October 19, 2010

Origin of Neolithic N1a

My comments on this paper will be posted in this space; needless to say, I have serious objections to the idea that N1a is a Mesolithic European mtDNA.

UPDATE (Oct 20):

The paper makes a significant contribution to the phylogeny of N1a by full sequencing of mtDNAs belonging to individuals from different populations, ascribing the Neolithic sequences to several identified clades. That part of the paper is solid and interesting.

My main problems with the paper are threefold:

1. The paper attempts to infer the origin of mtDNA founders using TMRCA estimates. But, as I have pointed out, TMRCAs of clades tell us nothing about where the common ancestors lived. The TMRCA of Latin American R1b, for example, predates the arrival of Western Europeans into the Americas by thousands of years.

2. The paper attempts to infer the origin of mtDNA founders using modern populations. Given the clear evidence for discontinuity between the Mesolithic and Neolithic and present in Central Europe, due to either demography or selection, I find it a very questionable proposition. If N1a turns up in dated human remains associated with a Mesolithic culture of Europe prior to the arrival of the Neolithic economy, then the hypothesis that it is a forager lineage is unsubstantiated.

3. Finally, the paper uses the frequency of identified subclades to infer the location of the founders. In addition to the above 2 criticisms, I must point out that this puts the cart before the horse. In a uniform landscape, we do expect present-day frequency to be related to the place of origin of a mutation. But, we are not dealing with such a landscape. In particular, we are dealing with an expanding population exploiting new territory. As an analogy, the few people who made the crossing into the Americas left a few relatives in the Old World, and produced a plethora of new ones (descendants) in the New World. They did so by exploiting their new environment.

An additional objection, is, of course, that the idea of Mesolithic N1a in Europe requires a virtual partition of pre-Neolithic European populations, to account for its non-existence among north/central Mesolithic North/Central Europeans. That is hard to swallow, unless by "Mesolithic" one means "very shortly before the Neolithic", which would make it possible for a lineage to establish itself in e.g., the Balkans shortly prior to the onset of the Neolithic, and begin expanding shortly thereafter. However, I don't see how an argument could be made for such a scenario, and, of course, I doubt that genetic dating methods in modern populations have enough precision to allow one to distinguish between late Mesolithic and early Neolithic intrusions.


BMC Evolutionary Biology 2010, 10:304doi:10.1186/1471-2148-10-304

Mitochondrial haplogroup N1a phylogeography, with implication to the origin of European farmers

Malliya GOUNDER Palanichamy et al.

Background
Tracing the genetic origin of central European farmer N1a lineages can provide a unique opportunity to assess the patterns of the farming technology spread into central Europe in the human prehistory. Here, we have chosen twelve N1a samples from modern populations which are most similar with the farmer N1a types and performed the complete mitochondrial DNA genome sequencing analysis. To assess the genetic and phylogeographic relationship, we performed a detailed survey of modern published N1a types from Eurasian and African populations.

Results
The geographic origin and expansion of farmer lineages related N1a subclades have been deduced from combined analysis of 19 complete sequences with 166 N1a haplotypes. The phylogeographic analysis revealed that the central European farmer lineages have originated from different sources: from eastern Europe, local central Europe, and from the Near East via southern Europe.

Conclusions
The results obtained emphasize that the arrival of central European farmer lineages did not occur via a single demic diffusion event from the Near East at the onset of the Neolithic spread of agriculture into Europe. Indeed these results indicate that the Neolithic transition process was more complex in central Europe and possibly the farmer N1a lineages were a result of a 'leapfrog' colonization process.

Link

18 comments:

Gioiello said...

A few years ago I found on SMGF and put on Mitosearch an Italian U2d, thought typical from South Asia. Palanichamy asked me for a FGS, but of course I wasn't able to give a sample.

Then I think that many mtDNA (and Y), thought not European (above all Italian), are here from more thousands of years than many are thinking.

Anonymous said...

What matters is not where N1a originated but the frequencies in reasonable samples of Mesolithic, early Neolithic farmers not just from Central Europe but from Portugal or Cyprus or Italy since in the Mediterranean zone the Neolithics did colonize in a saltatory manner and used watercraft to move around complete with all their plants and animals.

Correct me if I am wrong but N1a is essentially missing in Mesolithic Europeans, found in reasonable frequencies in Neolithic farmers from Central Europe but a rare haplogroup in modern Europeans. There appears to be a discontinuity between the three Europeans as far as this haplogroup is concerned.

eurologist said...

