Detection of White Root Rot in Avocado Trees by Remote Sensing.

Area of Science:

• Plant Pathology
• Agricultural Remote Sensing
• Machine Learning in Agriculture

Background:

• White root rot, caused by *Rosellinia necatrix*, significantly impacts woody crop yields, including avocados.
• Current detection methods (microbial, molecular) are difficult to implement at the orchard scale.
• Need for efficient, large-scale disease detection methods in agriculture.

Purpose of the Study:

• To evaluate the potential of physiological parameters derived from imaging techniques for detecting white root rot.
• To apply machine learning algorithms to analyze imaging data for disease prediction in avocado trees.

Main Methods:

• Analysis of physiological parameters: Normalized Difference Vegetation Index (NDVI) and normalized canopy temperature.
• Application of machine learning algorithms, including logistic regression analysis (LRA).
• Training and testing algorithms using imaging data from avocado orchards.

Main Results:

• Logistic Regression Analysis (LRA) trained on NDVI data demonstrated the highest sensitivity.
• The NDVI-based LRA achieved the lowest rate of false negatives among tested algorithms.
• Normalized canopy temperature was also tested but showed lower predictive performance compared to NDVI.

Conclusions:

• NDVI, analyzed by LRA, presents a promising, rapid, and feasible approach for detecting potential white root rot infections in avocado orchards.
• This imaging-based machine learning method can overcome limitations of traditional detection techniques at the orchard scale.
• Further validation could lead to practical tools for disease management in avocado production.