Kaspias
08-09-2019, 08:09 PM
All credits goes to Ryukendo (https://anthrogenica.com/member.php?3493-Ryukendo), i'm just copying his work here so we can talk about it.
I split the populations to the following sets:
SET 1: Altai-centered Turkics
SET 2: Central Asian Turkics
SET 3: Tajiks.
SET 4: Uralic-like Turkics, such as Chuvash and Tatars, + Lipka Tatars.
SET 5: Samoyed-like Uralics.
SET 6: Volga Uralics.
SET 7: Finnics and Saami
East set showed similar lists of closest populations and similar lists of contributors, while differing systematically between themselves.
These are the sets of contributors used to fit each set:
SET 1: Altai-centered Turkics
SET 2: Central Asian Turkics
SET 3: Tajiks.
SET 4: Uralic-like Turkics, such as Chuvash and Tatars, + Lipka Tatars.
SET 5: Samoyed-like Uralics.
These populations could not be fit without Ket and Nganassan. The nearest aDNA sample, Karasuk_Outlier, were at distance 0.06 from them, i.e. the nearest population was at 6% difference in distance. So Ket and Nganasan were added back to improve the fit, and later we break down Ket and Nganasan through fits at a second step.
For the next two sets of populations, I did not think that they received direct gene flow from populations like Dai in S China, Ulchi in Manchuria etc after the IA (i.e. it was mediated by some ENA-admixed population from the IA and later, instead of being airlifted across Siberia from far Easterm Asia) so all East Asian populations were purged:
SET 6: Volga Uralics.
SET 7: Finnics and Saami
SET 1: Altai Turkics. Note the close distances between aDNA samples and these populations. Note also the consistent appearance of Scythians (especially Scythian_Pazyryk), Altai_IA, Mongola, and Karasuk_Outlier among the closest populations.
TUVINIAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Karasuk_outlier
0.03137152 0.04251899
Altai_IA Scythian_ZevakinoChilikta
0.04568074 0.04910874
Scythian_AldyBel Mongola
0.05677704 0.06354492
Sarmatian_Pokrovka Scythian_Samara
0.07472056 0.07539667
[1] "distance%=2.9007 / distance=0.029007"
Tuvinian
Scythian_Pazyryk 72.8
Itelmen 10.3
Karasuk_outlier 8.9
Mongola 7.9
KHAKASSIAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Karasuk_outlier
0.01634770 0.02464753
Altai_IA Scythian_ZevakinoChilikta
0.02681109 0.03452210
Scythian_AldyBel Sarmatian_Pokrovka
0.04073022 0.05703954
Scythian_Samara Srubnaya
0.05873087 0.06077001
[1] "distance%=1.3475 / distance=0.013475"
Khakass
Scythian_Pazyryk 55.15
Karasuk_outlier 37.00
Mongola 7.85
ALTAIAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Scythian_ZevakinoChilikta
0.02285839 0.03645670
Altai_IA Karasuk_outlier
0.03752331 0.04321664
Scythian_AldyBel Mongola
0.05141050 0.05272910
Tu Sarmatian_Pokrovka
0.06170647 0.06655838
Altaian
Scythian_Pazyryk 79.7
Mongola 18.4
Itelmen 1.9
BURYAT (they are mongolic, but autosomally virtually identical to Turkics. Historically, they are known to descend from Kurykans, who are Turkics. Their language may be due to linguistic shift after the 13th century, when they were conquered by Mongols.)
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Scythian_ZevakinoChilikta
0.03610198 0.04968497
Altai_IA Mongola
0.05186752 0.05526476
Karasuk_outlier Scythian_AldyBel
0.06010071 0.06395234
Tu Iran_IA
0.06594780 0.07885052
[1] "distance%=3.2282 / distance=0.032282"
Buryat
Scythian_Pazyryk 70.7
Mongola 25.0
Itelmen 4.3
The patterns seem very consistent: Turkics around the Altai are Scythian_Pazyryk+Mongola (inner Mongolian Mongols, almost pure ENA with very little West Eurasian ancestry) at approximately 4:1 ratio. As we move into Siberia, Itelmen (beringian-like) and Karasuk_Outlier ancestry starts to appear.
Since all fits were satisfactory, I did not pursue any alternate models with inclusions or exclusions of other populations.
SET 2: Central Asian Turks. Scythian_AldyBel starts to appear, together with more West Eurasian ancestry from the Caucasus and West Asia, but the Pazyryk+Mongola pattern still dominates. Closest populations still tend to be Scythians, Mongola, Karasuk_outlier followed by other East Asians (at a much further distance away), resembling the pattern for SET 1.
KYRGYZ
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Scythian_ZevakinoChilikta
0.02845224 0.03408819
Altai_IA Mongola
0.03849174 0.03896629
Tu Scythian_AldyBel
0.04762992 0.04901515
Karasuk_outlier Naxi
0.05192305 0.06266666
[1] "distance%=1.4539 / distance=0.014539"
Kyrgyz
Scythian_Pazyryk 54.60
Mongola 39.45
Chechen 2.75
Ulchi 2.75
Punjabi 0.45
KAZAKH
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Scythian_ZevakinoChilikta
0.02424266 0.02979828
Altai_IA Mongola
0.03345902 0.04304158
Scythian_AldyBel Karasuk_outlier
0.04389642 0.04791227
Tu Sarmatian_Pokrovka
0.05116875 0.05716372
[1] "distance%=1.2048 / distance=0.012048"
Kazakh
Scythian_Pazyryk 54.30
Mongola 30.35
Iran_IA 7.90
Itelmen 2.05
Punjabi 1.70
Chechen 1.35
Ulchi 1.25
Nordic_IA 1.10
KARAKALPAK
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Scythian_ZevakinoChilikta
0.02423779 0.03028732
Altai_IA Scythian_AldyBel
0.03358975 0.04048377
Mongola Karasuk_outlier
0.04539390 0.04737290
Srubnaya Chechen
0.05555522 0.05568733
[1] "distance%=1.2968 / distance=0.012968"
Karakalpak
Scythian_Pazyryk 53.25
Mongola 27.50
Ulchi 5.00
Chechen 4.80
Nordic_IA 4.30
Punjabi 3.75
Iran_IA 1.40
UZBEK
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_ZevakinoChilikta Altai_IA
0.02649022 0.02785061
Scythian_Pazyryk Scythian_AldyBel
0.03199633 0.03549600
Sarmatian_Pokrovka Iran_IA
0.04356709 0.04400175
Chechen Ulchi
0.04438390 0.04438390
[1] "distance%=0.7791 / distance=0.007791"
Uzbek
Scythian_Pazyryk 19.80
Iran_IA 19.45
Scythian_AldyBel 17.75
Mongola 12.40
Punjabi 10.60
Naxi 8.50
Ulchi 4.40
Chechen 4.35
Nordic_IA 1.80
Scythian_ZevakinoChilikta 0.50
Itelmen 0.45
There is one population that doesn't fit anywhere; these are the Yakuts. They are quite far from any other population (the closest I could find were Evenks, who were at 0.08 distance away, still very far) and I could not find good fits for them. The Siberian part is probably badly represented by the samples we have now.
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Evenk Scythian_Pazyryk
0.08718705 0.09910875
Altai_IA Scythian_ZevakinoChilikta
0.10842466 0.10906728
Karasuk_outlier Scythian_AldyBel
0.10908411 0.11013517
Mongola Chechen
0.11420687 0.11651864
[1] "distance%=9.8642 / distance=0.098642"
Yakut
Scythian_Pazyryk 89.25
Dai 10.70 <--- this is very weird, so I dropped it for the next fit
Punjabi 0.05
[1] "distance%=7.994 / distance=0.07994"
Yakut
Evenk 63.15
Scythian_Pazyryk 26.70
Chechen 6.15
Mongola 3.10
Narva_Lithuania 0.90
SET 3: Tajiks. Closest populations are different yet again: not Pazyryk, ZevakinoChilikta, Karasuk_Outlier, but Sarmatian_Pokrovka, Scythian_Samara, Altai_IA (shared by Tajiks and Turkics it seems), Scythian AldyBel (also shared), Srubnaya. For the fits, Scythian_Aldybel + Sarmatian_Pokrovka + Srubnaya_Outlier are common.
