Discussion thread: New ancient DNA from Xinjiang (2022)

Bronze and Iron Age population movements underlie Xinjiang population history

Vikas Kumar, Wenjung Wang , Jie Zhang, Yongqiang Wang, Qiurong Ruan, Jianjun Yuxiao, Hong Wu, Xingjun Huxin, Hua Wu, Qiaomei Fu +17 authors Authors Info & Affiliations

SCIENCE • 31 Mar 2022 • Vol 376, Issue 6588 • pp. 62-69 • DOI: 10.1126/science.abk1534

Abstract:

5000 years of Xinjiang genetics

The Xinjiang region of China is bordered by mountains and represents an important historical region. Sampling ancient genomes, Kumar et al. investigated the changes in populations of this region over time from the Bronze Age, ~5000 to 3000 years before the present (BP), covering the Iron Age, ~3000 to 2000 years BP, and into the Historical Era, ~2000 years BP. This analysis identified that older individuals represented ancestries from Steppe cultures, and that a later inflow of East and Central Asian ancestry entered the region around the end of the Bronze Age toward the beginning of the Iron Age. During the Historical Era, mixing continued but retained a core Steppe component such that populations form a genetic continuum. This retention of genetic continuity in a central population is surprising because it represents patterns more typically observed in isolated populations. Furthermore, these genetic links identify a previously unknown lineage that could potentially explain the spread of the Indo-European languages.

My prayers have been answered, Dyeus bless! This will undoubtedly be one of my favourite articles to have come out this year. A big shoutout to the collaborative efforts of all the authors involved, and thank you for providing this data to us.

I was currently working on a rather extensive blog entry about the western tarim basin during the bronze and iron ages, and what I think the most likely origin for the khotanese language would be’. Many of the samples in this article actually come from sites I was discussing in that blog entry such as suodonbuluke sites, jirzankal and Sampula. The entry was what I would say 95% finished and sits at 45 pages and 10000 words, but I will have to postpone it a little bit in order to properly cover the data coming by way of genetics, because it was first only done on an archaeological basis. So far from what I’ve seen, what I wrote aligns perfectly with the results presented here.

They did address the topic of the Khotanese language in the article, this quote sums it up:

Further, although the spread of languages is not always congruent with population histories (32), the presence of Saka ancestry in Xinj_IA populations supports an IA introduction of the Indo-Iranian Khotanese language, which was spoken by the Saka and later attested to in this region (19).

This is something I would disagree with. I think that these findings of the article would actually be more relevant to the Khotanese language:

The IA also shows an increase in the frequency of BMAC ancestry in the f4-statistics comparisons (figs. S19 and S20) (21). Seven IA populations were found to contain BMAC ancestry (30 to 47%), and we observed four IA populations that could also be modeled using Indus periphery ancestry sources SPGT and two with Gonur_2BA (18 to 37%) (Fig. 3A and tables S9 and S10). The increase in the appearance of BMAC ancestry suggests a substantial movement of people from either BMAC- or Indus periphery–derived populations into the Xinjiang region during the IA (Fig. 3A), most likely through the IAMC route over the Pamir and Tianshan Mountains.

But I’ll explain my position in due time. In the meantime you could read the article Tocharian B etswe ‘mule’ and Eastern East Iranian by M. Peyrot, because the section on the origin of Khotanese allings quite closely with my view on the matter. 

Back to the article, here you can see the layout of the sampling locations:

Fig. 1. Xinjiang sampling locations along with their groupings used in this study. (A) Map showing the geographic sampling locations of the archaeological sites included in this study. (B) BA, IA, and HE time periods are depicted on a timeline in years before the present (BP). The number of individuals from each site and time period is given in parentheses. Abbreviations are defined in table S1 and the supplementary materials.

