Predicting movement of nursery hosts using a linear network model

There is widespread concern among scientists and land managers that Phytophthora ramorum may be accidentally introduced into oak-dominated eastern U.S. forests through the transfer of the pathogen from infected nursery plants to susceptible understory forest species (for example, Rhododendron spp.) at the forest-urban interface. Inspection programs can be made more efficient by identifying locations throughout the U.S. that are most likely to receive infected nursery stock. e develop a spatial network model framework utilizing potential interstate nursery stock movements on a bipartite network, adopting a Bayesian approach to model probabilities of transmission of P. ramorum from entry to destination nodes within the network. As the goal of this paper is to present a general model framework, no specific risk analysis is presented. In the face of future discoveries of P. ramorum infections in the eastern U.S., instances of models from this framework can be used to help identify sites with a high risk of infection based on observed infection patterns.