ISHKASHIM
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Sarmatian_Pokrovka Altai_IA
0.03287733 0.03523144
Scythian_Samara Karasuk
0.03739440 0.03858398
Scythian_AldyBel Iran_IA
0.03952654 0.03958612
Srubnaya Scythian_ZevakinoChilikta
0.03969207 0.04172257
[1] "distance%=1.3843 / distance=0.013843"
Tajik_Ishkashim
Iran_IA 31.25
Punjabi 21.30
Scythian_AldyBel 11.90
Sarmatian_Pokrovka 11.60
Srubnaya_outlier 11.40
Scythian_Samara 6.80
Scythian_ZevakinoChilikta 3.35
Chechen 2.40
SHUGNAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Sarmatian_Pokrovka Scythian_Samara
0.03059539 0.03372771
Karasuk Altai_IA
0.03599524 0.03653225
Srubnaya Scythian_AldyBel
0.03729174 0.03835506
Iran_IA Scythian_ZevakinoChilikta
0.03863595 0.04062046
[1] "distance%=1.345 / distance=0.01345"
Tajik_Shugnan
Iran_IA 36.50
Scythian_Samara 18.30
Punjabi 14.15
Srubnaya_outlier 12.85
Scythian_AldyBel 12.55
Sarmatian_Pokrovka 4.00
Mongola 1.65
RUSHAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Sarmatian_Pokrovka Scythian_Samara Karasuk
0.02915146 0.03349753 0.03518943
Altai_IA Srubnaya Scythian_AldyBel
0.03520315 0.03532160 0.03671922
Chechen Iran_IA
0.03938184 0.04015731
[1] "distance%=1.4672 / distance=0.014672"
Tajik_Rushan
Iran_IA 27.10
Scythian_AldyBel 19.60
Sarmatian_Pokrovka 18.95
Punjabi 11.20
Scythian_Samara 10.55
Chechen 7.25
Srubnaya_outlier 5.35
YAGNOBI
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Sarmatian_Pokrovka Iran_IA Scythian_Samara
0.03524610 0.03560888 0.03819717
Chechen Altai_IA Srubnaya
0.03856935 0.03916171 0.04043572
Scythian_AldyBel Karasuk
0.04104215 0.04129811
[1] "distance%=1.8059 / distance=0.018059"
Tajik_Yagnobi
Iran_IA 46.05
Scythian_AldyBel 17.95
Scythian_Samara 9.10
Srubnaya_outlier 8.85
Srubnaya 7.05
Chechen 6.15
Punjabi 3.35
Mongola 1.50
SET 4: Uralic-like Turkics. These populations, Chuvash and Tatars from the Volga, resemble the Uralic populations around them and do not share the turkic haplotypes centering on South Siberia identified by Yunusbaev et al. They may be due to linguistic shifts as well, because, judging from the recently uncovered part-hunnish genomes from Medieval Germany and the Balkans, the Turkic peoples, even at the earliest waves, were quite similar to South Siberians and others around the Altai in having high East Asian ancestry proper, and quite unlike present-day Volga peoples like Chuvash and Mishar Tatars, or Volga Uralics. Alternatively, if you do not believe the Huns in Europe represent the first para-Turkic incursions, the steppe was already quite East Asian at that point, so further issues from it are likely to retain this signal.
The closest population is Scythian_AldyBel (increasing closeness to AldyBel instead of Pazyryk as we move West, and occurrence of AldyBel at the edges of the Steppe while Pazyryk dominates in the centre, seems to be a consistent pattern), followed by Mezhovskaya (majority Andronovo and Sintashta-like population in C Siberia, identified by people like Parpola with some of the early Uralics), and Srubnaya.
CHUVASH
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_AldyBel Srubnaya
0.05332755 0.06003493
Scythian_Pazyryk Scythian_ZevakinoChilikta
0.06309898 0.06461331
Altai_IA Sarmatian_Pokrovka
0.06716522 0.06733498
Scythian_Samara Karasuk_outlier
0.06807143 0.06961130
[1] "distance%=5.0768 / distance=0.050768"
Chuvash
Scythian_AldyBel 73.45
Narva_Lithuania 6.75
Itelmen 5.40
Dai 5.20
Chechen 3.75
Ulchi 2.90
Comb_Ceramic 2.55
Dai is bad, so push out:
[1] "distance%=5.0557 / distance=0.050557"
Chuvash
Scythian_AldyBel 65.30
Mezhovskaya 18.15
Narva_Lithuania 5.55
Chechen 3.70
Evenk 3.70
Itelmen 2.65
Mongola 0.95
After fitting the Chuvash, I finalised the list of contributors for the other Volgaic populations (Uralics in SET 6), as well as the rest of this set.
TATAR
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_AldyBel Srubnaya
0.03080592 0.03752254
Scythian_Pazyryk Mezhovskaya
0.04248811 0.04396065
Nordic_IA Scythian_ZevakinoChilikta
0.04401260 0.04412678
Karasuk Altai_IA
0.04461660 0.04504374
[1] "distance%=2.3006 / distance=0.023006"
Tatar
Scythian_AldyBel 44.85
Nordic_IA 12.15
Comb_Ceramic 10.05
Chechen 9.20
Iran_IA 8.05
Mongola 7.25
Evenk 4.70
Itelmen 2.15
Narva_Lithuania 1.60
No Comb Ceramic:
[1] "distance%=2.3095 / distance=0.023095"
Tatar
Scythian_AldyBel 54.70
Nordic_IA 11.60
Chechen 11.05
Narva_Lithuania 6.60
Mongola 5.20
Iran_IA 5.10
Evenk 3.20
Itelmen 2.55
No Mongola:
[1] "distance%=2.3358 / distance=0.023358"
Tatar
Scythian_AldyBel 56.80
Nordic_IA 11.85
Chechen 11.45
Narva_Lithuania 6.70
Iran_IA 6.40
Evenk 3.65
Itelmen 3.15
TATAR_LIPKA (found in Poland and the Baltics)
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_AldyBel Scythian_Pazyryk
0.03337400 0.03853910
Scythian_ZevakinoChilikta Nordic_IA
0.04047207 0.04055040
Srubnaya Altai_IA
0.04166436 0.04233259
Karasuk Sarmatian_Pokrovka
0.04897414 0.04939444
[1] "distance%=1.767 / distance=0.01767"
Tatar_Lipka
Nordic_IA 26.45
Scythian_AldyBel 24.50
Mongola 20.05
Iran_IA 12.45
Chechen 6.00
Comb_Ceramic 5.60
Evenk 4.90
SET 5: Samoyed-like Uralics. These populations stretch across Northern Central Russia, and were very difficult to fit, as the closest aDNA sample was Karasuk_Outlier at 6% and fits with only ancients could not come close to representing them. Compare this:
KHAKASSIAN
KHAKASSIAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Karasuk_outlier
0.01634770 0.02464753
Altai_IA Scythian_ZevakinoChilikta
0.02681109 0.03452210
Scythian_AldyBel Sarmatian_Pokrovka
0.04073022 0.05703954
Scythian_Samara Srubnaya
0.05873087 0.06077001
with this:
MANSI
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Karasuk_outlier Scythian_AldyBel
0.06129125 0.07258879
Mezhovskaya Scythian_Pazyryk
0.07543525 0.07740380
Altai_IA Karasuk
0.07741589 0.08003623
Scythian_ZevakinoChilikta Scythian_Samara
0.08165920 0.08256958
Ket, Nganasan and Selkup, however, are very close to the populations in this cluster and also very close to each other:
KET
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Selkup Karasuk_outlier Scythian_AldyBel
0.02975829 0.06658063 0.08590533
Altai_IA Scythian_Pazyryk Mezhovskaya
0.08691091 0.08882178 0.09201866
Karasuk Scythian_Samara
0.09226259 0.09507147
So I added back Ket, Nganasan and Selkup to the contributors for this set. Before we get to modelling them, lets see what Ket, Nganasan and Selkup turn out to be:
KET
[1] "distance%=2.4243 / distance=0.024243"
Ket
Selkup 84.55
EHG 9.40
Comb_Ceramic 3.45
Narva_Estonia 1.50
Karasuk_outlier 1.10
[1] "distance%=5.4538 / distance=0.054538"
Ket
Karasuk_outlier 43.0
Nganassan 32.3
EHG 12.4
AfontovaGora3 12.3
Ancients only + Itelmen:
[1] "distance%=6.5437 / distance=0.065437"
Ket
Karasuk_outlier 82.3
AfontovaGora3 9.0
EHG 4.5
Itelmen 4.2
NGANASAN (this was still poorly fit, even with Ket and Selkup included):
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Itelmen Ket
0.08611205 0.10269269
Ulchi Karasuk_outlier
0.10510413 0.11700033