As well their principal component analysis (PCA) and charts containing ADMIXTURE and Qpadm Analyses: 

Fig. 2. PCA and ADMIXTURE analyses of Xinjiang populations. (A) Ancient Xinjiang and other ancient populations are shown as different colors and shapes.Present-day populations are shown as gray circles, and only major groups are included. Most of the ancient Xinjiang populations lie on the cline extending from European and Siberian to East Asian populations. The published populations of Dzungaria_EBA, Tarim_EMBA, and Shirenzigou_IA are depicted in various black shapes. IA north (IA_N), south (IA_S), west (IA_W), and east (IA_E) are the geographical locations of the IA individuals. (B) ADMIXTURE analysis of all the newly reported ancient individuals at K = 7. The four major components are maximized in the following populations: ANE (green), Iranian farmer (red), Anatolian farmer (violet), and East Asian hunter gatherer (yellow). The other three are maximized in Han (orange), Mixe (cyan), and Papuan (dark blue). Supplementary figures of PCA and ADMIXTURE show all the present-day and ancient populations (21).

Fig. 3. Inferred qpAdm models and summary of ancient Xinjiang with population movements. (A) The proximal qpAdm admixture proportions for all Xinjiang populations. Each bar represents admixture proportion of the listed subgroups for BA, LBA, IA, and HE populations. Subgroup details are provided in table S1 and the supplementary text, and qpAdm modeling results are provided in detail in tables S5 to S13 (21). The Swat Valley Protohistoric Grave Type IA populations are in the SPGT group, and the Yellow River basin Middle Neolithic population is in the YR_MN group. (B) Inferred scenarios of admixtures in BA, LBA, IA, and HE Xinjiang with possible population movements shown as arrows. Xinjiang BA populations can be mostly characterized with Steppe_EMBA and Xinj_BA1_TMBA1 (Tarim Basin EMBA) ancestries with additional ancestries of Central Asia (BMAC)—as observed in Chemurcheck culture (Steppe_EMA)—and Northeast Asia (Shamanka). MLBA Xinjiang populations contain additional Andronovo Steppe, Central Asian (BMAC), and East Asian ancestries, whereas the IA and HE populations show the major Xinjiang and Steppe MLBA ancestries with additional components from BMAC and East Asian (EA) sources, shown as a pie chart summarizing the qpAdm modeling of IA and HE populations using the Xinj_LBA population (table S13 and fig. S28) (21). Where possible, coloring corresponds to ancestry in (A).

As a self-admitted Scythian nerd I am incredibly happy as we have a whole bunch of antiquity period nomadic samples from the Ili valley region, something I wasn’t expecting soon. The Suodonbuluke culture sites is a Saka material culture closely linked to the Saka around the Tian Shan mountains.  Here is a little paragraph from the section about the Saka period in the Ili valley from my upcoming and currently postponed blog entry:

The iron age burials in the upper Ili valley constitute a material culture dubbed the Suodunbulake culture (or Tiemulike culture) , and this material culture covers the entirety of the first millennium before the common era. These archaeologists argued that the culture can be divided in three stages, each represented by the type sites of Qiongkeke, Suodonbulake and Yeshenkelieke. The first stage represents the arrival of the nomads, the second stage the prime of the culture, and the final stage represents the decline and arrival of new nomadic entities such as the Yuezhi and Wusun.

Here are some artefacts from the Saka period of the upper Ili Valley:

There are also several samples that are labelled as “Scythian, Yuezhi and Wusun related”. It might be a small detail but I really appreciate it that they didn’t just label these as Yuezhi or Wusun in similar fashion to other articles, because it is known from historical records that all three populations were present in the region during the period of Wusun rule. It is even described as such in Volume 96 of the Hanshu. There are also samples from Pazyryk-related sites of the southern Altai.

Unfortunately samples from the eastern half of the Tarim Basin, a region associated with Tocharian speakers, were not that prevalent in this article. Perhaps the two samples from the Xianshuiquangucheng (XSQG) and the two from Baiyanghe (BYH) could be linked to the Tocharians. On the map Xikakandasayi (XKKD) is shown in the same location as XSQC but according to the supplementary the site is on the southern rim of the Tarim Basin and proximate to Tibet, but the site would fall within the assumed distribution of “Tocharian C” on the southern rim of the Tarim Basin.

Map showcasing the proposed distribution of the Tocharian languages. Source: Wikimedia

However these are only a handful of genomes, and these regions were on the silk road. The Tocharian speaking cities were frequented by many foreigners, and in earlier times city states such as Loulan were allied with the Xiongnu empire. Thus it is not certain that these samples would have been Tocharian speakers or be genetically representative of Tocharian peoples. It may also be that some of the samples from different areas actually were Tocharians buried.

