Abstract: The objectives of this research were to 1) determine the mineral element contents and enrichment characteristics of Boletus edulis and 2) establish a model using partial least squares-discrimination analysis (PLS-DA) combined with spectral data fusion for identifying different geographical origins of B. eduils. The mineral elements of B. edulis and soil samples from six producing areas in Yunnan province were determined and the near-infrared and ultraviolet spectral information was collected. The accumulation characteristics and enrichment capacity of mineral elements by B. edulis were analyzed based on mineral element contents and enrichment coefficient. Savitzky-Golay (SG), second derivatives (2D), standard normal variables (SNV) and multiple-scattering correction (MSC) were used to pre-process spectral data. A monospectral data model was established using PLS-DA. The results showed that B. edulis was rich in K and P, and all tested minerals varied with the geographical origin with the highest Na level being 20.70 times as great as the lowest level. The content of Fe in soils from the six producing areas was the largest, which may be related to the acidity of the soils. The enrichment capacity of P, K and Zn in B. edulis was strong, with an enrichment factor of 4.63 for P. The optimal pretreatment methods for near-infrared and ultraviolet spectra were SG + 2D and SG + MSC, giving an accuracy for the prediction set of 88.46% and 96.15%, respectively. The intermediate fusion model had the best prediction performance. The Holtelling T2 test revealed that all samples fell within the 95% confidence interval, and the accuracy for the training and prediction sets was 100.00% and 92.31%, respectively. The results obtained in this study will be useful for the rational utilization of wild edible fungi resources in Yunnan.

Key words: Boletus edulis; mineral elements; enrichment coefficient; data fusion; geographical origin identification