基于组成特性的肉骨粉种属鉴别标志性变量挖掘

为了全面表征不同种属肉骨粉的组成特性，并进一步挖掘肉骨粉种属鉴别标志性变量，研究基于166个来源可靠的不同种属肉骨粉样本（猪、鸡、牛、羊源），从基本组分、元素组成、脂肪酸组成和氨基酸组成4个方面全面获取物料组成特性信息。对比分析不同种属肉骨粉的69个组成变量，其中31个组成变量在种属间差异性显著（P<0.05）。主成分分析（Principal Component Analysis，PCA）结合偏最小二乘判别分析（Partial Least Square-Discriminant Analysis，PLS-DA）对肉骨粉种属间特异性进行探索性分析。结果表明，元素组成和脂肪酸组成可以为猪、鸡、牛、羊肉骨粉提供特异性组成标志变量；氨基酸组成是反刍动物肉骨粉的特异性组成标志变量来源。综合PLS-DA和单因素方差分析结果，以VIP值大于1，P<0.05为指标，研究获取了不同种属肉骨粉之间的特异性组成标志变量，分别为：C10∶0、C18∶0、C18∶2n6c（猪肉骨粉）；Ca、K、Zn、C18∶0、C18∶2n6c（鸡肉骨粉）；Sr、C14∶1、C17∶0、C17∶1、C18∶0、C18∶2n6t（牛肉骨粉）；H、Mg、Sr、C10∶0、C16∶0、C17∶0、C17∶1、C18∶0（羊肉骨粉）；Sr、Ba、C14∶1、C17∶0、C15∶0、C17∶1、C18∶0、C18∶2n6t、C18∶2n6c、丝氨酸（反刍动物肉骨粉）；K、Zn、C18∶0、C18∶2n6c（哺乳动物肉骨粉）。该研究可以为肉骨粉种属鉴别方法及其多元应用机理分析提供数据支持，并可以为肉骨粉多元应用细化至不同种属提供理论基础。

Markers mining for species discrimination based on component characteristics of meat and bone meal

Abstract: This study aims to comprehensively characterize the composition of meat and bone meal, and further to identify the specific variables of various species using a data mining method. Based on the component characteristics data, a comprehensive comparison and markers mining study were carried out for the meat and bone meal that produced by various species. 166 samples of meat and bone meal were produced from various species (55 swine, 43 poultry, 36 bovine, and 32 ovine) in different factories of China. Composition characteristics in the samples of meat and bone meal were detected from four aspects, including the proximate component, element, fatty acid, and amino acid composition. The results of proximate component show that there was a complex variation in the samples of meat and bone meal, leading to the difference in four species was not considered statistically significance. An one-way Anova variance analysis was conducted for the composition data of element, fatty acid, and amino acid. 69 component variables were compared, incuding 14 variables from element composition, 37 variables from fatty acid composition, 18 variables from amino acid composition, in the meat and bone meal from different species. Consequently, there were significant differences among species (P<0.05) for 31 component variables, including 10 variables from element composition, 20 variables from fatty acid composition, 1 variable from amino acid composition. It infers that the component characteristics of meat and bone meal varied significantly in different species, particularly on the specific component variables. A Principal Component Analysis (PCA) combined with the Partial Least Square Discrimination Analysis (PLS-DA) was used to explore the species specificity of meat and bone meal. The results showed that the composition variables of element and fatty acid can serve as markers to idnetify the swine, poultry, bovine, as well as ovine meat and bone meal. The composition variables of amino acid were mainly marker sources of ruminant and non-ruminant meat and bone meal. Comprehensively characterization using the PLS-DA and one-way Anova variance analysis demonstrated that, taking the VIP value greater than 1, while P < 0.05 as the united indicator, the specific variables were achieved in the meat and bone meal for the species of: (1) swine were C10:0, C18:0 and C18:2n6c, (2) poultry were Ca, K, Zn, C18:0 and C18:2n6c, (3) bovine were Sr, C14:1, C17:0, C17:1, C18:0 and C18:2n6t, (4) ovine were H, Mg, Sr, C10:0, C16:0, C17:0, C17:1 and C18:0, (5) ruminant and non-ruminant were Sr, Ba, C14:1, C17:0, C15:0, C17:1, C18:0, C18:2n6t, C18:2n6c and serine, and (6) mammal and non-mammal were K, Zn, C18:0 and C18:2n6c. These selected specific variables can provide a sound theoretical basis for the multi-application of meat and bone meal from different species. The finding can also offer sinificant data support for the mechanism analysis and further application of meat and bone meal, particularly on the species identification method.