June 26, 2011

Modeling spread of R1b1b2 into Europe

A few comments:

I don't really get how the authors came up with their three models:
  • the "Paleolithic" model assumes beginning of expansion at 21 kBP; as this was unsupported under any mutation rate, I won't bother with it further.
  • the "Neolithic" models had expansion beginning at 10.5kBP, that is about 2k prior to the known earliest colonization of Europe from Anatolia, that occurred around 6,700 years BC.
  • the "recent" model has the expansion starting at 3kBP, but already at 3kBP R1b1b2 makes its earliest appearance at Lichtenstein in Germany, and indeed 3kBP takes us to the Iron Age, a period extremely unlikely to have been one of major dispersals into Europe, dispersals that could not have gone unnoticed by the literate civilizations of the time.

Here is what the authors have to say:
Our results show that an expansion in Neolithic or Mesolithic times (350 generations ago or 10 ky) leads to a lower sum of squared errors than post-glacial re-expansion started 700 generations ago (21 ky ago), regardless of assuming a GMR or EMR model (Figure 2 and Table 2). Using GMRs, simulations of recent (100 generations ago) and rapid expansions from three distinct origins provided a better fit to the geographical distribution of microsatellite diversity than did models with expansion started 350 generations ago. Although models of recent origins using GMRs provided poorer fit than a model of Neolithic expansion using the EMR (Figure 2 and Table 2), the small observed difference makes them however difficult to discriminate (odd ratio = 1.7; Figure 3).
So, basically the "evolutionary" rate stands its own against the "germline" rate assuming that the Neolithic expansion started 2,000 years before it did, and using the "germline" rate for an expansion at a much later time than anyone would believe.

Not directly related to the paper, I looked at the ongoing Dodecad v3 results to possibly correlate the spread of R-M269 to Western Europe with the autosomal evidence. It appears that the "West European" autosomal component shows a stronger relationship to the "West Asian" one than the "East European" one. This is consistent with an episode of gene flow into Europe from West Asia that affected Western more than Eastern Europe, which parallels the R-M269 distribution in Europe.

More interestingly, I had previously traced a peculiar "Dagestan component" in Europe that, counter-intuitively, seemed to be maximized in the Northwest. Looking at the recent Caucasus Y-chromosome paper, I notice the presence of R-M269 in the Lezgins of Dagestan (~29.6%), as well as the Abkhaz (12.1%). Looking at my Dodecad v3 results, I obtain a value of 24.3% of the "West European" component in the Lezgins, and a value of 15.7% in the Adygei, who are linguistically related to the Abkhaz. By contrast, other R1b-rich populations (namely Armenians and Turks) from West Asia show no substantial evidence of the "West European" component.


PLoS ONE 6(6): e21592. doi:10.1371/journal.pone.0021592

Wave-of-Advance Models of the Diffusion of the Y Chromosome Haplogroup R1b1b2 in Europe

Per Sjödin1, Olivier François

Abstract
Whether or not the spread of agriculture in Europe was accompanied by movements of people is a long-standing question in archeology and anthropology, which has been frequently addressed with the help of population genetic data. Estimates on dates of expansion and geographic origins obtained from genetic data are however sensitive to the calibration of mutation rates and to the mathematical models used to perform inference. For instance, recent data on the Y chromosome haplogroup R1b1b2 (M269) have either suggested a Neolithic origin for European paternal lineages or a more ancient Paleolithic origin depending on the calibration of Y-STR mutation rates. Here we examine the date of expansion and the geographic origin of hgR1b1b2 considering two current estimates of mutation rates in a total of fourteen realistic wave-of-advance models. We report that a range expansion dating to the Paleolithic is unlikely to explain the observed geographical distribution of microsatellite diversity, and that whether the data is informative with respect to the spread of agriculture in Europe depends on the mutation rate assumption in a critical way.

Link

15 comments:

Average Joe said...

Dienekes:

How and when do you think that R1b1b2 (M269) came to Europe?

princenuadha said...

The question I have, along with the origin of expansion, is by what route R1b1b2(a) took to western Europe.

(Unfortunately this modeling doesn't seem to have looked at northeast anatolia, the caucus, or the steps.)

I suspect that r1b1b2(a) came to western Europe by a route that went north of the black see, as opposed to going directly from anatolia to modern Greece. Also I think spread of m269 in western Europe was correlated to the spread of indo European in western Europe, and that when indo European travelled to Europe it took a fought north of the black see also.

The reason I think indo European came to most of Europe by a route north of the black see is because this chart. http://3.bp.blogspot.com/-UK6uv6RZBlw/TZr4U1kHkKI/AAAAAAAADbY/jvqITDejMG0/s1600/nature.jpg
That says that the non Greek indo European languages in Europe are more related to Indian than to Greek which makes it hard to imagine those non Greek IE languages in Europe not entering Europe north of the black sea.

