基于矿物元素技术的品种、产区葡萄酒的判别分析

李彩虹,开建荣,王彩艳,王芳,闫玥,张静,杨春霞,葛谦*

(宁夏农产品质量标准与检测技术研究所,宁夏 银川,750002)

摘 要 通过分析宁夏贺兰山东麓不同品种,不同产区葡萄酒中矿物元素含量差异,结合多元统计分析,筛选有效的溯源指标,构建葡萄酒品种和原产地判别模型。该研究采集了宁夏贺兰山东麓产区6个单品葡萄酒样品54份,甘肃武威产区和河北沙城产区葡萄酒样品10份,利用电感耦合等离子体质谱仪(inductively coupled plasma mass spectrometry,ICP-MS)测定了样品中58种矿物元素含量,结合方差分析、主成分分析和Fisher判别分析方法建立了葡萄酒品种和产地判别模型。结果表明,不同品种葡萄酒中有35种矿物元素含量存在显著差异;经过主成分分析,从58种矿物元素可提取出10个主成分,代表了总指标80.64%的信息;通过Fisher判别分析,回代检验的整体正确判别率为100%,但交叉检验的整体正确判别率仅为38.9%,说明基于矿物元素的差异性不能有效鉴别不同品种的葡萄酒。结合武威和沙城产区样品,经Fisher判别分析,回代检验和交叉检验的整体正确判别率分别为100.0%和98.4%,基本实现了不同产区葡萄酒的判别。研究证明矿物元素技术可用于葡萄酒的原产地判别。

关键词 葡萄酒;矿物元素;品种;产地;溯源

宁夏贺兰山东麓葡萄酒产区自2003年被国家列为“国家地理标志产品保护区”以来,所酿葡萄酒在国际上屡次获奖,已成为中国葡萄酒的代表性产区。葡萄酒的商业价值主要来源于产地和葡萄酒酿造生产的年份[1],因此,有些不法商贩为了牟取利益,通过伪造地理标签的方式误导消费者,从而损害了消费者的利益,给品牌竞争带来信任危机。我国葡萄酒产业起步较晚,对于产区、品种葡萄酒的鉴别,手段更是匮乏,故葡萄酒市场比较混乱,亟需可靠、实用的葡萄酒原产地溯源技术和判别方法[2-3]

农产品产地溯源主要是分析表征不同地域来源农产品的特异性指标,目前主要采用质谱、光谱和分子生物学等技术,通过分析农产品的矿物元素、挥发性成分、同位素含量与比率、DNA图谱等特征成分或指标,结合化学计量法,建立区分农产品产地来源的特征指纹图谱,从而对不同种类农产品进行产地溯源[4]。在影响农产品品质的自然条件因素中,原产地土壤环境的差异使其在很大程度上产生了独有的特点与个性[5],因此,矿物元素分析技术被认为是植源性食品产地判别较为有效的方法[6-7],已被广泛应用[8],例如枸杞[9]、中药材粉葛[10]、新疆红枣[11]、茶叶[12-14]等的产地溯源,均取得了良好的判别效果。

本研究以宁夏贺兰山东麓不同品种(霞多丽、美乐、蛇龙珠、马瑟兰、赤霞珠、黑比诺)葡萄酒和我国不同产区(贺兰山东麓、武威产区和沙城产区)赤霞珠葡萄酒为研究对象,采用电感耦合等离子体质谱技术,分析葡萄酒中Ag、Al、As、Ba、Be、Bi等58种矿质元素,研究矿质元素在品种和产地葡萄酒判别中的可行性,研究成果可为葡萄酒品种、产地溯源提供科学方法和理论依据。

1 材料与方法

1.1 材料与试剂

实验标准溶液选用美国Perkin Elmer公司的4组57种混合标准溶液(Ag、Al、As、Ba、Be、Bi、Ca、Cd、Co、Cr、Cs、Cu、Fe、Ga、Li、Mg、Mn、Na、Ni、Rb、Se、Sr、Tl、U、Zn;Au、Hf、Ir、Pd、Pt、Ru、Sb、Sn;B、Ge、Mo、Nb、P、Re、Ta、Ti、W、Zr;Ce、Er、Eu、Gd、Ho、Nd、Pr、Sc、Sm、Tb、Th、Tm、Y、Yb)和中国计量科学研究院的汞单元素标准溶液(Hg),共计58种元素;质控样品选用生物成分分析标准物质胡萝卜(GBW 10047),中国地质科学院地球物理地球化学勘查研究所;硝酸、盐酸(优级纯),国药集团;水为实验室一级用水。

1.2 仪器与设备

ELAN DRC-e型电感耦合等离子体质谱仪(inductively coupled plasma mass spectrometry,ICP-MS),美国Perkin Elmer公司;Mars6 Xpress微波消解仪,美国CEM公司;AL104型电子天平,梅特勒-托利多。

1.3 实验方法

1.3.1 样品采集

供试葡萄酒采集自宁夏(银川、青铜峡、红寺堡等)(n=54),涉及6个酿酒葡萄品种,包括霞多丽(n=9)、美乐(n=9)、蛇龙珠(n=6)、马瑟兰(n=11)、赤霞珠(n=9)和黑比诺(n=10);甘肃(n=6);河北(n=4)。其中甘肃的1个葡萄酒样本及河北的4个葡萄酒样本为多品种酿酒葡萄混合酿制的葡萄酒,不纳入品种间元素差异分析,样本采集时间为2020年。

1.3.2 样品前处理

葡萄酒样品先挥发乙醇,采用微波消解法消解,具体操作步骤:量取5.00 mL葡萄酒样品于微波消解管中,置于赶酸仪上120 ℃挥干乙醇,待冷却后加入硝酸10 mL,常温静置预消解3~4 h后,随后置于微波消解仪中进行消解。选择温度控制,10 min爬升至120 ℃,保持10 min;10 min爬升至150 ℃,保持20 min;10 min爬升至180 ℃,保持30 min,消解完毕完全冷却后,去掉盖子,置于赶酸仪上120 ℃赶酸3 h,然后冷却至室温,用水完全转移至25.0 mL刻度试管中,定容至刻度,混匀;同时做试剂空白。

