The genetic structure of the world’s first farmers

2fdrZf8.png
It is interesting that Jews plot with Neolithic Levant and Palestinians with Bronze Age Levant. Like Jews were sheltered from Bronze Age genetic influence. Well, they really must have hidden in a desert, not for 40 but 1,000 years. ;) Bedouins and Saudis are closer to Natufians though.
 
Looking at PCA, post above, it is interesting how Steppe Late Neolithic/Early Bronze is composed pretty much of two elements the EHG and Iranian N/Chl (Steppe EMBA). No EEF farmer input what so ever. Also in Extended figure 5.
By mid Bronze Age the Steppe group is strongly influenced only by Anatolian/European Neolithic farmer genetics at strong 30% level, Steppe MLBA. Strong EEF genetic flow. I'm assuming this could have been done by mixing heavily with Cucuteni farmers in West Yamnaya. Or EMBA was East Yamnaya and MLBA West Yamnaya. If I understood the paper this latter MLBA became dominant in the Steppe. However there is still this IA sample which is much closer to the EMBA population. I guess stepp was mixed and varied by cultures.
 
Any idea why Figure 1c is not in tune with Figure 4b? According to 1c Anatolian N is mixture of mostly Natufian/Levant farmer with WHG. According to 4b mostly Iranian N, then Levant N, and some WHG. 1b PCA chart also plots Anatolian and European farmer between position of Natufians and WHG. Iranian N being very far away.

If I understand it all correctly ADMIXTURE takes a whole lot of f3 stats and plots them. If it finds some samples clustering (i.e. a number of instances all with very low different f3 stats) it will consider it a population and measure the "distance" of others towards it as admixture proportions. You can imagine that for very old samples this will not work. They will have contributed for instance to population A as well as population B and C. The old samples will also have their own - lost - drift. This will be calculated as a mixture of A, B and C even if it is not.

ADMIXTURE will try to find a new cluster each step it takes.

So as long as Levant_N and Anatolia_N aren't considered different populations 1c is valid. As soon as both are considered separate populations 4b. In other words: 1c means K=11, 4b means K>11.
 
I have been busy and only saw this paper now. Finally some Natufian and Early Levantine Neolithic genomes! That's great news. I haven't had time to read the paper yet. Only the comments posted here and quickly browsing through the Y-DNA and mtDNA results.

I am surprised like most of you not to find any trace of Y-haplogroup G among the Pre-Pottery Neolithic Levantines, but I am pleased that we finally have some Y-DNA T and that it was indeed as I thought found among the earliest farmers. Note that the Mesolithic Natufians appear to belong exclusively to CT and E1b1b, while early pre-pottery farmers also have H2 and T. So it could be that H2 and T colonised the southern Levant from further north (e.g. northern Mesopotamia or Syria).

I do expect to find G2a among Early Farmers further north though, either in Syria or southern Turkey. They had to come from somewhere, and it is extremely unlikely that cereal farming spread almost exclusively through diffusion from the southern Levant to Anatolia while it spread almost exclusively by migration from Anatolia to all Europe. That wouldn't make any sense. Anyway it is sure that some T and H2 people were among the predominantly G2a Early European Farmers, but it's somewhat odd that only one T and no E-M123 have been found in Neolithic Europe or Anatolia to date. H2 looks like the strongest unifier of Near Eastern and European Neolithic people, which makes even less sense as it is almost extinct everywhere today.

Just as I expected there was no R1b, J1 or J2 in the Levant at the time. R1b-V88 would have arrived from the northern Fertile Crescent with cattle herders a bit later in the Neolithic, with J1 and J2 could have come with some goat herders in the Neolithic and/or Chalcolithic. More of them certainly came during the great Bronze Age expansions of the Indo-Europeans (R1a, R1b) and of the Kura-Araxes culture (J1, J2, G2a).

I wasn't sure about when E-M123 had entered the Levant from Africa. We now know that it was at least since the Mesolithic with the Natufians. That's interesting because it would mean that E-M123 wasn't brought by Proto-Semitic people during the Chalcolithic. Or else perhaps Proto-Semitic did really arrive in the Chalolithic but with a different wave of E-M34 people. In that case the E-M123 or other E1b1b would only be distantly related and all/most of these Natufian E1b1b lineages are now extinct, just like CT. This second hypothesis seems the most likely.

