The Arctic home of the Aryans

By Razib Khan 43 Comments

The Fatyanovo culture flourished between 2800 and 1900 BC. It seems they were part of a Central European “reflux” migration. That is, their forebears were related Yamna agro-pastoralists who migrated west out of the steppe and mixed with Central European farmers. Eventually, some of these people moved back east along the edge of the forest-steppe boundary.

The Fatyanovo is the name for a group of people who seem to have introduced agro-pastoralism to the region nearly up to the Urals in northeastern European Russia. A new preprint, Genetic ancestry changes in Stone to Bronze Age transition in the East European plain, confirms what we assumed:

Transition from the Stone to the Bronze Age in Central and Western Europe was a period of major population movements originating from the Ponto-Caspian Steppe. Here, we report new genome-wide sequence data from 28 individuals from the territory north of this source area – from the under-studied Western part of present-day Russia, including Stone Age hunter-gatherers (10,800-4,250 cal BC) and Bronze Age farmers from the Corded Ware complex called Fatyanovo Culture (2,900-2,050 cal BC). We show that Eastern hunter-gatherer ancestry was present in Northwestern Russia already from around 10,000 BC. Furthermore, we see a clear change in ancestry with the arrival of farming – the Fatyanovo Culture individuals were genetically similar to other Corded Ware cultures, carrying a mixture of Steppe and European early farmer ancestry and thus likely originating from a fast migration towards the northeast from somewhere in the vicinity of modern-day Ukraine, which is the closest area where these ancestries coexisted from around 3,000 BC.

The Fatyanovo culture seems to have given rise to the rival and later successor Abashevo culture, which flourished a bit further east (beyond the Urals in part). The Abashevo in their turn gave rise to the Sintashta culture, which flourished even further east, and somewhat south.

There are two things I want to highlight. First, the Y chromosome:

Then, we turned to the Bronze Age Fatyanovo Culture individuals and determined that their maternal (subclades of mtDNA hg U5, U4, U2e, H, T, W, J, K, I and N1a) and paternal (chrY hg R1a-M417) lineages…were ones characteristic of CWC individuals elsewhere in Europe…Interestingly, in all individuals for which the chrY hg could be determined with more depth (n=6), it was R1a2-Z93…a lineage now spread in Central and South Asia, rather than the R1a1-Z283 lineage that is common in Europe.

Here is the modern distribution of Z-93:

The reason Z283 is found where in ancient times Z93 was found is that over the past 500 years ethnic Russians have expanded eastward, retracing the biogeographic route of the earlier peoples along the forest-steppe frontier.

The steppe people seem to be highly patriarchal. Though there are some non-modal lineages, samples from a specific location are often dominated by a single haplogroup, indicative of a broader kinship-based society focused around descent from an ancestor. In contrast, the origins of females as evidenced by mtDNA, diversity seems to be rather catholic. Some of the mtDNA lineages above, and later in the Sintashta, seem to derive from farmer populations in Europe whose ultimate origins were in Anatolia.

Let me define gotra from Wikipedia:

In Hindu culture, the term gotra (Sanskrit: गोत्र) is considered to be equivalent to lineage. It broadly refers to people who are descendants in an unbroken male line from a common male ancestor or patriline. Generally the gotra forms an exogamous unit, with the marriage within the same gotra being prohibited by custom, being regarded as incestThe name of the gotra can be used as a surname, but it is different from a surname and is strictly maintained because of its importance in marriages among Hindus, especially among the higher castes

The second point is to show this table:

This group has been assembling a lot of data on phenotypic SNPs over time transects in Northeast Europe. One has to take these results with a grain of salt because the predictions are trained on modern samples. I do not think, for example, that European hunter-gatherers had “black skin.” I suspect that the Mesolithic populations were genetically different enough that their “light alleles” may not be in our panels, though my suspicion is that they’d be of darker hue as Inuit people are. That being said, selection work aligns with these results that Europeans, in particular, seem to have been getting lighter in many areas down to the present.

