Beyond Correlation to Causation in Hunter–Gatherer Ritual Landscapes: Testing an Ontological Model of Site Locations in the Mojave Desert, California

Why are rock art sites found in certain places and not others? Can locational or environmental variables inform an understanding of the function and meaning of the art? How can we move beyond observed patterning in spatial associations to a credible explanation of such meanings and ensure that we are not confusing correlation with causation? And what variables were most relevant in influencing site locational choices? These and related problems, whether recognized or not, are the subtext of the last three decades of rock art site distributional and landscape studies. They are now especially important to resolve given the need for accurate predictive modeling due to the rapid transformation of certain regions from undeveloped rural areas into rural industrial landscapes. Partly with this problem in mind, Whitley developed a descriptive model that provides an explanation for the location of Native Californian rock art in the Mojave Desert. It identifies the variables most relevant to site locations based on ethnographic Indigenous ontological beliefs about the landscape. These concern the geographical distribution of supernatural power and its association with certain landforms, natural phenomena and cultural features. His analysis further demonstrated that this model can account for two unusually large concentrations of sites and motifs: the Coso Range petroglyphs and the Carrizo Plain pictographs. But unanswered was the question of whether the model is applicable more widely, especially to smaller sites and localities made by different cultural groups. We documented and analyzed three petroglyph localities with seven small petroglyph sites in the southern Mojave Desert, California, to test this model. These sites are attributed to the Takic-speaking Cahuilla and Serrano tribes. Our study revealed a good fit between the expected natural and cultural variables associated with rock art site locations, with the number of such variables present at any given locale potentially correlated with the size of the individual sites. In addition to the research value of these results, this suggests that the model may be useful in the predictive modeling of rock art site locations for heritage management purposes.