It's a bit disappointing that they only managed to test a single Natufian mtDNA sample. It's hard to believe that they couldn't sequence any when mtDNA is the easiest part of the genome to sequence. It turns out to belong to haplogroup N1b, which is almost exclusively found in the Middle East today.

The mtDNA from the Pre-Pottery Neolithic in the Levant is K1a4b, R0a and T1a2, three almost exclusively Middle Eastern lineages today (K1a4b is found chiefly among the Druzes). This, combined with the Y-DNA results, strongly suggests that the Levantine Neolithic is not the source of any European Neolithic culture.

Early Iranian Neolithic samples only yielded mt-haplogroup X2 (found all over Europe today) and J1c10, (found nowadays in central and western Europe, continental Italy, Sardinia and Morocco). Both were found among European Neolithic farmers.

Therefore, based on the (very scarce) Y-DNA and mtDNA data, it looks like European farmers came from Anatolia and Iran, or more probably a common source in the northern Fertile Crescent around modern Kurdistan, but not from the Levant.

Iranian Chalcolithic samples belonged to I1c, K1a12a (2x), H29, U3a'c and U7a, the Armenian Chalcolithic belonged to K1a8 (2x), H2a1 and U4a, while the Anatolian Chalcolithic had just one K1a17. Those lineages are still typical of the northern Middle East today and are much rarer in Europe than the Neolithic mtDNA from Anatolia and Iran. The only exceptions are I1c, H2a1 and U4a, which are found especially in central and eastern Europe today, but did not show up in Europe until the Indo-European invasions with the Corded Ware culture. It is highly interesting that these were found in Chalcolithic Armenia and Iran, as it appears to confirm that Proto-Indo-Europeans originated in that region before crossing the Caucasus to found the Yamna culture in the Pontic-Caspian Steppe. More intriguing still, these three lineages (H2a1, I1c and U4a) are actually linked to the R1a-dominant Corded Ware and Catacomb cultures, not the R1b-dominant Yamna culture. We still have to determine what could be the link between Armenia-Iran and the northern forest steppe of Russia. It could be the same population that brought CHG admixture to both Yamna and Corded Ware and it does appear to have been maternally mediated, as clearly Chalcolithic Armenians and Iranians had completed different Y-DNA from Steppe people (L1a, G1, G2a and J vs R1a and R1b).
 
Last edited:
Looking at PCA, post above, it is interesting how Steppe Late Neolithic/Early Bronze is composed pretty much of two elements the EHG and Iranian N/Chl (Steppe EMBA). No EEF farmer input what so ever. Also in Extended figure 5.
By mid Bronze Age the Steppe group is strongly influenced only by Anatolian/European Neolithic farmer genetics at strong 30% level, Steppe MLBA. Strong EEF genetic flow. I'm assuming this could have been done by mixing heavily with Cucuteni farmers in West Yamnaya. Or EMBA was East Yamnaya and MLBA West Yamnaya. If I understood the paper this latter MLBA became dominant in the Steppe.

That's how I read it as well. We don't have any yDna from West Yamnaya, correct? Maybe that's where R1b L51+ has been hiding.
 
@LeBrok,
Do you mean the Ashkenazim? They don't plot near ancient Levantine farmers from what I can see. They plot right near the Sicilians and Maltese. If I'm reading the modern PCA correctly, it's parallel to the Anatolian Chalcolithc sample. It's true though that a lot of Jews carry E-M123, but so do some Palestinians, I think.

Another handy graphic:
311pgrq.jpg
 
Ancestral = Negative, He doesn't belong to P322.

yes , you are correct................so there in no link between the T (i1707) in the levant and the 2 early neolitihc T found in germany.

It means the German T's came from another area.


currently the highest % of T in this area apart from the kurds are the syrian who declare themselves assyrian in ethnicity
 
It's a bit disappointing that thy didn't test (or didn't report yet?) any Natufian mtDNA. It's hard to believe that they couldn't sequence any when mtDNA is the easiest part of the genome to sequence.

