The thirteenth-century human occupants of Promontory Cave, Utah, distinguished themselves from surrounding Fremont populations by being successful hunting specialists of bison in a region normally peripheral for that species. The hunters’ success is evident from the abundance of faunal remains excavated from the dry caves, which has facilitated zooarchaeological study of bison hunting strategies. The dry cave conditions also preserved hundreds of worn moccasins, which are of particular interest because of the Canadian-Subarctic style in which they are made and for their potential to reveal more about hunting strategies of their wearers. Here, we isolate ancient DNA from 38 Promontory Cave 1 (42BO1) moccasin and hide fragments and use these data to determine the species and sex of animals used to construct the moccasins. We found moccasins were mainly made from bison and most (87%) were females. The strong female bias in our data, which we demonstrate is a significant departure from sex ratios in present-day bison herds, suggests that the occupants of the cave were purposefully targeting female bison for moccasin manufacture. Our study is the first to our knowledge to determine faunal sex ratios from an assemblage of archaeological leather and highlights another potential avenue for ancient DNA technologies to augment what can be learned from the archaeological record.

We extracted DNA from hides and prepared the extracted DNA into genomic libraries for sequencing following protocols developed for working with degraded DNA (Fulton and Shapiro, 2019). Prior to extraction, we washed all samples with ultrapure water to remove surface debris. We extracted DNA following Dabney et al. (2013) and included one DNA extraction negative (no sample). Following extraction, we purified darkly colored extracts with polyvinylpolypyrrolidone in columns to remove PCR inhibitors following Arbeli and Fuentes (2007).

We generated Illumina sequencing libraries following either Meyer and Kircher (MK) (2010) or, for the most recently processed samples, the Santa Cruz Reaction (SCR) (Kapp et al., 2021) (Table 1). The SCR is a single-stranded library preparation approach that more efficiently converts extracted DNA into sequenceable molecules compared to double-stranded approaches like MK (Kapp et al., 2021); we therefore prepared SCR libraries from samples initially prepared following MK but with remaining extract. For three samples the MK preparation exhausted the DNA extract. For MK libraries, we labeled each molecule on both ends via a dual indexing PCR (iPCR) using KAPA Hifi (Roche, Pleasanton, CA, USA), unique indexes, and TruSeq Illumina sequencing primers, which we amplified in the modern DNA lab for 25 cycles. Following SCR library preparation, we performed quantitative PCR for each library to determine the optimal cycle number for dual indexing PCR, as described in Kapp et al. (2021). We cleaned all indexed libraries with a 1.5x concentration of SPRI beads (Beckman, Indianapolis, IN, USA). We quantified library concentration with a Qubit fluorometer (Thermo Fisher, Waltham, MA, USA) and pooled libraries at equimolar ratios. We sequenced all libraries and negative controls across several Illumina Miseq 2x75bp and NextSeq 2x150bp runs at the PGL.

We trimmed reads of adapters, removed low quality reads (-q 15) , removed reads smaller than 25 base pairs (-L 25), and merged reads using SeqPrep2 (https://github.com/jeizenga/SeqPrep2). We then used prinseq v.0.20.4 (Schmieder and Edwards, 2011) for complexity filtering and removal of duplicate reads. We mapped all reads with a minimum quality threshold of 30 (-q 30 ) using BWA (aln -l 1024;  version. 0.7.12-r1039) to cattle (Bos taurus ARS-UCD1.2_Btau5.0.1Y - bosTau9+Y), bighorn sheep (Ovis canadensis; NCBI CP011912.1), and roe deer (Capreolus capreolus CCMK01) nuclear genomes.

Raw data is not uploaded as it could contain human DNA of the Promontory Cave occupants. Only alignments of the hide samples to cattle, deer, and sheep are uploaded here.