Scythian_Pazyryk Scythian_AldyBel
0.11926721 0.12623954
Altai_IA Scythian_ZevakinoChilikta
0.12766151 0.12795882
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Selkup Itelmen
0.08098218 0.08611205
Ulchi Karasuk_outlier
0.10510413 0.11700033
Scythian_Pazyryk Scythian_AldyBel
0.11926721 0.12623954
Altai_IA Scythian_ZevakinoChilikta
0.12766151 0.12795882
[1] "distance%=7.0066 / distance=0.070066"
Nganassan
Itelmen 59.9
Ket 40.0
Ulchi 0.2
[1] "distance%=8.2133 / distance=0.082133"
Nganassan
Itelmen 76.20
Karasuk_outlier 23.65
Ulchi 0.10
Narva_Lithuania 0.05
[1] "distance%=6.2835 / distance=0.062835"
Nganassan
Selkup 53.50
Itelmen 46.35
Ulchi 0.15
SELKUP
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Ket Karasuk_outlier
0.02975829 0.07368082
Nganassan Scythian_Pazyryk
0.08098218 0.09240397
Scythian_AldyBel Altai_IA
0.09283311 0.09322548
Mezhovskaya Scythian_ZevakinoChilikta
0.09755882 0.09924904
[1] "distance%=1.7742 / distance=0.017742"
Selkup
Ket 73.45
Nganassan 23.90
AfontovaGora3 2.65
[1] "distance%=6.8956 / distance=0.068956"
Selkup
Karasuk_outlier 67.40
Itelmen 22.85
AfontovaGora3 9.75
Then, moving on to Khanty and Mansi:
KHANTY
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Ket Karasuk_outlier
0.03953348 0.06273628
Scythian_AldyBel Scythian_Pazyryk
0.07573712 0.08013188
Mezhovskaya Altai_IA
0.08014730 0.08052845
Karasuk Scythian_ZevakinoChilikta
0.08435784 0.08532921
[1] "distance%=2.9562 / distance=0.029562"
Khanty
Ket 58.35
Nganassan 18.25
AfontovaGora3 8.55
Srubnaya_outlier 8.45
Scythian_ZevakinoChilikta 4.15
Scythian_AldyBel 2.25
Without Ket, Nganasan or Selkup:
[1] "distance%=5.8237 / distance=0.058237"
Khanty
Karasuk_outlier 66.7
AfontovaGora3 17.2
Itelmen 15.8
EHG 0.3
MANSI
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Karasuk_outlier Scythian_AldyBel
0.06129125 0.07258879
Mezhovskaya Scythian_Pazyryk
0.07543525 0.07740380
Altai_IA Karasuk
0.07741589 0.08003623
Scythian_ZevakinoChilikta Scythian_Samara
0.08165920 0.08256958
[1] "distance%=3.2698 / distance=0.032698"
Mansi
Ket 56.25
Nganassan 15.05
Mezhovskaya 10.40
Srubnaya_outlier 7.70
AfontovaGora3 5.70
Scythian_ZevakinoChilikta 4.75
Scythian_AldyBel 0.15
Without Ket, Nganasan, or Selkup:
[1] "distance%=5.7804 / distance=0.057804"
Mansi
Karasuk_outlier 58.0
Srubnaya_outlier 20.6
Evenk 18.6
Itelmen 2.8
SET 6: Volga Uralics. These populations are close to Scythian_AldyBel, Mezhovskaya, Comb Ceramic, and Srubnaya and have varying levels of Nordic_IA as we move west into Europe proper. The appearance of Mezhovskaya is a very nice confirmation of the suspicions of the likes of Parpola and etc.
KOMI
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_AldyBel Srubnaya Mezhovskaya
0.03428793 0.03691489 0.04030518
Karasuk Nordic_IA Sarmatian_Pokrovka
0.04315808 0.04347712 0.04604267
Scythian_Samara Scythian_Pazyryk
0.04800987 0.04843802
[1] "distance%=2.3631 / distance=0.023631"
Komi
Scythian_AldyBel 32.90
Nordic_IA 20.10
Comb_Ceramic 15.70
Nganassan 9.65
Iran_IA 8.05
Samara_Eneolithic 6.85
Mongola 3.95
Chechen 1.40
Itelmen 1.05
Punjabi 0.35
[1] "distance%=2.4836 / distance=0.024836"
Komi
Scythian_AldyBel 42.25
Nordic_IA 16.10
Comb_Ceramic 15.05
Itelmen 9.25
Iran_IA 7.50
Mezhovskaya 4.00
Chechen 2.85
Narva_Lithuania 1.40
Samara_Eneolithic 1.20
Mongola 0.40
UDMURT
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_AldyBel Mezhovskaya Srubnaya
0.03459012 0.04080253 0.04316311
Karasuk Karasuk_outlier Sarmatian_Pokrovka
0.04639066 0.04743340 0.04903342
Scythian_Samara Scythian_Pazyryk
0.04910755 0.04974005
[1] "distance%=2.5721 / distance=0.025721"
Udmurt
Scythian_AldyBel 49.70
Mezhovskaya 16.95
Nganassan 12.70
Samara_Eneolithic 9.70
Chechen 4.60
Ket 3.45
Comb_Ceramic 2.90
[1] "distance%=2.8946 / distance=0.028946"
Udmurt
Scythian_AldyBel 54.70
Mezhovskaya 29.15
Itelmen 9.20
EHG 5.35
Chechen 1.60
MARI (the pattern looks similar to the populations around them, especially Komi, but fit is poor for some reason)
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_AldyBel Mezhovskaya
0.08122688 0.08454560
Scythian_Pazyryk Srubnaya
0.08830828 0.08903658
Scythian_ZevakinoChilikta Karasuk_outlier
0.09006742 0.09263677
Altai_IA Karasuk
0.09294106 0.09332671
[1] "distance%=7.8202 / distance=0.078202"
Mari
Scythian_AldyBel 57.0
Mezhovskaya 24.6
Nganassan 13.0
Comb_Ceramic 5.5
[1] "distance%=7.9324 / distance=0.079324"
Mari
Scythian_AldyBel 59.50
Mezhovskaya 30.45
Itelmen 7.80
Comb_Ceramic 1.40
Narva_Lithuania 0.85
SET 7: Last of all, the Finnics and Saami. Very similar to the Volga Uralics, similarity to Mezhovskaya, Scythian_AldyBel, and so on, except more Nordic_IA. When belarusian or Slavic is added back, the remainder generally aggregates to Mezhovskaya.
SAAMI
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Mezhovskaya Srubnaya Scythian_AldyBel
0.03980681 0.04278583 0.04393002
Nordic_IA Karasuk Scythian_Samara
0.04669220 0.04986604 0.05072782
Scythian_Pazyryk Sarmatian_Pokrovka
0.05346288 0.05416290
[1] "distance%=3.1638 / distance=0.031638"
Saami
Scythian_AldyBel 43.3
Mezhovskaya 31.4
Nordic_IA 15.7
Narva_Lithuania 9.6
Add an EEF source to let that vary freely:
[1] "distance%=3.1551 / distance=0.031551"
Saami
Scythian_AldyBel 41.95
Mezhovskaya 31.90
Nordic_IA 11.80
Narva_Lithuania 7.80
Globular_Amphora 3.80
Comb_Ceramic 2.70
EHG 0.05
Add non-Finnic Europeans like Slavic_Bohemia:
[1] "distance%=2.9006 / distance=0.029006"
Saami
Slavic_Bohemia 38.00
Mezhovskaya 25.00
Scythian_AldyBel 16.80
Comb_Ceramic 11.00
Karasuk_outlier 6.95
Narva_Lithuania 1.95
Nordic_IA 0.30
INGRIAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Nordic_IA Srubnaya Mezhovskaya
0.03442590 0.03588778 0.04400809
Karasuk Scythian_AldyBel Sarmatian_Pokrovka
0.04559721 0.04592725 0.04850874
Scythian_Samara Scythian_Pazyryk
0.05091807 0.05861495
[1] "distance%=2.5148 / distance=0.025148"
Ingrian
Nordic_IA 50.60
Comb_Ceramic 18.45
Iran_IA 11.05 <----Don't like this, no post-IA flow from Iran for sure. Purge
Samara_Eneolithic 9.30
Mongola 6.65
Itelmen 3.90
[1] "distance%=2.7026 / distance=0.027026"
Ingrian
Nordic_IA 55.05
Scythian_AldyBel 24.75
Samara_Eneolithic 9.70
Narva_Lithuania 8.05
Sarmatian_Pokrovka 2.45
Add EEF source:
[1] "distance%=2.6888 / distance=0.026888"
Ingrian
Nordic_IA 44.5
Comb_Ceramic 18.1
Scythian_AldyBel 17.1
Sarmatian_Pokrovka 13.6
Globular_Amphora 6.6
Add Europeans:
[1] "distance%=1.1782 / distance=0.011782"
Ingrian
Belarusian 49.30
Slavic_Bohemia 17.90
Mezhovskaya 13.35
Nordic_IA 8.80
Srubnaya_outlier 5.75
Comb_Ceramic 2.70
Narva_Lithuania 2.20
VEPS
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Nordic_IA Srubnaya Scythian_AldyBel
0.03946320 0.04187050 0.04715007
Mezhovskaya Karasuk Sarmatian_Pokrovka
0.04877829 0.05119489 0.05416271
Scythian_Samara Scythian_Pazyryk
0.05587790 0.05897818
[1] "distance%=3.1157 / distance=0.031157"
Vepsian
Nordic_IA 44.40
Scythian_AldyBel 40.35
Narva_Lithuania 9.40
Comb_Ceramic 5.40
Samara_Eneolithic 0.45
Add EEF Source:
