Methane emissions from the Florida Everglades: Patterns of variability in a regional wetland ecosystem

Abstract

Natural wetlands are presumed to be major sources of atmospheric methane, but current estimates of the global wetland emission vary by almost a factor of 20. Estimates of global source strengths are based on extrapolation of in situ flux measurements to large areas occupied by broad classes of wetland environments, and recent efforts at refinement of these estimates have concentrated on improving inventories of the global distribution of major wetland types. An additional potential source of uncertainty which has not been quantified is regional scale variability in emission rates within the major wetland types. We conducted an experiment which examined the spatial variability of methane flux within a large regional wetland system, the Florida Everglades. We also investigated the association of flux variability with relevant surface characteristics such as soil thickness, water depth, soil temperature, and vegetative community distribution. Unit area methane flux to the atmosphere from water‐saturated Everglades environments, measured in situ, varied over more than an order of magnitude (4.2 to 81.9 mg CH4/m2/d), depending on which habitat component of the ecosystem was sampled. Observed physical characteristics of the surface (water and soil depth, soil temperature) were not quantitatively associated with the variability in flux rates. However, the distribution of vegetative community types provided an empirical indicator of flux, permitting an inventory of emissions to be based on mapping of regional vegetation patterns. Use of high‐resolution, orbital remote sensing data helped reduce uncertainty in the emission inventory of the Everglades by directing in situ sampling efforts to important habitat types and by providing a means for calculating area‐weighted mean flux for the system as a whole. The results indicated that spatial variability in flux within a major wetland ecosystem can introduce significant uncertainty in extrapolations to larger areas, even if the extent of the major ecosystem itself is well known. The results also suggested that the response of total ecosystem flux to changing water level is not a linear function of flooded area, but is damped, with regional flux at lowered water levels decreasing proportionally less than flooded area. Both sources of variability can be addressed by the combination of remote sensing and in situ techniques we have employed in the Everglades.