Published online by Cambridge University Press: 17 March 2021
The extent to which Clovis peoples hunted proboscideans is debated. Convention requires that for a proboscidean butchery site to be accepted, contemporaneous artifacts must be spatially associated with faunal remains, and there must be evidence of use of the remains. Fourteen sites in North America currently meet those criteria; at least 31 do not. While these are reasonable requirements for avoiding false positives, such an approach risks identifying false negatives—rejecting spatial associations that are systemic associations. Given the known distributions of Clovis and proboscidean sites, how likely is it that artifacts are coincidentally associated with proboscidean remains? Conversely, how many spatial associations could be unrecognized butchery sites? To answer these questions, we simulated chance associations by plotting empirically informed densities and sizes of archaeological and proboscidean sites on simulated landscapes in which people and animals are (a) uniformly distributed and (b) tethered to water sources. The simulated frequencies of coincidental associations were compared to the observed frequency of co-occurrences. Our results suggest that of the 31 indeterminate empirical associations, at least 17 and as many as 26 are likely systemic associations, more than doubling previous estimates and revealing a greater role of humans in Pleistocene proboscidean exploitation than previously recognized.