1.3.3 矿物元素含量测定

电感耦合等离子体质谱仪测定元素采用标准模式。优化后的工作条件为:发生器功率:1 300 W;检测器模拟阶电压:-2 350 V;离子透镜电压:6.00 V;雾化器流量:0.98 L/min;等离子炬冷却气流量:17.0 L/min;辅助器流量:1.20 L/min。上机测定葡萄酒样品及GBW 10047中58种元素,GBW 10047的各元素测定结果均在标准参考值范围内。各元素标准曲线相关系数均>0.99,检出限为0.000 1 μg/L~3.66 mg/L(各元素测定的标准曲线、检出限和定量限见附表1,https://kns.cnki.net/kcms/detail/11.1802.TS.20211106.1444.002.html)。

1.4 数据处理

每个葡萄酒样本平行2次测定,平行测定结果的相对相差<10%的平均值作为最后结果进行数据处理分析。采用SPSS 25.0软件进行单因素方差、主成分分析和Fisher判别分析。

2 结果与分析

2.1 贺兰山东麓产区不同品种葡萄酒中矿物元素含量差异分析

由表1可知,葡萄酒中58种矿物元素均有检出,且含量差异较大。Mg元素含量最高(>100 mg/L)Na、P含量为10~100 mg/L,Al、B、Ca、Mn、Fe、Sr元素含量为1~10 mg/L,其次为Ce、Cr、Cu、Rb、Ti、Zn元素,(<1 mg/L),其余元素含量均处于μg/L级。

通过对霞多丽、美乐、蛇龙珠、马瑟兰、赤霞珠和黑比诺6个单品葡萄酒中58种元素含量进行方差分析。其中As、B、Ba、Be、Cd、Co、Cr、Cu、Er、Gd、Hg、Li、Mo、Na、Nb、Ni、Sb、Sn、Sr、Tb、Ti、Th、Tl、Tm、U、Y、Yb、Fe、W、Hf、Pd、Ga、Zr、Re、Ca 35种元素含量在6个品种间存在显著差异(P<0.05),元素Al、Bi、Ce、Cs、Eu、Ge、Ho、Mg、Mn、Nd、Pr、Rb、Sc、Se、Sm、Zn、P、Ru、Au、Ta、Ir、Pt、Ag 23种元素含量在品种间差异不显著(P>0.05)(表1)。霞多丽酿制的葡萄酒中Be、Er、Gd、Nb、Tb、Tl、Tm、Y、Yb、Hf、Pd、Zr、Ca 13种元素含量明显高于其他品种,而Cr、Cu、Ni、Sn、Sr、Fe、Ga、Re 8种元素含量低于其他品种;美乐酿制的葡萄酒中Cr、Cu和Re元素高于其他品种,As、Be、Ba、Co、Er、Gd、Hg、Li、Mo、Nb、W、Tb、Tl、Tm、Y、Yb、Pd 17种元素低于其他品种;蛇龙珠酿制的葡萄酒中As、Ba、Cd、Co、Na、U、W 7种元素含量明显高于其他品种酿制的葡萄酒;马瑟兰酿制的葡萄酒中B和Mo元素高于其他品种,Cd、Na、Hf、Zr低于其他品种;赤霞珠酿制的葡萄酒中Hg、Li、Ni、Sb、Sn、Sr、Ga、Fe 8种元素高于其他品种;黑比诺酿制的葡萄酒中B、Ca元素低于其他品种。宁夏种植的不同品种的酿酒葡萄酿制的葡萄酒中矿物元素含量有其各自的特征,矿物元素在不同品种之间具有较大的差异性。

2.2 葡萄酒中矿物元素的主成分分析

主成分分析技术是重要的指纹分析技术之一,它将多项指标重新组合成一组新的互相无关的几个综合指标,能够用较少指标反应较多信息的一种无监督分析方法[15-21]。本研究对宁夏产区不同品种葡萄酒存在显著差异的35种矿物元素进行主成分分析,KMO统计量为0.646(>0.5),各元素之间具有显著相关性,可以进行主成分分析,结果见表2(详见附表2,https://kns.cnki.net/kcms/detail/11.1802.TS.20211106.1444.002.html)。第1主成分方差贡献率为28.645%,综合了Er、Y、Tb、Tm、Yb、Gd、Pd、Hf、Zr、Th 10种元素信息,这些元素均为碱金属和过渡金属;第2主成分方差贡献率为14.325%,综合了Sb、Na元素信息;第3主成分方差贡献率为7.401%,综合了Ni、Cr元素信息;第4主成分方差贡献率为6.658%,代表了Ba、Be、Ga元素信息;第5主成分方差贡献率为4.790%,代表了Re、Sr元素信息;第6主成分方差贡献率为4.559%,综合了Li、Sn元素信息;第7主成分方差贡献率为3.916%,代表了Mo元素信息;第8主成分方差贡献率为3.820%,代表了B元素信息;第9主成分方差贡献率为3.586%,代表了Cu元素信息;第10主成分方差贡献率为 2.940%,代表了Ca元素信息;前10个主成分累计方差贡献率为80.640%。筛选出Er、Y、Tb、Tm、Yb、Gd、Pd、Hf、Zr、Th、Sb、Na、Ni、Cr、Ba、Be、Ga、Re、Sr、Li、Sn、Mo、B、Cu、Ca 25种葡萄酒的特征矿物元素。

表1 不同品种酿酒葡萄酿制葡萄酒中矿物元素含量
Table 1 Content of mineral elements in wine made from different grape varieties

