spatially dense 3d morphometric study of human craniofacial sexual dimorphism with respect to ancestry
Open Access
- Author:
- Yao, Wei
- Graduate Program:
- Genetics
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- None
- Committee Members:
- Mark Shriver, Thesis Advisor/Co-Advisor
- Keywords:
- Face Sexual Dimorphism
- Abstract:
- Numerous studies have found population variations in craniofacial traits and sexual dimorphism of the human face, but little research has been done to investigate the impact of ancestry on sexual dimorphism. This thesis quantified the degree of sexual dimorphism across seven population samples (West African, East Asian, Irish, Italian, Polish, Portuguese, and South Asian) and examined the inter- and intra-population sex variation patterns. A standard x, y, z coordinate system for 7,150 quasi-landmarks was applied to 3D faces of 3739 people aged from18 to 40. Principal Component Analysis (PCA) reduced the dimension to 44 feature variables explaining 98% of total variations. 773 individuals in the seven populations were extracted from the overall principal component (PC) space to be the focus of this study. MANOVA test on the scores of the top five PCs shows overall main effects of sex and ancestry (p-value = 0) and a sex*ancestry interaction term (p-value = 0). Univariate Two-way ANOVA further confirms the existence of main effects in every single PC (p value =0) and that interaction effect is present only in PC2 (p-value = 0.0017, family-wise alpha rate = 0.01). Then the levels of sexual dimorphism were calculated in two different ways: 1) as sex differences between averaged male- and female faces of a population group, illustrated by the visual patterns of sexual dimorphism between average faces; 2) as averaged individual sex differences, supplemented by intra-sex facial variability. These sample estimates were computed using two distance methods, Mahanalobis and Root-mean-square error (RMSE). The ancestry differences in the expression of sexual dimorphism were discovered by generating the 95% bootstrap confidence intervals for the sample statistics from 10,000 bootstrap replicates, One-way ANOVA and multiple comparisons of unequal group means on the bootstrap distributions of sex distances. Lastly, the population groups are ranked based on their Mahanalobis and RMSE sexual dimorphism distance sample mean, bootstrap mean validated by ANOVA and Tukey’s HSD test. We found that in the level of sexual dimorphism between consensus faces, East Asians rank the highest in both Mahanalobis and RMSE distance measures. Irish ranks the lowest in Mahanalobis distance, while West Africans rank the lowest in RMSE distance. When sexual dimorphism is calculated as the average individual face sex differences, East Asians rank the lowest and West Africans the highest in Mahanalobis distance, whereas West Africans rank the lowest and South Asians the highest in RMSE distance.