Abstract
Models for determining country of origin of one grain of black Tanbaguro soybean using inorganic element composition and linear discriminant analysis (LDA) were proposed and assessed for applicability.
Test samples consisting of 66 Tanbaguro varieties were collected from Japan and China. Approximately 100 grains of each type was ground to powder. Test samples of approximately 0.5g were digested with nitric acid using microwave digestor and 24 elements in the sample were determined by ICP-AES and ICP-MS. An LDA model for six of 24 elements (Ba, Ca, Mn, Nd, W, Ni) selected by backward stepwise regression could be used to correctly identify the country of production (Japan or China) of all 66 samples.
Ratios of inorganic element concentrations to potassium concentrations in samples were introduced to simplify pre-treatment. An LDA model using concentration ratios of the six elements (Ba, Ca, Mn, Nd, W, Ni) to potassium could be used to identify two countries.
The ratios of the six elements : K in one grain from each of the 97 samples (the 65 samples used to construct the LDA model and the 32 samples collected to validate the model) were analyzed in the same way and 97 samples were correctly classified to levels of approximately 84%. In addition, 97 samples were classified approximately 94% correctly using the LDA model with the ratios of the concentrations of three elements (Cd, Cs, V) selected from 15 elements (determined by ICP-MS) to that of potassium.
