Identification of the causal agent of Guar leaf blight and development of a semi-automated method to quantify disease severity

Abstract

Guar (Cyamopsis tetragonoloba) is an annual crop from which guar gum, a valuable biopolymer in industry, is extracted. The crop is affected by Alternaria spp. causing leaf spots. Accurate identification of the causal agent and semi-automated quantification are important in improving disease management. The objective of this study was to identify the causal agent of leaf spot in Guar, as well as to design an indirect tool using images to quantify severity and identify symptomatic plants. Guar plants showing leaf spot symptoms were collected in fields in Guasave, Sinaloa, Mexico, and fungal isolates were obtained from symptomatic leaves. A representative isolate was characterized by morphology, as well as phylogenetic analysis using partial sequences of three genes (tef1-α, gapdh, and rpb2). Subsequently, using photographs of healthy and diseased leaves with different levels of severity, a six-class scale was designed to represent severity using traditional, semiautomated, and automated image analysis methods such as ImageJ, segmentation using the pliman library of R, and fitting of a convolutional neural network model to detect diseased plants, quantify and classify the areas affected by the disease. The fungus Alternaria alternata was associated with the disease and was characterized. Image analysis methods allowed for the semi-automation of severity quantification by reducing the time and cost involved in the evaluation and with greater accuracy and precision with respect to visual methods.