The eye color prediction I somewhat trust since it’s quasi-Mendelian (~75% of the variance is due to one genetic location in Europeans). For the pigmentation, I would focus on the trend, not the absolute value. Anyone who has been to the Northeast Baltic (I have) knows that these are amongst the fairest people in the world. It is very unsurprising that these people have been getting paler over time.

There have been various arguments on this blog and elsewhere as to what the Sintashta people would look like.  I’ve posted the Narasimhan et al. data before. The results are broadly similar to the ones above for the Fataynovo.

The Fataynovo do not have the pigmentation genetic architecture that is similar to Nordic people. But, neither are they out of keeping with some European peoples. The Sintashta would be ~25% blue-eyed according to Narasimhan et al.’s data. In the 1000 Genomes about 10% of the alleles in Punjabis, Gujaratis, and Bengalis is the derived variant so common in Northern Europe, giving a recessive frequency ~1% of so blue-eyed, which is too high since other genes have an influence in these cases (though this allele is found in West Asia at appreciable frequencies, including in very old ancient DNA).

On the whole, these results confirm that the Aryans when they arrived in India were fair-skinned people. But, they were likely not as rosy-cheeked as the English who arrived thousands of years later, nor were their eyes quite often pale.


43 Replies to “The Arctic home of the Aryans”

  1. “In contrast, the origins of females as evidenced by mtDNA, diversity seems to be rather catholic.”

    I presume this is “chaotic” or something because… catholic doesn’t even begin to make sense.

      1. No. catholic means conservative, stable . In there, it means maternal lineage mostly intact. This is because of wars are between men and for women

  2. I have seen people with blue,grey eyes and people who phenotypically would pass a Italian and even Russian in some cases.

    Look at the Bollywood actor : Neil Nitin Mukesh . He said that US emigration officer couldn’t believe he was Indian.

    But after all one should acknowledge that there are huge variation of phenotype within groups.

  3. I guess you were right, about Indo Iranian pigmentation. I don’t know why Davidski got different results than you for Sintashta.

  4. There are some insights from mtDNA……

    Andronovo mtDNA is pretty inbred. Fatyanovo shares quite a few specific mtDNA lineages confirming the linkage between them.

    Of course, you find these specific mtDNA lineages in modern Southcentral Asia, especially ones with high Steppe ancestry, which is a cool way to show the connection is real.

    Also, Andronovo shares a few farmer mtDNA lineages with Bell Beaker, suggesting they have common farmer ancestors not just Steppe ancestors.

    The common farmer ancestor should be Globular Amphora, who admixed with Steppe tribes (Corded Ware) beginning in 2800 BC.

  5. I’m guessing the pigmentation of Sintashta is the product of contamination. Less likely is that there was selection for light pigmentation like in Europe.

    Also I’m curious on the mental gymnastics OIT proponents are going to go through to explain this. There’s also a 10,000 year old R1a in the paper.

      1. Although this comment is a joke, I do have a serious question from it : is there evidence that Fatyanovo (or Abashevo) had the chariot yet? David Anthony knew Sintashta would invent or import it.
        Fatyanovo seems too northerly (snow, mud) and foresty for chariots to work.

  6. Why do @Razib Khan didn’t ask Kushal Mehra about OIT VS AIT. He actually has listened to both theories by Professors he has invited on his podcast.
    I don’t know which party did he belong to AIT vs OIT?

  7. So Razib – 2 questions.

    1. Was the region we now know as western russia full of r1a-z93 males and they were eventually replaced by z283 males? When did that occur – are you saying thats a more recent phenomenon?

    2. Given the phenotype discussion, we are still pretty much in line with your earlier post about the swarthy sintashta? It sounds like the sintashta looked phenotypically like pashtuns or something. i.e. varying from light brown to light skin and the occasional light eyes.

    Anyways this is all making a lot more sense. It never made sense that Indo-Aryans would phenotypically look like modern-day eastern europeans.

  8. Why do @Razib Khan didn’t ask Kushal Mehra about OIT VS AIT.

    kushal is OIT. honestly i’m not super interested in debating the topic. we’re learning a lot. more interested in learning details than convincing other people

  9. Was the region we now know as western russia full of r1a-z93 males and they were eventually replaced by z283 males?

    really eastern european russia is the better way to put it. but yes, i think the replacement did happen. the replacement of scytians/sarmatians/steppe iranian peoples with slavs.

    BUT, i think a lot happened earlier as various turkic people displaced the steppe iranian people (though lots of the turks seem to be turkicized iranians; the kirghiz have VERY HIGH % of r1a-z93, as do the altaic people)

    1. The Kyrgyz if I seem to recall have a high R1a frequency but with some of the lowest diversity indicating that its just a founder effect. I remember people saying they almost all descend from the same line with a relatively recent TMRCA. They don’t seem to be descendants of. large number of Iranians but just the result of one Turkified Iranian lineage that got really lucky for whatever reason. Which is in line with their autosomal dna being mostly East Asian like. Same for the Altai region Turks. Its only Uzbeks and Turkmen in Central Asia who have significant Iranian admixture (as in slightly less than, equal to or even exceeding their East Eurasian Turkic admixture) but even then none of them are probably more than 60% Central Asian Iranian like.

  10. two notes

    1) i said corded ware are from ‘yamna-related’ ppl cuz the latter were r1b, and cw is r1a. there is ancient unsampled structure in the proto-steppe groups. should not surprise us if they are physically varied if they mixed with local women all over

    2) look at the mtdna. some of them are unique enough you note frequencies in south asia. e.g., some south indian communities are european branch of n1a (though havik brahmins are the arab branch). w is 8% freq in parts of pakistan.

    the migration was sex-biased. but, some indo-aryan women did make to south asia…

    1. I posted this on Eurogenes too but R1a in Punjabi Jatts is ~40% and 25% in Harynavi Jatts. Given these groups are about 30-40% steppe there must be some significant maternal contribution no?

      1. Where are these figures for Jatt haplogroups from? And what are the other Y group percentages for Punjabi and Haryana Jatts?

        1. anthropogenica has lots of ppl talking about this, but look at eupedia. in any case, google scholar the literature.

          the main issue is jat’s have a lot more haplogroupd Y H than they “should” based on their autosomal.

          1. I think you mean L-M20, right? Because based on the 80 Sikh Jat and 84 Haryana Jat samples from Nagy et al 2007, the frequency of H was 1.25% in Sikh Jats and 0% in Haryana Jats. I found that strange, the sample size is not big, and it is an old study. It listed no C in Haryana Jats which is in stark contrast with the presence of both C and H in Rors from the 2018 Indus population study (Pathak I think) which had a similar sample size.

            More details about the Nagy results in this thread:

          2. @DaThang
            That study by Nagy et al (2007) is super old and using 11 Y-str markers to predict haplogroup frequency in any population is not reliable.

  11. Given these groups are about 30-40% steppe there must be some significant maternal contribution no?


    though please note mtdna may imperfect reflect that.

  12. I think you mean L-M20, right? Because based on the 80 Sikh Jat and 84 Haryana Jat samples from Nagy et al 2007, the frequency of H was 1.25% in Sikh Jats and 0% in Haryana Jats. I found that strange, the sample size is not big, and it is an old study. It listed no C in Haryana Jats which is in stark contrast with the presence of both C and H in Rors from the 2018 Indus population study (Pathak I think) which had a similar sample size.

    i guess.

    if it’s not r1a1a-z93 it’s kind of all the same to me 😉

  13. ppl really care about this. sintashta freq for blue eye allele is .42.

    here is some equivalent frequency in modern pops (with chromosome counts) btwn .5 to .3

    Bosnian 0.5 15 30
    ItalyJew 0.5 10 20
    Samaritians 0.5 3 6
    Ket 0.5 2 4
    Romanians 0.4688 15 32
    Kumyks 0.4643 13 28
    Macedonian 0.4615 12 26
    Tatar 0.4565 21 46
    Georgian 0.45 27 60
    Chechens 0.45 18 40
    Italian_S 0.45 9 20
    Ossetian 0.4444 16 36
    MoroccanJews 0.4333 13 30
    Sicily 0.4286 12 28
    Tuscan 0.4167 95 228
    LibyaJew 0.4167 5 12
    TunisiaJew 0.4167 5 12
    Bashkir 0.4091 18 44
    French_Basque 0.3958 19 48
    Balkar 0.3864 17 44
    Nogais 0.375 12 32
    SephardicJews 0.3684 14 38
    Armenians 0.3571 25 70
    Greek 0.35 14 40
    Syrians 0.3438 11 32
    Druze 0.3404 32 94
    Cypriots 0.3333 8 24
    Kumyk 0.3333 2 6
    Tuvan 0.3333 2 6
    Spaniard 0.3319 79 238
    Adygei 0.325 13 40
    AzerbaijaniJews 0.3125 5 16
    Burusho 0.3 15 50

    i think we’d say the sintashta were white. but they probably did not look northern european, though some of them did have blue eyes

    1. How do those frequencies of the derived allele translate into actual phenotype?

      Obviously 30% of the Burusho are not light eyed, neither are 35% of Greeks and Armenians or 41% of Bashkirs. And certainly not 55% of Samaritians.

      The Spaniards being 33% light eyed is the only thing that might be legit.

      1. hey, remember you need to square it. blue eyes is a recessive phenotype.

        but yeah, it’s not perfect outside of non-euro groups cuz other background genes.

        (my kids vary in eye color from medium brown to hazel with golden flecks; they’re all heterozygote on that locus)

    2. Some entries are a real joke. Who are Bosnians, Macedonians, Romanians?

    3. People don’t usually consider Libyan Jews or Armenians as ‘white’. I think you need to stop applying modern terms with socio cultural meanings to genetic data.

      They were possibly having the phenotype of Armenia/Iran/Georgia is my guess.
      A people genetically equidistant to modern Indians and Europeans.

      Also didnt blue eyes originate from the western hunter gatherers, who are actually a non west-Asian lineage (crown north eurasians, related to other hunter gatherers like AASI) that got absorbed into the predominant west Asian lineages of Eurasia?

  14. also, i think it is plausible that they mixed with ppl in eastern iran prior to arriving in subcontinent. so the steppe % is a *floor* of indo-aryan, not the ceiling

  15. Tajik have a lot of Steppe mtDNA. It is very cool to see their direct links with Andronoovo, and with these new Fatyanovo samples. They also have significant non-Steppe Y DNA.

    Pashuten don’t have a lot of Steppe mtDNA. So there could be serious gender bias admix there.

  16. Pashuten don’t have a lot of Steppe mtDNA. So there could be serious gender bias admix there.

    pashtun OCA2-herc2 is 20%. not 40%. so lots of admixture with substrate. like 50% haplgroup M on the maternal line last i checked?

      1. @Jatt_Scythian

        Looking at a few studies here, it doesn’t seem that any Pashtun group comes even close to 50% haplogroup M.

        In the one with the biggest, most diverse sample set: Pakistani Pashtuns are at 25% for South Asian mtDNA M (sampled from all across KPK + FATA, no tribal information, 240 individuals).

        In the next largest study, Pakistani Pashtuns are at 28% for maternal haplogroup M (4 tribes were sampled, 100 individuals).

        In another, a Pakistani Pashtun tribe has only 11% haplogroup M (one tribe was sampled, 27 individuals).

        And in the fourth paper at my disposal (smallest sample size), Pakistani Pashtuns are at 6% for haplogroup M (no information on tribe(s), 19 individuals).

        So pretty safe to say that depending on tribe, Pakistani Pashtuns will range between 0% and 30% mtdna haplogroup M, and we might reasonably guess that most will hover around 15%-25%.

      2. With respect to Afghan Pashtuns, we can’t enjoy the benefit of multiple studies (or rather, it could be that I’m having trouble finding said studies, lol).

        But on the brightside, there is that Di Cristofaro et al. paper from a few years back.

        In it, we see Afghan Pashtuns (kinda redundant, since “Afghan” means “Pashtun” in Farsi/Dari) at 13% mtdna haplogroup M. So they don’t really seem to differ from the pattern seen with Pakistani Pashtuns.

        (90 samples, no tribal information/breakdown. But I have a source that tells me that these samples are mostly Ghilzai)

  17. @Razib,

    How do I read your Hirisplex spreadsheets.


    I understand MAF=Minor allele frequency. What does MAC stand for?

    Where does it tell you how many samples the allele frequency is based on?

  18. One detail missing here is the fact that most of R1a-Z93 in modern India belongs to a sister clade of the one found in the steppe (R1a-L657 vs. R1a-Z2124). With the large number of samples from the steppe carrying the latter while none the former, we can safely say that the former was a very rare lineage in the steppe populations (frequency probably below 1%?).

    So while it must have arrived with someone from the steppe, its modern frequency in India is completely irrelevant to measure any impact of steppe migrations. The rise in frequency is a local phenomenon that has been happening probably during the last 2500 years. The male who started this founder effect could have been 1% steppe admixed, or it could have been a recent migrant with 90% steppe ancestry, but it’s highly irrelevant either way.

    For all we know, R1a (as R1b) is not relevant for the spread of IE languages, since ancient DNA has disproved this outdated connection that was based on modern DNA. In India we still don’t know the fine details, but the evidence available suggests a late arrival, with minimal impact before 500 BC or even 1 AD.

    This is all old news, anyway, known for 2 years already. Isn’t it time to catch up with the data and drop obsolete theories that go against the evidence?

  19. “the aryans when they arrived in India were…..”
    Lmao, the aryans didnt arrive, the steppe people did. Find me one y hg R-Y3/Y2/L657 from these cultures – fatyanovo, sintashta, corded ware whatever.

      1. Yes razib, theres so much r1a in swat, forget even if its not the L657 kind whose tmrca is much prior to 1200bce. and that steppe ancestry in swat is so male mediated right?

  20. Looks like Alberto covered my point very nicely in the post above me.

    Although I do find some issue with the first part of one statement of his, which is still a hypothesis and is not yet confirmed because of lack of samples from india, but otherwise he is spot on.

    “So while it must have arrived with someone from the steppe, its modern frequency in India is completely irrelevant to measure any impact of steppe migrations. “

  21. I mentioned before that similar papers which do not refer on Vinca are little relevant. The paper says nothing about indigenous people who lived in Europe and who at that time already had the alphabet for 3000+ years. Everyone is afraid to even speculate which language they spoke and if this was so-called an Indo-European language. I am waiting for one paper which will give a big picture and follow logical timeline. First, where people lived during the Ice Age? After the IA, they started moving from the south to the north (not in the opposite direction) following the Ice frontier which was retreating to the north. Where Fatyanovo people lived before 2800BC? It is a bit tense that people came from the east to the Europe and maybe because did not like Europe or because of nostalgia, some of them went back to the east. The term ‘Slavs’ on the map, appeared 3500 years later, it should be replaced with ‘Serbs’. The fact is that the most of people during the IA lived in Vinca and about 6000 years ago started spreading around the Europe. About 4500 years ago, one stream (future Russians) crossed Carpathian and it took them about 500 years to reach the SA and today’s China. Someone noticed one r1a which is 10000 years old. In fact, Balkan (Serbian) r1a is 12000 years old, European r1a is 6-7000, Russian 4500 and SA is 3850 years old.

Comments are closed.