JIANG Haochen, WANG Shicheng, LI Bo, HAO Linlin, SUN Ci, CUI Jiehua, LI Guochen, XU Zhixiang
To carry out geographical origin identification of featured agricultural products, 10 kinds of green turnip with geographical indication certification were selected from Tianjin, Shandong and Henan provinces. Elemental Analyzer-Stable isotope mass spectrometry (EA-IRMS) was used to determinate the ratio of carbon and nitrogen isotopes in green turnip. Simultaneously, inductively coupled plasma mass spectrometry (ICP-MS) were used to determinate the contents of 18 mineral elements, including Co, Cd, As, Ni, Mo, Cr, B, Ba, Ca, Cu, Fe, Mg, Mn, Na, P, Sr, Zn and Se. The results of the box plot showed that stable isotope ratio has no significant effect on the identification of origin. Moreover, the one-way test of variance showed that there were significant obvious differences in the content of Na, Ca, Mg, Fe, Sr, Zn, B, Mn., as in green turnip among different provinces. The model established based on Partial least squares discriminant analysis (PLS-DA) detection results showed that different degrees of misjudgment in the green turnip samples from Henan, Tianjin and Shandong respectively. The overall discriminant rate of PLS-DA model was 86.44% when it was trained. However, when linear discriminant analysis (LDA) algorithm was applied to modeling, the misjudgment rate of green turnip samples decreased, and the accuracy of initial verification and cross-training reached 92.09% and 93.79%, respectively. Compared with PLS-DA model, the model established by LDA algorithm had higher discrimination accuracy and better effect on green turnip samples. This indicates that the application of LDA algorithm combined with mineral elements is feasible to identify the origin of green turnip, and it also provides a reference for the application of stable isotope ratio and mineral element content in the identification of origin of turnip in the future.