Relethford J H
Division of Epidemiology, New York State Department of Health, Albany 12237.
Hum Biol. 1991 Apr;63(2):155-65.
The effect of genetic drift on the genetic structure of seven Irish populations was investigated using anthropometric data collected during the 1890s on 259 adult males. These populations ranged in size from 769 to 3757, were relatively stable over time, and were located within 119 km of one another. Two populations are known to have experienced considerable English admixture. Data on ten anthropometric variables (three body measures and seven craniofacial measures) were adjusted for age and used to compute a relationship (R) matrix. The R matrix was converted into a distance measure and compared with a potential genetic drift distance measure, defined as (1/Ni + 1/Nj), where Ni and Nj are the effective population sizes of groups i and j (derivation of this formula is presented). Distances were rank-transformed, and the correlation between their pairwise elements was computed using matrix permutation methods to assess significance. Under the hypothesis that drift affects anthropometric variation, these correlations are expected to be positive. The correlation between anthropometric distance and potential genetic drift distance is 0.123, which is not significantly different from 0 (P = 0.368). When a multiple regression model is used to adjust for geographic distance and English admixture, the partial correlation (0.369) is significant (p = 0.021). As part of further analysis of the genetic structure of these populations, the same analyses were repeated using a distance matrix derived from surname frequencies. The correlation of surname distance and potential genetic drift distance is 0.164, which is not significant (p = 0.264). When the multiple regression model is applied, the correlation is 0.401, which is borderline significant (p = 0.055). These results show the influence of genetic drift, local migration, and admixture on Irish population structure.
利用19世纪90年代收集的259名成年男性的人体测量数据，研究了遗传漂变对爱尔兰七个群体遗传结构的影响。这些群体的规模从769人到3757人不等，随时间相对稳定，且彼此相距不超过119公里。已知有两个群体经历了大量的英国血统混合。对十个人体测量变量（三个身体测量指标和七个颅面测量指标）的数据进行了年龄调整，并用于计算关系（R）矩阵。将R矩阵转换为距离度量，并与潜在的遗传漂变距离度量进行比较，该度量定义为(1/Ni + 1/Nj)，其中Ni和Nj是群体i和j的有效种群大小（给出了此公式的推导）。对距离进行秩变换，并使用矩阵排列方法计算其成对元素之间的相关性以评估显著性。在遗传漂变影响人体测量变异的假设下，预计这些相关性为正。人体测量距离与潜在遗传漂变距离之间的相关性为0.123，与0无显著差异（P = 0.368）。当使用多元回归模型调整地理距离和英国血统混合时，偏相关性（0.369）显著（p = 0.021）。作为对这些群体遗传结构进一步分析的一部分，使用从姓氏频率得出的距离矩阵重复相同的分析。姓氏距离与潜在遗传漂变距离的相关性为0.164，不显著（p = 0.264）。应用多元回归模型时，相关性为0.401，接近显著（p = 0.055）。这些结果显示了遗传漂变、局部迁移和血统混合对爱尔兰群体结构的影响。
