ECOLOGICAL CONTROLS ON METHANE EMISSIONS FROM A NORTHERN PEATLAND COMPLEX IN THE ZONE OF DISCONTINUOUS PERMAFROST, MANITOBA, CANADA

Methane emissions were measured by a static chamber technique in a div erse peatland complex in the Northern Study Area (NSA) of the Boreal E cosystem Atmosphere Study (BOREAS). Sampling areas represented a wide range of plant community and hydrochemical gradients (pH 3.9 - 7.0). E missions were generally larger than those reported from other boreal w etland environments at similar latitude. Seasonal average fluxes from treed peatlands (includiug palsas) ranged from 0 to 20 mg CH4 m(-2) d( -1) compared with 92 to 380 mg CH4 m(-2) d(-1) in open graminoid bogs and fens (with maximum single fluxes up to 1355 mg CH4 m(-2) d(-1)). P ermafrost-related collapse scars had similarly high CH4 emissions, par ticularly in the lagg areas where continuous measurements of water tab le, peat surface elevation, and peat temperature showed that the peat surface adjusted to a falling water table in the abnormally dry 1994 s eason, maintaining warm, saturated conditions and high CH4 flux later into the season than nonfloating sites. A predictive model for CH4 flu x and environmental variables was developed using multiple stepwise re gression. A combined variable of mean seasonal peat temperature at the average position of the water table explained most of the spatial var iability in log CH4 flux (r(2) = 0.64), with height above mean water t able (HMWT), water chemistry (K-corr, pH, Ca), tree cover, and herbace ous plant cover explaining additional variance (r(2) = 0.81). Canonica l correspondence analysis (CCA) of combined vascular and bryophyte dat a with environmental variables showed that CH4 flux was negatively cor related with HMWT, the second axis of vegetation variability, and was only weakly correlated with chemistry, the first axis. Sedge and tree cover were correlated with high and low CH4 fluxes, respectively, whil e shrub cover was of less predictive value. Microtopographic groupings of hummocks and hollows were separated in terms of CH4 flux at the in termediate ranges of the moisture gradient. These data show that multi variate vegetation analyses may provide a useful framework for integra ting the complex environmental controls on CH4 flux and extrapolating single point chamber measurements to the landscape scale using remote sensing.