Insect outbreaks play a major role in forest destruction, in some cases turning forests from carbon sinks to carbon sources. Effective intervention requires an understanding of the factors regulating outbreaks, which in turn requires mechanistic models. Classical models of insect defoliators include only density-dependent attacks by natural insect enemies, but variation in host plant quality can also strongly affect defoliator survival. Incorporating plant quality into the models, however, requires experimental data that can distinguish between mortality due to natural predators and mortality due to host plant quality, as well as nonlinear fitting routines to estimate model parameters.
We combined model fitting, field experiments, and observational data to identify the mechanisms driving the complex dynamics of the forest defoliator jack pine budworm (Choristoneura pinus). We collected data in outbreaking budworm populations from 2012-2015, recording budworm density and rates of parasitism, as well as measures of tree quality, temperature, precipitation, and forest composition. We also experimentally excluded parasitoids from budworm larvae, to distinguish between the effects of host-tree quality and parasitoid attacks, and to measure rates of ongoing parasitism.
Results/Conclusions
We first analyzed our data using generalized linear models, confirming that rates of parasitism are density-dependent. Parasitism increased each year of the outbreak, and is likely an important driver of budworm outbreak collapse. Age of the host tree has a parabolic relationship with insect survival, such that budworms are least likely to survive on young or overmature trees. We then fit mechanistic models to our observational and experimental data, using stochastic non-linear differential equations. Our models made different assumptions about the effects of parasitoids and host tree quality. We chose between models using AIC scores. The best model allows for direct density-dependence, density-dependent parasitoid attacks, and plant quality (tree age) (from the next best model, Δ AIC = 69.9; from the null model, Δ AIC = 259). Plant quality does not, however, affect parasitoid attack rates directly. Extending this larval model to allow for multiple generations of jack pine budworm shows that plant quality strongly affects outbreak dynamics in the long-term. Our work shows evidence that host tree quality strongly affects jack pine budworm outbreaks, suggesting that host plant quality plays a general role in insect population cycles.