Multivariate Statistical Analysis-Based Identification of Fake Seeds for Rapid Argo Forensic Application

Abstract

Utilising Near-Infrared Reflectance Spectroscopy (NIRS) with chemometric tools like Principal Component Analysis (PCA) and Partial Least Squares Discriminant Analysis (PLS-DA), to differentiate viable and non-viable castor seeds (GCH 7 hybrid). Traditional seed viability tests are accurate but time-consuming and resource-intensive. NIRS offers a rapid, non-destructive alternative for assessing seed viability. Spectral data from 200 viable and 200 non-viable seeds were collected and analysed using PCA and PLS-DA to develop a Linear Discriminant Analysis (LDA) model. The model achieved 99% accuracy in classifying seed viability, demonstrating its potential as a reliable tool for on-spot seed quality assessment. Key spectral markers related to fatty acids, proteins, and functional groups related to castor seed oil were identified. The research highlights the feasibility of integrating NIRS with advanced data analytics for rapid seed viability testing, offering significant benefits to seed testing agencies and farmers by reducing time and resource requirements while maintaining high accuracy.