Perhaps it will follow. IIRC other pre-print papers added more data as well. Possibly they will update the Y-DNA P1(Xa whole lot) as well.
 
Some of you may have seen this tweet from Lazaridis already:
Iosif Lazaridis ‏@iosif_lazaridis
I1635 (Armenia_EBA) is R1b1-M415(xM269). We'll be sure to include in the revision. Thanks to the person who noticed!

This is Kura Araxas.
 
Some of you may have seen this tweet from Lazaridis already:
Iosif Lazaridis ‏@iosif_lazaridis
I1635 (Armenia_EBA) is R1b1-M415(xM269). We'll be sure to include in the revision. Thanks to the person who noticed!

This is Kura Araxas.
HA antoher one, I know I sound selfish and like a broken recorder but I pointed out the similarities between Kura Araxes and Steppic folks. I did point out Kura Araxes had both Kurgan and pit graves and horses there predate horses in Sintashta by several hundred years.

So far Kura Araxes period Haplogroups are L1a and R1b.
 
@LeBrok,
Do you mean the Ashkenazim? They don't plot near ancient Levantine farmers from what I can see. They plot right near the Sicilians and Maltese. If I'm reading the modern PCA correctly, it's parallel to the Anatolian Chalcolithc sample. It's true though that a lot of Jews carry E-M123, but so do some Palestinians, I think.

Another handy graphic:
Nope, all the other Jews in the area, especially the Libyan and Tunisian Jews. Look at modern population PCA in Extended Figure 1. Compare it to PCA Figure 1b
 
Imo if Kura Araxes had R1b, than Maykop which is roughly from the same timeframe further North had it too.
 
Just as I expected there was no R1b, J1 or J2 in the Levant at the time. R1b-V88 would have arrived from the northern Fertile Crescent with cattle herders a bit later in the Neolithic, with J1 and J2 could have come with some goat herders in the Neolithic and/or Chalcolithic. More of them certainly came during the great Bronze Age expansions of the Indo-Europeans (R1a, R1b) and of the Kura-Araxes culture (J1, J2, G2a).

I wasn't sure about when E-M123 had entered the Levant from Africa. We now know that it was at least since the Mesolithic with the Natufians. That's interesting because it would mean that E-M123 wasn't brought by Proto-Semitic people during the Chalcolithic. Or else perhaps Proto-Semitic did really arrive in the Chalolithic but with a different wave of E-M34 people. In that case the E-M123 or other E1b1b would only be distantly related and all/most of these Natufian E1b1b lineages are now extinct, just like CT. This second hypothesis seems the most likely.

It's a bit disappointing that they only managed to test a single Natufian mtDNA sample. It's hard to believe that they couldn't sequence any when mtDNA is the easiest part of the genome to sequence. It turns out to belong to haplogroup N1b, which is almost exclusively found in the Middle East today.

The mtDNA from the Pre-Pottery Neolithic in the Levant is K1a4b, R0a and T1a2, three almost exclusively Middle Eastern lineages today (K1a4b is found chiefly among the Druzes). This, combined with the Y-DNA results, strongly suggests that the Levantine Neolithic is not the source of any European Neolithic culture.Early Iranian Neolithic samples only yielded mt-haplogroup X2 (found all over Europe today) and J1c10, (found nowadays in central and western Europe, continental Italy, Sardinia and Morocco). Both were found among European Neolithic farmers.

Therefore, based on the (very scarce) Y-DNA and mtDNA data, it looks like European farmers came from Anatolia and Iran, or more probably a common source in the northern Fertile Crescent around modern Kurdistan, but not from the Levant.
The only squeaky wheel in it is that Anatolian farmers have 34% of autosomal Levant Neolithic in them. I see two possibilities to explain. Either Levant Y and mtDNA got deleted by better fit Iranian and local types uniparental DNA, which is not impossible knowing how fast they can bloom and die due to bottlenecking (8.2k event) and other evolutionary forcings. Second solution is that 34% of what appears to be Levant DNA is actually from a group of very related to them HGs in Central Anatolia, possibly carriers of G2a. Therefore all farmers genes come from Iranian Farmers, the rest from local hunters. That's quite a twist to what we assumed just days ago, but it is what it is. Well, possible by now, till dust settles.

samples belonged to I1c, K1a12a (2x), H29, U3a'c and U7a, the Armenian Chalcolithic belonged to K1a8 (2x), H2a1 and U4a, while the Anatolian Chalcolithic had just one K1a17. Those lineages are still typical of the northern Middle East today and are much rarer in Europe than the Neolithic mtDNA from Anatolia and Iran. The only exceptions are I1c, H2a1 and U4a, which are found especially in central and eastern Europe today, but did not show up in Europe until the Indo-European invasions with the Corded Ware culture. It is highly interesting that these were found in Chalcolithic Armenia and Iran, as it appears to confirm that Proto-Indo-Europeans originated in that region before crossing the Caucasus to found the Yamna culture in the Pontic-Caspian Steppe. More intriguing still, these three lineages (H2a1, I1c and U4a) are actually linked to the R1a-dominant Corded Ware and Catacomb cultures, not the R1b-dominant Yamna culture. We still have to determine what could be the link between Armenia-Iran and the northern forest steppe of Russia. It could be the same population that brought CHG admixture to both Yamna and Corded Ware and it does appear to have been maternally mediated, as clearly Chalcolithic Armenians and Iranians had completed different Y-DNA from Steppe people (L1a, G1, G2a and J vs R1a and R1b).
Well, if we assume that proto IE is actually Iranian Neolithic, which fed Anatolian Neolithic, which fed European Neolithic, it makes European Neolithic proto IE too. Meaning all old Europe spoke proto IE language, plus half of Middle East. It makes irrelevant whichever farmers, Maykop or Cucuteni, taught language to Steppe people via farming and mingling, it was all proto IE. Actually it would make it easier to understand why the steppe invaders were so successful teaching all Europe and half Asia to speak their language. It was already very similar to local languages.
Well, one of possibilities at the moment.
 
@Maciamo

I do expect to find G2a among Early Farmers further north though, either in Syria or southern Turkey. They had to come from somewhere, and it is extremely unlikely that cereal farming spread almost exclusively through diffusion from the southern Levant to Anatolia while it spread almost exclusively by migration from Anatolia to all Europe.

The model i have in my head is multiple farmer groups in the region expanding out from their start point until they bump into each other and with none stronger than the others they create an equilibrium - so in this interior regions crops and animals diffuse across the borders by trade.

It is their expansion *away* from each other where they are competing with HGs which happens by migration.

#

@Holderlin

Also I think Basal Eurasian comes from South Asia.

Flooded Persian Gulf is my current guess.

#

edit: example of first point

say the start points were
- aegean (G?)
- nile delta (E?)
- persian gulf (BE?)
- south caspian (J?)

and then they all expanded they'd bump into each other in the middle

but their outward expansion would be over HGs
 
If I understand it all correctly ADMIXTURE takes a whole lot of f3 stats and plots them. If it finds some samples clustering (i.e. a number of instances all with very low different f3 stats) it will consider it a population and measure the "distance" of others towards it as admixture proportions. You can imagine that for very old samples this will not work. They will have contributed for instance to population A as well as population B and C. The old samples will also have their own - lost - drift. This will be calculated as a mixture of A, B and C even if it is not.
Thanks I have to look into f3 stats to understand better.
ADMIXTURE will try to find a new cluster each step it takes.

So as long as Levant_N and Anatolia_N aren't considered different populations 1c is valid. As soon as both are considered separate populations 4b. In other words: 1c means K=11, 4b means K>11.
Hmmm, in 1c figure 1 they distinguish Iranian N by green colour, as distinct admixture. I think if you enlarge the chart you should see a little squares of green in Anatolian Neolithic. Anyway, Iranian Ch contains both Levant and Iranian N, blue and green. Why wouldn't Anatolian then?
Check this latest paper:
http://www.eupedia.com/forum/thread...the-world’s-first-farmers?p=482077#post482077
Figure C, GD13A represents Iranian Neolithic, Anatolian Neolithic has maybe 5% of it.

I think there is some disagreement here in calculation of influence of Iranian Neolithic in Anatolia, perhaps in what they call prediction.
 
The only squeaky wheel in it is that Anatolian farmers have 34% of autosomal Levant Neolithic in them. I see two possibilities to explain. Either Levant Y and mtDNA got deleted by better fit Iranian and local types uniparental DNA, which is not impossible knowing how fast they can bloom and die due to bottlenecking (8.2k event) and other evolutionary forcings. Second solution is that 34% of what appears to be Levant DNA is actually from a group of very related to them HGs in Central Anatolia, possibly carriers of G2a. Therefore all farmers genes come from Iranian Farmers, the rest from local hunters. That's quite a twist to what we assumed just days ago, but it is what it is. Well, possible by now, till dust settles.

Admixtures are not as reliable as haplogroups. Admixtures are only simulation based on a researcher's selection and comparison of specific alleles. You should be especially careful about the naming of admixtures, as they can be misleading. If you collect ancient samples haphazardly from various sites in different regions, how are you supposed to know where one admixture originated. In most cases it will be with an ancient population that hasn't been sampled yet.

In the case of Basal Eurasian admixture, we have no idea if it matches exactly a specific Mesolithic or Early Neolithic population, or if it represents an admixture that was already widely distributed and blended with other admixtures in the Mesolithic or Early Neolithic Near East. It is dangerous to assume that Basal Eurasian necessarily represents the very first farmers. Then how do we know that there was only one ethnic group of original farmer? Wheat, barley, oats, flax, lentils, peas and chickpeas, the so-called Neolithic founder crops, could very well have been domesticated each by a different Mesolithic tribe. That is why we see Early Neolithic farmers belonging to a variety of Y-haplogroups like CT, C1a2, E1b1b, G1, G2a1, G2a2, H2, T, which would most likely have been distinct tribes (i.e. extended families) during the Mesolithic. But farmers doesn't only mean cereal and legume farmers, but also goat, sheep, pig and cattle herders, and all the evidence at present suggest that these were domesticated by yet other tribes - R1b for cattle, J1 or J2 for goats and sheep, and possibly one of the above, like G2a, for pigs.

The way I see it now is that many independent tribes inhabiting the Fertile Crescent in the Mesolithic domesticated various crops. It took there centuries, or even millennia of trading and intermarriages to progressively merge with one another, but leaving large regional disparities. H2 and T may have been domesticating Emmer wheat and barley in modern Syria, while CT and E1b1b domesticated legumes around modern Israel/Palestine and Jordan, G2a domesticated Einkorn wheat pigs in the Taurus, and R1b rounded up the first cow herds just south of the Taurus. If that kind of scenario took place, which population matches the "pure" Basal Eurasian? Or to phrase it another way, isn't Basal Eurasian the result of the merger of several of these tribes already? Shouldn't we have separate admixtures for each group? After all at first there was only one Mesolithic European admixture, but we now clearly distinguish between WHG and EHG, and there could be a third distinct one in the Balkans or Italy for all we know.


Well, if we assume that proto IE is actually Iranian Neolithic, which fed Anatolian Neolithic, which fed European Neolithic, it makes European Neolithic proto IE too. Meaning all old Europe spoke proto IE language, plus half of Middle East. It makes irrelevant whichever farmers, Maykop or Cucuteni, taught language to Steppe people via farming and mingling, it was all proto IE. Actually it would make it easier to understand why the steppe invaders were so successful teaching all Europe and half Asia to speak their language. It was already very similar to local languages.
Well, one of possibilities at the moment.

That's the point. H2a1, U4 and I1c were not found in the European Neolithic, nor in the Anatolian or Iranian Neolithic. The only pre-Bronze Age samples we have are from Chalcolithic eastern/central Europe, Armenia and Iran.
 
@Gravetto-Danubian

I have also something like that in my head, the first different tribes involved in agriculture and herdering would have reached a population density high enough as to need more land; there were two solutions left: fight for land agaisnt their farmer neighbours (already with a dense population also), or the easy solution, to occupy lands of the neighbour HG, almost empty.

