Bronze Age population dynamics and the rise of dairy pastoralism on the eastern Eurasian steppe

Abstract

Recent paleogenomic studies have shown that migrations of Western steppe herders (WSH) beginning in the Eneolithic (ca. 3300-2700 BCE) profoundly transformed the genes and cultures of Europe and central Asia. Compared with Europe, however, the eastern extent of this WSH expansion is not well defined. Here we present genomic and proteomic data from 22 directly dated Late Bronze Age burials putatively associated with early pastoralism in northern Mongolia (ca. 1380-975 BCE). Genome-wide analysis reveals that they are largely descended from a population represented by Early Bronze Age hunter-gatherers in the Baikal region, with only a limited contribution (∼7%) of WSH ancestry. At the same time, however, mass spectrometry analysis of dental calculus provides direct protein evidence of bovine, sheep, and goat milk consumption in seven of nine individuals. No individuals showed molecular evidence of lactase persistence, and only one individual exhibited evidence of >10% WSH ancestry, despite the presence of WSH populations in the nearby Altai-Sayan region for more than a millennium. Unlike the spread of Neolithic farming in Europe and the expansion of Bronze Age pastoralism on the Western steppe, our results indicate that ruminant dairy pastoralism was adopted on the Eastern steppe by local hunter-gatherers through a process of cultural transmission and minimal genetic exchange with outside groups.

Keywords: LC-MS/MS; dental calculus; paleogenomics; α-S1-casein; β-lactoglobulin.

Fig. 1.

Map of the Eurasian steppes. (A) Distribution of the Western (brown) and Eastern (green) steppes and the locations of ancient (red) and modern (black) populations discussed in the text. Population codes are provided in the Dataset S1. A box indicates the location of the LBA burial mounds surveyed in the Arbulag soum of Khövsgöl aimag. (B) Enhanced view of LBA burial mounds (white circles) and burial clusters selected for excavation (boxes a–f) with the number of analyzed individuals in parentheses (SI Appendix, Table S1). (C) Photograph of burial 2009-52 containing the remains of ARS026, a genetic outlier with Western steppe ancestry.

Fig. 2.

The genetic profile of LBA Khövsgöl individuals summarized by PCA and ADMIXTURE. (A) Khövsgöl (Kvs, ARS017, and ARS026) and other ancient individuals (colored symbols) are projected onto the top PCs of modern Eurasian and Native American individuals. Contemporary individuals are marked by gray circles. Mean coordinates for each of the contemporary populations are marked by three-letter codes and by colors assigned to the associated geographic regions. Population codes are provided in Dataset S1 and SI Appendix, Fig. S8. (B) ADMIXTURE results for Khövsgöl and other ancient individuals with K values 9 and 17. In K = 17, the Khövsgöls main cluster is mainly modeled as a mixture of components most enriched in modern northeast Asians (e.g., Nivh) and ancient Siberians (e.g., AG3, Botai, and Okunevo).

Fig. 3.

The genetic affinity of the Khövsgöl clusters measured by outgroup-f₃ and -f₄ statistics. (A) The top 20 populations sharing the highest amount of genetic drift with the Khövsgöl main cluster measured by f₃(Mbuti; Khövsgöl, X). (B) The top 15 populations with the most extra affinity with each of the three Khövsgöl clusters in contrast to Tuvinian (for the main cluster) or to the main cluster (for the two outliers), measured by f₄(Mbuti, X; Tuvinian/Khövsgöl, Khövsgöl/ARS017/ARS026). Ancient and contemporary groups are marked by squares and circles, respectively. Darker shades represent a larger f₄ statistic. Population codes are provided in Dataset S1; see also SI Appendix, Figs. S11–S14 for further details.

Fig. 4.

Admixture modeling of Altai populations and the Khövsgöl main cluster using qpAdm. For the archaeological populations, (A) Shamanka_EBA and (B and C) Khövsgöl, each colored block represents the proportion of ancestry derived from a corresponding ancestry source in the legend. Error bars show 1 SE. (A) Shamanka_EBA is modeled as a mixture of Shamanka_EN and AG3. The Khövsgöl main cluster is modeled as (B) a two-way admixture of Shamanka_EBA+Sintashta and (C) a three-way admixture Shamanka_EN+AG3+Sintashta. Details of the admixture models are provided in SI Appendix, Tables S5 and S6.

Fig. 5.

Presence of ruminant β-lactoglobulin and α-S1-casein milk protein in LBA Khövsgöl dental calculus. (A) B- and Y-ion series for one of the most frequently observed β-lactoglobulin peptides, TPEVD(D/N/K)EALEKFDK, which contains a genus-specific polymorphic residue: D, Bos; N, Ovis; K, Capra. See SI Appendix, Fig. S16 for peptide and fragment ion error distribution graphs. (B) Alignment of observed peptides to the 178 amino acid β-lactoglobulin protein, with peptide taxonomic source indicated by color. Trypsin cut sites are indicated by gray ticks. The position and empirically determined observation frequency of BLG peptides for bovine milk are shown as a heatmap scaled from least observed peptides (light gray) to most frequently observed peptides (dark red), as reported in the Bovine PeptideAtlas (34). Inset displays a 3D model of the β-lactoglobulin protein with observed peptide positions highlighted in black. (C) Taxonomically assigned β-lactoglobulin (black) and α-S1-casein (gray) peptides presented as scaled pie charts on a cladogram of Mongolian dairy domesticates. Bracketed numbers represent the number of peptides assigned to each node. Ruminant milk proteins were well supported, but no cervid, camelid, or equid milk proteins were identified.