The role of temporal variation in host stage abundance in the transmission of West Nile virus (with Kevin Caillouet)

West Nile virus (WNV) has remained an annual public health concern in the United States since its introduction in 1999, yet the ecological triggers leading to seasonal outbreaks are not well understood. While the annual occurrence of WNV in humans has been widely associated with the end of the avian nesting season, no specific mechanism has yet been demonstrated to describe if and how the end of nesting causes amplification of the virus. To address this question, we develop a stage-structured differential equation model for WNV incorporating vector preference for specific host life stages, and investigate the impact of host selectivity as well as vector and host abundance on the timing of enzootic and epizootic WNV activity.

How does the risk of infectious disease affect habitat selection? (with Ian Hamilton)

When animals have a choice of habitats in which to reside, the risks and benefits offered by each patch must be taken into account in order to arrive at an optimal decision. This choice is often assumed to be one that maximizes individual fitness. When balancing resource quality with competition for resources, Ideal Free Distribution (IFD) theory predicts input matching, an equilibrium in which the distribution of the population over the spatial landscape matches the distribution of resources. No individual can improve its fitness by changing patches. Field data and experiments commonly find an undermatching distribution, where the higher quality patch is underused compared to IFD predictions. Additional risks, such as predation or kleptoparasitism, have been incorporated into habitat selection theory and can change equilibrium spatial distributions. These risks are often greater in higher quality patches that support higher densities of prey. Patches with higher densities of individuals may also support higher levels of disease, presenting a greater threat of infection by pathogens or parasites. We are investigating how this risk of infectious disease affects ideal free habitat selection, and what role it may play in explaining observed undermatching.

Modeling Waterborne Disease: The Role of Multiple Transmission Pathways (with Joe Tien and Marisa Eisenberg)

Many waterborne diseases have transmission pathways incorporating multiple timescales. Cholera, caused by the pathogen Vibrio cholerae, can be transmitted directly via person-person contact, or indirectly through ingestion of contaminated water from an environmental pathogen reservoir. While infected individuals can shed pathogen into sources of water, V. cholerae is also naturally occuring in many parts of the world where cholera is endemic, found even when there are no cases of cholera in the local population.

Cholera dynamics are well described by the Susceptible-Infected-Water-Recovered (SIWR) model, a modified SIR model with an additional equation for the concentration of pathogen in the water that incorporates both transmission pathways as well as feedback from the infected class into the water. Our identifiability results and parameter estimates from a recent outbreak in Angola indicate both transmission pathways are important to the spread of cholera. However, many factors influencing each pathway (i.e. handwashing, sanitation in food preparation, availability of clean or treated water) vary within and across populations. We are investigating the effect of heterogeneity in both sources of transmission (individually, as well as their interaction) on the spread of cholera by extending the SIWR model to multiple patches connected through either movement of people, water, or both. We are investigating how the distribution of transmission among patches, as well as the network structure, affect the basic reproductive number (R0) of the system.

Spatial Patterns in Structured Populations with Density Dependent Dispersal (with Jim Cushing and Bob Costantino)

For my doctoral dissertation, I used integrodifference equations with density dependent dispersal kernels to investigate how interactions between life stages can lead to spatial patterns in stage-structured populations, applying these models to understand the separation of pupae and adults in the inhibiting flour beetle Tribolium brevicornis (see right) and the differences in depth distributions for larvae and adults in T. castaneum.

Under certain initial conditions, our model admits an attractor with T. brevicornis adults and pupae spatially segregated. This pattern is a result of life stage interactions alone, as our domain is spatially homogeneous, and does not occur in the absence of inhibition (delayed pupation of larvae in presence of high numbers of adults).