关键词: 阿尔巴斯山羊；矿物质；正交偏最小二乘判别分析；有监督的聚类分析；地理标志

Abstract: To systematically collect mineral data from Albas goat meat for feature analysis and authenticity discrimination modeling based on machine learning, this study collected a total of 149 samples of longissimus dorsi and biceps femoris muscle from Albas goats in Ordos City and goats and sheep from five surrounding counties or banners. The contents of 28 mineral elements in each sample were determined, and supervised cluster analysis and descriptive statistics were conducted. A model for evaluating the authenticity of Albas goat meat was established. The orthogonal partial least squares-discriminant analysis (OPLS-DA) model showed clear separation between Albas goats and other breeds, between goats and sheep, among the six production areas, and between grazing and confinement feeding. The effect of cluster analysis based on the selected characteristic elements was better than that of the 28 elements. The contents of Na, P, Zn, Cr and Sr in the muscle of Albas goats were the highest among the five goat and sheep breeds. The contents of 13 minerals such as Na, Ca and P were significantly higher in goat meat than in sheep meat. The contents of Se and Li in Albas goat meat from Wurenduxi Gacha, Otog Banner (a core production area) were the highest among the six core and non-core production areas. The contents of 14 minerals such as Na, Fe and Cu were significantly higher in the meat of grazing sheep and goats compared with their penned counterparts. The external validation of the OPLS-DA models based on the 28 mineral elements and the selected elements showed discrimination between Albas goat meat and the meat of other goat breeds as well as sheep with 100% accuracy, and revealed 95.71% and 90.0% accuracy in origin traceability, respectively, and discrimination between the meat of grazing goats and sheep and that of penned goats and sheep with 94.29% accuracy.

Key words: Albas goat; minerals; orthogonal partial least squares-discriminant analysis; supervised cluster analysis; geographical indication