元素霞多丽美乐蛇龙珠马瑟兰赤霞珠黑比诺Al2.78±1.3a1.92±1.1a2.69±1.1a2.07±1.2a2.83±1.5a2.97±2.3aAs∗1.82±0.8b1.77±0.7b2.62±0.3a2.05±0.7ab1.98±0.7ab1.85±1.0bB2.18±0.7b3.14±1.4ab2.74±1.0ab3.50±1.2a3.45±0.8a2.24±0.8bBa0.057±0.04b0.059±0.01ab0.082±0.02a0.082±0.02a0.080±0.01a0.082±0.02aBe∗0.50±0.9a0.064±0.03b0.12±0.06b0.078±0.03b0.12±0.1b0.20±0.16abBi∗0.13±0.1a0.094±0.2a0.027±0.004a0.19±0.5a0.043±0.03a0.048±0.06aCd∗0.16±0.1abc0.092±0.1bc0.25±0.1a0.076±0.04c0.18±0.1ab0.11±0.07bcCe∗1.11±0.7a0.53±0.4a0.72±0.6a0.51±0.4a0.87±1.5a0.73±0.6aCo∗2.65±0.5ab1.94±0.7b3.13±0.9a2.27±0.7ab2.61±1.4ab2.50±0.9abCr∗0.20±0.05b0.30±0.01a0.27±0.06ab0.23±0.05ab0.31±0.09a0.27±0.1abCs∗2.11±1.4a2.56±1.5a3.30±1.0a3.32±2.7a3.31±2.2a3.18±3.3aCu∗0.37±0.2b1.54±1.5a0.70±0.3b0.79±0.5b0.74±0.5b0.48±0.3bEr∗0.12±0.08a0.032±0.02b0.054±0.04b0.041±0.03b0.053±0.05b0.055±0.03bEu∗0.032±0.02a0.025±0.008a0.042±0.02a0.032±0.01a0.038±0.03a0.036±0.02aGd∗0.15±0.1a0.057±0.05b0.10±0.09ab0.065±0.06ab0.097±0.1ab0.078±0.06abGe∗0.037±0.01a0.043±0.02a0.052±0.01a0.042±0.02a0.049±0.03a0.045±0.02aHg∗0.025±0.01ab0.012±0.01b0.023±0.01ab0.027±0.02ab0.033±0.02a0.018±0.01abHo∗0.035±0.03a0.010±0.007a0.015±0.01a0.012±0.01a0.014±0.01a0.015±0.009aLi0.038±0.01b0.035±0.008b0.047±0.01ab0.036±0.03b0.066±0.04a0.036±0.02bMg452±428a438±422a625±297a726±671a865±369a557±342aMn3.77±3.1a3.48±3.2a6.02±3.1a5.51±4.8a6.88±2.8a6.55±4.7aMo∗2.14±0.8bc1.47±0.7c3.62±1.2a3.74±1.8a2.81±1.5ab1.91±0.5cNa18.1±16ab12.0±13b30.5±15a10.6±9.0b20.9±13ab21.2±17abNb∗2.36±2.9a0.52±0.4b0.62±0.4b0.83±0.8b0.73±0.3b0.58±0.4bNd∗0.51±0.3a0.21±0.2a0.41±0.4a0.23±0.3a0.42±0.7a0.30±0.3aNi0.015±0.003b0.019±0.009ab0.020±0.004ab0.020±0.005ab0.024±0.005a0.020±0.007abPr∗0.12±0.08a0.049±0.04a0.096±0.09a0.055±0.06a0.11±0.2a0.074±0.07aRb0.40±0.01a0.46±0.01a0.62±0.1a0.62±0.1a0.66±0.3a0.69±0.5aSb∗1.12±0.4ab0.85±0.4b1.12±0.3ab0.90±0.4b1.66±0.9a1.30±1.0abSc∗2.34±2.1a2.31±2.5a2.04±1.2a2.93±1.9a2.85±2.5a1.63±0.7aSe∗0.80±0.2a0.75±0.2a0.82±0.2a0.79±0.2a0.90±0.1a0.77±0.2aSm∗0.086±0.05a0.044±0.02a0.079±0.06a0.054±0.03a0.075±0.09a0.065±0.04aSn∗4.67±3.6b8.06±8.0ab10.1±1.8ab5.12±3.8b12.5±12a7.65±6.6abSr0.58±0.1b1.16±0.1a1.37±0.3a1.30±0.6a1.41±0.6a1.20±0.4aTb∗0.025±0.02a0.008 3±0.007b0.013±0.01ab0.009 5±0.01b0.013±0.02ab0.012±0.01abTi0.17±0.04a0.16±0.04a0.15±0.04a0.15±0.04a0.17±0.03a0.14±0.05aTh∗0.61±0.5a0.24±0.2a0.35±0.2a0.50±0.8a0.60±0.7a0.37±0.6aTl∗0.16±0.1a0.083±0.02b0.13±0.02ab0.10±0.04b0.11±0.03ab0.11±0.04abTm∗0.018±0.01a0.004 8±0.002b0.006 7±0.004b0.005 8±0.005b0.006 6±0.006b0.008 0±0.004bU∗0.52±0.5ab0.78±1.0ab1.73±0.8a0.32±0.5b1.19±2.6ab0.77±0.5ab

