Journal of Geophysical Research: Atmospheres

Determinants influencing seasonal variations of methane emissions from alpine wetlands in Zoige Plateau and their implications

Authors


Abstract

[1] To understand the seasonality of methane flux from alpine wetlands in Zoige Plateau, 30 plots were set to measure the methane emissions in the growing and nongrowing seasons in three environmental types: dry hummock (DH), Carex muliensis (CM), and Eleocharis valleculosa (EV) sites. There were clearly seasonal patterns of methane flux in different environmental types in the growing and nongrowing seasons. Mean methane emission rate was 14.45 mg CH4 m−2 h−1 (0.17 to 86.78 mg CH4 m−2 h−1) in the growing season, and 0.556 mg CH4 m−2 h−1 (0.002 to 6.722 mg CH4 m−2 h−1) in the nongrowing season. In the growing season, the main maximum values of methane flux were found in July and August, except for a peak value in September in CM sites. In the nongrowing season, the similar seasonal variation pattern was shared among all the three sites, in which the methane emissions increased from February to April. In the growing season, the determining factors were surface temperatures (r2 = 0.55, P < 0.05), standing water depths (r2 = 0.32, P < 0.01) and plant community heights (r2 = 0.61, P < 0.01), while in the nongrowing season, ice thickness (r2 = 0.27, P < 0.05; in CM and EV sites) was found most related to flux. In our understanding, the seasonality of methane emissions in our study areas was temperature- and-plant-growth-dependent, and the water table position was also very important to shape the temperature-and-plant-growth-dependent seasonal variation of flux with its vigorous variations in alpine wetland ecosystems. Different environmental types within the wetland also influenced the seasonal pattern of methane flux. For an accurate estimate of the global methane source strength of alpine wetlands, the pronounced seasonal or even temporal variability in methane emission from alpine wetlands should be taken into consideration.

1. Introduction

[2] Methane is a biogenic trace gas that plays a crucial role in the chemistry of Earth's atmosphere. Its atmospheric concentration is one of the important factors controlling the earth's climate [Hansen et al., 1988; Arn Teh et al., 2005]. Atmospheric CH4 concentration has been increasing at the rate of 0.5–0.8% annually since the industrial revolution, and noteworthily, at a rate of 4.9 ppb a−1 over the period 1992–1998 [Intergovernmental Panel on Climate Control (IPCC), 2001]. In the past 150 years, CH4 contribution to the radiative forcing has been 35% that by CO2 and about 22% that by all long-lived greenhouse gases [Lelieveld et al., 1998]. There are many sources for methane both anthropogenic and natural. Because of the prevalence of waterlogged and anoxic conditions, natural wetlands become an important source for CH4 emission, contributing an estimated 24.8% of the global budget [IPCC, 2001].

[3] Methane emission rates reported in the literature vary widely, partly due to large diurnal and seasonal variations [Batjes and Bridges, 1994]. Methane emission from wetlands results from the interaction of several biological and physical processes in the soil [Conrad, 1989]. Methane production is a microbiological process, which is predominantly controlled by the absence of oxygen and the amount of easily degradable material [Amaral and Knowles, 1995]. The vertical distribution of methane production is related to the seasonal average standing water level [Moore and Dalva, 1997], usually reaching a maximum just beneath the standing water level. Furthermore, the methane production is strongly regulated by the amount and quality of available substrate, pH and temperature [Yavitt and Lang, 1990; Westermann, 1996]. The wetland plant also plays a very important role in the main three aspects of methane emissions, including providing the conduit for methane transportation, supplying substrates for methanogens through root exudation and delivering O2 to oxidize methane through roots to rhizosphere [Dacey and Klug, 1979; Schütz et al., 1991]. Therefore, possible causes for seasonal variations of methane emissions are variations in temperature, substrates, plant and methanogen biomass [Yavitt et al., 1987; Schütz et al., 1991; Saarnio et al., 1997].

