Modeling the population dynamics of jack pine budworm Choristoneura pinus under accelerating climate change
Deforestation is a global problem that can cause desertification, climatic changes, and displacement or extinction of plant and animal species. Insect outbreaks play a major role in forest destruction. Forests act as carbon sinks, but insect defoliation can turn forests into carbon sources. Effective intervention requires an understanding of the environmental and biological factors regulating outbreaks. We are combining mathematical modeling and field experiments to identify the mechanisms driving the cyclical dynamics of the jack pine budworm (Choristoneura pinus). We collected data in outbreaking jack pine budworm populations at three sites in Wisconsin in 2012 and seven sites in Michigan in 2013 and 2014. We recorded budworm density and rates of parasitization, and measures of tree quality, temperature, precipitation, and forest composition. We used experiments to distinguish between the effects of host-tree quality and parasitoid attacks, by excluding parasitoids from the larval budworms. We covered jack pine branches with cloth bags for a number of weeks, and compared rates of parasitization and survival between protected and exposed branches.
Analysis of our observational data shows a positive correlation between jack pine budworm survival and density of pollen cones. We confirmed that rates of parasitization are density-dependent, and are likely an important driver of budworm outbreak collapse. Parasitism caused up to 95% mortality in collapsing budworm populations in Wisconsin, while outbreak-level Michigan populations suffered 30-75% mortality from parasitism. We used our experimental parasitoid exclusion data to fit nonlinear models of budworm survival, and used maximum likelihood to choose between models. The model including the effect of parasitoid exclusion is significantly better than the null model (ΔAIC=89). We then expanded this model to construct a preliminary population model that includes equations for budworm, jack pine, and parasitoid populations, and allows for jack pine age structure. The model produces complex long-period, large-amplitude fluctuations that appear to mimic long-term jack pine budworm population dynamics, but are strikingly different from the dynamics of host-parasitoid interactions alone. We are fitting competing models of budworm and parasitoid populations to data, and estimating the parameters using a Markov-Chain Monte Carlo routine. The models combine high-performance computing with field experiments in order to gain insight into the complex interactions between insect outbreaks, parasitoids, and host plant quality.