drawdown area. However, selection of crop species, e.g. peanut, and best practices in farming, e.g. prohibiting N-fertilization, could potentially mitigate GWPs in drawdown area. In the net GHG emissions evaluation in the TGR, farming practices in the drawdown area shall be taken into consideration. © 2016 Elsevier Ltd. All rights reserved.

Keywords: Three Gorges Reservoir Drawdown area Farming practices Greenhouse gas emissions Global warming potential

1. Introduction

Abbreviations: GHG, Greenhouse gas; GHGI, Greenhouse gas intensity; GHGIb, GWPr of each treatment type is divided by full biomass with dry weight, which measures the intensity of GHG emissions per unit biomass; GHGIg, GWPr is divided by the total grain harvest of corn and peanut field, which gave the GHGI of corn and peanut production; GWP, Global warming potentials; GWPf, GWP from farming practices in drawdown area; GWPr, Net GWP from reservoir construction and operation; TGR, Three gorges reservoir; VWC, Volumetric water content. * Corresponding author. CAS Key Lab on Reservoir Environment, Chongqing Institute of Green and Intelligent Technology, Chinese Academy of Sciences, Chongqing, 400714, China. E-mail address: [email protected] (Z. Li). http://dx.doi.org/10.1016/j.jenvman.2016.05.080 0301-4797/© 2016 Elsevier Ltd. All rights reserved.

Excess emissions of greenhouse gases (GHGs) from anthropogenic activities, e.g., CO2, CH4, N2O and others, are regarded as the main drivers of global climate change (Ciais et al., 2013). Quantifying carbon footprints from all human activities and assessing their global warming potentials (GWPs) have been widely done in recent decades (Ciais et al., 2010). Dam construction and reservoir creation are also human products with carbon footprints that potentially contribute to global warming (Hertwich, 2013; Oud, 1993; Rudd et al., 1993). Reservoir carbon footprints are not solely based on the initial carbon stock before impoundment (Teodoru

Z. Li et al. / Journal of Environmental Management 181 (2016) 64e73

et al., 2012). Carbon inputted from upstream and terrestrial ecosystems, as well as increased in reservoir primary production, also significantly contribute to the net effect of reservoir construction and impoundment (Hertwich, 2013). Designed for multiple purposes, some reservoirs, e.g. the Three Gorges Reservoir (TGR) in China, are commonly operated with substantial water level variations (Li et al., 2014). Artificial operation of reservoirs not only perturbs carbon budgets in aquatic ecosystems, but may also create a large nearshore drawdown area with complicated carbon biogeochemical processes (Backeus, 1993). Without quantification on GHG emissions, research on human carbon footprints from reservoir operation remains limited. Created by the Yangtze Three Gorges Dam project, the TGR is the largest reservoir in China. Field surveys on water surface GHG emissions in the TGR indicated that gross emission of CH4 in the reservoir is comparable to those from other temperate reservoirs but significantly less than those from tropical reservoirs (Chen et al., 2011; Yang et al., 2013; Zhao et al., 2013). However, the TGR performs ~30 m vertical and seasonal water level variations for both hydropower generation and flood control, creating about 393 km2 of nearshore area in the form of seasonal drawdown and submersion. Chen et al. (2009) showed that newly created reservoir drawdown area resulted in remarkable CH4 emissions. Yang et al. (2012) estimated that CH4 emissions from reservoir drawdown area accounted for about 42e54% of the total CH4 emissions from the water surface of the TGR, which indicated that GHG emissions from drawdown area had significant contribution in gross GHG emissions of the reservoir. Besides, drawdown area in the TGR during summer has become popular for cultivation by local farmers in recent years. Impoundment of the TGR directly lose about 278 km2 nearshore arable farmland along the Yangtze River and its tributaries, which is approximately about 43.9% of total inundated area (632 km2) in the reservoir (Tullos, 2009; Xu et al., 2013). Resettlement of local people to higher land in this mountainous region significantly reduces the accessibility of fertile arable lands for farming. Exposure time in certain upper parts of reservoir drawdown area (elevation 165 m or higher) is approximately 180 days. This duration is feasible for cultivating crops for at least one growing season (Fig. 1). Living near the shore in the TGR, local farmers wish to use parts of the reservoir drawdown area as arable farmland for cultivation. A conservative estimation from incomplete survey in 2010

Fig. 1. Water level, inflow and outflow of the Three Gorges Reservoir (indicated as “TGR” in the figure) in 2012. Area of hatching indicates the period from March to September and elevation range for crop cultivation in drawdown area. The normal water level of the TGR is 175 m. For flood control, water level in the TGR gradually decreases from February to May and prepares about 22.0 km3 capacity for incoming floods.