续表1

元素霞多丽美乐蛇龙珠马瑟兰赤霞珠黑比诺Y∗1.07±0.7a0.30±0.2b0.50±0.4b0.37±0.3b0.45±0.4b0.46±0.3bYb∗0.18±0.1a0.048±0.02b0.072±0.04b0.059±0.04b0.080±0.06b0.089±0.04bFe0.81±0.2b1.47±0.4a1.87±0.8a1.46±0.5a2.07±0.5a1.77±0.9aZn0.18±0.06a0.12±0.05a0.16±0.06a0.15±0.07a0.14±0.05a0.16±0.09aP52.6±16a72.6±35a60.6±15a81.8±34a73.9±26a53.9±30aRu∗0.22±0.2a0.29±0.1a0.24±0.1a0.28±0.2a0.23±0.09a0.23±0.1aAu∗0.099±0.05a0.13±0.1a0.15±0.1a0.22±0.4a0.32±0.6a0.30±0.6aTa∗0.074±0.04a0.067±0.04a0.059±0.03a0.053±0.03a0.058±0.04a0.048±0.01aIr∗0.59±0.2a0.69±0.3a0.83±0.3a0.60±0.3a0.64±0.2a0.76±0.2aPt∗0.27±0.09a0.32±0.1a0.29±0.1a0.29±0.1a0.31±0.2a0.32±0.07aW∗0.29±0.3ab0.17±0.1b0.37±0.2a0.23±0.2ab0.27±0.1ab0.28±0.2abHf∗0.49±0.5a0.14±0.09b0.22±0.1b0.099±0.06b0.17±0.1b0.21±0.1bPd∗0.46±0.5a0.11±0.07b0.15±0.05b0.12±0.1b0.15±0.1b0.15±0.1bGa∗1.15±0.7b1.13±0.3b1.76±0.5a1.70±0.7ab1.82±0.4a1.71±0.6abZr∗10.6±8.3a2.26±1.5b3.41±2.3b1.87±0.9b2.82±1.8b3.92±4.8bRe∗0.025±0.02b0.056±0.02a0.044±0.01ab0.047±0.03ab0.046±0.02ab0.046±0.04abCa4.43±0.8a3.94±0.8ab3.35±0.3b4.06±0.8ab3.84±1.0ab3.46±1.1bAg∗0.11±0.05a0.11±0.02a0.076±0.03a0.098±0.04a0.082±0.06a0.11±0.06a

注:元素*表示元素质量浓度单位为μg/L,未标*的表示元素质量浓度单位为mg/L;表中数据均为平均值±标准差;同行不同的小写字母表示差异显著(P<0.05)

表2 前10个主成分的载荷矩阵及方差贡献率
Table 2 Load matrix and variance contribution rate of the first 10 principal components

统计主成分1主成分2主成分3主成分4主成分5主成分6主成分7主成分8主成分9主成分10特征值10.0265.0142.5902.3301.6761.5961.3711.3371.2551.029方差贡献率/%28.64514.3257.4016.6584.7904.5593.913.8203.5862.940累计方差贡献率/%28.64542.97050.37157.02961.81866.37770.29374.11377.70080.40

2.3 不同品种葡萄酒的判别分析

矿物元素含量的差异揭示了不同品种酿酒葡萄酿制的葡萄酒存在差异,但不足以对不同品种葡萄酒进行准确判别,为了验证品种是否影响产地判别,采用Fisher判别分析法对宁夏贺兰山东麓产区6个品种酿酒葡萄酿制的葡萄酒进行品种判别。

建立基于Fisher判别函数的一般判别方法对葡萄酒样本进行多变量判别分析,以58种矿物元素作为判别分析的自变量,进行逐步判别分析。结果显示,Al、As、B、Ba 等40种对品种判别显著的元素被引入到判别模型中。不同品种葡萄酒判别函数模型系数见表3,判别分类结果见表4。提取模型前5个典型判别函数,Willks′ Lambda检验结果进一步证实,在α=0.05的显著性水平下,5个函数对分类效果均为显著,其中判别函数1和判别函数2累积解释判别模型能力为73.2%,且相关系数均>0.92,表明判别函数1和判别函数2对6个葡萄酒品种分类占主要贡献作用,利用判别函数1和判别函数2的得分值作散点图。如图1所示,霞多丽、黑比诺、马瑟兰3个品种容易区分,并分别位于不同空间,赤霞珠、美乐和蛇龙珠3个品种样本有部分重叠。分类结果表明:回代检验的整体正确判别率分别为100%,回代检验是针对所有训练样本进行的检验,样品的错判率是相应总体率的偏低估计,而交叉检验比较真实地体现了模型的判别能力[22],交叉检验整体正确判别率仅为38.9%,每个品种均有大部分样本被误判,说明基于矿物元素的差异不能有效鉴别不同品种的葡萄酒。

图1 不同品种葡萄酒前2个典型判别函数得分散点图
Fig.1 Scattering points of the first two typical discrimination functions of wines from different varity

表3 不同品种酿酒葡萄酿制的葡萄酒判别函数模型系数
Table 3 Wine discriminant function model coefficients of different varieties of wine grapes

元素霞多丽美乐蛇龙珠马瑟兰赤霞珠黑比诺Al-13.514-16.764-13.975-16.289-8.953-6.58As-8.9416.55710.38310.5167.423.222B26.58330.76527.86132.46927.88722.344Ba-919.201-23.751216.073187.711148.8737.841Be-30.3-6.074-7.044-13.71-20.097-47.902Bi59.22265.34864.69980.35368.01851.12Cd-148.314-213.764-210.737-266.967-193.588-177.72Ce0.02-35.205-38.47-24.741-57.322-8.177Co14.89110.869.9338.0771.269-3.98Cr2.8182.9651.7092.4522.5012.262Cs9.1733.588-0.2843.2331.707-1.546Cu-0.139-0.106-0.145-0.174-0.134-0.222Er-6 276.104-6 311.514-5 871.957-6 894.947-5 099.506-4 787.506Eu-966.828-1 238.585-511.707-848.181-1 806.094-191.76Gd-121.791116.044262.915-511.088-917.632-239.897Ge-349.921-81.01686.518-91.043-23.697-225.621Hg1 337.2861 349.2161 200.6231 750.2381 429.785854.132Ho2 447.5324 086.4366 101.3276 106.5491 966.2522 783.936Li-0.789-0.801-0.764-0.925-0.504-0.456Mg-0.0220.0110.0290.0360.0250.001Mn0.648-2.288-6.803-6.505-4.673-2.529Mo6.5383.8737.5018.6575.8546.316Na0.1390.8511.5211.1280.6360.66Nb5.5712.15-2.7121.8311.931-1.627Nd14.45159.252252.497258.011330.605131.993Ni-1.832-1.858-0.868-1.259-1.531-1.233Rb-35.4767.04454.49637.57333.20944.733Sb25.52129.67224.84528.93928.62620.79Sc-8.173-3.949-5.086-4.016-3.155-5.08Se134.38231.53111.81717.08328.03142.821Sm2 290.0691 256.44794.8271 122.1771 558.032891.561Sn0.757-0.58-1.247-1.395-0.857-0.603Sr-10.0314.58419.86925.3816.9137.897Tb-9 523.174-7 848.902-9 103.827-7 729.693-5 280.211-6 364.523Ti119.511182.967266.484227.77290.757192.552Tm17 496.88414 004.85410 993.78712 269.85513 175.76911 733.599Y486.424403.788358.713448.99363.284341.385Zn166.8727.649-31.25712.604-19.8757.223Ir-9.93340.71566.12349.80144.34947.174Pt49.2269.719-20.134-24.763-4.69940.145(常量)-136.879-130.642-143.153-171.639-150.11-108.136

