PhD project. University of Toronto. Supervisor: Marie-Josée Fortin
Collaborators: Greg Dwyer (University of Chicago), Christina Cobbold (University of Glasgow), Kyle Haynes (University of Virginia)
2015. The American Naturalist 185(5): E30-152 http://www.jstor.org/stable/10.1086/680860.
Quick summary: We investigated the effects of landscape configuration and other factors on a spatial model of host-parasitoid (forest tent caterpillar) population dynamics. We also developed a new approximation technique to simplify model analysis and parameter estimation.
Understanding how cycles of forest-defoliating insects are affected by forest destruction is of major importance for forest management. Achieving such an understanding with data alone is difficult, however, because population cycles are typically driven by species interactions that are highly nonlinear. We therefore constructed a mathematical model to investigate the effects of forest destruction on defoliator cycles, focusing on defoliator cycles driven by parasitoids. Our model shows that forest destruction can increase defoliator density when parasitoids disperse much farther than defoliators because the benefits of reduced defoliator mortality due to increased parasitoid dispersal mortality exceed the costs of increased defoliator dispersal mortality. This novel result can explain observations of increased outbreak duration with increasing forest fragmentation in forest tent caterpillar populations. Our model also shows that larger habitat patches can mitigate habitat loss, with clear implications for forest management. To better understand our results, we developed an approximate model that shows that defoliator spatial dynamics can be predicted from the proportion of dispersing animals that land in suitable habitat. This approximate model is practically useful because its parameters can be estimated from widely available data. Our model thus suggests that forest destruction may exacerbate defoliator outbreaks but that management practices could mitigate such effects.