The genomic origins of the Bronze Age Tarim Basin mummies

Abstract

The identity of the earliest inhabitants of Xinjiang, in the heart of Inner Asia, and the languages that they spoke have long been debated and remain contentious¹. Here we present genomic data from 5 individuals dating to around 3000-2800 BC from the Dzungarian Basin and 13 individuals dating to around 2100-1700 BC from the Tarim Basin, representing the earliest yet discovered human remains from North and South Xinjiang, respectively. We find that the Early Bronze Age Dzungarian individuals exhibit a predominantly Afanasievo ancestry with an additional local contribution, and the Early-Middle Bronze Age Tarim individuals contain only a local ancestry. The Tarim individuals from the site of Xiaohe further exhibit strong evidence of milk proteins in their dental calculus, indicating a reliance on dairy pastoralism at the site since its founding. Our results do not support previous hypotheses for the origin of the Tarim mummies, who were argued to be Proto-Tocharian-speaking pastoralists descended from the Afanasievo^1,2 or to have originated among the Bactria-Margiana Archaeological Complex³ or Inner Asian Mountain Corridor cultures⁴. Instead, although Tocharian may have been plausibly introduced to the Dzungarian Basin by Afanasievo migrants during the Early Bronze Age, we find that the earliest Tarim Basin cultures appear to have arisen from a genetically isolated local population that adopted neighbouring pastoralist and agriculturalist practices, which allowed them to settle and thrive along the shifting riverine oases of the Taklamakan Desert.

Fig. 1. Overview of the Xinjiang Bronze Age archaeological sites analysed in this study.

a, Overview of key Eurasian geographic regions, features and archaeological sites discussed in the text; new sites analysed in this study are shown in grey. b, Enhanced view of Xinjiang and the six new sites analysed in this study. c, Timeline of the sites in a. The timeline is organized by region, and the median date for each studied group is shown. The base maps in a and b were obtained from the Natural Earth public domain map dataset (https://www.naturalearthdata.com/downloads/10m-raster-data/10m-cross-blend-hypso/). In the group labels, the suffixes represent the archaeological time periods of each group: N, Neolithic; EN, MN and LN, Early, Middle and Late Neolithic, respectively; EN, Eneolithic for Geoksyur, Parkhai and Sarazm; CA, Chalcolithic Age; BA, Bronze Age; MBA, Middle Bronze Age; EIA, Early Iron Age. MA-1, Mal'ta; EHG, Eastern European hunter-gatherers.

Fig. 2. Genetic structure of ancient and present-day populations included in this study.

Principal component analysis of ancient individuals projected onto Eurasian and Native American populations; the inset displays ancient individuals projected onto only Eurasian populations.

Fig. 3. Genetic ancestry and admixture dating of ancient populations from Xinjiang and its vicinity.

a, qpAdm-based estimates of the ancestry proportion of Dzungaria_EBA and Tarim_EMBA from three ancestry sources (AG3, Afanasievo and Baikal_EBA) (Supplementary Data 1D, E). Unlike Dzungaria_EBA individuals, Tarim_EMBA individuals are adequately modelled without EBA Eurasian steppe pastoralist (for example, Afanasievo) ancestry. b, Genetic admixture dates for key Bronze Age populations in Inner Asia, including Dzungaria_EBA1 (n = 3), Chemurchek (n = 3), Kumsay_EBA (n = 4), Mereke_MBA (n = 2), Dali_EBA (n = 1) and Tarim_EMBA1 (n = 12). The blue shade represents the radiocarbon dating range of the Yamnaya and Afanasievo individuals. The orange circles and the associated vertical bars represent the averages and standard deviations of median radiocarbon dates, respectively. The circles above each orange circle represent the estimated admixture dates with a generation time of 29 years, and the vertical bars represent the sum of standard errors of the admixture date and the radiocarbon date estimate. c, Representative qpAdm-based admixture models of ancient Eurasian groups (Supplementary Data 1D–I). For Dzungaria_EBA1 and Geoksyur_EN, we show their three-way admixture models including Tarim_EMBA1 as a source. For later populations in Xinjiang, IAMC and nearby regions, we used them as sources, and allocated a colour to each of them (blue for Dzungaria_EBA1; magenta for Geoksyur_EN). The base map in c was obtained from the Natural Earth public domain map dataset (https://www.naturalearthdata.com/downloads/10m-raster-data/10m-gray-earth/).

Extended Data Fig. 1. Burial goods excavated from the Xiaohe cemetery.

A, a wooden sculpture excavated from the upper layer of a double-layer mud coffin of XHM75. B, an oar-plank placed in front of a male burial. C, a wooden pole placed in front of a female burial. D, Burial XHM66 from layer 4 of the Xiaohe cemetery illustrating typical features of early burials, including boat-shaped coffins and mummified remains dressed in woollen garments. This burial style is common at Bronze Age cemeteries throughout the Tarim Basin, including Beifang and Gumugou. E, Side view of the Xiaohe cemetery showing wooden grave markers and fencing.

