Doctoral Dissertation Improvement: If and How Many "Races"? The Application of Mixture Modeling to World-Wide Craniometric Variation

在生物人类学中，根据表型特征将现代人分类的方法经历了漫长而复杂的历史。虽然了解人类生物变异的模式对于揭示人口结构、人口历史和现代人类进化的轨迹至关重要，但这种分类即使基于生物差异，也是有争议的，因为它过去与“种族”概念和相关的意识形态议程有关联。在体质人类学中，对“种族概念”的批评和对科学种族主义的焦虑引发了人们对任何使用形态变异作为探索人类群体之间进化关系的方法的研究的有效性的担忧。有人认为，“种族”、“族裔”或“地理血统”概念获取的任何生物信息必然会被个人信仰和有缺陷的方法混淆，而反过来，又不可能实现对人类变异的公正解释。因此，一些生物人类学家使用颅骨测量变异作为在表型距离研究中获取过去种群结构或在法医背景下估计个人祖先的手段，与那些将任何关于群体分化的讨论与种族分型联系起来的人之间存在着紧张关系。然而，关于文化和生物学之间相互作用的新的遗传学研究表明，在骨骼研究中可以探索其他方法，以科学地检测群体差异，并确定人口统计信息的统计和实际意义。最近的基因组学工作已经成功地将基于模型的聚类应用于世界范围内的大样本中的常染色体微卫星基因座，并识别出与主要地理区域、民族文化可识别的群体甚至混合群体密切对应的簇。这些结果表明按大陆血统进行种群分类的可行性，以及“种族”、“民族”和“祖先”信息在种群结构研究中的重要性。因此，核心问题仍然是：类似的嵌套层次子结构，或“种群集群”，能否通过使用现代人类样本的表型数据，特别是颅骨测量数据来揭示？该项目将基于模型的聚类、几何形态测量和传统多元统计学应用于W.W.Howells和T.Hanihara全球范围的颅骨测量数据集和来自人类学数据库欧登塞大学(ADBou)的大型中世纪丹麦墓地样本。使用传统的线性测量、地标间的桁架距离值和三维坐标数据(1)来测试这些新的、潜在更强大的量化方法在理解基于人群的颅面形状差异方面的有效性，以及(2)使用头面部测量来研究组内和组内歧视的统计学意义。该项目还探索了在法医和医学法律背景下使用颅骨测量学估计“地理血统”及其与“种族”的关系的证据相关性，以及在田纳西大学法医人类学数据库(FDB)的现代美国样本和美国人口普查2000年人口统计学的补充下，根据当前的Daubert专家证词标准产生的分类结果的实用价值。集群与地理区域、民族文化群体和混合人口的关联具有重大的学科和社会政治意义。这项研究将展示按大陆血统进行人口分类的潜力，并权衡“种族”、“民族”和“血统”信息对人口结构研究的重要性。它还将就改进地理血统估计的统计方法以及在法医人类学和法医-法医学背景下根据目前的杜伯特科学证据标准估计血统的证据意义提出建议。为这项研究获得的头面部测量将提供一个机会来解决人类生物变异的问题，特别是世界范围的颅骨测量多样性的量化。在检验混合建模与传统形态计量学和几何形态计量学相结合的统计效果以及聚类算法对种群间和种群内分类问题的适用性方面，本研究将有助于人类学的定量应用。此外，在提供来自人类学数据库欧登塞大学(ADBOU)收集的新的三维颅骨测量信息用于未来的形态测量研究时，该项目不仅将促进对“形状”的更广泛的研究，还将增加我们对这些过去丹麦人的生活史的了解，从而促进骨骼生物学家、人口遗传学家、生物考古学家和人口学家关注的领域的研究，并支持更多的大学之间的合作。由于为这项研究编写的原始数据和统计计算机代码将向其他学者免费开放，该项目将鼓励未来的研究人员调查这里采用的方法(因此促使对该项目所涉及的理论进行改进，并改进我们的科学方法)，并对使该项目成为可能的资源提出新的和不同的问题。

The classification of modern humans into categories using phenotypic traits has experienced a lengthy and complex history in biological anthropology. While understanding the patterns of human biological variation is essential to revealing population structure, demographic history and the trajectory of modern human evolution, such categorization, even if based on biological differences, is contentious given its past association with concepts of "race" and related ideological agendas. Within physical anthropology, criticisms of the "race concept" and anxieties over scientific racism have generated concerns regarding the validity of any study that uses morphological variation as a way to explore evolutionary relationships among human groups. It is argued that any biological information captured by concepts of "race", "ethnicity" or "geographic ancestry" is necessarily confounded by personal beliefs and flawed methodologies and that, in turn, it is impossible to achieve an unbiased explanation of human variation. A tension exists, therefore, between those biological anthropologists who use craniometric variation as a means to access past population structure in phenotypic distance studies or to estimate individual ancestry in forensic contexts, and those who associate any discussions of group differentiation with racial typing. New genetic research on the interaction between culture and biology suggests, however, that there are alternative methods that can be explored in skeletal studies for scientifically detecting group differences and determining the statistical and practical significance of demographic information. Recent genomic work has successfully applied model based clustering to autosomal microsatellite loci in large, world-wide samples, and has identified clusters which closely corresponded to major geographic regions, ethno-culturally identifiable groups, and even admixed populations. These results show the feasibility of population classification by