@Maciamo

If you suggest that R1b was carrying PIE from Iran... it would have similar problems as the IE from Anatolia: what to do then with the hundreds of no-IE languages left in the Caucasus? Why the way what this new R1b guy does is in fact is adding up another signal for a "yellow alarm", in this case one that may give with more data the final game over to R1b / IE; if the paper puts clear that peoples from Iran arrived to the steppes, it was through Armenia / Caucasus, and there we find now (again before any IE expansion) a R1b person, in this case under a stone kurgan from a developed culture that expanded over a region that the first language known in the area was non-IE Hurrian (and then the related Urartian). Once the Urartian kingdom was beaten by the Assyrians the IE Armenians were able to profit the kaos to get in from the west in the VI BC. Other R1b's remained there, as in the sample of 1800 BC (a R1b-ht35 Hurrian) and 1000 BC (a R1b-Z2013 Uratian) as to continue with actual R1b Armenians being of the same R1b clade as their Urartian ancestors. But the worst with that is that the first clade is the same found in 1 Yamnayan and the "son" clade is found in 10 Yamnayans... so what to do? to think that R1b was added up to the northwards expansion with metals and wheels reaching Yamnaya, or to figure out that the same R1b Yamnayans clades came back to where they came? (and only such R1b specific clade among those that is supposed to have Yamnayans). As the new Armenian sample is xM269 it is reasonable to think that he was M73? The case is that M73 are now in the area that spans from Samara to Kabul, peaking in Bashkirs, is so another possible case of a clade carried northwards to add up? So the yellow alarm is that there are enough data to point that a bunch of R1b clades came to the steppe from the south... but it is doubtful that they were speaking IE when settled there.
 
I don't think there's any way of knowing at this stage where the Basal Eurasian was hiding, but one of the points against an Arabian refugia is that Levant Neolithic has less of it than Iranian hunter-gatherer (44% to 64%). On the other hand, both Levant Neolithic and Iranian Neolithic have about the same amount-44%-so the argument could be made that Basal was diluted in the west by something WHG like and in the east by something EHG or maybe ANE like.

The other alternative often mentioned is a Persian Gulf refugia, which is actually I think what Alan said?

@Holderlin,
I've considered that too. It would certainly explain why we haven't stumbled upon it yet, given that we have nothing really old from that region.

@Alan,
That makes sense to me, and I think the statements in the paper alluding to perhaps more "proximate" contacts was included. So, not that there wasn't actual steppe movement to India, but that there are older processes which may be inflating the figures.


My poor old brain.All these simulations! Everyone of them is based upon human choice. No divine revelation helas as always in autosomes evluations.
I' ve to read and read again before synthetize. But yes I think (it's almost evident) exchanges of genes occurred between CHG ancient people and EHG of Steppes, even before Neolithic without speaking of an ancient common "ANE" or/and "proto-partly-gedrosia" anterior layer of old genes common all over central Asia. What is from deepcommon ancestry, what is from newly admixtures? We are sometimes speaking as if the genes states were a stable and "frozen" thing.
admixtures simulations are simulations (I repat myself, the age!), we need other tools as IBS, IBD things and what more?...
Concerning Basal Eurasian, seemingly Arabia is not the best choice; maybe as you say something around Persic Gulf?
By the way concerning Koweit people and surroundings, Coon said they presented some 'veddoid' traits more than central Arabs. In admixture we have South-Asian (for the worth it has) too. Here too, same problem: Is the ASI found among Near-Eastern Iranian Hindu Kush people the result of recent events linked to Harrapan introgression, or to gradual osmosis after Iranian/South Caucasus introgressions into ex-Harappa (Neolithoc, BA?) or for a part to deep ancestry? Perhaps no links with true (recent it's true) ASI but B-E could be linked to Elamits of Southern Iran (dravidian speaking)? All that is very complicated for me.
I would like to have the supposed 'basal-eurasian' estimations for Southern India.

&: apart: diverse papers I red always show a slight but ever present taste of Northern European in Hindu Kush and surroundings regions. WHat depth???
 

This thread has been viewed 217608 times.

Back
Top