In this study, we use a nested eddy covariance tower setup at Scotty Creek near Fort Simpson, Northwest Territories, Canada, flux footprint modelling, and spectral decomposition (i.e. singular spectrum analysis [SSA]) of FCH4 time series to assess the impact of permafrost-thaw induced boreal forest loss on growing season FCH4. The nested tower setup consists of a 2-m tower measuring FCH4 over a permafrost-free wetland and a 15-m landscape tower with FCH4 flux footprints originating from a heterogeneous boreal forest-wetland landscape. During the study period (May October 2014 & 2015), wetlands contributed 51 % to the landscape-tower flux footprints while drier, raised forested permafrost peatlands contributed 46 %. The remaining contributing 3 % were attributed to a shallow, permafrost-free lake.
We find that with 13.0 g CH4 m-2 the total growing season wetland FCH4 in 2014 and 2015 (May October) were about twice as large as the emissions over the heterogeneous boreal forest-wetland landscape (6.3 g CH4 m-2). Maximum monthly FCH4 were observed in August with 1.8 g CH4 m-2 (2014) and 2.3 g CH4 m-2 (2015) at the landscape tower and with 3.4 g CH4 mm-2 (2014) at the wetland tower.
Using SSA, we find that the seasonal FCH4 signal (i.e. 10-day window) accounted for 42 % and for 74 % of the total variance in FCH4 at the landscape and wetland tower, respectively. With 21 %, the sub-weekly FCH4 signal (i.e. 6h to 7-day window) accounted for a larger part of the total variance in FCH4 at the landscape tower than at the wetland tower (i.e. 9 %), likely a result of a more pronounced footprint heterogeneity at the landscape tower. The seasonal FCH4 signal closely followed an exponential function of the soil temperature in the wetland explaining 80 % and 75 % of the seasonal FCH4 signal at the landscape and wetland tower, respectively. Adding water level as explanatory variable to the model only increased the coefficient of determination slightly by 6 % and 1 % for the landscape and the wetland tower, respectively. Wetland contributions to landscape tower fluxes showed a positive relationship with the sub-weekly, landscape FCH4 signal and explained on average 23 % of its variance, supporting the observed larger FCH4 at the wetland tower compared to the landscape tower.
Our results indicate that permafrost-free wetlands represent a significant growing season source of FCH4 to the atmosphere, while CH4 of the forested permafrost peatlands seems to be negligible. A recent study has shown that wetlands at Scotty Creek have increased in coverage by about 0.25 % per year since 1977. Combining these change rates and our eddy covariance observations, we conclude that the landscape-scale growing season source may have grown accordingly by ~0.03 g per m2 per year over the past 40 years as a response to continued forest loss. Using a simple model of atmospheric CH4 and CO2 pools, we show that a concurrent continuous CO2 sink of ~250 g CO2 per m2 per year is required to neutralise the positive radiative forcing of these increasing CH4 emissions within a time frame of 50 years. Widespread permafrost thaw-induced forest loss in the southern Taiga Plains and other lowland boreal forest regions may therefore enhance the radiative forcing of terrestrial CH4 emissions and may be partly responsible for the observed positive trends in global atmospheric CH4 concentrations.