The nature and significance of worldwide human dermatoglyphic variation in finger ridge-counts

Within humans, dermatoglyphic features of friction ridge skin found on fingers, palms, soles, and toes, exhibit a wide range of variation across populations and data types that is maintained over time but ostensibly serves no direct functional purpose. Unlike any other aspect of human biology, dermatoglyphic data can both discriminate between individuals and remain unchanged throughout life due to fixation during intrauterine development. However, no agreement exists on the nature and significance of the observed variation in dermatoglyphic features partly because traditional statistical analyses cannot accommodate the structure of dermatoglyphic data. The goals of this dissertation were to identify the patterns of human dermatoglyphic variation in worldwide populations by comparing finger ridge-counts, directional asymmetry, fluctuating asymmetry, and summary response variables at different geographic and biological scales. The worldwide dermatoglyphics data set was introduced in entirety for the first time and contained more than 51,000 individuals from over 300 ethnic groups, and modeling strategies from data mining were applied like description, clustering, and prediction. Univariate finger ridge-counts performed better than summary total ridge-count and absolute ridge-count response variables, likely due to modeling constraints. Summary indices of asymmetry and diversity performed better than univariate Directional Asymmetry variables for each digit or Fluctuating Asymmetry variables for each digit, likely due to the structure of the data variables. From the supervised methods, the support vector regression models performed better at predicting the variation in dermatoglyphic features compared to distribution-dependent generalized linear models. However, these modeling frameworks often did not perform better vii than baseline models estimating the central tendency or performing linear regression. From the unsupervised methods, the k-means clustering assignments consistently captured the most variation in relevant dermatoglyphic features, but the remaining explanatory variables differentially captured variation depending on the specific dermatoglyphic feature. Principal component analysis loadings revealed developmentally linked drivers of variation in finger ridge-counts like finger pattern and size, digit 1 variation, radial-ulnar gradients, and 2D:4D contrasts. This dissertation on the nature and significance of worldwide human dermatoglyphic variation provides a framework for resurgence in anthropological dermatoglyphics to discover the evolutionary forces acting on and maintaining variation in discriminatory and persistent dermatoglyphic features.