Ninth International Conference on Permafrost ... - IARC Research

Methane Ebullition During Field-Simulated Lake Expansion andPermafrost DegradationOlivier MazéasUniversity of California, Berkeley, Dept. of Geography 507 McCone Hall #4740, Berkeley, CA 94720-4740, USAJoseph von FischerColorado State University, Dept. of Biology, Fort Collins, CO 80523, USARobert RhewUniversity of California, Berkeley, Dept. of Geography 507 McCone Hall #4740, Berkeley, CA 94720-4740, USAIntroductionThe Arctic accounts for 30% of the global emissions ofmethane (CH 4), a potent greenhouse gas, to the atmosphere(Christensen 1993). Within the Arctic, tundra and lakes aremajor sources, although they exhibit exceptionally highspatial variability. In arctic lakes, CH 4ebullition is the majortransport mechanism from the sediments to the atmosphere(95%). Ebullition rates are the greatest near the edges of thelakes, where active erosion occurs (Walter et al. 2006). Inregions of continuous permafrost, arctic lakes have beenexpanding in recent decades, owing to permafrost meltingand development of thermokarst (Smith et al. 2005). Lakeexpansion occurs when margins erode into water, supplyinglarge amounts of organic rich material to the sedimentwaterinterface. This allows carbon that was previouslystored in the soil (permafrost and active layer) to becomebioavailable and subject to decomposition. An increase inarctic CH 4emissions as a result of permafrost thawing andlake expansion would constitute a positive feedback to arcticwarming.In order to better understand processes associated withlake CH 4emissions, we conducted an experiment in a thawlake on the Arctic Coastal Plain during the summer and fallof 2007. Different layers of tundra soil were incubated inchambers at the bottom of the lake, and methane ebullitionwas monitored.Material and MethodsEleven incubations were initiated in mid-July 2007 at adepth of 1 m in Cake Eater Lake, on the Barrow EnvironmentalObservatory (BEO), Alaska. Each experimental chamberconsisted of a bucket (yielding an exposed surface area of0.07 m 2 ) fixed beneath an inverted funnel equipped with asampling port to capture and collect the emitted gases.The nearby tundra soil was vertically stratified in 3 distinctlayers, which we extracted separately. First, the unfrozen(upper) section of soil, hereafter called the active layer, wassampled down to the depth of frozen soil, and included liveplants and decaying peat material (3 adjacent points, ~20 cmdeep and 13 kg each, n = 3). Next, the seasonally frozen layersoil (~12 cm thick) was sampled with a mechanical auger,homogenized, and divided into separate buckets (about 8 kgeach, n = 3). Finally, a layer of permafrost was sampled (~12cm deep), homogenized, and placed into buckets (about 8 kgeach, n = 3). Although these soils were initially frozen, thesamples thawed before the initiation of the experiment.In addition to these 9 incubations, 2 others were added forcomparative purposes: a control incubation using an emptybucket and another one containing sawdust mixed in withthawed permafrost. This sawdust provided cellulose, whichis a major component of plant tissue, and its fermentationwas expected to yield substrates for methanogenesis.Ebullition gas volume determination and sampling wereperformed at variable time points along an 11-week period,ending on the days of initial lake freeze-up at the beginningof October.From each gas sample, a 0.5 ml subsample was analyzedusing a laser-based analyzer (Los Gatos Research) formethane and carbon dioxide determination. Nitrogen (N 2)was used as the carrier gas, and a calibration curve was runfor each sequence (Matheson Tri-Gas grade CH 4).The initial carbon content was determined using a Carbon-Nitrogen analyzer (NC2100, Carlo Erba). Water and sedimenttemperatures were recorded using in situ dataloggers, andwind speed and atmospheric pressure were also monitoredthroughout the experiment.Results and DiscussionOnly the active layer consistently emitted gases viaebullition throughout the experiment, with an average rateof 15 ml day -1 . The seasonally frozen and permafrost layerssporadically emitted small volumes of gas (typically 0 to 1ml per day), with cumulative volumes ~20 times smaller thanthe active layer. The ebullition frequency was highly variable,and significant ebullition events could not be correlated withany of the environmental parameters monitored.Daily ebullition events were consistently observedfrom the active layer incubations from the first day of theexperiment, with large ebullition rates (> mean) occurringwithin 7–15 days and some of the highest rates startingwithin 3–4 weeks. The replicates show a similar overalltrend; however, one replicate could not be monitored afterthe fifth week of incubation due to technical problems.Ebullition significantly decreased or ceased during the last7 to 10 days prior to lake freeze-up, a period when the watertemperature dropped to between 1 and 0°C. In contrast, notemporal pattern of ebullition rates was observed for thefrozen soil layers.The gas composition of the collected bubbles also differed205