Permafrost preservation reveals proteomic evidence for yak milk consumption in the 13th century

The study investigates whether proteomic analysis of ancient human dental calculus can provide direct evidence for yak (Bos grunniens/mutus) milk consumption in medieval Mongolia and, by extension, offer insights into yak domestication, dispersal, and use. Yaks are crucial to high-altitude lifeways across the Tibetan Plateau and Mongolian mountains, supplying high-fat milk, fibers, dung fuel, and transport in extreme cold. Despite their importance, zooarchaeological evidence for early domesticated yak in Mongolia is scarce, with only a single identifiable specimen (a cranium) preliminarily dated to the Late Bronze/Early Iron Age and limited references in archaeological reports from Mongol-era sites. Historical sources (Xiongnu and Mongol periods) emphasize yak products (e.g., hair, felts) as politically and economically significant goods, yet explicit mentions of yak dairying or milk consumption are lacking. The center of yak domestication is broadly placed on the Tibetan Plateau (~3000 BCE), with proposed dispersal northward toward the Altai and Khangai or westward towards the Pamirs, but details remain unresolved. Given these gaps, ancient proteomics offers a route to detect milk proteins and identify taxa consumed, potentially clarifying the presence and role of yak in medieval Mongolia.

Previous paleoproteomic studies from northern Eurasia have documented consumption of ruminant and horse milk using milk proteins such as beta-lactoglobulin (BLG), caseins, and alpha-lactalbumin. However, prior work in Mongolia and broader Eurasia had not recovered evidence of yak milk from dental calculus. Taxonomic identification of yak milk via BLG is challenging because yak BLG shares over 90% sequence identity with other Bos species (e.g., Bos taurus, B. bison), and yaks produce two paralogous BLG variants: BLG-A (similar to Bos taurus and less taxonomically informative) and BLG-E, which differs by a single amino acid in the final tryptic peptide. Therefore, detecting yak-specific peptides relies on recovery of the final tryptic peptide of BLG-E, making yak identification more difficult than for other dairy taxa. Archaeological and historical records attest to yak products’ importance (hair, felt) from the Xiongnu through the Mongol period, but archaeological remains and textual references do not clearly document yak dairying or milk consumption. This study builds on established calculus proteomics methods to address these gaps and to evaluate whether yak dairy proteins can be identified in medieval Mongolian individuals.

• Sampling: Dental calculus was collected from 11 adult individuals from the elite Khorig 1 and Khorig 2 cemeteries (Khovsgol province, Mongolia) during 2018–2019 excavations. Calculus was removed with sterilized scalers into 1.5 mL tubes, with care to avoid contamination. Samples were freeze-dried and stored sterile until analysis. Research permits were obtained from the National Museum of Mongolia.
• Authentication and screening: Protein profiles were assessed using a custom screening approach referencing a curated database (OSSD) to evaluate oral/origin authenticity and potential contaminants. Oral microbiome, host immune, and environmental proteins were tallied to gauge preservation and authenticity.
• Protein extraction and preparation: Calculus was demineralized and proteins extracted using a phosphate buffered saline (PBS)-modified protocol in a dedicated clean room. Personal protective equipment and sterile procedures were used; extraction blanks and a post-excavation control (archaeological sheep bone) accompanied batches. Proteins were digested (trypsin), and peptides cleaned for MS analysis.
• LC-MS/MS: Peptides were analyzed at the Functional Genomics Center Zurich on a Thermo Q Exactive HF with heated electrospray ionization. Nano-LC used C18 trapping and analytical columns with a gradient from 5–40% acetonitrile in 0.1% formic acid at 200 nL/min. Data-dependent acquisition collected MS1 (300–1700 m/z) at 300,000 resolution (200 m/z) and MS2 at 15,000 resolution, with HCD fragmentation (isolation width 2.0 m/z, AGC 50,000, max IT 100 ms).
• Database searching: Raw MS/MS data were converted to MGF (ConSpect from Prot. 3.0 v3.0.1178) and searched with Mascot v2.6.0 against SwissProt, custom ancient databases, and additional in-house databases (HODM), with trypsin enzyme specificity, up to two missed cleavages, variable deamidation (N/Q) and oxidation, precursor tolerance 10 ppm, fragment tolerance 0.1 Da. Peptide and protein FDRs, ion scores, e-values, and post-translational modifications were evaluated. Distinct peptides were validated via NCBI Protein BLAST to confirm taxonomic specificity.
• Radiocarbon dating: Associated human bone was dated at the UZBRC Radiocarbon Laboratory (Uzbekistan), yielding calibrated ranges (95.4%) including 1170–1270 CE (HOR-27) and 1287–1397 CE (HOR-01); additional ranges include 1187–1397 CE and 1068–1213 CE for other samples. Collagen quality control followed standard C/N criteria.
• Data availability: Raw proteomics data deposited in ProteomeXchange (PXD024510; project DOI https://doi.org/10.6084/m9.figshare.1452100).

