【作者】 夏立娅；

【导师】 孙汉文； 申世刚；

【作者基本信息】 河北大学 ， 分析化学， 2013， 博士

【摘要】 食品产地溯源技术是保护地理标志产品的重要技术支撑之一。食品产地溯源的有效方法是选择能够表征地域信息的特征因子，并通过化学计量学方法分析其“指纹”特征，从而识别食品的原产地。为探索特征因子的有效选择方法，分析不同指纹图谱技术在大米产地溯源中的可行性，本研究利用电感耦合等离子体质谱（ICP-MS）、近红外光谱、高效液相色谱和离子色谱等技术，检测了不同产地大米中矿物元素、有机组分、阴离子的含量以及产地土壤中矿物元素和阴离子的含量，结合多元统计学方法，系统研究了不同产地大米的特征属性，建立了多种大米产地的溯源方法。1.利用ICP-MS和原子吸收分光光度计测定了四产地大米样本中Mg、K、Ca、Na、Be、Mn、Ni、Cu、Cd、Fe、Al、Cr、Zn、Sb和Pb共15种元素含量以及其产地土壤中交换态Mg、K、Ca、Na、Mn，有效态Mn、Co、Ni、Cu、Zn、Cd、Pb、Fe，全态Be、Al、Sb、Pb含量和pH值。结果表明，大米中Mg、K、Ca、Na、Be、Mn、Ni、Cu和Cd9种元素的含量在不同产地之间有显著差异，土壤中17种矿物元素含量及pH在不同产地间均有显著差异，大米中Mg、K、Ca、Na、Be、Mn、Ni、Cu、Cd、Pb含量与土壤元素存在显著相关性。利用大米中Mg、K、Ca、Na、Be、Mn、Ni、Cu和Cd9元素含量建立的产地溯源方法准确度高于全部元素，线性判别方法对大米产地的检验判别正确率为100%，交叉检验判别正确率为93.8%。对大米中15种元素的R型系统聚类分析结果表明：Mg、Cu、K、Ni、Be、Mn和Ca7种元素聚为一类，具有较显著的共同特征，均与土壤中元素分布显著相关，是合适的产地鉴别元素。2.利用近红外光谱和模式识别技术建立了大米产地的快速鉴别方法。大米的近红外光谱经过一阶导数和平滑处理后，利用主成分分析法（PCA）对数据进行了降维。通过前三个主成分的载荷图确定了与产地相关性较大的特征波段为7700~6700cm-1与5700~4300cm-1。在全波段内，凝聚层次聚类和线性判别鉴别方法都可以100%正确的鉴别响水大米和非响水大米；对于非响水地区大米的具体产地判别，聚类分析正确率为91.9%，线性判别分析的总体正确率为96.7%。利用特征波段，凝聚层次聚类和线性判别鉴别方法都可以100%正确的鉴别响水大米和非响水大米；对于非响水地区大米的具体产地判别，聚类分析正确率为95.7%，线性判别分析的总体正确率为91.6%。证实了特征波段是大米产地溯源的有效波段。3.利用《中药指纹图谱相似度评价系统》分析了大米的乙酸乙酯提取物液相色谱数据，建立了由75个指纹峰和18个共有峰组成的响水大米指纹图谱，其中7、10、11和14号共有峰分别为没食子酸、邻苯二酸、对羟基苯甲酸和阿魏酸。利用该指纹图谱对大米样本进行分析，以相似度为0.9为标准，可以准确判别响水大米和非响水大米。大米乙酸乙酯提取物液相色谱数据的主成分分析结果表明，保留时间在40.66min，5.39min，22.06min，39.66min等色谱峰是分析大米产地的重要色谱峰。逐步线性判别分析结果表明，利用所筛选的23个特征峰建立判别方程，对大米产地的判别正确率为100%，交叉验证的正确率为97.7%；保留时间为22.06min，38.00min，40.66min，35.93min，17.87min和5.39min的特征峰对于产地判别起到主要作用，与主成分分析的结果较为一致。利用逐步多元线性回归分析，筛选了22个特征峰建立大米产地的预测方程。产地判别正确率为100%，标准残差为0.699。与逐步判别分析筛选的特征峰结果相比较，除保留时间为44.97min的色谱峰外，其余特征色谱峰的选择一致。利用逐步多元线性回归分析确定的22个特征峰数据进行K最近邻近分析，k=2时对于大米产地的鉴别正确率最高，总体正确率为96.9％。进一步证实，上述方法确定的特征峰在产地溯源分析方面是非常有效的。4.利用离子色谱法分析了四个产地大米和土壤中F-、Cl-、NO2-，NO3-和SO42-含量，结果表明，大米中F-、Cl-、NO2-和NO3含量在四产地间存在显著性差异，-大米中F-、Cl-、NO2-，NO3-和SO42-含量与土壤中阴离子含量均存在显著相关性，具有较强的原产地特征。利用大米中F-、Cl-、NO2-，NO3-和SO42-含量建立的Bayls判别分析模型对产地的判别的回代检验正确率为100%，交叉检验正确率为96.9%，Q型系统聚类分析的正确率为81.3%。从上述研究结果可以得出，通过大米中矿物元素、有机成分和阴离子的产地特征因子，结合化学计量学分析方法对大米产地的溯源是行之有效的。更多还原

