Dissertations, Theses, and Capstone Projects

Date of Degree

6-2024

Document Type

Dissertation

Degree Name

Ph.D.

Program

Anthropology

Advisor

Eric Delson

Committee Members

Christopher Gilbert

William Harcourt-Smith

Stephen Frost

Subject Categories

Biological and Physical Anthropology

Keywords

Anthropology, Cercopithecidae, Old World Monkeys, Geometric Morphometrics, Yushe, Molar

Abstract

The present study is an attempt to take steps towards quantifying 3D molar shapes of the primate family Cercopithecidae to construct a framework for the automation of identifying the anatomical position and taxonomic affiliations of isolated molars. I introduce a new landmark protocol and present a proposed workflow model which can be applied in the future. For the purpose of this study, I collected and compiled the largest sample of 3D surface scans of varying wear grades of all six cercopithecid molar positions (i.e., M1-M3 and M1-M3). The sample comprises 2641 specimens from 865 individuals, encompassing wear grades A through D. Four-fifths of the specimens belong to extant species and the majority of the sample was independently identified by the academic community. The teeth are virtually isolated, cleaned, and, if necessary, reconstructed with Geomagic Studio software and landmarked in Landmark Editor. Geometric transformations and statistical tests are performed in R. A within- and between-user study shows that the landmark protocol can easily and consistently be applied, and the LaSEC function illustrates its utility. Multiple statistical tests (e.g., Shapiro-Wilk, Levene's test) show that, although the sample does not fulfill the underlying assumptions of the applied parametric statistical tests (univariate / multivariate normal distribution, homoscedasticity, absence of collinearity), the results are still positive as they match, and even surpass, the results of equal, non-parametric tests. Due to the large sample size, it is possible to separate the specimens into a training set (70 %) and an independent test set (30 %) to verify the success rate of the model, unlike prior studies, which were limited in how they could confirm their identification successes. Although the Linear Discriminant Analysis (LDA) produces promising results with the position identification approach, the positive genera identifications are biased towards the subfamily Cercopithecinae. About 70 % of all four-cusped molars in the training and test sets are identified to their correct position. Overall generic identification success ranges from 70 % (M3) to 80 % (M2) for the training sets and 47 % (M3) to 64 % (M1) for the test sets. The certainty of identification success increases when the identification results of multiple tooth positions of an individual are combined. Over 60 % of the individuals with more than one tooth in the study are identified correctly by all or the majority of their teeth. Some biases remain and for many of the specimens identification results are not better, and in some cases even worse, than Delson's (1973) identification to the four morphological groups (papionins, cercopithecins, colobines, Theropithecus). I used three sets of fossils to showcase the steps of the identification process, the results of the methods, and the necessary assessment of the data: colobine and guenon molars from a probably Middle-Late Pleistocene locality at Taung, South Africa; a collection of 106 isolated molars from Baxian cave in China; and specimens from two localities in the Yushe basin, China. The teeth from Yushe (including molars and premolars) are described here for the first time, and the LDA identification results are compared to other quantitative identification approaches.

This work is embargoed and will be available for download on Monday, June 01, 2026

Graduate Center users:
To read this work, log in to your GC ILL account and place a thesis request.

Non-GC Users:
See the GC’s lending policies to learn more.

Share

COinS