[1] "distance%=3.0995 / distance=0.030995"
Vepsian
Nordic_IA 38.95
Scythian_AldyBel 37.15
Comb_Ceramic 13.35
Globular_Amphora 5.50
Narva_Lithuania 5.05
Add Europeans:
[1] "distance%=1.3069 / distance=0.013069"
Vepsian
Belarusian 68.35
Mezhovskaya 17.75
Slavic_Bohemia 7.55
Comb_Ceramic 5.75
Narva_Lithuania 0.55
FINNISH EAST
[1] "distance%=2.5918 / distance=0.025918"
Finnish_East
Nordic_IA 55.30
Scythian_AldyBel 24.70
Comb_Ceramic 7.40
Samara_Eneolithic 6.05
Narva_Lithuania 5.55
Mezhovskaya 1.00
Add EEF:
[1] "distance%=2.5993 / distance=0.025993"
Finnish_East
Nordic_IA 52.50
Scythian_AldyBel 23.30
Comb_Ceramic 13.95
Mezhovskaya 7.10
Narva_Lithuania 2.85
Globular_Amphora 0.30
Add Europeans:
[1] "distance%=1.5965 / distance=0.015965"
Finnish_East
Belarusian 35.20
Slavic_Bohemia 22.05
Nordic_IA 19.10
Comb_Ceramic 10.00
Mezhovskaya 6.65
Srubnaya_outlier 6.45
Narva_Lithuania 0.55
FINNISH
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Nordic_IA Srubnaya Scythian_AldyBel
0.02584531 0.03270961 0.04330134
Mezhovskaya Karasuk Scythian_Samara
0.04368180 0.04646210 0.05008357
Sarmatian_Pokrovka Scythian_Pazyryk
0.05031018 0.05771340
[1] "distance%=2.0815 / distance=0.020815"
Finnish
Nordic_IA 65.30
Scythian_AldyBel 24.05
Narva_Lithuania 5.35
Samara_Eneolithic 3.30
Comb_Ceramic 2.00
Add EEF:
[1] "distance%=2.0803 / distance=0.020803"
Finnish
Nordic_IA 61.70
Scythian_AldyBel 22.05
Comb_Ceramic 8.40
Mezhovskaya 3.25
Narva_Lithuania 2.30
Globular_Amphora 2.30
Add Europeans:
[1] "distance%=1.1816 / distance=0.011816"
Finnish
Nordic_IA 35.40
Belarusian 31.70
Slavic_Bohemia 17.05
Mezhovskaya 8.20
Comb_Ceramic 4.95
Srubnaya_outlier 2.70
General thoughts:
Very surprised at the levels of specificity demonstrated here! E.g.
Mezhovskaya: Very Finnic and Volga Uralic
Scythian AldyBel: Very Uralic and Iranic and tendency to increase to the West and at the edges of the steppe.
Scythian_Samara and Sarmatian_Pokrovka: Very Iranic.
Scythian Pazyryk: Very Turkic.
Mongola: Very Turkic.
Altai_IA: Equally related to Turkics and Iranics
Karasuk_Outlier: Found in Central and Northeast Siberia
Comb Ceramic, Narva Lithuania, Narva Estonia: Found in European Siberia and Finnic-Baltic region.
I'd wager that the Iron Age Finnish genomes, when released, will look like Mezhovskaya (mostly Andronovo/Sintashta plus small fraction local EHG-ENA type thing) + a minority fraction of Comb Ceramic/Narva + maybe a little bit more of Karasuk_outlier type ancestry.
It seems to me that a part of the forest-Steppe was "Uralicised" after Andronovo and Sintashta, and this "Mezhovskaya" gene pool forms the core of the Uralics, from Saami to the Volga region, and further that this steppic population spread East before the whole horizon started pushing north into the Taiga in Siberia like parallel teeth on a comb (we know from linguistics that this is the case). Among the Eastern Uralics, i.e. Mansi-Khanty and Samoyed-like and Samoyed Uralics in Central and Northeastern Siberia, this "Mezhovskaya" population was diluted by some AG3-EHG-ENA mix, akin to the Nganasans today, for which we still have no ancient representative. Later waves of Iranic Scythians, like AldyBel, no doubt further ENA-ised the gene pool.
The Iranic Scythians probably existed in a continuum between Sarmatian_Pokrovka and Scythian_AldyBel, with Srubnaya_Outlier ancestry as well, in SC Asia. Then a Pazyryk Scythian+Mongola mix formed in South Siberia, possibly the Mongola is Baikal HGs or alternatively a population issuing from South of the Gobi (seen from the Slab Grave culture crania), and these were the Turkics who then splashed back West, occupying all the lowlands of C Asia, and pushed the AldyBel types to the edges of the Steppe.
Now, we can confirm these tendencies (Mezhovskaya looks Uralic, Pazyryk and Eastern Scythians look Turkic) using shared segments, with the caveats that this is haploid data and unphaseable, so IBD is not 100% reliable. All the same, the patterns are very striking. I quote from this thread:
Last of all, we can attempt to confirm that the analysis is correct by looking at clines in PCA itself, corresponding to patterns of gene flow.
We can't use Global25 because we cannot be sure there are PCs that so happen to load onto the dimensions of variation that most distinguish Siberians from each other. There are PCAs from published literature, however, which will do very well.
We see that the split of the populations into seven sets distinguished by different behaviour in nMonte is exactly corroborated by the positions of the populations in PCA space:
The sets of populations are:
SET 1: Altai-centered Turkics
SET 2: Central Asian Turkics
SET 3: Tajiks.
SET 4: Uralic-like Turkics, such as Chuvash and Tatars, + Lipka Tatars.
SET 5: Samoyed-like Uralics.
SET 6: Volga Uralics.
SET 7: Finnics and Saami
PCA of Pankratov et al on Lipka Tatars:
https://anthrogenica.com/attachment.php?attachmentid=22259&d=1521684028
Note the overlap of Chucash, Tatar, and Volga Uralics in cluster 4 + 6. Also note the close clustering of Altai Turkics + Buryat.
Note that Lipka Tatars differ from the rest of the Chuvash+Tatar by having double the Iran_IA and prescence of 20% Mongola. This can be seen in the above PCA, where Lipka Tatars are in purple.
[1] "distance%=1.767 / distance=0.01767"
Tatar_Lipka
Nordic_IA 26.45
Scythian_AldyBel 24.50
Mongola 20.05
Iran_IA 12.45
Chechen 6.00
Comb_Ceramic 5.60
Evenk 4.90
PCA of Pugach et al on Siberians:
https://anthrogenica.com/attachment.php?attachmentid=22260&d=1521684101
Note the clustering of Khanty, Samoyeds (Nenets, Selkup and etc.) and Ket in cluster 5. Interestingly, this PCA makes me think that Evenk may be a good source for Nganasan. Maybe will try this later.
The new PCA in the paper suggests that Yukagir and Nganasan form a pole of their own variation in Siberia, being the most "Northeastern" of all Siberians, which probably explains why Nganasan received such horrible fits (because we could not "reach" Nganasan as it was outside the convex hull formed by the contributing populations). They are at the very top-right of the PCA.
https://anthrogenica.com/attachment.php?attachmentid=22266&d=1521746746
Given that (and also a tip-off from Shaikorth, thanks brah) I decided to re-run the models with Yukagir/Nganassan as a contributing source. It seems the Bolshoi samples will justify this too.
You can see that the new PCA is absolutely beautiful, it traces each linguistic migration as a line of populations suspended between East and West Eurasia. The "Uralic cline" is in purple stretching from Nganasan and Yukagir to Saami and Scandinavians, the Turkic cline streches from Buryat and Altaian to Turkish and Abkhazian. They are marked in the PCA below. You can also see that Mongola did not seem optimal for the end of the Turkic line, rather Hezhen and Daur are closer, and Hezhen+Daur+Ulchi looks best. These are Tungusic and Mongolic populations in Manchuria, on the far Eastern edge of the Steppes.
