Emerging inequality in the San Pedro de Atacama oases (Chile): An investigation of entheseal patterns using ANCOVA and factorial ANOVA

Abstract

Bioarcheologists have focused extensively over the past few decades on how to best investigate past activity, often concentrating on data collection protocols and more recently focusing on statistical approaches. Here, we complement ongoing studies focusing on emerging inequality during the Middle Period (AD 400-1000) in the San Pedro de Atacama oases (Chile) by investigating entheseal patterns among individuals (n = 210) interred in four cemeteries. This period represents a time of demographic expansion and the development of interregional networks and formalized social inequalities, all of which would have shaped the lived experiences of local inhabitants. The four cemeteries studied here allow investigations of potential differences between individuals living in a close geographic area: Solcor 3 and Casa Parroquial represent "elite" sites connected with the Tiwanaku state, the somewhat later site of Coyo 3 is associated with mining activities, and Quitor 6 Tardío represents individuals from relatively lower status. Using two complementary multifactorial tests, factorial ANOVA and ANCOVA, we identified differences in entheseal scores in several joints as a proxy for activity patterns among the cemeteries (left and right shoulders and wrists, and left elbow and hip; p < 0.05), with individuals from Casa Parroquial demonstrating lower entheseal scores in most joints. Results highlight how we can infer differences in entheseal patterns among individuals interred in cemeteries that were in close geographic proximity and in use over similar periods, highlighting possible differences in lived experiences in the sites categorized as "elite" as well as the effects of cemetery location, either within the core oases or outside them. Our analyses further suggest that ANCOVA and factorial ANOVA can identify more nuanced differences among the cemeteries while accounting for covariates in a single test, making them more robust inferential statistical approaches for this type of study.