65

indicated that over 70 km2 of drawdown area in the TGR was restored to agriculture (He et al., 2011). About 43% of them (~30 km2) were used for corn (Zea mays L.) cultivation, and 34% of them (~23.8 km2) of them were applied for growing peanut (Arachis hypogaea L.) (He et al., 2011). Other corps in farming practices in drawdown area in the TGR were rice (Oryza sativa) and green-leaf vegetables such as Chinese cabbage (Brassica rapa) and Chinese lettuce (Lactuca sativa var. asparagina). Even some drawdown area that were not historically farmlands, e.g., woodland and urban area before impoundment, were also reclaimed for cultivation. Farming and management practices can change soileair GHG emissions. Crop species, crop planting, field tillage, fertilization, application of herbicide and pesticide are the major factors. In recent years, global concern regarding climate change has fostered rapid growth of research on carbon footprints of agriculture (Robertson et al., 2000; Sainju et al., 2012, 2014b). Through changes in soil moisture, structure, porosity and other physico-chemical parameters, agricultural activities (in particular management of fields) are regarded as drivers of soileair GHG emissions. Reduction of N2O and CH4 emissions and increasing in C sequestration in the form of soil organic carbon could be achieved through effective practice, such as no tillage, low level of fertilization, and diversified crop rotation (Liebig et al., 2010; Robertson et al., 2000). Recently, Yang et al. (2012) reported that land uses, i.e. farming practices, played an important role in CH4 emissions during various water levels in the TGR. The study partially supported the hypothesis that land use change due to farming in drawdown area significantly impact GHG emissions of the TGR. However, lack of information on GHG emissions before impoundment, i.e. preimpoundment GHG emissions, and land use history, as well as the status GHG emissions among different farming practices still impeded the evaluation of net GHG emissions of the TGR. The objectives of the present study were to: (1) compare flux differences of CO2, CH4 and N2O between farming practices and fallow grassland during reservoir drawdowns; (2) evaluate GWP and GHGI among different types of crops in drawdown area; (3) address the implications of farming practices in drawdown area for net GHG emissions evaluation in the TGR. 2. Methods and materials 2.1. General description of farming practices and experimental design Because the drawdown area available to farming is relatively small and in patches (see photos of S1 in supplementary materials), in most cases, crop harvests in drawdown area are mainly supplied parts of daily food requirements. The remainder goes to feeding livestock, with little for sale in local farmers’ markets. No machinery are invested for planting, irrigation or fertilization. General procedures of corn and peanut cultivation in drawdown area has the following characteristics. 1) After reservoir drawdown in March (Fig. 1), fields of corn and peanuts are planted. 2) N fertilizer with urea is manually applied immediately after corn planting with amounts 40e50 g m
2, and no fertilizers were applied to peanut fields. 3) A second round of N-fertilization is carried out in corn fields at the end of April, with the same amounts as that in March. 4) No-till, no-pesticide applications, with a few herbicides, are served manually as needed. 5) Corn harvests are normally in July and peanut harvests normally in August. 6) After harvest, most corn straw are cut, collected outside fields and burned for cooking, and plant bodies of peanuts are mainly used for feeding livestock. Underground parts of both crops and other residues are left in drawdown area and submerged by the subsequent impoundment in October.

66

Z. Li et al. / Journal of Environmental Management 181 (2016) 64e73

As a first attempt to evaluate GWP and GHGI of farming practices in drawdown area in the TGR, corn field and peanut field were selected as cases and examples of farming practices in this study to compare with fallow grassland in drawdown area and unflooded grassland at elevations above 175 m which was not impacted by the reservoir operation. Two key points were carefully considered in the experimental design: 1) corn field, peanut field and fallow grassland were adjacent to each other horizontally. Sampling sites in these experimental fields were all at a narrow range of elevation i.e. 167e169 m, which assured that all the sampling sites in drawdown area had the same duration of reservoir inundation (~180 days). Impact of reservoir operation was identical among the experimental fields and fallow grassland in drawdown area; 2) all the treatments in the present study had the same land use history which assured that pre-impoundment GHG emissions and soil properties were the relatively same. 2.2. Site description and pre-impoundment status of soil properties Experiments were carried out in reservoir drawdown area at Lijiaba Bay, Gaoyang Lake in Pengxi River (Fig. 2). Pengxi River is one of the largest tributaries of the Yangtze River in the TGR. Located in the downstream of the Pengxi River backwater area, Gaoyang Lake (3106.1630 N, 108 39.9460 E; indicated as GY in Fig. 2; see photos in supplementary materials) was historically natural river reach and flat farmland for rice and fruit cultivation. It was then formed as an overflow lake with water surface area between 3 and 8 km2 and a drawdown area about 2e3 km2 along the shoreline regulated by the reservoir’s initial impoundment and follow-up seasonal operation. The study area was selected in drawdown area at Lijiaba Bay of Gaoyang Lake. Historical land use in the drawdown area was forest for fruit, e.g., navel oranges. Thorough reservoir clearance by wood cutting was done in the mid and late 1990s before initial impoundment of the TGR, creating about 3000 m2 fallow grassland with an average slope of 15 e18 near shore. About 600 m2 of the

drawdown area with an elevation range between 165 m and 175 m were used for corn cultivation, and about 300 m2 of the drawdown area with the same elevation range were used for peanut cultivation. In 2008, field survey at the study sites was carried out for preimpoundment soil properties (Zhang et al., 2012a, 2012b). Differences of pre-impoundment soil properties among experimental fields and fallow grassland were not significant, and preimpoundment soil heterogeneity and the potential impact to experiments could be ignored (see table of S2 in supplementary materials). In the present research, corn field, peanut field and fallow grassland for sampling and field measurement was selected at the elevation range from 167 m to 169 m as indicated above. Fallow grassland in drawdown area was initially muddy shoreline along the reservoir. Vegetation gradually recovered in fallow grassland after April and May with reduced water level and increased temperature. Unflooded grassland which is above the maximum water level, i.e. 175 m, is covered by vegetation year round. Grass species in both fallow grassland of drawdown area and unflooded grassland are mainly Cynodon dactylon and Vetiveria zizanioides, which are widespread in the region. Neither fallow grassland in drawdown area nor unflood grassland was disturbed by farming. 2.3. Field sampling and measurements Soileatmospheric CO2, CH4 and N2O fluxes were measured monthly using closed vented static chambers from March through September 2011. Sampling time was selected as 9 a.m. through 11 a.m., when instantaneous fluxes approximately represent mean flux of the day. Gas samples from the chamber were collected in airsampling bags (0.5 l, Hedetech, Dalian, China) seven times every 5 min over a half-hour period using a polypropylene medical syringe with triple valve, and analyzed by Agilent® 7890A gas chromatography within 24 h. Fluxes of CO2, CH4 and N2O were calculated by the change rate of gas sample concentration as follows:

Fig. 2. Three gorges reservoir and the location of research site. Pengxi River has a watershed area 5172.5 km2 and extents 3100ʹe3142ʹN and 107 56ʹe108 54ʹE. It is about 250 km upstream of the Three Gorges Dam.

