We re-visited a seven-stand boreal chronosequence west of Thompson, Manitoba, Canada, in which coarse woody debris (CWD) and its instantaneous decomposition were measured in 2000. New CWD measurements were performed in 2007, and tree inventories updated to provide mortality and snag failure data. These data were used to model CWD changes, compare methods of estimating decomposition, and infer possible fragmentation rates. Measured CWD was between 9.7 (in both the 77- and 43-year-old stands) and 80.4 (in the 18-year-old stand) Mg ha(-1) in 2007. Spatial variability was high; at most stands CWD levels had not changed significantly from 2000 to 2007. Tree mortality was a significant flux only in older stands, whereas snag fall rate varied by an order of magnitude, from 2.9% y(-1) (0.2 Mg ha(-1) y(-1)) in the 9-year-old stand to 9.8% y(-1) (2.3 Mg ha(-1) y(-1)) in the 12-year-old stand. A one-pool model based on these inputs underestimated actual 2000-2007 CWD decomposition in the younger stands, suggesting that fragmentation could be an important part of the carbon flux exiting the CWD pool. We compared three independent measures of annual decomposition (k): direct measurements of CWD respiration, rates based on the 7-year re-sampling effort described here, and rates inferred from the chronosequence design itself. Mean k values arrived at via these techniques were 0.06 +/- 0.03, 0.05 +/- 0.04, and 0.05 +/- 0.05 y(-1), respectively. The four-pool model suggested that the transition rate between decay classes was 0.14-0.19 y(-1); the model was most sensitive to initial CWD values. Although the computed k values implied a problem with chronosequence site selection for at least one site, the overall CWD trend was consistent with a larger number of sites surveyed in the region.
