By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
 We investigated a possible mechanism underlying observed correlations between net ecosystem carbon dioxide exchange and methane emission. Using the technique of 14C pulse-labeling, we traced the movement of carbon fixed by photosynthesis as it moved through wet sedge and moist tussock tundra plant-soil mesocosms and was emitted as methane to the atmosphere. The 14C tracer provided a definitive way of quantifying the fate of recently fixed carbon. Carbon fixed by photosynthesis was measured as emitted methane from both moist tussock and wet sedge tundra mesocosms within 2 hours after labeling. Integration of time series measurements of methane emission showed that recent photosynthates are an important source of carbon for methane production. Approximately 2% of carbon fixed by photosynthesis at peak growing season was subsequently emitted as methane from moist tussock tundra, and approximately 3% was emitted as methane from wet sedge tundra. Measurements of soil pore water carbon pools demonstrate rapid transfer of 14C from plant carbon to dissolved forms and subsequently to the atmosphere as carbon dioxide or methane.
 Attempts to estimate current methane (CH4) emissions to the atmosphere and to predict future CH4 emissions from natural and managed ecosystems have been hampered by a lack of information about the key controls on CH4 production and oxidation which interact to control net emission. Many studies have examined factors that are correlated with CH4 emission, such as water table depth, soil temperature, water chemistry, and vegetation cover [Frolking and Crill, 1994; Moore et al., 1994; Shannon and White, 1994; Bubier et al., 1995; Mikkelä et al., 1995; Johnson et al., 1996; Moosavi et al., 1996]. However, modeling efforts to predict net CH4 emissions are complicated by the fact that numerous factors influence net CH4 emission, and no single-factor approach can fully explain the high spatial and temporal variability in observed CH4 emission [Whalen and Reeburgh, 1992]. Models based only on environmental factors such as water table depth and soil temperature cannot capture observed CH4 emission dynamics, and more recent process-based models [Cao et al., 1995, 1996; Walter et al., 1996; Walter and Heimann, 2000] appear to match observations only under a limited set of environmental conditions.
 Identifying the carbon sources of CH4 emissions is key to understanding spatial and temporal variations and to predicting future changes in the global CH4 budget. The primary substrates for microbial production of CH4 are acetate (CH3COOH) and carbon dioxide (CO2). These substrates become available to methanogens through a variety of pathways, including decomposition of soil organic matter (“old carbon”) and root exudation and root respiration (“new carbon,” also referred to as recently fixed carbon or recent photosynthates). Quantifying the contributions of old versus new sources of carbon is important in modeling the current and future global carbon budgets [Trumbore et al., 1995] and global CH4 budgets. Radiocarbon measurements of CH4 in soil pore water and emitted CH4 indicate that CH4 is produced mainly from new carbon [Chanton et al., 1988; Wahlen et al., 1989; Aravena et al., 1993; Chanton et al., 1995]. The purpose of this study was to directly measure the contribution of recently fixed carbon to CH4 emission.
 Past pulse-labeling studies have focused on measuring the carbon isotopic tracer in organic carbon compounds in the soil and plants and also in CO2 in the soil [Martin and Kemp, 1986; Milchunas and Lauenroth, 1992; Swinnen et al., 1994]. The flux of carbon in the form of CH4 remains relatively poorly characterized. Two research groups have investigated the contribution of photosynthates to CH4 emission from rice agroecosystems using pulse-labeling techniques. Using 13C as a tracer, Minoda and Kimura  and Minoda et al.  estimated that photosynthesized CO2 contributed 29% of the total CH4 emissions during a growing season and 22% of the total CH4 emissions if rice straw was applied to the soil. Dannenberg and Conrad  performed a 14C pulse-labeling study on rice plants and found that 3–6% of the assimilated 14CO2 was emitted as 14CH4 within 16 days after labeling. Watanabe et al.  measured contributions of carbon to CH4 emission from sources other than plants and estimated by difference that carbon supplied by rice plants made up approximately 80–85% of the total CH4 emission when rice straw was not applied to the soils. In the only published study of a natural ecosystem, Megonigal et al.  conducted a 14C pulse-labeling study on a single wetland plant and found that 14CH4 represented less than 1% of the 14C recovered 17 days after labeling. This paper reports the results of the first replicated mesocosm study of a natural wetland system to examine carbon fluxes in the form of CH4.
