PDA

View Full Version : [split] Yazidis & co.: A Mitanni Signal?



pegasus
06-06-2020, 03:34 PM
Alkaevli

Overfitting is usually a situation when You have too much similar source populations. Which create an overfit. Here it is not the case.
This type of a imbalances usually occur when an important population is missing.

In Iranians case it could be for example a Chaff face ware period sample prior KA. Or a Khabur ware sample from MBA. Those cultures are related to North Mesopotamia.

In this particular case an addition of East Asian pop from Mongolia changes the proportions.

Target: Kurdish
Distance: 1.1347% / 0.01134660
27.6 SYR_Ebla_EMBA
22.6 TKM_Gonur1_BA
11.6 RUS_Sintashta_MLBA
10.2 IRN_Seh_Gabi_C
9.2 IRN_Hajji_Firuz_C
8.8 TUR_Arslantepe_EBA
6.6 Kura-Araxes_ARM_Talin
2.0 Yamnaya_RUS_Samara
1.4 MNG_Hovsgol_BA
0.0 RUS_Darkveti-Meshoko_En
0.0 TUR_Kaman-Kalehoyuk_MLBA

And here what happens when Azerbaijan LC is added. I didn't check it but it must be a Chaff faced eare period.

Target: Kurdish
Distance: 1.0278% / 0.01027815
26.0 TKM_Gonur1_BA
23.6 SYR_Ebla_EMBA
12.2 IRN_Hajji_Firuz_C
11.6 TUR_Arslantepe_EBA
8.8 AZE_Caucasus_lowlands_LC
8.8 RUS_Sintashta_MLBA
4.2 Kura-Araxes_ARM_Talin
3.2 Yamnaya_RUS_Samara
1.6 MNG_Hovsgol_BA
0.0 IRN_Seh_Gabi_C
0.0 RUS_Darkveti-Meshoko_En
0.0 TUR_Kaman-Kalehoyuk_MLBA

The most Parsimonious models for Ezids I got was this, I think other Kurds esp those in Turkey will show more diversity and show a combination of Hittite/Luwian and Indo Iranian ancestry. Though in both our models it suggests there was a massive movement from Central Asia to this region.

Target: Ezid:EZI-010
Distance: 1.4403% / 0.01440266
52.2 SYR_Ebla_EMBA
26.4 UZB_Sappali_Tepe_BA
21.4 KAZ_Dali_MLBA


Target: Ezid
Distance: 0.4908% / 0.00490822
51.6 SYR_Ebla_EMBA
31.0 UZB_Sappali_Tepe_BA
17.4 KAZ_Dali_MLBA



"sample": "Kurdish:Average",
"distance": 1.1331,
"Alalakh_MLBA": 50,
"Sappali_Tepe_BA": 36,
"Oy_Dzhaylau_MLBA": 14




With genoplot

DMXX
06-06-2020, 05:06 PM
Very good fits, but I really doubt if there is any significant Hittite or Luwian ancestry in Kurds. I think they are more relevant to Anatolian Turks, Greeks, Syrians maybe followed by Armenians and Assyrians. I think Kurds are mostly Haji Firuz related with additional BMAC + Steppe.

I agree.

I think the combo of Ebla_EMBA and Sappali Tepe just so happens to approximate - By chance - the sort of agriculturalist admixture that typifies West Iranians in the absence of a reliable BA proxy.

Ebla_EMBA is, roughly speaking, about 75% Levant-Anatolia_N and 20% Iran_N. In turn, Sappali looks overwhelmingly Parkhai_EN derived, which in turn is mostly Iran_N related.

So, really, that statistical fit is just telling us that the underlying "base agriculturalist" population for groups like Kurds (and the others) is an Iran_N shifted variant of Ebla_EMBA... Which is basically a continuation of Hajji Firuz (45-60% Levant-Anatolia_N, the rest Iran).

pegasus
06-06-2020, 06:57 PM
Very good fits, but I really doubt if there is any significant Hittite or Luwian ancestry in Kurds. I think they are more relevant to Anatolian Turks, Greeks, Syrians maybe followed by Armenians and Assyrians. I think Kurds are mostly Haji Firuz related with additional BMAC + Steppe.

I agree, it was just my assumption for Kurds living in the Anatolian Highlands.
Ezids are fascinating since they are highly endogenous, based on those outliers at Megiddo they seem to harbor ancestry from Mittani and Median related peoples. Ezids model exceedingly well with those Mittani outlier populations it proves a surge of Steppe MLBA and BMAC ancestry was already present in the MLBA. Though the same model does not work for the main Kurd cluster on Nmonte.




"sample": "Ezid:Average",
"distance": 1.0555,
"Levant_Megiddo_MLBA_o1": 45.5,
"Alalakh_MLBA_o": 30.5,
"Ebla_EMBA": 24


"sample": "Kurdish:Average",
"distance": 1.612,
"Levant_Megiddo_MLBA_o1": 36,
"Ebla_EMBA": 32.5,
"Alalakh_MLBA_o": 31.5


"sample": "Kurdish:Average",
"distance": 0.9846,
"Alalakh_MLBA": 35.5,
"Sappali_Tepe_BA": 27.5,
"Hajji_Firuz_C": 24,
"Oy_Dzhaylau_MLBA": 13



"sample": "Ezid:Average",
"distance": 1.8713,
"Hajji_Firuz_C": 60.5,
"Seh_Gabi_C": 18,
"Dali_MLBA": 17,
"Sappali_Tepe_BA": 4.5


"sample": "Ezid:Average",
"distance": 1.3299,
"Ebla_EMBA": 46,
"Sappali_Tepe_BA": 28,
"Dali_MLBA": 14.5,
"Seh_Gabi_C": 11.5


"sample": "Ezid:Average",
"distance": 1.0498,
"Ebla_EMBA": 38,
"Sappali_Tepe_BA": 22,
"Seh_Gabi_C": 21.5,
"Oy_Dzhaylau_MLBA": 18.5

Magnetic
06-06-2020, 07:09 PM
The most Parsimonious models for Ezids I got was this, I think other Kurds esp those in Turkey will show more diversity and show a combination of Hittite/Luwian and Indo Iranian ancestry. Though in both our models it suggests there was a massive movement from Central Asia to this region.



I dont think that we have central asian ancestry . it is rather that iran neo moved towards central asia and mixed with north/east euro hg types . and we have our steppe most likely from southern russia area . it wouldnt make any sense for iran neo to move to central asia and then move back to our habitated lands .

SampleFitTepecik Ciftlik NGanj Dareh NRUS CatacombOroqenAGUsers►Magnetic2.878549.53314.53

Johnny ola
06-06-2020, 07:11 PM
We can say that Kurds have assilimated armenians/assyrians???

Magnetic
06-06-2020, 07:13 PM
I thought this thread is about armenians

pegasus
06-06-2020, 07:34 PM
I dont think that we have central asian ancestry . it is rather that iran neo moved towards central asia and mixed with north/east euro hg types . and we have our steppe most likely from southern russia area . it wouldnt make any sense for iran neo to move to central asia and then move back to our habitated lands .

SampleFitTepecik Ciftlik NGanj Dareh NRUS CatacombOroqenAGUsers►Magnetic2.878549.53314.53

To the contrary , it strongly looks like some population on a gradient between TKM IA and Sappali Tepe outlier is bringing Steppe ancestry, in that case, you will see Iran_N related ancestry rise which is what you see with Hasanlu. Actually most of the Steppe ancestry looks like the Yaz type so it is Steppe MLBA related not Yamnaya, I think as Mountain pointed out Yamnaya/Steppe EMBA is more applicable to non Iranic populations.

Magnetic
06-06-2020, 07:37 PM
To the contrary , it strongly looks like some population on a gradient between TKM IA and Sappali Tepe outlier is bringing Steppe ancestry, in that case, you will see Iran_N related ancestry rise which is what you see with Hasanlu. Actually most of the Steppe ancestry looks like the Yaz type so it is Steppe MLBA related not Yamnaya, I think as Mountain pointed out Yamnaya/Steppe EMBA is more applicable to non Iranic populations.

but that doesnt make any sense . if sappali tepe brought steppe to us . how is sappali tepe Iran N rich then ? :D did iran n walk there ...made picnic with steppe people and moved back to anatolian N and mixed with them to give birth to us ? hehe

DMXX
06-06-2020, 07:46 PM
"sample": "Ezid:Average",
"distance": 1.0555,
"Levant_Megiddo_MLBA_o1": 45.5,
"Alalakh_MLBA_o": 30.5,
"Ebla_EMBA": 24


As wizardly as this model appears to be, it doesn't make a whole lot of historical or geographical sense. Megiddo and Ebla are both distally shifted from the populations that are currently considered as Kurdish - Particularly the Yazidis from Armenia. A back-migration of Indo-Aryans from Megiddo or forward migration of ?Hittites from Alalakh into the SE Turkey-Armenia-N Iraq region... That's reaching.

IMO, as is the case with the Ebla+Sappali pairing for W. Iranian-related pops in general, this looks like another case of chance "goodness of fit" owing to the inclusion of accidentally perfect ingredients. Having said that, stranger things have happened in the world of population genetics - Could the Yazidis be "Iranicised" descendants of a Mitanni-Hittite-Kassite-like hybrid population... Maybe?

One wrench in that proposal's machinery is the observation that the Talysh of Azerbaijan form a genetic continuum of sorts with the Yazidi. By extension, that'd implicate most Kurds, Azeris, Lors, Hasanlu_IA etc. as being a product of a similar ancient multi-Indo-European-speaking blend. That conclusion is special pleading of a unique kind that I haven't seen since the good ol' days of overreach, when Balochis were described as being "90% Kurdish".

DMXX
06-06-2020, 07:50 PM
but that doesnt make any sense . if sappali tepe brought steppe to us . how is sappali tepe Iran N rich then ? :D did iran n walk there ...made picnic with steppe people and moved back to anatolian N and mixed with them to give birth to us ? hehe

Yes, actually.

Azeris and Kurds are either more Iran_N-shifted than our Chalcolithic predecessors were (Hajji Firuz), or, have the same amount (despite having significant admixture from outside of the region).

As one progresses from the Chalcolithic through to the Iron Age (Hasanlu) through to the present, three consistent patterns are observed:

1. Decrease in Levant_N admixture
2. Increase in Iran_N admixture
3. Increase in EHG-related admixture

The latter two ostensibly suppressed the former, which is in line with a hybrid MLBA steppe-BMAC population (represented by something resembling Turkmenistan_IA/Sintashta plus Sappali Tepe) making a hefty contribution to the genepool of the nascent regional "family" that gave rise to Azeris, Kurmanji Kurds, the Talysh and Lors.

[edit]: Sappali Tepe was predominantly Iran_N, per Narasimhan et al. - All of the Central Asian/Turanian agriculturalists were predominantly so (though culture-specific variation in the frequencies of Iran_N, Anatolia_N and WSHG were reported).

[edit 2]: Correction, remembered the outcome of a recent model I'd tried...

Magnetic
06-06-2020, 07:53 PM
Yes, actually.

Azeris and Kurds are more Iran_N-shifted than our Chalcolithic predecessors were (Hajji Firuz).

As one progresses from the Chalcolithic through to the Iron Age (Hasanlu) through to the present, three consistent patterns are observed:

1. Decrease in Levant_N admixture
2. Increase in Iran_N admixture
3. Increase in EHG-related admixture

The latter two ostensibly suppressed the former, which is in line with a hybrid MLBA steppe-BMAC population (represented by something resembling Turkmenistan_IA/Sintashta plus Sappali Tepe) making a hefty contribution to the genepool of the nascent regional "family" that gave rise to Azeris, Kurmanji Kurds, the Talysh and Lors.

SampleFitSappali Tepe BAGRC MycenaeanOroqenAGUsers►Magnetic2.737251.546.52

DMXX
06-06-2020, 07:59 PM
SampleFitSappali Tepe BAGRC MycenaeanOroqenAGUsers►Magnetic2.737251.546.52

You clearly won't have ancestry from the Mycenaeans.

Per Genoplot (https://genoplot.com/g25):



[
{
"sample": "AGUsers:Magnetic",
"distance": 2.5072,
"Ebla_EMBA": 54,
"Sappali_Tepe_BA": 29,
"Sintashta_MLBA": 13,
"Hovsgol_BA": 4
}
]


Removing Sappali Tepe (your current position):



[
{
"sample": "AGUsers:Magnetic",
"distance": 5.2985,
"Ebla_EMBA": 73.5,
"Sintashta_MLBA": 20.5,
"Hovsgol_BA": 6
}
]


Humouring the proposal that you actually just need Hajji Firuz added:



[
{
"sample": "AGUsers:Magnetic",
"distance": 3.0188,
"Hajji_Firuz_C": 73,
"Sintashta_MLBA": 13,
"Ebla_EMBA": 9,
"Hovsgol_BA": 5
}
]


It's quite obvious that you (as is the case with practically every Iranian-related pop in W Asia outside of the Ossetians) require a BMAC source.

Magnetic
06-06-2020, 08:09 PM
You clearly won't have ancestry from the Mycenaeans.

Per Genoplot (https://genoplot.com/g25):



[
{
"sample": "AGUsers:Magnetic",
"distance": 2.5072,
"Ebla_EMBA": 54,
"Sappali_Tepe_BA": 29,
"Sintashta_MLBA": 13,
"Hovsgol_BA": 4
}
]


Removing Sappali Tepe (your current position):



[
{
"sample": "AGUsers:Magnetic",
"distance": 5.2985,
"Ebla_EMBA": 73.5,
"Sintashta_MLBA": 20.5,
"Hovsgol_BA": 6
}
]


Humouring the proposal that you actually just need Hajji Firuz added:



[
{
"sample": "AGUsers:Magnetic",
"distance": 3.0188,
"Hajji_Firuz_C": 73,
"Sintashta_MLBA": 13,
"Ebla_EMBA": 9,
"Hovsgol_BA": 5
}
]


It's quite obvious that you (as is the case with practically every Iranian-related pop in W Asia outside of the Ossetians) require a BMAC source.

why not ?

I wanted to use some anatolian reference but couldnt find it on genoplot

what is ebla ?

hajji firuz has Anatolian , Iran N , and if I am not wrong a little CHG in it . I used sappali tepe so I used bmac . ?!

I tried to model with 2 pops one being bmac one mediterranean

my modelling has a better fit than your last one . I dont know what ebla is

DMXX
06-06-2020, 08:15 PM
why not ?


You're (heritage-wise) a Turkish Kurd (ergo ancestry from C to SE Anatolia).

The Mycenaeans were Iron Age folk who preceded the classical Greeks.

There's no historical mechanism through which you could be 20% Mycenaean.

Reminds me of that one time I (somehow) managed to get a GD ~2.5 using modern English people as a ref pop for myself.



I wanted to use some anatolian reference but couldnt find it on genoplot


Doesn't look like the meta-data for that has been added just yet. The samples are there though. Look for Titris f.ex.



my modelling has a better fit than yours

No. Check the first model I posted.

Magnetic
06-06-2020, 08:21 PM
You're (heritage-wise) a Turkish Kurd (ergo ancestry from C to SE Anatolia).

yes . according to 23andme I also have ancestry from west and northwest anatolia . I dont know how accurate that is though



The Mycenaeans were Iron Age folk who preceded the classical Greeks.

There's no historical mechanism through which you could be 20% Mycenaean.

not 20% but 47% ^_^

of course I know that I dont have mycenaen ancestry . I just asked you why because I was curious to your answer . but the mycenaean is a good proxy for the native mediterranean in us nevertheless .

Italy_Tivoli Renaissance gives a decent fit too

SampleFitTivoli RenaissanceSappali Tepe BAOroqenAGUsers►Magnetic2.81095345.51.5



Reminds me of that one time I (somehow) managed to get a GD ~2.5 using modern English people as a ref pop for myself.


how ? :D






No. Check the first model I posted.

yeah sorry I meant the third one . I edited my post but you already quote my previous

DMXX
06-06-2020, 08:27 PM
Behold:

[
{
"sample": "AGUsers:DMXX",
"distance": 2.3948,
"Hajji_Firuz_C": 46,
"Sappali_Tepe_BA": 28,
"English": 20.5,
"Korean": 5.5
}
]

Magnetic
06-06-2020, 08:28 PM
Behold:

[
{
"sample": "AGUsers:DMXX",
"distance": 2.3948,
"Hajji_Firuz_C": 46,
"Sappali_Tepe_BA": 28,
"English": 20.5,
"Korean": 5.5
}
]

ah it is because of the EHG / steppe related stuff that is masked as english in this case I guess

lol

Magnetic
06-06-2020, 08:29 PM
u wot m8

SampleFitHajji Firuz CSappali Tepe BAEnglishOroqenAGUsers►Magnetic2.564160.51917.53

Johnny ola
06-06-2020, 08:31 PM
yes . according to 23andme I also have ancestry from west and northwest anatolia . I dont know how accurate that is though



not 20% but 47% ^_^

of course I know that I dont have mycenaen ancestry . I just asked you why because I was curious to your answer . but the mycenaean is a good proxy for the native mediterranean in us nevertheless .

Italy_Tivoli Renaissance gives a decent fit too

SampleFitTivoli RenaissanceSappali Tepe BAOroqenAGUsers►Magnetic2.81095345.51.5



how ? :D




yeah sorry I meant the third one . I edited my post but you already quote my previous



Try to use a MBA or Copper Age anatolian sample.Not even me as a Pontic Greek myself and i got not more than 5% admixture from Myceneans and you expect Kurds to have xD?Your anatolia N shift indicates Native Anatolian(hattic-hittite) ancestry witch is decent among Kurdish people.

Magnetic
06-06-2020, 08:34 PM
Try to use a MBA or Copper Age anatolian sample.Not even me as a Pontic Greek myself and i got not more than 5% admixture from Myceneans and you expect Kurds to have xD?Your anatolia N shift indicates Native Anatolian(hattic-hittite) ancestry witch is decent among Kurdish people.

how can you not have more than 5% ? my modelling shows 47% for me . I am confused

how do you know you dont have any ? I mean yes I dont have any I know . I just used it because I didnt find the anatolian references . it seems that the "Anatolia" is removed and only the rest of the names are there lol . kaman kalehöyük fits better . but the mycanaean works well as a med proxy still . and I am expecting you to score more than me in the modelling

Johnny ola
06-06-2020, 08:44 PM
The Mycenean samples have very bad quality IMO and low coverage.If you want to use a Med sample try the: Iberia_Northeast_Empuries2

It is from Greek colonizers in Catalonia Spain.They are almost identical to Myceneans.But IMO Kurds and the majority of west asian's Barcin-ANF ancestry is coming from Hattic-Hittites-Hurrians etc.Armenians,Assyrians,laz,Pontic Greeks,Kurds,Turks are all ANF admixed.You can also use the recent samples from Arslantepe and Ikitzepe as well.

Look my model for example.As a Pontic Greek i am using both the Empuries sample and the Ikitzepe(witch is from Samsun city).

Target: Ioannis_scaled
Distance: 1.5780% / 0.01578004
40.2 TUR_Ikiztepe_LC
32.4 Kura-Araxes_ARM_Kaps
16.2 Levant_Baqah_BA
11.2 Iberia_Northeast_Empuries2


My mother for example is not that far away genetically from Kurds.She got Kurdish on her list(G25 averages).We share pretty much the same components..thought in different amounts!!!!

Distance to: Theodora_scaled
0.02635265 Turkish_Trabzon
0.02706294 Greek_Trabzon
0.03125680 Armenian_Hemsheni
0.03400412 Armenian
0.04149149 Georgian_Laz
0.04284557 Greek_Central_Anatolia
0.04378576 Assyrian
0.04587846 Georgian_Jew
0.05388215 Iraqi_Jew
0.05696881 Druze
0.05771673 Iranian_Jew
0.05843571 Cypriot
0.05848924 Lebanese_Druze
0.06193891 Mountain_Jew_Chechnya
0.06303160 Georgian_Imer
0.06387379 Lebanese_Christian
0.06396132 Syrian_Jew
0.06457593 Karaite_Egypt
0.06520764 Lebanese_Muslim
0.06600926 Greek_Kos
0.06664057 Turkish_Kayseri
0.06989302 Kurdish
0.07218288 Palestinian_Beit_Sahour
0.07267641 Talysh_Azerbaijan
0.07289398 Syrian

Magnetic
06-06-2020, 08:49 PM
SampleFitSappali Tepe BAIberia Northeast Empuries2OroqenAGUsers►Magnetic2.75845245.52.5

well it is no surprise since you are geographically close people . I also get Pontic Greeks at 0.06478491 in my list . but we have noticable differences still . we have more Iran N and more Steppe while you are more CHG

why do you use kura araxes along with ikiztepe etc. ?

kura araxes already has anatolian etc in it

Johnny ola
06-06-2020, 08:56 PM
SampleFitSappali Tepe BAIberia Northeast Empuries2OroqenAGUsers►Magnetic2.75845245.52.5

well it is no surprise since you are geographically close people . I also get Pontic Greeks at 0.06478491 in my list

why do you use kura araxes along with ikiztepe etc. ?

kura araxes already has anatolian etc in it

I know.All the transcaucasian cultures(Kura-Araxes,Maykop) etc they have a decent amount of Anatolian N admixture.But when i am modeling my self i need both a transcaucasian sample(i am getting good fits with both K.A and Maykop) and an Anatolian Copper age/Bronze age sample.If i want a good distance i need to put a kura-araxes sample.I also need a Greek and Jewish sample to model my self because of my Levant/EEF shift.My ancestry with a few words has to do with a colchian-Kartvelian(similar to laz-Georgian),a native anatolian population(hattian-hittite) and East Med ethnic groups like Jews and Greeks.

Oreo
06-06-2020, 09:32 PM
The Mycenean samples have very bad quality IMO and low coverage.If you want to use a Med sample try the: Iberia_Northeast_Empuries2

It is from Greek colonizers in Catalonia Spain.They are almost identical to Myceneans.But IMO Kurds and the majority of west asian's Barcin-ANF ancestry is coming from Hattic-Hittites-Hurrians etc.Armenians,Assyrians,laz,Pontic Greeks,Kurds,Turks are all ANF admixed.You can also use the recent samples from Arslantepe and Ikitzepe as well.

Look my model for example.As a Pontic Greek i am using both the Empuries sample and the Ikitzepe(witch is from Samsun city).

Target: Ioannis_scaled
Distance: 1.5780% / 0.01578004
40.2 TUR_Ikiztepe_LC
32.4 Kura-Araxes_ARM_Kaps
16.2 Levant_Baqah_BA
11.2 Iberia_Northeast_Empuries2


My mother for example is not that far away genetically from Kurds.She got Kurdish on her list(G25 averages).We share pretty much the same components..thought in different amounts!!!!

Distance to: Theodora_scaled
0.02635265 Turkish_Trabzon
0.02706294 Greek_Trabzon
0.03125680 Armenian_Hemsheni
0.03400412 Armenian
0.04149149 Georgian_Laz
0.04284557 Greek_Central_Anatolia
0.04378576 Assyrian
0.04587846 Georgian_Jew
0.05388215 Iraqi_Jew
0.05696881 Druze
0.05771673 Iranian_Jew
0.05843571 Cypriot
0.05848924 Lebanese_Druze
0.06193891 Mountain_Jew_Chechnya
0.06303160 Georgian_Imer
0.06387379 Lebanese_Christian
0.06396132 Syrian_Jew
0.06457593 Karaite_Egypt
0.06520764 Lebanese_Muslim
0.06600926 Greek_Kos
0.06664057 Turkish_Kayseri
0.06989302 Kurdish
0.07218288 Palestinian_Beit_Sahour
0.07267641 Talysh_Azerbaijan
0.07289398 Syrian

What is Ikitzepe sample? A native Anatolian? Where does it plot?

pegasus
06-07-2020, 06:45 AM
As wizardly as this model appears to be, it doesn't make a whole lot of historical or geographical sense. Megiddo and Ebla are both distally shifted from the populations that are currently considered as Kurdish - Particularly the Yazidis from Armenia. A back-migration of Indo-Aryans from Megiddo or forward migration of ?Hittites from Alalakh into the SE Turkey-Armenia-N Iraq region... That's reaching.

IMO, as is the case with the Ebla+Sappali pairing for W. Iranian-related pops in general, this looks like another case of chance "goodness of fit" owing to the inclusion of accidentally perfect ingredients. Having said that, stranger things have happened in the world of population genetics - Could the Yazidis be "Iranicised" descendants of a Mitanni-Hittite-Kassite-like hybrid population... Maybe?

One wrench in that proposal's machinery is the observation that the Talysh of Azerbaijan form a genetic continuum of sorts with the Yazidi. By extension, that'd implicate most Kurds, Azeris, Lors, Hasanlu_IA etc. as being a product of a similar ancient multi-Indo-European-speaking blend. That conclusion is special pleading of a unique kind that I haven't seen since the good ol' days of overreach, when Balochis were described as being "90% Kurdish".

Yes, I feel the BMAC component is a bit inflated, we need BA genomes from Western Iran that would account for that excess Iran_N without inflating BMAC values though I don't feel it will drastically change, in that final Ezidi model , the BMAC;Steppe ratios approximate to a TKM IA like population, and the BMAC component went down 6% . The Megiddo outlier does not model with Megiddo it prefers Ebla. Also, the Mittani outlier model fails with Kurds it works with Ezids. By the Ezidi's own accord some claim to have come from the Rojava area in Northern Syria, how accurate that is remains to be seen but the model supports that and this is Mittani territory. Ezidis tend to have more Steppe ancestry as well compared to the main Kurdish cluster. I will look at Talysh people and compare with Ezidis

DMXX
06-07-2020, 06:29 PM
Yes, I feel the BMAC component is a bit inflated, we need BA genomes from Western Iran that would account for that excess Iran_N without inflating BMAC values though I don't feel it will drastically change, in that final Ezidi model , the BMAC;Steppe ratios approximate to a TKM IA like population, and the BMAC component went down 6% .


IMO, in the absence of suitable genomes, it's probably sensible at this point to tentatively presume that the various Indo-Iranian waves which reached the Zagros were between a 50:50 (TRK_IA) or 60:40 (Megiddo_o1's non-Levant admix) of BMAC:MLBA steppe, ancestry wise.

All Iranians and Kurds seem to show a larger proportion of BMAC-related ancestry if that ratio is applied, which is in accordance with your ~6% value for Yazidis. We may consider that surplus as evidence of the pre-II interactions (f.ex. BMAC traders).

The confounder here for populations like the Lurs, Feylis and Persians from Fars province is they may well harbour Elamite ancestry, which in all likelihood was rich in Iran_N-related admix.
Observe this unusually good model for Lurs:



[
{
"sample": "Iranian Lor:Average",
"distance": 1.2307,
"Ezid": 68.5,
"Seh_Gabi_C": 16.5,
"Talysh_Azerbaijan": 15
}
]


Ergo, the above "soft rule" probably applies mostly to NW-N Iranian-derived populations (Talysh, Iranian Azeris, Yazidis, other Kurmanji Kurds, Zazas, potentially the Sorani Kurds, things become shaky once one hits Kermanshah province IMO).

That'd roughly place the sum Iron Age Indo-Iranian contribution to the first five pops described above as being between 30-50% overall.



The Megiddo outlier does not model with Megiddo it prefers Ebla.


More evidence in favour of the Indo-Aryan Mitanni origin for this sample, I'd wager.



Also, the Mittani outlier model fails with Kurds it works with Ezids. By the Ezidi's own accord some claim to have come from the Rojava area in Northern Syria, how accurate that is remains to be seen but the model supports that and this is Mittani territory.


It could well be the case that Yazidis specifically were a Hurrian-Mitanni hybrid population that became Iranicized at a later time (or "Kurmanjified", if that's a word).

Would certainly explain the highly unusual (and potentially syncretic) nature of their religion.

I'm quite open to this being a possibility. Though, an explanation as to why they superficially form a "continuum" of sorts with the Talysh requires explaining.



Ezidis tend to have more Steppe ancestry as well compared to the main Kurdish cluster.


Yes, they seem to represent the top half shelf of the Kurmanji Kurdish-Iranian Azeri range for that.

The Yazidis most certainly aren't to KK's and IA's as what Pamiris are to Pashtuns or the Kalash - The highest-scoring Yazidis are in the same general bracket as the highest-scoring non-Yazidi Kurmanji Kurds (e.g. StarDS9), Azeris (several, including the halfie yours truly), Zoroastrians and the like.



I will look at Talysh people and compare with Ezidis

Please do. Show us more of your wizardry (no sarcasm, you produce superb models on a frequent basis).

pegasus
06-07-2020, 07:12 PM
IMO, in the absence of suitable genomes, it's probably sensible at this point to tentatively presume that the various Indo-Iranian waves which reached the Zagros were between a 50:50 (TRK_IA) or 60:40 (Megiddo_o1's non-Levant admix) of BMAC:MLBA steppe, ancestry wise.

All Iranians and Kurds seem to show a larger proportion of BMAC-related ancestry if that ratio is applied, which is in accordance with your ~6% value for Yazidis.

The confounder here for populations like the Lurs, Feylis and Persians from Fars province is they may well harbour Elamite ancestry, which in all likelihood was rich in Iran_N-related admix.
Observe this unusually good model for Lurs:



[
{
"sample": "Iranian Lor:Average",
"distance": 1.2307,
"Ezid": 68.5,
"Seh_Gabi_C": 16.5,
"Talysh_Azerbaijan": 15
}
]


Ergo, the above "soft rule" probably applies mostly to NW-N Iranian-derived populations (Talysh, Iranian Azeris, Yazidis, other Kurmanji Kurds, Zazas, potentially the Sorani Kurds, things become shaky once one hits Kermanshah province IMO).

That'd roughly place the sum Iron Age Indo-Iranian contribution to the first five pops described above as being between 30-50% overall.



More evidence in favour of the Indo-Aryan Mitanni origin for this sample, I'd wager.



It could well be the case that Yazidis specifically were a Hurrian-Mitanni hybrid population that became Iranicized at a later time (or "Kurmanjified", if that's a word).

Would certainly explain the highly unusual (and potentially syncretic) nature of their religion.

I'm quite open to this being a possibility. Though, an explanation as to why they superficially form a "continuum" of sorts with the Talysh requires explaining.



Yes, they seem to represent the top half shelf of the Kurmanji Kurdish-Iranian Azeri range for that.

The Yazidis most certainly aren't to KK's and IA's as what Pamiris are to Pashtuns or the Kalash - The highest-scoring Yazidis are in the same general bracket as the highest-scoring Kurds (e.g. StarDS9), Azeris (several, including the halfie yours truly), Zoroastrians and the like.



Please do. Show us more of your wizardry.

Talysh have significant Caucasus ancestry which is absent in the Ezidi. I think the continuum between Talysh and Ezidis is because the CHG is compensating for the relative lack of Iran compared with Ezidis and conversely the excess Iran_N in Ezidis compensates for the lack of CHG relative to Talysh so they kind of cancel each other out.

"sample": "Ezid:Average",
"distance": 0.8163,
"Ebla_EMBA": 30,
"Sappali_Tepe_BA": 27,
"Hajji_Firuz_C": 24.5,
"Krasnoyarsk_MLBA": 17,
"Mongola": 1.5,
"Kura-Araxes_ARM_Kalavan": 0
},
{
"sample": "Talysh Azerbaijan:Average",
"distance": 0.9651,
"Ebla_EMBA": 21.5,
"Sappali_Tepe_BA": 21.5,
"Hajji_Firuz_C": 19,
"Krasnoyarsk_MLBA": 12,
"Mongola": 2.5,
"Kura-Araxes_ARM_Kalavan": 23.5
}

Coldmountains
06-07-2020, 07:15 PM
The most Parsimonious models for Ezids I got was this, I think other Kurds esp those in Turkey will show more diversity and show a combination of Hittite/Luwian and Indo Iranian ancestry. Though in both our models it suggests there was a massive movement from Central Asia to this region.

Target: Ezid:EZI-010
Distance: 1.4403% / 0.01440266
52.2 SYR_Ebla_EMBA
26.4 UZB_Sappali_Tepe_BA
21.4 KAZ_Dali_MLBA


Target: Ezid
Distance: 0.4908% / 0.00490822
51.6 SYR_Ebla_EMBA
31.0 UZB_Sappali_Tepe_BA
17.4 KAZ_Dali_MLBA



"sample": "Kurdish:Average",
"distance": 1.1331,
"Alalakh_MLBA": 50,
"Sappali_Tepe_BA": 36,
"Oy_Dzhaylau_MLBA": 14




With genoplot

I am probably late here but wanted to say that i agree and get with a a different model on vahaduo basically the same results. Ezidi definetly have unusual high steppe/Mitanni-like ancestry even compared to Iranians way more in the east. But it seems it is very hard to differentiate BMAC from Iran_C here so a lot of local Iran_C ancestry is in BMAC here.

Target: Iranian_Fars

Distance: 0.3848% / 0.00384845
35.6 SYR_Ebla_EMBA
34.8 BMAC
13.2 Steppe
8.2 IRN_Seh_Gabi_C
4.4 KAZ_Botai
3.8 IRN_Shahr_I_Sokhta_BA2

Target: Iranian_Lor
Distance: 0.4585% / 0.00458490
46.2 SYR_Ebla_EMBA
39.6 BMAC
8.0 Steppe
3.0 KAZ_Botai
2.2 IRN_Seh_Gabi_C
1.0 IRN_Shahr_I_Sokhta_BA2

Target: Ezid
Distance: 0.1970% / 0.00196999
44.0 SYR_Ebla_EMBA
29.6 BMAC
17.8 Steppe
5.4 IRN_Seh_Gabi_C
2.4 IRN_Shahr_I_Sokhta_BA2
0.8 KAZ_Botai

Target: Kurdish
Distance: 0.5900% / 0.00589956
38.8 SYR_Ebla_EMBA
36.0 BMAC
15.4 Steppe
9.8 IRN_Seh_Gabi_C

Coldmountains
06-07-2020, 07:19 PM
I dont think that we have central asian ancestry . it is rather that iran neo moved towards central asia and mixed with north/east euro hg types . and we have our steppe most likely from southern russia area . it wouldnt make any sense for iran neo to move to central asia and then move back to our habitated lands .

SampleFitTepecik Ciftlik NGanj Dareh NRUS CatacombOroqenAGUsers►Magnetic2.878549.53314.53

Even Syrians have some ancestry from Central Asia/Iron Age Iran. But the existence of R1a-Z93 and especially L657 among them pointed to this already before we had ancient dna.

Target: Syrian
Distance: 0.7655% / 0.00765511
66.8 SYR_Ebla_EMBA
11.4 Steppe
10.8 BMAC
4.4 ETH_4500BP
2.0 IRN_Seh_Gabi_C
2.0 ITA_Rome_Latini_IA
1.6 GRC_Mycenaean
1.0 MNG_North_N

DMXX
06-07-2020, 07:19 PM
Talysh have significant Caucasus ancestry which is absent in the Ezidi. I think the continuum between Talysh and Ezidis is because the CHG is compensating for the relative lack of Iran compared with Ezidis and conversely the excess Iran_N in Ezidis compensates for the lack of CHG relative to Talysh so they kind of cancel each other out.


Yup, the data looks convincing here. The wizard strikes again.

Looks like that continuum was an artifact, which opens things up for the implications concerning the Yazidis (i.e. their background isn't bound by the Talysh).

DMXX
06-08-2020, 08:23 AM
So one of the samples has kurdish background.

Excellent catch.

On the basis of what's demonstrated via aDNA, it's quite inconceivable that a fully Syrian Arab individual would have a preferentially closer GD with Iranian plateau-derived pops than those from Anatolia.

deuterium_1
06-08-2020, 06:37 PM
I dont think that we have central asian ancestry . it is rather that iran neo moved towards central asia and mixed with north/east euro hg types . and we have our steppe most likely from southern russia area . it wouldnt make any sense for iran neo to move to central asia and then move back to our habitated lands .

SampleFitTepecik Ciftlik NGanj Dareh NRUS CatacombOroqenAGUsers►Magnetic2.878549.53314.53

The issue with finding definitive Mittani samples is that they assimilated rapidly into the neighbouring Hurrian population as they were an elite.

DMXX
06-08-2020, 09:32 PM
The issue with finding definitive Mittani samples is that they assimilated rapidly into the neighbouring Hurrian population as they were an elite.

I think our best luck is sampling burials from the Early Western Grey Ware, an Early Iron Age culture located along the rim of the eastern Urmia shoreline in Northwest Iran, which several archaeologists have linked to the Mitanni precursors.

Otherwise, you're right - It'll be next to impossible to locate bona fide Mitanni anywhere within the range of the Hurrian empire.

DMXX
06-09-2020, 01:52 AM
A bit of hypothesis testing, here.

Using Vahaduo, I ran the Yazidis through a battery consisting of three "base" agriculturalist pops and Hovsgol_BA (catch-all for ENA admix), followed by a cycling of EMBA steppe (Catacomb), MLBA steppe (Sintashta) and Turkmenistan_IA. With the former two steppe pops, I then added Gonur1_BA immediately afterwards to see if there's any improvement.




Target: Ezid
Distance: 1.9848% / 0.01984751
64.2 IRN_Hajji_Firuz_C
16.0 RUS_Catacomb
15.6 AZE_Caucasus_lowlands_LC
3.2 MNG_Hovsgol_BA
1.0 TUR_Titris_Hoyuk_EBA

Target: Ezid
Distance: 1.8381% / 0.01838086
42.2 IRN_Hajji_Firuz_C
17.0 AZE_Caucasus_lowlands_LC
14.6 RUS_Catacomb
13.6 TUR_Titris_Hoyuk_EBA
9.6 TKM_Gonur1_BA
3.0 MNG_Hovsgol_BA

Target: Ezid
Distance: 2.2178% / 0.02217813
71.8 IRN_Hajji_Firuz_C
17.2 RUS_Sintashta_MLBA
7.4 AZE_Caucasus_lowlands_LC
3.6 MNG_Hovsgol_BA
(no TH_EBA)

Target: Ezid
Distance: 1.5695% / 0.01569453
40.8 IRN_Hajji_Firuz_C
16.6 TKM_Gonur1_BA
15.6 RUS_Sintashta_MLBA
12.4 TUR_Titris_Hoyuk_EBA
11.8 AZE_Caucasus_lowlands_LC
2.8 MNG_Hovsgol_BA

Target: Ezid
Distance: 1.6133% / 0.01613277
49.4 IRN_Hajji_Firuz_C
29.0 TKM_IA
12.4 AZE_Caucasus_lowlands_LC
6.6 TUR_Titris_Hoyuk_EBA
2.6 NG_Hovsgol_BA



The single-best model was MLBA steppe + BMAC, with a near 50:50 split between the two and the total coming to around 32%.
Turkmenistan_IA, which we already know is almost 50:50 itself, yields a highly similar fit (unsurprisingly).

Repeating this same battery with the Eurogenes Kurdish average:




Target: Kurdish
Distance: 1.6928% / 0.01692798
68.8 IRN_Hajji_Firuz_C
14.4 AZE_Caucasus_lowlands_LC
14.4 RUS_Catacomb
2.4 MNG_Hovsgol_BA

Target: Kurdish
Distance: 1.3479% / 0.01347883
39.4 IRN_Hajji_Firuz_C
16.6 AZE_Caucasus_lowlands_LC
16.4 TUR_Titris_Hoyuk_EBA
13.0 TKM_Gonur1_BA
12.4 RUS_Catacomb
2.2 NG_Hovsgol_BA

Target: Kurdish
Distance: 2.1505% / 0.02150507
74.8 IRN_Hajji_Firuz_C
14.8 RUS_Sintashta_MLBA
7.4 AZE_Caucasus_lowlands_LC
3.0 NG_Hovsgol_BA

Target: Kurdish
Distance: 1.2232% / 0.01223166
38.4 IRN_Hajji_Firuz_C
19.0 TKM_Gonur1_BA
15.2 TUR_Titris_Hoyuk_EBA
13.0 RUS_Sintashta_MLBA
12.4 AZE_Caucasus_lowlands_LC
2.0 NG_Hovsgol_BA

Target: Kurdish
Distance: 1.3521% / 0.01352133
56.2 IRN_Hajji_Firuz_C
26.2 TKM_IA
11.6 AZE_Caucasus_lowlands_LC
4.2 TUR_Titris_Hoyuk_EBA
1.8 NG_Hovsgol_BA



Perhaps surprisingly given the recent speculation, the pattern is pretty much the same for the Kurds (MLBA steppe + BMAC worked best, near 50:50 split, total comes to around 28%).

Repeating the same procedure with the Megiddo outlier (using Alalakh, Megiddo_MLBA and Ebla_EBA as the agriculturalist groups):




Target: Levant_Megiddo_MLBA_o1
Distance: 3.1924% / 0.03192421
56.0 TUR_Alalakh_MLBA
29.0 RUS_Catacomb
15.0 Levant_Megiddo_MLBA

Target: Levant_Megiddo_MLBA_o1
Distance: 3.1924% / 0.03192421
56.0 TUR_Alalakh_MLBA
29.0 RUS_Catacomb
15.0 Levant_Megiddo_MLBA
(no need for Gonur1_BA)

Target: Levant_Megiddo_MLBA_o1
Distance: 4.1657% / 0.04165673
55.4 TUR_Alalakh_MLBA
32.2 RUS_Sintashta_MLBA
12.4 Levant_Megiddo_MLBA

Target: Levant_Megiddo_MLBA_o1
Distance: 3.5952% / 0.03595212
30.0 TUR_Alalakh_MLBA
28.8 Levant_Megiddo_MLBA
28.8 RUS_Sintashta_MLBA
12.4 TKM_Gonur1_BA

Target: Levant_Megiddo_MLBA_o1
Distance: 3.9331% / 0.03933143
41.6 TKM_IA
39.8 TUR_Alalakh_MLBA
18.6 Levant_Megiddo_MLBA



Very different pattern. The Megiddo_o1 outlier clearly prefers an EMBA steppe source (sans BMAC) over MLBA steppe + BMAC, with a proportion of around 29% (converting this to an MLBA steppe equivalent: ~44%, which is about as MLBA steppe-shifted as modern Pamiri Tajiks or some SE European populations).

Including multiple other steppe-related pops (including all Sintashta outliers) yields a similar picture (pre-MLBA steppe pops are preferred):




Target: Levant_Megiddo_MLBA_o1
Distance: 3.2612% / 0.03261226
53.2 TUR_Alalakh_MLBA
28.0 RUS_Progress_En
18.8 Levant_Megiddo_MLBA

Target: Levant_Megiddo_MLBA_o1
Distance: 3.5379% / 0.03537930
77.6 TUR_Alalakh_MLBA
22.4 RUS_Khvalynsk_En

Target: Levant_Megiddo_MLBA_o1
Distance: 3.3117% / 0.03311683
54.4 TUR_Alalakh_MLBA
28.4 RUS_Afanasievo
17.2 Levant_Megiddo_MLBA

Target: Levant_Megiddo_MLBA_o1
Distance: 3.9231% / 0.03923074
80.4 TUR_Alalakh_MLBA
19.6 RUS_Sintashta_MLBA_o3

Target: Levant_Megiddo_MLBA_o1
Distance: 3.6571% / 0.03657092
72.4 TUR_Alalakh_MLBA
27.0 RUS_Sintashta_MLBA_o2
0.6 Levant_Megiddo_MLBA

Target: Levant_Megiddo_MLBA_o1
Distance: 4.4499% / 0.04449924
81.0 TUR_Alalakh_MLBA
19.0 RUS_Sintashta_MLBA_o1



Two immediate conclusions:
1) The steppe sources for Yazidis now appears common to the Kurds in general (there is no significant exceptionalism in this regard)
2) Megiddo_o1, as misnomer's earlier modelling shows, doesn't fit the general pattern seen in Yazidis or other Kurds (we know from other models that Lurs, Azeris and other NW Iranian-derived pops comply with this same pattern as well). Therefore, by and large, Megiddo_o1's origins don't appear to be intimately tied with modern NW Iranian-derived pops.

This is the first part of a lengthy algorithm-style procedure I'd undertaken for several W. Asian groups of interest. Click here for Part 2. (https://anthrogenica.com/showthread.php?20507-Thread-for-the-Israel-Megiddo-outliers&p=674480&viewfull=1#post674480)

Disclaimer: Please treat these outputs and conclusions as suggestive until someone can corroborate the findings via formal statistics.

DMXX
06-09-2020, 02:20 AM
Follow-up from Part 2, where I've come to the tentative conclusion that Alalakh_MLBA_o may be a Mitanni sample.

Applying them to the models with Turkmenistan_IA (which is used here as a catch-all substitute for the Iron Age Iranians):



Target: Kurdish
Distance: 1.3521% / 0.01352133
56.2 IRN_Hajji_Firuz_C
26.2 TKM_IA
11.6 AZE_Caucasus_lowlands_LC
4.2 TUR_Titris_Hoyuk_EBA
1.8 NG_Hovsgol_BA
...
Target: Kurdish
Distance: 1.1907% / 0.01190672
42.4 IRN_Hajji_Firuz_C
21.8 TKM_IA
12.8 AZE_Caucasus_lowlands_LC
12.8 TUR_Titris_Hoyuk_EBA
8.6 TUR_Alalakh_MLBA_o
1.6 MNG_Hovsgol_BA




Target: Ezid
Distance: 1.6133% / 0.01613277
49.4 IRN_Hajji_Firuz_C
29.0 TKM_IA
12.4 AZE_Caucasus_lowlands_LC
6.6 TUR_Titris_Hoyuk_EBA
2.6 NG_Hovsgol_BA
...
Target: Ezid
Distance: 1.5969% / 0.01596938
45.0 IRN_Hajji_Firuz_C
27.6 TKM_IA
12.6 AZE_Caucasus_lowlands_LC
9.4 TUR_Titris_Hoyuk_EBA
3.0 TUR_Alalakh_MLBA_o
2.4 MNG_Hovsgol_BA


Intriguingly, the non-Yazidi Kurds from Eurogenes observe the best fit from the inclusion of Alalakh_o. Conversely, the Yazidis see no real improvement (and don't register a significant amount of Alalakh_o).

Extending this test to the pan-Azeri dataset and the Talysh:



Target: Talysh_Azerbaijan
Distance: 1.2784% / 0.01278396
39.4 IRN_Hajji_Firuz_C
20.4 TKM_IA
16.6 TUR_Titris_Hoyuk_EBA
12.6 TUR_Alalakh_MLBA_o
8.0 AZE_Caucasus_lowlands_LC
3.0 MNG_Hovsgol_BA


As one would expect, all of these populations select Turkmenistan_IA over Alalakh_o.

Ran myself (1/2 Iranian Azeri, 1/2 northern Persian) out of curiosity:



Target: scaled
Distance: 2.3499% / 0.02349916
25.6 TKM_IA
24.0 TUR_Titris_Hoyuk_EBA
18.8 AZE_Caucasus_lowlands_LC
12.8 IRN_Hajji_Firuz_C
12.2 TUR_Alalakh_MLBA_o
6.6 MNG_Hovsgol_BA


In line with the Talysh, it appears.

Running individual NW Iranian-derived pops against the above would be useful to see where the range and top values are.

StarDS9
06-09-2020, 10:42 AM
I don't think there is any mystery to the Ezidi results. Like Zoroastrian samples, there is likely a bottle neck in these groups.

Iranians of Fars also get high steppe, so maybe be high in some and little lower in others.

We also need Kurds from Iran on G25 specially from Kurdistan province to get a better picture as some kurds from there got high North European on Gedmatch.

DMXX
06-11-2020, 08:34 PM
Repeating the above setup via Genoplot, albeit with individual samples.
Here, I assume (with reason) that Alalakh_MLBA_o is not a native Anatolian BA individual and is, instead, one of the early unadmixed Mitanni (please see here (https://anthrogenica.com/showthread.php?20507-Thread-for-the-Israel-Megiddo-outliers&p=675375#post675375) for the work-through; also here (https://anthrogenica.com/showthread.php?20507-Thread-for-the-Israel-Megiddo-outliers&p=674480&viewfull=1#post674480); also misnomer's convincing qpAdm output in that thread, which fully supports my G25 modelling here).
Notation - "A:" = average, "R:" = range.

Beginning with Hasanlu_IA:

Sample:Hasanlu IA ► Average
Fit:1.9787
Results:Hajji Firuz C43
Caucasus lowlands LC23
Alalakh MLBA o13.5
TKM IA10.5
Titris Hoyuk EBA9.5
Hovsgol BA0.5

Great fit. They come out as ~76% Anatolian-Caucasus-Iranian (ACI) agriculturalist and ~24% Central Asian Indo-Iranian (CAII).
Intriguingly, and perhaps not unexpectedly, they're registering more Alalakh_o than Turkmenistan_IA, which I take (per the work-through above) as evidence of proportionally greater precursor Mitanni admix versus Median ancestry in this sample.

The Talysh:

SampleFitHajji Firuz CTKM IATitris Hoyuk EBAAlalakh MLBA oHovsgol BACaucasus lowlands LCTalysh Azerbaijan ► Talysh52.8260.523.58530
Talysh Azerbaijan ► Talysh31.6855.519.571233
Talysh Azerbaijan ► Talysh12.145518.577.539
Talysh Azerbaijan ► Talysh42.144813.57.513315
Talysh Azerbaijan ► Talysh22.484813.56.511.5416.5

A: ~69.5% ACI agriculturalist, 27.5% CAII and ~3% ENA.
R: 65.5-71% ACI agriculturalist, 25-31.5% CAII and 3-4% ENA.
Acceptable fit for all.
All (barring Talysh4) are Turkmenistan_IA > Alalakh_o.

Iraqi and Turkish Kurds (including our AG friends StarDS9 and mori yek+family):

SampleFitHajji Firuz CTKM IAAlalakh MLBA oHovsgol BACaucasus lowlands LCTitris Hoyuk EBAAGUsers ► Mori yek father2.669.514.5103.52.50
AGUsers ► StarDS92.566273400
Kurdish ► KRD-401.8265.5175.55.506.5
AGUsers ► Mori yek1.5263189.5360.5
Kurdish ► KRD-142.3959.512.591.598.5
Kurdish ► kurd11562.1584.563.523.54.5
Kurdish ► KRD-82.5756.513.58.52181.5
Kurdish ► KRD-52.2255.5178.52125
Kurdish ► KRD-432.2953.516.582200
Kurdish ► KRD-312.5650.515.5111.5210.5
AGUsers ► Mori yek mother1.7250.518.512397
Kurdish ► kurd11011.54919.58211.510
Kurdish ► KRD-452.54531.511813.5
Kurdish ► KRD-492.9643.527.540.513.511
Kurdish ► kurd11591.8142.521.59219.55.5

A: ~72% ACI agriculturalist, ~26% CAII and 2.5% ENA
R: 55-86% ACI agriculturalist, 10.5-32.5% CAII and 0.5-5.5% ENA
Acceptable fit for all.
The majority of these Kurds are <77% ACI agriculturalist; sample kurd1156 is the most (86%) and also happens to be the least CAII (10.5%).
A handful of Kurds are quite a bit more Titris Hoyuk_EBA than the average, whereas others score none, which is in accordance with the inference of StarDS9 and co. regarding Anatolian admix.
As with most of the Talysh, all Kurds are Turkmenistan_IA > Alalakh_o.

The Yazidis (or Ezid):

SampleFitHajji Firuz CCaucasus lowlands LCTitris Hoyuk EBATKM IAAlalakh MLBA oHovsgol BAEzid ► EZI-0602.4461.515.5013.572.5
Ezid ► EZI-0102.155.593.528.503.5
Ezid ► EZI-0132.555320.5024.511
Ezid ► EZI-0122.47501802453
Ezid ► EZI-0293.0447.515.511.5156.54
Ezid ► EZI-0202.944.510926.573
Ezid ► EZI-0012.784413.57.526.57.51
Ezid ► EZI-0742.76372011.5199.53

A: ~70% ACI agriculturalist, ~27.5% CAII and ~2.5% ENA
R: 63.5-77% ACI agriculturalist, 20.5-34% CAII and 1-4% ENA
Acceptable fits for all (sample 029 is just out of 3%).
Half of these samples score <5% TH_EBA, whereas two score 11%. Perhaps not unexpected, given their self-attested ancestry further west.
All are Turkmenistan_IA > Alalakh_o; the range of Alakakh_o admix is also narrower than what's observed in the non-Yazidi Kurds above.

I haven't performed this run on Azeris, as Genoplot's "Azeri" cohort appears to be a mixture between Republic and Iranian Azeris, who are quite different in terms of admixture proportions.

In summary, among the various Kurdish groups and the Talysh, the average amount of agricultural ancestry is 70-72% ACI and the CAII admix looks to be a remarkably consistent 26-27.5%.

Regarding Alakah_o ancestry specifically, the averages for these groups are:
Talysh - 9.8%
Non-Yazidi Kurds - 6.8%
Yazidi Kurds - 5.4%

Therefore, if we treat Alalakh_o as an early Mitanni sample, it's apparent that the Yazidis don't have any remarkable amount of Mitanni ancestry among them.
Instead, the majority of non-Yazidi Kurds exceed the Yazidi average, which may be expected, given the Yazidis ostensibly represent a founder effect of the broader diversity observed in the erstwhile and unattested Iranian speakers from the Classical period onwards in that part of West Asia.

Note: I don't consider these numbers authoritative (qpAdm is required to definitively calculate these). However, I do expect these patterns to hold in future analyses.

StarDS9
06-13-2020, 11:20 AM
I think Yezidis from Iraq will differ from Yezidis from Armenia. On Dodecad 12b one Yezidi from Iraq was closer to Sorani and Feyli average http://kurdishdna.blogspot.com/2012/06/kurdish-autosomal-dna-based-on-dodecad.html?m=1

Me and Mori yek ancestors are believed to be recent arrivals from Iran, so might not represent all Anatolian Kurds. Mori yek clusters closer to Lors where as my mother seems near identical to Azeris from iran might also explain my mtdna C4b. Most of my 23andme relatives from iran seem to be from Azeri provinces in Iran. I think there were Kurmanjis around there and closer to North Central Iran and has something to do with the early Alevis..

Halgurd
06-17-2020, 12:42 AM
Non-Yazidi Kurds - 6.8%
Yazidi Kurds - 5.4%

These aren't remarkably different numbers. I don't understand why there's this common misconception that Yazidis are genetically unique, when it has been proven before that they cannot be genetically distinguished to other Kurds.

MfA illustrated this very well at Corduene.blogspot.com

And also, it's interesting how Yazidi Kurds from Afrin, Tur Abdin, Batman etc (basically the Yazidi Kurds outside Singal, Sheikhan and Lalish) are given literally no attention for some apparent reason.

DMXX
06-17-2020, 01:21 AM
These aren't remarkably different numbers.


Correct. They're effectively the same. The modelling is unambiguous on this point - Yazidi Kurds are basically just a drifted subset of the diversity one observes in the non-Yazidi Kurdish realm.

FWIW, I'm personally not convinced that the Talysh are more "Mitanni-derived" than Kurdish groups, despite the models demonstrating that.

The Mitanni precursors (following the Anthony-Kuz'mina type model) apparently migrated from NE Iran (Mazandaran province) towards the east Urmia coastline. There's no evidence of settlement in the SE Caspian (at least, that I'm aware of).
There's a clear historical precedent to expect such ancestry to be present among Kurdish groups - The same doesn't hold for the Talysh.

As such, I think the Alalakh_o elevation in the Talysh is probably representative of a richer Iran_N agriculturalist stream than what Hajji Firuz ChL offers.



I don't understand why there's this common misconception that Yazidis are genetically unique, when it has been proven before that they cannot be genetically distinguished to other Kurds.


It's due to a cognitive bias (specifically, outcome bias).

Observers note that the Yazidis have a very different religious identity from other Kurds, resulting in a cognitive leap of faith concerning their ancestry (i.e. Yazidis must have different ancestry, given they have a different religious denomination).

The exact same thing happened with Zoroastrians several years ago - A flurry of individuals from across the community settled on models which met their outcome bias supporting the implicit conclusion that "Indo-Iranian religion = higher steppe ancestry".

We now know, however, that when modeled correctly (specifically the inclusion of additional sources of Iran_N ancestry), Zoroastrians from central Iran aren't that remarkable.
Although they're slightly more steppe-shifted compared to neighbouring pops (due to an apparent deficiency in the minor but detectable Sub-Saharan, Arabian or E. Eurasian admix that's observed south of central Iran), their amount of MLBA steppe ancestry isn't exceptional (several other Persian, Azeri and Kurdish individuals from the existing datasets either meet or exceed the Zoroastrian average in some models).

[Edit]: Just to emphasise - I don't actually think that NW Iranian-related pops (Azeris, Kurds, Talysh) are 70-72% agriculturalist and only 26-28% C. Asian IIr. I think they're/we're substantially more than that, given the cross-comparisons with qpAdm.
Davidski's qpAdm run on Hasanlu_IA revealed they're approx. 16% MLBA steppe (https://eurogenes.blogspot.com/2017/08/the-iron-age-iranian.html). nMonte G25 has substantially diminished that (Turkmenistan_IA's half MLBA steppe, yet the sample only scored 10% here). Similarly, I seldom break above 21% MLBA steppe in G25, but Davidski had me at around 26% via qpAdm a couple years ago.
It's the general patterns I was interested in fleshing out here.

DMXX
06-17-2020, 01:44 AM
Me and Mori yek ancestors are believed to be recent arrivals from Iran, so might not represent all Anatolian Kurds. Mori yek clusters closer to Lors where as my mother seems near identical to Azeris from iran might also explain my mtdna C4b. Most of my 23andme relatives from iran seem to be from Azeri provinces in Iran. I think there were Kurmanjis around there and closer to North Central Iran and has something to do with the early Alevis..

Similarly, I have a lot of Kurdish relatives on 23andMe. In fact, most of the Turkish relatives I've discovered there ended up being of Kurdish origin.

The ties between Iranian Azeris and Kurds are multi-factorial, if we're going by history:
- Shared Median ancestry (IMO the biggest factor)
- The Alevi/Qizilbash connection
- The Donboli/Dimili (https://en.wikipedia.org/wiki/Khoy_Khanate)-Afsharid connection

Of course, generations of small-scale intermixing between the speech communities (the majority of the people living in my father's province in Iran are either Azeri or Kurmanji Kurdish). However, I'd think that's a very rare occurrence these days (or at least since the Safavid imposition of Shi'ism).

Halgurd
06-17-2020, 02:12 AM
Similarly, I have a lot of Kurdish relatives on 23andMe. In fact, most of the Turkish relatives I've discovered there ended up being of Kurdish origin.

The ties between Iranian Azeris and Kurds are multi-factorial, if we're going by history:
- Shared Median ancestry (IMO the biggest factor)
- The Alevi/Qizilbash connection
- The Donboli/Dimili (https://en.wikipedia.org/wiki/Khoy_Khanate)-Afsharid connection

Of course, generations of small-scale intermixing between the speech communities (the majority of the people living in my father's province in Iran are either Azeri or Kurmanji Kurdish). However, I'd think that's a very rare occurrence these days (or at least since the Safavid imposition of Shi'ism).

Yes I seem to be getting a few strong Azeri matches too, for some reason from Zanjan especially. I am partially from Naqadeh which is Kurdish-Azeri mixed so I guess that may explain it. A lot of towns in the region are quite multi ethnic.

DMXX
06-17-2020, 02:37 AM
Yes I seem to be getting a few strong Azeri matches too, for some reason from Zanjan especially. I am partially from Naqadeh which is Kurdish-Azeri mixed so I guess that may explain it. A lot of towns in the region are quite multi ethnic.

Very nice - Likewise, I'm partially from the regions around Khoy (which are, by and large, Azeri-Kurdish mixed villages of varying proportions). My father's village of birth was a near 50:50 mix of the two (used to have an Armenian minority). Do you get any Assyrian matches on 23andMe? The south Urmia region seems to be where they were once more commonly found.

Zanjan's a strange place to pick up a signal - IIRC the city's mostly a mix of Persian and Azeri.

Halgurd
06-17-2020, 09:55 AM
Very nice - Likewise, I'm partially from the regions around Khoy (which are, by and large, Azeri-Kurdish mixed villages of varying proportions). My father's village of birth was a near 50:50 mix of the two (used to have an Armenian minority). Do you get any Assyrian matches on 23andMe? The south Urmia region seems to be where they were once more commonly found.

Zanjan's a strange place to pick up a signal - IIRC the city's mostly a mix of Persian and Azeri.

As far as I know the Kurds there are mainly Shikak Kurds, which are plentiful in Kurdish cities in Turkey too. That might explain your matches there?

Yes actually I do have a few Assyrian matches with some heritage from Urmia.

StarDS9
06-17-2020, 08:24 PM
The Alevi tribes in Maras/Malatya and Alevi Kurds in Sivas are known by as Sinemili/.Sinaminli From what I have read it's named after a Alevi leader in Maras. I believe they probably had a different tribal name before they adopted Sinemili, which sounds more Turkish.

There are Turks all around Maras and Malatya, my mother says her meternal grandmother was from South maras in Pazarcik.

Most of the people from my tribe have mixed with Circassians, their tribal names are Kaberdey, Cechen, Abzeh http://www.cerkezdernegi.com/?pnum=59&pt=Çerkes+Köyleri

My mother model's as 1/4 Circassian/Turkish mix, but she says far as she knows she does not know of non kurdish ancestry. Plus her test was done with 23andme v5 which might give weird results as she gets higher circassian then me on G25 but not on Gedmatch calculators.

So far I as I know I am the only one who has C4b from the Middle East so far.

DMXX
06-17-2020, 08:39 PM
StarDS9 and I happen to have the same general situation with our maternal lines - Both belong to East Eurasian-originating haplogroups where subclade diversity is maximum in the EC Asian-Siberian-Beringian zone.

Within our current repository of aDNA, mtDNA C4b1 was found in four Iron Age to Classical period Kazakh and Kyrgyz steppe nomads (3x C4b1*, 1x C4b1a). C4b was also found in Hungarian Magyar burials. It hasn't been found in Andronovo, Afanasievo, Tagar or Karasuk (yet?), nor do any of the European Scythians carry it.

Iron Age Kazakhs were (per Greek and Persian accounts) within the realm of what we'd consider to be Scythians. Whether these Iron Age Kazakhs were ancestral to the same Scythians who displaced the Cimmerians (and chased them deep into West Asia and Mesopotamia specifically)... Who knows? There's insufficient data to make a convincing argument either way. It's possible, particularly as the displacement trajectory due to the Xiongnu confederacy was east-to-west.

The distribution of C4 around the Zagros doesn't help matters much - Derenko et al. found ~4.5% (n=22) of Iranian Azeris and ~2% of Qashqais (n=~120) had C4 (no further subclade assessment given). It was absent in their Persian cohort.
Iranian Azeris (as folks like myself and renowned Kurdish poster 'Palisto' have opined for several years now) come from the same genetic milieu as Kurmanji Kurds and other NW Iranian-related peoples, so they/"we" offer no further circumstantial elaboration on StarDS9's line's origins (those Cimmerian-chasing Scythians traversed through Armenian and Median territory; the Turkish influx). Ditto, Qashqais are considered by several Iranologists as an offshoot of Azeris (linguistically at least).

A broader view helps orient things. C4b* has also been found in various parts of Eastern Europe (e.g. Ukraine). However, C4b1 is quite firmly rooted in the EC Asian-Siberian-Beringian zone, both in terms of subclade diversity and frequency.

If Star's subclade status is resolved (i.e. confirmation he's C4b* or a downstream group like C4b1a), the origin of his line could be better contextualised.

As I see it now, IMO, the options (in order of descending likelihood)
Scythian = Turkish (both equally likely until subclade status resolution has been achieved or more aDNA emerges)
Parthian (we lack their aDNA, uncertain if E. Eurasian mtDNA reached Khorasan during the Classical period before Christ's time)
Median (quite unlikely, given the absence of E. Eurasian mtDNA in ChL NW Iran, or indeed, around the BMAC or in Sintashta based on current data)
Hurrian-Urartian, Sumerian (one needs to OD on baqlava to entertain those)

[edit]: The info regarding Star's mum (known Circassian admix, ?Turkish based on models) wouldn't help clarify the above dilemma, unfortunately, as the Circassians are located in the approximate area through which the Scythians (in pursuit of the Cimmerians) entered West Asia (though it isn't known whether they followed a Black or Caspian coastline route).

StarDS9
06-17-2020, 09:29 PM
I have recently done ftdna for my mother and currently waiting for it to arrive at ftdna (hoping clears customs/Saliva is sufficient) I will upgrade for a mtdna test from FTdna and find which subclade of c4b my mtdna is.

Is ftdna mtdna testing good?

StarDS9
06-17-2020, 10:15 PM
I have checked my DNA relatives on Gedmatch and all my top matches come out Kurdish with some Turkish mixture. Weirdly some of them show Jewish like mixture. I had read there were Jews in Maras before.

Also only me and one other is the most Kurdish/Iranian shifted.

StarDS9
06-17-2020, 11:45 PM
Here is a run I did that will give best results. The Azeri on the table is single Azerb72 as it's highly likely from iran. I use Turkish Balakesir as unlikely to have Kurdish mixture and used Greeks from Central Anatolia.

SampleFitIranian LorGreek Central AnatoliaArmenianKabardinAzeriTurkish BalikesirKurdish ► Average1.0479.5875.500
Ezid ► Average1.0577.564.57.513.5
AGUsers ► StarDS92.0359.50.5016.5221.5
AGUsers ► StarDS9 Mom1.8833.51712.53412
Turkish Trabzon ► Average1.8102965600

MMerlin
06-18-2020, 01:01 AM
If it benefits anyone, here are 3 semi-relevant G25 Coords, from individuals in my immediate family.

Of note, SampleA is half Kurdish, half Azeri, descended from landowners.

SampleB is southern Iranian, Elamite rich ancestry.



SampleA,0.092197,0.090382,-0.063733,-0.032946,-0.040315,-0.012271,0.002115,-0.007384,-0.033542,-0.021139,0.002111,0.004646,-0.004608,-0.010872,0.001221,0.021877,0.001565,-0.001647,0.007542,-0.007003,-0.010731,-0.011005,-0.000493,0.000602,0.000359

SampleB,0.078538,0.102568,-0.064865,-0.026163,-0.042777,-0.001116,-0.0047,-0.005077,-0.011249,-0.008747,-0.001949,0.002098,0.003122,0.006468,0.005293,0.018 032,-0.005737,0.006588,0.003645,-0.01038,0.000374,-0.004081,0.001849,-0.005181,0.004191

SampleC,0.079676175862048,0.102568471276509,-0.0641105139583204,-0.0335921609903263,-0.0421617360885436,-0.00920339176608276,-0.000940042552228355,-0.00184607692147429,-0.0227021459338099,-0.0138499444042205,-0.00276060319495577,0.000149866607354674,0.0063924 0956134696,0.00206434008826065,0.0058359729265993, 0.013126369642822,-0.0106914919445323,0.000506754378372797,0.00615920 44940885,-0.0116305786614424,-0.00262037592722876,-0.00395688766583031,0.00209521120653742,-0.00554295228195228,-0.000478998956157526

StarDS9
06-18-2020, 05:20 PM
If it benefits anyone, here are 3 semi-relevant G25 Coords, from individuals in my immediate family.

Of note, SampleA is half Kurdish, half Azeri, descended from landowners.

SampleB is southern Iranian, Elamite rich ancestry.



SampleA,0.092197,0.090382,-0.063733,-0.032946,-0.040315,-0.012271,0.002115,-0.007384,-0.033542,-0.021139,0.002111,0.004646,-0.004608,-0.010872,0.001221,0.021877,0.001565,-0.001647,0.007542,-0.007003,-0.010731,-0.011005,-0.000493,0.000602,0.000359

SampleB,0.078538,0.102568,-0.064865,-0.026163,-0.042777,-0.001116,-0.0047,-0.005077,-0.011249,-0.008747,-0.001949,0.002098,0.003122,0.006468,0.005293,0.018 032,-0.005737,0.006588,0.003645,-0.01038,0.000374,-0.004081,0.001849,-0.005181,0.004191

SampleC,0.079676175862048,0.102568471276509,-0.0641105139583204,-0.0335921609903263,-0.0421617360885436,-0.00920339176608276,-0.000940042552228355,-0.00184607692147429,-0.0227021459338099,-0.0138499444042205,-0.00276060319495577,0.000149866607354674,0.0063924 0956134696,0.00206434008826065,0.0058359729265993, 0.013126369642822,-0.0106914919445323,0.000506754378372797,0.00615920 44940885,-0.0116305786614424,-0.00262037592722876,-0.00395688766583031,0.00209521120653742,-0.00554295228195228,-0.000478998956157526

Where in South Iran is the individual from? Ilam has mostly Feyli Kurds. Are all 3 Kurds?

J Man
06-18-2020, 05:42 PM
Where in South Iran is the individual from? Ilam has mostly Feyli Kurds. Are all 3 Kurds?

Do you know of any Feyli Kurd Y-DNA results?

StarDS9
06-18-2020, 06:06 PM
Do you know of any Feyli Kurd Y-DNA results?

From what I can find, here is a few.

YDNA:
5x J1
2x J2a1a
1x E1b1b1c1
2x E1b1b1a
1x G2a4

mtDNA:
2x H*.
2x K1a4.
2x HV1.
1x K1a.
1x HV1a.
1x U8b
1x J1d
1x X2e
1x T1
1x H13a1a
1x L3d3a
1x HV
1xJ2b1

hartaisarlag
06-18-2020, 06:19 PM
Any further resolution on the E1b1b1c1? Is that a 23andMe call?

StarDS9
06-18-2020, 06:28 PM
Any further resolution on the E1b1b1c1? Is that a 23andMe call?

Honesty do not know of that.

StarDS9
06-18-2020, 06:30 PM
Here is a run I did of the provided samples, sample A clearly shows it's half Azeri.

SampleFitSeh Gabi CSintashta MLBATepecik Ciftlik NHovsgol BALevant PPNBShahr I Sokhta BA3Caucasus lowlands LCYoruba► sampleA2.776915.54.54.5321.50
Iranian Lor ► Average1.1668146.5243.520
► SampleC1.6965.515.53.52.57321
Iranian Bandari ► Average2.23641702014.502.5
Kurdish ► Average1.456317.53.51.560.580
Iranian Zoroastrian ► Average1.46321.521.53.544.50
Iranian Fars ► Average1.26319.52.52.5561.50
Ezid ► Average1.0357.520417.53.56.50
► sampleB2.2356.516.552.56.55.56.51

Coldmountains
06-18-2020, 06:35 PM
Very interesting thread. I looked at their Z93 clades and they and Kurds seem to lack or have very low amounts of Indo-Aryan R1a.L657 what is making me sceptical about Kurds or Yazidi showing much direct Mitanni ancestry. But yeah maybe it became low because of founder effects or Indo-Aryan Mitanni belonged rather to Z2124 clades.

StarDS9
06-18-2020, 06:50 PM
Very interesting thread. I looked at their Z93 clades and they and Kurds seem to lack or have very low amounts of Indo-Aryan R1a.L657 what is making me sceptical about Kurds or Yazidi showing much direct Mitanni ancestry. But yeah maybe it became low because of founder effects or Indo-Aryan Mitanni belonged rather to Z2124 clades.

There are some Kurds with Z282, inluding 1 Ezidi. And a few with Z2124
http://corduene.blogspot.com/2016/03/kurdish-tribes-y-dna-haplogroups.html

Coldmountains
06-18-2020, 07:10 PM
There are some Kurds with Z282, inluding 1 Ezidi,
http://corduene.blogspot.com/2016/03/kurdish-tribes-y-dna-haplogroups.html

These R1a-Z282 Kurds seem to belong to a basal R1a-Z282 clades (R1a-YP4858 ). R1a-Z283 was already found in Srubnaya so i would not be surprised if it arrived actually from the east with Indo-Iranians too.

