December 08, 2012

Main orientations of human genetic differentiation (Jay et al. 2012)

Mol Biol Evol (2012) doi: 10.1093/molbev/mss259

Anisotropic isolation by distance: the main orientations of human genetic differentiation

Flora Jay et al.

Genetic differentiation among human populations is greatly influenced by geography due to the accumulation of local allele frequency differences. However, little is known about the possibly different increment of genetic differentiation along the different geographical axes (north-south, east-west, etc). Here we provide new methods to examine the asymmetrical patterns of genetic differentiation. We analyzed genome-wide polymorphism data from populations in Africa (n = 29), Asia (n = 26), America (n = 9) and Europe (n = 38), and we found that the major orientations of genetic differentiation are north-south in Europe and Africa, east-west in Asia, but no preferential orientation was found in the Americas. Additionally, we showed that the localization of the individual geographic origins based on SNP data was not equally precise along all orientations. Confirming our findings, we obtained that in each continent, the orientation along which the precision is maximal corresponds to the orientation of maximum differentiation. Our results have implications for interpreting human genetic variation in terms of isolation by distance and spatial range expansion processes. In Europe for instance, the precise NNW-SSE axis of main European differentiation can not be explained by a simple Neolithic demic diffusion model without admixture with the local populations because in that case the orientation of greatest differentiation should be perpendicular to the direction of expansion. In addition to humans, anisotropic analyses can guide the description of genetic differentiation for other organisms and provide information on expansions of invasive species or the processes of plant dispersal.



shenandoah said...

"but no preferential orientation was found in the Americas."

That seems a little strange to me. Doesn't make much sense, in light of what we know already. I would expect differentiation between North and South American continents; and also between East and West of the Mississippi River.

However, seems to me that ~parentage would have far more influence than geography.

eurologist said...

Not sure how much this paper adds to the literature. I found Europe in Fig. 5 instructive: basically, the Finns and Russians destroy the picture - I would have liked to see an analysis without them.

These two pull most Northern Europeans south, and most Eastern population west, because the other populations can't be as "north" as Finns, and Russians of course are abnormally east for their make-up. Slovakia is a weird outlier, but the Swiss make good sense.

genefan said...

As seen on Figure 5 of this study, Europe makes almost the same picture we saw recently on 23andMe and on Davidski's blog.
The only populations more or less matching their current geographical location, presumably living there for a long time are Italians, French and Spanish.
On the Balkans true for Albanians, Kosovar have a little shift towards Serbia, where they used to live 150 y.a.
Very strangely Serbians also have very little shift, while Bulgarians and Macedonians merge, but to the West of the Serbs!
Greeks again show strong affinity towards the Italians and are placed in the middle of the Adriatic, while historically should have some Anatolian bias not only from the Neolithic framers, but also with the recent influx of refugees from Turkey.
The Turkish are also very displaced unless they used Balkan Turks who have some Tartar admixture.
Eastern Europe - again a proof Poles, Russian and Ukrainians are from the same stock, recently expanded to the East with the Russian Empire.
However Slovakia, maybe because they used only 1 sample(Ashkenazi?), should be some mistake going to South Italy.

Davidski said...

Eurologist, you forgot to take your pills again.

The POPRES Russians are genetically very western. I don't know who they are exactly, but most of them could easily be Eastern Poles. The Vologda Russians from the HGDP are very different.

Anyway, if we ignore the Slovakian sample, this analysis matches the SNP structure of R1a.

Subclades like M458 and L365 are native to Poland (West Slavic), while others like Z92 just to the east of the current Polish border (East Slavic and Baltic).

So it looks like there's been great stability in what is now Poland in terms of genetics since the Bronze Age. That's despite all the border changes and a wide variety of ethnic minorities living in some Polish-Lithuanian cities at one time or another.

Anonymous said...

The point of the study eludes me. I say this as any person can see how people vary in their physical characteristics with longitude and latitude. If you that Eurasia, then it goes from people who can be categorized as wholly European in the western extremity to wholly Asian in the eastern extremity. And everyone knows that in Europe, the Pyrenees, the Alps and to a smaller extent, the Dinaric Alps are genetic barriers hence the restriction of the Italians to south of the Alps, the Spanish to south of the Pyrenees, and origin points of the Greeks outside of the Dinaric Alps near to the Italian peninsula. The mathematics, statistics and genetic theory is not needed to explain what is there to be seen.

I was more interested in a timeline of colonization. Obviously, the Europeans did not spring out of Europe's soils but originated from antecedents elsewhere. The study just assumes all those people other than those whose migrations are known, the Xhosa for instance, just sprung out of the ground where they are found today. Frankly it is rather poor science.

eurologist said...

Eurologist, you forgot to take your pills again


Why would you even belch such an insult without truly understanding what I said?

Russians in this analysis are placed far east, but Slavic people on average are not. Here, Russians (and Finns) define "Far East" although the former are not, as much, really.

Similarly, here, Finns define "North," although they are just different in a different dimension and are thus destroying the Isles and other Scandinavian North reality.

eurologist said...