 High-latitude wetland ecosystems currently represent approximately 30% of the natural wetland source of CH4 to the atmosphere [Reeburgh et al., 1998]. Because of their high soil moisture and carbon contents, arctic ecosystems are potentially important as future sources of atmospheric CH4. Developing an understanding of the controls on carbon cycling in arctic tundra is critical to estimating current and future global CH4 budgets, as well as current and future arctic carbon budgets. This work was part of an integrated study using 14C pulse-labeling of tundra mesocosms to investigate C cycling in moist tussock and wet sedge tundra ecosystems [Loya et al., 2002; G. W. Kling et al., Gaseous and dissolved carbon production in Arctic ecosystems: The role of belowground processes, manuscript in preparation, 2002; K. J. Nadelhoffer et al., Pulse-labeling studies of carbon cycling in Arctic tundra ecosystems: Photosynthate allocation in moist tussock and wet sedge tundra vegetation, manuscript in preparation, 2002]. In this paper we present the results of measurements aimed at tracing the contribution of recent plant photosynthates to CH4 emission. We address the following questions: How quickly is the carbon fixed by photosynthesis emitted as CH4? How much carbon fixed by photosynthesis is emitted as CH4 in the same growing season?
2.1. Experimental Setup and Design
 Intact cores of soil plus vegetation were taken from moist tussock and wet sedge tundra sites at the Toolik Lake Long Term Ecological Research site on the North Slope of Alaska (68°38′N, 149°36′W) during August 1997. Each core was 27 cm in diameter and approximately 31 cm deep. The wet sedge tundra mesocosms were dominated by Carex chordorrhiza, and the moist tussock tundra mesocosms were dominated by Eriophorum vaginatum. Further description of these tundra types is provided by Shaver and Chapin . The cores (plant-soil mesocosms) were placed in 20 L polyethylene containers and transported to Woods Hole, Massachusetts.
 We conducted the experiment in growth chambers (model PVG36, Controlled Environments, Inc.) at the Ecosystems Center of the Marine Biological Laboratory over the period of a simulated arctic growing season. To induce plant senescence we gradually reduced the air temperature to −4°C and decreased the photoperiod from 12 hours of light to complete darkness. After one week of continuous darkness at −4°C, we simulated the start of the growing season. Over the period of one week, we gradually increased the air temperature to 10°C and increased the photoperiod to 24 hours of full light. The average flux of photosynthetically active radiation (PAR) at the leaf level under full light conditions was 800–1000 μmol photons m−2 s−1. The average soil temperature under 10°C and full light conditions was 15.2°C. The water table was kept constant at approximately 5 cm below the soil surface in moist tussock tundra mesocosms and at 2 cm above the soil surface in wet sedge tundra mesocosms through addition of distilled water. After 9.5 weeks under these conditions we began to simulate the end of the growing season by lowering the temperature and light levels. Over a period of three weeks we gradually decreased the photoperiod to 10 hours of light and reduced the air temperature to 6°C, with overnight freezes.
 Twelve each of moist tussock and wet sedge tundra mesocosms were pulse-labeled with 8 MBq (216 μCi) of 14C. We began to pulse-label the mesocosms on the 52nd day of our simulated growing season when the plants were close to maximum biomass. Three mesocosms (referred to as long-term mesocosms) each of moist tussock and wet sedge tundra were sampled intensively for gas fluxes for approximately 10 weeks after they were labeled. By the end of the experiment, the plants had senesced, and the mesocosms were harvested for analysis of 14C remaining in each mesocosm. The nine other mesocosms (short-term mesocosms) of each vegetation type were analyzed to examine the allocation of 14C carbon in all carbon pools, including plant and soil pools, at several intermediate times during the experiment. Three mesocosms of each vegetation type were destructively harvested at 1 day, 1 week, and 3 weeks after labeling. The 14C label distributions in the short-term mesocosms were used to examine changes in label distribution in plant and soil pools through time. The time series gas flux measurements from the original three long-term mesocosms of each vegetation type were used to estimate the gas fluxes from the short-term mesocosms harvested at intermediate time points in the experiment.