I also checked the closest populations for the Scythians, which allowed me to mark their position (since they were not included on this PCA) using a purple lozenge.
https://anthrogenica.com/attachment.php?attachmentid=22267&d=1521748539
I split the populations to the following sets:
SET 1: Altai-centered Turkics
SET 2: Central Asian Turkics
SET 3: Tajiks.
SET 4: Uralic-like Turkics, such as Chuvash and Tatars, + Lipka Tatars.
SET 5: Samoyed-like Uralics.
SET 6: Volga Uralics.
SET 7: Finnics and Saami
East set showed similar lists of closest populations and similar lists of contributors, while differing systematically between themselves.
These are the sets of contributors used to fit each set:
SET 1: Altai-centered Turkics
SET 2: Central Asian Turkics
SET 3: Tajiks.
SET 4: Uralic-like Turkics, such as Chuvash and Tatars, + Lipka Tatars.
SET 5: Samoyed-like Uralics.
These populations could not be fit without Ket and Nganassan. The nearest aDNA sample, Karasuk_Outlier, were at distance 0.06 from them, i.e. the nearest population was at 6% difference in distance. So Ket and Nganasan were added back to improve the fit, and later we break down Ket and Nganasan through fits at a second step.
For the next two sets of populations, I did not think that they received direct gene flow from populations like Dai in S China, Ulchi in Manchuria etc after the IA (i.e. it was mediated by some ENA-admixed population from the IA and later, instead of being airlifted across Siberia from far Easterm Asia) so all East Asian populations were purged:
SET 6: Volga Uralics.
SET 7: Finnics and Saami
SET 1: Altai Turkics. Note the close distances between aDNA samples and these populations. Note also the consistent appearance of Scythians (especially Scythian_Pazyryk), Altai_IA, Mongola, and Karasuk_Outlier among the closest populations.
TUVINIAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Karasuk_outlier
0.03137152 0.04251899
Altai_IA Scythian_ZevakinoChilikta
0.04568074 0.04910874
Scythian_AldyBel Mongola
0.05677704 0.06354492
Sarmatian_Pokrovka Scythian_Samara
0.07472056 0.07539667
[1] "distance%=2.9007 / distance=0.029007"
Tuvinian
Scythian_Pazyryk 72.8
Itelmen 10.3
Karasuk_outlier 8.9
Mongola 7.9
KHAKASSIAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Karasuk_outlier
0.01634770 0.02464753
Altai_IA Scythian_ZevakinoChilikta
0.02681109 0.03452210
Scythian_AldyBel Sarmatian_Pokrovka
0.04073022 0.05703954
Scythian_Samara Srubnaya
0.05873087 0.06077001
[1] "distance%=1.3475 / distance=0.013475"
Khakass
Scythian_Pazyryk 55.15
Karasuk_outlier 37.00
Mongola 7.85
ALTAIAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Scythian_ZevakinoChilikta
0.02285839 0.03645670
Altai_IA Karasuk_outlier
0.03752331 0.04321664
Scythian_AldyBel Mongola
0.05141050 0.05272910
Tu Sarmatian_Pokrovka
0.06170647 0.06655838
Altaian
Scythian_Pazyryk 79.7
Mongola 18.4
Itelmen 1.9
BURYAT (they are mongolic, but autosomally virtually identical to Turkics. Historically, they are known to descend from Kurykans, who are Turkics. Their language may be due to linguistic shift after the 13th century, when they were conquered by Mongols.)
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Scythian_ZevakinoChilikta
0.03610198 0.04968497
Altai_IA Mongola
0.05186752 0.05526476
Karasuk_outlier Scythian_AldyBel
0.06010071 0.06395234
Tu Iran_IA
0.06594780 0.07885052
[1] "distance%=3.2282 / distance=0.032282"
Buryat
Scythian_Pazyryk 70.7
Mongola 25.0
Itelmen 4.3
The patterns seem very consistent: Turkics around the Altai are Scythian_Pazyryk+Mongola (inner Mongolian Mongols, almost pure ENA with very little West Eurasian ancestry) at approximately 4:1 ratio. As we move into Siberia, Itelmen (beringian-like) and Karasuk_Outlier ancestry starts to appear.
Since all fits were satisfactory, I did not pursue any alternate models with inclusions or exclusions of other populations.
SET 2: Central Asian Turks. Scythian_AldyBel starts to appear, together with more West Eurasian ancestry from the Caucasus and West Asia, but the Pazyryk+Mongola pattern still dominates. Closest populations still tend to be Scythians, Mongola, Karasuk_outlier followed by other East Asians (at a much further distance away), resembling the pattern for SET 1.
KYRGYZ
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Scythian_ZevakinoChilikta
0.02845224 0.03408819
Altai_IA Mongola
0.03849174 0.03896629
Tu Scythian_AldyBel
0.04762992 0.04901515
Karasuk_outlier Naxi
0.05192305 0.06266666
[1] "distance%=1.4539 / distance=0.014539"
Kyrgyz
Scythian_Pazyryk 54.60
Mongola 39.45
Chechen 2.75
Ulchi 2.75
Punjabi 0.45
KAZAKH
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Scythian_ZevakinoChilikta
0.02424266 0.02979828
Altai_IA Mongola
0.03345902 0.04304158
Scythian_AldyBel Karasuk_outlier
0.04389642 0.04791227
Tu Sarmatian_Pokrovka
0.05116875 0.05716372
[1] "distance%=1.2048 / distance=0.012048"
Kazakh
Scythian_Pazyryk 54.30
Mongola 30.35
Iran_IA 7.90
Itelmen 2.05
Punjabi 1.70
Chechen 1.35
Ulchi 1.25
Nordic_IA 1.10
KARAKALPAK
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Scythian_ZevakinoChilikta
0.02423779 0.03028732
Altai_IA Scythian_AldyBel
0.03358975 0.04048377
Mongola Karasuk_outlier
0.04539390 0.04737290
Srubnaya Chechen
0.05555522 0.05568733
[1] "distance%=1.2968 / distance=0.012968"
Karakalpak
Scythian_Pazyryk 53.25
Mongola 27.50
Ulchi 5.00
Chechen 4.80
Nordic_IA 4.30
Punjabi 3.75
Iran_IA 1.40
UZBEK
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_ZevakinoChilikta Altai_IA
0.02649022 0.02785061
Scythian_Pazyryk Scythian_AldyBel
0.03199633 0.03549600
Sarmatian_Pokrovka Iran_IA
0.04356709 0.04400175
Chechen Ulchi
0.04438390 0.04438390
[1] "distance%=0.7791 / distance=0.007791"
Uzbek
Scythian_Pazyryk 19.80
Iran_IA 19.45
Scythian_AldyBel 17.75
Mongola 12.40
Punjabi 10.60
Naxi 8.50
Ulchi 4.40
Chechen 4.35
Nordic_IA 1.80
Scythian_ZevakinoChilikta 0.50
Itelmen 0.45
There is one population that doesn't fit anywhere; these are the Yakuts. They are quite far from any other population (the closest I could find were Evenks, who were at 0.08 distance away, still very far) and I could not find good fits for them. The Siberian part is probably badly represented by the samples we have now.
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Evenk Scythian_Pazyryk
0.08718705 0.09910875
Altai_IA Scythian_ZevakinoChilikta
0.10842466 0.10906728
Karasuk_outlier Scythian_AldyBel
0.10908411 0.11013517
Mongola Chechen
0.11420687 0.11651864
[1] "distance%=9.8642 / distance=0.098642"
Yakut
Scythian_Pazyryk 89.25
Dai 10.70 <--- this is very weird, so I dropped it for the next fit
Punjabi 0.05
[1] "distance%=7.994 / distance=0.07994"
Yakut
Evenk 63.15
Scythian_Pazyryk 26.70
Chechen 6.15
Mongola 3.10
Narva_Lithuania 0.90
SET 3: Tajiks. Closest populations are different yet again: not Pazyryk, ZevakinoChilikta, Karasuk_Outlier, but Sarmatian_Pokrovka, Scythian_Samara, Altai_IA (shared by Tajiks and Turkics it seems), Scythian AldyBel (also shared), Srubnaya. For the fits, Scythian_Aldybel + Sarmatian_Pokrovka + Srubnaya_Outlier are common.
