AN INVESTIGATION INTO INTRAINDIVIDUAL VARIATION IN HUMAN SKELETAL SEXUAL TRAIT MORPHOLOGY

Author

Kaleigh Christine Best, Southern Illinois University CarbondaleFollow

Date of Award

8-1-2023

Degree Name

Doctor of Philosophy

Department

Anthropology

First Advisor

Shimada, Izumi

Abstract

AN ABSTRACT OF THE DISSERTATION OFKaleigh C. Best, for the Doctor of Philosophy degree in Anthropology, presented on May 12, 2023, at Southern Illinois University Carbondale. TITLE: AN INVESTIGATION INTO INTRAINDIVIDUAL VARIATION IN HUMAN SKELETAL SEXUAL TRAIT MORPHOLOGYMAJOR PROFESSOR: Dr. Izumi ShimadaSex estimation is the most vital component of the biological profile assessment as several other subsequent analyses across biological anthropology rely on the correct identification of biological sex. This is especially important in times of commingling, fragmentary and partial remains, in inconsistent sex estimation, or in studies of sex estimation variation. While a multitude of studies have contributed to the documentation of similarities and differences in the morphological expression of a specific trait and how well this trait can be used to classify the sex of an unknown individual, how the expression of these morphological traits relate to each other within the individual has been poorly studied. Further, although several intrinsic and extrinsic factors are known to contribute to sexual trait expression, how they interact to produce morphology, and how that morphology may change in response to different biological conditions is not well understood. This dissertation utilizes 3D scans of 235 individuals collected from the University of Tennessee Donated Skeletal Collection (UTK) and the Robert J. Terry Osteological Collection (Terry) to examine sexual trait expression across the cranium, mandible, os coxae, sacrum and scapula. Specifically, it evaluates: 1) if sexual trait expression is correlated between skeletal regions including localized anatomical regions, through matrix correlation and Pearson correlation tests, 2) if age affects morphological sexual trait expression through M/ANOVA and if so, how and when these changes occur using ANOVA and Tukey post-hoc tests, 3) if the magnitude and pattern of sexual trait expression varies between populations through M/ANOVA and matrix correlations, 4) using discriminate function analyses, how well geometric morphometric analyses capture sexual trait expression compared to published rates, and 5) through M/ANOVA and discriminant function analyses, which intraindividual skeletal regions are the most accurate in sex classification analyses compared to others. For each skeletal element, landmarks were placed on each scan and sexual trait expression and an intra observer placement study was conducted. The landmark configurations were then subjected to a generalized Procrustes analysis, variance- covariance matrices were generated, and a principal component (PC) analyses in MorphoJ for combined sex data, derived male and females from combined sex data, and for independently input males and females. PCs were visually interpreted and subjected to M/ANOVA for three age groups: 18-29, 30-45, and 45 and older. PCs were then subjected to stepwise linear discriminant function analyses with leave- one-out cross validation. The first five PCs for each element were then utilized for a matrix correlation test and evaluated using a strength of correlation test (Chan et al 2003) and for significance using a Pearson’s correlation test. Results indicate that for intraobserver error is low, with scores being mostly considered excellent in reliability. For UTK data, shape changes associated with sexual trait classification were found for each of the five elements. Within these tests, many influential regions were known to sex estimation analyses, but other, such as the basicranium are currently underutilized. M/ANOVA revealed that sex was significant for all five skeletal elements, with some contributions from age and ancestry impacting sexual trait expression, however these interactions were not within a consistent pattern. Through the results of the discriminant function analyses, it appears some skeletal elements, such as the os coxae, scored higher correct classification rates than those reported in the literature, while others were on par or below reported rates. The os coxae and regions around it, were the best elements for correct sex classification accuracy, with the scapula rated as the worst. In the Terry data, sexual trait related shape changes are presented for each of the skeletal elements studied. Known morphological regions, as well as some more subtle ones, were found to be influential. M/ANOVA revealed that all elements, but the scapula had statistically significant shape changes present for sex estimation. Age and ancestry appear to contribute minimally to sexual trait morphology, but again, not in a way that is consistent, either within the collection or similarly to UTK. By examining the discriminant function analyses, some correct sex-estimation rates were on higher or on par with those reported in the literature, but others, such as the scapula were lower. This may be due the size being a larger contributor for sex estimation than shape, and the geometric morphometric analyses minimizing its effects in these analyses. Also using these tests, the os coxae was selected as the best element for high levels of sex classification accuracy, but which element was next best was different than the UTK results. This likely indicates biocultural influences are affecting sexual trait morphology and should be considered in future analyses. Significant inter-element correlations were detected in both skeletal collections, with UTK having more pairings than Terry. The most common and most statistically significant inter-element correlation across collections was between the cranium-mandible, which may be reflective of the integrated nature of these two elements. Different element correlations existed between each collection, with no clear pattern in pairings present in both collections, which may suggest that biocultural influences are impacting sexual trait expression. However, in both collections, females had more overall correlations, and more statistically significant correlations than males, although there was less of a difference between sexes in Terry than in UTK. This result may be due to circulating hormones present during puberty, which may be contributing to more correlation between elements in females and not males. It also suggests that once again, biocultural differences between the Terry and UTK collections may be impacting the magnitude and patterns of correlation. This research highlights the complex dynamic nature of sexual trait expression in bone in relation to several intrinsic and extrinsic factors. It supports the idea that sexual trait expression is not concordant between skeletal regions among the same individual and that these inter-element correlations appear to be different between skeletal collections. Likewise, it suggests that age and ancestry may minimally be affecting sexual trait expression, but not in a consistent way across skeletal elements or between collections. It also suggests that while geometric morphometrics may help in capturing shape variation, it does not always translate to higher correct classification rates. This research also supports previous literature that the os coxae is the best element for sex estimation, although is not clear on what element is next best. Future studies in sexual trait expression should account for biocultural influences, as this research suggests that sexual trait expression is influenced by poorly understood intrinsic and extrinsic factors.