Z. Li et al. / Journal of Environmental Management 181 (2016) 64e73

Flux ¼

Sl$P$ F1 $F2 $V ; Sp$R$ð273:15 þ TÞ$S

where Flux (mmol m
2 h
1) is the flux of CO2, CH4 or N2O; Sl (mg L
1 s
1) is slope in a concentration-time graph. R2  0.90 was required in linear regression. P (kPa) is atmosphere pressure. F1 is a conversion coefficient from mg L
1 to mmol m
3and F2 from min to h. V (m3) refers to air volume in the chamber. Sp is normal atmospheric pressure 101.325 kPa, and R is a constant at 8.314. S (m2) represents soileair area covered by the chamber and T ( C) is chamber temperature. Surface soil samples were taken as in Zhang et al. (2012b). Soil pH, temperature, and ORP in sampling sites was measured using an IQ 150 (HACH®, USA). Soil volumetric water content and conductivity were measured by a TDR 100 (Fieldscout®, Spectrum Technologies, USA). Soil nitrogen (Soil TN) and organic matter (SOM) were measured by Kjeldahl and potassium dichromate-external heating method, separately (Zhang et al., 2012b). And soil phosphorus (Soil TP) was determined by processing soil samples at 450  C, followed by hydrochloric acid extraction. Phosphate concentrations in the supernatant of the extraction were analyzed by using the molybdenum blue method (Zhang et al., 2012a). Daily meteorological data, i.e., solar radiation, air temperature, relative humidity, and rainfall were collected by onsite meteorological sensors with a CR1000 meteorological control module (Campbell Scientific®, USA). Corn and peanut production were weighed immediately after harvest in July and August. To estimate total vegetation biomass, the grain-to-straw ratio was used. According to the Statistical Yearbook of Chongqing 2012 (Huang and Tong, 2012), this ratio for corn was 1.2 in 2011 and that of peanuts was 0.8. After harvesting, corn field and peanut field were left until the next impoundment. We ignored little vegetation recovered after harvesting in these fields. For fallow grassland in drawdown area and unflooded grassland, weighting was used for evaluating biomass production at the end of September. Biomass per square meter of vegetation, i.e., all plant bodies above and under the ground in fallow grassland of drawdown area and unflooded grassland were weighed after heating at 105  C for 15 min and then dried at 80  C for 48 h until constant weight.

2.4. Evaluation of GWP and GHGI Theoretically, there are two net effects of GWP in farming practices in the drawdown area: 1) that associated with reservoir impoundment and operation, whose controlled background is preimpoundment GHG emissions; 2) that associated with farming during the drawdown period, whose controlled background is the case without farming, e.g., GHG emissions from fallow grassland in drawdown area. Since we did not have the data of preimpoundment GHG emissions and there were no significant differences between pre-impoundment soil properties of experimental sites in drawdown area and those in unflooded grassland, we assumed that simultaneous measurements of soileatmospheric GHG fluxes in unflooded grassland as a background of preimpoundment GHG emissions, whereas data from fallow grassland were considered as control between farming practices in drawdown area. Therefore, net GWP from reservoir construction and operation (GWPr) was calculated as GWPr ¼ CO2 equivalents in drawdown area (soil respiration þ soil N2O flux þ soil CH4 flux)
CO2 equivalent in unflooded grassland

67

Net GWP from farming practices in drawdown area (GWPf) was calculated as GWPf ¼ CO2 equivalents in fields (soil respiration þ soil N2O flux þ soil CH4 flux)
CO2 equivalent in fallow grassland Given that crop cultivation in drawdown area was from March through September, annual GWPr and GWPf were the sums of monthly values from that period. Farming practices in drawdown area involved no machinery or energy input, e.g., pumping for irrigation, fertilization and harvest transportation. Excess energy consumption and GHG emissions from fuel or power were not taken into consideration. CO2 equivalents of CH4 and N2O fluxes were calculated by multiplying their values by 298 and 25, respectively (Solomon, 2007). Since data on root respiration for corn, peanuts and grass were limited in drawdown area, to determine soil respiration, we multiplied soil CO2 emission by a conversion rate of 0.52 to eliminate root respiration, as suggested by Mosier et al. (2006) and applied by Sainju et al. (2014b). In agro-systems, GHGI was used to calculate GHG production from crop harvests (Mosier et al., 2006; Sainju et al., 2014a). This helps evaluate GHG effects in various types of farming. To compare GHG emissions between farming and natural vegetation recovery, and between different types of grain harvest, we extended the concept of GHGI and used it to evaluate GHG effects of farming in drawdown area. Since GWPr indicated the net GHG effect of reservoir operation, we divided GWPr of each treatment type by full biomass with dry weight to obtain GHGIb, which measures the intensity of GHG emissions per unit biomass. GHGIg was then calculated by dividing GWPr by the total grain harvest of corn and peanut field, which gave the GHGI of corn and peanut production. 2.5. Statistical analysis Variance analysis (ANOVA) with Fisher’s Least Significant Difference (LSD) significance was used to test GHG fluxes and environmental parameters differences in different land uses. Redundancy analysis, which was performed on CANOCO and validated by Monte Carlo permutation methods with 199 runs, was applied to elucidate major environmental parameter drivers of GHG emissions in both the drawdown area and unflooded grassland. A value of p  0.01 was considered acceptable. All datasets were inputted to Origin® software for figure construction. 3. Results 3.1. GHG fluxes in drawdown area and unflooded grassland In the study site, mean soil CO2 emission for all land-use types from March through September 2011 was 10.38 ± 0.97 mmol m
2 h
1 (mean ± standard error). Corresponding CH4 and N2O fluxes were
8.61 ± 2.15 mmol m
2 h
1 and 3.42 ± 0.80 mmol m
2 h
1. In the drawdown area, soil CO2 emissions did not show significant differences among corn field, peanut field and fallow grassland (analysis of variance (ANOVA) with Fisher’s Least Significant Difference (LSD) significance (sig.) > 0.05). However, soil CO2 emissions in the drawdown area were much greater than those in unflooded grassland (ANOVA with LSD, sig.  0.05). Mean CH4 fluxes in drawdown area and unflooded grassland were negative during the study. CH4 fluxes in peanut field among the sampling sites showed a maximum CH4 sink, whereas the fluxes in corn field indicated the minimum CH4 sink (ANOVA with LSD, sig.  0.05). N2O fluxes in drawdown area and unflooded grassland were positive, indicating a N2O source. N2O fluxes in corn field were

