Search

Balsam fir sawfly (Neodiprion abietis (Harris)) has become a serious pest of young managed balsam fir (Abies balsamea (L.) Mill.) stands in western Newfoundland, Canada. During 1991–2008, a total area of 561 000 hectares was moderate to severely defoliated. We quantified impacts (growth and survival) using data from permanent sample plots (PSPs) and dendrochronology and related these impacts to defoliation severity determined from aerial defoliation data, in order to provide input into a Decision Support System. We analyzed 67 Newfoundland Forest Service PSPs, selected based on severity of defoliation (classes 1–6), stand age and management interventions (pre-commercially thinned vs natural) and measured before and after defoliation (1996–2008). We used Bayesian statistics to combine information from different sources, each having their own limitations and associated uncertainty. No mortality was observed in immature plots 12 years after defoliation, but survival was 54 per cent lower in mature defoliated than in non-defoliated plots. Plots in defoliation class 1 (1 year of moderate, 30–70 per cent, defoliation) showed 22 per cent cumulative growth reduction and complete recovery to pre-defoliation growth increment after 5 years. Plots in defoliation classes 2–6 (one to three consecutive years of severe, 71–100 per cent, defoliation) had mean cumulative growth reductions of 26–40 per cent and did not recover to pre-defoliation levels even 9 years after defoliation ceased. Natural and thinned plots responded similarly to defoliation severity. These results suggest that proactive control measures need to be implemented since impacts are severe, even with only 1 year of severe defoliation. [ABSTRACT FROM AUTHOR]

Predictions of defoliation are an important component of planning aerial insect control programs, especially for defoliators such as balsam fir sawfly (Neodiprion abietis (Harris)) that cause severe impacts on forest growth and yield. Currently, defoliation prediction is done manually based on field observations and experience, but it is a good candidate for a Bayesian network (BN), a flexible tool for combining available expert knowledge and empirical data. We created a BN model and linked it to a geographic information system to map predicted defoliation for balsam fir sawfly in western Newfoundland over an area of 5.7 million ha from 2001 to 2008. Based on expert knowledge, probabilistic influence of egg counts, previous defoliation, and stand characteristics (species composition, stand age, and management intervention) on subsequent-year defoliation was quantified. For validation purposes, maps created using the BN model were compared with manual defoliation predictions and with measured aerial defoliation survey maps. BN model defoliation prediction maps were found to be in moderate agreement (mean Kappa value of 0.59) with conventional manual prediction maps. Overall, the BN model showed similar accuracy to manual predictions, but with benefits of automating the process and of providing more spatial detail in predictions. [ABSTRACT FROM AUTHOR]