PDA

View Full Version : How closely related and genetically similar are Hazara and Uzbek people?



Sikeliot
07-18-2019, 10:01 PM
Does anyone know?

DMXX
07-18-2019, 10:41 PM
Superficially, lukewarm:



Uzbek:Average
21 Uygur_Averaged Averaged 5.33
39 Hazara_Averaged Averaged 6.87
80 Hazara_Afghanistan_Averaged Averaged 7.993


A couple of reference points for what a 5.5-8 fit looks like in Europe, W Asia and S Asia:



English:Average
820 Croatian_Averaged Averaged 5.287
988 Polish_Averaged Averaged 6.396
1310 Finnish_Averaged Averaged 8.041




Greek:Average
151 Ashkenazi_Jew_Averaged Averaged 5.142
249 Portuguese_Averaged Averaged 6.439
517 Hungarian_Averaged Averaged 8.049




Iranian_Persian:Average
159 Turkish_Adana_Averaged Averaged 5.222
193 Assyrian_Averaged Averaged 6.428
363 Adygei_Averaged Averaged 8.115




Brahmin_Uttar_Pradesh:Average
168 Gujarati_Averaged Averaged 5.488
271 Punjabi_Jat_Averaged Averaged 6.795
370 Kamboj_Averaged Averaged 8.052


They (and Uyghurs) are effectively a three-way mixture of pre-Medieval SC Asian (represented here by the Yaghnobi-Kalash clade), Medieval Turkish and Medieval Mongolian.



Sample Details Fit Map KAZ Karluk Kalash Mongolian Tajik Yagnobi
1 Hazara:Average 2.1237 Open Map 48.2 18.2 32.4 1.2
2 Hazara_Afghanistan:Average 2.3893 Open Map 46.6 16.8 36 0.6
3 Uygur:Average 2.9446 Open Map 56.6 14.2 24.4 4.8
4 Uzbek:Average 2.0895 Open Map 61.2 15.6 10 13.2


As an aside, I've never been able to get a fit below 4.0 for Hazaras before today (haven't seen anyone else manage that, either). It looks like our preconceived notion was that the Hazaras were actually a two-way mixture of a Tajik-Pashtun like group and Mongolian, which clearly isn't so (they require a hefty source of something Medieval Turkish, which isn't the same as Mongolian). As such, your question's actually helped us resolve one of those strange modelling enigmas that've appeared several times in recent years.

The fit for Uyghurs isn't great, because they do require something more East Asian:



Sample Details Fit Map Han KAZ Karluk Kalash Mongolian Tajik Yagnobi
1 Uygur:Average 1.9043 Open Map 13 51.6 16.4 10.4 8.6


They look about 13% Han Chinese (which doesn't surprise me at all - The Han Chinese have had a physical and cultural influence on the Tarim Basin since ~200 B.C.; that's a full millennia before the Uyghurs even arrived there in ~800 A.D.). Note the radical improvement in fit (2.9->1.9).

Hazaras from Afghanistan and Pakistan basically look like one population here (80%+ Turko-Mongol, 16-18% Yaghnob-Kalash).

Hope you've found that useful.

Sikeliot
07-18-2019, 10:44 PM
Superficially, lukewarm:



Uzbek:Average
21 Uygur_Averaged Averaged 5.33
39 Hazara_Averaged Averaged 6.87
80 Hazara_Afghanistan_Averaged Averaged 7.993


A couple of reference points for what a 5.5-8 fit looks like in Europe, W Asia and S Asia:



English:Average
820 Croatian_Averaged Averaged 5.287
988 Polish_Averaged Averaged 6.396
1310 Finnish_Averaged Averaged 8.041




Greek:Average
151 Ashkenazi_Jew_Averaged Averaged 5.142
249 Portuguese_Averaged Averaged 6.439
517 Hungarian_Averaged Averaged 8.049




Iranian_Persian:Average
159 Turkish_Adana_Averaged Averaged 5.222
193 Assyrian_Averaged Averaged 6.428
363 Adygei_Averaged Averaged 8.115




Brahmin_Uttar_Pradesh:Average
168 Gujarati_Averaged Averaged 5.488
271 Punjabi_Jat_Averaged Averaged 6.795
370 Kamboj_Averaged Averaged 8.052


They (and Uyghurs) are effectively a three-way mixture of pre-Medieval SC Asian (represented here by the Yaghnobi-Kalash clade), Medieval Turkish and Medieval Mongolian.



