PARENT SESSION
Symposium 3: Emerging Diseases: Stressing the Union of Community Ecology and Epidemiology
Organized by: S Collinge and C Ray
Monday, August 4. 8:00 AM to 11:30 AM, SITCC Chatham Ballroom C.

Modeling plague in prairie communities: using sensitivity analysis to guide data collection and model construction.

Ray, Chris^{*,1, 2}, ¹ University of Colorado, Boulder, CO² University of Nevada, Reno, NV

ABSTRACT- Sylvatic plague, the source of human bubonic plague, has invaded wildlife communities across western North America. The plague bacterium, Yersinia pestis, is typically transmitted between mammals by fleas. The effective virulence of Y. pestis varies among host species, and the efficiency of Y. pestis transmission varies among vector species. No single host-vector association has been identified as the predominant reservoir for plague in North America. Instead, the plague reservoir appears to involve several host and vector species and frequent disruption of normal host-vector associations. In prairie communities, the most obvious impacts of plague include infrequent, isolated human infections and frequent, extensive prairie dog die-offs. These die-offs may affect the ecological and epidemiological function of prairie communities. Many species may benefit from the ecological services provided by prairie dog colonies, so the density of many plague hosts and the contact rates between hosts and vectors may be altered by prairie dog presence. Given these unknown and potentially complex relationships between hosts and vectors, no single model of plague dynamics can yet be proposed for prairie communities. But I have used a type of sensitivity analysis—familiar from population viability analysis—to analyze the behavior of many potential plague models. This analysis begins with a matrix model to project dynamics of infectious and non-infectious segments of several host and vector populations. Each matrix element represents a transition rate governed by a (linear or nonlinear) relationship between segments of the community (e.g., the rate at which 'mouse fleas' become infectious through contact with infectious prairie dogs). Model predictions are sensitive to changes in each matrix element, and this sensitivity can be used to suggest appropriate model structures and data collection efforts. Appropriate models should predict observed patterns, and research priorities should focus on the more sensitive elements of these models.

Key words: epidemiology, Cynomys, epizootic, matrix