continental ancestry as well as the importance of "race", "ethnicity" and "ancestry" information to studies of population structure. The central question remains therefore: can similar nested hierarchical sub-structuring, or "population clusters", be revealed from phenotypic, and specifically, craniometric data using modern human samples? This project applies model based clustering, geometric morphometry, and traditional multivariate statistics to the W.W. Howells and T. Hanihara world-wide craniometric datasets and large Medieval Danish cemetery samples from the Anthropological Data Base Odense University (ADBOU) collection. Traditional linear measurements, trussed inter-landmark distance values, and three-dimensional coordinate data are used (1) to test the efficacy of these new and potentially more powerful quantitative approaches to understanding population-based differences in craniofacial shape and (2) to investigate the statistical significance of among and within-group discrimination using craniofacial measurements. This project also explores the evidentiary relevance of estimating "geographic ancestry" and its relationship to "race" using craniometrics in forensic and medico-legal contexts and the practical value of the classification results produced given current Daubert standards for expert testimony with the addition of modern American samples drawn from the University of Tennessee's Forensic Anthropology Data Bank (FDB) and demographics from the United States Census 2000.The association of clusters with geographic regions, ethno-cultural groups, and admixed populations has significant disciplinary and socio-political implications. This research will demonstrate the potential of population classification by continental ancestry and weigh the importance of "race", "ethnicity" and "ancestry" information to studies of population structure. It will also provide recommendations for improved statistical methodologies for geographic ancestry estimation and for the evidentiary significance of estimating ancestry in forensic anthropological and medico-legal contexts given current Daubert standards for scientific evidence. The craniofacial measurements obtained for this study will offer an opportunity to address questions of human biological variation, and specifically the quantification of world-wide craniometric diversity. In testing the statistical effects of combining mixture modeling with traditional and geometric morphometrics and the applicability of clustering algorithms to questions of among and within-population classification, this research will contribute to quantitative applications in anthropology. Moreover, in making new three-dimensional craniometric information from the Anthropological Data Base Odense University (ADBOU) collection available for future morphometric research, this project will not only advance the broader study of "shape" but will also increase our knowledge of the life history of these past Danish peoples, thus facilitating studies in areas of concern to skeletal biologists, population geneticists, bioarchaeologists and demographers alike and supporting additional inter-collegiate collaborations. As the raw data and the statistical computer code written for this study will be made freely accessible to other scholars, this project will encourage future researchers to investigate the methodology here employed (therefore prompting refinements to the theories addressed by this project and improvements to our scientific methods) and bring new and different questions to bear on the resources that made this project possible.