2.4 不同产地葡萄酒的判别分析

为了验证矿物元素分析技术对葡萄酒产地判别的可行性,建立基于Fisher判别函数的一般判别方法对宁夏贺兰山东麓、甘肃武威、河北沙城产区葡萄酒样本进行多变量判别分析,以58种矿物元素作为判别分析的自变量,进行逐步判别分析,结果显示,Al、B、Cs、Na、Rb、Sr、Ti、Fe、Zn、Pt、Re 11种对产地判别显著的元素被引入到判别模型中。不同产地葡萄酒判别函数模型系数见表5,判别分类结果见表6。提取模型前2个典型判别函数对分类效果均为显著,利用判别函数1和判别函数2的得分值作散点图,如图2所示,宁夏贺兰山东麓、河北沙城、甘肃武威3个产地容易区分,并分别位于不同空间。分类结果表明:回代检验和交叉检验的整体正确判别率分别为100%和98.4%,说明基于矿物元素指纹的差异可有效鉴别不同产地的葡萄酒。

表4 不同品种酿酒葡萄葡萄酒的一般判别分析结果
Table 4 Results of general discriminant analysis of different varieties of wine grape wine

方法品种预测组成员信息霞多丽美乐蛇龙珠马瑟兰赤霞珠黑比诺整体正确判别率/%霞多丽(n=9)900000美乐(n=9)090000蛇龙珠(n=6)006000回代检验马瑟兰(n=11)0001100100.0赤霞珠(n=9)000090黑比诺(n=10)0000010正确率/%100.0100.0100.0100.0100.0100.0霞多丽(n=9)331101美乐(n=9)031122蛇龙珠(n=6)012102交叉验证马瑟兰(n=11)11252038.9赤霞珠(n=9)102330黑比诺(n=10)021025正确率/%33.333.333.345.433.350.0

表5 不同产地葡萄酒判别函数模型系数
Table 5 Coefficient of discriminant function model for wines from different provenances

元素 宁夏甘肃河北Al-1.09510.339-7.532B1.097-3.4270.749Cs-1.759-7.243.133Na0.128-0.3360.164Rb19.04755.42412.593Sr-5.62413.99-26.479Ti146.535-13.904154.611Fe3.783-13.14211.82Zn24.613-31.873128.986Pt15.7455.1615.612Re129.489488.74884.646(常量)-22.543-56.487-56.001

表6 不同产地葡萄酒的一般判别分析结果
Table 6 Results of general discriminant analysis of wines from different regions

方法原属产区预测组成员信息宁夏甘肃河北整体正确判别率/%宁夏(n=54)5400回代检验甘肃(n=6)060100.0河北(n=4)004正确率/%100.0100.0100.0宁夏(n=54)5301交叉验证甘肃(n=6)06098.4河北(n=4)004正确率/%98.1100.0100.0

图2 不同产地葡萄酒前2个典型判别函数得分散点图
Fig.2 The scatter points of the first two typical discriminant functions were obtained

3 结论

通过分析宁夏贺兰山东麓产区的霞多丽、美乐、蛇龙珠、马瑟兰、赤霞珠、黑比诺6个单品葡萄酒中58种矿物元素含量及组成特征,明确了As、B、Ba、Be、Cd、Co、Cr、Cu、Er、Gd、Hg、Li、Mo、Na、Nb、Ni、Sb、Sn、Sr、Tb、Ti、Th、Tl、Tm、U、Y、Yb、Fe、W、Hf、Pd、Ga、Zr、Re、Ca 35种元素含量存在显著差异(P<0.05)。通过主成分分析确定了Er、Y、Tb、Tm、Yb、Gd、Pd、Hf、Zr、Th、Sb、Na、Ni、Cr、Ba、Be、Ga、Re、Sr、Li、Sn、Mo、B、Cu、Ca 25品种葡萄酒的特征矿物元素。Fisher判别分析确定了Cd、Ce、Co、Cu、Gd、Hg、Mg、Se、Zn、P 10种葡萄酒的有效溯源指标。Fisher判别分析方法构建的判别模型的回代检验的整体正确判别率为97.3%,但是交叉检验的整体正确判别率仅为35.6%,故基于矿物元素的差异不能有效鉴别不同品种的葡萄酒。

通过基于Fisher判别函数的一般判别方法对宁夏贺兰山东麓产区、甘肃武威产区和河北沙城产区葡萄酒样本中58种矿质元素进行多变量判别分析显示,Al、B、Cs、Na、Rb、Sr、Ti、Fe、Zn、Pt、Re 11种对产地判别显著的元素被引入到判别模型中,回代检验和的交叉检验整体正确判别率分别为100%和98.4%,说明基于矿物元素指纹的差异可有效鉴别不同产地的葡萄酒。

综合以上分析,矿物元素技术结合多元统计分析方法对产地葡萄酒的判别有效可行,建立的基于Fisher判别模型可用于葡萄酒原产地的识别,对地理标志产品葡萄酒及消费者合法权益的保护提供了有效的技术支撑。

参考文献

[1] 申雪, 袁玉伟, 聂晶, 等.稳定同位素在葡萄酒产地溯源中的研究进展[J].中外葡萄与葡萄酒, 2020(6):60-65.