Does anybody have evidence that m269 came to western Europe north of the black sea?

Unknown said...

Hmmmm Interesting paper. I wish they had shown all the data though.

They are using diversity in 9 STRs not SNPs. So Haplotypes not Haplogroups. Not what I expected from the title.

(1) Figure 3 (D) shows the diversity in the STRs across Europe which I can't recall previously seeing. Basically there is lower diversity in Western Europe (yellow) than Eastern Europe (red). So Western Europe has a lot more R1b (as we know from other studies) but less diversity.

(2) The authors think the data fits better in models using the Evolutionary rate (older ages) over the germline rate, so I am going to work with that.

(3) Figure 1 shows that the current diversity matches EQUALLY Evolutionary rate models of:

Neolithic, Anatolia origin
Paleolithic, Anatolia Origin
Paleolithic, Italy origin
Paleolithic, Iberia origin
Recent, Anatolia origin

So no conclusion there.

(4) The geographic data (Figure 3, A, B and C) only shows expansion from Anatolia in the neolithic or more recently, and the neolithic expansion from Anatolia is closer than a more recent expansion from Anatolia. But the other data is annoyingly missing.

So not much to be gleamed from Figure 3, certainly nothing on the source of the expansion, or the paleolithic/neolithic problem.

(4) The bulk of the data is summarized in Figure 2 which is a bugger to get your head around.

I am assuming that the area under each curve is a measure of how bad each model is. Ideally the line should run along the bottom. Presumable this is the averaged value in Table 2, although I must say the areas look pretty similar to me by eye.

Figure 2 is difficult as it is not clear what "density" is. I even checked the references (31,32). If it is population density (the only density mentioned in the text) then these graphs show:

The high density populations have lower deviations from actual in the recent/neolithic models than the paleolithic models. The models have similar errors for low density populations.

For the paleolithic models the out-of-Anatolia model had lower errors at peak population than Spain or Italy, but these should have the same peak heights (as they should be identical models) and they dont. So something is wrong.

I am troubled by their "rejection sampling".

I find myself losing faith in this paper.

(5)Table 2 clearly states that the Neolithic model is the best fit, but the standard deviations are huge and again, irritatingly most of the data is missing.

STRs are normally used for shorter term studies than this as they mutate rapidly. I think perhaps 9 STRs was insufficient for this kind of study. The only parts that made sense were inconclusive.

And this statement in the abstract "We report that a range expansion dating to the Paleolithic is unlikely to explain the observed geographical distribution of microsatellite diversity" seems to have nothing to do with the contents of the paper at all. No paleolithic geographical distribution data was presented in the paper. The obvious place to do so was Figure 3.

Davidski said...

R1a1 was found at Eulau you donkey.

Also, this comment doesn't make any sense considering the Fst distances between these components.

"It appears that the "West European" autosomal component shows a stronger relationship to the "West Asian" one than the "East European" one."

West Euro is closer to West Asian, than East Euro is.

But West Euro and East Euro are both closer to each other than either is to West Asian.

Charles Nydorf said...

Could the "northwest European component" among Ashkenazic Jews be largely the Daghestan component?

Charles Nydorf said...

Perhaps the fact that Ashkenazic Jews register in ADMIXTURE as higher in the West European than East European components actually reflects the Daghestan rather than West European component.

Dienekes said...

R1a1 was found at Eulau you donkey.

You don't have to be an ass (pun intended). I meant Lichtenstein which is 3ky, and the point of the post holds (namely that the expansion of R-M269 into Europe cannot have started when it was already deep into Europe).


But West Euro and East Euro are both closer to each other than either is to West Asian.


That is irrelevant for the issue at hand. Clearly neither West Euro or East Euro is of deep antiquity, as all major West Eurasian components have similar Fst divergences to each other and there is no room for any of them to have diverged from each other in a Paleolithic time frame while others did so in a Bronze Age time frame.

What the relative position of W/E European components compared to W Asian shows is that "something" from West Asia affected Western Europe more than it did Eastern Europe, and R1b is quite a good candidate for what that something might be.


Could the "northwest European component" among Ashkenazic Jews be largely the Daghestan component?


A high W/E European ratio is found in many populations, so it's not clear if the W European among Ashkenazi Jews was all acquired in Europe. One would have to compare A Jews to other Jews (who had no opportunity of mixing with Western Europeans) as well as Near Eastern populations.

I suspect that r1b1b2(a) came to western Europe by a route that went north of the black see, as opposed to going directly from anatolia to modern Greece. Also I think spread of m269 in western Europe was correlated to the spread of indo European in western Europe, and that when indo European travelled to Europe it took a fought north of the black see also.

