Using spread models to optimise surveillance for Xylella fastidiosa
Although surveillance is of central importance to the prevention and control of pests and pathogens, the complex nature of natural biological systems means that it can be challenging to fully account for the ecological and epidemiological characteristics of the pathosystem in question when planning surveillance. Our previous work has demonstrated that failing to account for these issues can lead to suboptimal surveillance performance, both in terms of pathogen detection and cost effectiveness. Using the example of the emerging pathogen Xylella fastidiosa CoDiRO in the Apulia region of southern Italy, we demonstrate how to design a surveillance strategy which incorporates both biological and statistical considerations by linking a spatially explicit model of pathogen spread with a statistical model of a sampling process. Using this model, we apply a simple optimisation routine to identify where best to sample olive tree hosts in the currently “uninfected” region towards the north of Apulia, in order to detect any new incursions at an early stage. By investigating the impact of different spread patterns, we have found that accurately characterising the long distance spread of the pathogen (resulting largely from vector movement) will be key to improving the spatial targeting of early detection surveillance efforts and maximise the probability of early detection of spread beyond the infected region.