SHEN X,YUAN Y W,NIE J,et al.Research progress of stable isotopes in wine origin tracing[J].Sino-Overseas Grapevine & Wine, 2020(6):60-65.

[2] 杜晓宁, 张鹏帅, 雷雯, 等.稳定同位素技术在我国食品安全检测领域的应用进展[J].同位素, 2019, 32(3):231-243.

DU X N, ZHANG P S, LEI W, et al.Application of stable isotope technique in food safety field[J].Journal of Isotopes, 2019, 32(3):231-243.

[3] 张龙. 植源性农产品溯源以及鉴别技术研究[D].杭州:浙江大学, 2012.

ZHANG L.The trace and authentication technologies on plant derived agricultural products[D].Hangzhou:Zhejiang University, 2012.

[4] 卢诗扬, 张雷蕾, 潘家荣, 等.特色农产品产地溯源技术研究进展[J].食品安全质量检测学报, 2020,11(14):4 849-4 855.

LU S Y, ZHANG L L, PAN J R, et al.Research progress on origin traceability technology of characteristic agricultural products[J].Journal of Food Safety & Quality, 2020, 11(14):4 849-4 855.

[5] 李长文, 魏纪平, 孙素琴.白酒宏观红外指纹三级鉴定[J].酿酒科技, 2006(6):35-38.

LI C W, WEI J P, SUN S Q.Liquor grading identification by three- step IR macro- fingerprint method[J].Liquor-Making Science & Technology, 2006(6):35-38.

[6] 钱丽丽, 邱彦超, 李殿威, 等.基于产地、品种和年份影响矿物元素含量的大米判别[J].食品科学, 2021, 42(16):322-327.

QIAN L L, QIU Y C, LI D W, et al.Influence of geographical origin, variety and crop year on mineral element contents of rice and geographical origin discrimination based on mineral elements[J].Food Science, 2021, 42(16):322-327.

[7] GONZLVEZ A, LLORENS A, CERVERA M L, et al.Elemental fingerprint of wines from the protected designation of origin Valencia[J].Food Chemistry, 2009, 112(1):26-34.

[8] 马楠, 鹿保鑫, 刘雪娇, 等.矿物元素指纹图谱技术及其在农产品产地溯源中的应用[J].现代农业科技, 2016(9):296-298.

MA N, LU B X, LIU X J, et al.Technology of mineral elements fingerprint and its application in origin traceability of agricultura l production[J].Modern Agricultural Science and Technology, 2016(9):296-298.

[9] 开建荣, 石欣, 李彩虹, 等.基于矿物元素技术的中宁不同产区枸杞的判别分析[J/OL].食品与发酵工业, 2022.https://doi.org/10.13995/j.cnki.11-1802/ts.027625.

KAI J R, SHI X, LI C H, et al.Discriminant analysis of Lycium barbarum from different areas in Zhongning based on mineral element technique[J/OL].Food and Fermentation Industries, 2022.https://doi.org/10.13995/j.cnki.11-1802/ts.027625.

[10] 贺媛媛, 孙倩倩, 郭波莉,等.基于矿质元素指纹的粉葛产地溯源研究[J].核农学报, 2021, 35(7):1 565-1 573.

HE Y Y, SUN Q Q, GUO B L, et al.Traceability of pueraria thomsonii radix(fenge) geographical origin based on mineral element fingerprint[J].Journal of Nuclear Agricultural Sciences, 2021, 35(7):1 565-1 573.

[11] 何伟忠, 赵多勇, 范盈盈, 等.新疆红枣营养品质与稳定同位素及矿物元素特征产地溯源比较[J].核农学报, 2021, 35(5):1 099-1 112.

HE W Z, ZHAO D Y, FAN Y Y, et al.Comparison of the nutrient quality, stable isotope and multi-element characteristics of Xinjiang jujube for origin traceability[J].Journal of Nuclear Agricultural Sciences, 2021, 35(5):1 099-1 112.

[12] 宋向飞. 基于矿质元素指纹图谱技术的碧螺春茶产地溯源研究[D].南京:南京农业大学, 2018.

SONG X F.Study on the geographical origin of biluochun tea based on mineral element fingerprint[D].Nanjing:Nanjing Agricultural University, 2018.

[13] 王洁, 伊晓云, 倪康, 等.基于稀土元素指纹的扁形茶产地判别分析[J].浙江农业科学, 2016, 57(7):1 118-1 124.

WANG J, YIN X Y, NI K, et al.Discriminant analysis of the origin of flat tea based on fingerprint of rare earth elements[J].Journal of Zhejiang Agricultural Sciences, 2016, 57(7):1 118-1 124.

[14] MA G C, ZHANG Y B, ZHANG J Y, et al.Determining the geographical origin of Chinese green tea by linear discriminant analysis of trace metals and rare earth elements:Taking Dongting Biluochun as an example[J].Food Control, 2016, 59:714-720.

[15] 李艳敏, 张立严, 狄红梅.主成分和判别分析在清香型白酒产地溯源中的应用[J].中国酿造, 2018, 37(1):145-148.

LI Y M, ZHANG L Y, DI H M.Application of principal component analysis and discriminant analysis in origin traceability of light-flavor Baijiu[J].China Brewing, 2018, 37(1):145-148.

[16] 高云, 郁志芳.基于主成分分析的芹菜品质评价[J].食品工业科技, 2020, 41(3):308-314;320.

GAO Y, YU Z F.Quality evaluation of celery based on principal component analysis[J].Science and Technology of Food Industry, 2020, 41(3):308-314;320.

[17] LIU Z, ZHANG W X, ZHANG Y Z, et al.Assuring food safety and traceability of polished rice from different production regions in China and Southeast Asia using chemometric models[J].Food Control, 2019, 99:1-10.

[18] 张腊腊, 韩明虎, 胡浩斌, 等.基于主成分分析的苹果品质综合评价[J].江苏农业科学, 2020, 48(3):209-213.

