Resilience and sensitivity of ecosystem carbon stocks to fire-regime change in Alaskan tundra

Abstract

Fire disturbance has increased in some tundra ecosystems due to anthropogenic climate change, with important ramifications for terrestrial carbon cycling. Assessment of the potential impact of fire-regime change on tundra carbon stocks requires long-term perspectives because tundra fires have been rare historically. Here we integrated the process-based Dynamic Organic Soil version of the Terrestrial Ecosystem Model with paleo-fire records to evaluate the responses of tundra carbon stocks to changes in fire return interval (FRI). Paleorecords reveal that mean FRIs of tundra ecosystems in Alaska ranged from centennial to millennial timescales (200–6000 years) during the late Quaternary, but projected FRIs by 2100 decrease to a few hundred years to several decades (70–660 years). Our simulations indicate threshold effects of changing FRIs on tundra carbon stocks. Shortening FRI from 5000 to 1000 years results in minimal carbon release (<5%) from Alaskan tundra ecosystems. Rapid carbon stock loss occurs when FRI declines below 800 years trigger sustained mobilization of ancient carbon stocks from permafrost soils. However, substantial spatial heterogeneity in the resilience/sensitivity of tundra carbon stocks to FRI change exists, largely attributable to vegetation types. We identified the carbon stocks in shrub tundra as the most vulnerable to decreasing FRI because shrub tundra stores a large share of carbon in combustible biomass and organic soils. Moreover, our results suggest that ecosystems characterized by large carbon stocks and relatively long FRIs (e.g. Brooks Foothills) may transition towards hotspots of permafrost carbon emission as a response to crossing FRI thresholds in the coming decades. These findings combined imply that fire disturbance may play an increasingly important role in future carbon balance of tundra ecosystems, but the net outcome may be strongly modulated by vegetation composition.