Model Analysis of Spatial Patterns in Mountain Pine Beetle Outbreaks

The mountain pine beetle [MPB, Dendroctonus ponderosae Hopkins (Coleoptera: Scolytidae)] is an aggressive bark beetle, one that typically needs to kill host trees in order to successfully reproduce. This ecological adaptation has resulted in an organism that is both economically important and ecologically significant. Even though significant resources have been expended on MPB research, and a great deal of knowledge exists regarding individual aspects of MPB ecology, some of the most basic questions regarding outbreaks remain unanswered. In our opinion, one reason for the lack of synthesis and predictive power is the inadequate treatment of spatial dynamics in outbreak theories. This paper explicitly addresses the role of spatial dynamics in the precipitation and propagation of MPB outbreaks. We first describe a spatially dynamic model of the MPB/forest interaction that includes chemical ecology, spatial redistribution of beetles, attack, and resulting host mortality. The model is a system of 6 coupled, partial differential equations with 7 state variables and 20 parameters. It represents an attempt to capture the relatively complex predator/prey interaction between MPB and host trees by including the minimum phenomenological descriptions necessary for ecological credibility. This system of equations describes the temporal dynamics of: beetle attraction as a function of pheromone concentration; the change in numbers of flying and nesting beetles; tree resistance/susceptibility; and tree recovery from attack. Spatial dynamics are modeled by fluxes due to gradients in pheromones and kairomones, and the random redistribution of beetles in absence of semiochemicals. We then use the parameterized model to explore three issues central to the ecology of MPB/forest interaction. The first of these is in response to the need for objective ways to compare patterns of successful beetle attacks as they evolve in space. Simulation results indicate that at endemic levels, the pattern of successful attacks are determined almost exclusively by the underlying distribution of susceptible host trees (environmental determinism). As an outbreak develops, the pattern of successfully attacked trees switches to one that is dynamically driven by the self-generated semiochemical landscape (dynamic determinism). This switch from an environmentally determined spatial pattern to a dynamically driven pattern is the hallmark of an outbreak. We discuss the application of a spatial correlation coefficient that can be used to differentiate between the spatial distribution of killed trees in endemic and outbreak phases. The second issue we address through simulation is synchrony in adult emergence. Synchronous adult emergence is critical for the mass attack strategy necessary for overcoming tree defenses. Results from these simulations indicate that the degree of synchrony in adult emergence can have important consequences for assessing the risk of an outbreak. The final issue we investigate through simulation is the effect of spatial pattern of nurse trees (those successfully attacked the previous year) on outbreak potential. Simulations indicated that the spatial proximity of nurse trees was an important determinant of subsequent successful attacks. We conclude with a discussion of the general implications of our simulation experiments. Copyright 1998 Academic Press.