That is doubtful, as R-M269 is more diverse in southeastern Europe than it is in the northern Caucasus or eastern Europe. As far as Y-STR diversity goes, the pattern seems to point to Anatolia (and beyond, not sampled here) as the origin of R-M269. I don't exclude that some of it may have spread via the Caucasus to eastern Europe, but it's fairly hard to imagine a purely steppe solution.

eurologist said...

Apart from the weird dates used in the study (e.g., their "neolithic" is closer to the post-Younger Dryas expansion in Europe then the agricultural one), the authors make the same mistake as others by treating together the known, two largely distinct subgroups of M269 (mostly European versus mostly Anatolian) and calculating a "diversity" from that. This is analogous to treating R1a and R1b (or I and J) the same and calculating a diversity from that. You simply get such a mix of time frames and expansion waves that the outcome is completely useless.

Also, why not use models that capture the known expansions and low population densities of the past (pre-LGM, post LGM, post younger Dryas, agricultural)? If the last two of these involved much higher population densities, they could dominate today's signatures even if M269 is paleolithic in Europe.

n/a said...

"R-M269 is more diverse in southeastern Europe than it is in the northern Caucasus or eastern Europe."

Incorrect. Following Belaresque, these authors failed to include any Eastern European samples in the relevant analyses.

Some recent amateur calculations for R-L23*:

Western Europe - British Isles, France, Spain, Germany, Netherlands
Northeastern Europe - Czech Republic, Poland, Ukraine, Lithuania
Southeastern Europe - Italy, Hungary, Croatia, Greece, Bulgaria

Intraclade age estimates (30 yrs./G):
Western Europe - G=124+/-11 or 3720+/-330 ybp, n=15
NE Europe - G=124+/-11 or 3720+/-390 ybp, n=9
SE Europe - G=118+/-15 or 3540+/-450 ybp, n=8
All Europe - 133/13 or 3990+/-390 ybp, n=32
Turkey - 135/13 or 4050+/-390 ybp, n=22


And:

Intraclade method:
Caucasus, G=164 or 4920 ybp
All L23*, G=161 or 4830 ybp
All non-Europe, G=154 or 4620 ybp
Europe, G=139 or 4170 ybp
Turkey, G=134 or 4020 ybp

The larger interclade sigmas extend the range from 6130 BC to 1390 BC. When considering the ancestry of these members, mostly non-European, this doesn't look as much like an early neolithic distribution into SE Europe as it does a 4th millenium BC expansion out of the Caucasus. The interclade TMRCA's center quite nicely around the time of the Maikop and Kura-Araxes cultures of the Caucasus.


Correct answer is north of the Black Sea.

Jim said...

"The reason I think indo European came to most of Europe by a route north of the black see is because this chart. http://3.bp.blogspot.com/-UK6uv6RZBlw/TZr4U1kHkKI/AAAAAAAADbY/jvqITDejMG0/s1600/nature.jpg
That says that the non Greek indo European languages in Europe are more related to Indian than to Greek which makes it hard to imagine those non Greek IE languages in Europe not entering Europe north of the black sea. "

First off, that chart is not authoritative in its branchings. It conflicts in several places with Ethnologue's chart for Indo-European. And no one died and left Ethnologue in charge either, and it oftne diagerees with itself, but that's the point really.

Second, the tree model as a mdel for the structure of IE has been under a cloud for decades. It just hides a lot of important historical information relevant to your particular point. For instance it is inadequate for showing the realtionships between the clooge of lngugaes that used to be in the balkans - proto-Greek, Illyrian, Thraco-Phriugian,, even that lash-up of Italo-Celto-Germanic in your chart. A tree model is inadequate for dealing with Sprachbunds, and as it happens there is a modern Balkan Sprachbund involving Greek, Romanian and some Slavic languages, none of which are closely related.

So no, that chart isn't enough support for saying that what you are trying to say. Not that you aren't still right, at least partially. IE entry through Anatolia is a long way from a done deal.

princenuadha said...

"I obtain a value of 24.3% of the "West European" component in the Lezgins, and a value of 15.7% in the Adygei, who are linguistically related to the Abkhaz. By contrast, other R1b-rich populations (namely Armenians and Turks) from West Asia show no substantial evidence of the "West European" component."

Then n/a saying that m269 came to western European primarily by a route north of the black sea, and from the caucasus no less, hugely bolsters your connection between m269, the caucasus, and "West European".

Now you can explain why western Europeans have m269 and why the Turks don't have WE!!!

As a side not I too another look at the chart and I think it would be better to call "West European", "Northwest European" instead. WE peaks in Ireland and there is actually more WE in Scandinavia, Finland, and Germany than in Iberia.

The one surprise to how I was previously talking about WE is why the Fins would have so much. Maybe WE is some mix of Caucasian and "other" and maybe the fins have a lot of this "other"...