There appears to be a discontinuity between the three Europeans as far as this haplogroup is concerned.

I think that is just the nature of the neolithic in most of Europe: outside of the (southeast) Balkans, it largely involved local groups suddenly expanding. Danubian is just one such bubble of a limited set of haplogroups that happened to live in the area. After several such bubbles, and after surrounding regions could gain a foothold again, things could come to more of an equilibrium. But that probably took 2,000-3,000 years, or so.

And I don't find the ages provided dubious, at all. If anything, they are a bit on the young side, given that there were migrations out of Europe but none documented into Europe between LGM and the advent of agriculture.

Marnie said...

Dienekes, thanks for posting this. Unfortunately, I won't be able to read this open source paper for another week or so, so I won't comment explicitly on it now.

However, I will say that I've recently stumbled across several references that suggest that human populations may have sought out dynamic ice-marginal zones for their abundance of big game. See "Ice Age Peoples of North America: An Introduction to the Peopling of the Americas."

Also, the Curdy paper:

http://www.mtsn.tn.it/pubblicazioni/7/42/08_Curdy.pdf

I'm increasingly of the opinion that some European and Caucasian populations gravitated to these Ice margin climates, rather than fully retreating southward.

Andrew Oh-Willeke said...

I agree that the conclusions reached from the data aren't very convincing. The five ancient mtDNA greater LBK farmer haplotype breakdown was as follows:

DEB3 and FLO1 are N1a1a1 and this haplotype "reaches a maximum in eastern Europe and central Asia, and decreases in the direction of central and northern Europe. The same occurs in the southern direction, towards Anatolia and the Caucasus...one of the
N1a1a1 subclade, N1a1a1a...is restricted to Kazakhstan, Altai and Buryat Republic, and European part of Russian Federation and it shows significant heterogeneity in the haplotype distribution, attesting that these areas were a center of expansion."

HAL2 is N1a1a2 is "observed in Denmark, Poland, Scotland, Norway, Switzerland, France, Portugal, Hungary, Austria, and Volga-Ural region."

USW4 is N1a1a3 "which is widespread
in Italy, Yemen, Arabian Peninsula, Austria, Germany, Slovakia, Sweden, and Norway."

DEB1 and ECS1 are N1a1b "which is found in the Arabian Peninsula, Armenia, and Italy...[and there is] both a high frequency and a high degree of diversity of this lineage in the Arabian Peninsula."

Note that the "Near Eastern" N1a1b and the "European" N1a1a1a are both found at the same DEB location in the time LBK time period, and that half of the N1a types found in LBK farmers are found in the Near East today. Yet, it is unlikely that the N1a in the Near East today got there via back migration (the Crusades and the Roman Empire could provide back migration, but that seems like a less likely hypothesis).

In short, the diversity of the N1a types among LBK farmers, which include types found in the Near East and South Asia today, argues for a mix of N1a types in the source LBK gene pool that experienced serial founder effects as subgroups of the whole expanded in different directions from the Near East. (Similarly, many regions where N1a is found have multiple subhaplotypes of N1a today.)

This is further supported by the overall diversity of haplotypes found in LBK farmers generally of which N1a made up only a quarter of the whole. The LBK farmers were not overwhelmingly of a single, or even a small number of matrilines, and the fact that we have only 24 samples, suggests that there were probably other low frequency matrilines that were not observed in the sample in the source LBK population.

Of course, if the original LBK population was matriline diverse, then the haplotypes observed probably all arose (with the possible exception of final mutations like the break of N1a1a1a from N1a1a1) in the pre-LBK Neolithic area, rather than in places like Central Europe and Central Asia.

Surely, whatever caused N1a to go from 25% to a fraction of one percent of the population could also have made subtypes of N1a rare enough that they don't appear in the Near Eastern part of the sample. A 99% diminishment is going to locally eliminate some subtypes.

aargiedude said...

The study overlooked something crucial that they themselves uncovered. We can now see that Haak's 7500 year old samples from Europe already belonged to 4 different lineages of N1a that were separated from each other by several coding region mutations:

N1a (Palanichamy, 2010).pdf
[page 29, note that the 6 Haak samples from 7500 years ago are listed at the top of the chart, such as DEB3, FLO1, etc.]

