October 21, 2012

Post-LGM expansion of mtDNA?

This issue keeps appearing and re-appearing. It is perhaps due to a tendency of conflating spatial population expansions with the proliferation of descendants within a lineage. The two are not necessarily related. Genetic-only methods can pick up on the signal of common descent and population growth, but cannot do the same for the signal of spatial expansion. Whether this growth happens (i) in situ for a long time, and only lately becomes a spatial expansion, or (ii) at the same time as the spatial expansion, or indeed (iii) long after it, will result in coalescences that precede, coincide with, or follow the actual spatial expansion event.

It is difficult to see how Europe was being filled up for thousands of years by a population taking advantage of post-glacial warming conditions, and, yet, when we actually look at ancient DNA from   Europeans who lived just before the advent of farming, they show little evidence of possessing (m)any of the lineages that had been supposedly expanding in Europe since the LGM.

SCIENTIFIC REPORTS doi:10.1038/srep00745

MtDNA analysis of global populations support that major population expansions began before Neolithic Time

Hong-Xiang Zheng et al.

Agriculture resulted in extensive population growths and human activities. However, whether major human expansions started after Neolithic Time still remained controversial. With the benefit of 1000 Genome Project, we were able to analyze a total of 910 samples from 11 populations in Africa, Europe and Americas. From these random samples, we identified the expansion lineages and reconstructed the historical demographic variations. In all the three continents, we found that most major lineage expansions (11 out of 15 star lineages in Africa, all autochthonous lineages in Europe and America) coalesced before the first appearance of agriculture. Furthermore, major population expansions were estimated after Last Glacial Maximum but before Neolithic Time, also corresponding to the result of major lineage expansions. Considering results in current and previous study, global mtDNA evidence showed that rising temperature after Last Glacial Maximum offered amiable environments and might be the most important factor for prehistorical human expansions.

Link

12 comments:

Charles Nydorf said...

Patrilocal marriage patterns may make women, and their mtdna, relatively more mobile.

Annie Mouse said...

What exactly is your point? So this method cannot distinguish between spacial and local population expansions. This still does not take away from the main point. Which is that the population expansion did not start with the archaelogical neolithic, in Europe Asia OR Africa.

Studies like these provide us with evidence of ancient population expansions even if only a few of the progeny survive.

A massive neolithic spacial replacement expansion of population into Europe should show as the dominant expansion in European mitochondrial trees. This paper says this did not happen in Europe, or elsewhere.

We only have a few samples of ancient mitochondrial DNA. And the types of DNA may be distorted by variations in funereal practices in different cultures. There is also the issue of U's apparent disease susceptibility.

This paper is stuffed with lots of lovely stuff and is well worth a careful read.

Six major expansions are reported for the Americas.

Most European lineages expand before the neolithic except J1c3 and T2b, which seem to correspond to the neolithic in Anatolia/Caucasus.

The age of H1 and H3 expansions are specifically examined (pre neolithic).

Also:

"The H lineage in the Middle East was estimated ~15 kya, which was younger than European H (~18 kya). Although haplogroup H was thought to have a Middle East origin, previous work also supported that it expanded in Europe22. In Europe, H expanded 18–16 kya, which is definitely in Paleolithic Time."

GailT said...

I have several criticisms on the European section of this report, and I expect this casts doubt on the conclusions from other regions as well.

First, the samples are not random (as the authors claim). In fact, the samples are selected by the 1000 Genomes Project from narrowly defined regions and many large regions are not evaluated at all. It would make more sense to include results from other studies to represent data from other eastern Europe, Near East and western Asia.

The approach used to define star patterns is not described but is obviously flawed. For example, H1 and U5a are both treated as "star patterns", even though H1 has nearly 100 unique daughters while U5a has only two. There is something very obviously different in the patterns of these two groups that is entirely ignored.

The 39 U5b samples were not considered at all, and they only reported results for 26 U5a samples.