Extended Data Fig. 2. F-statistics for the ancient Xinjiang and the Eurasian steppe populations.

A, we show top 5 outgroup f3-statistics of the form f3(Target, X; Mbuti) for the 361 world-wide populations as contrast populations X, and 8 populations from this study and the Eurasian Steppe as target: Dzungaria_EBA1, Dzungaria_EBA2, Chemurchek, Dzungaria_EIA, Okunevo_EMBA, Kazakhstan_EMBA, Botai_CA, West_Siberia_N, horizontal bars represent ± 1 standard error measure (s.e.m.) calculated by 5 cM block jackknifing. B, f4-statistics of the form f4(Mbuti, X; Dzungaria_EBA1, Tarim_EMBA1), horizontal bars represent ± 3 (thin) and ± 1 (thick) s.e.m. calculated by 5 cM block jackknifing, and C, f4-statistics of the form f4(Mbuti, X; Dzungaria_EBA2, Tarim_EMBA1), where X is 361 world-wide populations. We show the top and the bottom 15 f4 statistics. Horizonal bars represent the point estimate ± 3 (thin) and ± 1 (thick) s.e.m., respectively, as estimated using 5 cM block jackknifing. F4 statistics deviating three s.e.m. or more from zero are marked in red.

Extended Data Fig. 3. Unsupervised ADMIXTURE plot for the Bronze Age Xinjiang individuals.

We plot ancestry component estimates for K = 8 using ‘AncestryPainter’ (https://www.picb.ac.cn/PGG/resource.php). Dzungaria_EBA individuals show an ancestry pattern close to Afanasievo and Yamnaya, while Tarim_EMBA individuals show a pattern similar to AG3, West_Siberia_N and Botai_CA from the Eurasia steppe.

Extended Data Fig. 4. Reduced genetic diversity of the Tarim_EMBA individuals.

A, a comparison of individual outgroup f3-statistics for the ancient Xinjiang populations and their neighboring populations from Inner Asia, including Tarim_EMBA1 (n = 12), Tarim_EMBA2 (n = 1), ANE (n = 3), Dzungaria_EBA1 (n = 3), Dzungaria_EBA2 (n = 2), West_Siberia_N (n = 3) and Botai_CA (n = 3), which Tarim Basin individuals show the highest affinity to each other. In each boxplot, the box marks the 25^th and 75^th quartiles of the distribution, respectively, and the horizontal line within the box marks the median. The whisker delineates the maximum and the minimum. B, the cumulative distribution of ROH tracts shows that Tarim_EMBA individuals did not descend from close related parents. C, pairwise mismatch rate (pmr) between individuals in the ancient populations of Xinjiang and its neighboring regions, including all pairs of individuals within the Afanasievo (n = 27), ANE (n = 3), Baikal_EBA (n = 9), Baikal_EN (n = 15), Botai_CA (n = 3), Dzungaria_EBA (n = 5), Dzungaria_EIA (n = 10), Sintashta_MLBA (n = 51), Tarim_EMBA (n = 13), West_Siberia_N (n = 3), as well as present-day isolated populations such as Papuan and Karitiana. Tarim_EMBA individuals uniformly show a much reduced pmr value that is equivalent to the first-degree relatives in Afanasievo or Sintashta_MLBA. The red dotted lines mark the expected pmr value for the given coefficient of relationship (r), ranging from 0 (unrelated) and 1/4 (second degree relatives) to 1/2 (first degree relatives), based on the mean value of pmr among these populations, respectively. In each box plot, the box represents the interquartile range (the 25^th and 75^th quartiles), and the horizon line within the box represents the median. Black-filled and open circles represent outliers (1.5 times beyond the IQR) and extreme outliers (3 times beyond the IQR), respectively. The whisker delineates the smallest and the largest non-outlier observations. D, Y chromosome phylogeny of the Bronze Age Xinjiang male individuals. Xiaohe male individuals fall into a branch distinct from western Bronze Age steppe pastoralists, such as Afanasievo and Yamnaya. One individual from Beifang falls in a position that is more basal than Xiaohe, but its phylogenetic position cannot be fixed due to low coverage, and its proximate position(s) are instead indicated with an asterisk.

Extended Data Fig. 5. Proteomic evidence for dairy consumption in Xiaohe dental calculus, ca. 2000-1800 BCE.

A, B- and Y-ion series for the frequently observed β-lactoglobulin peptide TPEVD(D/N/K)EALEK, which contains a taxon-specific polymorphic residue: D, Bovinae; N, Ovis; K, Capra. See SI Appendix. B, Taxonomically assigned β-lactoglobulin (black), α-S1-casein (dark grey), and α-lactalbumin peptide spectral matches (PSMs) presented as scaled pie charts on a cladogram of dairy livestock. Bracketed numbers represent the number of PSMs (excluding duplicates) assigned to each node. †Included on the Bovidae node are: 13 PSMs assigned to Bovidae; 21 PSMs assigned to Bovidae but excluding Capra.