ISHKASHIM
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Sarmatian_Pokrovka Altai_IA
0.03287733 0.03523144
Scythian_Samara Karasuk
0.03739440 0.03858398
Scythian_AldyBel Iran_IA
0.03952654 0.03958612
Srubnaya Scythian_ZevakinoChilikta
0.03969207 0.04172257
[1] "distance%=1.3843 / distance=0.013843"
Tajik_Ishkashim
Iran_IA 31.25
Punjabi 21.30
Scythian_AldyBel 11.90
Sarmatian_Pokrovka 11.60
Srubnaya_outlier 11.40
Scythian_Samara 6.80
Scythian_ZevakinoChilikta 3.35
Chechen 2.40
SHUGNAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Sarmatian_Pokrovka Scythian_Samara
0.03059539 0.03372771
Karasuk Altai_IA
0.03599524 0.03653225
Srubnaya Scythian_AldyBel
0.03729174 0.03835506
Iran_IA Scythian_ZevakinoChilikta
0.03863595 0.04062046
[1] "distance%=1.345 / distance=0.01345"
Tajik_Shugnan
Iran_IA 36.50
Scythian_Samara 18.30
Punjabi 14.15
Srubnaya_outlier 12.85
Scythian_AldyBel 12.55
Sarmatian_Pokrovka 4.00
Mongola 1.65
RUSHAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Sarmatian_Pokrovka Scythian_Samara Karasuk
0.02915146 0.03349753 0.03518943
Altai_IA Srubnaya Scythian_AldyBel
0.03520315 0.03532160 0.03671922
Chechen Iran_IA
0.03938184 0.04015731
[1] "distance%=1.4672 / distance=0.014672"
Tajik_Rushan
Iran_IA 27.10
Scythian_AldyBel 19.60
Sarmatian_Pokrovka 18.95
Punjabi 11.20
Scythian_Samara 10.55
Chechen 7.25
Srubnaya_outlier 5.35
YAGNOBI
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Sarmatian_Pokrovka Iran_IA Scythian_Samara
0.03524610 0.03560888 0.03819717
Chechen Altai_IA Srubnaya
0.03856935 0.03916171 0.04043572
Scythian_AldyBel Karasuk
0.04104215 0.04129811
[1] "distance%=1.8059 / distance=0.018059"
Tajik_Yagnobi
Iran_IA 46.05
Scythian_AldyBel 17.95
Scythian_Samara 9.10
Srubnaya_outlier 8.85
Srubnaya 7.05
Chechen 6.15
Punjabi 3.35
Mongola 1.50
SET 4: Uralic-like Turkics. These populations, Chuvash and Tatars from the Volga, resemble the Uralic populations around them and do not share the turkic haplotypes centering on South Siberia identified by Yunusbaev et al. They may be due to linguistic shifts as well, because, judging from the recently uncovered part-hunnish genomes from Medieval Germany and the Balkans, the Turkic peoples, even at the earliest waves, were quite similar to South Siberians and others around the Altai in having high East Asian ancestry proper, and quite unlike present-day Volga peoples like Chuvash and Mishar Tatars, or Volga Uralics. Alternatively, if you do not believe the Huns in Europe represent the first para-Turkic incursions, the steppe was already quite East Asian at that point, so further issues from it are likely to retain this signal.
The closest population is Scythian_AldyBel (increasing closeness to AldyBel instead of Pazyryk as we move West, and occurrence of AldyBel at the edges of the Steppe while Pazyryk dominates in the centre, seems to be a consistent pattern), followed by Mezhovskaya (majority Andronovo and Sintashta-like population in C Siberia, identified by people like Parpola with some of the early Uralics), and Srubnaya.
CHUVASH
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_AldyBel Srubnaya
0.05332755 0.06003493
Scythian_Pazyryk Scythian_ZevakinoChilikta
0.06309898 0.06461331
Altai_IA Sarmatian_Pokrovka
0.06716522 0.06733498
Scythian_Samara Karasuk_outlier
0.06807143 0.06961130
[1] "distance%=5.0768 / distance=0.050768"
Chuvash
Scythian_AldyBel 73.45
Narva_Lithuania 6.75
Itelmen 5.40
Dai 5.20
Chechen 3.75
Ulchi 2.90
Comb_Ceramic 2.55
Dai is bad, so push out:
[1] "distance%=5.0557 / distance=0.050557"
Chuvash
Scythian_AldyBel 65.30
Mezhovskaya 18.15
Narva_Lithuania 5.55
Chechen 3.70
Evenk 3.70
Itelmen 2.65
Mongola 0.95
After fitting the Chuvash, I finalised the list of contributors for the other Volgaic populations (Uralics in SET 6), as well as the rest of this set.
TATAR
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_AldyBel Srubnaya
0.03080592 0.03752254
Scythian_Pazyryk Mezhovskaya
0.04248811 0.04396065
Nordic_IA Scythian_ZevakinoChilikta
0.04401260 0.04412678
Karasuk Altai_IA
0.04461660 0.04504374
[1] "distance%=2.3006 / distance=0.023006"
Tatar
Scythian_AldyBel 44.85
Nordic_IA 12.15
Comb_Ceramic 10.05
Chechen 9.20
Iran_IA 8.05
Mongola 7.25
Evenk 4.70
Itelmen 2.15
Narva_Lithuania 1.60
No Comb Ceramic:
[1] "distance%=2.3095 / distance=0.023095"
Tatar
Scythian_AldyBel 54.70
Nordic_IA 11.60
Chechen 11.05
Narva_Lithuania 6.60
Mongola 5.20
Iran_IA 5.10
Evenk 3.20
Itelmen 2.55
No Mongola:
[1] "distance%=2.3358 / distance=0.023358"
Tatar
Scythian_AldyBel 56.80
Nordic_IA 11.85
Chechen 11.45
Narva_Lithuania 6.70
Iran_IA 6.40
Evenk 3.65
Itelmen 3.15
TATAR_LIPKA (found in Poland and the Baltics)
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_AldyBel Scythian_Pazyryk
0.03337400 0.03853910
Scythian_ZevakinoChilikta Nordic_IA
0.04047207 0.04055040
Srubnaya Altai_IA
0.04166436 0.04233259
Karasuk Sarmatian_Pokrovka
0.04897414 0.04939444
[1] "distance%=1.767 / distance=0.01767"
Tatar_Lipka
Nordic_IA 26.45
Scythian_AldyBel 24.50
Mongola 20.05
Iran_IA 12.45
Chechen 6.00
Comb_Ceramic 5.60
Evenk 4.90
SET 5: Samoyed-like Uralics. These populations stretch across Northern Central Russia, and were very difficult to fit, as the closest aDNA sample was Karasuk_Outlier at 6% and fits with only ancients could not come close to representing them. Compare this:
KHAKASSIAN
KHAKASSIAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_Pazyryk Karasuk_outlier
0.01634770 0.02464753
Altai_IA Scythian_ZevakinoChilikta
0.02681109 0.03452210
Scythian_AldyBel Sarmatian_Pokrovka
0.04073022 0.05703954
Scythian_Samara Srubnaya
0.05873087 0.06077001
with this:
MANSI
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Karasuk_outlier Scythian_AldyBel
0.06129125 0.07258879
Mezhovskaya Scythian_Pazyryk
0.07543525 0.07740380
Altai_IA Karasuk
0.07741589 0.08003623
Scythian_ZevakinoChilikta Scythian_Samara
0.08165920 0.08256958
Ket, Nganasan and Selkup, however, are very close to the populations in this cluster and also very close to each other:
KET
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Selkup Karasuk_outlier Scythian_AldyBel
0.02975829 0.06658063 0.08590533
Altai_IA Scythian_Pazyryk Mezhovskaya
0.08691091 0.08882178 0.09201866
Karasuk Scythian_Samara
0.09226259 0.09507147
So I added back Ket, Nganasan and Selkup to the contributors for this set. Before we get to modelling them, lets see what Ket, Nganasan and Selkup turn out to be:
KET
[1] "distance%=2.4243 / distance=0.024243"
Ket
Selkup 84.55
EHG 9.40
Comb_Ceramic 3.45
Narva_Estonia 1.50
Karasuk_outlier 1.10
[1] "distance%=5.4538 / distance=0.054538"
Ket
Karasuk_outlier 43.0
Nganassan 32.3
EHG 12.4
AfontovaGora3 12.3
Ancients only + Itelmen:
[1] "distance%=6.5437 / distance=0.065437"
Ket
Karasuk_outlier 82.3
AfontovaGora3 9.0
EHG 4.5
Itelmen 4.2
NGANASAN (this was still poorly fit, even with Ket and Selkup included):
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Itelmen Ket
0.08611205 0.10269269
Ulchi Karasuk_outlier
0.10510413 0.11700033
Scythian_Pazyryk Scythian_AldyBel
0.11926721 0.12623954