• Sample preservation and dietary signals: Of 11 calculus samples, 10 yielded oral proteins; among those, 8/10 (80%) showed evidence of animal product consumption (milk and/or meat/blood). Total proteins per sample were comparable to or exceeded prior Central/Inner Asian studies.
• First yak dairy identification: Yak-specific milk peptides (Bos grunniens/mutus; BLG-E variant) were identified in dental calculus from two individuals: HOR-27 (1170–1270 CE) and HOR-01 (1287–1397 CE), providing the earliest global proteomic evidence for yak milk consumption. HOR-27 preserved eight milk proteins (including BLG/BLG1, alpha-lactalbumin, lysozyme C, beta-, alpha-S1-, alpha-S2-, and kappa-caseins) and 23 taxonomically informative peptides, including one specific to Bos mutus, plus peptides assignable to Equus (n=12), Bovinae (n=2), Bovidae (n=4), and Pecora (n=4). HOR-01 had ruminant milk proteins with peptides specific to Bos grunniens/mutus (n=1) and broader groups (Bos, Bovinae, Pecora) and two ruminant-specific blood proteins.
• Diversity of dairy proteins: Across individuals, multiple milk proteins were detected: BLG (including yak BLG-E), caseins (beta-, alpha-S1-, alpha-S2-, kappa-), alpha-lactalbumin, and lysozyme C. Several individuals (HOR-40, HOR-13, HOR-21, HOR-25) had equine milk proteins (BLG1, lysozyme C, alpha-lactalbumin).
• First archaeological identification of horse caseins: In HOR-08, equine-specific alpha-S2- and beta-caseins were identified (first recovery of horse caseins from archaeological samples). HOR-08 also had BLG and alpha-S1-casein assigned to Ovis/Caprinae.
• Blood/tissue consumption evidence: Hemoglobin subunit beta peptides specific to Pecora/Bovidae were recovered, indicating ruminant meat/blood intake. HOR-30 exhibited three peptides of Equus serum albumin (a tissue protein), suggesting ingestion of horse-derived tissues/fluids.
• Preservation context: Exceptional permafrost conditions at Khorig enabled the recovery of rarely preserved proteins (e.g., caseins, blood/tissue proteins) and a broad proteomic diversity.

The findings directly address the question of whether yak milk was consumed in medieval Mongolia by demonstrating yak-specific BLG-E peptides in two elite individuals dated ca. 1170–1397 CE. This constitutes the earliest global proteomic evidence for yak dairy consumption and suggests that yak products were part of elite diets in Mongolia’s mountainous west during the Mongol period. The presence of diverse dairy proteins and blood/tissue proteins reflects both dietary breadth (ruminant and equine products) and exceptional biomolecular preservation. The identification of horse caseins expands the known repertoire of equine dairy markers in archaeological contexts. These results reinforce historical narratives that communities at the empire’s frontiers controlled production and exchange of high-altitude yak commodities (hair, felt, dairy), moving them into Mongol urban centers. However, given the taxonomic challenges of distinguishing yak from other Bos taxa via BLG (reliance on BLG-E’s final tryptic peptide) and the limited sample, the broader chronology and prevalence of yak dairying remain to be clarified. The permafrost context underscores both the scientific potential and the vulnerability of such sites to climate change, with implications for cultural heritage management.

This study provides the first proteomic identification of yak dairy consumption, from two elite individuals at Khorig in Khovsgol, Mongolia, dating to the 13th–14th centuries CE. It demonstrates that permafrost-preserved dental calculus can retain an unusually diverse suite of dietary proteins, including yak-specific BLG-E peptides, multiple caseins, and blood/tissue proteins, and it documents the first archaeological recovery of horse caseins. These results highlight yak products’ significance in Mongol-period economies and the utility of paleoproteomics for detecting yak management. Future work should expand sampling across social contexts and regions, refine yak-specific peptide reference databases and taxonomic markers beyond BLG-E, and systematically target permafrost and other high-preservation contexts to build a broader temporal and geographic picture of yak dairying and consumption.

• Taxonomic resolution: Yak milk identification relies heavily on recovery of the final tryptic peptide of BLG-E due to high BLG sequence similarity among Bos species, making identifications sensitive to peptide preservation and coverage.
• Sample scope and context: Only 11 individuals were analyzed (with 10 yielding proteins), from elite burials at a single permafrost site; results may not generalize to broader populations, periods, or regions.
• Database and search variability: Differences between database search spaces (SwissProt, custom ancient, HODM) can shift peptide spectral matches; high-scoring modern contaminants can outcompete ancient proteins in searches, potentially affecting identifications if not carefully controlled.
• Preservation/authentication: Although permafrost enhanced preservation, the proportion of proteins consistent with an authentic ancient oral profile indicates variable preservation across samples; environmental and lab contaminants must be considered.
• Broader inference: Earliest evidence at Khorig does not preclude earlier yak dairying elsewhere; absence of earlier detections may reflect identification challenges rather than true absence.