[4] Methane emissions from high-altitude wetlands are also of great importance because of the prevalence of waterlogged, anoxic conditions in seasonally thawed layers. However, our knowledge of CH4 emissions in alpine or subalpine wetlands is mainly confined to wetlands on the American continent [West et al., 1999; Wickland et al., 2001], besides some sporadic reports from Qinghai–Tibetan Plateau [Jin et al., 1999; Ding et al., 2004; Hirota et al., 2004]. Zoige alpine wetlands, located in eastern edge of Qinghai–Tibetan Plateau, a very important and sensitive area for climatic change [Xu and Yao, 2001; Yao et al., 2002], are typical alpine wetlands of great importance as hot spots for biodiversity in the world [Wu, 1997]. Zoige alpine wetlands were one of the three major methane emission centers of Qinghai–Tibetan Plateau [Jin et al., 1999]. However, research on CH4 emission in Zoige alpine wetlands has lagged behind, so has systematic investigation on them. To scrutinize the CH4 efflux in Zoige wetlands, more research is urgently needed.

[5] The present paper aims (1) to understand the seasonal variation of methane emissions in growing and nongrowing seasons in an alpine wetland in Zoige Plateau and (2) to find out the key factors influencing the seasonal variations of methane emissions.

2. Materials and Methods

2.1. Site Description

[6] The investigations were carried out in an alpine wetland of Wetland National Nature Reserve of Zoige (33°56′N, 102°52′E, 3430 m above sea level (asl)), located on the northeast edge of the Qinghai–Tibetan Plateau (Figure 1), from June 2005 to April 2006. The altitude of Zoige Plateau is 3500 m on average, and alpine lakes and peatlands are well developed. The alpine wetlands of this region cover an area of 6180 km2, which is 31.5% of the whole plateau of Zoige. This region belongs to cold Qinghai–Tibetan climatic zone. The range of mean annual temperature is about 0.6 to 1.2°C. The highest monthly mean temperature is 9.1 to 11.4°C in July, while the lowest one is −8.2 to 10.9°C in January. Air pressure is low, about 668.8 hPa, due to its high altitude. The range of mean annual precipitation is approximately 560 to 860 mm. There are two peaks of rainfall, in May and August separately.

Figure 1.

Location of the sampling site, an alpine wetland of Zoige Plateau.

[7] A typical close organic flat wetland (which is a kind of bog, with a unique microtopography, numerous hummocks scattered) was chosen for this study, which covered 28% of the whole Zoige wetlands, according to the Zoige wetlands classification of Mires of the Zoige Plateau [Cai et al., 1965]. The wetland consisted of waters about 5 cm deep on average and dry hummocks (the shape not very regular) about 20 cm high over the standing water level. At our measuring site, the dry hummocks were predominantly composed of Kobresia tibetica, Cremanthodium pleurocaule, Potentilla bifurca, Pedicularis sp. and with 90 to 95% coverage rate, covering about 45% of the whole site. Scattered in the hollow area were two predominant emergent plant populations: Carex muliensis and Eleocharis valleculosa. Therefore, for a comprehensive representation of the wetland, dry hummock, Carex muliensis and Eleocharis valleculosa sites were chosen for measurement. In all sites, the greatest root density was found in the 0 to 20 cm soil depth. Beneath this layer, there was a peat layer about 50 to 80 cm in depth. A 75 × 75 m sampling square has been fenced from May 2005.

2.2. Sampling Plots Establishment and Methane Flux Measurement

[8] Thirty plots in the study site were established along the installed boardwalk to minimize disturbance to the peatland during sampling. They included the three above mentioned environmental types: Carex muliensis (CM), Eleocharis valleculosa (EV) and dry hummock (DH). Among the 30 plots, 10 were for dry hummock sites, 9 for Carex muliensis sites, and 11 for Eleocharis valleculosa sites. The distance between any two plots is more than 5 m and less than 10 m, depending on the distribution of Carex muliensis, Eleocharis valleculosa, and dry hummock patches in the sampling area.

[9] The methane flux was measured with vented closed chambers [Hutchinson and Mosier, 1981; Mosier et al., 1991]. The chambers (30 cm in diameter, 50 cm in height) were made of cylindrical polyvinyl chloride (PVC) pipe. Through the top surface of the chamber, there was a pipe (0.5 mm in diameter) to connect with the ambient atmosphere, with a spiral part inside the chamber. The chamber anchors (20 cm in height) were driven 8–15 cm (depending on the stability of soils) into the soil 48 h prior to the flux measurement to maintain balance of the system. In order to minimize heating, aluminum foil was employed to cover the whole chamber, except for the driven-in-soil part. When the measurements began, we connected chamber tops and anchors with a tight rubber belt (2.7 mm in thickness) to make sure that the whole chamber was airtight.