ZHANG L L, HAN M H, HU H B, et al.Comprehensive evaluation of apple quality based on principal component analysis[J].Jiangsu Agricultural Sciences, 2020, 48(3):209-213.

[19] 王甜甜, 廖小勇.基于主成分分析的长江流域水质评估[J].现代商贸工业, 2020, 41(6):194-195.

WANG T T, LIAO X Y.Water quality assessment of the Yangtze river basin based on principal component analysis[J].Modern Business Trade Industry, 2020, 41(6):194-195.

[21] CHENG H, QIN Z, GUO X, et al.Geographical origin identification of propolis using GC-MS and electronic nose combined with principal component analysis[J].Food Research International, 2013, 51(2):813-822.

[21] 张梦潇, 周文化, 莫华, 等.不同品种紫薯粉鲜湿面条的品质特性及主成分分析[J].食品工业科技, 2020, 41(1):79-85;124.

ZHANG M X, ZHOU W H, MO H, et al.Quality characteristics and principal component analysis of fresh noodles with purple sweet potato flour of different varieties[J].Science and Technology of Food Industry, 2020, 41(1):79-85;124.

[22] 李安, 陈秋生, 赵杰, 等.基于稳定同位素与稀土元素指纹特征的大桃产地判别分析[J].食品科学, 2020, 41(6):322-328.

LI A, CHEN Q S, ZHAO J, et al.Discriminations of the geographical origin of peach based on stable isotope and rare earth element fingerprint characteristics[J].Food Science, 2020, 41(6):322-328.

附表1 ICP-MS仪器测定矿物元素标准曲线、检出限和定量限
Enclose Table 1 Standard curve, detection limit and quantification limit of mineral elements determined by ICP-MS

元素工作方程相关系数检出限/(mg·L-1)定量限/(mg·L-1)元素工作方程相关系数检出限/(mg·L-1)定量限/(mg·L-1)Al27Y=22.849 2x0.999 720.012 60.042 77777777Rb85∗Y=56.771 4x0.999 8020.001 90.006 5As75Y=4.209 3x0.999 900.0050.016 7Sb121Y=4.434 08x0.999 9960.000 700.002 3B11Y=3.986 5x0.999 640.0310.103Sc45Y=24.658x0.999 9080.000 380.001 27Ba138Y=32.081x0.999 980.000 110.000 37Se78Y=0.987 727x0.999 9340.1140.379Be9Y=3.955 56x0.999 480.007 80.026Sm152Y=7.149 82x0.999 9950.000 1720.000 573Bi209Y=30.170 9x0.999 990.000 010.000 03Sn118Y=6.344 48x0.999 9890.000 480.001 6Cd114Y=13.523 8x0.999 920.000 070.000 22Sr88Y=110.688x0.999 5760.000 00420.000 014Ce140Y=23.650 2x0.999 990.000 010.000 03Tb159∗Y=71.122 7x0.999 9670.000 60.001 9Co59Y=12.295 7x0.999 980.000 220.000 73Ti48Y=11.888 8x0.999 9530.001 980.006 6Cr52Y=13.804 2x0.999 950.000 640.002 1Th232Y=52.208 3x0.999 9850.025 40.084 7Cs133∗Y=33.609 6x0.999 990.000 40.001 3Tl205Y=90.182 1x0.999 8420.000 001 50.000 004 9Cu63Y=11.755 4x0.999 900.000 280.000 93Tm169∗Y=136.841x0.999 8830.000 10.000 5Er166Y=27.982x0.999 950.000 010.000 03U238∗Y=89.630 5x0.999 9630.001 00.003 5Eu153∗Y=35.385 6x0.999 970.000 310.001 2V51Y=17.996 9x0.999 9470.001 70.005 67Gd133Y=12.423 4x0.999 980.000 030.000 11Y51∗Y=97.176 9x0.999 7630.000 50.001 8Ge74Y=7.887 32x0.999 910.000 780.002 6Yb174∗Y=46.533 1x0.999 8760.001 50.005 1Hg202∗Y=134.757x0.993 430.000 710.002 3Fe57Y=2.397 89x0.998 4622.157.17Ho165∗Y=140.913x0.999 870.000 120.000 41Zn66Y=6.887 32x0.999 6250.008 40.028La139Y=16.892 8x1.000 000.000 10.000 3P31Y=0.659 051x0.999 9583.6612.19Li7Y=9.353 04x0.999 800.000 420.001 4Au197Y=35.316 8x0.999 7730.000 0200.000 066 7Mg24Y=11.382 8x0.999 880.005 80.019 33Ta181∗Y=134.574x0.999 8610.000 80.002 7Mn55Y=14.900 2x0.999 980.000 340.001 1Pt195Y=10.265 1x0.999 9680.000 1140.000 38Mo98Y=21.107 2x0.999 720.000 084 00.000 28W184Y=22.252 1x0.999 950.000 01580.000 052 7Na23Y=58.550 3x0.998 470.002 600 00.008 677Hf180∗Y=45.878 8x0.999 8980.002 80.009 3Nb93Y=70.750 8x0.999 790.000 001 10.000 003 7Pd106Y=7.593x0.999 9670.000 440.001 47Nd142Y=18.718 8x0.999 970.000 026 00.000 086 7Ga69Y=18.820 3x0.999 9290.000 140.000 467Ni60Y=4.879 06x0.999 910.006 00.020Zr90Y=32.594 9x0.999 8730.000 0240.000 080Pb208Y=17.102 2x0.999 990.000 028 00.000 093 3Te130Y=12.925 4x0.999 6980.000 220.000 733Pr141Y=18.756 2x0.999 9970.000 010 20.000 034 0Re187∗Y=33.348 4x0.999 9730.0030.010

注:元素右上角*表示元素含量为μg/L,未标*的元素质量浓度为mg/L

附表2 前10个主成分的载荷矩阵及方差贡献率
Enclose Table 2 Load matrix and variance contribution rate of the first 10 principal components

