Movements and genetic change between the Chalcolithic and Bronze Age in the Levant

[Ygorcs]

Hi! I have previously read about the surprising Chalcolithic Israel individuals with a lot of ANF and presence of genes for blue eyes in relation to the later Bronze Age samples from the same region. I have also often been very curious about how Proto-Semitic and its immediate descendants spread in the Near East, considering an unusual correlation: right between the Neolithic and the Bronze Age Proto-Semitic appears, expands and splits into many different languages from the Mediterranean to the fringes of the Zagros (North Mesopotamia), but at the same time it seems to have been a time of big increase of Iranian/Caucasian ancestry, not an expansion of Natufian/Levant_N components as you'd expect from a language that has its closest language groups in North & Northeast Africa and had a major presence in the Levant since its very beginning (Eblaite, in Ebla, Northwestern Syria). Considering all of that, I decided to try and make some models of geentic ancestry just to either reinforce or debunk some of my admittedly amateur speculations (by the way, I ended up finding out that the results tend to be much more coherent and plausible if you just use about 10 dimensions instead of the 25 or 50 you can find in databases). See:

Using mostly Mesolithic/Early Neolithic references (distance rates ~0.8-1.1%)

MESOLITHITC-NEOLITHIC	NATUFIAN	LEVANT_N	LEVANT_CHL	LEVANT_BA_S
Natufian	100	53,0	45,2	47,5
Barcin		44,7	41,0	23,5
Abdul Hosein			10,6	13,8
CHG		2,3	3,2	15,2
Iberomaurusian
Ethiopia Mota
EHG
Onge				0

There is a clear trend that parallels what happened in other parts of the Fertile Crescent. Firstly, a huge increase of Iranian (Abdul_Hosein_N) and Caucasian (CHG) ancestry after the Neolithic era. In this model, CHG increased particularly between the Levant_Chl (those ~49% blue-eyed Israelis) and Levant_BA_S; and secondly, an increase in the Natufian:Barcin ratio between the Chalcolithic and the Bronze Age. There are two possibilities I can see as most likely:
1) Levant_BA_S mostly descends from a Chalcolithic South Levant population that remained much more Natufian-like than even Levant_Neolithic, with much lower Anatolian gene flow, and which later received some moderate, but heavily CHG and Iranian-related influxes;
2) or Levant_BA_S was quite like Levant_Chalcolithic, but was profoundly transformed by at least two streams of different ancestry, one more enriched in Natufian, the other more enriched in Iran_Neolithic and particularly CHG.

Could this indicate, perhaps, that the earliest Proto-Semitic speakers were mainly a mix of Natufian+CHG/Iranian_N, with little Anatolia_N? Assuming a simple two-way admixture from Levant_Chl all the way to Levant_BA_S, the difference in Natufian, CHG and Iranian_Neolithic (Abdul_Hosein) admixtures could be theoretically (just mathematical speculations, okay? Nothing scientifically reliable here) explained by a mix of 40% Levant_Chalcolithic + 60% of a population with ~49% Natufian, ~39% Iranian+CHG and ~12% Barcin/ANF ancestral components. Could that have been roughly the genetic makeup of the earliest Proto-Semitic speakers in Northeast Levant/North Mesopotamia (assuming the earliest evidences of Semitic are also near the place where Proto-Semitic spread from)?

Or rather the Levant received several different streams of different populations, maybe one much more Natufian-like (Ancient Egyptians? People of the Arabian peninsula? Saudis are often modeled as very Natufian-like if compared to Lebanese or Palestinians even today) and the other much more Iranian/Caucasian-like.

Now, using a much more chronologically proximate model, with mostly Chalcolithic reference populations (but Natufian and CHG to account for possible unsample populations that had more of those components than the scarce Chalcolithic DNA samples we have as of now)... Again, Levant_BA_South looks like a Levant_Chalcolithic population heavily (>50% of genetic replacement) changed by Natufian-rich (additional 23.6%) and Iranian/Caucasian-rich (16.6% Seh Gabi_CHL + 10.8% CHG) populations. Either this was a population descended from a much more Natufian-enriched and less Anatolian-enriched Levant_Chalcolithic people, or they did indeed receive a major stream of very Natufian-admixed people.

Together with Iranian/Caucasian or independently? I don't know, but there's something interesting in Levant_BA_N, also associated with a Canaanite city:

Levant_BA_N, under those same constraints, is modeled as simply Levant_Chl (60.9%) with a lot of additional CHG/Iranian_Chalcolithic-related ancestry (15% Seh_Gabi + Tepe Hissar + 14% CHG), as well as minor EEF-related (Varna) and North African (Morocco_EN) influences, which are maybe just inaccurate proxies for some other population rich in EEF and indigenous North African ancestry (something North African_Chalcolithic perhaps?). But what matters is that there is none of that big chunk of extra Natufian that exists in the South.

