AbstractEpidemiological surveillance for important wildlife diseases often relies on samples obtained from hunter-harvested animals. A problem, however, is that although convenient and cost-effective, hunter-harvest samples are not representative of the population due to heterogeneities in disease distribution and biased sampling. We developed an agent-based modeling framework that i) simulates a deer population in a user-generated landscape, and ii) uses a snapshot of the in silico deer population to simulate disease prevalence and distribution, harvest effort and sampling as per user-specified parameters. This framework can incorporate real-world heterogeneities in disease distribution, hunter harvest and harvest-based sampling, and therefore can be useful in informing wildlife disease surveillance strategies, specifically to determine population-specific sample sizes necessary for prompt detection of disease. Application of this framework is illustrated using the example of chronic wasting disease (CWD) surveillance in Missouri’s white-tailed deer (Odocoileus virginianus) population. We show how confidence in detecting CWD is grossly overestimated under the unrealistic, but standard, assumptions that sampling effort and disease are randomly and independently distributed. We then provide adjusted sample size recommendations based on more realistic assumptions. These models can be readily adapted to other regions as well as other wildlife disease systems.
An agent-based framework for improving wildlife disease surveillance: A case study of chronic wasting disease in Missouri white-tailed deer
