Quick summary: We used hierarchical generalized linear mixed regression models to distinguish effects of defoliation on pollen cone production from spatially synchronous background variation in cone production.
Patterns of jack pine (Pinus banksiana Lambert) pollen cone production are of interest because they may help explain jack pine budworm (Choristoneura pinus pinus Freeman) outbreak patterns. We used generalized linear mixed models to analyze pollen cone production in 180 permanent plots in Ontario, Canada between 1992 and 2008. Pollen cone production increased with stand age, and large trees in sparsely-populated stands produced more pollen cones. Defoliation decreased the propensity of trees to produce pollen cones for at least two years. We also identified important patterns that are not explained by defoliation and stand characteristics. Pollen cone production is spatially synchronized among years, trees in central Ontario produced more pollen cones than trees in northwestern Ontario, and background cone production increased over time in the central region but not in more northwestern plots. Synchronized reproduction is common among tree species, but has not previously been noted for jack pine pollen cones. Increasing cone production in central Ontario may be evidence of changing forest and (or) climatic conditions and deserves further investigation. Our model can be used to quantitatively predict pollen cone production and assess the risk of jack pine budworm defoliation.