European population substructure…Finns in the corner again
Dienekes has a long post on a new paper, Correlation between Genetic and Geographic Structure in Europe. I took the figure and decided to just label the geographic provenance of the primary clusters which emerged when one plotted them along the two largest dimensions of variation (Y axis is 1st component, X is 2nd component) for easy gestalt absorption. To a large extent genetics does seem to follow geography. Obviously the labels for Italy and Spain really underestimate the area these two samples span, so they are meant to be general pointers, not precise indicators of the center of a given cluster. Note Finland…too terrified to join the party I assume?
Update: Also, see what Sandman sayeth.
Update II: And Genetic Future.
Labels: Finn baiting, Genetics
A Finn strikes back! Against Sweden, at least…aise.blogspot.com/2008/07/learning-swedish-is-essential.html
http://fourthcheckr
Razib
I’m way out of my depth on this subject. All I see is pretty coloured dots. I’m going to ask the obvious silly questions. Be gentle, someone’s got to do it.
1. You don’t mark Germany. Do the turquoise dots over both the British Isles and an area to the left/west of Poland represent the “Germanic” peoples?
2. Is that the same Celtic green over the British Isles as over, err, Poland? Are the Poles, in fact, Celts?
3. Are the black dots in the Spanish silver Basque?
4. The Finnish language supposedly bears some very distant relationship to Hungarian. But I can’t see any rogue yellow dots in the area where, I assume, Hungary should be. Any info on this?
Thanks in advance…
georgesdelatour, the colors represent the country that individual samples come from. The position on X and Y represent how similar/different their SNP compliment is, to be crude.
http://www.gnxp.com/blog/2008/04/european-population-substructureagain.php
http://www.gnxp.com/blog/2007/11/european-american-population.php
http://www.gnxp.com/blog/2008/04/snps-dont-lie.php
didn’t mark germany cuz they’re all smeared on the ‘inside.’ also, look at the *shapes* as well as color. the ticks aren’t very distinctive, but they’re different. yeah to 4. the hungarians aren’t genetically close to the finns, they’re more like czechs and south slavs. that makes sense, as it seems likely that the magyar language was adopted by the peasants which the original magyar tribes conquered. the tribes weren’t numerous. language *correlates* with genes, but it is an imperfect correlation. this is why you do studies like this. each dot is a person.
p.s. some of the ticks occlude others. this is why their shapes can be confusing. but there’s nothing SUPER surprising in this. if sardinians were included they’d likely be as weird as the finns; being on an island is just as isolating as being a shy freak.
For a long time I’ve thought that three of the biggest genetically-isolating factors for most of history would be geography, religion, and language. Basically, people tend to marry and have children with others who match on two or three of those factors.
For example, a little while back Razib nailed that puzzle about why Greeks are genetically close to Near Easterners. For a thousand-odd years, the Greek/Byzantine Empire had its population center-of-gravity in the Near East, resulting in heavy gene flow. And the Greeks there also tended to intermarry with the Greeks in Greece, or actually went there after the Greek/Turkish expulsions.
It isn’t surprising that there’s a gap between Poles and Finns in the results if there’s a gap in the samples. Until they do some analysis for Lithuanians, Latvians and Estonians, the best assumption is that they’d fill that gap with that sequence of clusters and Finns would not seem so odd.
georgesdelatour,
there’s a key on the right side of the graph. (I missed it, too.)
It isn’t surprising that there’s a gap between Poles and Finns in the results if there’s a gap in the samples. Until they do some analysis for Lithuanians, Latvians and Estonians, the best assumption is that they’d fill that gap with that sequence of clusters and Finns would not seem so odd.
the gap between your kind and your erstwhile herrenvolk was a bit larger than i expected….
The light blue squares are listed as IE. Does IE represent the Irish or some other group?
irish. that’s why you see those squares around “british isles.” though some of those aren’t “squares” because of occlusion by the UK inverted triangles.
I’ve uploaded the file to forum.
[very much like an introverted freak to make threats over the internet]
Norway (Førde) NO
Sweden (Uppsala) SE
Finland (Helsinki) FI
Ireland IE
UK (London) UK
Denmark (Copenhagen) DK
Netherlands (Rotterdam) NL
Germany I (Kiel) DE1
Germany II (Augsburg) DE2
Austria (Tyrol) AT
Switzerland (Lausanne) CH
France (Lyon) FR
Portugal PT
Spain I ES1
Spain II (Barcelona) ES2
Italy I IT1
Italy II (Marches) IT2
Former Yugoslavia YU
Northern Greece EL
Hungary HU
Romania RO
Poland (Warsaw) PO
Czech Republic (Prague) CZ
Dienikes says:
clear clustering of individuals from different ethnic groups within the European continuum, indicating that ethnic groups are not only cultural, but to some extent biological entities.