成分12345678910Er0.9710.1280.0290.123-0.038-0.0390.0220.0110.0030.006Y0.960.150.0240.03-0.07-0.0570.0360.0280.0450.002Tb0.9370.1820.113-0.0930.038-0.1290.086-0.010.074-0.092Tm0.92-0.097-0.0270.2910.045-0.060.0040.106-0.0630.027Yb0.880.017-0.0280.369-0.054-0.015-0.0510.075-0.1140.064Gd0.8770.2710.18-0.0390.044-0.1510.079-0.0590.077-0.09Pd0.859-0.015-0.094-0.109-0.0660.150.0690.157-0.144-0.019Hf0.8490.095-0.0420.081-0.0150.155-0.0090.196-0.1180.141Zr0.6640.078-0.1250.375-0.2050.168-0.1050.163-0.2010.271Th0.605-0.0130.18-0.066-0.223-0.1470.218-0.1430.021-0.246Nb0.4090.07-0.047-0.238-0.3960.1220.0350.201-0.190.237Sb0.1270.8050.0610.105-0.0610.079-0.0620.013-0.0510.019Na0.1290.7690.167-0.003-0.2080.2290.1390.055-0.135-0.058Cd0.2140.5290.2490.030.0430.0490.3230.078-0.402-0.054Ni-0.1510.0460.8110.102-0.0120.1840.094-0.067-0.2960.102Cr0.0390.120.7920.0370.140.0850.0340.370.062-0.034Fe0.0470.3930.6140.1820.1610.035-0.077-0.1320.409-0.192Co0.340.3960.5370.1550.0480.1210.095-0.159-0.026-0.17Ba-0.0010.1820.340.8380.1030.0080.106-0.0850.021-0.011Be0.381-0.126-0.0870.811-0.0670-0.0550.143-0.120.174Ga-0.0330.4520.360.67-0.0690.0710.149-0.2160.1220.08Tl0.53-0.004-0.1280.571-0.0830.1060.2970.203-0.166-0.026Re-0.06-0.1280.139-0.1030.8870.0980.009-0.074-0.0740.028Sr-0.18-0.0460.2810.1340.6350.4550.128-0.0410.22-0.101Hg-0.024-0.0670.2510.042-0.60.4370.278-0.227-0.10.035Li-0.0590.1980.0860.0250.0190.8440.112-0.2390.008-0.026Sn0.0440.2280.2190.0230.1350.6610.0540.3950.14-0.131Mo-0.0320.01-0.0430.086-0.0460.1230.855-0.0220.109-0.07As0.280.3990.2060.0540.2790.0960.5660.01-0.1290.046W0.2740.2120.3510.102-0.2840.0110.5630.347-0.13-0.002B-0.3640.134-0.074-0.0070.2350.2110.062-0.6990.1290.265U0.0670.5030.020.0110.118-0.0180.1860.5490.154-0.067Cu-0.171-0.193-0.098-0.1210.0540.1280.0530.0150.7610.154Ca-0.082-0.113-0.1240.067-0.072-0.09-0.093-0.2720.0690.852Ti0.2840.0610.2790.2520.04-0.1080.0660.1270.4180.568特征值10.0265.0142.5902.3301.6761.5961.3711.3371.2551.029方差贡献率/%28.64514.3257.4016.6584.7904.5593.913.8203.5862.940累计方差贡献率/%28.64542.97050.37157.02961.81866.37770.29374.11377.70080.40

Discriminant analysis of wine variety and origin based on the content of mineral elements

LI Caihong,KAI Jianrong,WANG Caiyan,WANG Fang,YAN Yue,ZHANG Jing, YANG Chunxia,GE Qian*

(Ningxia Research Institute of Quality Standards and Testing Technology of Agricultural Products, Yinchuan 750002, China)

ABSTRACT Based on the analysis of the mineral element content in wines made from different varieties of grapes and from different producing areas, in the eastern foot of Helan Mountain in Ningxia, combined with multivariate statistical analysis, the effective traceability index was selected to establish the wine variety and origin discrimination model. In this study, 54 samples of six single grape wine samples from Helan Mountain in Ningxia, and ten wine samples from Wuwei in Gansu and Shacheng in Hebei were collected. The contents of 58 mineral elements in the samples were determined by ICP-MS. Based on analysis of variance, principal component analysis and Fisher discriminant analysis, a wine variety and origin discrimination model were established. The results showed that there were significant differences in the contents of 35 mineral elements in different varieties of wines. After principal component analysis, ten principal components and 25 mineral elements were extracted from 58 mineral elements, representing 80.64% of the total index information. Fisher discriminant analysis showed that the overall correct discriminant rate of back generation test was 100%, but the overall correct discriminant rate of cross test was only 38.9%, indicating that differences in mineral elements could not effectively identify different varieties of wine. Fisher discriminant analysis showed that the overall correct discriminant rates of back generation test and cross test were 100.0% and 98.4%, respectively, which basically realized the discrimination of wines from different regions. It is proved that mineral element origin tracing technology can be used in wine origin discrimination.

Key words wine; mineral elements; variety; origin; tracing

DOI:10.13995/j.cnki.11-1802/ts.029266

引用格式:李彩虹,开建荣,王彩艳,等.基于矿物元素技术的品种、产区葡萄酒的判别分析[J].食品与发酵工业,2022,48(12):281-287.LI Caihong,KAI Jianrong,WANG Caiyan, et al.Discriminant analysis of wine variety and origin based on the content of mineral elements[J].Food and Fermentation Industries,2022,48(12):281-287.

第一作者:高级实验师(葛谦助理研究员为通信作者,E-mail:278842005@qq.com)

基金项目:宁夏回族自治区自然科学基金项目(2020AAC03282);宁夏回族自治区自然科学基金重点项目(2021AAC02023);宁夏农林科学院农业高质量发展和生态保护科技创新示范项目(NGSB-2021-5);宁夏农林科学院先导资金项目(NKYJ-20-01);宁夏回族自治区自然基金项目(2021AAC03282)

收稿日期:2021-09-09,改回日期:2021-09-27