Predicting the past: archaeological predictive modeling in Central Texas

Abstract

Currently there is a debate raging among archaeologists regarding the origin and timeframe of the first inhabitants of the New World. Clovis archaeological sites are currently the oldest widely accepted occupation, and the Gault site in Central Texas has a well-stratified assemblage of Clovis artifacts. The discovery of additional sites like the Gault site could provide valuable information to resolve the debate. Two logistic regression models are created to locate areas with a high probability for site presence. The environmental characteristics are the independent variables and the probability of site presence is the dependent variable in these regressions. The first model compared the location of Paleo-Indian sites to the location of other sites within the study area. This model was moderately successful and showed that Paleo-Indian sites are found near ecotones, springs, and chert sources. The first model also indicated that Paleo-Indian sites are found preferentially on northern aspects and on steeper slopes than other sites. The limitations of this model are attributed to the similarities among all prehistoric locational patterns. The second model compared the locations of all archaeological sites in the area to the background environment of Central Texas. This model was more successful than the first and shows a higher probability for site presence near streams, near ecotones, and in association with various geological units. This model also shows that sites are more likely to be found at locations with northern aspects and steeper slopes. The improvement of the second model over the first is an indicator that the relatively large size of the study area is problematic for discerning specific occupation groups, but is suitable for modeling generic site locations.