Repositori URV

Identificador: imarina:9225571

Handle: https://hdl.handle.net/20.500.11797/imarina9225571

Autors:
Foschi, MartinaBiancolillo, AlessandraVellozzi, SimonaMarini, FedericoD'Archivio, Angelo AntonioBoque, Ricard

Resum:
Emmer is a traditional Italian wheat species attracting growing attention for the high-nutritive and dietary value. The growth of emmer consumption and the recent spreading even in areas where production was not traditional pose a risk to biodiversity and to the geographical identities. Thus, the present work aims to develop a nondestructive and routine-compatible method to discriminate three Italian landraces and lay the basis for a possible authentication method. One-hundred and forty-seven emmer samples, harvested in 2019 in three traditional production areas (Garfagnana, Monteleone di Spoleto, Gran Sasso and Monti della Laga National Park), were investigated by Mid-Infrared (MIR) and Near-Infrared (NIR) spectroscopy. Two different approaches of multiclass Partial Least Squares-Discriminant Analysis (PLS-DA) were applied on the collected fingerprinting profiles. Eventually, Data-Fusion strategies have been employed to combine the different information sources and classify the samples according to the geographical origin. The most accurate predictions were provided by the Sequential and Orthogonalized-Partial Least Squares-Discriminant Analysis (SO-PLS-DA) model, which misclassified only one test sample over 44 (in external validation). Finally, a chemical interpretation of the most discriminant variables was performed.