This shoots down the possibility that N1a's bizarrely massive presence in those 7500 year old central European samples was due to a founder effect. Coding mutations occur about once every 5000 years.

This also means this "package" of N1a lineages had to already exist in the original population back in Anatolia/Levant.

The predominantly European N1a1a lineages couldn't have arisen as a founder effect amongst the N1a* being carried into Europe by the farmers from Anatolia because even back then N1a1a had branches that were separated from each other by several coding mutations, indicating a huge time depth, possibly as much as 10,000 years prior to that time. So these N1a1a lineages had to already exist in the original population back in Anatolia/Levant. And they should have made up a big part of Anatolian N1a, just as they did in their descendants in central Europe 7500 years ago.

Yet today, of 6 N1a samples from Anatolia/Levant, none of them belongs to any of the 4 lineages found in Haak's study, they belong instead to N1a* and the Central Asian branch N1a1a1a, while in Europe, the modern compositon of N1a is notably similar to the composition of Haak's N1a from 7500 years ago.

aargiedude said...

There appears to be a discontinuity between the three Europeans as far as this haplogroup is concerned.
.....................
I think that is just the nature of the neolithic in most of Europe: outside of the (southeast) Balkans, it largely involved local groups suddenly expanding.


So you're saying N1a was a founder effect? That's what my post is about. It couldn't have been that. Haak found 4 lineages of N1a, which this new study has determined were already deeply split from each other even back then.




there were migrations out of Europe but none documented into Europe between LGM and the advent of agriculture

How exactly do we know this?!




Note that the "Near Eastern" N1a1b and the "European" N1a1a1a are both found at the same DEB location in the time LBK time period, and that half of the N1a types found in LBK farmers are found in the Near East today.

No, you got things mixed up. Haak found 2 samples of N1a1a1* (see page 29 of the pdf), which today is typical of Europe. You confused that with N1a1a1a, which is instead the predominant lineage of Central Asian. It's this Central Asian lineage that is found today in half of Near East N1a, that's why you noted incorrectly that half of Near East N1a was found in the LBK farmers. In fact, none of the LBK N1a share the same lineage to the current Near East N1a, a total disconnect between the two. On the other hand, modern European N1a is notably similar to LBK farmer N1a.



a mix of N1a types in the source LBK gene pool that experienced serial founder effects

Are you saying that N1a* went into Europe and through several serial founder effects produced N1a1 and then N1a1a and its several branches? If so, that's what this post is about. The authors of the study missed a crucial thing that they themselves uncovered: that there are several coding mutations between the 4 lineages of the ancient Haak samples, meaning that the Haak samples couldn't be the product of a founder effect in an expanding population, they had to be already formed and well established back in the "home base" of Anatolia and/or Levant, where today they are completely missing.




Surely, whatever caused N1a to go from 25% to a fraction of one percent of the population could also have made subtypes of N1a rare enough that they don't appear in the Near Eastern part of the sample. A 99% diminishment is going to locally eliminate some subtypes.

You're taking this for granted, when it is in fact something extremely, exceptionally unlikely. It's the only way the pieces can fit, barring a fundamental flaw in ancient mtdna analysis, which is looking more and more likely given the recent ancient mtdna study of Denmark.

aargiedude said...
This comment has been removed by the author.
aargiedude said...
This comment has been removed by the author.
Dienekes said...

aargiedude, you quadruple-posted. In the future, combine your comments into 1-2.

eurologist said...

The study overlooked something...

...This shoots down the possibility that N1a's bizarrely massive presence in those 7500 year old central European samples was due to a founder effect.


I have a very hard time understanding your argument, but will try to look into it the next several days.

aargiedude said...

eurologist, the new study did a full genome sequence of N1a and discovered that it has quite a few branches, each with several coding region mutations separating it from all other branches. Coding mutations appear on average every 5,000 years, so these are long term separations between these branches. Yet the Haak study shows that the central European N1a from 7500 years ago could already be assigned to 4 of these branches on the basis of their HVR haplotypes, as this study has done by noting at the top of the chart of page 29 of their pdf to which branch each of the 6 Haak N1a samples belongs. This means that even back then there was still a huge time depth to the separation between the N1a lineages that made up central European N1a, possibly as deep as 10,000 years prior to that time. One obvious explanation for the strange massive presence of N1a in central Europe back then is that it was the result of a founder effect, something that could have happened amongst the local population or even amongst the expanding farmer population. Well, this new finding says that's not possible. The N1a samples belong to 4 separate lineages that even back then were separated from each other by a huge timespan.

