Bronze and Iron Age population movements underlie Xinjiang population history:
The Xinjiang region in northwest China is a historically important geographical passage between East and West Eurasia. By sequencing 201 ancient genomes from 39 archaeological sites, we clarify the complex demographic history of this region. Bronze Age Xinjiang populations are characterized by four major ancestries related to Early Bronze Age cultures from the central and eastern Steppe, Central Asian, and Tarim Basin regions. Admixtures between Middle and Late Bronze Age Steppe cultures continued during the Late Bronze and Iron Ages, along with an inflow of East and Central Asian ancestry. Historical era populations show similar admixed and diverse ancestries as those of present-day Xinjiang populations. These results document the influence that East and West Eurasian populations have had over time in the different regions of Xinjiang.
This is a hard paper to read. The terminology could be clearer, and the narrative crisper. Perhaps this is just due to the need to compress into the length science wanted, I don’t know. But I find it hard to read, I doubt others will find it easier.
Here’s the main figure:
Here are my general summaries
1 – ~5,000 years ago the indigenous people of “Xinjiang” (northern half = Dzungaria, southern half = Uighurstan) were mostly distantly related to “Ancient North Eurasians” (the Tarim Basin mummies)
2 – Around this time early Indo-Europeans, the Afanasievo culture, begin to migrate south of the Altai pasturelands to the north. The Afanasievo are basically genetically 100% Yamnaya initially; their culture was really just an eastward migration of Yamnaya. They immediately start mixing with the indigenous ANE-related population, and also to some extent with small numbers of people from Northeast Asia.
3 – After 2000 BC two new entrants arrive on the scene. First, Indo-Iranians, second-wave Indo-Europeans who are different from the Afanesievo in having “Anatolian farmer” ancestry due to their provenance in the borderlands of Poland, Belarus and Ukraine. The Indo-Iranians, part of the Andronovo horizon, assimilate with the local populations, and almost certainly are responsible for some of the Iranian languages prominent in early historical records in the southwestern part of the Tarim Basin (e.g., Khotanese). Ad this time there is also a clear influx of Central Asian/Turanian ancestry related to the BMAC culture. These are basically pre-Indo-European Iranians, and you can think of them as strongly skewed to Zagrosian ancestry, along with some ANE/WSHG admixture, and minimal but nontrivial Anatolian farmer (the ancestors of the Indus Periphery population are zeroed out for Anatolian farmer).
If you look at admixture analyses of modern Uyghurs you see some Persian-related ancestry. This may in fact be Iranians from the medieval period, but perhaps more likely they’re due to the mixture between Andronovo and BMAC ancestry, which would perfectly mimic Persian ancestry.
At least period you also start seeing more East Asian ancestry with a southern affinity.
4 – By the historical period, you have the migration of populations from Central Asia, Siberia, and East Asia. Some of these you recognize. The Sakas and the Old Uyghurs for example. Others, you may not.
Looking at the admixture, one of the historical era samples is clearly Northwest Indian in ancestry. There is some evidence of Indian ancestry as early as the Bronze Age as well. This is a small component, but it is clearly not trivial. This particular result puts Kumarajiva in perspective.
The major finding of this paper seems to be that Xinjiang has not seen population replacement, but assimilation. The indigenous Tarim Basin ANE-derived population still make a substantial contribution to modern Uyghurs, as do the Afanasievo early Indo-Europeans, later Indo-Iranians, Xiongnu, Han Chinese and Central Asians.
Finally, it does look like the admixture between Afanasievo and ANE-derived indigenous populations dates to between 2500 and 3000 BC. Judging from the distinctiveness of Tocharian languages the authors claim that this means that it seems likely Tocharian is an Afanesievo language (and if that is true, then the original Yamnaya spoke something similar to Tocharian).
To try and provide a sort of companion, I’ve made a y-dna plot again time (CE) for the new samples here, using mean CE: https://imgur.com/a/JawUG9j
In terms of the phases:
1) From 3000 BCE – 2000 BCE, we see the Afanasievo and then the Chemurchek Culture that is largely are slightly later than Afanasievo (and don’t have any admixed Yamnaya ancestry individuals, unlike the Afanasievo group here which has 1 of 6) and unclassified BA samples, who are largely R1b1a (M269) with some Q2a individuals. This is not the Q1b that is seems more found in the Western Steppe as early as the Khvalynsk cemetary and is found among a large extended family of Afanasievo people from the Yenisei. However at this time there are some few males who have R1a2 (an R1a clade basal to the starburst clade), a Northeast Asian C2a1a (together with unadmixed NE Asian ancestry, as was previously found at an early Afanasievo site in Mongolia) although this very early on in the sequence, and there is a Q1b.
