Seasonality buffers carbon budget variability across heterogeneous landscapes in Alaskan Arctic tundra

Arctic tundra exhibits large landscape heterogeneity in microtopography, hydrology, and active layer depth. While many carbon flux measurements and experiments are done at or below the mesoscale (≤1 km), modern ecosystem carbon modeling is often done at scales of 0.25°–1.0° latitude, creating a mismatch between processes, process input data, and verification data. Here we arrange the naturally complex terrain into mesoscale landscape types of varying microtopography and moisture status to evaluate how landscape types differ in terms of CO2 and CH4 balances and their combined warming potential, expressed as CO2 equivalents (CO2-eq). Using a continuous 4 year dataset of CO2 and CH4 fluxes obtained from three eddy covariance (EC) towers, we investigate the integrated dynamics of landscape type, vegetation community, moisture regime, and season on net CO2 and CH4 fluxes. EC towers were situated across a moisture gradient including a moist upland tundra, a heterogeneous polygon tundra, and an inundated drained lake basin. We show that seasonal shifts in carbon emissions buffer annual carbon budget differences caused by site variability. Of note, high growing season gross primary productivity leads to higher fall zero-curtain CO2 emissions, reducing both variability in annual budgets and carbon sink strength of more productive sites. Alternatively, fall zero-curtain CH4 emissions are equal across landscape types, indicating site variation has little effect on CH4 emissions during the fall despite large differences during the growing season. We find that the polygon site has the largest mean warming potential (107 ± 8.63 g C–CO2-eq m−2 yr−1) followed by the drained lake basin site (82.12 ± 9.85 g C–CO2-eq m−2 yr−1) and the upland site (77.19 ± 21.8 g C–CO2-eq m−2 yr−1), albeit differences were not significant. The highest temperature sensitivities are also at the polygon site with mixed results between CO2 and CH4 at the other sites. Results show a similar mean annual net warming effect across dominant landscape types but that these landscape types vary significantly in the amounts and timing of CO2 and CH4 fluxes.