Annual carbon balance of Canada's forests during 1895–1996

This paper reports annual carbon (C) balance of Canada's forests during 1895–1996 estimated using the Integrated Terrestrial Ecosystem C‐budget model (InTEC) [ Chen et al. this issue]. During 1895–1910, Canada's forests were small sources of 30±15 Tg C yr −1 due to large disturbances (forest fire, insect‐induced mortality, and harvest) in late nineteenth century. The forests became large sinks of 170±85 Tg C yr −1 during 1930–1970, owing to forest regrowth in previously disturbed areas and growth stimulation by nondisturbance factors such as climate, atmospheric CO 2 concentration, and N deposition. In recent decades (1980–1996), Canada's forests have been moderate sinks of 50±25 Tg C yr −1 , as a result of a tradeoff between the negative effects of increased disturbances and positive effects of nondisturbance factors. The nondisturbance factors, in order of importance, are (1) atmospheric N deposition (measured by a national monitoring network), (2) net N mineralization and fixation (estimated from temperature and precipitation records), (3) growing season length increase (estimated from spring air temperature records), and (4) CO 2 fertilization (estimated from CO 2 records using a leaf‐level photosynthesis model). The magnitudes of modeled nondisturbance effects are consistent with simulation results by the Carnegie‐Ames‐Stanford Approach (CASA) and are also in broad agreement with flux measurements above mature forest stands at several locations in Canada. Results for the disturbance effects agree with a previous study [ Kurz and Apps , 1996]. The overall C balance from InTEC generally agrees with that derived from tree ring data [ Auclair and Bedford , 1997] and from forest inventories. The combination of our result and that of Houghton et al. [1999] for the United States suggests that North America (> 15°N) was probably a C sink of 0.2‐0.5 Pg C yr± −1 during 1980s, much less than that of 1.7 Pg C yr± −1 estimated by Fan et al. [1998] using an atmospheric inversion method.