A Morphometric Approach to the Study of Archaeological and Modern Capsicum spp. Seeds Using Elliptical Fourier Analysis and Machine Learning Methods

Summary

This is an abstract from the "SAA 2023: Individual Abstracts" session, at the 88th annual meeting of the Society for American Archaeology.

Traditional morphometric, or shape, analysis of archaeobotanical remains utilizes linear measurements taken in set axes of view (e.g., lateral) to generate quantitative assessments of morphological variation—mainly of carbonized disseminules—between taxa, or within a taxon. In contrast, landmark and semi-landmark analyses (LMA) apply statistical methods to a series of points, or landmarks, on homologous areas of a specimen to identify morphological variation across specimens. A limitation of LMA, however, is that its performance is contingent on the consistent availability of identifiable landmarks. Analyses of non-landmarked outlines of biological specimens with “unornamented” shapes using Elliptical Fourier Analysis (EFA) can accurately distinguish between taxa. In this poster, we present the results of the application of EFA and machine learning methods to more than 450 modern Capsicum spp. seeds from six different species (C. annuum, C. baccatum, C. chacoense, C. chinense, C. frutescens, and C. pubescens). Preliminary results indicate that EFA outperforms morphometric techniques reliant on linear measurements alone, but overlapping shape variation makes species detection difficult apart from two species (C. baccatum and C. chinense). Nevertheless, misclassified species are misidentified as phylogenetically related taxa, suggesting that this method could be useful for identifying specific Capsicum lineages in archaeological specimens.

Cite this Record

A Morphometric Approach to the Study of Archaeological and Modern Capsicum spp. Seeds Using Elliptical Fourier Analysis and Machine Learning Methods. Caleb Ranum, Alan Farahani, Katherine Chiou, Julia Sponholtz, Patricia Mathu. Presented at The 88th Annual Meeting of the Society for American Archaeology. 2023 ( tDAR id: 474885)

Keywords

Geographic Keywords
Other

Record Identifiers

Abstract Id(s): 37174.0