Altai_IA Scythian_ZevakinoChilikta
0.12766151 0.12795882
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Selkup Itelmen
0.08098218 0.08611205
Ulchi Karasuk_outlier
0.10510413 0.11700033
Scythian_Pazyryk Scythian_AldyBel
0.11926721 0.12623954
Altai_IA Scythian_ZevakinoChilikta
0.12766151 0.12795882
[1] "distance%=7.0066 / distance=0.070066"
Nganassan
Itelmen 59.9
Ket 40.0
Ulchi 0.2
[1] "distance%=8.2133 / distance=0.082133"
Nganassan
Itelmen 76.20
Karasuk_outlier 23.65
Ulchi 0.10
Narva_Lithuania 0.05
[1] "distance%=6.2835 / distance=0.062835"
Nganassan
Selkup 53.50
Itelmen 46.35
Ulchi 0.15
SELKUP
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Ket Karasuk_outlier
0.02975829 0.07368082
Nganassan Scythian_Pazyryk
0.08098218 0.09240397
Scythian_AldyBel Altai_IA
0.09283311 0.09322548
Mezhovskaya Scythian_ZevakinoChilikta
0.09755882 0.09924904
[1] "distance%=1.7742 / distance=0.017742"
Selkup
Ket 73.45
Nganassan 23.90
AfontovaGora3 2.65
[1] "distance%=6.8956 / distance=0.068956"
Selkup
Karasuk_outlier 67.40
Itelmen 22.85
AfontovaGora3 9.75
Then, moving on to Khanty and Mansi:
KHANTY
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Ket Karasuk_outlier
0.03953348 0.06273628
Scythian_AldyBel Scythian_Pazyryk
0.07573712 0.08013188
Mezhovskaya Altai_IA
0.08014730 0.08052845
Karasuk Scythian_ZevakinoChilikta
0.08435784 0.08532921
[1] "distance%=2.9562 / distance=0.029562"
Khanty
Ket 58.35
Nganassan 18.25
AfontovaGora3 8.55
Srubnaya_outlier 8.45
Scythian_ZevakinoChilikta 4.15
Scythian_AldyBel 2.25
Without Ket, Nganasan or Selkup:
[1] "distance%=5.8237 / distance=0.058237"
Khanty
Karasuk_outlier 66.7
AfontovaGora3 17.2
Itelmen 15.8
EHG 0.3
MANSI
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Karasuk_outlier Scythian_AldyBel
0.06129125 0.07258879
Mezhovskaya Scythian_Pazyryk
0.07543525 0.07740380
Altai_IA Karasuk
0.07741589 0.08003623
Scythian_ZevakinoChilikta Scythian_Samara
0.08165920 0.08256958
[1] "distance%=3.2698 / distance=0.032698"
Mansi
Ket 56.25
Nganassan 15.05
Mezhovskaya 10.40
Srubnaya_outlier 7.70
AfontovaGora3 5.70
Scythian_ZevakinoChilikta 4.75
Scythian_AldyBel 0.15
Without Ket, Nganasan, or Selkup:
[1] "distance%=5.7804 / distance=0.057804"
Mansi
Karasuk_outlier 58.0
Srubnaya_outlier 20.6
Evenk 18.6
Itelmen 2.8
SET 6: Volga Uralics. These populations are close to Scythian_AldyBel, Mezhovskaya, Comb Ceramic, and Srubnaya and have varying levels of Nordic_IA as we move west into Europe proper. The appearance of Mezhovskaya is a very nice confirmation of the suspicions of the likes of Parpola and etc.
KOMI
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_AldyBel Srubnaya Mezhovskaya
0.03428793 0.03691489 0.04030518
Karasuk Nordic_IA Sarmatian_Pokrovka
0.04315808 0.04347712 0.04604267
Scythian_Samara Scythian_Pazyryk
0.04800987 0.04843802
[1] "distance%=2.3631 / distance=0.023631"
Komi
Scythian_AldyBel 32.90
Nordic_IA 20.10
Comb_Ceramic 15.70
Nganassan 9.65
Iran_IA 8.05
Samara_Eneolithic 6.85
Mongola 3.95
Chechen 1.40
Itelmen 1.05
Punjabi 0.35
[1] "distance%=2.4836 / distance=0.024836"
Komi
Scythian_AldyBel 42.25
Nordic_IA 16.10
Comb_Ceramic 15.05
Itelmen 9.25
Iran_IA 7.50
Mezhovskaya 4.00
Chechen 2.85
Narva_Lithuania 1.40
Samara_Eneolithic 1.20
Mongola 0.40
UDMURT
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_AldyBel Mezhovskaya Srubnaya
0.03459012 0.04080253 0.04316311
Karasuk Karasuk_outlier Sarmatian_Pokrovka
0.04639066 0.04743340 0.04903342
Scythian_Samara Scythian_Pazyryk
0.04910755 0.04974005
[1] "distance%=2.5721 / distance=0.025721"
Udmurt
Scythian_AldyBel 49.70
Mezhovskaya 16.95
Nganassan 12.70
Samara_Eneolithic 9.70
Chechen 4.60
Ket 3.45
Comb_Ceramic 2.90
[1] "distance%=2.8946 / distance=0.028946"
Udmurt
Scythian_AldyBel 54.70
Mezhovskaya 29.15
Itelmen 9.20
EHG 5.35
Chechen 1.60
MARI (the pattern looks similar to the populations around them, especially Komi, but fit is poor for some reason)
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Scythian_AldyBel Mezhovskaya
0.08122688 0.08454560
Scythian_Pazyryk Srubnaya
0.08830828 0.08903658
Scythian_ZevakinoChilikta Karasuk_outlier
0.09006742 0.09263677
Altai_IA Karasuk
0.09294106 0.09332671
[1] "distance%=7.8202 / distance=0.078202"
Mari
Scythian_AldyBel 57.0
Mezhovskaya 24.6
Nganassan 13.0
Comb_Ceramic 5.5
[1] "distance%=7.9324 / distance=0.079324"
Mari
Scythian_AldyBel 59.50
Mezhovskaya 30.45
Itelmen 7.80
Comb_Ceramic 1.40
Narva_Lithuania 0.85
SET 7: Last of all, the Finnics and Saami. Very similar to the Volga Uralics, similarity to Mezhovskaya, Scythian_AldyBel, and so on, except more Nordic_IA. When belarusian or Slavic is added back, the remainder generally aggregates to Mezhovskaya.
SAAMI
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Mezhovskaya Srubnaya Scythian_AldyBel
0.03980681 0.04278583 0.04393002
Nordic_IA Karasuk Scythian_Samara
0.04669220 0.04986604 0.05072782
Scythian_Pazyryk Sarmatian_Pokrovka
0.05346288 0.05416290
[1] "distance%=3.1638 / distance=0.031638"
Saami
Scythian_AldyBel 43.3
Mezhovskaya 31.4
Nordic_IA 15.7
Narva_Lithuania 9.6
Add an EEF source to let that vary freely:
[1] "distance%=3.1551 / distance=0.031551"
Saami
Scythian_AldyBel 41.95
Mezhovskaya 31.90
Nordic_IA 11.80
Narva_Lithuania 7.80
Globular_Amphora 3.80
Comb_Ceramic 2.70
EHG 0.05
Add non-Finnic Europeans like Slavic_Bohemia:
[1] "distance%=2.9006 / distance=0.029006"
Saami
Slavic_Bohemia 38.00
Mezhovskaya 25.00
Scythian_AldyBel 16.80
Comb_Ceramic 11.00
Karasuk_outlier 6.95
Narva_Lithuania 1.95
Nordic_IA 0.30
INGRIAN
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Nordic_IA Srubnaya Mezhovskaya
0.03442590 0.03588778 0.04400809
Karasuk Scythian_AldyBel Sarmatian_Pokrovka
0.04559721 0.04592725 0.04850874
Scythian_Samara Scythian_Pazyryk
0.05091807 0.05861495
[1] "distance%=2.5148 / distance=0.025148"
Ingrian
Nordic_IA 50.60
Comb_Ceramic 18.45
Iran_IA 11.05 <----Don't like this, no post-IA flow from Iran for sure. Purge
Samara_Eneolithic 9.30
Mongola 6.65
Itelmen 3.90
[1] "distance%=2.7026 / distance=0.027026"
Ingrian
Nordic_IA 55.05
Scythian_AldyBel 24.75
Samara_Eneolithic 9.70
Narva_Lithuania 8.05
Sarmatian_Pokrovka 2.45
Add EEF source:
[1] "distance%=2.6888 / distance=0.026888"
Ingrian
Nordic_IA 44.5
Comb_Ceramic 18.1
Scythian_AldyBel 17.1
Sarmatian_Pokrovka 13.6
Globular_Amphora 6.6
Add Europeans:
[1] "distance%=1.1782 / distance=0.011782"
Ingrian
Belarusian 49.30
Slavic_Bohemia 17.90
Mezhovskaya 13.35
Nordic_IA 8.80
Srubnaya_outlier 5.75
Comb_Ceramic 2.70
Narva_Lithuania 2.20
VEPS
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Nordic_IA Srubnaya Scythian_AldyBel
0.03946320 0.04187050 0.04715007
Mezhovskaya Karasuk Sarmatian_Pokrovka
0.04877829 0.05119489 0.05416271
Scythian_Samara Scythian_Pazyryk
0.05587790 0.05897818
[1] "distance%=3.1157 / distance=0.031157"
Vepsian
Nordic_IA 44.40
Scythian_AldyBel 40.35
Narva_Lithuania 9.40
Comb_Ceramic 5.40
Samara_Eneolithic 0.45
Add EEF Source:
[1] "distance%=3.0995 / distance=0.030995"
Vepsian
Nordic_IA 38.95
Scythian_AldyBel 37.15
Comb_Ceramic 13.35
Globular_Amphora 5.50
Narva_Lithuania 5.05
