Habitat Risk Software

Abstract

The Habitat Risk Software leverages a Bayesian Hierarchical model to analyze publicly available geospatial (raster) data and surveillance disease testing data to estimate the risk that specific locations within a wildlife habitat will harbor a disease positive individual. Th three part scripts of the Habitat Risk Software (1) prepare the surveillance (testing) and geospatial (raster) data for model inclusion, (2) parameterize and estimate coefficient value for 25 predetermined candidate model structures, (3) select the model structure with the lowest Deviance Information Criterion (DIC) given the data, (4) gather diagnostic plots to verify modeling assumptions have been met, and then (5) display the results of the best model in geographical and tabular context via an interactive web user interface (UI). UI capabilities include interactive and downloadable maps of estimate risk (and associated error), and overlay map of disease data with the spatial covariates, and detailed statistical information about the best model and the model selection process. All computation of the Habitat Risk Software appear in three R scripts, which may be run individually in sequence. The fourth script, the "command center" script was written to allow the user to initiate all three scripts with a single "Run All" command. This Habitat Risk Software is adaptable for analysis of many wildlife diseases an din many study areas. However, the example code in this eCommons packet was designed to estimate risk of Chronic Wasting Disease (CWD) in White tailed deer (Odocoilues virginianus) from several US states: Alabama, Arkansas, Connecticut, Florida, Georgia, Indiana, Iowa, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, New Hampshire, New York, North Carolina, Ohio, Pennsylvania, Rhode Island, Tennessee, Virginia, and Wisconsin. Two data input types are necessary to use the Habitat Risk Software: (1) pre-processed spatial (raster) data and (2) surveillance disease testing data generated by wildlife disease agencies. Spatial (raster) data originate for the US from publicly available data and may be processed for immediate inclusion in the Habitat Risk Software using the software in Mitchell et al. (2021; https://doi.org/10.7298/2tt1-yy48). Surveillance disease testing data is generated externally by the state wildlife agency, formatted according to the template, and fed into the Habitat Risk Software. Due to information sensitivity, disease testing data for each state has been redacted from this software, but a data template is included such that a future user can furnish their own data for use in this software. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.