HU Guixian, SHAO Shengzhi, ZHANG Yongzhi, ZHU Jiahong, ZHAO Shouping, YUAN Yuwei
To explore the regional features and the traceability of geographical origin of Myrica rubra, the ratios of stable isotopes (e.g. δ15N, δ13C, δD and δ18O, etc.) and the content of multi-elements (e.g. Li, Be, Na, K and Fe, etc.) in Myrica rubra were determined using elemental analyzer-isotope ratio mass spectrometry (EA-IRMS) and inductively coupled plasma mass spectrometry (ICP-MS), respectively. Then, two pattern recognition methods, i.e. principal component analysis (PCA) and linear discriminant analysis (LDA) were applied to distinguish the geographical origins of Myrica rubra from five different provinces (including Zhejiang, Fujian, Yunnan, Guizhou and Jiangsu). One-way ANOVA showed that the stable isotope ratios and the contents of multi-elements in Myrica rubra show geographical characteristics at some content; however, the difference is not significant, the Myrica rubra from different provinces cannot be discriminated using single index. In PCA, the cumulative variance contribution rate of the first three principal components only comprised 42.77%, and the scores plots of all samples from different provinces appeared partial overlaps, which indicated that PCA cannot effectively discriminate the origin of the Myrica rubra in this context. Therefore, a novel PCA-LDA method was proposed, in which all samples were divided into three categories: Zhejiang, Fujian, and other provinces (including Yunnan, Guizhou and Jiangsu), and each category was randomly divided into training set and testing set via Monte-Carlo method for the modeling and accuracy validation. After 2 000 runs, the discriminant accuracy of the Myrica rubra using established PCA-LDA model was up to 99.6% for Zhejiang, and 90.3% for Fujian and 98.4% for other provinces, respectively. Therefore, the strategy combining EA-IRMS and ICP-MS detection with PCA-LDA method can be used in the traceability and discrimination of the geographical origin of Myrica rubra from different provinces.