Leaf hyperspectral reflectance as a potential tool to detect diseases associated with vineyard decline

Grape production in the Serra Gaucha region, south of Brazil, is severily constrained by several diseases such as the decline and death syndrome caused grapevine trunk (fungal) diseases (GTDs) and the grapevine leafroll-associated virus(GLRaV). As pathogens induce changes in leaf tissue that modify the reflectance, the spectral signature of asymptomatic and symptomatic grapevine leaves infected by GTDs and GLRaV was analyzed to check whether spectral responses could be useful for disease identification. This work aims at (a) defining the spectral signature of grapevine leaves asymptomatic and symptomatic to GTDs and GLRaV; b) analyzing whether the spectral response of asymptomatic leaves can be distinguished from symptomatic; and (c) defining the most useful wavelengths for discriminating spectral responses. For such, reflectance of leaves in either condition collected in a Merlot vineyard during three growing seasons was measured using a spectroradiometer. Principal components and partial least square discriminant analyses confirmed the spectral separation and classes discrimination. The average spectra, difference spectra, and first-order derivative (FOD) spectra indicated differences between asymptomatic and symptomatic leaves in the green peak (520-550 nm), chlorophyll-associated wavelengths (650-670 nm), red edge (700-720 nm), beginning of near-infrared (800-900 nm), and shortwave infrared. Hyperspectral data was linked to biochemical and physiological changes described for GTD and GLRaV. Variable importance in the projection (VIP) analysis showed that some wavelengths allowed to differentiate the tested pathosystems and could serve as a basis for further validation and disease classification studies.