[10] On the 16th of the four consecutive months from June to September 2005 and the same date from February to April 2006, the methane flux was measured at 0900 LT in Beijing standard time (GMT+8). Additional measurements were performed on 28 August 2005 and March 2006, in light of the consideration that August was the peak growing stage of plants and that March was the month when the soil was unfrozen and the ice was thawing quickly. For more synchronized results, when measurements were taken, each person was in charge of five plots. In 2006, not all of the 30 plots were measured due to the feasibility and possibility in that frigid environment. Eleven plots were chosen to measure, including five for dry hummock sites, three for Carex muliensis (CM) sites, and three for Eleocharis valleculosa (EV) sites. In the nongrowing season, both Carex muliensis and Eleocharis valleculosa sites were covered with ice.

[11] Four samples of the chamber air were manually pumped into 50 mL syringes at 10 min intervals over a 30 min period after enclosure. Samples were injected into gas collecting bags (made in Dalian, China) and delivered to Inner Mongolia Grassland Ecosystem Research Station, Chinese Academy of Sciences, for analysis in 2005. In 2006, the air samplings were analyzed in Chengdu Institute of Biology, Chinese Academy of Sciences. The weather was always fine when the measuring took place.

[12] Upon commencement of the experiment, on 13 June 2005, a trial was done on the study site to find out if the chamber temperature was different from the air temperature after a 30 min enclosure. We took three replicate chambers for this trial and measured both temperatures at 10 min intervals. The time chosen was at 1200 to 1230 LT, when the local temperature was rapidly increasing. The comparison showed that the mean air temperature was slightly lower than that in the chamber, but the difference was not statistically significant (P = 0.9 > 0.05). So we measured air temperatures instead of chamber temperatures. Air temperatures were simultaneously monitored by a mercury thermometer, while the temperatures at the vertical profile were measured by a digital meter (EcoScan, pH6).

[13] The CH4 concentration was determined by gas chromatography, Hewlett 5890 Packard Series II equipped with an injection loop, a FID (flame-ionization detector) operating at 200°C and a 2 m stainless steel column packed with 13 XMS (60/80 mesh). The column oven temperature was 55°C and the carrier gas was N2 with a flow rate of 30 mL min−1. Certified CH4 standard in 9.39 mL L−1 (China CH4 National Research Center for Certified Reference Materials, Beijing) was used for calibration [Wang and Han, 2005]. The rate of CH4 increase in the chamber air was calculated from a linear regression of concentration measured versus time with an average air temperature.

2.3. Environmental Factors

[14] A digital meter was introduced (EcoScan, pH6) for measurements of redox potentials and temperatures in the vertical profile. Redox potentials were measured at 5 cm intervals from 5 cm to 15 cm of the soil profile. The signal was considered to be constant when the drift of Eh was within 1 mV min−1. A reference electrode (200 mV) was employed to certify the meter before measuring. The water temperature in emergent plant sites and the ground surface temperature in dry hummock sites were recorded. Soil temperatures were measured at 5 cm and 10 cm of the soil profile. Soil samples were collected at 30 cm soil depth. Total carbon content, total nitrogen and phosphorus content were measured in the lab of Chengdu Institute of Biology, Chinese Academy of Sciences.

[15] Standing water depth and the hummock height over the standing water were recorded after the air sampling. The community height (the average height of vascular plants) of each plot was also recorded. In the nongrowing season, thaw depths and ice thickness were recorded at the same time.

2.4. Statistical Analysis

[16] Mean methane fluxes, surface and soil temperature, Eh, standing water depth, plant community height and aboveground biomass for each vegetation type were calculated by averaging the nine to eleven replicates for each sampling day. A full general linear model in which season was treated as an independent variable was used to compare the differences of environmental factors and methane fluxes in the growing season, and to assess the significance of the impacts of environmental type, season, and the combined effect of the two on methane fluxes and environmental factors.

