Fire Blight Research Papers

860 Words4 Pages

Fire blight is caused by the bacterium Erwinia amylovora Burrill (Winslow et al., 1920), which was recognized as the first plant bacterial pathogen by fulfilling Koch’s postulates in 1884. The earliest symptom of the disease occurs on flowers which appear water-soaked and discolored before they turn necrotic, and they may discharge bacterial ooze when the environmental conditions are conducive. Shoot blight is the most obvious symptoms of fire blight in which Leaves show blackening along the midrib and veins before becoming fully necrotic, and the tip of the infected shoots die and bend over into a shepherd’s crook shape (Sherman, 2000; Steiner, 2000). E. amylovora has been reported on the European and Mediterranean Plant Protection Organization …show more content…

On one hand to analyze comprehensively the whole spectral response to retain the whole information; and on the other one to identify synthetic indices such as optimum narrow bands and new Spectral Vegetation Indices (SVIs) which are able to characterize the status of the crop and the different levels of stress (Thenkabail, 2001; Thenkabail et al., 2004; Jain et al., 2007). Thus, many approaches have been proposed for discriminating disease presence including the use of multivariate statistical analysis techniques. The proposed procedures may allow both to eliminate the redundant information and to identify synthetic indices which maximize differences among levels of stress (Broge and Leblanc, 2001; Ray et al., 2010). Classification models are divided either based on the form of the decision boundaries among classes into linear or non-linear models, or based on the multivariate probability distribution of the data into parametric or non-parametric models. The parametric methods, like Nearest Mean Classifier (NMC), are commonly used when the studied dataset represents a sample from a multivariate normally-distributed population, whereas the Non-parametric methods, such as Partial Least Squares-Discriminant Analysis (PLS-DA), Artificial Neural Networks (ANNs) and Classification and Regression Tree (CART), are used when the multivariate distribution is different from the normal (Hand,

Open Document