【Abstract】 Food geographical origin traceability system was a powerful method to protectgeographical indication. The effective method for food geographical origin traceabilitywas to select the characteristic factors which was characterized by geographical origininformation, then to analyze the "fingerprint" characteristics by chemometric methodsto identify the origin of the food. The objective of this paper was to search effectivemethod to select characteristic factors, and investigate the feasibility of tracing ricegeographical origin by fingerprint techniques. In this study, inductively coupled plasmamass spectrometry (ICP-MS), near infrared spectroscopy (NIR), high performanceliquid chromatography (HPLC) and ion chromatography technology were used todetect the contents of mineral elements, organic components and anion content indifferent origin rice. Multivariate statistical methods were used to analysis the ricecharacteristics and to establish the methods of origin traceability.1. The contents of Mg, K, Ca, Na, Be, Mn, Ni, Cu, Cd, Fe, Al, Cr, Zn, Sb, and Pb inrice, exchange state Mg, K, Ca, Na, and Mn, and effective state Mn, Co, Ni, Cu, Zn, Cd,Pb, and Fe, as well as full state Be, Al, Sb, and Pb in soil were detected by ICP-MS andthe atomic absorption spectrometry. The results showed that the Mg, K, Ca, Na, Be, Mn,Ni, Cu, and Cd in rice had significant differences between different producing areas.The pH and17kind mineral elements in the soil had significant differences betweendifferent producing areas. The Mg, K, Ca, Na, Be, Mn, Ni, Cu, Cd, and Pb hadsignificant correlation between rice and soil. The discrimination correct rate basised on9elements was higher than that of all elements. In linear discriminant analysis (LDA),the discrimination correct rate and cross discriminant correct rate of rice origin were100%and93.8%, respectively. The result of R-type system clustering analysis basedon15elements showed that: Mg, Cu, K, Ni, Be, Mn, and Ca clustered to a class. Itindicated that they are appropriate origin identify elements with the commoncharacteristics related with soil elements.2. A rapid method was developed for discrimination of the geographical origins of ricewith pattern recognition technique by near infrared spectrocopy (NIRS). After first derivativeand smooth processing, principal component analysis (PCA) was used to reduct thedimensionality of the spectral datas. Through the loading graph of the first three principal components, characteristic wave band (7700-6700cm-1,5700-4300cm-1) with max-relativitywas determined. In whole wave, using agglomerative hierarchical cluster analysis andFisher’s linear discriminant, the discrimination of Xiangshui rice and Non-Xiangshui rice wasall100%. The correct rate of specific geographical origins of Non-Xiangshui rice was91.9%by cluster analysis and96.7%by discriminant anlysis. In characteristic wave band, thecorrect rate of specific geographical origins of Non-Xiangshui rice was95.7%by clusteranalysis and91.6%by discriminant anlysis. The results indicate that it is feasible todiscriminate the geographical origins of rice with pattern recognition technique by NIRS, andselecting characteristicwave band is one of the validated methods to improve the precision ofthe discrimination mode.3. HPLC fingerprint of the ethyl acetate extracts was established by Similarity EvaluationSystem for Chromatographic Fingerprint of Traditional Chinese Medicine (Version2004A)with18common peaks. The common peaks of7,10,11, and14were gallic acid, oxophenicacid, p-hydroxybenzoic acid and ferulic acid, respectively. It found that this method wasreliable to identify the Xiangshui rice with the criterion of similarity larger than0.9. Thepeaks at40.66min,5.39min,22.06min, and39.66min were deemed as characteristic peakswhich closely related to the origin information by principal component analysis (PCA). Totalof23characteristic peaks which closely related to the origin information were determined bystepwise LDA, and the peaks at22.06min,38.00min,40.66min,35.93min,17.87min, and5.39min were deemed as important peaks as same as PCA. The discrimination accuracies andcross-validation obtained by LDA achieved100.0%and97.7%, respectively. Total of22characteristic peaks which closely related to the origin information were determined bystepwise multiple linear regression (MLR). The discrimination accuracies and standarddeviation are100%and0.699, respectively. Compared with the results of LDA, thecharacteristic peaks were identical except44.97min peaks. The discrimination accuracies ofK-nearest neighbor method (KNN) were96.9%with K=2which based on22characteristicpeaks selected by MLR. The results indicated that the characteristic peaks could be efficientlyused to classification of the geographic origin of rice.4. The contents of F-, Cl-, NO2-, NO3-, and SO42-in rice samples and soil samplesfrom four provinces of China were analyzed by ion chromatography. The result ofvariance analysis and correlation analysis demonstrated that there was significantdifference in the contents for F-, Cl-, NO2-, and NO3-in the rice samples from4 provinces, and the anions in rice were closely connected with the anions in soil. Thepredictions of geographic origin made by linear discriminant analysis (LDA) based onanions gave an overall correct classification rate of100%and cross-validation rate of96.9%. The correct rate of Q-type hierarchical cluster analysis (Q-type CA) was81.3%.The above results showed that the characteristic factors of mineral element, organiccomponents and anion content which was characterized by geographical origincombined with multivariate statistical methods are effective in rice geographicaltraceability.更多还原