[17] Simple linear regression analyses were carried out with CH4 emissions as dependent variable, and soil and vegetation characteristics as independent variables. The effect of a certain variable was considered significant when P < 0.05 and highly significant when P < 0.01.

3. Results

3.1. Seasonal Variation of Physical Factors

[18] On the whole year scale, the physical factors varied greatly between growing and nongrowing seasons (Table 1). Temperatures in the profile of sediments and redox potentials were much higher in growing season than those in the nongrowing season.

Table 1. Averages of Variables Determined at Dry Hummock Sites, Carex muliensis Sites, and Eleocharis valleculosa Sites in Growing Season and Nongrowing Seasona
 Growing SeasonNongrowing Season
DHCMEVDHCMEV
  • a

    DH, dry hummock sites; CM, Carex muliensis sites; and EV, Eleocharis valleculosa sites. N/D means no data.

CH4 emission, mg CH4 m−2 h−16.79 ± 2.1722.85 ± 12.3314.10 ± 5.100.090 ± 0.0611.47 ± 2.400.50 ± 0.73
Water temperature or surface temperature, °C15.56 ± 4.8917.97 ± 4.4418.05 ± 5.126.79 ± 1.974.66 ± 5.805.75 ± 7.36
5 cm sediment temperature, °C16.10 ± 3.0015.09 ± 2.8313.82 ± 2.573.58 ± 0.676.77 ± 0.896.80 ± 0.66
10 cm sediment temperature, °C14.90 ± 2.1413.71 ± 1.6812.67 ± 1.331.52 ± 0.223.03 ± 0.803.10 ± 0.53
5 cm sediment Eh, mVN/D−146.08 ± 38.97−156.84 ± 70.87N/D−105.67 ± 12.70−118.00 ± 8.54
10 cm sediment Eh, mVN/D−153.34 ± 39.70−182.40 ± 55.50N/D−109.33 ± 9.01−116.00 ± 4.58
15 cm sediment Eh, mVN/D−160.36 ± 41.85−181.20 ± 74.51N/DN/D−114.33 ± 6.66
Standing water table, cm−18.36 ± 12.687.12 ± 3.9310.66 ± 5.67N/DN/DN/D

[19] In the growing season, physical factors showed obvious seasonal variation patterns (Table 2). Air temperature, surface temperature and temperatures in the profile of sediments showed similar seasonal variation patterns in which highest temperature recorded in August in all three sites (Figure 2 just listing the seasonal variation of the surface temperature). The standing water table showed two peaks in June and September respectively in all three sites (Figure 3). The plant community height reached its greatest values in July in all three sites (Figure 3). The redox potentials at different depths of the vertical profile showed similar seasonal variation in the growing season. The lowest values recorded were in August (Figure 4).

Figure 2.

Seasonal variations of ground or water temperature and methane emissions at dry hummock sites (DH), Carex muliensis sites (CM), and Eleocharis valleculosa sites (EV) in growing season and nongrowing season.

Figure 3.

Seasonal variations of plant community height and standing water table at dry hummock sites (DH), Carex muliensis sites (CM), and Eleocharis valleculosa sites (EV) in growing season.

Figure 4.

Seasonal variations of redox potentials at the vertical profile of sediments in both Carex muliensis sites (CM) and Eleocharis valleculosa sites (EV) in growing season.

Table 2. Significance of Impacts of Environmental Types, Season, and Their Combined Effect on CH4 Emission and Environmental Factors in Growing Seasona
 Environmental TypeSeasonCombined Effect of Environmental Type and Season
  • a

    SI, significant impact α < 0.05; HSI, highly significant impact, α < 0.01; NS, no significant impact.

