Discrimination of Musa banana genomic and sub-genomic groups based on multi-elemental fingerprints and chemometrics.

Abstract

The potential of unripe banana flour multi-elemental fingerprints for classifying banana genomic and subgenomic groups was assessed using chemometrics. The elemental concentration of N, P, K, Mg, Ca, Zn, Cu, Mn, Fe, and B in unripe banana flour from 33 banana varieties belonging to four genome groups and 11 subgenome groups were determined using Flame-atomic Absorption spectrometry and colorimetry. Principal component analysis (PCA) combined with linear discriminant analysis (LDA), support vector machine (SVM), and artificial neural network (ANN) was applied for classification with an 80:20 split between the calibration and verification sets (157 and 39 samples, respectively). The elements K, N, and Mg presented the highest mean concentrations of 1273 mg/100 g, 424 mg/100 g, and 132 mg/100 g, respectively. The classification model verification set samples were successfully classified based on their genome groups (100 % accuracy) and subgenome groups (78.95–100% accuracy) for PCA-LDA, PCA-ANN, and PCA-SVM models. The results demonstrate that multi-elemental fingerprinting combined with chemometrics can be employed as an effective and feasible method for classification of Musa genomic and sub-genomic groups.