Quantitatively distinguishing between bone surface modifications using confocal microscopy and scale–sensitive fractal analysis
Date
2023
Authors
Campbell, Ross M., author
Pante, Michael C., advisor
Du, Andrew, advisor
Krapf, Diego, committee member
Journal Title
Journal ISSN
Volume Title
Abstract
The damage found on fossilized bone surfaces resulting from the feeding behavior of various prehistoric taphonomic actors (hominins, carnivores, raptors, etc.) in archaeological assemblages is a crucial piece of evidence that provides an inferential framework within which archaeologists can reconstruct the ecological and behavioral contexts of our hominins ancestors. However, these reconstructions are only useful if the bone surface modifications (BSM) can be inferentially linked to the specific taphonomic actor which created the mark. The inability to do so in a standardized and replicable manner has sparked multidecade-long debates over the actors responsible for individual marks and has resulted in drastically different interpretations of site formation processes and hominin behavioral ecology. Therefore, the goal of this study is to determine whether variations in within-mark fractal variables, paired with the micromorphological variables presented in Pante et al. (2017), can aid in quantitatively distinguishing between four different taphonomic agents (cut, trample, tooth, and percussion marks). To achieve this goal, a sample of 100 experimentally - produced BSM were sampled from the existing collection in the 3D imaging and analysis laboratory at Colorado State University. Scans of individual marks were acquired using Sensofar's S-neox 3D scanner, while 3D models of the marks were analyzed with the Digital Surf® software. Quadratic discriminant and complimentary random forest models were created to identify relationships between the measured fractal variables and the taphonomic agents creating BSM. The results of the quadratic discriminant and random forest models classifying all 4 BSM agents result in low classification accuracies between 52% - 58%, thereby indicating the micromorphological and fractal variables could not be used to accurately identify taphonomic agents by their within-mark surface complexity/roughness measurements. However, sub - grouping the dataset into models discriminating between only pairs of BSM types (i.e., cutmark vs trample mark) increases the classification accuracy of the QDA and random forest models to the 60% - 86% range, thereby indicating the micromorphological variables presented in Pante et al. (2017), when paired with the fractal variables Smooth – Rough Crossover (SRC), Area Scale Fractal Complexity (Asfc) and the Scale of Max Fractal Complexity (Smfc), can discriminate between the known taphonomic agents in the sample with relative accuracy. This study is beneficial to the study of archaeological BSM as it aids in our understanding of hominin subsistence behavior in prehistoric contexts by continuing the development of an objective and standardized method of differentiating feeding traces which provides a platform for more scientific, i.e. testable inferences about hominin behavior in archeological sites.