CH4 emission, mg CH4 m−2 h−1HSIHSISI
Water temperature or surface temperature, °CNSHSINS
5 cm temperature, °CSIHSINS
10 cm soil temperature, °CHSIHSINS
5 cm soil Eh, mVNSSINS
10 cm soil Eh, mVHSINSNS
15 cm soil Eh, mVNSNSNS
Standing water table, cmHSIHSINS
Community height, cmNSHSINS

[20] In the nongrowing season, seasonal variation of physical factors was also apparent. Surface temperature and temperatures in the profile of sediments reached their highest values from February to April (Figure 2 just listing the seasonal variation of the surface temperature). The ice thickness recorded its highest value in February and then decreased to zero in April both in Carex muliensis sites (CM) and Eleocharis valleculosa sites (EV), while the thaw depth reached its highest value from February to March. In April the sediment almost thawed in the dry hummock sites (Figure 5).

Figure 5.

Seasonal variation of ice thickness in both Carex muliensis sites (CM) and Eleocharis valleculosa sites (EV) and seasonal variation of thaw depth in dry hummock sites (DH) in nongrowing season.

3.2. Seasonal Variation of Methane Emissions

[21] In the growing season, seasonal variation patterns of methane emissions were different among all three sites (Table 2 and Figure 2). In the Carex muliensis sites (CM), the maximum of methane emissions was recorded in September, while the secondary peak was recorded in July. In Eleocharis valleculosa sites (EV), the maximal methane emission was recorded in July and then methane emission decreased gradually. In the dry hummock sites (DH), however, the greatest value for the growing season was in August. And the methane emissions were higher in Carex muliensis (CM) and Eleocharis valleculosa (EV) sites than those in the dry hummock sites. In the nongrowing season, the similar seasonal variation pattern shared among all the three sites, in which the methane emissions increased from February to April (Table 1 and Figure 2).

3.3. Key Factors of Seasonal Variations in Methane Emissions

[22] In the growing season, surface temperatures were significantly related to methane emissions (r2 = 0.55, P < 0.05). The standing water depth and plant community height were found most correlated to methane emissions in the three sites (r2 = 0.32, 0.61, P < 0.01). In Eleocharis valleculosa sites (EV), only the plant community height was significantly related to methane emissions (r2 = 0.99, P < 0.05). In the dry hummock sites, 5 cm, 10 cm sediment temperatures and plant community height were found most related to methane emissions (r2 = 0.47, 0.39, 0.51, P < 0.01).

[23] In the nongrowing season, surface temperatures were significantly correlated to methane emissions in all main plots, however, the coefficient was relatively low (r2 = 0.12, P < 0.05). In Eleocharis valleculosa sites (EV), surface temperatures were significantly correlated to methane emissions (r2 = 0.31, P < 0.05). In both Carex muliensis sites (CM) and Eleocharis valleculosa sites (EV), ice thicknesses were significantly correlated to methane emissions (r2 = 0.27, P < 0.05). In the dry hummock sites, no measured factors were found significantly correlated to methane emissions (Table 3).

Table 3. Linear Regressive Equations Between Methane Emissions and Key Factors in Two Emergent Plants Sites (Eleocharis valleculosa Sites and Eleocharis valleculosa Sites) and Dry Hummock Sites in Growing and Nongrowing Seasonsa
FactorsEquation
Growing SeasonNongrowing Season
  • a

    N/D indicates no data.

Surface temperature, °CF = −2.620 + 1.011*T (r2 = 0.55, P < 0.05, in three sites)F = −0.06 + 1.102*T (r2 = 0.12, P < 0.05, in three sites)F = 0.44 + 0.075*T (r2 = 0.31, P < 0.05, Eleocharis valleculosa sites)
5 cm sediment temperature, °CF = −3.459 + 0.639*T (r2 = 0.47, P < 0.01, dry hillock sites)N/D
10 cm sediment temperature, °CF = −4.232 + 0.742*T (r2 = 0.39, P < 0.01, dry hillock sites)N/D
Standing water depth, cmF = 15.10 + 0.2.293*D (r2 = 0.32, P < 0.01, in three sites)N/D
Plant community height, cmF = −3.287 + 0.792*D (r2 = 0.61, P < 0.01, in three sites), F = −14.314 + 0.971*D (r2 = 0.99, P < 0.05, Eleocharis valleculosa sites), F = 2.372 + 0.375*D (r2 = 0.51, P < 0.05, dry hummock sites)N/D
Ice thickness, cmN/AF = 0.322–0.053*T (r2 = 0.27, P < 0.05, two emergent plants sites)

