Using a Big Data Model Integration Approach to Forecast the Geographic Range of an Invasive Disease at the Continental Scale
Abstract
Forecasting the geographic range of an invasive disease is challenging given the complex relationships among hosts, vectors, and the environment. Vesicular Stomatitis (VS) is the most common vesicular livestock disease in North America. Endemic to Central America, VS spreads through arthropod vectors infected by an RNA virus (VSV) to northern latitudes in a series of outbreak and expansion events. While infections are rarely fatal, VSV's symptoms are difficult to distinguish from foot-and-mouth disease resulting in mandatory reporting and quarantine periods which elevate the economic cost of an infection. Recent regional and landscape scale studies have identified a series of complex environmental relationships associated with VSV infection patterns. However, efforts have not yet focused on exploring broader environmental relationships to estimate the spatial maxima of infections. Since the geographic range of VS can be modified as land-use evolves and climate variability increases, coarse-scale investigations can offer insight into the spatial arrangement of VS under novel global change scenarios.
Our objective was to model the epidemic range of VS in the western US under current and potential future climate scenarios using a big data model integration (BDMI) approach that included human and machine learning supervised by a transdisciplinary team. We harmonized publicly available data characterizing the expected biologically meaningful environment. We constructed and evaluated a suite of predictive models of VS occurrence in the US using a database of known occurrences from the USDA. Alternative climate scenarios were simulated by modifying long-term means of environmental variables. Our results suggest infection locations are characterized by above average: livestock densities, summer and fall vegetation greenness, soil water storage capacity, and below average winter precipitation. Climate scenarios that modify stream flow, in particular, are expected to impact the range of VS the most in the future. These results can be used to define priority areas for research, monitoring, and mitigation efforts. Further, this analysis provides an illustration of the robustness and flexibility of a BDMI approach in exploring complex problems at multiple scales.- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFM.B41N2921B
- Keywords:
-
- 0414 Biogeochemical cycles;
- processes;
- and modeling;
- BIOGEOSCIENCESDE: 0439 Ecosystems;
- structure and dynamics;
- BIOGEOSCIENCESDE: 0466 Modeling;
- BIOGEOSCIENCESDE: 1910 Data assimilation;
- integration and fusion;
- INFORMATICS