Résumé
Highlights: • Use of 13 SSR markers allowed 66 allele detection and easy scoring. • 163 leaf samples collected throughout the Pacific corresponded to 36 distinct genotypes. • Multivariate analysis of SSR data differentiate five groups. • FTIR spectra analysis of 138 root samples differentiate Nobles from non-Nobles kavas. • FTIR spectra PLS-DA model prediction accuracy allows differentiation over 96.9%.Abstract: According to the FAO/WHO Codex Alimentarius regional standards for kava (Piper methysticum Forst. f.), only Noble cultivars shall be used as raw material. Other P. methysticum cultivars and wild forms (P. m. var. wichmannii C.DC.) are excluded from trade. There is a need for accurate identification tools as morpho-agronomic descriptors are influenced by growing conditions and once kava roots have been harvested, it is visually impossible to differentiate Noble cultivars from others. In the present study, 163 leaf samples originating from Papua New Guinea, the Solomon Islands, Vanuatu, Fiji, Tahiti, Hawaii and Pohnpei were analyzed using 13 microsatellite (Single Sequence Repeat, SSR) markers. Hierarchical Clustering Analysis (HCA) and Multiple Correspondence Analysis (MCA) clearly differentiated five groups: Nobles from Polynesia, Nobles from Vanuatu, Two-day cultivars from Vanuatu, var. wichmannii from Vanuatu and P. methysticum cultivars for Papua New Guinea. Fourier Transform InfraRed (FTIR) was used to assess its potential as a high throughput analytical tool to detect nobility. FTIR spectra (1800-800 cm−1) from 138 root powder samples were analyzed using Principal Component Analysis (PCA). Nobles from Vanuatu and from Polynesia were not differentiated from each other but were differentiated from poor quality P. methysticum cultivars and from var. wichmannii. Linear Discriminant Analysis (LDA) improved the differentiation of Nobles from others. Samples were then classified in two groups: Nobles (Vanuatu + Polynesia) and non-Nobles. A learning and a test sets were composed at random. PLS-DA (Partial Least Square Discriminant Analysis) allowed the differentiation of Nobles with a prediction accuracy over 96.9%. FTIR has potential for kava quality control. The method described here is simple, fast, eco-friendly, and requires less than 0.5 g of root powder. Its routine application could greatly contribute to guarantee kava quality and protect the international markets.