The effect of natural and anthropogenic disturbances on the uncertainty of large-area forest growth forecasts

Abstract

This study aimed to estimate the contribution of disturbances to the uncertainty of forest growth forecasts in the Bas-Saint-Laurent region in Quebec, Canada. We focused on two major disturbances affecting that region: spruce budworm (SBW) outbreaks and harvest activities. Growth forecasts were carried out for a period of 100 years (2003-2103) using ARTEMIS-2009, a stochastic individual-based model. Using the Monte Carlo technique, we simulated four scenarios: a baseline; a harvest scenario; a SBW scenario; and a scenario including both harvest and SBW. Uncertainty estimation was performed using a bootstrap variance estimator that applies to the context of hybrid inference. The results revealed that the total variances increased over time. For the scenarios including SBW, the variances were three to six times greater than those in the scenarios without outbreaks. Harvesting did not greatly contribute to the total variance. We conclude that to reduce the uncertainty of large-area growth forecasts in the Bas-Saint-Laurent, considering SBW dynamics is a crucial issue.

Close-to-nature silviculture in eastern Quebec: Advances over the last decade

“…The common ground among these studies is the use of a stochastic approach, which can be better applied to the reality of growth models available in Quebec. Additionally, the uncertainties presented in Melo et al (2019) show that relying on uncertain predictions to determine the longterm AAC calculations may not be as sustainable as would be expected. Potential strategies must promote discussion about short-term actions in forest management; deep uncertainty analysis recommends having flexible options and safety boundaries (Hallegatte 2009;Dittrich et al 2016).…”

“…Melo et al (2018) also decomposed the total model error, finding that the mortality submodel increased the variance of the predictions from 35 % to 60 %. Melo et al (2019), in the second step of their assessment of the Bas-Saint-Laurent region, included the effect of disturbance on the uncertainty of growth predictions. The authors estimated the contributions of SBW and harvesting and then compared these contributions to uncertainty with that arising from sampling error.…”

“…The scenarios accounting for SBW outbreaks increased total uncertainty by 80 % after the initial 30 years of the 100-year simulation horizon. Efforts to increase sampling size and develop efficient sampling designs decrease overall uncertainty, although according to Melo et al (2019), priority should be given to predicting the occurrence of natural disturbances. The authors propose a perspective of these uncertainties (Fig.…”

“…In (a), the solid line represents a growth scenario without disturbances, whereas the dashed line represents a scenario including only harvest. In (b), the solid line represents the scenario accounting for only SBW outbreaks, and the dashed line represents the scenario, including both SBW and harvesting (modified from Melo et al 2019).…”

An alternative simulation framework to evaluate the sustainability of annual harvest on large forest estates

“…The simulation framework can be summarized in three steps: (i) it starts with an FMU for which a sample of plots is available, (ii) the sample is projected using a stochastic ITBM that accounts for harvesting and (iii) the plot-level predictions and their uncertainty are scaled up to the FMU level using appropriate point and variance estimators. Melo et al (2019) relied on a similar framework to predict the future standing volumes and their uncertainty in the Bas-Saint-Laurent region, Quebec, Canada. The same approach has also been used in 9 D r a f t…”