68

Z. Li et al. / Journal of Environmental Management 181 (2016) 64e73

Table 1 Soileatmospheric GHG fluxes in drawdown area and unflooded grassland. Positive flux indicated source to atmosphere; negative flux indicated sink in soil. Land-use type Fallow grassland

Corn field

Peanut field

All data sets in the drawdown area

Unflooded grassland

Mean S.D. Range Mean S.D. Range Mean S.D. Range Mean S.D. Range Mean S.D. Range

CO2 fluxes (mmol m
2 h
1)

CH4 fluxes (mmol m
2 h
1)

N2O fluxes (mmol m
2 h
1)

10.65 ± 2.34 5.73 2.84e19.68 10.18 ± 0.80 1.96 7.31e13.63 10.31 ± 1.79 4.38 5.32e19.10 10.38 ± 0.97 4.32 2.84e19.68 5.13 ± 0.89 2.17 1.98e9.02


10.67 ± 4.11 10.07
22.70e6.15
2.15 ± 1.64 4.03
9.04e3.70
13.03 ± 3.95 9.68
25.47e2.04
8.61 ± 2.15 9.61
25.47e6.15
0.89 ± 0.69 1.69
2.32e2.21

2.18 ± 0.72 1.76 0.04e0.78 6.64 ± 1.75 4.29 2.59e15.94 1.44 ± 0.34 0.83 0.28e2.61 3.42 ± 0.80 3.56 0.04e15.94 0.35 ± 0.05 0.11 0.20e0.50

significantly greater than those in fallow grassland, peanut field and unflooded grassland (ANOVA with LSD, sig.  0.05), but differences in those fluxes between fallow grassland and peanut field were not significant (ANOVA with LSD, sig. > 0.05). Although mean N2O fluxes during the study were low in fallow grassland and peanut field, the difference did not pass the ANOVA significance test (LSD, sig. > 0.05) (Table 1). Monthly variations of GHG fluxes were apparent (Fig. 3). Soil CO2 emissions had a general increase from March through September 2011. In fallow grassland, corn field and unflooded grassland, soil CO2 emissions maximized between July and August and decreased in September. In peanut field, maximum soil CO2 emission was detected in May. CH4 fluxes were initially positive in March. Negative CH4 fluxes were detected all the treatments, indicating a transition of these fluxes from source to sink. During the entire summer season (June to August), negative CH4 fluxes varied among the sampling sites. N2O fluxes in fallow grassland and peanut field decreased from March to April. There was a significant decrease of N2O fluxes in fallow grassland in June and July. N2O fluxes in fallow grassland and peanut field minimized in September. N2O flux in corn field maximized in May. Unflooded grassland did not show a significant decrease of those fluxes at the end of summer, in contrast to the study site.

3.2. Variation of environmental parameters Variations of meteorological condition were discussed in supplementary material. Monthly variations of soil volumetric water content (VWC) in unflooded grassland had a trend very similar to rainfall, indicating that precipitation was the major source of soil water in the area. However, soil volumetric water content during March and April in all the treatments in drawdown area were significantly higher than in unflooded grassland (ANOVA with LSD, sig.  0.05). Intense evaporation in August reduced soil water content in all the treatments to low levels. Monthly ORP in drawdown area showed alternation between positive and negative values. However, in March, soil ORP there were generally negative whereas that in unflooded grassland was positive. Minimum soil conductivity in fallow grassland was detected in March, whereas that in corn field and peanut field were in August. Soil pH in unflooded grassland was significantly lower than in the treatments in drawdown area (t-test, sig.  0.05). From March through September, there were apparent decreases of soil TN in both unflooded grassland and drawdown area. There were general increases of soil organic matter in various land-use types of drawdown area, but its increase in unflooded grassland was not significant (Fig. 4).

Fig. 3. Monthly variations of GHG fluxes in corn field, peanut field, fallow grassland and unflooded grassland.

Z. Li et al. / Journal of Environmental Management 181 (2016) 64e73

69

Fig. 4. Environmental parameters in drawdown area and unflooded grassland during the study period.

The redundancy analysis provided potential impact factors that regulated GHG emissions for different land uses (Fig. 5). All those analyses using the full model passed Monte Carlo testing (p  0.01) for model validation. In fallow grassland, air temperature, photosynthetic active radiation, soil conductivity and ORP had significant positive effects on the increase of soil CO2 emission. Soil TN and volumetric water content had positive effects on CH4 emissions, whereas soil organic matter had a negative effect on that emission. In corn field, CO2 emission was positively correlated with soil organic matter, air temperature and photosynthetic active radiation. However, CH4 emission in corn field was negatively correlated with these environmental parameters. In corn field, N2O emission was stimulated by increase of soil volumetric water content and relative humidity. This was similar to the N2O emission in peanut field. CO2 emission in peanut field was negatively correlated with soil TN, whereas increase in air temperature and photosynthetic active radiation contributed to CO2 emission in peanut field. This was similar to corn field. Emission of CH4 in peanut field was

negatively correlated with soil organic matter, pH, air temperature, and photosynthetic active radiation. In unflooded grassland, increase of N2O emission was also stimulated by increase of soil volumetric water content and relative humidity, as in corn field. CO2 emission increased with increase of photosynthetic active radiation and air temperature but with decrease of soil organic matter and soil TN. Emission of CH4 was positively correlated with ORP and soil organic matter, but negatively correlated with pH and rainfall. Emission of N2O in unflooded grassland was again stimulated by increase of soil volumetric water content and relative humidity. 3.3. Grain harvest and biomass yields in the drawdown area In August, the corn harvest was 506.3 ± 34.6 g m
2 and the peanut harvest was 175.4 ± 20.8 g m
2. Although both harvests were slightly smaller than those in typical arable fields of the area before impoundment, the differences were not significant (t-test,

70

Z. Li et al. / Journal of Environmental Management 181 (2016) 64e73

Fig. 5. Redundancy analysis between environmental parameters and GHG emissions in drawdown area and unflooded grassland. Red arrows are environmental factors. Blue arrows are different types of GHG. In the figure, RH ¼ relative humidity; VWC ¼ soil volumetric water content; PAR ¼ photosynthetic active radiation; EC ¼ soil conductivity; Atemp ¼ Air temperature; TN ¼ soil total nitrogen. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

sig.  0.05) and were in the normal statistical range grain harvests of the TGR region (Huang and Tong, 2012). For vegetation biomass (dry weight) of all land uses, that in corn field was 1113.9 ± 89.1 g m
2, among the greatest biomass production. Biomass in unflooded grassland was 247.7 ± 19.8 g m
2, among the lowest during the study period (Fig. 6).