To clarify and expand on this, if you take a European group of populations and include Finns in PCA, what happens is that Finns, being so far North and East, will dominate that geographical position. However, Finns are besides of being NE.

Thus, the true signature of Europeans without influence over the past ~3,000 years gets fogged.

Same for Russians. They certainly picked up far eastern Europeans, and more East Asians/Siberians/ Amerindians than many other Eastern Europeans. But when weighted geographically, they are incredibly far east. Thus, they taint the picture by moving all other Slavic populations artificially west.

Onur Dincer said...


You should not take the Figure 5 results so seriously, most notably due to the very unbalanced sampling of the populations (the excessively sampled populations such as the Italians dominate the positioning of all the other populations).

BTW, very few Balkan Turks have any known or probable Tatar admixture.

genefan said...

Onur, I can't talk about all Balkan Turks, but Bulgarian Turks definitely have a big Tartar input and many remember it.
Settling of different Tartar groups in what is now Bulgaria started maybe in the 16 century, it is documented a group of Volga Tartars found Tatar Pazarjik. With the expansion of the Russian empire more and more Moslims from there took refuge in the Ottoman empire.
There were big population exchange agreements in 1829 and 1860. Probably tens of thousand if not more Crimean Tartars, Nogays, Circassian, etc, moved to the Rum part of the empire. Nowadays only about 2000 claim Tartar ancestry in Bulgaria. Where are the others - mixed with the other Turks, of course.

Kristiina said...

I am an eager reader of this blog but as a Finn I got somewhat insulted by the tone of some of your comments. Nonetheless, I am quite used to hear this kind of stuff. I think that we should broaden our views and not stick to stereotypes. Genetic studies should make us all feel closer to each other and not make some of us seem aliens.
The East Asian ancestry of the Finnish people is not very extensive and it consists of Central Asian and Siberian markers. On maternal side, we share haplogroups Z and D5, but the share of Z is only 2,5 % in Finnish people and D5 is even more rare. In China these are not common haplogroups either; they are clearly Siberian in their distribution. On paternal side, we have loads of N, the brother haplogroup of O, but these both haplogroups are derivatives of MNOPS and go back to the same node as haplogroup R. They might all be of Central Asian origin.
I see the Finnish and Saami people northern people; we combine the northernmost genes from the West and the East. IMO, for example, the haplogroup I1 is a paleoeuropean marker that is not connected to Indoeuropean or Finno-Ugric languages but instead represents the first settlers of Northern Europe. The Saami people have among the highest levels of haplogroup I in the world.
I have seen studies and opinions (also here) where the Finnish people are considered more archaic and more representatives of Mesolithic Europeans than Modern Europeans in general. From this point of view, I think that Finnish people deserve the position they have been given in the comparative genetic studies. I have the impression that isolated people often harbour older haplogroups that can be very illuminating in reconstructing our trails on the globe.
By the way, my own MtDNA haplogroup is I.

Onur Dincer said...


I am well aware of the Tatar emigration to Rum (i.e., Anatolia + the Balkans) after the Russian advancements in Crimea. The impact of the Tatar emigration is not as big as you claim and it is territorially pretty restricted (whether in the Balkans or in Anatolia). BTW, one of my grandfathers is a Turk from Bulgaria and I have enough information about Turks from Bulgaria.

Onur Dincer said...

Ponto, I was referring to the sampling of just this study, not the literature as a whole.

Unknown said...

Well said

Onur Dincer said...

Lastly, the Tatar emigration to Rum (Anatolia + the Balkans) is so recent (during the last 200 years) that any Turk with any Tatar descent is aware of the Tatar descent. Turks with any Tatar descent are a very small minority in the whole Turkish population (whether in the Balkans or Anatolia). Even in Bulgaria Tatars or other peoples with any Tatar descent are largely restricted to parts of the Dobruja region (a region far away from my grandfather's region).

eurologist said...

For those who did not actually read the paper, I think there are three important points:
1. Some populations are only represented by one (Slovakia, Ukraine) or a few individuals (Russia, Turkey, Cyprus) others by many tens or >100 (UK, Spain, Portugal, Swiss, Italy; the two Finns combined = 81)
2. Figure 5 is not just based on the two main PCA components, but on 17 components.
3. It appears that the regression did not compensate for the different weight of the populations.

So, as I said before, Finland for example makes every other northern population move south and partially west, because the others do not possess much of what trains the model on N and NE. Likewise, most northern populations don't have what makes Turkey and Cyprus define E, so they move W. The Isle populations don't have what makes Spain and Portugal define W, so they move E.

Lastly, the problem in Europe is that clines are not the same in the N and S nor in the W or E. For example, in the E 'West Asian' has a strong N->S cline, in the S it is a W->E cline. Because of that and the use of so many higher components defined by isolates (e.g., Finns or Iberians), a linear regression method is insufficient - they should have used at least a quadratic one. In the end, of course, Russians have little that would make the model move them much away from Poland or Ukraine - but a better model would make both their common origin and their differences (Baltic and German contribution levels) more apparent.