2.2. 14C Pulse Labeling
 Each plant-soil mesocosm was sealed with a clear acrylic headspace chamber during labeling. A 14C-labeled sodium bicarbonate solution (55 MBq/g C) was acidified with hydrochloric acid to produce 14CO2, which was taken up by plants through photosynthesis. Additional unlabeled CO2 was added as necessary to maintain the CO2 concentration near 400 ppmv inside the chamber. Each mesocosm was pulse-labeled for a period of 1.5 hours during which samples of headspace air were collected to determine uptake efficiency. The headspace chamber was removed after labeling. On average, 93% and 80% of the 14C label added (8 MBq) was taken up in moist tussock and wet sedge tundra mesocosms, respectively. Measurements in subsequent experiments showed that direct diffusion of label into the mesocosm soils during labeling was negligible (<4% of label added). Thus we assume that 14C measured in carbon pools such as emitted methane, soil pore water, and plant biomass was originally taken up through photosynthesis. The mesocosms and labeling methodology are described in greater detail by King .
2.3. Sampling and Analysis
 Gas fluxes of CO2, 14CO2, CH4, and 14CH4, and pore water samples were measured from each of the long-term mesocosms 1–2 times per day during the first week after labeling and 1–2 times per week thereafter. Gas fluxes were measured from the short-term mesocosms approximately every other week. Gas fluxes and pore water samples were also measured on each mesocosm immediately before it was destructively harvested.
 CH4 fluxes were measured using the static chamber method and by calculating the increase in chamber headspace CH4 concentrations over time [Whalen and Reeburgh, 1988]. Specifically, we sampled 10 mL of headspace air at 10-min intervals over 30 min (average chamber headspace volume = 35 L). A reflective covering was placed over the chamber to prevent excessive temperature changes during the flux measurement period. The chamber was placed over each mesocosm only during gas flux measurements. The air samples were analyzed for CH4 concentration on a gas chromatograph equipped with a flame ionization detector (described below). CH4 fluxes to the atmosphere, representing loss from the ecosystem, are considered positive.
 CO2 flux measurements were made using a portable infrared gas analyzer (LI-6200, Licor, Inc.). The photosynthesis system was calibrated using National Institute of Standards and Technology (NIST)-analyzed CO2 standards. The net ecosystem CO2 exchange rates were determined using a clear acrylic headspace chamber over a 3-min period. The chamber was darkened using a reflective covering for ecosystem respiration measurements, also made over a 3-min period. A gas chromatograph (GC-14A, Shimadzu Corp.) equipped with two 1-m Porapak Q columns, a flame ionization detector (FID) for CH4, and a thermal conductivity detector (TCD) for CO2 was used to analyze gas samples collected during CH4 flux measurements and gas samples extracted from soil pore water. Both gas chromatograph detectors were calibrated using standards that were referenced to NIST standards. Methane fluxes were calculated based on the linear regression of syringe samples that described the change in CH4 concentration over time. A linear regression of r2 <0.90 was rejected due to sampling error or bias of flux measurements due to other processes, such as experimenter-caused ebullition. Flux measurements of CO2 and CH4 were linearly interpolated and integrated over time using the trapezoidal rule to calculate the overall fluxes of CO2 and CH4 in each experiment. The fluxes of 14CO2 and 14CH4 (analyses described below) were integrated in the same way. These integrated fluxes of 14CH4 and the total amount of 14C taken up as 14CO2 (Section 2.2.) were used to calculate the fraction of assimilated 14CO2 emitted as 14CH4.
