For the greenhouse gas trajectories, we assumed a constant background CO2
concentration (380 ppm) and we used ocean impulse functions from the pre-industrial
period from Joos and Siegenthaler (24). These two adjustments were made (as compared
to the approach taken in section 1.4) so that we could examine radiative forcing from a
contemporary change in the fire regime in Fig. 3C. In a series of model runs, we varied
the final probability of fire between 0.0055 and 0.03 at increments of 0.001. The resulting
radiative forcing values are shown in Fig. 3C, assuming the stand age distribution starts at
steady state with a mean fire return time of 80 years. We arbitrarily chose a mean of 80
years here for consistency with the stand age of our control - across the boreal biome fire
return times vary widely (44). We could have added a different set of offsets so that the
sum of different forcing agents was zero with a mean fire return time at steady state at
100 or 120 years. These adjustments would not change the slope of the lines in Fig. 3C or
our conclusions. We also generated a series of stand age distributions assuming that the
probability of fire did not depend on stand age (exponential distributions). These latter
stand age distributions did not substantially change the slopes of the lines shown in Fig.
3C.