3.4. GWP and GHGI GWP is a relative measure of how much heat GHG traps in the atmosphere (Rosa and Schaeffer, 1995). Fig. 7 and Table 2 show that soil CO2 emission was the main contributor to GWP for all land uses, representing up to 98% of gross GWP. The methane sink in most months caused negative CO2 equivalent values of methane in the GWP calculation. The maximum ratio of methane in gross GWP was in fallow grassland during March, at 1.67%. Owing to N-fertilization, N2O contributed 49.5% of gross GWP in May 2011. However, this value significantly decreased to around 1%e2% after July in both the treatments of drawdown area and unflooded grassland. August had the greatest gross GWP in unflooded grassland, accounting for 24.9% of annual gross GWP. Fallow grassland had a similar monthly variation as unflooded grassland, where the maximum gross GWP in July was among the highest monthly values. However, because of the transition from methane source to sink, fallow grassland had the least gross GWP in April, leading to the minimum GWPr. N-fertilization in corn field increased gross GWP in both March and April. Highest monthly gross GWP in corn field was

Fig. 6. Biomass of vegetation after harvest in August. FG ¼ fallow grassland; CF ¼ corn field; PF ¼ peanut field; UG ¼ unflooded grassland.

in May, constituting 19.8% of annual GWP in corn field. Peanut field had monthly variations of gross GWP similar to corn field. Annual gross GWP of fallow grassland was 13,859.54 kg CO2eq ha
1. For corn field, annual gross GWP in NRG was 16,604.14 kg CO2eq ha
1, among the highest of all land-use types. Unflooded grassland had the least gross GWP of all landuse types, at 6259.80 kg CO2eq ha
1. Compared with unflooded grassland, GWP in drawdown area was significantly greater. This led to a positive value of GWPr. However, since peanut field had lower gross GWP compared with fallow grassland, GWPf in peanut field had the negative value
132.41 kg CO2eq ha
1. Results showed that corn field had the lowest GHGIb, at 0.13 kg CO2eq kg biomass
1. GHGIb of peanut field was 0.27 kg CO2eq kg biomass
1, while that in fallow grassland was 0.24 kg CO2eq kg biomass
1. GHGIg in corn field was 0.28 kg CO2eq kg grain
1. GHGIg in peanut field was 0.51 kg CO2eq kg grain
1, about 1.8 times greater than that in corn field. The low level of GHGIb and GHGIg in corn field was mainly because of its maximum biomass product and corn harvest in drawdown area.

4. Discussion 4.1. Factors impact GHG emissions in drawdown area and unflooded grassland In March and April, drawdown area was muddy with high water content, owing to initial exposure upon water level decline. This created a reductive soil environment in drawdown area. Soil organic matter in drawdown area was greater than that in unflooded grassland. These facilitated an environment for substantial CH4 and N2O emissions during initial drawdown area exposure (Beringer et al., 2013). From March to April, unflooded grassland exhibited a significant increase in soil water content (about 127%), but there was a slight decrease in drawdown area (about 4%e16%). There, N-fertilization in corn field promoted an increase N2O emission in April relative to fallow grassland and peanut field. Urea application in drawdown area resulted in a reductive environment, with decrease in ORP and increase in soil TN and soil conductivity following N-fertilization (Robertson et al., 2000; Sainju et al., 2012). Recent research by Tang et al. (2014) indicated that sedimentation rates in drawdown area in the TGR was around 0.5e10.0 cm yr
1. This supported the hypothesis of two major sources contributing to GHG emissions in the initial muddy period: 1) organic matter that settled in drawdown area during the preceding high water level period; 2) the prior year’s residues that are

Z. Li et al. / Journal of Environmental Management 181 (2016) 64e73

71

Fig. 7. CO2 equivalent of greenhouse gases in different treatments.

Table 2 Global warming potential and greenhouse gas intensities in various land uses. Land uses

CO2 equivalent CO2

Gross GWP CH4

kg CO2eq ha
1 Fallow grassland Corn field Peanut field Unflooded grassland

12,545.24 11,970.80 12,160.66 6031.75

GWPr

GWPf

kg CO2eq ha
1
218.65
43.74
267.71
18.36

GHGIb

GHGIg

kg CO2eq kg biomass
1

kg CO2eq kg grain
1

0.24 0.13 0.27 e

e 0.28 0.51 e

N2O

1532.95 4677.08 1015.78 246.41

13,859.54 16,604.14 12,908.73 6259.80

not completely released into water column during the last submersion. Soil organic matter stability and bioavailability are important to such emissions (Anderson, 1991) and flooding may contribute to surface soil organic matter (Sharifi et al., 2013). Therefore, supported by evidence in Fig. 5, we inferred that methane emissions in newly formed drawdown area (i.e., March and April) were regulated by various sources of bioavailable soil organic matter, most of which came from the sedimentation during the preceding high water level period. Rain events in the following May significantly increased soil water content in both drawdown area and unflooded grassland. Soil water content and pH in all treatments were in a similar range (~40% and ~7.5, respectively; Fig. 4). These major soil parameters gradually became similar between drawdown area and unflooded grassland. The second round of N-fertilization in corn field stimulated N2O emissions about 100% from April to May. N2O emissions in corn field were clearly greater than those in fallow grassland and peanut field, as was soil TN. Increase of air temperature and solar radiation enhanced soil and root respiration, potentially intensifying the sink flux of methane in drawdown area and unflooded grassland. CH4 fluxes showed a significant transition from source to sink in April or May, probably because of decrease in soil water content and change of ORP into an oxidative environment (Imer