 A sample of headspace air was collected in a glass bulb (175 cm3) for 14C analysis of the CO2 and CH4 at the end of each CH4 flux measurement. Bulbs were flushed and filled using a small diaphragm pump (model 801501, Sensidyne Inc.) at a flow rate of 600 mL min−1. The sample bulbs were sealed by two valves and were processed within 48 hours for analysis of 14CO2 and 14CH4.
 Soil pore water samples were collected using stainless steel needles inserted in the soil. Gas extraction from soil pore water was performed by a headspace equilibration of bubble-free pore water with nitrogen (N2, gas), as described by Kling et al. . Dissolved CH4 concentrations were calculated from measured headspace concentrations using solubility coefficients from Yamamoto et al. . Dissolved organic carbon content was determined using a TOC analyzer (TOC-5000, Shimadzu Corp.).
 Gas samples were separated and prepared for 14C analysis on an oxidation line [Whalen and Reeburgh, 1990; King, 1999]. The gas samples were introduced to the oxidation line, and CO2, including 14CO2, was trapped in a 1 M sodium hydroxide (NaOH) trap. The rest of the gas sample passed through a combustion tube (800°C) containing CuO catalyst. Methane and other hydrocarbons, present only in trace amounts, were converted to CO2 and trapped in a second NaOH trap. Calibration tests of the oxidation line demonstrated that 99% of CH4 passed through the combustion tube was oxidized to CO2. Subsamples of the NaOH solutions were added to scintillation cocktail (Scintiverse II, Fisher Scientific) and analyzed in a liquid scintillation counter (LS 3801, Beckman Instruments, Inc.) for 14C activity. Corrections were made for quench caused by NaOH.
3. Results and Discussion
3.1. Biomass, CO2 Exchange, and Methane Emission Measurements
 The mean (± standard error, n = 3) aboveground biomass of the moist tussock tundra mesocosms harvested at the end of the experiment was 1047 ± 108 g m−2; average aboveground biomass of the wet sedge tundra mesocosms harvested at the end of the experiment was 257 ± 20 g m−2. The aboveground biomass measurements for moist tussock tundra are consistent with field observations of biomass [Shaver and Chapin, 1991; Shaver et al., 1992; Chapin et al., 1996]. The aboveground biomass values for wet sedge tundra were slightly lower than those reported from field observations [Shaver et al., 1998]. The ratio of aboveground to belowground biomass was approximately 2.7:1 in the moist tussock tundra and 1:2.4 in the wet sedge tundra mesocosms. The ratio observed in moist tussock tundra mesocosms is high relative to field observations, perhaps due to incomplete separation of fine roots from the soil or to decreased root growth or increased root mortality (turnover) in the mesocosms. The ratio observed in wet sedge tundra mesocosms is similar to field observations [Shaver and Chapin, 1991].
 The mean (± standard error, n = 3) net ecosystem CO2 exchange in moist tussock tundra was greater than in wet sedge tundra over the length of the experiment, (1.3 ± 0.2 versus 0.1 ± 0.2 g C m−2 d−1, respectively). CO2 fluxes directed to the atmosphere from the ecosystem are negative. The mean ecosystem respiration in moist tussock tundra was −4.4 ± 0.5 g C m−2 d−1, and the mean ecosystem respiration in wet sedge tundra was −1.9 ± 0.1 g C m−2 d−1. The rates of net ecosystem CO2 exchange and ecosystem respiration were similar to rates observed in the field near the sites where the mesocosms were obtained [Shaver and Chapin, 1991; Shaver et al., 1998]. The differences in CO2 exchange between moist tussock tundra and wet sedge tundra are consistent with higher biomass and higher productivity of moist tussock tundra communities [Bliss and Matveyeva, 1992; Oechel and Billings, 1992; Oechel et al., 1997].
