Advanced Forecasting Modeling to Early Predict Powdery Mildew First Appearance in Different Vines Cultivars

Abstract

Eurasian grapevine is a widely cultivated horticultural plant worldwide, but it is more susceptible to powdery mildew. In recent years, the high cost and negative environmental impact of calendar-applied sulfur fungicides are leading research to find alternative remedies. In this study, the early prediction (three days) of the first appearance of powdery mildew infection, on two different Italian grapevine cultivars, was detected through a partial least squares discriminant analysis (PLSDA). The treatment indications of the “PLSDA” models (treatments according to the predictive model) were compared with those of the “Standard” (treatments according to the established agricultural practice of the area). This allowed the early containment of the disease, preventing its subsequent propagation. The model was built based on weather-climate data and phytopathological information collected on the “Untreated” control cultivar to monitor the natural spread of the disease (three years of training and two of tests). For both the cultivars and the two test years (2021 and 2022), the “PLSDA” models early predicted the first appearance of fungal disease, reducing the treatment number (about four) with respect to “Standard”. In addition, analyses of key fruit quality parameters were conducted to evaluate the effectiveness of treatment reduction.