五指毛桃活性成分含量近红外光谱建模研究  

Models for Determining Active Ingredients Content in Ficus hirta by Using Near Infrared Spectroscopy

作　　者：王颂 刘璐冰 王志宏 徐巧林[1] 陈颖乐 黄文妍 杨柳 曾雷[1] WANG Song;LIU Lubing;WANG Zhihong;XU Qiaolin;CHEN Yingle;HUANG Wenyan;YANG Liu;ZENG Lei(Guangdong Provincial Key Laboratory of Silviculture,Protection and Utilization/Guangdong Academy of Forestry,Guangzhou,Guangdong 510520,China)

机构地区：[1]广东省森林培育与保护利用重点实验室/广东省林业科学研究院,广东广州510520

出　　处：《林业与环境科学》2024年第4期43-50,共8页Forestry and Environmental Science

基　　金：广东省林业科技创新项目(2021KJCX004、2022KJCX012)。

摘　　要：基于近红外光谱技术(Near infrared reflectance spectroscopy,NIRS),对广东省主要五指毛桃Ficus hirta产地样品的水分及活性成分进行测定,建立定量模型并进行预测效果评价,实现五指毛桃成分含量的快速检测。研究在广东省湛江、茂名、云浮、河源、梅州5个主要产地采集五指毛桃292份,以高效液相色谱测定的质量分数作参比,采用NIRS结合偏最小二乘法,建立根中水分、补骨脂素及佛手柑内酯含量的快检定量模型。结果表明,五指毛桃水分含量、补骨脂素及佛手柑内酯质量分数的定量模型相关系数为0.97、0.91和0.83;校正均方根偏差为0.23、0.18、0.03;交互验证均方根偏差为0.27、0.25、0.04;预测均方根误差为0.20、0.20和0.03。五指毛桃NIRS定量模型中,水分模型稳定性和准确性较好,检测精度达到近红外快速检测设备通用要求;补骨脂素模型预测效果较好,相关性良好;佛手柑内脂模型效果较差,该指标含量较低,且分布梯度相对集中,建议作为初筛参考。By using near infrared reflectance spectroscopy(NIRS),the water content and active ingredi-ents of Ficus hirta collected from Guangdong province were determined,and the corresponding quantitative models were established,and the prediction effect of models was evaluated.The rapid determination of water content and active ingredient content in F.hirta was established.292 samples of F.hirta were collected from five major producing areas of Guangdong province,including Zhanjiang,Maoming,Yunfu,Heyuan and Meizhou.On the basis of high performance liquid chromatography(HPLC),using NIRS combined with par-tial least squares(PLS),NIRS quantitative models were established for the determination of water content,psoralen and bergapten content in the root of F.hirta.The correlation coefficients for water content,psoralen and bergapten content were 0.97,0.91 and 0.83 respectively,and RMSEC were 0.23,0.18,0.03 respec-tively,and RMSECV were 0.27,0.25 and 0.04 respectively,and RMSEP were respectively 0.20,0.20 and 0.03.The NIRS quantitative models of water content were efficient and robust,and the detection accuracy reached general requirements for near infrared rapid detection equipment,which can satisfy the daily needs of detection.The NIRS quantitative models of psoralen content have good effect and good correlation.The effect of the bergapten content model is poor,due to the low content of this index and the relatively concentrated distri-bution gradient.So it is suggested that NIRS can be used as the reference value of bergapten content initial screening in the root of F.hirta.

关 键 词：五指毛桃 活性成分 近红外光谱技术 定量模型 

分 类 号：S789[农业科学—林学]