 Rates of CO2 exchange were relatively constant throughout the experiment because constant environmental conditions were maintained in the growth chamber, except for stepwise changes in light and temperature near the end of the experiment to simulate the end of the growing season. The net ecosystem CO2 exchange rates measured at the time of labeling (data not shown) were not different (p = 0.8 and 0.5 for moist tussock tundra and wet sedge tundra, respectively) from the average net ecosystem CO2 exchange rates during the experiment. There was no evidence that the 14C-labeling had any effect on CO2 exchange rates of the mesocosms. Mean CO2 exchange rates in the long-term mesocosms tended to be higher than in the short-term mesocosms. This difference between the short-term and long-term mesocosms can be explained by differences in plant biomass; long-term mesocosms had higher biomass and were more productive.
 Mean (± standard error, n = 3) CH4 emission rates over the length of the experiment were 0.1 ± 0.02 g CH4-C m−2 d−1 (= 133 mg CH4 m−2 d−1) in moist tussock tundra mesocosms and 0.3 ± 0.04 g CH4-C m−2 d−1 (= 400 mg CH4 m−2 d−1) in wet sedge tundra mesocosms (Figure 1). The mean emission rates for moist tussock tundra and wet sedge tundra are significantly different from each other (p < 0.001) as expected based on field observations [Reeburgh et al., 1998].
 The CH4 fluxes from moist tussock tundra mesocosms were higher than fluxes observed in the field [Reeburgh et al., 1998] and consistent with other laboratory results [Johnson et al., 1996]. Our moist tussock tundra mesocosms were more representative of tussock rather than intertussock areas. Tussock areas typically exhibit higher methane emissions than do intertussock areas [Reeburgh et al., 1998], which is consistent with our observations. However, the CH4 emission rates from the moist tussock tundra mesocosms were still 2–4 times higher than in situ measurements above tussock-only sites, probably because the mesocosms had higher soil moisture than in the field. The higher CH4 emission rates measured in the moist tussock tundra mesocosms could be due to higher soil moisture in the mesocosms compared to in situ sites or to higher soil temperatures, as described below.
 The CH4 emission rates from wet sedge tundra mesocosms were 2–4 times higher than emission rates observed in the field for these vegetation types [Sebacher et al., 1986; Crill et al., 1988; Bartlett et al., 1992; Schimel, 1995; King et al., 1998; Reeburgh et al., 1998]. A possible explanation is that pore water movement was restricted within each mesocosm leading to more anaerobic conditions and a higher concentration of carbon compounds, which would increase CH4 production, decrease CH4 oxidation, and therefore increase net CH4 emission. Higher soil temperatures in the mesocosms (15°C) compared to average field soil temperatures (∼6°C) for 0–30 cm soil depth may also partially explain high CH4 emission rates from the mesocosms of both vegetation types. The temperature dependence of CH4 production is described by literature Q10 values of 1.3 to 28 [van Hulzen et al., 1999] and a range of 5.3–16 specifically for temperate and subarctic peat soils [Dunfield et al., 1993].
 The CH4 emission rates from moist tussock tundra mesocosms increased gradually over the first part of the experiment probably due to increasingly anaerobic conditions and increased belowground plant productivity (Figure 1a). The same pattern of increasing CH4 emission through the growing season has been observed from in situ measurements (W. S. Reeburgh, unpublished data, 1996). The rates of CH4 emission declined in both moist tussock and wet sedge tundra mesocosms at the end of the experiment due to decreased temperature and photoperiod in the growth chamber that simulated the end of the growing season.
3.2. Emissions of 14CO2 and 14CH4
 Emissions of 14CH4 were first detected in all long-term moist tussock and wet sedge tundra mesocosms within 2 hours after labeling (Figure 2). The maximum 14C activity of CH4 emitted from both vegetation types occurred within the first week after labeling (Figure 2). These results agree with results from rice plant and wetland plant studies which indicated detection of photosynthate-derived CH4 within 12–24 hours after labeling and maxima in 14CH4 emission rates within one week of labeling [Minoda and Kimura, 1994; Minoda et al., 1996; Dannenberg and Conrad, 1999; Megonigal et al., 1999]. Because the carbon assimilated during labeling still had to pass through several pools before emission as CH4, it is not surprising that the timing of maximum 14CH4 emission varied between mesocosms. The detection of 14CH4 emission within 2 hours of labeling indicates rapid conversion of plant-derived carbon to CH4 and rapid transport of CH4 out of the soil.
