Modelling plant disease spread and containment: Simulation and approximate Bayesian Computation for Xylella fastidiosa in Puglia, Italy

Abstract

Mathematical and computational models play a crucial role in understanding the epidemiology of economically important plant disease outbreaks, and in evaluating the effectiveness of surveillance and disease management measures. A case in point is Xylella fastidiosa, one of the world’s most deadly plant pathogens. Since its European discovery in olives in Puglia, Italy in 2013, there remain key knowledge gaps that undermine landscape-scale containment efforts of the outbreak, most notably concerning the year of introduction, the rate of spread, dispersal mechanisms and control efficacy. To address this, we developed a spatially explicit simulation model for the outbreak spreading among olive groves coupled to a simulation of the real surveillance and containment measures. We used Approximate Bayesian Computation to fit the model to surveillance and remote-sensing infection data, comparing the fits for three alternative dispersal mechanisms (isotropic, wind and road). The model accurately explained the rate and spatiotemporal pattern of the outbreak and found weak support for the wind dispersal model over the isotropic model. It suggests that the bacterium may have been introduced as early as 2003 (95% CI [2000, 2009]), earlier than previous estimates and congruent with anecdotal evidence. The isotropic model estimates the pathogen is spreading at 5.7 km y-1 (95% CI [5.4-5.9]) under containment measures, down from 7.2 km y-1 (95% CI [6.9-7.5]) without containment measures. Our estimate of an approximately 10-year lag between introduction and detection highlights the need for stronger biosecurity and surveillance for earlier detection of emerging plant pathogens. The outputs from simulations without any disease management also suggest that while containment measures have caused some slowing of X. fastidiosa spread, stronger measures will be required to contain the outbreak fully.