Correlation analysis between active components of Cornus officinalis and inorganic elements in rhizosphere soil and rapid analysis of origin quality by near-infrared spectroscopy combined with machine learning

Although it is rational to believe that the content of effective components and inorganic elements in Corni Fructus (CF) varies with the wide range of planting and the different environments of the bases, the effect of the inorganic elements of rhizosphere soil in different producing areas on them is still unknown yet. In this research, CF and its rhizosphere soil from 29 production areas in seven provinces of China were collected, the content of seven active components in CF was determined by High-Performance Liquid Chromatography (HPLC), and the levels of inorganic elements in CF and rhizosphere soil were measured by Inductively Coupled Plasma Atomic Emission Spectrometry (ICP-AES). As to the active components, Gallic acid was 0.179-1.458 mg/g, Protocatechuic acid 0.064-0.527 mg/g, 5-HMF 0.361-1.237 mg/g, Morroniside 10.326-24.110 mg/g, Loganin 4.539-11.620 mg/g, Sweroside 0.815-1.568 mg/g, and Cornuside 1.198-2.930 mg/g, respectively. The content of 14 inorganic elements in CF varied greatly between different regions, but the differences between samples from adjacent regions were relatively little. Most of the samples with a high content of inorganic elements in CF were from Shaanxi and Henan provinces. The content of 14 inorganic elements in rhizosphere soil samples from different areas, from high to low, were in the order of Fe > K > Ca > Na > Mg > Mn > Zn > Cr > Cu > Ni > Pb > As > Cd > Hg. The enrichment ability of CF to different metal elements in the soil varies, from strong to weak, in the order of K > Ca > Mg > Pb > Cu > Zn > Hg > As > Cd > Cr > Na > Ni > Fe > Mn. The correlation analysis results showed that there is a higher correlation between the content of inorganic elements in CF and the content of active components in CF than between the content of active components in CF and the content of inorganic elements in the rhizosphere soil. A place of origin identification model was established based on the Near Infrared (NIR) spectroscopy data combined with Support Vector Machine (SVM). The identification rate of CF from different producing areas was high, with a recognition rate of 96.55%. The validation results show that the model has a high recognition rate of 100.00% and 87.50% for CF in Shaanxi and Henan, respectively. The content prediction model established with a similar strategy had high accuracy in predicting the content of seven components in CF, and the model validation index R-2 ranges from 0.79 to 0.99. The validation results show that R-2 ranges from 0.95 to 0.99.