Add Europeans:
[1] "distance%=1.3069 / distance=0.013069"
Vepsian
Belarusian 68.35
Mezhovskaya 17.75
Slavic_Bohemia 7.55
Comb_Ceramic 5.75
Narva_Lithuania 0.55
FINNISH EAST
[1] "distance%=2.5918 / distance=0.025918"
Finnish_East
Nordic_IA 55.30
Scythian_AldyBel 24.70
Comb_Ceramic 7.40
Samara_Eneolithic 6.05
Narva_Lithuania 5.55
Mezhovskaya 1.00
Add EEF:
[1] "distance%=2.5993 / distance=0.025993"
Finnish_East
Nordic_IA 52.50
Scythian_AldyBel 23.30
Comb_Ceramic 13.95
Mezhovskaya 7.10
Narva_Lithuania 2.85
Globular_Amphora 0.30
Add Europeans:
[1] "distance%=1.5965 / distance=0.015965"
Finnish_East
Belarusian 35.20
Slavic_Bohemia 22.05
Nordic_IA 19.10
Comb_Ceramic 10.00
Mezhovskaya 6.65
Srubnaya_outlier 6.45
Narva_Lithuania 0.55
FINNISH
[1] "1. CLOSEST SINGLE ITEM DISTANCES"
Nordic_IA Srubnaya Scythian_AldyBel
0.02584531 0.03270961 0.04330134
Mezhovskaya Karasuk Scythian_Samara
0.04368180 0.04646210 0.05008357
Sarmatian_Pokrovka Scythian_Pazyryk
0.05031018 0.05771340
[1] "distance%=2.0815 / distance=0.020815"
Finnish
Nordic_IA 65.30
Scythian_AldyBel 24.05
Narva_Lithuania 5.35
Samara_Eneolithic 3.30
Comb_Ceramic 2.00
Add EEF:
[1] "distance%=2.0803 / distance=0.020803"
Finnish
Nordic_IA 61.70
Scythian_AldyBel 22.05
Comb_Ceramic 8.40
Mezhovskaya 3.25
Narva_Lithuania 2.30
Globular_Amphora 2.30
Add Europeans:
[1] "distance%=1.1816 / distance=0.011816"
Finnish
Nordic_IA 35.40
Belarusian 31.70
Slavic_Bohemia 17.05
Mezhovskaya 8.20
Comb_Ceramic 4.95
Srubnaya_outlier 2.70
General thoughts:
Very surprised at the levels of specificity demonstrated here! E.g.
Mezhovskaya: Very Finnic and Volga Uralic
Scythian AldyBel: Very Uralic and Iranic and tendency to increase to the West and at the edges of the steppe.
Scythian_Samara and Sarmatian_Pokrovka: Very Iranic.
Scythian Pazyryk: Very Turkic.
Mongola: Very Turkic.
Altai_IA: Equally related to Turkics and Iranics
Karasuk_Outlier: Found in Central and Northeast Siberia
Comb Ceramic, Narva Lithuania, Narva Estonia: Found in European Siberia and Finnic-Baltic region.
I'd wager that the Iron Age Finnish genomes, when released, will look like Mezhovskaya (mostly Andronovo/Sintashta plus small fraction local EHG-ENA type thing) + a minority fraction of Comb Ceramic/Narva + maybe a little bit more of Karasuk_outlier type ancestry.
It seems to me that a part of the forest-Steppe was "Uralicised" after Andronovo and Sintashta, and this "Mezhovskaya" gene pool forms the core of the Uralics, from Saami to the Volga region, and further that this steppic population spread East before the whole horizon started pushing north into the Taiga in Siberia like parallel teeth on a comb (we know from linguistics that this is the case). Among the Eastern Uralics, i.e. Mansi-Khanty and Samoyed-like and Samoyed Uralics in Central and Northeastern Siberia, this "Mezhovskaya" population was diluted by some AG3-EHG-ENA mix, akin to the Nganasans today, for which we still have no ancient representative. Later waves of Iranic Scythians, like AldyBel, no doubt further ENA-ised the gene pool.
The Iranic Scythians probably existed in a continuum between Sarmatian_Pokrovka and Scythian_AldyBel, with Srubnaya_Outlier ancestry as well, in SC Asia. Then a Pazyryk Scythian+Mongola mix formed in South Siberia, possibly the Mongola is Baikal HGs or alternatively a population issuing from South of the Gobi (seen from the Slab Grave culture crania), and these were the Turkics who then splashed back West, occupying all the lowlands of C Asia, and pushed the AldyBel types to the edges of the Steppe.
Now, we can confirm these tendencies (Mezhovskaya looks Uralic, Pazyryk and Eastern Scythians look Turkic) using shared segments, with the caveats that this is haploid data and unphaseable, so IBD is not 100% reliable. All the same, the patterns are very striking. I quote from this thread:
Last of all, we can attempt to confirm that the analysis is correct by looking at clines in PCA itself, corresponding to patterns of gene flow.
We can't use Global25 because we cannot be sure there are PCs that so happen to load onto the dimensions of variation that most distinguish Siberians from each other. There are PCAs from published literature, however, which will do very well.
We see that the split of the populations into seven sets distinguished by different behaviour in nMonte is exactly corroborated by the positions of the populations in PCA space:
The sets of populations are:
SET 1: Altai-centered Turkics
SET 2: Central Asian Turkics
SET 3: Tajiks.
SET 4: Uralic-like Turkics, such as Chuvash and Tatars, + Lipka Tatars.
SET 5: Samoyed-like Uralics.
SET 6: Volga Uralics.
SET 7: Finnics and Saami
PCA of Pankratov et al on Lipka Tatars:
https://anthrogenica.com/attachment.php?attachmentid=22259&d=1521684028
Note the overlap of Chucash, Tatar, and Volga Uralics in cluster 4 + 6. Also note the close clustering of Altai Turkics + Buryat.
Note that Lipka Tatars differ from the rest of the Chuvash+Tatar by having double the Iran_IA and prescence of 20% Mongola. This can be seen in the above PCA, where Lipka Tatars are in purple.
[1] "distance%=1.767 / distance=0.01767"
Tatar_Lipka
Nordic_IA 26.45
Scythian_AldyBel 24.50
Mongola 20.05
Iran_IA 12.45
Chechen 6.00
Comb_Ceramic 5.60
Evenk 4.90
PCA of Pugach et al on Siberians:
https://anthrogenica.com/attachment.php?attachmentid=22260&d=1521684101
Note the clustering of Khanty, Samoyeds (Nenets, Selkup and etc.) and Ket in cluster 5. Interestingly, this PCA makes me think that Evenk may be a good source for Nganasan. Maybe will try this later.
The new PCA in the paper suggests that Yukagir and Nganasan form a pole of their own variation in Siberia, being the most "Northeastern" of all Siberians, which probably explains why Nganasan received such horrible fits (because we could not "reach" Nganasan as it was outside the convex hull formed by the contributing populations). They are at the very top-right of the PCA.
https://anthrogenica.com/attachment.php?attachmentid=22266&d=1521746746
Given that (and also a tip-off from Shaikorth, thanks brah) I decided to re-run the models with Yukagir/Nganassan as a contributing source. It seems the Bolshoi samples will justify this too.
You can see that the new PCA is absolutely beautiful, it traces each linguistic migration as a line of populations suspended between East and West Eurasia. The "Uralic cline" is in purple stretching from Nganasan and Yukagir to Saami and Scandinavians, the Turkic cline streches from Buryat and Altaian to Turkish and Abkhazian. They are marked in the PCA below. You can also see that Mongola did not seem optimal for the end of the Turkic line, rather Hezhen and Daur are closer, and Hezhen+Daur+Ulchi looks best. These are Tungusic and Mongolic populations in Manchuria, on the far Eastern edge of the Steppes.
I also checked the closest populations for the Scythians, which allowed me to mark their position (since they were not included on this PCA) using a purple lozenge.
https://anthrogenica.com/attachment.php?attachmentid=22267&d=1521748539