1031.93 1409.13 899.52 e

e 377.20
132.41 e

et al., 2013; Mosier et al., 1997). Magnitudes of the CH4 sink in fallow grassland and peanut field were significantly larger than those in corn field and unflooded grassland. Whalen (2005) reported that seasonal change of temperature and soil water content could impact substrate precursors and microbial activities. Iqbal et al. (2013) inferred that agriculture in the TGR, e.g., in corn field of the present study, could disrupt the oxidation potential of methane in soils. Other studies showed that N-fertilization could inhibit CH4 oxidation (Stiehl-Braun et al., 2011; Willison et al., 1995). These findings supported the explanation of the relatively weak methane sink in corn field relative to peanut field and fallow grassland during our study period. From June to September, the flood season onset, rain events with increased solar radiation and temperature altered patterns of GHG emissions. Soil CO2 emissions peaked in August, indicating that both soil and root respiration increased with air temperature and solar radiation. Rain events and soil water content may not have changed soil CO2 emissions during the study period. The magnitude of methane sink in fallow grassland and corn field decreased about 50e60% from June to July, indicating that increase in soil water content would disrupt methane oxidation in the soil environment (Beringer et al., 2013). Continual dry and hot weather in August significantly increased the magnitude of the methane

72

Z. Li et al. / Journal of Environmental Management 181 (2016) 64e73

sink in fallow grassland and corn field. There was a significant drop in N2O emissions during summer in both drawdown area and unflooded grassland. Even in corn field with previous excess N input in March and April, there were steep declines of N2O emissions from June through September. In fallow grassland, N2O emissions decreased 95.0% during that period, among the highest of all land uses. In unflooded grassland, this decrease was ~45.4%, which was significantly smaller than that in drawdown area. N2O is an intermediate product of denitrification in soils (Saggar et al., 2013). Both the decrease in denitrification and increase in N2O oxidation may have contributed to the reduction of N2O emissions (Barneze et al., 2015; Senbayram et al., 2014). In our study, we could not furnish more evidence to distinguish which environmental factor was the main contributor to the decrease in N2O emissions from June through September. However, supported by Nelson and Terry (1996) and Gebauer et al. (1996), we inferred that water level fluctuation in drawdown area significantly altered soil physical properties, e.g., aggregating structure of soils and porosity. Soil parameters such as oxygen transferred and water content for microorganisms significantly changed. Therefore, compared with unflooded grassland with less disturbance by reservoir operation, we inferred that the significant drop of N2O emissions in drawdown area was mainly driven by change of soil physical properties caused by reservoir operation. Furthermore, since the decrease of N2O emissions in fallow grassland, corn field and peanut field were similar (Fig. 3) after June, we hypothesized that plant cover and agricultural activity did not significantly impact soil denitrification and N2O emissions in drawdown area. However, more evidence is needed to test this hypothesis. 4.2. GWP, GHGI and their implications for farming practices in drawdown area To compare GWP of different land uses in all the treatments and control, we considered the following constraints. C As indicated in Section 2.1, all historical land use in both drawdown area and unflooded grassland was woodland for fruits. This led to similar historical soil organic matter (supplementary material) and pre-impoundment GHG emissions. C Fallow grassland, corn field, and peanut field had similar elevations, i.e. 167e169 m (unflooded grassland was the exception) and therefore similar durations of submersion and exposure. Impacts of reservoir operation on soil physicochemical properties and soileair GHG emissions were independent of land use in drawdown area. We found that GWP for all land uses in drawdown area were about 50e60% greater than that in unflooded grassland, giving positive values of GWPr in drawdown area. Since the unflooded grassland and drawdown area had the same land-use history, we inferred that the operation of the TGR contributed to the net GHG effect in drawdown area. We discerned significant differences in gross GWP of fallow grassland, peanut field and unflooded grassland. Positive values of GWPf in corn field but negative ones in peanut field indicated the significant difference between carbon source and sink in drawdown area. N-fertilization was critical in the increase of gross GWP, resulting in a positive net GHG effect in corn field. Negative GWPf in peanut field showed that this crop field may also modulate GWP in drawdown area. We could not find any GWP and GHGI data on similar agriculture patterns, e.g., seasonal farming in drawdown area or riparian zone, nor we could find any reports with parallel comparison of GHG potentials of farming practices in drawdown area. Recently,

Jayasundara et al. (2014) showed that corn GHGI varied from 0.24 to 0.35 kg CO2eq kg grain
1 with regular agriculture and farmland management in Ontario, Canada; over 70% of GHGI there was associated with N-inputs. The result of our study (0.28 kg CO2eq kg grain
1) was in the lower part of the aforementioned range. Although farming in drawdown area limited the use of machinery that contributed about 10e20% to GHGI, it may also be inferred that farming there would not create excessive GHG emissions per unit grain harvest compared with harvests from normal farmland. Further research is required. In 2008, IPCC proposed a framework for assessment of net GHG emissions from reservoir systems (Edenhofer et al., 2011). An equation was established therein: Net emissions ¼ gross emissions
pre-impoundment emissions
emissions from unrelated anthropogenic sources The present study showed that reservoir operation produced net GHG emissions in drawdown area in the TGR compared to unflooded area. Various land uses in drawdown area, e.g., fallow grassland, corn field and peanut field in our study, should be included in calculating gross emissions in the above equation. Data from unflooded background with the same land-use history could be used for the pre-impoundment emissions. However, for farming practices in drawdown area, net emissions assessment is extremely complex. Certain types of crops, e.g., peanuts in the present study, generated less gross GWP in drawdown area. Moreover, harvesting grain and collecting grain straw for home use removed certain amounts of biodegradable carbon from drawdown area in the TGR. This apparently reduced carbon stock in drawdown area and potential GHG emissions after inundation during the subsequent high water level period. Farming in drawdown area would cause excessive soil erosion that potentially transports carbon into the reservoir during the low water level period (Tang et al., 2012). This indicated that more research is required for best farming practices in drawdown area, which would help reduce gross GWP or even mitigate net GHG emissions there. 5. Conclusion Soil water content, air temperature and rainfall were among the most influential factors of GHG emissions in drawdown area and unflooded grassland. Corn field in drawdown area had the maximum global warming potentials (GWP) mainly due to N-fertilization which promoted N2O emission from March to May. Gross GWP in peanut field in drawdown area was about 7% lower than that in fallow grassland. Compared to unflooded grassland, reservoir operation created positive net effect on GHG emissions and GWPs in drawdown area. However, crop species, e.g. peanut, and best practices in farming, e.g. prohibiting N-fertilization, could partially reduce GWPs in drawdown area. In the net GHG emissions evaluation in the TGR, farming practices in the drawdown area shall be taken into consideration. Acknowledgments This study was supported by the National Natural Science Foundation (Key Project, No. 41430750), the Chongqing Science and Technology Commission (Key Project, No. cstc2015jcyjBX0006), and the Chinese Academy of Sciences (Project No. KZCX2-XB314).We also thank Lin Wang, Chen Zhang, Zhiwei Sun and Hongtao Gao, who participated in field sampling and lab analysis. All the data in the manuscript are accessible, and could be requested from the corresponding author.