So, that seems to indicate that a lot of Natufian didn't come with the same source that brought a lot of Iranian+CHG admixture after the Chalcolithic. It would've been mostly Iranian+CHG or maybe a Levant_Chl-like + Iranian+CHG-enriched population... Or we can also think that there was very ancient substructure within the Levant, so that the Levant_South had always (since the Neolithic at least) been more enriched in Natufian ancestry, whereas the North experienced a lot more Anatolian and (initially) Iranian influence. In that case, the major change would've been brought just by a lot more Iranian_N/Iranian_Chl+CHG ancestry... which then may have been or not brought by Proto-Semitic speakers...

Therefore, another doubt appears: should we start to think of the possibility that the Iranian/CHG, J1-P58 and other things spread by Semites came from a successful and expansionist non-Semitic-speaking population, especially males, who just adopted the local Levant_Chalcolithic language? We know that even in historic times, with plenty of evidences of Semitic languages, Hurrians ruled much of the North Levant and were highly influential there.)

What do you think is most likely? What do these results that suggest a profound change in the Levant between the Neolithic, Chalcolithic and BA can tell us about the historic cultures? I think this is a very important discussion that probably true scientists will investigate further in the future, because it has a lot to do with several of the first great civilizations.

distance%=0.2393"

Levant_BA_South

Levant_ChL,48.7

Natufian,23.6

Seh_Gabi_ChL,16.6

CHG,10.8

Tanzania_Luxmanda_3000BP,0.3

Ethiopia_4500BP,0

Barcin_ChL,0

Maykop,0

Tepe_Hissar_ChL,0

Varna,0

Morocco_EN,0

Bantu_S.W.,0

[1] "distance%=0.1013"

Levant_BA_North

Levant_ChL,60.9

Seh_Gabi_ChL,15

CHG,14

Varna,5

Morocco_EN,3.1

Tepe_Hissar_ChL,2

Ethiopia_4500BP,0

Natufian,0

Barcin_ChL,0

Maykop,0

Tanzania_Luxmanda_3000BP,0

Bantu_S.W.,0

[LeBrok]

Do we have any BA and IA Levant samples in GedMatch?

[Ygorcs]

Do we have any BA and IA Levant samples in GedMatch?

I don't know, LeBrok. I'm using a different calculator software. But I haven't been able to find any IA Levant samples. Don't they have any DNA of them extracted yet? There are N, Chl and BA samples, but not IA ones.

[Ygorcs]

Incidentally, I compared those results for BA Levant (North and South) with New Kingdom Egypt (also LBA) and Anatolia_MLBA. It all seems to fit what I'd expect from the genetic clines in that region back then. It's interesting that there is only CHG, not Iranian_Chalcolithic (Seh Gabi_Chl) in LBA Egypt, and only Iranian_Chalcolithic in MBLA Anatolia, but this may be just an artifact. In any case, it's curious that the 2 probably Semitic-speaking samples have quite a bit of Seh Gabi-related ancestry (Seh Gabi is in the Zagros close to Central-North Mesopotamia - https://www.researchgate.net/figure/...fig2_259741522), whereas Egypt has none and Anatolia only very little of it.

Just like Levant_BA_South, the population of Egypt_New_Kingdom seems to derive from a much more Natufian-like Levant_Chl (therefore, something more Levant_N, but perhaps with even much less ANF than it had), but the Levant-derived population of Levant_BA_North and its partial ancestry in Anatolia_MLBA seems to me much more similar to Levant_Chl.

[1] "distance%=0.3995"

Egypt_New_Kingdom

Levant_ChL,55

Natufian,25.7

CHG,13.9

Tanzania_Luxmanda_3000BP,4.5

Bantu_S.W.,0.9

Ethiopia_4500BP,0

Barcin_ChL,0

Maykop,0

Tepe_Hissar_ChL,0

Seh_Gabi_ChL,0

Varna,0

Morocco_EN,0

[1] "distance%=0.6047"

Anatolia_MLBA

Barcin_ChL,78.5

Levant_ChL,16.7

Seh_Gabi_ChL,3.3

Varna,1.5

Ethiopia_4500BP,0

Natufian,0

CHG,0

Maykop,0

Tepe_Hissar_ChL,0

Tanzania_Luxmanda_3000BP,0

Morocco_EN,0

Bantu_S.W.,0

[LeBrok]

I don't know, LeBrok. I'm using a different calculator software. But I haven't been able to find any IA Levant samples. Don't they have any DNA of them extracted yet? There are N, Chl and BA samples, but not IA ones.

Actually I've found one Levant Neolithic and 3 BA samples on GedMatch. But nothing from Iron Age or Classical period.

In Harappa World

M115616	I0867		M291439	I1706
Levant Neolithic			Levant BA
Run time	9.93		Run time	13
S-Indian	-		S-Indian	0.26
Baloch	-		Baloch	3.57
Caucasian	25.97		Caucasian	37.26
NE-Euro	-		NE-Euro	-
SE-Asian	0.07		SE-Asian	0.62
Siberian	-		Siberian	-
NE-Asian	0.06		NE-Asian	0.43
Papuan	-		Papuan	-
American	-		American	0.34
Beringian	-		Beringian	-
Mediterranean	32.53		Mediterranean	12.01
SW-Asian	39.86		SW-Asian	44.73
San	-		San	-
E-African	1.52		E-African	-
Pygmy	-		Pygmy	0.4
W-African	-		W-African	0.38

Increased Caucasian and Baloch and reduced Med points to Iranian/Armenian influence. I would expect more Baloch though if it is Iranian, so maybe it is more through Armenian/Anatolian mediation.
Higher SW Asian might point to some migration from Arabian Peninsula or Egypt area?

[LeBrok]

Does someone have GedMatch "Egypt New Kingdom" number?

[markod]

Did you take a look at the Negev Bedouin samples (A but especially B)? They seem to be the link between Egypt and the Middle East, and they are significantly more southern than the Natufians.



My guess would be that these are indicative of trans-Sinai populations who in my view likely spread Afro-Asiatic languages.

[Angela]

Did you take a look at the Negev Bedouin samples (A but especially B)? They seem to be the link between Egypt and the Middle East, and they are significantly more southern than the Natufians.



My guess would be that these are indicative of trans-Sinai populations who in my view likely spread Afro-Asiatic languages.

I'm not sure. They have SSA which might have been picked up relatively recently and would pull them south. Do you know where the other Bedouin group plots? Just nearer Natufians?

[markod]

I'm not sure. They have SSA which might have been picked up relatively recently and would pull them south. Do you know where the other Bedouin group plots? Just nearer Natufians?

BedouinA is very close to Saudis and some Palestinians. The intetesting thing is that they seem to have more SSA affinity (pink), despite being more northern.



My guess is that the Mesolithic North African populations are responsible for the southern shift. Taforalt plots with Afar from Ethiopia/Djibouti, so quite deep in Africa.

[Pip]

My calculations suggest the addition to Levantine DNA in the Bronze Age is a better fit with Northern populations (Armenia/Georgia) than with Iranian. I don't know the dates of the samples here, but a heavy (40-45%) migration from Armenia/E Turkey principally during the 3rd millennium BC provides the best fit.

I suspect Natufian:Anatolian proportions might not have changed that much during the Bronze Age, but perhaps varied more between the different Levantine populations.

[Ygorcs]

Did you take a look at the Negev Bedouin samples (A but especially B)? They seem to be the link between Egypt and the Middle East, and they are significantly more southern than the Natufians.



My guess would be that these are indicative of trans-Sinai populations who in my view likely spread Afro-Asiatic languages.

I haven't analyzed their DNA makeup, but it should be interesting to know what's going on in their ancestry. In my (preliminary) opinion about Proto-Afro-Asiatic speakers, I think they were probably associated with a common immediate antecessor of Halfan/Kubbaniyan and Iberomaurusian cultures and probably located (at first) in Egypt. Halfan-derived people later spread from Upper Egypt to the Levant, creating the Kebaran culture and hence the Natufians, with more West Asian admixture. That's just an assumption, of course. I'd suppose this original population would be somewhat between Taforalt and Natufian autosomally. However, my hunch is that several of the Afro-Asiatic branches that are still extant derive from secondary expansion of later cultures, not from the first Proto-Afro-Asiatic speakers (maybe the Omotic do, they're so divergent linguistically that some even doubted they were really Afro-Asiatic at all).

At first I wouldn't like to test Ancient Egyptians based on Bedouin A or Bedouin B. I don't like to mix modern and ancient populations much, except when they are very distant populations (so not much overlap) and there is no reasonably proximate ancient DNA to use as a reference... That southern-shifted position and apparent similarity with Ancient Egyptians may derive from a shared higher, even if minor, Subsaharan influence, or simply from a higher Taforalt-related ancestry from the original natives of Egypt as they mixed with their Natufian relatives.

[Ygorcs]

BedouinA is very close to Saudis and some Palestinians. The intetesting thing is that they seem to have more SSA affinity (pink), despite being more northern.

That's intriguing, sounds paradoxical, but maybe the rest of their ancestry was more northern, so that "evens" their position.

But, interestingly, among the modern populations I compared the ancient Egyptian to, Saudis came as the closest, and right after them Yemenis. I think that's mainly because of the higher Natufian-related admixture.

[Ygorcs]

My calculations suggest the addition to Levantine DNA in the Bronze Age is a better fit with Northern populations (Armenia/Georgia) than with Iranian. I don't know the dates of the samples here, but a heavy (40-45%) migration from Armenia/E Turkey principally during the 3rd millennium BC provides the best fit.

I suspect Natufian:Anatolian proportions might not have changed that much during the Bronze Age, but perhaps varied more between the different Levantine populations.

Interesting. Do you still have the results of that calculation/the model of ancestral admixtures you used?

But in any case I do not think we should interpret these results literally because there are just so few samples to account for the genetic structure that certainly existed, with populations varying in clines from one extreme to the other, particularly in the closely related CHG-to-Iran_Neolithic cline. There were probably populations that were neither exactly CHG nor totally Abdul_Hosein. Maybe the increased proportions of Abdul_Hosein+CHG or, using Chalcolithic samples, Seh_Gabi+CHG in fact indicate a source in an intermediary unsampled population that didn't live neither in the Caucasus proper (Kotias & Satsurblia), nor in the central Zagros (Abdul_Hosein and Seh_Gabi), but more in the highlands near Lake Van or Urmia. That would in fact make total sense in "my"(not just mine, in fact) assumption that Proto-Semitic came from Upper Mesopotamia (North Syria/North Iraq/South Turkey), close to where it is first attested with Eblaite, Mari Eblaite and Akkadian.


The only problem I can see is that it seems like the proportion of Abdul_Hosein:CHG varies completely along the time, but maybe those lower Neolithic and Chalcolithic proportions of CHG/Iranian ancestry are not directly related to the wave that brought Proto-Semitic to the Levant, bringing lots of "Iranian/CHG" J1 & J2 with them.

[halfalp]

Interesting. Do you still have the results of that calculation/the model of ancestral admixtures you used?

But in any case I do not think we should interpret these results literally because there are just so few samples to account for the genetic structure that certainly existed, with populations varying in clines from one extreme to the other, particularly in the closely related CHG-to-Iran_Neolithic cline. There were probably populations that were neither exactly CHG nor totally Abdul_Hosein. Maybe the increased proportions of Abdul_Hosein+CHG or, using Chalcolithic samples, Seh_Gabi+CHG in fact indicate a source in an intermediary unsampled population that didn't live neither in the Caucasus proper (Kotias & Satsurblia), nor in the central Zagros (Abdul_Hosein and Seh_Gabi), but more in the highlands near Lake Van or Urmia. That would in fact make total sense in "my"(not just mine, in fact) assumption that Proto-Semitic came from Upper Mesopotamia (North Syria/North Iraq/South Turkey), close to where it is first attested with Eblaite, Mari Eblaite and Akkadian.


The only problem I can see is that it seems like the proportion of Abdul_Hosein:CHG varies completely along the time, but maybe those lower Neolithic and Chalcolithic proportions of CHG/Iranian ancestry are not directly related to the wave that brought Proto-Semitic to the Levant, bringing lots of "Iranian/CHG" J1 & J2 with them.

Agamemmnon from Anthrogenica who is apparently as specialist of Afro-Asiatic languages, believe that ( something like that ) ; Semitic languages wich are genetically closer to Berbero-Lybian and Cushitic, was formed by the input of maybe an Ergative language, wich modern exemple would the closest to Northwest Caucasian, but ancestraly could be related with languages such as Elamit, Sumerian, Hurrian...

I didn't found the thread were he said that, but it was just a few times ago.

[Pip]

Interesting. Do you still have the results of that calculation/the model of ancestral admixtures you used?

An example is Jordanian BA, which best-fits with the addition of 22% Kalavan (NE Armenia early BA) and 10% Satsurbia (Georgia PHG) to a mix of Levantine Neolithic and some Natufian.

Curiously, Kalavan was at least partly R1b (a brother clade of P297).

[Pip]

An example is Jordanian BA, which best-fits with the addition of 22% Kalavan (NE Armenia early BA) and 10% Satsurbia (Georgia PHG) to a mix of Levantine Neolithic and some Natufian.

Curiously, Kalavan was at least partly R1b (a brother clade of P297).

I've just come across Rothman's study at https://www.pnas.org/content/112/30/9190, which matches the aDNA analysis above, suggesting that the main changes in the Levant during the Bronze Age were principally a result of the spread of Kura Araxes culture and people from an area very near Kalavan in Northern Armenia and down the Levant as far as Jordan over the period 2,800 to 2,650 BC.

[Regio X]

Actually I've found one Levant Neolithic and 3 BA samples on GedMatch. But nothing from Iron Age or Classical period.
In Harappa World

M115616	I0867		M291439	I1706
Levant Neolithic			Levant BA
Run time	9.93		Run time	13
S-Indian	-		S-Indian	0.26
Baloch	-		Baloch	3.57
Caucasian	25.97		Caucasian	37.26
NE-Euro	-		NE-Euro	-
SE-Asian	0.07		SE-Asian	0.62
Siberian	-		Siberian	-
NE-Asian	0.06		NE-Asian	0.43
Papuan	-		Papuan	-
American	-		American	0.34
Beringian	-		Beringian	-
Mediterranean	32.53		Mediterranean	12.01
SW-Asian	39.86		SW-Asian	44.73
San	-		San	-
E-African	1.52		E-African	-
Pygmy	-		Pygmy	0.4
W-African	-		W-African	0.38

Increased Caucasian and Baloch and reduced Med points to Iranian/Armenian influence. I would expect more Baloch though if it is Iranian, so maybe it is more through Armenian/Anatolian mediation.
Higher SW Asian might point to some migration from Arabian Peninsula or Egypt area?

Any result in well built calculators, correctly interpreted, may be meaningful of ancestry or shared ancestry. Notice that you're running ancient samples against modern samples/clusters, which are generally the recipients when there are older populations involved. Apparently this approach has its use, as we could notice in the similarity tools, for example. The results themselves may be informative, especially when analysed as a whole. But I'm affraid the clusters, in isolation, may become more complex here, now seen mainly as destinations rather than sources. If a certain calculator built for modern samples says a certain Neolithic farmer individual is 5% North Sea, then it doesn't mean that he has 5% of Scandinavian ancestry. Rather, it'll probably mean that Scandinavians must have received genetic influence either from Neolithic farmers and/or from related pop. So, increased Baloch may mean that some Levant BA-like pop influenced genetically Baloch people. Or that a third pop(s), ancestral to Baloch, whatever they were, influenced Levant BA at some degree, yes. Or both.
Just something to be considered. Not a refutation.... At the end, it's all about shared ancestry anyway. Both ways.

ED: correction.

[Pip]

Any result in well built calculators, correctly interpreted, may be meaningful of ancestry or shared ancestry. Notice that you're running ancient samples against modern samples/clusters, which are generally the recipients when there are older populations involved. Apparently this approach has its use, as we could notice in the similarity tools, for example. The results themselves may be informative, especially when analysed as a whole. But I'm affraid the clusters, in isolation, may become more complex here, now seen mainly as destinations rather than sources. If a certain calculator built for modern samples says a certain Neolithic farmer individual is 5% North Sea, then it doesn't mean that he has 5% of Scandinavian ancestry. Rather, it'll probably mean that Scandinavians must have received genetic influence either from Neolithic farmers and/or from related pop. So, increased Baloch may mean that some Levant BA-like pop and likely few or non-related to Levant Neo influenced genetically Baloch people. What do Levant BA had that Levant Neo lacked? Or that a third pop(s), ancestral to Baloch, whatever they were, influenced Levant BA at some degree, yes. Or both.
Just something to be considered. Not a refutation....

Yes, I agree. Sometimes finding similarities between mixed-dated samples can give clues, but this is why I generally prefer fitting only with combinations of prior-dated populations.

I have found the Levantine Bronze Age fits remarkably well, and fairly simply, with an addition of South Caucasus Kura-Araxes to the existing aDNA, which also seems to match the archaeology. I don't see any significant signs of separate Iranian admixture.

Despite K-A having R1b, it does not seem to have had significant EHG, and looks pretty much like CHGs catalysed by some mixing with Anatolian farmers. Although EHG-rich R1b-Z2103 was already present in the Northern Caucasus 2,750 BC, it seems this probably did not filter through into the Southern Caucasus mix until later in the Bronze Age (after K-A had spread South, or possibly contributing to K-A moving South). Z2103 phylogeny and autosomality would suggest a fairly early Armenian expansion - so possibly some time in the mid or late third millennium BC (Trialeti?) in admixture with quite a lot of other Georgian DNA. The main spread of EHG into the Levant looks to have occurred some time later.

[Regio X]

Yes, I agree. Sometimes finding similarities between mixed-dated samples can give clues

Yes, it can. In my example, the two results together are meaningful. The absence of Levant Neo in Baloch is a clue that the Levant BA is the recipient of a third pop, common to both, since Levant BA has Levant Neo influence, meaning Baloch should get some Levant Neo if Levant BA-like ancestry were a vector.
Unless the calculator is cheating us. Lol For instance, my mother-in-law gets 0% of Eastern Euro at K15. Baltic component "eats" all of them. Just odd. Anyway, it happens, and other clusters/clues may occasionally compensate.

