基于数据融合的藏药五脉绿绒蒿品质评价及其环境影响因素研究

Quality Evaluation of Tibetan Medicine Meconopsis Quintuplinervia Regel. Based on Data Fusion and Study on Its Environmental Impact Factors

关键词:  红外光谱  数据融合  五脉绿绒蒿  总生物碱  品质评价  环境因素

Key words:  infrared spectroscopy  data fusion  Meconopsis quintuplinervia Regel.  total alkaloids  quality evaluation  environment factor

摘要:

目的 运用光谱数据融合策略建立藏药五脉绿绒蒿总生物碱含量快速检测模型,探讨环境因素对五脉绿绒蒿生物碱类物质积累的影响。方法 分别采集19个不同产地的五脉绿绒蒿的近红外光谱(near-infrared,NIR)、中红外光谱(mid-infrared,MIR)和中红外衰减全反射光谱(attenuated total reflectance,ATR),利用紫外分光光度法测定总生物碱的含量,采用主成分回归方法分别建立3个单一光谱模型和2个数据融合定量检测模型,通过Pearson相关性分析探讨地理因子、气候因子、土壤因子等不同环境因素与总生物碱含量的关系。结果 经数据融合后,NIR+ATR模型效果较好,其校正集相关系数、预测集相关系数、校正集均方根误差、预测集均方根误差值分别为0.980 3,0.997 2,0.060 3,0.063 3,且残差预测偏差>4。海拔、年均降水量与总生物碱含量呈显著正相关关系,纬度、年均温度与总生物碱含量呈显著负相关关系,土壤全氮和碱解氮与总生物碱含量呈极显著正相关关系。结论 光谱数据融合可以增强模型的预测能力,五脉绿绒蒿生物碱类物质的积累同时受大的地理环境因子和小环境土壤因子及气象因子的影响。本研究可为五脉绿绒蒿快速质量评价及资源合理利用规划提供科学依据。

Abstract:

OBJECTIVE To develop a model for the quick quality evaluation of total alkaloid content and to explore the effects of environmental factors on the accumulation of alkaloids in Meconopsis quintuplinervia Regel.. METHODS The total alkaloid contents of Meconopsis quintuplinervia Regel. from 19 different areas in Qinghai-Tibet Plateau were detected by ultraviolet spectrophotometry. The near-infrared(NIR), mid-infrared(MIR) and mid-infrared with attenuated total reflectance(ATR) spectroscopy information of Meconopsis quintuplinervia Regel. were determined and integrated, respectively. Combined spectroscopy information with the content of total alkaloid, three single-spectrum and two data fusion quantitative models were performed by principal component regression. Pearson correlation analysis was utilized to explain the relationship between geographical factors, climatic factors, soil factors, and total alkaloid content. RESULTS It was found that the data fusion model of NIR coupled with ATR was the best model, with the related coefficient of calibration 0.980 3 and prediction 0.997 2, the root mean square error of calibration 0.060 3 and prediction 0.063 3, and residual predictive deviation>4. And altitude and annual precipitation were significantly positively correlated with total alkaloid content, while latitude and annual temperature were significantly negatively correlated with total alkaloid content. Moreover, total nitrogen and alkali-hydrolyzable nitrogen were positively correlated with total alkaloid content. CONCLUSION The spectral data fusion can enhance the prediction ability of the model. The accumulation of alkaloids was affected by both large geographical environmental factor, small environmental soil factors and meteorological factors. This study can provide a scientific basis for rapid quality evaluation and rational planning of resource utilization of Meconopsis quintuplinervia Regel.