Spatiotemporal occupancy patterns of chronic wasting disease

Chronic wasting disease (CWD) among cervids in Kansas has seen a consistent rise over the years, both in terms of the number of infections and its geographical spread. In this study, we assessed the occupancy patterns of CWD among white-tailed deer and mule deer across the state.

Using surveillance data collected since 2005, we applied a dynamic patch occupancy model within a Bayesian framework, incorporating various environmental covariates. Using principal components analysis, 13 fully orthogonal components representing cervid habitat, soil, and elevation were derived. Competing models with different temporal patterns were fit, and the best model selected based on Watanabe-AIC values and AUC value of 0.89.

The occupancy pattern produced by this model revealed a steady progression of the disease toward the east and southeast of the state. A random forest analysis of covariates at annual intervals indicated that geographic location, elevation, areas occupied by mixed forests, and several soil attributes (pH, clay content, depth to restrictive layer, available water content, and bulk density) explained most of the variability in the surveillance data (R ² = 0.96).

The findings reported in this study are the first for the state of Kansas but are consistent with previous findings from other geographic jurisdictions in the US and Canada. This consistency underscores their value in designing surveillance and management programs.