RECONSTRUCTION AND ANALYSIS OF HISTORICAL CHANGES IN CARBON STORAGE IN ARCTIC TUNDRA

Abstract

Surface air temperature in arctic regions has increased since pre-industrial times, raising concerns that warmer and possibly drier conditions have increased soil de- composition rates, thereby stimulating the release to the atmosphere of the large stores of carbon (C) in arctic soils. We used a model (MBL-GEM, Marine Biological Laboratory General Ecosystem Model) of ecosystem C and nitrogen (N) dynamics to predict and analyze historical (1829-1990) changes in C storage in a N-limited, tussock-tundra ecosystem near Toolik Lake on the North Slope of Alaska. The model simulates stand-level photosynthesis and N uptake by plants, allocation of C and N to foliage, stems, and fine roots, respiration in these tissues, turnover of biomass through litterfall, and decomposition of litter and soil organic matter. We first calibrated the model by deriving a single parameter set that closely simulated the response of tussock tundra to decade-long experimental manipulations of nutrients, temperature, light, and atmospheric CO2. The calibrated model predicts that historical increases in temperature and atmospheric CO2 have increased total ecosystem C storage. Higher temperatures increased soil and plant respiration, but those losses of C were overcompensated by increased photosynthesis re- sulting from redistribution of N from soil to plants. This redistribution of N was due to increased net mineralization and uptake of N. Increases in atmospheric CO2 also increased photosynthesis, but consequent increases in C storage were constrained by limits on in- creases in the C:N ratio of vegetation. In contrast, hypothesized historical decreases in soil moisture substantially decreased simulated total ecosystem C storage as a result of large increases in soil respiration. With decreased soil moisture, increases in photosynthesis associated with redistribution of N from soil to plants only partially compensated for respiratory C losses, as plant uptake of N could not keep pace with increased N mineral- ization rates. Consequent losses of N from the ecosystem contributed to the declines in C storage under drier conditions. Based on the combined effects of reconstructed historical changes in atmospheric C02, mean growing-season temperature, and two alternative soil moisture scenarios, the model predicts a -5.4 to +2.3% change in ecosystem C from 1829 to 1990. These estimates are consistent with field evidence that historically recent changes in C storage of tussock tundra have been relatively small. However, the model also predicts that relatively large transient losses of ecosystem C (50-180 g.m-2.yr-4) may frequently have occurred since the late 1800s, with one of the largest simulated C losses of the 20th century occurring from 1988 to 1990 (3-yr average = 133 g C.m-2.yr-'). These simulated losses were similar to losses measured in tussock tundra at Toolik Lake from 1983 to 1990 (e.g., 1990 losses = 156 g C.m-2.yr-') but nonetheless suggest that these measured losses may be transitory in nature.