Dienekes said...

One obvious explanation for the strange massive presence of N1a in central Europe back then is that it was the result of a founder effect, something that could have happened amongst the local population or even amongst the expanding farmer population. Well, this new finding says that's not possible. The N1a samples belong to 4 separate lineages that even back then were separated from each other by a huge timespan.

Your inference is baseless, as it is based on the strawman argument against the idea that only a single N1a founder from West Asia spawned N1a Europeans.

The existence of multiple N1a types separated from each other by thousands of years does not disprove in any way that these types represent recently arrived founder lineages. Take a few dozen modern-day Anatolians and transplant them to Central Europe and you'll get plenty of mtDNA types separated from each other by thousands of years.

Andrew Oh-Willeke said...

"Are you saying that N1a* went into Europe and through several serial founder effects produced N1a1 and then N1a1a and its several branches?"

No. I'm saying that many subhaplotypes of N1a were present in the founding LBK population and that not every subhaplotype group was present in every branch of the LBK expansion. Folks took a left at the Danube had some lineages, folks who followed the Danube to its source had others, folks who took a right at the Danube had others. This in turn influenced the haplotype mix in subsequent generations in that region. Founder effects and not mutations in Europe probably produced the effect.

The only haplotype that looks likely to have developed in Europe is the split from N1a1a1 to N1a1a1a.

none of the LBK N1a share the same lineage to the current Near East N1a

According to the paper, USW4 is N1a1a3 "which is widespread
in . . . Yemen, [and the] Arabian Peninsula," and DEB1 and ECS1 are N1a1b "which is found in the Arabian Peninsula." So three of the six samples were either N1a1a2 or N1a1b which are currently found in the Near East. Three divided by six is half. What am I missing?

Surely, whatever caused N1a to go from 25% to a fraction of one percent of the population could also have made subtypes of N1a rare enough that they don't appear in the Near Eastern part of the sample. A 99% diminishment is going to locally eliminate some subtypes.

You're taking this for granted, when it is in fact something extremely, exceptionally unlikely.


Taking what for granted? That N1a once made up 25% of ancient LBK DNA and now is rare? I deliberate refrain from resolving the difficult issue of the mechanism by which this happened, but the fact that it did seems to be pretty much beyond dispute.

aargiedude said...

Your inference is baseless, as it is based on the strawman argument against the idea that only a single N1a founder from West Asia spawned N1a Europeans.

You're not getting the implications. Excluding N1a, the 7500 year old mtdna samples from Haak are similar to modern European mtdna. They don't look like they suffered a lot of founder effects. So how likely is it that 4 founder effects occured in N1a samples? Well, that depends. If the frequency of N1a in the original population was 0.3%, the odds are virtually zero. Even if N1a was 1% or 2% of the mtdna, it still looks extremely unlikely. Ergo, N1a had to be a significant haplogroup already, even before the first founder effect took place. Thus, Anatolia had to have a high frequency of N1a, and furthermore it had to consist to a large degree of the types found in their brothers that had just entered into Europe: N1a1b, N1a1a3, N1a1a2, and N1a1a1*. Today we find that modern European N1a is in fact remarkably similar to the 7500 year old central European N1a: more than 90% of them belong to the 4 lineages identified by Haak from 7500 years ago. But of the 6 modern Anatolian and Levant N1a samples detected so far, there isn't even a single match with the 7500 year old central European N1a. 3 are N1a* and 3 are N1a1a1a. The latter is from Mongolia, makes up a small minority of European N1a, is found only in the eastern parts of Europe, and in any case, with its very terminal position in the N1a phylotree, being in fact a daughter clade of one of the 7500 year old Haak lineages, it's a sideshow of no significance in the plot.

So we have a problem. The evidence continues to mount that we can't disregard N1a in ancient mtdna studies. Just recently there was a study of 3 very ancient samples from France, of a similar date as the Haak samples. 1 of them was N1a, with the haplotype strongly suggesting N1a1b. And now this new study is telling us it had to be already a significant and diverse part of the parent population itself in Anatolia and/or the Levant. N1a continues to snowball in importance in the Neolithic, apparently being the 1st or 2nd most common haplogroup in Europe and the Middle East, only to then, somehow, very quickly disappear by around 5,000 years ago. And the modern European and Anatolian N1a belong to different planets and don't reflect at all a possible ancient connection via the farmers bringing into Europe the N1a from Anatolia.

aargiedude said...

which is found in the Arabian Peninsula." So three of the six samples were either N1a1a2 or N1a1b which are currently found in the Near East. Three divided by six is half. What am I missing?