 The time course of 14CO2 loss from the mesocosms was the same in moist tussock tundra as in wet sedge tundra (data not shown). Large proportions of recently assimilated carbon were very quickly respired as CO2, probably through leaf respiration initially and then through root respiration. Decomposition of root exudates by soil microbes also contributed to fluxes of 14CO2. Detection of 14CO2 indicated that the pulse-labeling resulted in plant incorporation of the 14C label and that some photosynthates were immediately respired.
3.3. Proportion of 14C Emitted as Methane
 Overall, 1–5% of the plant productivity taken up as 14C during labeling was emitted as 14CH4 during the same growing season in moist tussock and wet sedge tundra mesocosms (determined by integration of traces in Figure 2). Emissions of 14CH4 from the long-term mesocosms were 1.3, 2.2, and 2.5% of 14C assimilation by moist tussock tundra and 1.3, 2.5, and 5.1% of 14C assimilation by wet sedge tundra. The mean percentage (± standard error, n = 3) of 14C label emitted as 14CH4 was 2.0 ± 0.4% from moist tussock tundra and 2.9 ± 1.1% from wet sedge tundra. These results suggest that recently fixed carbon makes up a large proportion of the total carbon emitted as methane from these ecosystems. Our measurements of respired CO2, emitted CH4, pore water dissolved organic and inorganic carbon, plant biomass, and soil carbon accounted for greater than 70% of the total 14C originally taken up. Our results are similar to results from other 14C pulse-labeling studies which reported that <1% to 6% of assimilated 14C was emitted as 14CH4 [Dannenberg and Conrad, 1999; Megonigal et al., 1999].
3.4. Origin of Methanogenic Substrates
 The carbon sources for CH4 production in soil pore water are the dissolved organic carbon (DOC) and dissolved inorganic carbon (DIC) pools. Measurements of these pools were made throughout the experiment to monitor possible changes in overall methanogen substrate availability. Average pore water concentrations of DOC were approximately 9.3 mM and 1.6 mM in moist tussock tundra and wet sedge tundra mesocosms, respectively. Average pore water concentrations of DIC were approximately 8.8 mM and 10.2 mM in moist tussock tundra and wet sedge tundra mesocosms, respectively. Wet sedge tundra mesocosms had higher pore water concentrations of dissolved CH4 (∼400 μM) than did moist tussock tundra mesocosms (∼100 μM; Figure 3), which is expected due to the anaerobic conditions in wet sedge tundra soils. Total pore water concentrations of DOC, DIC, and dissolved CH4 were relatively constant throughout the experiment (data not shown).
 The 14C activity of the DOC and DIC pools was measured to determine the availability of plant-derived carbon in the soil pore water for CH4 production. The measurements of 14C in DOC and DIC pools show that plant-derived carbon was quickly transferred to the soil pore water where it became available as a substrate for the soil microbial community [see Loya et al., 2002]. The 14C label was detected in all pore water carbon pools within hours of labeling. Dissolved 14CH4 also appeared in the pore water of moist tussock tundra mesocosms within hours of labeling, indicating that conversion of plant-derived carbon to CH4 occurred as soon as it was available in the soil. In moist tussock tundra mesocosms, the specific activity of pore water CH4 reached a maximum within the first day after labeling and generally decreased thereafter (Figure 4a). The maximum specific activity of DIC occurred 4–8 days after labeling, and the maximum specific activity of DOC occurred 8 or more days after labeling. The specific activities of all dissolved pore water components in moist tussock tundra mesocosms did not change in response to changes in light and temperature at the end of the experiment.
