UC Davis

The human pelvis plays a significant role in many critical biological processes. The angle of pelvic tilt impacts the curvature of the vertebral column which influences posture, the articulation of the pelvis with the femur affects locomotor ability, the width of the pelvis impacts thermoregulation, and the size and shape of the birth canal influence the mechanism and difficulty with which childbirth occurs. The relationship between individual pelvic bones to the femur, the vertebral column, and even to infant head size has been well documented, but due to the paucity of complete fossils, a robust analysis of the evolution of the shape and size of the pelvic girdle and the dimensions of the birth canal throughout hominin evolution is yet to be firmly established. Here, I seek to contribute to our understanding of the evolution of the human pelvis by confronting issues of bone fragmentation and the absence of soft tissue in the fossil record using virtual anthropology techniques including three-dimensional (3D) visualization and geometric morphometrics. I achieve this through three studies that focus on different challenges to paleoanthropological analysis created by the typically poor preservation of pelvic remains. In the pages that follow, the Introduction provides a review of the research that has contributed to the present understanding of hominin pelvic morphology and the methods that have previously been employed to elucidate patterns of pelvic variation. In Chapter 1, my co-authors and I present two new reconstructions of the Kebara 2 Neanderthal (KMH 2) pelvis using both virtual and physical techniques. The reconstruction process involved correcting the positions of misaligned fragments of the right hip bone and the sacrum, mirror imaging the right half of the sacrum and the right hip bone to create a new left side, articulating 3D printouts of the reconstructed elements, and using the physical articulations to align virtual renderings of the bones. Measurements taken on the new reconstructions were then compared to measurements taken from a previous reconstruction by Yoel Rak to assess interobserver reconstruction variation. In general, these new reconstructions tend to accentuate features noted previously in Kebara 2 and other Neanderthals, for example, they both present a long superior pubic ramus and an anteriorly positioned pelvic inlet; however, we also detect reconstructor interobserver error in the shape and size of the outlet. Relative to the previous reconstruction, in the first reconstruction, the outlet is small and has a smaller pelvic index (i.e., is more circular), while in the second reconstruction, the outlet is larger and retains the same shape (i.e., the same pelvic index). These two reconstructions not only help to improve our understanding of Neanderthal pelvic anatomy but also highlight the subjectivity inherent in fossil reconstructions. Chapter 2 examines the shape, size, and spacing of the adult human pubic symphysis. The goals of this study are to identify a statistical model that best predicts pubic symphysis morphology, and to assess the relationships between the anthropometric variables used to train the model and symphyseal shape. My co-authors and I trained several simple linear regressions and 2-stage linear regression models on a dataset of biometric data and landmark and semilandmark coordinate data for the purpose of identifying the best model to predict symphysis shape. The data were collected from recent CT scans taken from patients in the University of California health system. This study shows that linear regression modeling can be used to quantitatively estimate the shape of an individual’s pubic symphysis, and we propose that it can be used in addition to other reconstruction techniques to improve fossil pelvic reconstructions by more accurately estimating the shape of this joint and reducing the effects of researcher bias in the process. In Chapter 3, my co-authors and I trained another linear regression model to predict the translation and rotation matrix values that would transform a human right hip bone onto its left pair to create a new left side without referring to the sacrum. We trained the model on landmark and semilandmark coordinate data collected from the human dataset used in Chapter 2. We then applied this model to the 2 new reconstructions of the Kebara 2 pelvis presented in Chapter 1 to predict their left hip bones, and finally we assessed how well the model predictions corresponded with the human left sides and Kebara’s reconstructed left side. Our results showed that regression modelling can be used to reliably predict ‘missing’ human hip bones and Kebara 2’s left hip bones using a human training sample. We believe that this method can be employed in conjunction with researcher’s anatomical expertise and other techniques including the technique to predict the shape of the pubic symphysis which we present in Chapter 2, to reduce subjectivity in the fossil pelvic reconstructions and to elucidate more of hominin pelvic morphology. The Conclusion provides closing remarks, as well as some future directions we wish to pursue on this topic.