基于稳定同位素与矿物元素的蜂蜜产地溯源判别模型的构建

Construction of honey origin traceability discriminant model based on stable isotopes and mineral elements

中文关键词:  蜂蜜  稳定同位素  矿物元素  产地溯源

英文关键词:honey  stable isotopes  mineral element  origin traceability

中文摘要:

      目的 本研究通过对不同产地蜂蜜的稳定同位素、矿物元素进行含量差异分析,结合多元统计学技术,筛选出有效的产地溯源指标,构建蜂蜜的产地溯源判别模型。方法 通过采集甘肃、江苏、安徽、湖南、广西、广东和辽宁7个地区的蜂蜜样本,利用稳定同位素比率质谱(Isotope ratio mass spectrometry,IRMS)和电感耦合等离子体质谱(Inductively coupled plasma mass spectrometry,ICP-MS)测定稳定同位素δ13C、δ15N和22种矿物元素。结合单因素方差分析(Analysis of variance,ANOVA)、主成分分析(Principle components analysis,PCA)、正交偏最小二乘判别分析(Orthogonal partial least squares-discriminant analysis,OPLS-DA)、线性判别分析(Linear discriminant analysis,LDA)等多元统计分析方法探讨不同产地蜂蜜判别的可行性。结果 不同产地的稳定同位素δ13C、δ15N和22种矿质元素均存在一定的差异。PCA分析提取的7个主成分方差总贡献率为83.254%,并能够实现不同产地蜂蜜的初步分类。基于OPLS-DA建立的判别模型,能够较好地进行7个地区蜂蜜的产地区分,同时筛选出VIP值大于1的特征差异指标(Cr、As、K、Al、Ti、Cu、δ13C、Li、δ15N、Na、Zn、Be、Mn)并用于Fisher判别分析,构建的判别模型判别正确率为100%,交叉验证率为95.7%,基本实现了蜂蜜的产地判别。结论 本研究表明利用稳定同位素和矿质元素并结合多元统计学分析能够建立蜂蜜的产地溯源模型。

英文摘要:

      Objective In this study, effective origin traceability indicators were selected to establish a honey origin discrimination model by analyzing the difference of stable isotopes and mineral elements in honey from different places, combined with multivariate statistical analysis technology. Methods Isotope ratio mass spectrometry (IRMS) and Inductively coupled plasma mass spectrometry (ICP-MS) were used to determine stable isotopes δ13C, δ15N and 22 mineral elements by collecting honey samples from seven regions of Gansu, Jiangsu, Anhui, Hunan, Guangxi, Guangdong and Liaoning. Combined with Analysis of variance (ANOVA), Principle components analysis (PCA), Orthogonal partial least squares-discriminant analysis (OPLS-DA) and Linear discrimination analysis (LDA) to explore the feasibility of honey differentiation in different origin. Results The stable isotopes δ13C, δ15N and 22 mineral elements are different from each other. The total contribution rate of seven principal components extracted by PCA analysis is 83.254%, and it can realize the preliminary classification of honey from different origin. The discriminative model based on OPLS-DA can well distinguish the origin of honey in seven regions. At the same time, the characteristic difference indexes with VIP value greater than 1 (Cr, As, K, Al, Ti, Cu, δ13C, Li, δ15N, Na, Zn, Be, Mn) were screened and used in Fisher discriminant analysis. The accuracy of discriminant model constructed was 100% and the cross-validation rate was 95.7%, which basically realized the origin discrimination of honey. Conclusions This study shows that the honey origin traceability model can be established by using stable isotopes and mineral elements combined with multivariate statistical analysis.