 The specific activities of pore water components in wet sedge tundra were more variable than in moist tussock tundra. Detection of 14C in DOC and DIC pools soon after labeling indicated that recently fixed carbon was quickly available as substrate for methanogenesis. The maximum specific activity of dissolved CH4 occurred approximately 6 days after labeling (Figure 4b) and coincided with the maximum specific activity of DOC and DIC and with the highest emission of 14CH4. These results suggest that fluxes between pore water carbon pools in wet sedge tundra occur rapidly. Transient increases in the specific activity of pore water carbon pools may have been a byproduct of plant translocation of previously fixed carbon from aboveground tissues to belowground tissues in response to lowered temperature and light levels at the end of the experiment.
 The results of this pulse-labeling study show that transport of carbon through the plant-soil system and emission as CH4 occurs quickly. The 14C label first appeared as emitted CH4 within 2 hours of labeling in both wet sedge and moist tussock tundra. The maximum flux of 14CH4 occurred within one week after the pulse-labeling. Such fast turnover of carbon means that changes in plant uptake of carbon on short timescales (days to weeks) may be important controls on CH4 emissions. Similar results were reported for rice systems [Minoda and Kimura, 1994; Minoda et al., 1996; Dannenberg and Conrad, 1999], which suggests that the same processes leading to CH4 emission are occurring in very different ecosystems and also suggests that it may be possible to model CH4 emissions for various ecosystems using relationships between net ecosystem CO2 exchange and CH4 emission.
 Our results show that recent photosynthates are an important carbon source for emitted CH4. Overall, approximately 2–3% of 14C incorporated at mid season was emitted as CH4 in wet sedge and moist tussock tundra. Carbon taken up by plants as 14CO2 was traced through the plant-soil system of the mesocosms and was measured as emitted 14CH4. However, we did not find a simple relationship between plant productivity and CH4 emission, as described by Whiting and Chanton . That relationship predicts higher CH4 emission with higher plant productivity, but our measurements show lower CH4 emission from moist tussock tundra, which demonstrated relatively higher plant productivity. Artifacts caused by the laboratory setting of this study (e.g., stagnant water column) probably account for the observed pattern.
 By using a carbon tracer we have directly measured conversion of plant photosynthates to emitted CH4. These are among the first experimental results to quantitatively demonstrate a mechanism linking plant productivity and CH4 emission. The gas exchange measurements demonstrate that, although the amount of carbon exchanging with the atmosphere as CO2 or CH4 at any given point in time is small, the integrated amounts over the period of a growing season are significant. We measured the conversion of 2–3% of recent plant productivity to CH4 emission in this study. The average measured CH4 emission from wet sedge and moist tussock tundra was 12.75 g C m−2 for the growing season. Average plant productivity in these tundra ecosystems is 647 g biomass m−2 [Shaver and Chapin, 1991], and biomass is composed of approximately 50% carbon. Converting 3% of average plant productivity to CH4 emission results in a flux of CH4 originating from recent plant productivity of 9.71 g C m−2 during a growing season. Thus, using our results and average values for plant productivity in back-of-the-envelope calculations, we estimate that greater than 75% of average CH4 emissions from these ecosystems originate from recently fixed carbon. These observations support radiocarbon evidence that recently fixed carbon is a significant carbon source of emitted CH4 [Wahlen et al., 1989; Chanton et al., 1995; King and Reeburgh, 2002]. Our observations suggest that the relatively small pool of recently fixed carbon is a primary source of carbon for the CH4 cycle in arctic tundra, whereas the large carbon stores in tundra soil (21.8 kg m−2, [Billings, 1987]) appear to contribute minimally to CH4 emission over seasonal timescales. Our results imply that recently fixed carbon represents a significant fraction of annual net CH4 emission from these ecosystems and that environmental changes affecting ecosystem productivity may heavily impact the global CH4 cycle.
 We thank Wes Clapp, Tamara Clark, Marty Downs, and Jesse Nippert for laboratory assistance. We thank the University of Alaska, Fairbanks and Toolik Field Station for field support. We greatly appreciate comments from three anonymous reviewers. This work was supported by the NSF Office of Polar Programs (OPP 96-15942 and OPP 93-18531).