4. Discussion

4.1. Seasonal Variation of Methane Emissions

[24] Mean methane emission rate was 14.45 mg CH4 m−2 h−1 (0.17 to 86.78 mg CH4 m−2 h−1) in the growing season, and 0.556 mg CH4 m−2 h−1 (0.002 to 6.722 mg CH4 m−2 h−1) in the nongrowing season. Methane emission from Zoige Plateau wetlands was higher than that from other alpine wetlands [Jin et al., 1999; West et al., 1999; Wickland et al., 1999, 2001; Hirota et al., 2004; Hu et al., 2005] (summarized in Table 4). This was attributable to the fact that the soil of alpine wetlands on Qinghai–Tibetan Plateau is extremely rich in organic matter [G. X. Wang et al., 2002] and that Zoige peatlands were one center of CH4 release on Qinghai–Tibetan Plateau [Jin et al., 1999]. Because of its well-developed alpine wetlands [Cai et al., 1965; Zhao, 1999] and a high emission rate, Zoige Plateau was considered to be playing an important role as a global CH4 source. However, because of Zoige alpine wetlands' sensitivity and frangibility to climatic changes, even on the same plateau, another research reported mean methane emission rate as 4.51 mg CH4 m−2 h−1 [D. X. Wang et al., 2002], which was much lower than the value in the present paper. Therefore, long-term multisite methane flux monitoring research is urgently needed to make more rational estimate of methane flux from alpine wetlands on Zoige Plateau.

Table 4. Comparison of Mean Methane Flux in Alpine Wetlandsa
Alpine Wetland EcosystemsVegetationMethane FluxStudy PeriodReference
Milligrams CH4 per Square Meter per HourMilligrams CH4 per Square Meter per Day
  • a

    N/D indicates no data available in each paper.

Alpine Wetlands on Zoige Plateau
In Zoige County (3430 m asl)Carex and Eleocharis14.45347Jul–Aug 2005this study
In Hongyuan County (3470 m asl)Carex wetlands4.51 (0.36–10.04)N/DMay–Sep 2001D. X. Wang et al. [2002]
 
Other Alpine Wetlands on Qinghai–Tibetan Plateau
In the Lanhaizhi wetland (3250 m asl)Carex, Scirpus, Hippuris and Potamogeton2.4657.74 Jul to 15 Sep 2002Hirota et al. [2004]
In the Huashixia Region (4300–4500 m asl)Carex, Hippuris and CalthaND25.4Jul–Aug 1996Jin et al. [1999]
Seasonal_flooded alpine wetlands (3280 m asl)Carex and Hippuris2.91–16.25N/D30 Jun to 4 Sep 2003Hu et al. [2005]
 
Alpine Wetlands in Rocky Mountains
In tundra, Colorado Front Range (3500 m asl)Carex meadowND8.45 (1.30–26.4)Jun–Sep 1992 and 1993West et al. [1999]
At southern rocky mountains (3200 m asl)Carex and EleocharisND251Jun–Sep 1998Wickland et al. [2001]

[25] A pronounced seasonal variation of methane emissions, i.e., maximal emissions in July and August and low but obvious emissions in winter, was found in the alpine wetland of Zoige Plateau, which was also observed in boreal peatlands [Dise et al., 1993; Alm et al., 1999], littoral zones of boreal lakes [Kankaala et al., 2004], tundra wetlands [Nakano et al., 2000], temperate wetlands [Kim et al., 1999] and lakes [Duan et al., 2005]. The seasonal variations of methane emissions were ascribed to ecological determinants, e.g., vegetation, climate and water regime. Using the eddy covariance technique (which was employed with a tunable diode laser spectrometer to quantify methane flux) to monitor methane emissions in a prairie marsh, Kim et al. [1999] found that maximum methane emission was recorded in the late summer (August) and the overall seasonal variation of methane emissions was significantly correlated to the sediment temperature. After two-summer measurements of methane flux in tundra wetlands, Nakano et al. [2000] reported that temporal variation in methane flux in waterlogged sites in permafrost areas was controlled by the thermal regime of a seasonal thaw layer; for summer season variation, methane flux was significantly correlated to centimeter degrees, the product of temperature and thaw depth. On the basis of 3-year measurements in a boreal lake, Kankaala et al. [2004] reported that the seasonal variation in methane emissions was significantly related to sediment temperature, but more weakly related to plant biomass.

