Rare earth elements minimal harvest year variation facilitates robust geographical origin discrimination: The case of PDO "Fava Santorinis"

This study examines the trace and rare earth elemental (REE) fingerprint variations of PDO (Protected Designation of Origin) "Fava Santorinis" over three consecutive harvesting years (2011-2013). Classification of samples in harvesting years was studied by performing discriminant analysis (DA), k nearest neighbours (j-NN), partial least squares (PLS) analysis and probabilistic neural networks (PNN) using rare earth elements and trace metals determined using ICP-MS. DA performed better than j-NN, producing 100% discrimination using trace elements and 79% using REEs. PLS was found to be superior to PNN, achieving 99% and 90% classification for trace and REEs, respectively, while PNN achieved 96% and 71% classification for trace and REEs, respectively. The information obtained using REEs did not enhance classification, indicating that REEs vary minimally per harvesting year, providing robust geographical origin discrimination. The results show that seasonal patterns can occur in the elemental composition of "Fava Santorinis", probably reflecting seasonality of climate. (C) 2016 Elsevier Ltd. All rights reserved.