Dienekes, are the D finish from western Finland, the area colonized by Swedes?

Dienekes said...

Now you can explain why western Europeans have m269 and why the Turks don't have WE!!!

There is no "global" relationship between M269 and "WE". Armenians have practically no "WE" and they have 27% R1b1 which is also quite diverse in them it seems. Given the 29% in Lezgins, higher value in N Iran than in S Iran, and about 15% in Anatolian Turks, I would say the origin of R-M269 is to be sought in the Transcaucasus. I strongly suspect R-M269 was also formerly more frequent in the Balkans (see Bosch Aromun paper), and it is also quite diverse in southeastern Europe (again see Y-STRs in Bosch paper).

I don't exclude the possibility that some R-M269 may have gone from the Transcaucasus via Kura-Araxes and Maikop north of the Black Sea and into Europe, but in my opinion most of the Greek R-M269 and some of the Italian one came from Anatolia which it closely resembles, and most of the European R-M269 expanded from Central Europe; whether that ultimately followed a northern or southern route to Central Europe remains to be seen.

Andrew Oh-Willeke said...

Is there anything other than genes and recent historical records to provide any sense of the sequence in which the respective linguistic peoples of the Transcaucasus arrive there?

As I understand it, even very basic questions, like whether speakers of the proto-Armenian IE dialect are in the Transcaucasus pre-Hittite (i.e. prior to ca. 1800 BCE) or only post-Bronze Age collapse (i.e. after ca. 1200 BCE), are controversial

Answer to questions like the origins of the Kartevlians v. NE Caucasians v. NW Caucasians v. Armenians v. anybody else in the Transcaucasus seem utterly obscure which is problematic when so many genetic origin arrows seem to point to that general region, so it seem like more evidence is needed than just genetics and language to flush out what happened in pre-history. How are these ethnicities culturally distinct apart from their adoptions of Islam and Orthodox Christianity in historical times and their languages? Do they differ in food production approaches? In matrilineal v. patriarchial organization? In the technology levels indicated by proto-language reconstructions? In artistic and architectural styles? In mythology and origin myths? In preferred climate niches?

Geographically close populations in the region seem to be surprisingly genetically distinct, which indicates either something enforcing strong endogamy pressures beyond language (e.g. Hasua and Fulani are starting to experience ethnic merger despite their very different languages as a result of colocation, similar lifestyle and shared religion), or recent arrivals that haven't had time to mix (which doesn't fit the stereotype of what we think we know of the region as a relict area). Is there anything besides transportation barriers that keeps these populations so ethnically fractured?

The fact that the Transcaucus populations retained their ethnicity and language in the face of what appear to have been pretty high frequency and regular IE contacts while so many other cultures was assimilated is also notable.

It also isn't at all clear to me how much we know about which subsets of Transcaucasus populations are associated with the multiple Transcaucasus firsts in metallurgy. Was copper/bronze/steel metallurgy areal or ethnic? I've seen some association of this with Hurrian/NW Caucuasian populations, but the extent that this overlaps into Kartevelian and NE Caucasian and Armenian proto-societies is not at all clear to me.

princenuadha said...

'There is no "global" relationship between M269 and "WE"'

That was never the point. The point was that the Caucasians, who have W.A., W.E., and Dagestan, brought m269 to most of Europe including western Europe.

Anyways I was only trying interpreting your suggested connection between W.A., m269, and Caucasian in light of m269 evidence given by n/a.

"I don't exclude the possibility that some R-M269 may have gone from the Transcaucasus via Kura-Araxes and Maikop north of the Black Sea and into Europe"

Well if n/a's data is correct he basically proved it.

As to whether or not m269 came to western Europe mostly north or south of the black sea, n/a's data suggests north too.

Dienekes said...

Thankfully, a German group has announced it is analyzing Bronze Age samples from eastern Europe and the north Caucasus, so we will know soon enough if there was any R-M269 going that way in the period in question.

As I said above, it's not an "either-or" thing. R-M269 occurs in Anatolia, and I don't see why it would have gone from the Transcaucasus to the eastern European plain, down the Balkans and into Anatolia, making roughly a full circle. If, on the other hand, we accept that it entered Anatolia directly from the east, then its occurrence in Greece and Italy can be explained economically from there, which would also explain the J2/R1b correlation in southeastern Europe.

In short, I am all but convinced by the absence (so far) in ancient Neolithic remains of R-M269 and its apparent young Y-STR age that it is a late Neolithic or Bronze Age phenomenon. Moreover, the most economical explanation for its origin (taking into account both R-V88 of Africa and R-M73 of Central Asia, as well as R1a and the various paragroups) is that it spread from Asia to Europe and Africa. By what exact route(s) and from which exact source is a matter that only ancient DNA will answer convincingly IMO.