Sample Details Fit Map KAZ Karluk Kalash Mongolian Tajik Yagnobi
1 Hazara:Average 2.1237 Open Map 48.2 18.2 32.4 1.2
2 Hazara_Afghanistan:Average 2.3893 Open Map 46.6 16.8 36 0.6
3 Uygur:Average 2.9446 Open Map 56.6 14.2 24.4 4.8
4 Uzbek:Average 2.0895 Open Map 61.2 15.6 10 13.2


As an aside, I've never been able to get a fit below 4.0 for Hazaras before today (haven't seen anyone else manage that, either). It looks like our preconceived notion was that the Hazaras were actually a two-way mixture of a Tajik-Pashtun like group and Mongolian, which clearly isn't so (they require a hefty source of something Medieval Turkish, which isn't the same as Mongolian). As such, your question's actually helped us resolve one of those strange modelling enigmas that've appeared several times in recent years.

The fit for Uyghurs isn't great, because they do require something more East Asian:



Sample Details Fit Map Han KAZ Karluk Kalash Mongolian Tajik Yagnobi
1 Uygur:Average 1.9043 Open Map 13 51.6 16.4 10.4 8.6


They look about 10% Han Chinese (which doesn't surprise me at all - The Han Chinese have had a physical and cultural influence on the Tarim Basin since ~200 B.C.; that's a full millennia before the Uyghurs even arrived there in ~800 A.D.). Note the radical improvement in fit (2.9->1.9).

Hazaras from Afghanistan and Pakistan basically look like one population here (80%+ Turko-Mongol, 16-18% Yaghnob-Kalash).

Hope you've found that useful.



I don't understand those numbers or models. Are you saying Uzbeks and Hazara are, or are not, similar?

DMXX
07-18-2019, 10:46 PM
I don't understand those numbers or models. Are you saying Uzbeks and Hazara are, or are not, similar?

I've provided you with the quantitative data (outputs) and cross-comparison with other populations to enable you to make your own subjective valuation on how "similar" they are.

My subjective answer regarding how "similar" they are was "lukewarm".

Sikeliot
07-18-2019, 10:49 PM
I've provided you with the quantitative data (outputs) and cross-comparison with other populations to enable you to make your own subjective valuation on how "similar" they are.

My subjective answer regarding how "similar" they are was "lukewarm".

I would imagine they have a similar ratio of Iranic to Mongolian type ancestry. But I do know I have seen some Hazaras on Gedmatch scoring Uzbek first. I just didn't know if they are the norm or not.

DMXX
07-18-2019, 10:55 PM
I would imagine they have a similar ratio of Iranic to Mongolian type ancestry...

Please view the comparison table in the 6th boxed code again. This isn't the case.

Hazaras, on average, are almost twice as Mongolian-derived as they are derived from a SC Indo-Iranian-type source. In turn, they're ~25-30% more Medieval Turkish-derived than they are Mongolian.

The ratio of Turko-Mongol:SC Indo-Iranian ancestry is in the order of ~5-5.5:1 here.

Censored
07-18-2019, 11:51 PM
When dealing with populations like these which have very divergent ancestries i.e. a mix of east and west eurasian or west Eurasian and SSA, relatively small differences in one component can lead to a much larger difference than it would if those components were more similar(all east eurasian or all west eurasian). In that case it may be worth applying a different standard(if you want) to a place like central Asia wrt to fits compared to say Europe. It is all subjective tho.

Case in point, just adjusting for that extra east Asian ancestry in Hazaras compared to Uzbeks(not even touching the other components) makes the fit much better.

"sample": "Hazara:Average",
"fit": 1.4362,
"Uzbek": 82.5,
"Mongola": 17.5,

"sample": "Hazara_Afghanistan:Average",
"fit": 1.7683,
"Uzbek": 79.17,
"Mongola": 20.83,

newtoboard
07-18-2019, 11:57 PM
What about Tajiks in Uzbekistan?

Ajeje Brazorf
07-19-2019, 12:07 AM
They are not that close to each other, Uzbeki has more Steppe and Hazara has more East Eurasian admixture.

[1] "distance%=2.5457"

Hazara

RUS_Devils_Gate_Cave_N_NEO240,25
HUN_LBK_MN_I4196,11.8
IRN_Ganj_Dareh_N_I1954,10.6
LAO_Hoabinhian_La368,9.2
RUS_Devils_Gate_Cave_N_NEO238,9.2
IRN_Tepe_Abdul_Hosein_N_AH1,9
RUS_Karelia_HG_I0211,8
GEO_CHG_KK1,6.4
RUS_Shamanka_N_DA249,6.2
RUS_West_Siberia_N_I1960,4.2
MKD_N_I0676,0.4


[1] "distance%=2.4698"

Uzbek

RUS_Devils_Gate_Cave_N_NEO240,23.2
HUN_LBK_MN_I4196,16
RUS_Karelia_HG_I0211,11.6
IRN_Tepe_Abdul_Hosein_N_AH1,11
GEO_CHG_KK1,9.4
LAO_Hoabinhian_La368,8.2
IRN_Ganj_Dareh_N_I1954,7.6
RUS_West_Siberia_N_I1960,5.2
RUS_Devils_Gate_Cave_N_NEO238,4.8
RUS_Shamanka_N_DA249,3

DMXX
07-19-2019, 12:20 AM
When dealing with populations like these which have very divergent ancestries i.e. a mix of east and west eurasian or west Eurasian and SSA, relatively small differences in one component can lead to a much larger difference than it would if those components were more similar(all east eurasian or all west eurasian). In that case it may be worth applying a different standard(if you want) to a place like central Asia wrt to fits compared to say Europe. It is all subjective tho.

Case in point, just adjusting for that extra east Asian ancestry in Hazaras compared to Uzbeks(not even touching the other components) makes the fit much better.

"sample": "Hazara:Average",
"fit": 1.4362,
"Uzbek": 82.5,
"Mongola": 17.5,

"sample": "Hazara_Afghanistan:Average",
"fit": 1.7683,
"Uzbek": 79.17,
"Mongola": 20.83,

Great fits, but at face value, the use of Uzbeks as a source pop for Hazaras doesn't make much historical sense. An Uzbek-like population most likely didn't exist in Medieval southern Afghanistan or Pakistan, and we have linguistic evidence (the eastern Persian base of the Hazaragi language) that the "indigenous" strata of the Hazara were a Persianate population (and not a Turkish one). Also, there appears to be some Mongolian ancestry in modern Uzbeks, so these outputs can't be interpreted literally.

Having said that, we can accurately gauge precisely how Mongolian-shifted the Hazara are by dissecting the Uzbeks through an equally tight model (cycle=1k bat=500 pen=def):



Sample Details Fit Map KAZ Karakhanid Mongola Pashtun Tajik Yagnobi
1 Uzbek:Average 1.6022 Open Map 55 7.6 24 13.4


Transposing the above back into your models (~80% Uzbek and ~20% Mongolian) means that the Hazara definitively look around ~45% Medieval Turkish, ~25% Mongolian, and the remainder along the Pashtun-Yaghnobi cline (~30%).

Ajeje Brazorf
07-19-2019, 12:26 AM
Other models for the ancients of Central Asia

[1] "distance%=2.1919"

KAZ_Botai_BOT2016 (3500 BC)

RUS_West_Siberia_N_I1960,87.4
RUS_Shamanka_N_DA251,7.4
LAO_Hoabinhian_La368,2.2
HUN_ALPc_Szakalhat_MN_I2743,1
RUS_AfontovaGora3_AfontovaGora3,1
CHN_Tianyuan_TY,0.6
RUS_Kolyma_Meso_Kolyma_River,0.4


[1] "distance%=4.1444"

TJK_Sarazm_En_I4910 (3578 BC)

IRN_Ganj_Dareh_N_I1954,66.6
GEO_CHG_KK1,16.4
RUS_West_Siberia_N_I1960,11.2
RUS_AfontovaGora3_AfontovaGora3,4.8
LAO_Hoabinhian_La368,0.8
IRN_Tepe_Abdul_Hosein_N_AH1,0.2


[1] "distance%=1.8427"

TKM_Geoksiur_En (3500 BC)

IRN_Ganj_Dareh_N_I1954,61.2
GEO_CHG_KK1,16.6
RUS_AfontovaGora3_AfontovaGora3,9.6
HUN_LBK_MN_I4196,3.2
IRN_Tepe_Abdul_Hosein_N_AH4,3
IRN_HotuIIIb_Mesolithic_I1293,2.4
IRN_Tepe_Abdul_Hosein_N_AH1,1.6
RUS_Karelia_HG_UzOO77,1
RUS_Karelia_HG_I0211,0.8
DEU_LBK_N_I2026,0.6


[1] "distance%=2.75"

UZB_Sappali_Tepe_BA_I7414 (1973 BC)

IRN_Tepe_Abdul_Hosein_N_AH1,31.4
IRN_Ganj_Dareh_N_I1947,24.8
IRN_HotuIIIb_Mesolithic_I1293,14.2
IRN_Tepe_Abdul_Hosein_N_AH4,6.8
Iberia_N_mur,6.6
HRV_Sopot_MN_I5077,6.4
GEO_CHG_KK1,5.6
HUN_ALPc_III_MN_I3535,1.4
LAO_Hoabinhian_La368,1.2
RUS_MA1_MA1,1.2
USA_Spirit_Cave_11000BP_AHUR770c,0.4

Censored
07-19-2019, 12:37 AM
Great fits, but at face value, the use of Uzbeks as a source pop for Hazaras doesn't make much historical sense. An Uzbek-like population most likely didn't exist in Medieval southern Afghanistan or Pakistan, and we have linguistic evidence (the eastern Persian base of the Hazaragi language) that the "indigenous" strata of the Hazara were a Persianate population (and not a Turkish one). Also, there appears to be some Mongolian ancestry in modern Uzbeks, so these outputs can't be interpreted literally.

Having said that, we can accurately gauge precisely how Mongolian-shifted the Hazara are by dissecting the Uzbeks through an equally tight model (cycle=1k bat=500 pen=def):



Sample Details Fit Map KAZ Karakhanid Mongola Pashtun Tajik Yagnobi
1 Uzbek:Average 1.6022 Open Map 55 7.6 24 13.4


Transposing the above back into your models (~80% Uzbek and ~20% Mongolian) means that the Hazara definitively look around ~45% Medieval Turkish, ~25% Mongolian, and the remainder along the Pashtun-Yaghnobi cline (~30%).

Oh no I wasnt implying that Hazaras are descended from an Uzbek like population, I was just trying to illustrate how the differences in diverged ancestral component(in this case east Eurasian vs. west Eurasian) results in greater total distances between two populations, as opposed to differences in less diverged components. And to show how just adjusting for that difference would make those groups a lot closer.

As you said above, it is kind of subjective as to who is "close" to what and I think it's important to note that a fit which may count as close in one geographic realm, say South or Central Asia, may not be that close in the context of another like in Europe for example, as these areas have different amounts of internal variation.

DMXX
07-19-2019, 12:39 AM
Gotcha, though by accident, you've helped push the Hazara fit down to a truly excellent 1.4-1.7 (barely hours after I broke the ~3.0 floor too). You're the current record holder now!

I remain pleasantly surprised by the fact that the Hazara are more Medieval Turkish-shifted than they are Mongolian-shifted, though it isn't a massive surprise (the establishment of the Turko-Mongol cultural entity in the Middle Ages is precisely the milieu from which the Hazara are thought to have descended from).

Censored
07-19-2019, 12:54 AM
Gotcha, though by accident, you've helped push the Hazara fit down to a truly excellent 1.4-1.7 (barely hours after I broke the ~3.0 floor too). You're the current record holder now!

I remain pleasantly surprised by the fact that the Hazara are more Medieval Turkish-shifted than they are Mongolian-shifted, though it isn't a massive surprise (the establishment of the Turko-Mongol cultural entity in the Middle Ages is precisely the milieu from which the Hazara are thought to have descended from).

Yep, and going off that we can split Uzbek into its source components and add Mongola to it, probably giving a similarly good fit:
At default penalty:
"sample": "Hazara:Average",
"fit": 1.3175,
"KAZ_Karakhanid": 45.83,
"Pashtun": 24.17,
"Mongola": 23.33,
"Tajik_Yagnobi": 6.67,

"sample": "Hazara_Afghanistan:Average",
"fit": 1.5431,
"KAZ_Karakhanid": 50.83,
"Mongola": 23.33,
"Pashtun": 17.5,
"Tajik_Yagnobi": 8.33,

There we go, Mongola seems to work better than Mongolian.

DMXX
07-19-2019, 12:58 AM
Well, how about that. We've cracked the Hazara together.

I think I've found a new modelling partner. :beerchug:

Censored
07-19-2019, 02:02 AM
I would be honored :D

pegasus
07-19-2019, 02:57 AM
What about Tajiks in Uzbekistan?

They have not been tested but the ones form Sughd , where most of them live in Uzbekistan should range from Yaghnobi/Pamiri like to the more ENA admixed kinds , there has been limited studies on Tajiks save for Pamiris ( who are not even Tajiks to begin with) but East Iranic speakers.

pegasus
07-19-2019, 03:07 AM
Does anyone know?

https://journals.plos.org/plosgenetics/article/figure/image?size=large&id=10.1371/journal.pgen.1005068.g002

Uzbeks are more SC Asian mixed ( ie more Iran_N +ANF related ancestry) but are still very much related with Hazaras.