Late Quaternary Paleoenvironments and Archaeology in the San Pedro Basin, Southeastern Arizona, U.S.A.

Abstract

One of the most challenging questions surrounding the Clovis colonization of North America is the character and structure of terminal Pleistocene environments, including floral and faunal communities. A series of cores in the mouth of an arroyo revealed late Pleistocene - early Holocene wetland sediments buried 12 meters below surface, at the approximate elevation of the entrenched modern San Pedro River channel. A suite of ¹⁴C dates show that wetlands of the ancestral San Pedro River occupied portions of the inner valley coincident with the Younger Dryas (13,000 - 11,500 cal yr BP) and the early Holocene (10,000 - 9,500 cal yr BP). A lack of Sporormiella fungal spores indicates that mammoths were rare or absent when Clovis populations appeared in the valley around 12,800 cal yr BP. Palynological and stable carbon isotope analyses show that C₄ grasses increased at 9,940 cal yr BP, just prior to frequent burning after 9,510 cal yr BP and rapid erosion at 9,470 cal yr BP. δ ¹⁸O values from soil carbonates vary but do not record a systematic shift in precipitation source or temperature during the late Pleistocene - early Holocene transition. The establishment of C₄ grasslands in the inner valley correlates with widespread changes in the Chihuahuan Desert flora around 10,000 cal yr BP. A relatively dense accumulation of Clovis-mammoth associations in San Pedro Basin contrasts the lack of megaherbivores indicated at Palominas Arroyo. The upper San Pedro Basin of southeastern Arizona contains a minimum of four Clovis-mammoth associations, making it possibly one of the densest concentrations of human-proboscidean sites on earth in terms of time and space. I use the Younger Dryas-age stratum known as the "black mat" to compare the Clovis-age archaeofaunal record of the basin to its paleontological background in order to measure the level of human predation that created this remarkable record. This analysis indicates that Clovis people were affecting the last mammoth populations to a significant degree, a situation expected only in the presence of abundant mammoths. I argue that this condition was met in the San Pedro Basin, possibly in the form of a terminal Pleistocene refugium. If the refuge hypothesis indeed explains mammoth predation, then Clovis-mammoth associations should occur as clusters as they do in the San Pedro Basin rather than as isolates as they are known to occur elsewhere. The use of radiocarbon frequency distributions to reconstruct prehistoric human and animal populations must account for taphonomic loss and other factors that affect the archaeological and paleontological records. Surovell et al. (Journal of Archaeological Sciences, 36, 1715–1724) have recently proposed a volcanic-based correction factor for removing "taphonomic bias" from temporal frequency distributions. Analysis of 718 radiocarbon dates sampled from the alluvium of the San Pedro and Santa Cruz Rivers and their tributaries in southeastern Arizona shows that discovery and scientific biases play an equally important role in the creation of radiocarbon frequency distributions, and that "taphonomic bias" has not been systematic through time. The latter principle is further demonstrated using a sample of 123 Pliocene to Clovis-age proboscideans from the San Pedro Valley. We propose an alternative model that is based on the nature of the stratigraphic record, with discovery bias, scientific bias, taphonomic loss, and the shape of the calibration curve all operating to influence the temporal frequency distribution of prehistoric phenomena.