Burned Rock Middens, Settlement Patterns, and Bias in the Lower Pecos Canyonlands of Texas




Koenig, Charles W.

Journal Title

Journal ISSN

Volume Title



Burned rock middens (BRMs) are one of the most common archaeological features encountered in the Lower Pecos Canyonlands of southwest Texas and Coahuila, Mexico. BRMs form from the repeated use of a single location for constructing earth ovens. Based in part upon interpretations of BRM accumulations, two models of Archaic settlement patterns have been hypothesized for the Lower Pecos: the semi-sedentary rockshelter and canyon collectors model and the nomadic foragers model. However, these two settlement pattern models have never been tested using site survey data. In order to test these two competing settlement pattern models, a new area within the Lower Pecos was surveyed: Dead Man’s Creek (a tributary to the Devils River). Observations regarding BRM site location data along Dead Man’s Creek (DMC) indicate that there could be a connection between BRM site location and the availability of naturally occurring sediment. Through the use of GIS, site frequency and density was analyzed using Buffer analysis to determine site patterns in relation to the Devils River. The patterns observed within the DMC data were then compared to three additional datasets: the Lower Pecos regional site data, site data from Seminole Canyon State Park and Historic Site, and site data from Devils River State Natural Area – North Unit (DRSNA-NU). The DMC data could only be compared to the Seminole Canyon and DRSNA-NU data because the regional data are too biased towards the main river canyons. Patterns within the frequency and density data for DMC, Seminole, and DRNSA-NU indicate that more earth oven cooking was occurring as distance away from the major rivers increased. This pattern of increased earth oven cooking away from the major river canyons conflicts with the canyon collector settlement pattern model, but there is too little site data to fully evaluate either the canyon collector or the nomadic forager models of Lower Pecos settlement pattern models. Further, the site data for the Lower Pecos is heavily biased in two ways, both of which impact settlement pattern modeling. First, nearly all of the surveys have occurred along the major river canyons. Second, there is a recording bias towards recent sites found on the surface. Based on the analysis of the limited geoarchaeological investigations, there is the potential for buried archaeology in the three main topographic settings in the region (uplands, rockshelters, and canyon bottoms). Further, due to geomorphic processes the most common sites present on the surface date to the last 3,000 RCYBP. These two biases have severely impacted previous settlement pattern hypotheses, and until we collect additional site data from areas greater than 7 kilometers from the major rivers and conduct extensive geoarchaeological investigations, settlement pattern models will remain biased. Only through multi-disciplinary, systematic studies can data be objectively collected to test previous hypotheses and build new, better-grounded settlement pattern models for the Lower Pecos Canyonlands.



Archaeology, Burned rock middens, Earth ovens, Settlement patterns, GIS


Koenig, C. W. (2012). <i>Burned rock middens, settlement patterns, and bias in the Lower Pecos Canyonlands of Texas</i> (Unpublished thesis). Texas State University-San Marcos, San Marcos, Texas.


Rights Holder

Rights License

Rights URI