Predicting the Lyme disease vector range expansion: A modeling approach for New York State

Emerging diseases have become a critical public health issue during the last century. Often, they are consequence of range and density increases in the populations of arthropods vectoring diseases. Accurate prediction of range and density through and after these increases provides valuable information for prevention and management of disease risk. Multiple modeling methods are available; some predict the equilibrium distribution and density but are not useful during the non-equilibrium process of expansion. In order to be applicable to the non-equilibrium phase, the models need to use spatially- and temporally-aware datasets that explicitly incorporate time and space as factors. When those non-equilibrium models are used in correspondingly spatial and temporal contexts they efficiently predict the density at particular time intervals. Nevertheless, such models may deviate sharply when applied to novel geographies or when projected to time intervals well beyond their calibration time. Thus, identifying a priori the specific conditions where a non-equilibrium predictive model needs to be replaced is extremely difficult. We use field density estimates of Ixodes scapularis, the tick vector of Lyme disease, in the New York State to analyze the performance of non-equilibrium and equilibrium modeling approaches. We used an extensive sampling scheme that temporally overlaps with the rapid expansion to the North of ticks in the area. We evaluate non-equilibrium and equilibrium models according to their predictability and report the effect of geography and time over predictability for each approach. Finally, we discuss the implications of these findings over the modeling and study of the expansion of ticks across the northern limit of their distribution.