CLIMATIC EFFECTS ON TUNDRA CARBON STORAGE INFERRED FROM EXPERIMENTAL-DATA AND A MODEL

We used a process-based model of ecosystem carbon (C) and nitrogen (N) dynamics, MEL-GEM (Marine Biological Laboratory General Ecosystem Mod el), to integrate and analyze the results of several experiments that examined the response of arctic tussock tundra to manipulations of CO2 , temperature, light, and soil nutrients. The experiments manipulated these variables over 3- to 9-yr periods and were intended to simulate anticipated changes in the arctic environment. Our objective was to us e the model to extend the analysis of the experimental data so that un measured changes in ecosystem C storage and the underlying mechanisms controlling those changes could be estimated and compared. Using an in verse calibration method, we derived a single parameter set for the mo del that closely simulated the measured responses of tussock tundra to all of the experimental treatments. This parameterization allowed us to infer confidence limits for ecosystem components and processes that were not directly measured in the experiments. Thus, we used the mode l to estimate changes in ecosystem C storage by inferring key soil pro cesses within the constraints imposed by measured components of the ec osystem C budget. Because tussock tundra is strongly N limited, we hyp othesized that changes in ecosystem C storage in response to the exper imental treatments would be constrained by several key aspects of C-N interactions: (1) changes in the amount of N in the ecosystem, (2) cha nges in the C:N ratios of vegetation and soil, and (3) redistribution of N between soil (with a low C:N ratio) and vegetation (with a high C :N ratio). The model results reveal widely differing patterns of chang e in C-N interactions and constraints on change in ecosystem C storage among treatments. For example, after 9 yr the elevated CO2 (2 x ambie nt) treatment and the N fertilized (10 g N.m(-2) yr(-1)) treatment inc reased ecosystem C stocks by 1.4 and 2.9%, respectively. Whereas the i ncrease in the CO2 treatment was due solely to an increase in the C:N ratios of vegetation and soil, the increase in the fertilized treatmen t was due to increased ecosystem N content and a shift of N from soil to vegetation. In contrast, the greenhouse (3.5 degrees C above ambien t) and shade (one-half ambient light) treatments decreased ecosystem C stocks by 1.9 and 2.7%, respectively. The primary reason for the net C losses in these treatments was an increase in respiration relative t o photosynthesis, with a consequent decrease in the ecosystem C:N rati o, However, when we simulated the elevated temperatures in the greenho use treatment without the confounding effects of decreased light inten sity (an artifact of the greenhouse structures), there was a long-term increase in ecosystem C stocks because of increased photosynthetic re sponse to the temperature-induced shift of N from soil to vegetation. If our simulated changes in ecosystem C storage are extrapolated for t he approximate to 43 Pg C contained in arctic tundras globally, the ma ximum net gain or loss (approximate to 0.3% per yr) from tundra would be equivalent to 0.13 Pg C/yr. Although fluxes of this magnitude would have a relatively minor impact on current changes in atmospheric CO2, the long-term impact on tundra C stores could be significant. The syn thesis and insights provided by the model should make it possible to e xtrapolate into the future with a better understanding of the processe s governing long-term changes in tundra C storage.