Exposure to biting ticks and insects on Department of Defense lands puts military personnel at risk of contracting arthropod-borne diseases, many of which can have debilitating complications. However, because most North American vector-borne diseases are maintained in wildlife reservoirs, disease transmission is often sensitive to environmental conditions, including seasonality and global change. In addition, there is a lack of models that account for the detailed mechanisms and differences among potential control methods. This makes it difficult to predict when and where human disease will appear, and how it should be managed when it does. Climate warming, in particular, can complicate predictive efforts, because it compromises the ability to extrapolate from current conditions or ‘rules-of-thumb’ that have been acquired over years of experience. One of the most important aspects in this regard is phenology change – i.e., the change in the timing of plant and animal lifecycles. Shifting phenologies are not only one of the earliest indicators of climate change. This can then affect disease transmission, fundamentally altering disease dynamics, and impacting everything from disease risk to management options.
To aid in prediction of human disease risk, and to develop on-the ground management strategies for reducing disease spread, the research team developed modeling frameworks that incorporated phenology and global change impacts on phenology as well as novel model formulations that account for the specifics of different insecticide management strategies. The frameworks are broad enough to accommodate a range of different vectors (mosquitos, ticks) and pathogens (virus, bacteria) and are general enough to apply to both well studied diseases, and diseases that have not been as well characterized. For this reason, researchers focused heavily on the mosquito side of control, where a handful of mosquitoes are responsible for the majority of disease spread, including both common and rare diseases. Using novel models, researchers additionally developed a graphical user interface (GUI) that allows practitioners to input time series data on mosquito abundances (e.g. from United States Centers for Disease Control light traps) and then outputs optimal timing of management strategies. The underlying optimization scheme that was developed for this GUI is broad and flexible, thus management decisions can be made based on the combination of historically observed phenology, cost constraints, and efficacy of the chosen management strategy. In addition, the GUI can be re-run at any point in the season, allowing land managers to update their predictions partway through the year, allowing for incorporation of current year conditions.
Using a range of different modeling approaches, the research team developed a number of improved models of vector-borne disease management. These models include pertinent mechanisms that differentiate the mode of action of different management strategies, thereby improving all models of vector-borne diseases. Researchers also clearly delineated the difference between approximate, implicit modeling of control, and accurate, mechanistically corrected explicit modeling of control, highlighting when and where the former can be used in place of the latter. Next, researchers developed models for two diseases – La Crosse Encephalitis and Zika – that have been detected in North America. For the La Crosse Encephalitis model, researchers began extending results to include phenology, which suggests that some of the observed increase in La Crosse Virus may be a result of warming conditions. In addition, these results provide a stark warning for the future. Finally, researchers examined some general models of vector-borne disease systems, including vector and host populations, in the context of changing phenology due to global change. These models suggest that the basic reproduction number will depend on the various perturbations to different vector and host populations, and that effects on dynamics may be complex and non-intuitive, including the appearance of year-to-year cycles and chaos in vector-borne disease outbreaks. Finally, researchers extended their phenologically explicit modeling to consider specific vectors involved in the most notorious North American vector-borne diseases, including La Crosse, West Nile Virus, and Eastern Equine Encephalitis. Researchers are currently using these models, interfaced to a genetic algorithm, to best predict the timing of management strategies, both under current conditions and in the future. Finally, the researchers capped their work off by developing a GUI that puts our complex mathematical formulations under the cap, allowing management practitioners to input their data on mosquito abundances, as well as their economic and social constrains in order to identify optimal timing of larvicide and adulticide treatments based on their requirements and desires.
Predictions from the modeling efforts identify some of the key management mechanisms influencing disease transmission, as well as the relevant ecological processes associated with disease spread. This information allows the researchers to identify the system properties that should be monitored to best predict disease, both now and in the future. Another key contribution from the research is the first ordinary differential equations epidemiology model for LAC. This is an emerging disease that is now one of the most commonly reported arbovirus infections. Finally, application of optimal control to the phenologically explicit models and, in particular, the highly parameterize vector models for Aedes, Culex and Culiseta enables us to predict management strategies that successfully balance the competing needs of vectorborne disease management on military installations. Ultimately the researchers have transitioned the modeling approaches to a GUI interface that can be used by practitioners based on their own data and systems.