The farmers didn't come from Saudi Arabia. [Also, the Near East is Anatolia, then comes the Middle East, the Arabian Peninsula isn't normally considered to be even part of the Middle East]

Taking what for granted? That N1a once made up 25% of ancient LBK DNA and now is rare? I deliberate refrain from resolving the difficult issue of the mechanism by which this happened, but the fact that it did seems to be pretty much beyond dispute.

It's so incredibly unlikely that this actually happened that it raises the possibility that there is a fundamental problem with ancient mtdna analysis and the results are incorrect.

Dienekes said...

aargiedude, your argument is based on the assumption that N1a was extremely rare in the Anatolian Neolithic population that spawned the European Neolithic.

You are forgetting though, that the European Neolithic was not launched by a random sample of the entire current Anatolian population, but by a sample of a limited number of Neolithic communities that had the idea to cross the Aegean and/or the Bosporus.

The offspring of these communities that went to Europe would've fared much better than those that stayed behind, as they had a new underexploited territory to farm, while Anatolia was already thousands of years old in terms of its farming economy. As a result genetic types that were rare in Anatolia became excessively abundant in Europe. The original types might even become extinct in the homeland (as drift weeds out lineages more effectively in static than in expanding populations), but as Andrew points out the so-called European types are still found in the Near East, albeit at lower frequencies, which is exactly what we expect.

These Arabian matches of Neolithic farmers are not recent back-migration, because if it was recent back-migration they would be dominated by European H-types and the like, as N1a is extremely rare in Europe in the last few centuries.

Andrew Oh-Willeke said...

The farmers didn't come from Saudi Arabia.

We know as a matter of near scientific certainty from the ancient DNA that about 1/8th of the LBK farmers shared common ancestors with the mtDNA N1a haplotype people who live in Saudia Arabia today.

We know that there have been multiple waves of migration in West Asia over the last 8,000 years; almost all of the migration waves of the last 5,500 years in the region are reflected in historical records which are more ancient and complete no where else in the world. It also isn't unreasonable to consider the possibility that there may also have been one or more waves of migration in the 2,500 years from the earliest LBK sites to the earliest Sumerian records of West Asia for which a historical record is absent.

We don't know why N1a is so much more rare in West Eurasia than it was among the LBK. Natural selection against it, and subsequent migration of a people who lacked N1a at meaningful fequencies into their territory (and many surrounding areas) either replacing them or outnumbering them in the gene pool are really the only mechanisms that make sense. But, we don't have any one event that we can pin that on in the last 8,000 years.

Dienekes is right in suggesting that recent back-migration doesn't look like a very plausible possibility. Since there are no known selective effects associated with N1a strong enough to account for such a precipitous decline in its frequency, some sort of migration from a low N1a population with a haplotype makeup similar to that found today seems most plausible.

Since such a dramatic change in population makeup ought to leave some sign, and since recent history is pretty well ruled out, we probably ought to be looking for a cause in the Bronze Age or Iron Age, prior to which there was relative continuity from LBK cultures.

Alternately, one could argue for gradual follow on migration from Europe pre-Bronze Age from places that lacked N1a, contra the original wave of migration.

[Also, the Near East is Anatolia, then comes the Middle East, the Arabian Peninsula isn't normally considered to be even part of the Middle East]

In modern English, the terms "Near East" and "Middle East" are bascially synonomous, and both (are certainly the term "Middle East") would include the Arabian Peninsula. As I understand it, the coinage of the terms is from British colonial and military usage.

Until World War II, it was customary to refer to areas centered around Turkey and the eastern shore of the Mediterranean as the "Near East", while the "Far East" centered on China, and the Middle East then meant the area from Mesopotamia to Burma, namely the area between the Near East and the Far East. In the late 1930s, the British established the Middle East Command, which was based in Cairo, for its military forces in the region. After that time, the term "Middle East" gained broader usage in Europe and the United States, with the Middle East Institute founded in Washington, D.C. in 1946, among other usage.

The term "Middle East" no longer includes South Asia in common usage.

Near East and Middle East overlap heavily with "West Asia" which is now popular as a less stereotype loaded term in the academic literature.