From Scatterplots to Statistics: Identifying the Local Isotope Range in Multivariate Data
In recent decades, isotopic assays of strontium, lead, and oxygen in biological remains have revolutionized archaeological migration studies by providing direct evidence for the occurrence, timing, and geographic origins of individual residence change. Such research requires the clear identification of ‘local’ isotopic baselines for comparison against assayed individuals, and yet no single method to accomplish this task has emerged as best practice. Some researchers advocate the use of commensal fauna for determining ‘local’ isotopic ranges whereas others look to the structure (e.g., modality) of the dataset under investigation. The use of multiple isotopes complicates the matter further, requiring new approaches to reduce the subjectivity and arbitrariness that characterize many earlier methods like scatterplots and standard deviations. Here we suggest a novel approach using standard statistical methods. We build on the assumption that the central tendency of faunal isotope ratios from habitation areas largely overlaps with the isotope ratios of ‘local’ humans. Furthermore, we use likelihood estimates of clustering solutions to systematically eliminate ‘non local’ fauna in multi isotope data sets. We present strontium and lead isotope case studies based on multiple taxa from Indus Civilization faunal assemblages in order to highlight the advantages of a standardized faunal metric.
Cite this Record
From Scatterplots to Statistics: Identifying the Local Isotope Range in Multivariate Data. Benjamin Valentine, Penny Jones, Erik Otárola-Castillo. Presented at The 81st Annual Meeting of the Society for American Archaeology, Orlando, Florida. 2016 ( tDAR id: 403096)
This Resource is Part of the Following Collections
min long: 59.678; min lat: 4.916 ; max long: 92.197; max lat: 37.3 ;