杂草疫情分析在大豆产地溯源中的辅助性应用研究

Study on the Auxiliary Application of Weed Epidemic Analysis to Trace Soybean Origin

中国是进口大豆主要消费国,近15年,累计进口大豆90 875万t。由于各国大豆品质不一,征税标准不同,国际大豆贸易中的掺假行为日益凸显。为了进一步完善大豆溯源技术,本研究对2005-2019年进口大豆中携带杂草籽的种类、截获频次、来源国等信息进行整合分析,采用主成分分析和聚类分析方法提取累计贡献率高的主成分,并分析各类杂草是否存在地域差异。结果显示:确定杂草种类981种,其中单一来源国杂草251种。主成分分析提取的前2个因子贡献率达95.41%,可反映进口大豆中携带杂草疫情的绝大多数信息。以上述2个因子为分类变量,对综合因子得分前100的杂草进行聚类分析的结果显示,进口大豆携带的部分杂草地域特征明显,杂草种类与地理位置密切相关,按其来源特征可分为一般性杂草、南美区和北美区3类。如圆叶牵牛(Ipomoea purpurea)等24种杂草在南美洲国家进口大豆中截获率更高,而苘麻(Abutilon theophrasti)等14种杂草在北美洲国家进口大豆中则更加常见,可为进口大豆产地溯源提供辅助性指标,为商品归类属性鉴别业务提供参考依据。

China is a major consumer of imported soybean, with a total of 908.75 million t imported in the past 15 years. The quality and the tax standard of soybean varies from countries makes adulteration profitable in global trade. So it is of great significance for traceability technology. This study integrated and analyzed the data of weeds carried in imported soybean such as species, intercepted frequency and country of origin from 2005 to 2019. Principal component analysis and cluster analysis were used to extract principal components with a high cumulative contribution and to analyse whether there were geographical differences between the various types of weeds. The result showed that, 981 weed species were screened from the imported soybean from 2005 to 2019, including 251 species from a single source country. The contribution rate of the first two factors extracted by principal component analysis was 95.41%, reflecting most of the information. The results of cluster analysis showed that the top 100 weed species could be divided into three categories: General, South American and North American. Part of the weed species had obvious regional characteristics in this study, such as Ipomoea purpurea and Abutilon theophrasti were common in imported soybeans from South America and North America respectively. Therefore, the weeds data analysis can be support to trace soybean origin, and provide reference to the commodity classification and attribute identification.