[26] In our study, for the growing season, the seasonal variation of methane flux was found significantly correlated to surface temperatures, standing water depths and plant community heights; for the nongrowing season, ice thickness was found best related to flux (Table 3). In our understanding, the seasonality of methane emissions in the study area was temperature- and-plant-growth-dependent.

4.2. Temperatures

[27] Temperature accounted for the seasonal variation of methane emissions due to its effect on methanogenesis, methane transportation around roots [Hosono and Nouchi, 1997], plant growth, freezing and thawing, etc. In temperate wetlands, the sediment temperature was significantly correlated to methane flux [Kim et al., 1999]. About high-latitude wetlands (>60°N), several researches illustrated the relations between temperatures and methane flux. A correlation between flux and temperature was found only for wetter sites, while drier sites showed no such correlation [Svensson and Rosswall, 1984; Nykänen et al., 1998]. Bartlett and Harriss [1993] reported that seasonal changes in flux were closely correlated to ground temperature for both wet and dry sites. The best correlations were found between flux and centimeter degree, the product of temperature and thaw depth [Whalen and Reeburgh, 1992; Nakano et al., 2000].

[28] These findings were generally accorded with ours. However, in the wet sites (emergent plant sites), the result of the present paper showed that it's not sediment temperature, but surface temperature (ground or water temperature) that was found significantly correlated to methane emission in both the growing season and nongrowing season (Table 3). In our understanding, the special alpine climate, especially the changeable thermal condition, should be considered as a rational explanation. In that case, more detailed research is needed to prove it. For the dry hummock sites (drier sites), 5 cm and 10 cm sediment temperatures were best predictors for the seasonal variation of methane emissions in the growing season (Table 3) due to their deeper anaerobic layers (which is sensitive to temperature variation) without standing water. In the nongrowing season, surface temperatures were significantly correlated to methane emissions in all main plots. However, the correlation coefficient (r2 = 0.12) was low and therefore hard to explain the methane flux in nongrowing season. In tundra wetlands, Nakano et al. [2000] reported that temporal variation in methane flux in waterlogged sites in permafrost areas was controlled by the thermal regime of a seasonal thaw layer. We tried to do some measurements on ice temperature and thaw layer depth, but because of limited data collection and the sensitivity and fragility of alpine ecosystems, rational relation between thaw layer depth and methane emission was not found in this study. But it remains an interesting issue in alpine wetlands. In the high-frigid wetlands, such as boreal wetlands, tundra and alpine wetlands, temperature was a key ecological factor that limited the activity of methanogens (J. Q. Tian et al., unpublished manuscript, 2006) and then greatly influenced methane emission. Therefore, the seasonal thermal variation would result in the relatively seasonal variation pattern of methane emission. In other words, the seasonal variation of methane flux in the alpine wetland of Zoige Plateau was temperature-dependent.

4.3. Plant Community Height

[29] Plant growth has often been shown to control seasonal variations of methane emissions in vegetated wetlands [van der Nat and Middelburg, 2000; Joabsson and Christensen, 2001]. In wetlands, this is related both to effectively pressurized ventilation and oxidation of methane in rhizosphere of the actively growing plants and to emissions of substrates from rhizomes easily available for methanogens [Kim et al., 1999; Brix et al., 2001; Kankaala et al., 2004]. Kim et al. [1999] found the peak methane emission occurred 2–3 weeks after the peak shoot biomass; however, Brix et al. [2001] observed the maximal methane emission before the peak plant biomass, due to high water table and high availability of labile organic compounds for methanogens. Kankaala et al. [2004] found a more weakly relation between seasonal dynamics of methane emissions and plant growth.

