GSA 2020 Connects Online

Paper No. 79-8
Presentation Time: 3:30 PM

THE PHENOMIC TOOLKIT AND PALEONTOLOGY: A CASE STUDY USING PALEOGENE MARSUPIALS (Invited Presentation)


VITEK, Natasha S., Department of Ecology & Evolution, Stony Brook University, 650 Life Sciences Building, Stony Brook, NY 11794-5245, BOYER, Doug M., Department of Evolutionary Anthropology, Duke University, 130 Science Drive, Durham, NC 27708, STRAIT, Suzanne G., Biological Sciences, Marshall University, 1 John Marshall Drive, Huntington, WV 45619 and BLOCH, Jonathan I., Florida Museum of Natural History, University of Florida, Gainesville, FL 32611-7800

Automated three dimensional geometric morphometric analysis (auto3dgm) is a promising high-throughput tool for studying shape variation across space and time. It is important to recognize that such high-dimensional methods complement, rather than displace, other morphometric approaches. Here, we use the fossil record of peradectid marsupial molars from before, during, and after the Paleocene-Eocene Thermal Maximum (PETM) in the Bighorn Basin, Wyoming, to exemplify uses and challenges of auto3dgm. The question of whether PETM peradectids belong to the same taxon as non-PETM peradectids remains open, complicated by long-standing difficulties discriminating isolated second and third lower molars (M2, M3) in order to compare homologous tooth positions.

Using 3D digital models, we assembled a dataset of lower molars (N = 65) following previously published protocols for surface creation and alignment. Most (N = 52) were peradectid M2s or M3s. A minority (N = 13) were either of another family or tooth position included to demonstrate the outcome of analyzing misidentified specimens. In a principal components analysis (PCA) those “mistakes” were not clearly differentiated from correctly identified specimens in PCs 1-2 but occupied the edges of specimen clusters. These results illustrate that while auto3dgm is not immune to the “garbage in, garbage out” problem, a careful check of which surfaces occupy the margins of distributions can help avoid spurious results in downstream analyses.

After misidentified specimens were removed, PCA decomposed the dataset into two non-overlapping clusters of PETM vs. non-PETM specimens along PC1. These results support a two-taxon interpretation across this interval. Within each cluster, PC2 partially separated known M2s from known M3s. Heat maps of mean differences between known M2s and M3s highlighted trigonid proportions as important for discrimination. A resulting trigonid length-to-width ratio became a simple tool for identifying isolated molars, including broken and worn specimens, to tooth position. Overall, auto3dgm forms one part of a broader morphometric toolkit necessary to document faunal change.