Resumen de Spatial models for the analysis of plant disease epidemics and their control
Martina Cendoya Martínez
The exponential growth of global trade and international travel, to with climate change, has had a significant impact on the introduction and spread of pests and pathogens affecting important crops around the world. This situation represents a growing threat to global food security and socioeconomic stability. Therefore, understanding the factors involved in the entry, establishment and spread of pathogens and disease dynamics, is crucial to effectively prevent and manage new outbreaks and minimize their impact. In the last decade, the plant pathogenic bacterium Xylella fastidiosa has caused a phytosanitary crisis in the European Union, affecting, among others, two of the most important Mediterranean crops, olive and almond trees. The general objective of this PhD Thesis is to provide different spatial modeling tools applicable to the study of plant disease epidemics, to analyze their spatial distribution and related factors, as well as to serve as a reference for their control. In particular, we address several issues on X. fastidiosa through advanced statistical methods that are novel in the field of plant pathology, with special focus to the analysis of the spatial distribution. In this context, the following main objectives are proposed: i) the analysis of climatic and spatial effects on the distribution of X. fastidiosa; ii) the effects of the incorporation of dispersal barriers on the distribution of the pathogen; iii) the implementation of a spatially explicit individual-based spread model; and iv) the evaluation of the effectiveness and efficiency of different outbreak management plans, including surveillance and control strategies. Spatial modeling is approached from the perspective of Bayesian hierarchical spatial models using the integrated nested Laplace approximation (INLA) for inference and prediction. This approach allows to deal with different types of spatial data and highlights the importance of considering spatial structure in models. As an extension of these models, a non-stationary model is used for incorporation of dispersal barriers, illustrating the strong influence of these structures on the probability of pathogen presence. Moreover, a simulation model of disease spread at individual level is proposed, where a correlation function establishes their spatial relationship. This model provides a detailed perspective on disease spread and establishes the basis for the implementation and evaluation of surveillance and control strategies, with early surveillance being of particular importance. The models proposed and applied to X. fastidiosa are flexible and can be extended to other plant diseases.
