Two-step Authentication of Argentinian Rice Cultivars by Near-infrared Spectroscopy Combined with One-class and Discriminant Chemometric Methods
Academic Article in Scopus
Authenticity verification of rice cultivars in Argentina is essential for ensuring genetic quality, international competitiveness, and market confidence, especially since the country exports cultivars recognized for their yield and industrial quality. Among the cultivars, IRGA-424 is the most widely cultivated due to its high yield potential and strong disease resistance. This study introduces a fast, and nondestructive strategy to authenticate the IRGA-424 cultivar by combining near-infrared spectroscopy with a two-step chemometric approach (one-class and discriminant methods). A total of 196 rice samples from four cultivars, long-grain rice (IRGA-424, IC-109, and SCS-121) and medium-grain rice (Fortuna), were analyzed. In the first authentication step, one-class models were developed using one-class partial least squares (OC-PLS) and data-driven soft independent modeling of class analogy (DD-SIMCA). Initially, all long-grain cultivars were defined as the target class and medium-grain cultivars as the non-target class. The DD-SIMCA model achieved the best performance, with sensitivity and specificity of 98.2% and 100%, respectively, and accuracy of 98.5%. In a subsequent step, the model was refined by designating IRGA-424 as the target class, enabling its differentiation from the remaining long-grain cultivars. The DD-SIMCA model yielded sensitivity, specificity, and accuracy values of 91.3%, 95.1%, and 94.4%, respectively. In the second step, a support vector machine (SVM) was employed to develop and validate a multiclass model that discriminated IRGA-424 from other long-grain cultivars, achieving 100% classification accuracy. In conclusion, NIR spectroscopy integrated with a two-step chemometric approach provides a powerful tool for authenticating rice cultivars produced in Argentina.