Uh? I see the exact opposite. The data points are smeared all over each other like an Impressionist painting. The data seem to match geographical distribution almost perfectly (if you squint your eyes, you can even see the Adriatic in the graph!)
Any remaining “clustering” might just as well be a result of sampling effect, For example, the British Isles seem “closer” to Scandinavia than to France – but would you get the same results if the French sample included Britanny? Also, look at the Spanish data: if the Catalans (ES1) had been removed, you would see a completely separate Spanish cluster – but since they are included, Spain morphs gracefully into France.
In short, from this graph, national borders do not seem to introduce remarkable discontinuities.
Rather, the graph seems to confirm the conventional wisdom that the best predictor of genetic distance is geographical distance, that gene variation is gradual over space, and that (almost) no large-scale gene flow (other than homogeneous diffusion) seems to have occurred since Stone Age (possible exceptions include Greece / Southern Italy).
razib: and your erstwhile herrenvolk
Dude, I think you took that “stop using latin words” thing a bit overboard…
I’m too lazy (and/or busy) to read the original paper, so excuse me if the following point is already answered.
In this kind of ethno-genetic study, to the best of my knowledge, the samples are seldom strictly random samples of the inhabitants of an area. In the UK, for example, they would usually deliberately exclude people with any known non-European ancestry. Sometimes there would be additional restrictions, e.g. a local sample would be confined to people all of whose grandparents came from the same locality. Presumably similar restrictions would apply, mutatis mutandis, in other countries, e.g. a sample of ‘Basques’ might exclude people known to have migrated to the Basque area from elsewhere in Spain. And – maybe – in Finland it might exclude people whose first language is Swedish.
My point is simply that without knowing exactly what the samples include and exclude it is risky to draw conclusions, as to some extent the sampling methodology is often biased towards identifying ‘pure’ or ‘indigenous’ types.
It isn’t surprising that there’s a gap between Poles and Finns in the results if there’s a gap in the samples. Until they do some analysis for Lithuanians, Latvians and Estonians, the best assumption is that they’d fill that gap with that sequence of clusters and Finns would not seem so odd.
Yeah, but that doesn’t explain the gap between the Finns and the adjacent Swedes. You’d expect the Baltic peoples to be north of the Poles, northeast of the Germans, and east of the Swedes and Finns.
Interesting if Estonians and Lithuanians differentiate, though.
Yeah, but that doesn’t explain the gap between the Finns and the adjacent Swedes. You’d expect the Baltic peoples to be north of the Poles, northeast of the Germans, and east of the Swedes and Finns.
here’s a banal explanation: the swedish samples were from the very far south of sweden, which was historically long under danish rule. it may be that northern sweden would be far closer to finland (a lot of these people are swedishized anyway).
i think that this is kind of like the race debate. i agree with dienekes and not toto, but perhaps we have different ideas of what we should expect and what “distinct” means….
“here’s a banal explanation: the swedish samples were from the very far south of sweden, which was historically long under danish rule. it may be that northern sweden would be far closer to finland (a lot of these people are swedishized anyway).”
The swedish samples were from Upsala north of Stockholm what is pretty north.
The swedish samples were from Upsala north of Stockholm what is pretty north.
ah, thanks. ok, finns are just weird.
The weird thing is, it actually looks like a map of Europe – Germany in the middle, UK, NL, NO, SE, FI, across the top, in that order; France to the south of Germany; Spain, Italy, and Greece to the south of France, in that order; YU to the north of Greece…
I still don’t understand why this is surprising. With the sampling Finns are effectively an island population and unlike with the British Isles, our “original home” is not represented at all.
The reason there’s a gap between Finns and Swedes not like what’s seen with other neighbours is that we’re not like the other neighbours. Greeks and South Slavs likely mostly come from the same waves of settlement, Danes, Swedes and Norwegians largely come from the same waves of settlement that emanated from Denmark and so on… but Finns and Swedes have a fundamentally different origin. This is where two waves of settlement collided.
Sweden was populated from what’s now Denmark, Finland was populated from what’s now Russia. Whatever similarities there are come from later admixture (and for that: Finnish admixture in Swedes will be underestimated by sampling Uppsala, Swedish admixture in Finns will be overestimated by sampling Helsinki) or the general similarity between Eastern and Western Europeans.
I agree with Jaakeli. Finns are like Mordvins, more or less, and no one even knows who the Mordvins are.
Uppsala, where this sample came from, is Central East Sweden. Northern Swedes would indeed cluster much closer to Finns.
In fact, if we had more regions sampled from each country, as well as smaller countries like the Baltic states, the overlapping would be much, much stronger.
The Finns from eastern Finland would still be the furthest out, but there would be no space between them and the rest.
And then, of course, if we were looking at inter-continental markers, these Europeans would be even closer to each other, and the more exotic samples from far Eastern Europe, West Asia and North Africa would be the outliers.