Non-Destructive NIR Spectrometric Cultivar Discrimination of Castor Seeds Resulting from Breeding Programs

Abstract

This article proposes a near-infrared (NIR) diffuse reflectance spectrometric method for non-destructive discrimination of castor seeds from the two cultivars most commonly employed in Brazilian plantations (BRS Nordestina and BRS Paraguacu). For this purpose, two classification techniques are compared, namely SIMCA (soft independent modelling of class analogies) and PLS-DA (partial least squares discriminant analysis). By applying the SIMCA classifier to a test set comprising 150 seeds, the BRS Nordestina and BRS Paraguacu class models yielded sensitivity/specificity values of 0.91/0.99 and 0.71/1.00, respectively. Better results were obtained by using PLS-DA, which correctly classified all test samples, i.e., yielded sensitivity and specificity values of 1.00. These findings suggest that the proposed method is a promising approach to identify castor seed genotypes, either in seed lots or for breeding purposes, prior to being planted.