Methane (CH4) emissions from Arctic tundra are an important feedback to global climate. Currently, modelling and predicting CH4 fluxes at broader scales are limited by the challenge of upscaling plot-scale measurements in spatially heterogeneous landscapes, and by uncertainties regarding key controls of CH4 emissions. In this study, CH4 and CO2 fluxes were measured together with a range of environmental variables and detailed vegetation analysis at four sites spanning 300 km latitude from Barrow to Ivotuk (Alaska). We used multiple regression modelling to identify drivers of CH4 flux, and to examine relationships between gross primary productivity (GPP), dissolved organic carbon (DOC) and CH4 fluxes. We found that a highly simplified vegetation classification consisting of just three vegetation types (wet sedge, tussock sedge and other) explained 54% of the variation in CH4 fluxes across the entire transect, performing almost as well as a more complex model including water table, sedge height and soil moisture (explaining 58% of the variation in CH4 fluxes). Substantial CH4 emissions were recorded from tussock sedges in locations even when the water table was lower than 40 cm below the surface, demonstrating the importance of plant-mediated transport. We also found no relationship between instantaneous GPP and CH4 fluxes, suggesting that models should be cautious in assuming a direct relationship between primary production and CH4 emissions. Our findings demonstrate the importance of vegetation as an integrator of processes controlling CH4 emissions in Arctic ecosystems, and provide a simplified framework for upscaling plot scale CH4 flux measurements from Arctic ecosystems.

Methane (CH4) emissions from the northern high-latitude region represent potentially significant biogeochemical feedbacks to the climate system. We compiled a database of growing-season CH4 emissions from terrestrial ecosystems located across permafrost zones, including 303 sites described in 65 studies. Data on environmental and physical variables, including permafrost conditions, were used to assess controls on CH4 emissions. Water table position, soil temperature, and vegetation composition strongly influenced emissions and had interacting effects. Sites with a dense sedge cover had higher emissions than other sites at comparable water table positions, and this was an effect that was more pronounced at low soil temperatures. Sensitivity analysis suggested that CH4 emissions from ecosystems where the water table on average is at or above the soil surface (wet tundra, fen underlain by permafrost, and littoral ecosystems) are more sensitive to variability in soil temperature than drier ecosystems (palsa dry tundra, bog, and fen), whereas the latter ecosystems conversely are relatively more sensitive to changes of the water table position. Sites with near-surface permafrost had lower CH4 fluxes than sites without permafrost at comparable water table positions, a difference that was explained by lower soil temperatures. Neither the active layer depth nor the organic soil layer depth was related to CH4 emissions. Permafrost thaw in lowland regions is often associated with increased soil moisture, higher soil temperatures, and increased sedge cover. In our database, lowland thermokarst sites generally had higher emissions than adjacent sites with intact permafrost, but emissions from thermokarst sites were not statistically higher than emissions from permafrost-free sites with comparable environmental conditions. Overall, these results suggest that future changes to terrestrial high-latitude CH4 emissions will be more proximately related to changes in moisture, soil temperature, and vegetation composition than to increased availability of organic matter following permafrost thaw.