Detection of Xylella fastidiosa in almond orchards by synergic use of an epidemic spread model and remotely sensed plant traits

• DOI: 10.1016/j.rse.2021.112420
• Publication Date: 2021-04-28T00:24:09Z
• Authors: C. Camino; R. Calderón; S. Parnell; H. Dierkes; Y. Chemin; M. Román-Écija; M. Montes-Borrego; B.B. Landa; J.A. Navas-Cortes; P.J. Zarco-Tejada; P.S.A. Beck
• Keywords: Xylella fastidiosa; VNIR

The early detection of Xylella fastidiosa (Xf) infections is critical to the management this dangerous plan pathogen across world. Recent studies with remote sensing (RS) sensors at different scales have shown that Xf-infected olive trees distinct spectral features in visible and infrared regions (VNIR). However, further work needed integrate plant disease epidemics. Here, we research how changes picked up by sets RS traits (i.e., pigments, structural or leaf protein content), can help capture spatial dynamics Xf spread. We coupled a spread model probability Xf-infection predicted RS-driven support vector machine (RS-SVM) model. Furthermore, analyzed which contribute most output prediction models. For that, almond orchards affected (n = 1426 trees), conducted field campaign simultaneously an airborne collect high-resolution thermal images hyperspectral visible-near-infrared (VNIR, 400–850 nm) short-wave (SWIR, 950–1700 nm). best performing RS-SVM (OA 75%; kappa 0.50) included as predictors content, nitrogen indices (NIs), fluorescence indicator (Tc), alongside pigments parameters. Leaf content together NIs contributed 28% explanatory power model, followed chlorophyll (22%), parameters (LAI LIDFa), indicators photosynthetic efficiency. Coupling epidemic increased accuracy 80%; 0.48). In where presence was assayed qPCR 318 combined RS-spread yielded OA 71% 0.33, higher than RS-only visual inspections (both 64–65% 0.26–31). Our demonstrates combining epidemiological models lead highly accurate predictions distribution.