A stochastic characterisation of the natural disturbance regime of the boreal mixedwood forest with implications for sustainable forest management

The annual area burned on an 8.6 × 10⁶ ha study area in the boreal mixedwood forest of northeastern Alberta, Canada, was characterised as a serially independent random draw from a lognormal distribution. This characterisation was applied in Monte Carlo simulations, which showed that estimates of the mean annual burn rates, even with long sample periods, are highly imprecise. Monte Carlo simulation was also used to simulate the development of a forest subject to lognormally distributed annual burn rate in an attempt to characterise the equilibrium age-class structure. No equilibrium age-class structure could be identified from the simulation results. The validity of equilibrium age-class distribution models (e.g., the negative exponential and Weibull) and analysis that relies on these models is questioned for forests where the annual burn rate is highly variable.