Author's School

Graduate School of Arts & Sciences

Author's Department/Program

Earth and Planetary Sciences


English (en)

Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)

Chair and Committee

Jill Pasteris


The rarity of good fossil samples throughout geologic time frequently makes fossil identification difficult. This dissertation presents a new, multivariate, statistically validated method to identify tetrapods based on quantification of the shapes of microstructural features in cortical bone of the postcranial skeleton. The ultimate goal is to reduce the reliance on rare, near-complete fossil skeletons. The method is validated on a set of 15,745 mammalian microstructural features from eleven diverse species. An additional set of 21,122 microstructural features from one species serve to examine microstructural variation within a single skeleton. Microstructural measurements were made on thin-sections using optical microscopy. Initial tests of the method were applied to extant mammalians whose taxonomic affinities were known. Three case studies comparing: 1) the left tibiae from 11 mammals,: 2) the mid-body of each left rib in Odocoileus virginianus, and: 3) five cross-sections from left rib seven of O. virginianus represented tests of inter-taxonomic, intra-skeletal, and intra-bone microstructural variation, respectively. Principal Component Analysis of measurements on the tibiae of 11 mammals was successful in discerning a taxonomic signal in the shape and size characteristics of primary vasculature, secondary osteons, Haversian canals, primary lacunae, and secondary lacunae. No single microstructure or measurement is sufficient to account for taxonomic variation. Rather, size, shape, and orientation of various microstructural features, in combination, define and distinguish the taxa. Soft Independent Modeling of Class Analogy properly reassigned test samples from several taxa. In contrast with the results from the multi-species set, analysis of the intra-skeletal and intra-bone case studies revealed no pattern of microstructural variation. The data suggest that the microstructural variation within a skeleton is small compared to variation between taxa and that intra-skeleton variation will not affect the overall taxonomic designation. All principal component analyses were tested and found to be significant at the 95% confidence level using Multiple Discriminant Analysis. This work establishes a methodology for using bone microstructural features as a means for reconstructing taxonomic identity and supports continued research on this methodology, with the goal of applying it to rare fossil specimens in order to enable a next-generation approach to paleoecological analysis.


Permanent URL: