November 03, 2012

Recent admixture in Altaic populations: a legacy of Empire?

Continuing my experiments with ALDER, I took every single Altaic population publicly available, i.e., the following 25 populations:
Altai, Balkars_Y, Buryat, Chuvashs_16, Daur, Dolgan, Evenk_15, Hezhen, Kumyks_Y, Kyrgyz_Bishkek_Ho, Mongol, Mongola, Nogais_Y, Oroqen, Tu, Turkish_Aydin_Ho, Turkish_Istanbul_Ho, Turkish_Kayseri_Ho, Turkmens_Y, Turks, Tuva, Uygur, Uzbeks, Xibo, Yakut
I also took three West Eurasian populations unlikely to have historical East Asian admixture (French, French_Basque, and Sardinians), and three East Eurasian populations unlikely to have historical West Eurasian admixture (Dai, She, Miaozu). I merged all of the above in PLINK with a --geno 0.03 flag, and extracting SNPs present in the Rutgers recombination map for Illumina chips (a total of 524,822 SNPs).

I then ran ALDER for all 25 Altaic populations using any of the 3*3 West/East Eurasian reference pairs, or a total of 25*3*3= 225 runs. I retained only those 2-ref admixture analyses for which ALDER reported "success" with no warnings.

I then converted reported times to calendar dates: a generation of 29 years was assumed; lacking information about the age of the sampled individuals, I assumed that the "present" is 1980; finally, I report the earliest and latest -/+ limits of any confidence interval, as well as the median of all estimates.

The results can be seen below; for 11 of the 25 populations there was at least one test which was successful with no warnings. This does not mean that the other populations are unadmixed, but the following cases appear to be most "well-behaved":


Now, these appear to make excellent sense.

Of the Dolgans:
There also existed a group of Russian settlers on the River Heta, who, by the end of the 19th century, had become Dolganized and had gradually adopted the way of life of nomadic reindeer breeders. ... The tribes forming the nucleus of the Dolgans migrated from the banks of the River Lena at the end of the 17th century. One of the reasons for migration was the fact that Russian goods, flour, for instance, were coming to the Taimyr Peninsula by the boats on the Lena.
The 1770-1860AD range for the admixture appears to coincide with the period where the Dolgans came under Russian influence.

Of the Evenks:
The history of the Evenks' habitation can be traced in detail from the 17th century on. At that time the Evenks left several of their previous territories, for instance, the River Angara, when the Yakut, the Buryat and the Russians appeared in the province. The Evenks had especially bad relations with the Yakuts, who had settled in the river basin of the Lena in the 13th century. In the 18th and 19th centuries the Evenks living there adopted the Yakut language. In the Baikal area the Evenks began to speak the Buryat and the Mongolian languages, and even converted to lamaism. The southern Evenk -- the Manegir, the Birar, the Solon -- were influenced by the Manchu, Daur and Chinese cultures. The arable lands in Siberia were occupied by Russian settlers, migrating there in the 17th century, and those Evenks, living in the vicinity on the upper reaches of the Lena and near Baikal, were russified.
Again, the  1630-1800AD admixture range seems consistent with the time when Evenks came into contact with Russians.

Of the Nogais:
 In the first half of the 17th century a number of Nogay tribes were nomadic on the steppes between the Danube and the Caspian. The invasion of the warlike Kalmyks forced several of the Nogay tribes to leave their home steppes and withdraw to the foothills of the North Caucasus. By the River Kuban they met with the Cherkess.  In the Moscow chronicles from the 16th and 17th centuries there are several mentions of the Nogay, including the two Nogay Hordes, the Great and the Small. The former roamed beyond the River Volga, the latter somewhat to the west. Both had numerous military encounters with the Russians. In the 17th century some of the Nogay chiefs entered into an alliance with Moscow and fought at times together with the Russians against the Kabardians, the Kalmyks and peoples of Dagestan. 
 The 1610-1730AD range intersects the period when the Nogais settled in the North Caucasus and interacted with North Caucasians and Russians.

Not much needs to be said for the admixture signal in the Uygur, Uzbek, Kyrgyz, and Mongols which collectively ranges from 1260-1500AD. This was a period of Mongol power when Mongolian and Turkic speaking peoples assumed control over Central Asia and replaced to a great degree the previous inhabitants of the area.

The origin of the Balkars is less certain, because they are an old Turkic group that settled in the Caucasus, but the admixture (830-1220AD) date seems plausible. So does, of course, that of the Turks from Caesaria (990-1260AD) which parallels those of my recent experiment, and can be associated with the takeover of Anatolia following the Battle of Manzikert. Finally, I don't have a read explanation for the 11-12th century signal of admixture in the Siberian Altai and Buryat, but presumably it has something to do with the expansions of Altaic peoples around that time that were also felt in the west during this period; presumably, this involved some type of mixture with Caucasoid groups in Siberia.

The admixture dates are quite helpful in helping us better interpret other signals of admixture such as those of ADMIXTURE analyses (e.g., globe13). For example, the Dolgan have 13.1% North_European in that experiment, and the Altai have 13.2%, but apparently this occurred centuries apart and may have involved different groups of West Eurasian people.

In conclusion, ALDER seems to find some quite plausible dates for major admixture episodes in the history of Altaic populations that are compatible with fairly recent historical events.

23 comments:

Anatolian Turkmen said...

Can you check Aydin Turks with Turkmens and Uzbeks with respect to Greeks and Armenians?

Anatolian Turkmen said...

Dienekes,
One more point regarding your previous post:

You say:

Pre-test: Does Turks have a 1-ref weighted LD curve with Armenians? and the answer is no because were too similar.

Do you do a Pre-test with Aydin Turks and Turkmens for example?

Pre-test: Does Aydin Turks have a 1-ref weighted LD curve with Turkmens?

What is the answer to that question, I wonder.

Thank you very much for your investigation and to be open minded to alternatives.

Dienekes said...

Pre-test: Does Aydin Turks have a 1-ref weighted LD curve with Turkmens?


They do; 23.72 +/- 7.46 generations and 13.3 +/- 4.4 admixture percentage

JVR said...

Hi Dienekes!

Your blog is really interesting and your work, very useful, for us who try to understand the evolution of the human diversity. I've been searching on your 'glob13' and other spreadsheets, and I want to now if there is any way of dissecting the North European component, in west-east gradient (I mean trying to separate a west-Scandinavian one and an east-Balto/Slavic one), and the Mediterranean into a northwest-Iberian, a east-Greek? and a southwest-Maghrebian ones. Again, thanks for your work!

Anatolian Turkmen said...

What about 2 way results for Aydin Turks?
Specifically:

Greek-Uzbek

And

Greek- Turkmen?

Valikhan said...

Northern Altaians are admixed Turkic tribes with Samoyedic and Ket tribes. It wasn't one-shot event rather steady and long admixing and the timing you get could be just average. Same for Buryats, they formed from admixture of different Mongolian, Tungus and Turkic tribes.

Anatolian Turkmen said...

Dienekes:
I thank you for repeating your analyses.

To me after looking at these results it is clear that the Turkification is not due to elite dominance.

There are multiple methods now that suggest an upwards of 30% Turkic admixture in Anatolia. This is also after all the Balkan, Caucasus and World wars.

I think your 1/7 Turkic admixture analysis initially assumed a much more Asiatic genetic source for incoming Turkic tribes. However we see that Turkmens, the closest linguistic cousins of Anatolian Turks, and Uzbeks (who also claim an Oghuz background: ogUZ Bek) have significant West Asian genetic content.

Kipchaks (Kazakh and Kirghiz) speak a different Turkic dialect than the Southern Turks.

Turkmens, Azeri Turks, Anatolian Turks, Iranian Turks, Balkan Turks all speak the same Oghuz dialect.

Uyghur and Uzbek are more intelligible to an Anatolian Turk than say Kirgiz despite a larger distance to Kashgar.

An Aydin Turk and a Turkmenistan Turkmen have the same amounts of Siberian genes.

Turkification seems to be not due to elite dominance but a significant Turkish migration in the 1070-1270 period which probably was supported until the late 1500's with additional Central Asian migrations.

The exact amount is hard to know but it seems to be in the 25% to 40% range depending on the source population as evidenced by your latest rolloff/alder analyses.

Ortu Kan said...

@ libya:

I don't know how definitive we can be about Iranian admixture or lack thereof in the proto-Salar. The origin accounts all seem to contain a healthy dose of apocryphal elements, but "Samarqand" is a point of general consensus, and this city was of course an old Iranian stronghold that first came under Western Turk rule many centuries before the Yuan. Even if we hold back from taking this waypoint at face value and assume they were originally centered somewhere else in the region, Oghuz speakers in western Turkestan of the 13th or 14th century would have more than a few opportunities for admixture with Iranians under their belts.

Secondly, you're ignoring the fact that the taking of non-Muslim brides by Muslim men (contingent on the women's conversion to Islam or at least the raising of their children as Muslims) was and remains commonplace in many contact environments. We know the Salars have significantly intermarried with Tibetans, Mongols, and, yes, to some extent, Han. In any event you're also forgetting admixture with largely Han-like (though perhaps less so in the Yuan) but nonetheless Muslim ethnicities like the Hui -- which continues to this day.

In sum, the modern Salars are a poor proxy for the Turkic component in Anatolian Turks.

Dienekes said...

There are multiple methods now that suggest an upwards of 30% Turkic admixture in Anatolia. This is also after all the Balkan, Caucasus and World wars.

The fact that a variety of Central Asian Altaic groups mixed with Caucasoids during Mongol Empire times rather supports the idea that they were originally more Mongoloid than they are today.

libya said...

Uzbeks too are a far worse proxy for 11 th century Turks since they are the result of more than 1000 years intermixing between an Iranian majority and Turk newcomers

Non muslim Turk populations like Khakass are surely a better proxy

The migration of Turks to Anatolia was very rapid one starting from western Kazakstan and according to the book below ‍‍‍‍‍(the part written by Yuri Bregel), the number for all Turk newcomers to Iran and Anatolia is about 80 thousands (whereas population of Anatolia only is estimated to as much as 6 to 12 mlns=>this estimation is not from the book I gave the link but from other sources)

http://www.amazon.com/Turko-Persia-Historical-Perspective-American-Research/dp/0521522919

Valikhan said...

Dienekes,
You are missing one key fact - those were Qipchaks who arrived during Mongol reign over Central Asia. That you are talking about is indeed the case for Qipchaks. Oghuz Turks are absolutely different story, they arrived to there in 6-7th century ad, 700 hundreds years before Qipchaks and Mongols. Plenty time to mix and dilute for them.

Valikhan said...

As for Balkars' admixture. This is the time when Khazar khaganate was broken and Bulgars, which were part of khaganate population, split apart, one part heading to Volga river basin, another one to Balkans and the rest in Caucasus.

libya said...

@Valikhan
By 6-7 th century , there was still not a bifurcation of Turks into Oghuz and Kipchaks

Oghuz tribe migrated at once and rapidly from eastern siberia (lena river) to eastern kazakstan to the kazakstanian peninsula that stretches onto caspian sea (I forgot its name) then again at once and very rapidly to Iran and Anatolia by the 1000's =>those Turks were certainly nearly 100% mongoloid and carriying the Altaic lineage C3c and were nomads

Sedentary Turkic speaking populations of tarim basin and other adjacent regions did not migrate anywhere since they were sedentary

Slumbery said...

Valikhan

This is beside the point, but the separation of Danube and Volga Bulgars happened way before the Khazar empire was broken. Actually the split of the Bulgars is connected to the rising of the Khazars, not to their falling, since the Khazars broke the united Bulgar "empire".

Onur said...

Dienekes,
You are missing one key fact - those were Qipchaks who arrived during Mongol reign over Central Asia. That you are talking about is indeed the case for Qipchaks. Oghuz Turks are absolutely different story, they arrived to there in 6-7th century ad, 700 hundreds years before Qipchaks and Mongols. Plenty time to mix and dilute for them.


Qipchaqs were then the northern neighbors of the Oghuz, as they were living in southwestern Siberia, right north of what is now Kazakhstan. The two Turkic groups had close relationship and spoke at most dialects of the same Turkic language back then.

Onur said...

By 6-7 th century , there was still not a bifurcation of Turks into Oghuz and Kipchaks

Seems to be very true. The Oghuz and Qipchaq confederations seem to have formed only after the dissolution of the Turkic Qaghanate (>750).

Valikhan said...

By 6-7 century ad there were at least 3 branches of Turkic languages - Oghuz, Qipchak, Karluk. The earliest versions of Oghuz languages in form of OghuR branch, existed somewhat by 3 century ad.

Valikhan said...

Saljuks appeared in Asia Minor in 11th century. By that time their ancestors were ruling over Chach, Sogdiana, etc over 300 years.

Onur said...

By 6-7 century ad there were at least 3 branches of Turkic languages - Oghuz, Qipchak, Karluk. The earliest versions of Oghuz languages in form of OghuR branch, existed somewhat by 3 century ad.

The Oghuz branch and the Oghur branch of the Turkic language family have nothing to do with each other. The Oghuz branch is a branch of the Common Turkic (=Shaz Turkic) branch of the Turkic language family. The Common Turkic branch also includes the Qipchaq, Qarluq (=Uyghuric) and Siberian branches of the Turkic language family. In contrast, the Oghur (=Lir Turkic) branch is a branch of the Turkic language family that is separate from the Common Turkic branch.

http://en.wikipedia.org/wiki/Common_Turkic_languages

http://en.wikipedia.org/wiki/Oghur_languages

By the 6th-7th centuries, Common Turkic was a single language and had not yet broken into separate languages. The Oghur branch, on the other hand, was already separate from Common Turkic by then.

Saljuks appeared in Asia Minor in 11th century. By that time their ancestors were ruling over Chach, Sogdiana, etc over 300 years.

No, the Oghuz were then almost exclusively living in what is now Kazakhstan. Sogdiana and Chach were then sometimes ruled by various Turkic groups (apparently none of them Oghuz until the 11th century invasion of the Seljuqs from what is now Kazakhstan), but their inhabitants were almost exclusively Iranic-speaking.

Valikhan said...

One should not rely on wikipedia. Please read professional linguists like Dybo A.V., who says that ProtoTurkic language started to split away by 1 century BC-0.

Onur said...

One should not rely on wikipedia. Please read professional linguists like Dybo A.V., who says that ProtoTurkic language started to split away by 1 century BC-0.

Yes, the Proto-Turkic language probably started to split away around that time (many other sources - including Wikipedia - agree with that). But during that split the Proto-Turkic language split into the Common Turkic and Oghur (not Oghuz) branches. The split of the Common Turkic language into the Oghuz, Qipchaq, Qarluq and Siberian branches would start many centuries later. For instance, during the Turkic Qaghanate era (6th-8th centuries) Common Turkic was still a single language and whatever internal divisions it had were then at most in the dialectal level. In the 11th century a Turkic linguist, Mahmud al-Kashgari, could still - quite deservedly - treat Common Turkic as a single language and note the similarity and high mutual intelligibility of its dialects.

BornBeforeWW2 said...

Can anyone explain the possible century or date of admixture between my 13.9% Nogai ancestry as population 2 and the 86.1% of my population 1 ancestry (a mix of N. European, Mediterranean, West Asian and SW Asian?), what century this admixture may have occurred and in what geographic location possibly? I'm number DOD 807. Thanks for any information as how I might have gotten nearly 14% Nogai_Y ancestry? Thanks.

BornBeforeWW2 said...

Would anyone have any suggestions how I might have acquired 13.9% Nogai ancestry as population 2 mixed with 86.1% population 1 ancestry mixed between Mediterranean, N. European, W. Asian and SW Asian? Thanks. I'm DoD-807.