2) In the Middle to Late Bronze Age, from 2000 BCE to 1000 BCE, sampling is sparse, but we see our first R1a1a samples around 1500 BCE, together with another Q1b.
3) Iron Age from 750 BCE to 250 BCE, where the samples are most dense. R1a1a at this time is the most prominent single haplogroup, but does not appear to constitute a majority of samples. Note that Q1b remains strong in this era, while we see a lot more appearance of western / Iran+Caucasus related haplogroups, and our first appearance of East Asian O2a2b (from typically further south and east in East Asia than previously found East Asian haplogroups). Also one South Asian related L1a2 has appeared. R1b-M269 is still sporadically present at this time, though a different clade of R1b which is very minor in West Eurasia seems more common.
(As a quick check, if I look at the 84 males who I’ve labelled following the supplement as Iron Age, Scythian, Yuezhi or Wusan, Suodunbulake-related, Nomad, Zaghunluq, Sampula, and who all average around 270 BCE, then R1a1a at this time constitute 31 of these samples. So at this time already R1a1a is probably around a third, and the largest plurality but not a majority? There were a few samples at this time who were not called under the Y-Leaf data I chose to use from the supplement, but these were considered to be haplogroup N under the other methods. This may suggest less R1a1a than MLBA steppe ancestry at this time, following the qpAdm proportions in graphical abstract).
4) Later historical era are quite thin on the ground, so hard to speculate on.
I think this all does suggest quite the interesting picture where we probably would say that early BMAC related groups and their successors were quite likely to be trading through Central Asia into Xinjiang… but they perhaps didn’t settle there much, all through the Bronze Age! That makes sense though; it’s not the most promising place to seek a wife or subsistence. Likewise for the sheep-herder pastoralists from the Southwest along the IAMC, these probably were not the most desirable of locations? It’s when the Iron Age probably establishes more settlements that we then see proper influence from the Southwest in robust quantity. Albeit, it does look like even at this time, the non-R1a1a haplogroups seems more rooted in Central Asia than the Southwest (which are still relatively minor).
You might need to have a proper look at the structure under Q1b, more than I’ve given it, to speculate where it is coming from, whether it is more East Asian / more ANE / more EHG admixed groups.
“and if that is true, then the original Yamnaya spoke something similar to Tocharian”
More properly “similar to ancient Tocharian” since the Tocharian fragments that we have come from much later time (open to being corrected)
Off topic but recently published and may be of interest –
Paper: https://assets.researchsquare.com/files/rs-604881/v1_covered.pdf?c=1631871420 – “Genetic mechanisms underlying East Asian and European Facial differentiation”
Facial morphology, the most conspicuous feature of human appearance, is highly heritable. Previous studies on the genetic basis of facial morphology were mainly performed in European populations. Applying a proven data-driven phenotyping and multivariate genome-wide scanning protocol to the largest collection of 3D facial images of an East Asian population to date, we identified 244 leading variants associated with normal-range facial variation, of which 130 are novel. A newly proposed polygenic shape analysis indicates that the effects of the variants on East Asian facial shape can be generalized into the European population. Based on this analysis, we further identified 13 variants mainly related to differences between European and East Asian facial shape. Natural selection analyses suggest that the difference in European and East Asian nose shape is caused by a directional selection, mainly due to a local adaptation in Europeans. Our results expand the knowledge of human facial genetics and illustrates for the first time the underlying genetic basis for facial differences across populations.
The paper got uploaded to that preprint server ages ago, but as it got some press coverage on the responsible university’s site ((https://www.fudan.edu.cn/en/2022/0409/c344a130843/page.htm ) and recently published in Nature I thought it was worth mentioning.
Possibly this can be applied to some of these ancient genomes to predict facial shape.
It’s interesting to see the suggestion that European adaptation drives the difference (based on a tree branch model with Yoruba as root; Population Branch Statistic). It’s often thought to be the case that East Asian face morphology is what is particularly adapted, and derived, from a more generalized early OoA face shape. Quoting the university’s site (“Based on various natural selection analyses, the study genetically explored selected facial regions and revealed that nose differences are likely due to adaptive selection in Europe, but genetic drift in East Asia.”).
On the other hand, they note that they lack the power to detect variants that are near fixed in East Asians, or which are fixed in Europeans and East Asians, so this may change if more variants are picked up by a similar analysis in Europeans or in admixed EUR-EAS.