2
We can get an extremely good fit for Chuvashes by modeling them as 60% Mari and 40% Mishar (d=.009) or 70% Mari and 30% Mordovian (d=.013). Doesn't it mean that even if Maris are drifted, their drift is shared by Chuvashes?
Like Maris, Chuvashes are also far from every other population average in G25:
Distance to: Chuvash
.048 Besermyan
.049 Udmurt
.056 Mari
.064 Tatar_Kazan
.069 FIN_Levanluhta_IA
.071 Saami
.072 RUS_Chalmny-Varre
.073 Komi
.077 Saami_Kola
.086 VK2020_NOR_North_VA_o2
.087 Tatar_Mishar
.090 RUS_Mezhovskaya
.092 Tatar_Lipka
.096 MDA_Cimmerian
.097 RUS_Tagar
.097 VK2020_NOR_North_VA_o1
Therefore it is surprising that my two-way models for Chuvashes have such good fit.
In the model below, if I changed Baltic_LVA_MN into NOR_N_HG, it reduced the average distance from .0567 to .0561. The combined average proportion of Barcin and CHG stayed as 28.1% (from 21.3 Barcin + 6.8% CHG to 20.9% Barcin + 7.2% CHG).
Above some Saami individuals randomly have 0% RUS_Bolshoy_Oleni_Ostrov_o but others have 10-20%. Also one Chuvash has 42% RUS_Bolshoy_Oleni_Ostrov_o but another has 13%. The individuals with lower BOO_o have higher EHG (Volga-Kama_N) and higher kra001 (Krasnoyarsk_BA).
RUS_Bolshoy_Oleni_Ostrov_o is just half EHG and half RUS_Krasnoyarsk_BA, so it's not needed as an additional component in modeling Uralics:
Target: RUS_Bolshoy_Oleni_Ostrov_o
d=.018 - 79% RUS_Bolshoy_Oleni_Ostrov + 21% RUS_Krasnoyarsk_BA
d=.029 - 47% RUS_Samara_HG + 53% RUS_Krasnoyarsk_BA
d=.034 - 48% RUS_Karelia_HG + 52% RUS_Krasnoyarsk_BA
d=.034 - 48% RUS_Sidelkino_HG + 52% RUS_Krasnoyarsk_BA
d=.035 - 47% RUS_Khvalynsk_En + 53% RUS_Krasnoyarsk_BA
d=.036 - 52% RUS_Sintashta_MLBA_o3 + 48% RUS_Krasnoyarsk_BA
d=.038 - 47% RUS_Volga-Kama_N + 53% RUS_Krasnoyarsk_BA
d=.040 - 46% RUS_Veretye_Meso + 54% RUS_Krasnoyarsk_BA
I think we're still pure enough... I'm terrified to imagine what even purer Uralics would be like.
Bookmarks