食品科学 ›› 0, Vol. ›› Issue (): 0-0.

• 安全检测 •    下一篇

基于矿物元素指纹图谱技术的松江大米产地溯源

石春红,胡桂霞   

  1. 上海市松江食品药品检验所
  • 收稿日期:2019-08-27 修回日期:2020-05-29 出版日期:2020-08-25 发布日期:2020-08-19
  • 通讯作者: 胡桂霞 E-mail:xia19860302@163.com
  • 基金资助:
    2018上海市食品药品监督管理系统项目课题研究专项

Geographical Traceability of Songjiang Rice By Mineral Elements Fingerprints

Chun-Hong SHI1,Gui-Xia HU   

  • Received:2019-08-27 Revised:2020-05-29 Online:2020-08-25 Published:2020-08-19
  • Contact: Gui-Xia HU E-mail:xia19860302@163.com

摘要: 摘摘要:松江大米是国家地理标志保护产品,溯源技术是其品牌保护的重要手段。采用电感耦合等离子体质谱技术(ICP-MS)分析180件松江大米与非松江大米样本中40种矿物元素含量,并结合多元统计方法(方差分析、相关性分析、因子分析和判别分析)分析矿物元素指纹特征,建立判别模型以溯源大米样品的松江与非松江产地。结果表明,大米样本的矿物元素含量在松江与非松江地域间具有显著性差异,元素之间具有显著相关性,因而将筛选后16种元素降维至5个公因子,公因子得分散点图能够明确判别大米产地;以筛选后的B、Na、Fe、Co、Ni、Zn、As和Se共8种矿物元素指标建立的溯源模型对训练集大米产地的整体判别正确率为93.0%,灵敏度为95.2%,特异性为86.8%,交叉检验判别正确率为84.2%~92.3%。验证集样本验证已建立的溯源模型准确度,发现松江与非松江大米产地整体判别正确率为92.1%,灵敏度为96.0%,特异性为84.6%。该模型对训练集与验证集的判别统计学参数基本一致,证明该溯源模型具有优异的判别正确率、灵敏度、特异性和溯源稳定性,因此在判别松江大米与非松江大米的产地上具有切实可行性。结果表明,大米样本的矿物元素含量在松江与非松江地域间具有显著性差异。

关键词: 关键词: 松江大米, 矿物元素指纹, 判别分析, 产地溯源

Abstract: Abstract: Rices from Songjiang are subjected to protected geographic indication (PGI) according to Chinese legislation. Forty trace mineral elements have been determined and analyzed by inductively coupled plasma mass spectrometry (ICP-MS) in data set of rice samples with two geographical origins: Songjiang and Non-Songjiang. The discrimination model for classifying rice samples with Songjiang PGI and Non-Songjiang based on metal fingerprints has been developed. The obtained data were analyzed by variance analysis (ANOVA), principal component analysis and discriminant analysis. The results showed that the samples have different mineral profiles when they have been produced in different geographical origins . Discriminant model based on eight characteristic mineral elements for B, Na, Fe, Co, Ni, Zn, As and Se was constructed by the linear discriminant analysis , which achieved adequate classification rate(92-96%) for Songjiang samples against other origins . The discrimination model has overall correct of 93.0% for training set. Besides, the discrimination accuracy was 95.2% (sensitivity) for Songjiang PGI rices and 86.8% (specificity) for Non-Songjiang rices. The cross validation accuracy of discrimination were 84.2%~92.3% in training set,respectively. The overall discrimination accuracy for validation set was 84.6%~92.1% in back substitution test. The overal discrimination accuracy for validation set were closed with the test set proving that all eight metal indicators carry sufficient information of geographical traceability .The model constructed for Songjiang rices against Non-Songjiang rices achieved adequate classfication accuracy, good sensitivities and acceptable specifities for detecting the fraud in the Songjiang PGI label.

Key words: Key words: Songjiang rice, mineral elements, characteristic indicators, geographical traceability

中图分类号: