Saturday, June 18, 2016

Insights from New Pre-Historic Middle Eastern mtDNA

Lazardidis et al. 2016 collected genome-wide data, along with mtDNA, from 44 ancient Middle Easterners. 27 of 44 had enough mtDNA coverage to get haplogroup results. I added the new ancient mtDNA results to this spreadsheet Ancient Middle East which includes all of the ancient Middle Eastern mtDNA results I've collected that are at least 3,000 years old.

By the way I have upcoming posts on analysis of mito-genomes from specific haplogroups and regions, just like the one I did of JT recently. Analysis of mitogenomes of Southern Africa and haplogroups K1 and H1 are coming up next.

Preview: The new mtDNA results show that modern regional mtDNA diversity in West Eurasia had already began to form by the Neolithic and support the conclusions of Lazardidis et al. 2016 that Levant Neolithic and Anatolian Neolithic did not have a lot of recent common ancestry.

I broke up my post into two sections.

1. What "Middle Eastern mtDNA" is. 
2. The mtDNA relationship between Paleo/Neo/Chalcolithic Middle Easterners and modern West Eurasians.

Lazardidis et al. 2016 was an absolute masterpiece like all of the other ancient DNA papers created by the same team of researchers who appear to have dedicated their careers to unlocking the origins of humanity as a whole and the diversity within humanity by taking DNA from old bones.

Below listed regions and time periods  the new mtDNA results come from. .

Levant(Israel and Jordan).
Paleolithic 12,000-14,000 years old. N1b
Neolithic 9,000-10,000 years old. K1a4b, T1a, T1a2, R0a, R0a2
Bronze age 4,000-4,5000 years old H14a, X2m

SouthWestern Iran
Mesolithic 11,000 years old
Neolithic 10,000 years old X2, J1c10
Chalcolithic 6,000-8,000 years old K1a12a, K1a12a, U7a, U3a'c, I1c, H29, X2
Bronze age 3,500 years old U1a1

Chalcolithic 5,500-6,500 years old K1a8, K1a8, H, H2a1, U4a
Early Bronze age 4,500 years old H1u, X2f
Late Bronze age 3,500 years old T1a1'3

Western Turkey
Chalcolithic 6,000 years old K1a17

1. "Middle Eastern mtDNA"

I consider the below mtDNA hapologroups "Middle Eastern mtDNA".
R0(inlu. HV, H, V), U1, U3, U7, K(aka U8b2), U9, JT, N1, N2, X.

This is why I consider them Middle Eastern: We have over 100 mtDNA samples from European hunter gatherers ranging 38,000 to 8,000 years old. All but two belonged to haplogroup U(xK, U1, U3, U8, U9). Just about all of them belonged to European-specific forms of U. They contributed 10-25%(ranges by region) mtDNA to modern Europeans. The rest of European mtDNA falls under haplogroups that first appear in Europe when people from the Middle East migrated there starting 9,000 years ago. These newcomers to Europe also carried with them a new type of ancestry that didin't exist in Europe prior. It;s a Middle Eastern-specfic type of ancestry that all our new ancient Middle Eastern genomes share. Everywhere this Middle Eastern-specific ancestry is found today so are the above mtDNA haplogroups.

All of these Middle Eastern haplogroups are definitely over 30,000 years old. By 10,000-12,000 years ago they had gained many subclades and regional diversity. There was already a high amount of sharing between distant regions in the Middle East among distantly related populations and many "Expansion Point Lineages"(Definition of term is here). Most "Expansion Point Lineages" that exist today certainly already existed 10,000 years ago. This prediction of mine is supported by the new ancient Middle Eastern mtDNA results. By the time Middle Eastern mtDNA migrated into Europe, South Asia, and Africa there was a lot of mtDNA sharing within the Middle East and so today you'll find subclades of haplogroups(like T2a1a) that show no differentiation in Ireland and Yeman and Ethopia and India. You can't tell apart Ethopian T2a1a from Irish T2a1a.

Today a great majority of European mtDNA is "Middle Eastern". Don't let the term "Middle Eastern" confuse you. They were only in the Middle East 10,000 years ago but since then they have spread. Europeans, East Africans, and South Asians are to a large extent of Middle Eastern decent. Especially Europeans, most are over 50% and the impact is much greater on mtDNA than Y DNA.

2. The mtDNA relationship between Paleo/Neo/Chalcolithic Middle Easterners and modern West Eurasians.

Below is a comparison of the new ancient Middle Eastern mtDNA and already published mtDNA results from Neolithic Anatolia and Europe to modern mtDNA. In the end I give a concluding remark as to what this tells us about the origins of regional mtDNA diversity in West Eurasia today.

Natufian and Neolithic Levant mtDNA: N1b, K1a4b, T1a, T1a2, R0a, R0a2.

N1b  and R0a peaks in SouthWest Asia today are rarely found outside of the Middle East. T1a2 and K1a4b are Middle Eastern-specific haplogroups today(however T1a2 has a strong presence in Italy) and peaks in frequency in the Levant. Most of my SouthWest Asian K1a mito-genomes so far are K1a4b.

So in conclusion Natufian and Neolithic Levant mtDNA so far looks similar to mtDNA in the Levant today.

Neolithic Anatolia/Europe:

We have over 300 samples from them so there's no point in listing all those results. I will explain though how Neolithic Anatolian/European mtDNA was similar to modern European mtDNA.

>J1c, T2b, J2b1a: Today these are all European-specific clades of JT(see here). A similar ratio of Neolithic Anatolia/Europe's JT belonged to these subclades as modern Europeans, and in fact they had a higher frequency of them than any modern Europeans. Of the few high coverage J1c, T2b, and J2b1a genomes from Neolithic Anatolia/Europe we can see they had the same subclade combustion as modern Europeans. This includes popular subclades that are at >1% outside of Europe.

>K1a4a, K1a1b1, K1a2a, K1a3a: These are all K1a subclades that are rare or non-existent in my Middle Eastern mito-genomes and very popular in my European mito-genomes. They've all been found in the few K1a mito-genomes we have from Neolithic Europe. K1a4a is the most important one. It's the main K1a subclade in at least Denmark and White Americans, and was found in a 7,000 year old woman from Neolithic Spain. Her ancestors arrived in Spain only a few hundred years earlier from Anatolia.

>HV6-17, HV0(mostly V): These are the only popular forms of HV(xH) in Europe. They both are very rare outside of Europe. West Asia, especially SouthWest Asia, has a more diverse array of popular HV(xH). No HV(xH) sample has been found in Neolithic Anatolia/Europe except European-specific HV6-17 and HV0(mostly V) and both were fairly popular.

>H1, H3: H1 is very popular in all of Europe reaching 15% or higher outside of Italy and the Balkans. It is under 5% in all of the Middle East. H3 is at just over 5% in the British Isles, Scandinavia, France, and Iberia but is only at a few percent in the rest of Europe. H3 is at a >1% frequency in the Middle East.

Neolithic Central Europeans had a low frequency of H1(~5%) and even lower for H3(less than 1%). However Neolithic French and Spanish had high frequencies of both and most importantly H3. Most of their H was either H1 or H3. In some Neolithic grave sites in Spain H3 was more popular than H1. Notably typical Basque H3c was found in Neolithic Spain and typical Danish H1c was found in Neolithic Sweden.

Neolithic and Chalcolithic Iranian mtDNA: X2, J1c10, K1a12a, K1a12a, U7a, U3a'c, I1c, H29, X2

J1c10 is a very rare subclade of J1c. My J1c10s are from Bedouin, Italy, Sardinia, and Morocco. Nothing can be concluded about J1c10 from that. However I can say that J1c is the primary form of J in Europe and rare outside of Europe. It arrived in Europe 9,000 years ago from Neolithic Anatolia. It's very surprising to see J1c10, instead of J1b1b, in Iran 10,000 years ago. Very surprising.

K1a12 is almost completely absent in close to 1,000 K1 mitogenomes from Denmark, Finland, and White Americans. That covers small parts of Europe. My only K1a12(all but two are K1a12a) samples are from Armenia, Kuwait, Iran, Turkey, Druze(Levant), and Italy. It isn't particularly popular but my data so far supports the idea it's mostly West Asian and even Northern West Asia(where Iran is).

U7 doesn't reach 1% in frequency in any region of Europe. It's nonexistent in close to 500 mtDNA samples from pre-historic Europe. U7 is at 1.3% in the Levant, 3-5% in Turkey and Iraq, and a whopping 10% in Iran. U7 is also one of the most popular West Eurasian haplogroups in India. The U7a result from Chalcolithic Iran directly connects them maternally to modern Iranians and South Asians.

U3 varies from 1-3% in Europe, is at 3% in Iran, and 5% in the Levant. It peaks in SouthWest Asia but has a decent presence in all of West Eurasia. I do have a large collection of U3 mito-genomes but haven't looked at them in detail yet. Most are U3b not U3a'c in every region, including Iran.

Chalcolithic, Bronze age Armenia and Anatolia mtDNA: K1a8, K1a8, K1a17, H, H2a1, U4a, H1u, X2f, T1a1'3

K1a8 and K1a17: Both are rare. My only examples of K1a8 are from the Levant(several countries. My only examples of K1a17 are from Levant, Arabia, Italy, Egypt, and Kuwait.

U4a: U4a is of "EHG"(Mesolithic East Europe) origin no doubt about it. It's consistent with the EHG ancestry that Lazardidis et al. 2016 modeled Chalcolithic Armenia as having. Looking at the evidence presented by  Lazardidis et al. 2016 I'm very confident Chalcolithic Armenians had EHG ancestry.

H2a1: It found at least 1% in all of my West Eurasian regional sample sets except Iran, Spain, Balkans. It peaks in my samples from the NorthEastern corner of Europe(Baltic, BeloRussia, Karelia and Russia) stats at about 3%. Frequencies of H2a1 can't tell you anything about its history. Only a large amount of H2a1 mito genomes can.

Something significant I can say though looking at ancient mtDNA that H2a1 is absent in over 300 samples from Neolithic Europe but existed in Eneolithic and Bronze age Russia and then later in Bronze age Central Europe after people from Russia migrated there. It's existence in Chalcolithic Armenia, suggests it originated in the Caucasus region and then migrated to Russia and from there Central Europe.

H1u: Very rare subclade of H1. I have over 1,000 H1 mito genomes from Basque, Danish, Finnish, Italy, and White Americans. Two are H1u, one from Denmark and one from a White American. I have a much smaller amount of H1s(maybe 20) from the Middle East and have found two examples of H1u and both are from Druze.

X2f: It is absent in my European samples. It exists in just about every Northern West Asian population I have examples from; Georgia, Armenia, Iran. I also have an example from Druze.

T1a1'3: T1a1'3 is the common ancestor of T1a1 and T1a3. Chances are this sample had T1a1 or T1a3, but because of low coverage it couldn't be discerned which one he had. T1a3 is unheard of today, my only examples are from three White Americans and one Isreali. T1a1 on the other hand is the most popular form of T1a in Europe and Iran. In Northern Europe <90% of T1a is T1a1, while in Iran and Italy T1a1 is most popular but you'll commonly find other forms of T1a. Many examples of T1a1 have been found in Bronze age Europe and Central Asia(immigrants from Europe). It was popular in Bronze age Steppe populations and likely brought to Europe from the Steppe, and ultimately from ancient Northern West Asia.

Nothing can be deciphered from a X2 results(instead of X2b, X2f, etc). I don't have any I1c mito-genomes as of far. I1 is the most popular I subclade in most of West Eurasia and first appears in Europe with Steppe admixture(ultimately from populations similar to Chalcolithic Iran?).

Concluding Remark: These new results and already published result suggest that by the Early Neolithic much of the mtDNA diversity in West Eurasia had formed. We see Levant-specific mtDNA in Neolithic Levant, Northern West Asian-specific mtDNA in Neolithic/Chalcolithic Iran, and European-specific mtDNA in Neolithic Anatolia.

The mtDNA differentiation between Anatolia Neolithic and Levant Neolithic are consistent with conclusions by Lazardidis et al. 2016, that farming spread to Anatolia without much gene flow from the Levant.


  1. What can you say about the similarities/differences between the mtDNA of Steppe_EMBA and Iran_ChL+N?

    1. All they share is hg I(I1 is found in LNBA Europe) which is absent from Neolithic Europe.

      The differences are U7/U3 in Iran and U5a/U4 in Steppe. U7 and U3 are absent from Steppe and LNBA Europe but are pretty popular in Iran today. U5a and U4 are EHG-lineages so had a strong presence in Steppe but were non-existent all non-hunter gatherers in the world in 3000 BC.

      To understand ancient Iran's mtDNA relationship to the Steppe we need more mtDNA especially from Neolithic Iran because Chalcolithic Iran had significant Levant_Neolithic-type ancestry and definitely isn't ancestral to Steppe people. The U7 and U3 is reflective of the fact Chalcolithic Iran didn't contribute to the Steppe.

      Chalcolithic Armenia wasn't ancestral to Steppe either but it's important to remember they shared H2a1 and U4a with Steppe people and LNBA Europe. Interestingly both were found in Chalcolithic Samara. Armenia and Steppe got H2a1(CHG) and U4a(EHG) from the same sources, but one didn't contribute to the other.

  2. Krefter

    Nice summary. I would like to hear Your opinion why there is no typical WHG mtdna there despite the claims that 50% of Levant is WHG. Also the same in Iran_N. No mtDNA typical to high ANE levels. Like W.

    Btw do You update this spreadsheet of ancient Y DNA?
    Or You have new link.

    1. The Levant is not 50% WHG. They're about 50% from people related to WHG. As far as we know WHG only lived in Western Europe(maybe originally Italy). This is why today mtDNA U5b(Over 90% of WHG had U5b) is rarely found outside of Europe.

      mtDNA U is popular in ancient Middle Easterners and is related to WHG's U5b. Haplogroups such as K(aka U8b2) and U3 are the best candidates for WHG-related mtDNA in the Levant.

      "Btw do You update this spreadsheet of ancient Y DNA?"

      I'll update it soon.

  3. This comment has been removed by the author.

  4. Hi Krefter--my own mt-DNA haplgroup is H1u...fascinating to learn about your theories on its origins! My maternal line is from Calabria, Italy--in the Sila Piccola mountains, on the border of Catanzaro and Cosenza provinces. GENBank: JN224991

  5. You analysis would suggest that mtDNA T made its way into Egypt and East Africa from the Levant via the Nile rather than by sea from Mesopotamians who had sea commerce very early on.