It seems most Kurds are Z2125 but interestingly there is some Y40 among them which is otherwise very rare. Maybe Y40 can be linked with Mitanni but this is all of course very speculative

DMXX
06-18-2020, 07:14 PM
Very interesting thread. I looked at their Z93 clades and they and Kurds seem to lack or have very low amounts of Indo-Aryan R1a.L657 what is making me sceptical about Kurds or Yazidi showing much direct Mitanni ancestry. But yeah maybe it became low because of founder effects or Indo-Aryan Mitanni belonged rather to Z2124 clades.

It may be possible that R1a-L657 was specific to the S. Asian Indo-Aryans. The Mitanni may have belonged to a different subclade altogether.

If my previous G25 modelling and inference is correct regarding Alalakh_o being an unadulterated Mitanni individual, then, the key to predicting their subclade assignment may lie with Srubnaya (or indeed, the outlier groups from the Kazakh steppe).

The modelling I did in the Megiddo outliers thread using Genoplot (https://genoplot.com/) indicated that Alalakh_o really favoured Srubnaya_o over any other source of EHG-rich ancestry. I established an archaeological narrative (the known Srubnaya colony off the Kopet Dag foothills in Khorasan) that is plausible.

Srubnaya_o was a lady (mtDNA U5a1), so we can't glean much from her "outlier community"'s Y-DNA directly. Indirectly, however, we may infer that they might've carried Y-DNA Q - I determined that Srubnaya_o seemed to work unusually well as a combination of all the Sintashta outliers, two of which carried that subclade.

Core Srubnaya also had several individuals with Y-R1a parahaplogroups (including R1a1*-SRY1532.2(xR1a1a) and R1a1a*-M17(xZ2123,Z93). They also carried R1a1a-Z2123.

Ergo, it's quite possible that - If this line of reasoning is correct (a Srubnaya_o-type group->Kopet Dag colony->Marlik->Urmia->traditional Mitanni) - they might've carried an R1a parahaplogroup, R1a-Z2123 or even Q rather than R1a-L657. All of these subclades have been found irregularly across this particular part of West Asia among modern groups.

By the way, in reference to an earlier post of mine, even Feyli Kurds (who appear near-indistinguishable from Lurs) are around 19-21% MLBA steppe in qpAdm. The inferred range for other Iranian groups is around 20-24%. Further evidence in support of my proposition that the values seen in G25 modelling are a substantial underestimation.

Coldmountains
06-18-2020, 09:01 PM
It may be possible that R1a-L657 was specific to the S. Asian Indo-Aryans. The Mitanni may have belonged to a different subclade altogether.

If my previous G25 modelling and inference is correct regarding Alalakh_o being an unadulterated Mitanni individual, then, the key to predicting their subclade assignment may lie with Srubnaya (or indeed, the outlier groups from the Kazakh steppe).

The modelling I did in the Megiddo outliers thread using Genoplot (https://genoplot.com/) indicated that Alalakh_o really favoured Srubnaya_o over any other source of EHG-rich ancestry. I established an archaeological narrative (the known Srubnaya colony off the Kopet Dag foothills in Khorasan) that is plausible.

Srubnaya_o was a lady (mtDNA U5a1), so we can't glean much from her "outlier community"'s Y-DNA directly. Indirectly, however, we may infer that they might've carried Y-DNA Q - I determined that Srubnaya_o seemed to work unusually well as a combination of all the Sintashta outliers, two of which carried that subclade.

Core Srubnaya also had several individuals with Y-R1a parahaplogroups (including R1a1*-SRY1532.2(xR1a1a) and R1a1a*-M17(xZ2123,Z93). They also carried R1a1a-Z2123.

Ergo, it's quite possible that - If this line of reasoning is correct (a Srubnaya_o-type group->Kopet Dag colony->Marlik->Urmia->traditional Mitanni) - they might've carried an R1a parahaplogroup, R1a-Z2123 or even Q rather than R1a-L657. All of these subclades have been found irregularly across this particular part of West Asia among modern groups.

By the way, in reference to an earlier post of mine, even Feyli Kurds (who appear near-indistinguishable from Lurs) are around 19-21% MLBA steppe in qpAdm. The inferred range for other Iranian groups is around 20-24%. Further evidence in support of my proposition that the values seen in G25 modelling are a substantial underestimation.

I suspect that most of Kurdish Indo-Iranian Y-dna and autosomal is from a rather later Iranic wave. The Y40 seems to be based on his presence from South Asia to West Asia the best candidate for Mitanni input in my opinion. There is also some non-Indo-Iranian IE ancestry among them similar to Haji_Firuz_BA and Armenia_BA rich in R1b-Z2103. The high frequency of R1b-Z2103 among Kurds points very much to that.

Mitanni Indo-Aryans unlike later Iranics were quickly fully assimilated without having a lasting linguistic and cultural impact so we should not necessarily expect them to have a big genetic impact in the region. It is possible that their Y-DNA was (almost) fully replaced in the region.

About Srubnaya_O we should remember that she had a lot of WSHG ancestry, which was also high in many other BMAC and steppe samples before and after Srubnaya. This models are often showing different results based on the source groups and the best fits are not always the most realistic and historical fits.

pegasus
06-18-2020, 10:28 PM
I suspect that most of Kurdish Indo-Iranian Y-dna and autosomal is from a rather later Iranic wave. The Y40 seems to be based on his presence from South Asia to West Asia the best candidate for Mitanni input in my opinion. There is also some non-Indo-Iranian IE ancestry among them similar to Haji_Firuz_BA and Armenia_BA rich in R1b-Z2103. The high frequency of R1b-Z2103 among Kurds points very much to that.

Mitanni Indo-Aryans unlike later Iranics were quickly fully assimilated without having a lasting linguistic and cultural impact so we should not necessarily expect them to have a big genetic impact in the region. It is possible that their Y-DNA was (almost) fully replaced in the region.

About Srubnaya_O we should remember that she had a lot of WSHG ancestry, which was also high in many other BMAC and steppe samples before and after Srubnaya. This models are often showing different results based on the source groups and the best fits are not always the most realistic and historical fits.

This is true, other samples in the IAMC have plenty of Eneolithic Steppe and WSHG related ancestry, Srubnaya_o is not needed and proximal. Interestingly Alalakh outlier did not need any Steppe MLBA related source. The other Megiddo outliers, on the other hand, do have a large amount of Steppe MLBA related ancestry.




"sample": "Alalakh MLBA o:Average",
"distance": 1.834,
"Sappali_Tepe_BA": 72.5,
"Aigyrzhal_BA": 19,
"Chemurchek_EBA_2": 8.5



"sample": "Alalakh MLBA o:Average",
"distance": 1.8821,
"Sappali_Tepe_BA": 72,
"Aigyrzhal_BA": 23,
"Okunevo_BA": 5



"sample": "Bustan BA o1:Average",
"distance": 1.664,
"Sappali_Tepe_BA": 46.5,
"Aigyrzhal_BA": 40.5,
"Gonur2_BA": 13

DMXX
06-19-2020, 06:25 PM
I suspect that most if not all of Kurdish Indo-Iranian Y-dna is from a rather later Iranic wave. Also there is probably some non-Indo-Iranian IE ancestry among them similar to Haji_Firuz_BA and Armenia_BA rich in R1b-Z2103. The high frequency of R1b-Z2103 among Kurds points very much to that.


I agree fully with this.



Mitanni Indo-Aryans unlike later Iranics were quickly fully assimilated without having a strong and lasting linguistic and cultural impact so we should not expect them to have a big genetic impact in the region. It is possible and not unlikely that there Y-DNA was fully replaced in the region.

An alternative proposal:
- Mitanni were indeed incorporated quite rapidly into the Hurrian hierarchy (they appear to have been a warrior caste of sorts)
- Given the effective population size of the Hurrian empire was likely self-sustaining and numbering in the tens of thousands, I agree that the Mitanni likely didn't leave a very large autosomal impact on the region
- However, the current discussion involves uniparental markers, which are liable to selection factors
- As we can infer the Mitanni had a lofty position in the Hurrian hierarchical structure (particularly for newcomers), they very well could've experienced greater demographic success on the Y and possibly the mt lines (i.e. the uniparental contribution was disproportionately larger than the autosomal)

I do agree that, either way, the Iranian "wing" of the IIr family had a greater contribution on the region, irrespective of the Mitanni contribution's magnitude or type. The consistent favouring of MLBA steppe admixture in the discussed pops alone indicates this.

As for your and pegasus' comments regarding Srubnaya_o - I'll grant that to both of you. I'll admit that I'd forgotten about the approx. 9% WSHG-EHG-like admix that was present in the BA Turanian agriculturalists, even those in Gonur. Ergo, Alalakh_o very well could've fully been of BMAC stock.

This wouldn't necessarily change the chronology of the migration route highlighted in a previous post - Srubnaya (or a Sintashta-derived population) would've still been implicated in a cultural diffusion (this would be needed to account for the obvious Eurasian steppe material goods seen in Marlik).
Ergo, per the CM-pegasus proposal, Alalakh_o's ancestors may have been BMAC individuals who'd partook in a cultural exchange with steppe pops, inherited elements of the language (particularly those associated with theology and the steppe-related material goods themselves) and migrated westwards thereafter.
This is actually consistent with the "guest-host" model that Anthony and Mallory support for the early dissemination of the IE languages.

It would be useful to have a simple qpAdm model that investigated this (e.g. lpops Alalakh_o, Gonur1_BA, Sarazm_EN/any other WSHG-rich Tur agri, Srubnaya_o/any other pop around the BA Kazakh steppe with that weird quasi-WSHG stuff). Maybe we could ask Kale or Davidski?

Halgurd
06-19-2020, 08:02 PM
According to Kurdish historians (which I personally find more credible than European orientalists), Kurds are an amalgamation of mainly the Medes and Hurrians. Now history, of the Middle East especially, has been full of population assimilation.

Given that the Medes eventually dominated much of the Hurrian heartland it would make sense that the Hurrians pretty much were assimilated by the Medes and became Median speaking people. Most importantly the Cyrtians, who were a Median speaking tribe according to Minorsky, most likely gave the Kurds their name.

The Indo-Iranian speaking elite of Mitanni most likely got assimilated into Hurrian society after their downfall.
If we accept this view then,
Mitanni elite -> Hurrians -> Medes -> Kurds and also other NW Iranian people.

I know it’s much more complex than that but, if we are to accept this view, by stage 3 the Mitanni autosomal DNA would barely be detectable let alone with their descendants.

DMXX
06-29-2020, 05:58 AM
The OT concerning the classification of the Kurdish languages has been split to here:

https://anthrogenica.com/showthread.php?20853-split-Kurdish-Languages-How-Should-They-Be-Classified