Geographical origin classification of peanuts and processed fractions using stable isotopes
Syed Abdul Wadood, Jing Nie, Chunlin Li, Karyne M. Rogers, Yongzhi Zhang, Yuwei Yuan
Abstract
This study investigates the use of stable isotopes (C, N, H, and O) to characterize the geographical origin of peanuts along with different peanut fractions including whole peanut kernel, peanut shell, delipidized peanuts and peanut oil. Peanut samples were procured in 2017 from three distinctive growing regions (Shandong, Jilin, and Jiangsu) in China. Peanut processing significantly influenced the δ13C, δ2H, and δ18O values of different peanut fractions, whereas δ15N values were consistent across all fractions and unaffected by peanut processing. Geographical differences of peanut kernels and associated peanut fractions showed a maximum variance for δ15N and δ18O values which indicated their strong potential to discriminate origin. Different geographical classification models (SVM, LDA, and k-NN) were tested for peanut kernels and associated peanut fractions. LDA achieved the highest classification percentage, both on the training and validation sets. Delipidized peanuts had the best classification rate compared to the other fractions.