[LeBrok]

Here is composition of Ancient Levant with today's residents and neighbors of Levant.

			S Indian	Baloch	Caucasian	NE Euro	SE Asian	Siberian	NE Asian	Papuan	American	Beringian	Mediterranean	SW Asian	San	E African	Pygmy	W African
Natufian					14%					1%			24%	57%		4%
Levant Neolithic, Jordan					26%								33%	40%		2%
Levant BA, Jordan				4%	37%		1%						12%	45%		0%
turk-istanbul	hodoglugil	20	1%	16%	45%	11%	1%	3%	3%	0%	0%	1%	11%	9%	0%	0%	0%	0%
assyrian	harappa	6	0%	19%	52%	1%	0%	0%	0%	0%	0%	0%	8%	18%	0%	0%	0%	0%
lebanese	behar	7	1%	11%	41%	3%	1%	1%	1%	0%	0%	0%	13%	23%	0%	3%	0%	1%
jordanian	behar	20	1%	9%	39%	1%	0%	0%	0%	0%	0%	0%	12%	28%	0%	5%	0%	2%
palestinian	hgdp	46	1%	7%	39%	1%	0%	0%	0%	0%	0%	0%	12%	31%	0%	5%	0%	2%
egyptian	behar	12	0%	3%	30%	0%	0%	0%	0%	0%	0%	0%	13%	33%	0%	12%	0%	6%
bedouin	hgdp	46	0%	5%	21%	2%	0%	0%	0%	0%	0%	0%	7%	56%	0%	5%	0%	3%
iraqi-arab	harappa	4	4%	19%	38%	2%	0%	1%	1%	1%	0%	1%	8%	22%	0%	2%	1%	1%

Seems like today's Jordaniens/Levant in general got more Iranian/Armenian (through Anatolia/Assyria?) and African (through Egypt?) influence since Bronze Age.

[Ygorcs]

Agamemmnon from Anthrogenica who is apparently as specialist of Afro-Asiatic languages, believe that ( something like that ) ; Semitic languages wich are genetically closer to Berbero-Lybian and Cushitic, was formed by the input of maybe an Ergative language, wich modern exemple would the closest to Northwest Caucasian, but ancestraly could be related with languages such as Elamit, Sumerian, Hurrian...

I didn't found the thread were he said that, but it was just a few times ago.

Fascinating, since Hurro-Urartian (before expanding southward) was just north of Upper Mesopotamia, and Sumerian just south of it. More and more I think those scholars who thought Semitic came from the Levant and preferably from the south (South Levant or even nearer to the Arabian peninsula) were wrong, and in fact Proto-Semitic was a "new" language that absorbed one of the earlier Afro-Asiatic languages of Natufian Levant (probably already subject to significant Anatolian_Neo influences) and changed it with an intensive foreign Iranian/CHG input. Genetically at least it's what makes most sense, and it also makes sense in the early attestation of the language family if we consider the possibility that its expansion was still going on, and the takeover of Sumer by Akkadians around 2400-2300 B.C. was part of that southward movement that would also have affected the South Levant and thence Arabia.

[Ygorcs]

An example is Jordanian BA, which best-fits with the addition of 22% Kalavan (NE Armenia early BA) and 10% Satsurbia (Georgia PHG) to a mix of Levantine Neolithic and some Natufian.

Curiously, Kalavan was at least partly R1b (a brother clade of P297).

[1] "distance%=0.2912 / distance=0.002912"

Kura-Araxes_Kalavan

CHG 54.0

Barcin_N 28.6

Abdul_Hosein_N 11.2

Natufian 6.2

But isn't Kalavan (2619-2465 B.C.) a bit too late to be a good proxy for the population that probably mixed with more Levantine-like people (Afro-Asiatic only makes sense, compared with all the African branches, if it came from the Levant and ultimately Natufian) and started to spread Proto-Semitic dialects? Proto-Semitic is dated to before ~3500 B.C. By ~2500 B.C. the Akkadians, already a specifically East Semitic branch, were already well known by the Sumerians and lived to their north and northwest. I assume the bulk of the ethnogenesis of that BA Levant population would've happened before Akkadians appear in the historic record, maybe even before 3000 B.C.

I keep getting a necessary mix of Iranian_Chalcolithic+CHG (maybe a much more CHG-enriched population with a lot of Iranian_Neolithic too?) as the main if I use only Chalcolithic and Neolithic samples, assuming that the Proto-Semitic-speaking population was already pretty much formed. Kalavan doesn't seem to fit that (too little Iran_Neo). Whatever, I guess that a so-called "very CHG-enriched population related to Iranian_Chalcolithic" would end up looking quite similar to Armenia_Kalavan, maybe just a bit more Iranian-like, so we both may be finding similar results in fact. Another population south of Armenia, closer to Upper Mesopotamia (and also to the Zagros) would fit that scenario perfectly. I just think we haven't found "the" right population (not that many samples from the Chalcolithic and EBA in that region) that introgressed into a much more Levant_Neo/Natufian-like population as it moved to the flatlands and formed the Proto-Semitic and early Semitic tribes which later contributed to other populations of the Levant, Mesopotamia and Arabia.

Using 10 dimensions (and I did include Armenia_Chl, Maykop, Hajji_Firuz_Chl and others):

[1] "distance%=0.2386 / distance=0.002386"

Levant_BA_South

Levant_ChL 38.85

Natufian 24.35

Seh_Gabi_ChL 18.90

CHG 10.55

Levant_N 7.35

Using more proximate 25 dimensions(often less reliable and realistic in my opinion, so I don't take its results too seriously), in one model adding EMBA Kalavan and in the other, not:

[1] "distance%=2.6335 / distance=0.026335"

Levant_BA_South

Levant_ChL 84.75

Tepe_Hissar_ChL 14.00

Kura-Araxes_Kalavan 1.25

[1] "distance%=2.2745 / distance=0.022745"

Levant_BA_South

Levant_ChL 23.7
Natufian 23.6
Seh_Gabi_ChL 21.6
Levant_N 20.9
Kura-Araxes_Kalavan 10.3

*** NOW WITHOUT KALAVAN...***

[1] "distance%=2.6339 / distance=0.026339"

Levant_BA_South

Levant_ChL 85.1
Tepe_Hissar_ChL 14.9

[1] "distance%=2.6727 / distance=0.026727"

Levant_BA_South

Levant_ChL 86.8
Namazga_Eneolithic 13.2


*** By the way, Tepe Hissar was like this, still mostly Iranian_Neolithic:

[1] "distance%=0.3705 / distance=0.003705"

Tepe_Hissar_ChL

Abdul_Hosein_N 72.8
Barcin_N 14.3
CHG 12.8


Let's not forget that much of the Levant was under Hurrian rule, but that was much later in the MLBA, nonetheless what matters is that it's possible that this CHG/Iranian input may have come with different ethnicities from different places, not all of it with the "primordial" Caucasus/Iran+Levant Semitic-speaking expansion that defined the post-BA Levant linguistically and culturally.

[markod]

But isn't Kalavan (2619-2465 B.C.) a bit too late to be a good proxy for the population that probably mixed with more Levantine-like people (Afro-Asiatic only makes sense, compared with all the African branches, if it came from the Levant and ultimately Natufian) and started to spread Proto-Semitic dialects? Proto-Semitic is dated to before ~3500 B.C. By ~2500 B.C. the Akkadians, already a specifically East Semitic branch, were already well known by the Sumerians and lived to their north and northwest. I assume the bulk of the ethnogenesis of that BA Levant population would've happened before Akkadians appear in the historic record, maybe even before 3000 B.C.

Why not derive AA directly from Africa? Natufian wouldn't be needed, and indeed it seems as though they suffered some male-biased replacement in the later Neolithic.

[Pip]

[1] "distance%=0.2912 / distance=0.002912"

Kura-Araxes_Kalavan

CHG 54.0

Barcin_N 28.6

Abdul_Hosein_N 11.2

Natufian 6.2

But isn't Kalavan (2619-2465 B.C.) a bit too late to be a good proxy for the population that probably mixed with more Levantine-like people (Afro-Asiatic only makes sense, compared with all the African branches, if it came from the Levant and ultimately Natufian) and started to spread Proto-Semitic dialects? Proto-Semitic is dated to before ~3500 B.C. By ~2500 B.C. the Akkadians, already a specifically East Semitic branch, were already well known by the Sumerians and lived to their north and northwest. I assume the bulk of the ethnogenesis of that BA Levant population would've happened before Akkadians appear in the historic record, maybe even before 3000 B.C.

I keep getting a necessary mix of Iranian_Chalcolithic+CHG (maybe a much more CHG-enriched population with a lot of Iranian_Neolithic too?) as the main if I use only Chalcolithic and Neolithic samples, assuming that the Proto-Semitic-speaking population was already pretty much formed. Kalavan doesn't seem to fit that (too little Iran_Neo). Whatever, I guess that a so-called "very CHG-enriched population related to Iranian_Chalcolithic" would end up looking quite similar to Armenia_Kalavan, maybe just a bit more Iranian-like, so we both may be finding similar results in fact. Another population south of Armenia, closer to Upper Mesopotamia (and also to the Zagros) would fit that scenario perfectly. I just think we haven't found "the" right population (not that many samples from the Chalcolithic and EBA in that region) that introgressed into a much more Levant_Neo/Natufian-like population as it moved to the flatlands and formed the Proto-Semitic and early Semitic tribes which later contributed to other populations of the Levant, Mesopotamia and Arabia.

Using 10 dimensions (and I did include Armenia_Chl, Maykop, Hajji_Firuz_Chl and others):

[1] "distance%=0.2386 / distance=0.002386"

Levant_BA_South

Levant_ChL 38.85

Natufian 24.35

Seh_Gabi_ChL 18.90

CHG 10.55

Levant_N 7.35

Using more proximate 25 dimensions(often less reliable and realistic in my opinion, so I don't take its results too seriously), in one model adding EMBA Kalavan and in the other, not:

[1] "distance%=2.6335 / distance=0.026335"

Levant_BA_South

Levant_ChL 84.75

Tepe_Hissar_ChL 14.00

Kura-Araxes_Kalavan 1.25

[1] "distance%=2.2745 / distance=0.022745"

Levant_BA_South

Levant_ChL 23.7
Natufian 23.6
Seh_Gabi_ChL 21.6
Levant_N 20.9
Kura-Araxes_Kalavan 10.3

*** NOW WITHOUT KALAVAN...***

[1] "distance%=2.6339 / distance=0.026339"

Levant_BA_South

Levant_ChL 85.1
Tepe_Hissar_ChL 14.9

[1] "distance%=2.6727 / distance=0.026727"

Levant_BA_South

Levant_ChL 86.8
Namazga_Eneolithic 13.2


*** By the way, Tepe Hissar was like this, still mostly Iranian_Neolithic:

[1] "distance%=0.3705 / distance=0.003705"

Tepe_Hissar_ChL

Abdul_Hosein_N 72.8
Barcin_N 14.3
CHG 12.8


Let's not forget that much of the Levant was under Hurrian rule, but that was much later in the MLBA, nonetheless what matters is that it's possible that this CHG/Iranian input may have come with different ethnicities from different places, not all of it with the "primordial" Caucasus/Iran+Levant Semitic-speaking expansion that defined the post-BA Levant linguistically and culturally.

Good post.

Yes, Kalavan is probably too late to be ancestral to the Levantine immigrants, but I think it still represents a reasonable proxy for them. I would just say that the aDNA that first admixed into the Levant looks to have been mostly from a Kalavan-like population, by my calculations more like Kalavan that other earlier Armenian or Iranian samples. To be more precise, it looks to be from a Kalavan-like population skewed towards mainstream Iranian, so might well be closer still to Chalcolithic NW Iran.

The Kura-Araxes culture appears to have spread into Iran before moving across to the Levant, so it would probably have picked up some Iranian admixture around Van/Urmia and perhaps moved on further from there. This could still identify Kura-Araxes as the main culture bringing these people into the Levant.

It might also be the case that this incoming population moved in gradually over a long period, as later Levantine populations have a higher proportion of its DNA; but difficult to tell with fairly few samples. If this were a gradual process, I suppose the incoming population might have assimilated an early Semitic language already spoken in the Levant.

Yes, the Hurrians came later, probably related people from a similar direction, but most likely with additional EHG from North of the Caucasus.

[Ygorcs]

Why not derive AA directly from Africa? Natufian wouldn't be needed, and indeed it seems as though they suffered some male-biased replacement in the later Neolithic.

Oh I probably expressed myself in an ambiguous way. My assumption is that AA came directly from Africa, but Semitic and possibly also Egyptian (and Cushitic perhaps too?) came from Natufians who in turn got North African ancestry probably via the Kebaran and ultimately from the Halfan/Khubanniyan (recent analyses have suggested that it wasn't that Iberomaurusians had a lot of Natufian ancestry, it was actually that both of them had a lot of North African West Eurasian-derived ancestry). Then the only thing clearly connecting West Asian AA (Semitic) and the AA branches is Natufians, not Anatolians, Iranians or Caucasians, so probably AA was picked up by those CHG/Iran-rich people in the Levant or in a heavily "Levantinized" population (which I believe lived in Upper Mesopotamia near the Zagros and Armenian/Kurdish Highlands). It's just a matter of chronology. I don't think all AA branches came directly from the Proto-AA source in a star-shaped expansion with no posterior connections, but rather from distinct later offshoots derived ultimately from that Proto-AA population.