Hierarchical Bayesian Modelling of Xylella fastidiosa spread in southern Italy and mainland Spain
| Main Authors: | Martínez-Minaya, Joaquín, Cendoya, Martina, Vicent, Antonio, Saponari, Maria, López-Quílez, Antonio, Conesa, David |
|---|---|
| Format: | Proceeding |
| Bahasa: | eng |
| Terbitan: |
, 2019
|
| Online Access: |
https://zenodo.org/record/3558754 |
Daftar Isi:
- In the last years, numerous epidemiological studies have been carried out to understand disease spread in humans, animals and plants. Epidemiological modelling may help to prevent, eradicate or contain disease spread under different scenarios. Here, we focus on diseases caused by the bacterium Xylella fastidiosa which was recently detected in the Mediterranean Basin. The olive quick decline, caused by X. fastidiosa subsp. pauca, has devastated extensive areas in the Salento peninsula, south east Italy. An outbreak of almond leaf scorch, caused by X. fastidiosa subsp. multiplex, was detected in 2017 in Alicante province, eastern Spain. The introduction and spread of X. fastidiosa in other regions could cause potential negative consequences, including yield losses and costly control measures, not only in olive or almond but also in other economically important crops such as grapes, citrus, or stone fruits. In this work, we present an analysis of the prevalence of X. fastidiosa in southern Italy and mainland Spain using species distribution modelling (Martínez-Minaya et al., 2018). These models may assist to associate the prevalence of these diseases (georeferenced data gathered from official surveys) with different climate variables based on monthly average data from 1970 to 2000 obtained from the WorldClim database. In particular, we present different Hierarchical Bayesian spatio-temporal models in order to better understand the epidemiological factors driving disease spread. As obtaining the posterior distribution of the parameters governing the models is not straightforward, we used Integrated Nested Laplace Approximation methodology (INLA) (Martins et al., 2013). The spatial effect was implemented in the models using the Matérn covariance family, approximated as a weak solution to a Stochastic Partial Differential Equation (SPDE) (Lindgren et al., 2011). As previous modelling studies on the same topic (White et al., 2017), results will assist risk managers to limit further disease spread in the European territory and to minimize the impact in the areas where the pathogen is no longer eradicable.