Multi-block approach for the characterization and discrimination of Italian chickpeas landraces.

An untargeted characterization was carried out to attempt the geographical discrimination of three high-valued Italian chickpeas (Cicer arietinum L.), harvested in 2019 in three traditional and relatively close production areas: Navelli (Abruzzo, Central Italy), Cicerale (Campania, South Italy), and Valentano (Lazio, Central Italy). The present study aimed to develop and validate a potentially non-destructive and routine-compatible method for the geographical traceability of chickpea landraces of high traditional value. The outer part of 146 kernels belonging to the three varieties was analysed by Attenuated Total Reflectance-Fourier Transform-Mid Infrared (ATR-FT-MIR) and FT-Near Infrared (NIR) spectroscopies. Eventually, each sample was cut in two, and the cross-sections (the internal and external parts) were analysed by the two spectroscopic techniques. Spectral information was organized in four data blocks (MIRout, MIRin, NIRout, and NIRin), and single-block Partial Least Squares-Linear Discriminant Analysis (PLS-LDA) was applied. Accurate results were obtained from the single-block-processing of the spectroscopic profiles for the outer kernel part (MIRout and NIRout) that, combined with interesting outcomes from a preliminary class modelling approach suggest the real possibility of implementing a non-destructive authentication method. Notwithstanding, Sequential and Orthogonalized (SO)-PLS-LDA and SO-Covariance selection (Covsel)-LDA were applied to interpret better the information in the four collected data blocks. In this context, VIP (Variable Importance in Projection) analysis was performed to identify the significant variables, leading to a direct chemical interpretation of the classification models. • Geographical discrimination of traditional Italian chickpeas using spectroscopy. • Single-block PLS-LDA yields accurate results for outer kernel analysis. • Promising results for a non-destructive method in chickpea landraces traceability. • VIP analysis and SO-CovSel revealed significant variables for the classification. • A better chemical insight was achieved by multi-block classification models. [ABSTRACT FROM AUTHOR]