Background/Question/Methods

Vesicular Stomatitis (VS) is a vector-borne livestock disease that emerges in the United States (US) approximately every 10 years. After emerging, this disease spreads across the western US in a generally northward direction from spring to winter with the possibility of over-wintering and then continuing to spread the following year. To characterize the spatiotemporal dynamics of this disease in relation to the landscape heterogeneity over which it spreads, we used a landscape connectivity approach in five regions of the western US. The landscapes are represented by a diverse set of geospatial data, including flowing and standing surface water, soil properties, land use and land cover, and topography. VS incidence records from 2004-2006 and 2014-2015 were used in conjunction with landscape properties data to build a resistance surface, i.e. a geospatial data layer representing the lack of suitability for the disease, and then to predict the connected corridors of VS spread where the cumulative cost to travel between incidences is minimized. We also devised an adaptive artificial intelligence (AI) system to integrate the knowledge gained from the multiple regions and incidence years to improve predictions of future emergence patterns.   Results/Conclusions The models indicate that suitability of VS spread is primarily driven by proximity to surface water, but that the relative importance of landscape properties varies between (i) the time of year, and (ii) the landscape region. The connected corridors estimated from these resistance surfaces indicate the likelihood of VS between known incidences. Cross-validation from withheld testing sets supports the estimated corridors. The AI system successfully integrated the collection of predicted corridors into an evolving system of VS spread. By predicting the likely routes of disease spread, mitigation strategies in future emergence cases can be designed to mitigate spread in these corridors.