View Full Version : For Maciamo: PCA vs. Map of Polish regions

10-12-17, 03:43

I collected Polish and neighbouring regional samples that I found (samples assigned to regions mostly based on genealogical info) and added some academic samples (Mazovia2 = seven samples from Warsaw from Behar 2013 study). I created a PCA. Then I superimposed this PCA with regional Polish and neighbouring (including East German) averages on a map.

Map1: https://i.imgur.com/Xx4lrBX.png


And after adding Lithuanian average to the PCA:

Map2: https://i.imgur.com/Vwk3mLW.png


Compare to my suggested division into regions:



10-12-17, 04:13
Sample sizes for Polish and German averages:

Neumark & Posen-Westpreussen - 17
Sachsen - 14 (including 8 from Leipzig)
Pommern and Vorpommern - 8
Sachsen-Anhalt - 2
Mecklenburg - 7
Brandenburg - 15

Greater Poland - 24 (including 15 from Poznan)
Upper Silesia - 18
Lusatian Sorbs - 8
Pomerelia (among them Kashubians) - 12
Podlachia (Podlasie) - 26
Podkarpacie (eastern Polish Mountains) - 15

Sudovia (SuwaƂki Region) - 3

Mazovia1 - 31 Mazovians from various areas
Mazovia2 - 7 Warsaw samples from Behar 2013

SE-Kresy (Poles from Western Ukraine) - 10
Red Ruthenia, or Lubelskie - 8 (5 from Lublin)
Northern Poland (Kujawsko-Pomorskie) - 11
NE-Kresy (Poles from Vilno-Grodno Regions) - 2

Czechs - 33 samples

I forgot to add Estonian Poles from Kushniarevich 2015.

10-12-17, 05:02
Vinnytsia, Poltava, Kharkiv, Chernigov, Boryslav and Zboriv are of course ethnic Ukrainians.

Polesye, Grodno, Vitebsk and Mogilev are Belarusians. Zakarpattia are Carpathian Rusyns.

10-12-17, 10:13
Thanks, Tomenable. That's very interesting.

10-12-17, 19:57
Thanks, Tomenable. That's very interesting.

Also let's add that a 2D PCA can be misleading, because in that PCA both Mediterranean admixtures and North-Western admixtures are pulling samples to the west (south-west and north-west respectively). So for example Zakarpattia is so far to the west not due to high level of North-West Euro, but rather due to high level of Iberian-like and Italian-like admixtures.

I think only a 3D PCA would reveal the whole picture.

10-12-17, 21:41
Differences between East German and West Slavic regional samples in Eurogenes K36:



10-12-17, 23:00
Here is my PCA with all individual samples. Czechs pull towards South-West Euro admixtures (Iberian, Italian, etc.) while East Germans pull towards North-West Euro admixtures (North Sea, North Atlantic, etc.):

Green - Poles, Red - Czechs, Blue (including Light Blue) - Germans



Here I added East Slavic labels (averages). Label "Lviv" = Poles from Lviv, not Ukrainians:

"Austria" = East Austrians (mix of regions but not Tyrol, which plots much further west):


As you can see East Slavs overlap only with Eastern Poles, not with Polish core.

Some of Western Ukrainians (Boryslav, Zboriv, Zakarpattia) are close to Czechs:


Edit: Of course UA Vitebsk should be BY Vitebsk (as Vitebsk is in Belarus).

What surprised me is that Poles from Lviv and surrounding areas plot more to the west than geography would suggest. They plot close to Lesser Poles and close to Upper Silesians. Map posted below show them as SE-Kresy (South-East Kresy). On the other hand, Poles from Wilno and Grodno region are clearly Baltic-shifted and plot close to Belarusians and Lithuanians.

19-12-17, 12:47
Thanks, Tomenable. That's very interesting.

This map is based on the proportions of Nordids to Baltids in former East Germany:



Apparently in East Prussia Samland and Memelland areas were particularly Baltid.

Which probably tells us about high % of Old Prussian ancestry in those areas.