Z. Li et al. / Journal of Environmental Management 181 (2016) 64e73

Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jenvman.2016.05.080. References Anderson, J.M., 1991. The effects of climate change on decomposition processes in grassland and coniferous forests. Ecol. Appl. 1, 326e347. Backeus, I., 1993. Ecotone versus ecocline - vegetation zonation and dynamics around a small reservoir in tanzania. J. Biogeogr. 20, 209e218. Barneze, A.S., Minet, E.P., Cerri, C.C., Misselbrook, T., 2015. The effect of nitrification inhibitors on nitrous oxide emissions from cattle urine depositions to grassland under summer conditions in the UK. Chemosphere 119, 122e129. Beringer, J., Livesley, S.J., Randle, J., Hutley, L.B., 2013. Carbon dioxide fluxes dominate the greenhouse gas exchanges of a seasonal wetland in the wet-dry tropics of northern Australia. Agric. For. Meteorol. 182, 239e247. Chen, H., Wu, Y., Yuan, X., Gao, Y., Wu, N., Zhu, D., 2009. Methane emissions from newly created marshes in the drawdown area of the three gorges reservoir. J. Geophys. Res. Atmos. 114. Chen, H., Yuan, X., Chen, Z., Wu, Y., Liu, X., Zhu, D., Wu, N., Zhu, Q.a, Peng, C., Li, W., 2011. Methane emissions from the surface of the three Gorges reservoir. J. Geophys. Res. Atmos. 116. Ciais, P., Sabine, C., Bala, G., Bopp, L., Brovkin, V., 2013. Chapter 6: Carbon and Other Biogeochemical Cycles e Final Draft Underlying Scientific-technical Assessment. U. K., Cambridge, pp. 3e74. Ciais, P., Wattenbach, M., Vuichard, N., Smith, P., Piao, S.L., Don, A., Luyssaert, S., Janssens, I.A., Bondeau, A., Dechow, R., Leip, A., Smith, P.C., Beer, C., van der Werf, G.R., Gervois, S., Van Oost, K., Tomelleri, E., Freibauer, A., Schulze, E.D., Team, C.S., 2010. The European carbon balance. 2: croplands. Glob. Change Biology 16, 1409e1428. Edenhofer, O., Pichs-Madruga, R., Sokona, Y., Seyboth, K., Kadner, S., Zwickel, T., €mer, S., von Stechow, C., 2011. Renewable EnEickemeier, P., Hansen, G., Schlo ergy Sources and Climate Change Mitigation: Special Report of the Intergovernmental Panel on Climate Change. Cambridge University Press. Gebauer, R.L.E., Tenhunen, J.D., Reynolds, J.F., 1996. Soil aeration in relation to soil physical properties, nitrogen availability, and root characteristics within an arctic watershed. Plant Soil 178, 37e48. He, J., Jiang, F., Wang, J., Wu, C., Han, M., Gao, X., Wang, Y., Li, J., 2011. A new approach to increase farmers’ income by scientifically balancing between farming practice and environmental protection: case study on farming practices in drawdown area in the three gorges reservoir. In: Proceedings of the Twelfth Challenge Cup National Competition of College Student’s Extracurricular Science and Technology Works. Chongqing University, Chongqing, pp. 1e64. Hertwich, E.G., 2013. Addressing biogenic greenhouse gas emissions from hydropower in LCA. Environ. Sci. Technol. 47, 9604e9611. Huang, K., Tong, Z., 2012. Chongqing statistical Yearbook 2012. In: E.B.o.C.S (Ed.), Yearbook. Chongqing. Imer, D., Merbold, L., Eugster, W., Buchmann, N., 2013. Temporal and spatial variations of soil CO2, CH4 and N2O fluxes at three differently managed grasslands. Biogeosciences 10, 5931e5945. Iqbal, J., Lin, S., Hu, R.-G., Zhao, J.-S., 2013. CH4 oxidation potentials of different land uses in three gorges reservoir area of central subtropical China. Pedosphere 23, 609e619. Jayasundara, S., Wagner-Riddle, C., Dias, G., Kariyapperuma, K.A., 2014. Energy and greenhouse gas intensity of corn ( Zea mays L.) production in Ontario: a regional assessment. Can. J. Soil Sci. 94, 77e95. Li, Z., Zhang, Z., Xiao, Y., Guo, J., Wu, S., Liu, J., 2014. Spatio-temporal variations of carbon dioxide and its gross emission regulated by artificial operation in a typical hydropower reservoir in China. Environ. Monit. Assess. 186, 3023e3039. Liebig, M.A., Tanaka, D.L., Gross, J.R., 2010. Fallow effects on soil carbon and greenhouse gas flux in central North Dakota. Soil Sci. Soc. Am. J. 74, 358e365. Mosier, A.R., Delgado, J.A., Cochran, V.L., Valentine, D.W., Parton, W.J., 1997. Impact of agriculture on soil consumption of atmospheric CH4 and a comparison of CH4 and N2O flux in subarctic, temperate and tropical grasslands. Nutr. Cycl. Agroecosystems 49, 71e83. Mosier, A.R., Halvorson, A.D., Reule, C.A., Liu, X.J., 2006. Net global warming potential and greenhouse gas intensity in irrigated cropping systems in northeastern Colorado. J. Environ. Qual. 35, 1584e1598. Nelson, S.D., Terry, R.E., 1996. The effects of soil physical properties and irrigation