[30] In contrast to the above mentioned papers, the present paper chose plant community height, which can directly indicate shoot biomass, as the predictor of plant growth [Ding et al., 1999]. Seasonality of methane flux was best correlated to seasonal dynamics of the plant community height (P < 0.01). In dry hummock (DH) and Eleocharis valleculosa sites (EV), the plant community height was also significantly related to emissions; while in the Carex muliensis sites (CM), the correlation was not significant. Single-factor environmental relationships are insufficient to amply explain the variation in methane flux [Whalen and Reeburgh, 1992; Christensen, 1993]. In Figure 2, the sequence of the standing water table is DH < CM < EV. The moderate standing water depth (in CM) would make the relation between plant growth and methane emission not significant, due to its great variation (CV: 55%). But the details of the threshold value of this water regime variation were not known. In the nongrowing season, the litters were not recorded.

4.4. Other Determining Factors

[31] The water table position determines the degree of aerobic and anaerobic metabolism in wetland sediments [Moore and Roulet, 1993]. But as shown in Figure 3 and 4, the seasonal dynamics of the standing water depth was not in accordance with the variation of redox potentials. In July and August, relatively high temperatures stimulated not only methanogenic activity but also activity of other aerobic microbes, with rapid consumption of oxygen. Hence there was a decrease in redox potentials and an increase in methane production [Ding et al., 2005]. The unlikeliness of an increase in redox potentials was probably due to the relatively low standing water depth in August (Figure 4). For high-latitude wetlands, Nykänen et al. [1998] found that in drier sites the seasonality of methane emission was significantly correlated to the water table. However, in the present study, it is standing water depths that was found most correlated to methane emissions in the whole studying area, despite the lack of significant relation between flux and the standing water depth in seasonal dynamics (Tables 2 and 3). This could be ascribed to spatial variation within the wetland. Therefore, standing water depth and redox potential variation should not be good predictors for seasonality of methane emissions due to their irregular fluctuation and indirect effects on methane emissions on the seasonal scale.

[32] In the nongrowing season, gas exchange between peat and the atmosphere can occur even through the frozen soil and snowpack [Dise, 1992; Alm et al., 1999]. Our results indicated that the ice thickness was best correlated to methane emissions in CM and EV sites which were covered with ice. This implied that methane emission should decrease with the increasing of the ice thickness.

[33] Environmental types also had significant impact on methane emission and even on its seasonal pattern (see Table 2). Large spatial variability of CH4 emissions from wetland ecosystems is a common phenomenon [Bartlett and Harriss, 1993; Bubier et al., 1993] due to the spatial variations of ecological factors, such as soil, vegetation, and water regime. On different scales, the variations of methane emissions were different. Therefore, for comprehensive presentation of the seasonal methane emission pattern in these special alpine wetlands with microhabitat variation, more attention should be paid to the above mentioned environmental types in sampling designs.

5. Conclusion

[34] Methane emission was measured in a temperate alpine wetland in Zoige Plateau. The emission rates showed a distinct seasonal variation in all three different environmental types of the wetland. The most important determining factors behind the seasonality of emissions are probably surface (ground or water) temperature and plant growth, while the water table position is also very important to shape the temperature-and-plant-growth-dependent seasonal variation of flux due to its vigorous variations in alpine wetland ecosystems. Different environmental types within the wetland also influenced the seasonal pattern of methane flux. The pronounced seasonal or even temporal variability in methane emission from alpine wetlands needs to be taken into account for an accurate estimate of the global methane source strength of alpine wetlands.

Acknowledgments

[35] This study was financially supported by Chinese Academy of Sciences (KZCX2-YW-418, KSCX1-07, KSCX2-01-09, joint-scholar project of the Bright of Western China), Chinese Science and Technology Ministry (2001BA606A-05), and Sichuan Science and Technology Bureau (03ZQ026-043). The Administrative Bureau of Wetland National Nature Reserve of Zoige and Zoige High-frigid Research Station, Chengdu Institute of Biology, Chinese Academy of Sciences, are thankful for the logistic assistance and guide in Zoige. We must give personal thanks to Zhang Ming and Li Hua for their suggestions and logistic arrangement on our field measurements, Yanbin Hao and Xiangzhong Huang for their help with analyses of gas samples, and Yucheng Yang and Qinchao Li for their assistance in gas sampling. Wan Xiong, an expert for ESP, is thanked for her great and patient help in our writing and reasoning. Three anonymous reviewers are thanked for their detailed evaluation and suggestion on our manuscript.

Ancillary