'The Transferability of Plant Pest Models' by Thomas Taylor

Plant diseases have increased dramatically in recent times with significant environmental and economic impacts. Surveillance for early detection of new epidemics is crucial for successful control and eradication. Parnell et al (2015) introduced a ‘rule of thumb’ to determine how much and how often surveillance should be performed to achieve early detection. However, the accuracy of the rule of thumb over large spatial scales with heterogenous and aggregated host population is not fully known, but is the key to its practical application. We run thousands of simulations on similar but not identical plant distributions, generating a disease outbreak. We experiment with the variables that affect the epidemic such as the dispersal scale of the pathogen, the aggregation of the landscape, and the transmission coefficient that contributes to the recently famous basic reproduction number. Using a computer simulated monitoring program, we calculate how far an epidemic has spread before being detected and compare this to our prediction prior to observation. Is there is a significant gap between our rule of thumb prediction and the simulated outcome? Can we predict how accurate our rule of thumb will be in different scenarios? For example, how will our confidence in the rule of thumb change for a forest disease (Ash dieback) as compared to a crop disease (Tomato blight)? With the significance of novel invasive plant pests increasing globally, tools such as the rule of thumb, are sorely needed to allocate appropriate surveillance resources.