73

method on denitrification. Soil Sci. 161, 242e249. Oud, E., 1993. Global warming: a changing climate change for Hydro. Water Power Dam Constr. May 20e23. Robertson, G.P., Paul, E.A., Harwood, R.R., 2000. Greenhouse gases in intensive agriculture: contributions of individual gases to the radiative forcing of the atmosphere. Science 289, 1922e1925. Rosa, L.P., Schaeffer, R., 1995. Global warming potentials - the case of emissions from dams. Energy Policy 23, 149e158. Rudd, J.W.M., Harris, R., Kelly, C.A., Hecky, R.E., 1993. Are hydroelectric reservoirs significant sources of greenhouse gases. AMBIO 22, 246e248. Saggar, S., Jha, N., Deslippe, J., Bolan, N.S., Luo, J., Giltrap, D.L., Kim, D.G., Zaman, M., Tillman, R.W., 2013. Denitrification and N2O: N-2 production in temperate grasslands: processes, measurements, modelling and mitigating negative impacts. Sci. Total Environ. 465, 173e195. Sainju, U.M., Barsotti, J.L., Wang, J., 2014a. Net global warming potential and greenhouse gas intensity affected by cropping sequence and nitrogen fertilization. Soil Sci. Soc. Am. J. 78, 248e261. Sainju, U.M., Caesar-TonThat, T., Lenssen, A.W., Barsotti, J.L., 2012. Dryland soil greenhouse gas emissions affected by cropping sequence and nitrogen fertilization. Soil Sci. Soc. Am. J. 76, 1741e1757. Sainju, U.M., Stevens, W.B., Caesar-TonThat, T., Liebig, M.A., Wang, J., 2014b. Net global warming potential and greenhouse gas intensity influenced by irrigation, tillage, crop rotation, and nitrogen fertilization. J. Environ. Qual. 43, 777e788. Senbayram, M., Chen, R., Wienforth, B., Herrmann, A., Kage, H., Muehling, K.H., Dittert, K., 2014. Emission of N2O from biogas crop production systems in Northern Germany. Bioenergy Res. 7, 1223e1236. Sharifi, A., Kalin, L., Hantush, M.M., Isik, S., Jordan, T.E., 2013. Carbon dynamics and export from flooded wetlands: a modeling approach. Ecol. Model. 263, 196e210. Solomon, S., 2007. Climate Change 2007-the Physical Science Basis: Working Group I Contribution to the Fourth Assessment Report of the IPCC. Cambridge University Press. Stiehl-Braun, P.A., Hartmann, A.A., Kandeler, E., Buchmann, N., Niklaus, P.A., 2011. Interactive effects of drought and N fertilization on the spatial distribution of methane assimilation in grassland soils. Glob. Change Biology 17, 2629e2639. Tang, Q., Bao, Y., He, X., Zhu, B., Zhang, X., 2012. Farmer’s adaptive strategies on land competition between societal outcomes and agroecosystem conservation in the purple-soiled hilly region, Southwestern China. J. Mt. Sci. 9, 77e86. Tang, Q., Bao, Y.H., He, X.B., Zhou, H.D., Cao, Z.J., Gao, P., Zhong, R.H., Hu, Y.H., Zhang, X.B., 2014. Sedimentation and associated trace metal enrichment in the riparian zone of the three gorges reservoir, China. Sci. Total Environ. 479, 258e266. Teodoru, C.R., Bastien, J., Bonneville, M.-C., del Giorgio, P.A., Demarty, M., Garneau, M., Helie, J.-F., Pelletier, L., Prairie, Y.T., Roulet, N.T., Strachan, I.B., Tremblay, A., 2012. The net carbon footprint of a newly created boreal hydroelectric reservoir. Glob. Biogeochem. Cycles 26. Tullos, D., 2009. Assessing the influence of environmental impact assessments on science and policy: an analysis of the three gorges project. J. Environ. Manag. 90 (Suppl. 3), S208eS223. Whalen, S.C., 2005. Biogeochemistry of methane exchange between natural wetlands and the atmosphere. Environ. Eng. Sci. 22, 73e94. Willison, T.W., Webster, C.P., Goulding, K.W.T., Powlson, D.S., 1995. Methane oxidation in temperate soils e effects of land-use and the chemical form of nitrogen-fertilizer. Chemosphere 30, 539e546. Xu, X., Tan, Y., Yang, G., 2013. Environmental impact assessments of the three gorges project in China: issues and interventions. Earth Sci. Rev. 124, 115e125. Yang, L., Lu, F., Wang, X., Duan, X., Song, W., Sun, B., Chen, S., Zhang, Q., Hou, P., Zheng, F., Zhang, Y., Zhou, X., Zhou, Y., Ouyang, Z., 2012. Surface methane emissions from different land use types during various water levels in three major drawdown areas of the three gorges reservoir. J. Geophys. Res. Atmos. 117. Yang, L., Lu, F., Wang, X., Duan, X., Song, W., Sun, B., Zhang, Q., Zhou, Y., 2013. Spatial and seasonal variability of diffusive methane emissions from the three gorges reservoir. J. Geophys. Res. Biogeosci. 118, 471e481. Zhang, B., Fang, F., Guo, J.S., Chen, Y.P., Li, Z., Guo, S.S., 2012a. Phosphorus fractions and phosphate sorption-release characteristics relevant to the soil composition of water-level-fluctuating zone of three gorges reservoir. Ecol. Eng. 40, 153e159. Zhang, B., Fang, F., Guo, J.S., Li, Z., Fu, C., 2012b. Distribution characteristics and evaluation of soil SOM, TN, TP in the water-level-fluctuating zone of three gorges reservoir. Fresenius Environ. Bull. 21, 3779e3785. Zhao, Y., Wu, B.F., Zeng, Y., 2013. Spatial and temporal patterns of greenhouse gas emissions from three gorges reservoir of China. Biogeosciences 10, 1219e1230.