Sokal R R, Uytterschaut H, Rösing F W, Schwidetzky I
Department of Ecology and Evolution, State University of New York, Stony Brook 11794.
Am J Phys Anthropol. 1987 Sep;74(1):1-20. doi: 10.1002/ajpa.1330740102.
We analyze the taxonomic structure of European populations at three time periods, the Early Middle Ages, the Late Middle Ages and the Recent Period. The data consist of sample means for 10 cranial variables based on 137, 108, and 183 samples for the three periods. Clustering by standard numerical taxonomic procedures reveals that the data are represented only poorly as hierarchic classifications. The clusters form significant and moderately strong associations with an arrangement of the samples by regions (geography) and by language family. Whereas during the early period, language family showed a stronger association with clusters based on cranial morphology, in the recent populations these clusters correspond better with geography than with language. Ordinations of these populations by means of nonmetric multidimensional scaling shows the continuity of the taxonomic structure at all three periods. Only a few populations are outliers. The relations between phenetic distances (cranial morphology), geography, and language are examined by means of multiple Mantel tests. At all three periods geography is correlated somewhat more strongly with phenetics than is language affiliation, but the correlation with the latter increases with time. When the data are pooled over the three periods, the populations tend to group by language affiliation more than they do by period. Ordination of the pooled data reveals language patterns rather than patterns due to period, showing strong shifts in cranial measurements through time. These analyses show that while there is no clear-cut taxonomic structure in European populations that would justify the traditional classifications based on the crania, there are significant and important associations with both language affiliation, geography, and time period, in this order. These patterns are likely to have become established through the migration and subsequent expansion of populations into their areas of occupation during the time interval studied rather than by geographic differentiation in situ.
我们分析了欧洲人群在三个时期的分类结构，即中世纪早期、中世纪晚期和近代。数据包括基于三个时期的137个、108个和183个样本的10个颅骨变量的样本均值。通过标准数值分类程序进行聚类分析表明，这些数据作为层次分类的代表性很差。聚类与按地区（地理）和语系排列的样本形成了显著且适度强的关联。在早期，语系与基于颅骨形态的聚类关联更强，而在近代人群中，这些聚类与地理的对应关系比与语系的对应关系更好。通过非度量多维尺度分析对这些人群进行排序，显示了所有三个时期分类结构的连续性。只有少数人群是异常值。通过多重曼特尔检验研究了表型距离（颅骨形态）、地理和语言之间的关系。在所有三个时期，地理与表型的相关性比语言归属与表型的相关性略强，但与后者的相关性随时间增加。当将三个时期的数据合并时，人群倾向于按语言归属分组，而不是按时期分组。对合并数据的排序揭示了语言模式，而不是时期模式，显示出颅骨测量随时间的强烈变化。这些分析表明，虽然欧洲人群中没有明确的分类结构来证明基于颅骨的传统分类是合理的，但与语言归属、地理和时期存在显著且重要的关联，顺序依次为语言归属、地理和时期。这些模式可能是在研究的时间间隔内，通过人群向其居住地区的迁移和随后的扩张而形成的，而不是通过原地的地理分化形成的。
