Previous year outbreak conditions and spring climate predict spruce budworm population changes in the following year

Abstract

Listen

Translate article

We determined effects of local spruce budworm (Choristoneura fumiferana Clem.; SBW) population level, proximity to sites with high SBW populations, insecticide spray, and environmental variables on SBW populations from 2014 to 2018, the outbreak initiation period in northern New Brunswick, Canada. SBW second instar larvae (L2) per branch data collected at 1100–2000 sample points per year were used to create annual interpolated population rasters. Fishnet sample points extracted from these rasters were overlaid with georeferenced layers of 46 possible predictor variables including forest composition, climate, topography, site quality, and insecticide treatment. Results showed that local SBW population in the previous year, proximity to sites with high SBW populations, and early spring climate were consistently the most important predictors over the 5 study years. Simultaneous autoregressive models were used to address spatial autocorrelation when forecasting the SBW L2 population, and a linear mixed effects model was fit to aggregate data for 2015–2018. The models reduced spatial dependence in the residuals, and explained 68–79% of variance in annual L2 levels and 53% of variance over the 4 years combined. Sensitivity analysis showed that locations with 5–10 more SBW L2 per branch than observed values, or 20–40 km closer to high population sites in the previous year could have up to 24 more L2 in the current year. Cumulative degree days in April helped to estimate the upper and lower bounds of the population. Expansion and retraction of SBW outbreak initiation were mathematically described. Understanding which variables influence SBW outbreak initiation and population level assists in design of small area target-specific insecticide spray applications and helps focus SBW L2 sampling on predicted outbreak hot spots.