The Swordbearers of Kizil, likely representing Tocharian noblemen.

What is fascinating however is that we have samples from the Zaghunluq cemetery, an iron age cemetery in the southern Tarim Basin that contains several hundred tombs. You might be familiar with some of the people buried here, because one of the most famous Tarim mummies, the Cherchen man, was from the Zaghunluq cemetery.

The Cherchen Man in all his glory

This burial site is one of the reasons why I developed such a fascination for the peoples and societies of the Taklamakan desert, and will always remain special to me.This article provided no less than seventeen sequenced genomes from this cemetery, which is amazing. Unfortunately it isn’t clear which individuals these samples were taken from, and if any belonged to the famous mummies of Cherchen. Here are some images of the mummies from Zaghunluq:

The samples date to 2491-1889 BP, which to my understanding would be a bit younger than the proposed dates for the burial of Cherchen Man, which is generally placed somewhere in between 1000 and 600 BC.  Eleven out of the sixteen samples were males, C3654 had Y-chromosome haplogroup O1a, C3674 and C837 had Y-chromosome haplogroup R1b1a and C840 had no further assignment beyond R1. The vast majority (7) of the individuals from Zaghunluq carried Y-chromosome haplogroup R1b2 or R-PH155. This haplogroup is a rather interesting one that has been seen in samples coming from Shirenzigou, The Xiongnu period, the Hunnic periods of Central Asia and Europe respectively. Last year we received genetic data on the earliest Tarim mummies, derived from a relict population of predominant Ancient North Eurasian origin, and all of their male paternal lineages were R1b-PH155. If you want to read more check out this older entry of mine.

Now given that Zaghunluq likely was a cemetery used by a local community over a period of centuries, it is likely that many of the people buried here shared patrilineal kinship with one another. Thus the high amount of R-PH155 at this site might not have been reflective of the rates of R-PH200 of their society as a whole. The idea that there is more than a thousand years of paternal continuity between some of the bronze age and some of the iron age mummies, despite all the population influxes that happened in between is rather amazing though.

C3674 Is listed as having R1b1a1b1a1a1c2b2b1a2 which I think translates to R-S21728, which I doubt is correct. The other call is just R1 so I think something is going on here. The other with R1b1a was C837 but there was no further designation beyond that clade. I’d say it would be more likely that if they both had R1b1a, it would have been under R-Z2103.

Although we don’t know the Y-DNA haplogroup of Cherchen man, I think R1b would be nearly certain. But which one? R-PH155 shows up in the highest frequency amongst these samples, making it likely that Cherchen man also would've belonged to this lineage. But it is also possible he was of the minority R1b1a or O1a lineages present here, or that he had a paternal origin unrelated to the other individuals at Cherchen, being perhaps R1a or Q.

Another debate surrounding Zaghunluq would be the ethnolinguistic identity of the people there. It is still rather unsolved if they would have been Khotanes speaking, Tocharian speaking, or speakers of a different language, whether Indo-European or not. Some of the materials at the site connect it to the peoples of the west, but the location of the burial would match up closer with “Tocharian C” speaking people than Khotanese speaking people. I think the autosomal ancestry of these samples could help out immensely here.

However we have far more samples to discuss than the Zaghunluq samples of course. I’m interested to have a closer look at the samples from Jirzankal, an iron age archaeological site near modern day Tashkurgan that may have been an ancient Zororastrian religious site based on the fire altars and burials that have been interpreted as exposure burials. Many of the individuals at the site weren’t local to the area, and may have been religious pilgrims. Jirzankal also is one of the oldest sites to have cannabis with elevated THC contents, suggesting selection for psychoactive purposes according to this article. 

What a place to spark one up eh? Here is one of the fire altars from the site:

There are also the northern samples dated to the Afanasievo and Chemurchek periods to check out, the Andronovo samples of the later bronze age, the many iron age samples across Xinjiang, and the samples from the antiquity and medieval periods.

I guess I will treat this thread as my spot to dump data and speculate about the samples and I would like to invite my fellow readers to partake. So definitely return to this thread and check out the comment section every now and then, because the real fun stuff hasn’t even started yet!