数量分类方法在蜂蜜孢粉学中的探讨:以山西中部地区为例

Numerical taxonomy in melissopalynology: A case study in the central region of Shanxi, China

蜂蜜孢粉学是研究蜂蜜中花粉的科学,是孢粉学的一门重要分支学科.本文选取山西中部地区19个蜂蜜样品作为分类运算单位,61种花粉类型作为分类性状,采用数量分类方法对蜂蜜样品进行了系统聚类分析(Q型聚类)和主成分分析.系统聚类将19个样品分成8类,而主成分分析将它们分为5类,并论述了每类的显著特点.结果表明,(ⅰ)数量分类方法用在蜂蜜孢粉学中蜂蜜样品的分类是可行的,而且分类结果直观易懂;(ⅱ)在系统聚类的实际分类过程中,需要特别注意分类阈值λ的取值问题,通常λ取值不同,分类结果也不同,因此λ取值要视具体情况而定,既把相似性大的样品归为同类,又把相似性小的样品尽量分开;(ⅲ)由于数量分类方法和蜂蜜孢粉定量分析方法都考虑了优势花粉类群的丰度,因此分类结果有相似之处;但由于前者同时还兼顾了非优势花粉类群的丰度,所以分类结果又有别于蜂蜜花粉定量研究结果.

Melissopalynology is an important branch of palynology that deals with the pollen in honey. In the present study, 19 honey samples from the central region of Shanxi and 61 pollen types from those honeys were selected as operational taxonomic units and unit characters, respectively. Hierarchical （Q-type） cluster analysis and principal component analysis were conducted on the samples to create a numerical taxonomy. The two analyses divided the 19 samples into 8 and 5 categories, respectively. The results showed that （i） numerical taxonomy is feasible for the melissopalynological classification of honey samples, and the taxonomical results are easy to understand. （ii） More attention should be paid to the threshold value （2） when using cluster analysis. Different 2 values may yield different taxonomical results. Therefore, the ）. value should be based on the actual situation so that very similar samples will be categorized together and dissimilar samples will be distinguished from one another. （iii） Finally, both the numerical taxonomy and melissopalynological analyses emphasize the frequencies of dominant pollen types, so their taxonomical results are similar. Numerical taxonomy also considers the frequencies of non-dominant pollen types, so the taxonomical results may differ from those of melissopalynological analysis.