The report presents results from a model with no context, no scholarship, no reference or discussion of ancient DNA, and no critical evaluation of the model to assess if the model outputs are meaningful. It was published as a "Report" with expedited peer review, and it might have benefited from better peer review.

bmdriver said...

The Dharmic centric migration from out of India into Europe, as apposed to the Western centric notion of Aryan Invasion of India.



https://skydrive.live.com/?cid=7539d6f345689b72&id=7539D6F345689B72!116

Download file, it will open up as PDF document.

andrew said...

Given the closeness of Epipaleolithic and early Neolithic populations is very close its hard to discern which is at fault.

terryt said...

"It is perhaps due to a tendency of conflating spatial population expansions with the proliferation of descendants within a lineage".

The two probably are closely connected though. If a population is spatially limited it is likely to be genetically limited too. It would only be with a population's geographic expansion that multiple lineages would survive.

eurologist said...

For example, H1 and U5a are both treated as "star patterns", even though H1 has nearly 100 unique daughters while U5a has only two.

Gail, are we reading the same paper? Fig. 4 shows a star pattern in U5a with 26 unique positions. H1 is not even plotted; S3 lists 70, not 100. I admit I am confused though, as it seems that every sample is unique - but perhaps that is to be expected due to the random selection and full genome resolution.

Given the closeness of Epipaleolithic and early Neolithic populations is very close its hard to discern which is at fault.

Andrew, a lot of the European dates are in the 15,000 to 20,000ya range - that is not close to Neolithic, at all. You would have to propose a factor of ~1/3 error. In fact, as always when many dates are in that time period, if anything, it seems to me the dates should be corrected by about a factor of 1.5 to make sense. There clearly was no expansion in Europe during that time, but there was before (post-LGM) and after (Gravettian, ~22,000 - 30,000ya huge pan-European expansion).

Since they apparently did the ground work, I would have liked to see a color-coded network of the joint European and Middle-Eastern H populations. I don't see how that would not give a clue to the H origin and timing in Europe.

GailT said...

"Gail, are we reading the same paper? Fig. 4 shows a star pattern in U5a with 26 unique positions. H1 is not even plotted"

If you look at first level subclades, H1 has approximately 70, from H1a to H1bz, and it also has a large number of H1* samples that have not yet been given a name - thus my reference to nearly 100 unique daughters. That is a true star patter, nearly 100 lines radiating out. This tells us this H1 was part of a population that was rapid growing, and the best fit for that type of growth is farmers expanding into new territory.

U5a has only 2 first level subclades, U5a1 and U5a2. So this indicates that U5a lived in a population that was growing very slowly. The other nodes in U5a descend from U5a1 and U5a2, and while the pace of expansion begins to increase at the next level, it is still very slow compared to the explosive population growth we see in H subclades. So there is something very qualitatively different about H1 and U5a.

terryt said...

"This tells us this H1 was part of a population that was rapid growing, and the best fit for that type of growth is farmers expanding into new territory. U5a has only 2 first level subclades, U5a1 and U5a2. So this indicates that U5a lived in a population that was growing very slowly. The other nodes in U5a descend from U5a1 and U5a2, and while the pace of expansion begins to increase at the next level, it is still very slow compared to the explosive population growth we see in H subclades"

But the time of expansion of those two haplogroups could coincide. The number of steps from basal R is much the same for each haplogroup. U is a basal R haplogroup and so the U1, U5, U6 and U2'3'4'7'8'9 split is ancient within U. So the expansion within U5 (and possibly the other U clades) would be early. On the other hand H1 is several steps removed from basal R for a start. From R0 we have R0a'b and HV. From there we have several branches of HV including H itself. From where we get H1 and its subsequent expansion.

eurologist said...

Gail,

I think you are too hung up on (the arbitrary) nomenclature. This says little about expansion, except perhaps whether it started softly or briskly. Then there are chance events. Look at the figures of the paper - they (and others) speak for themselves.

GailT said...

"But the time of expansion of those two haplogroups could coincide. The number of steps from basal R is much the same for each haplogroup."

Behar et al. 2012 have an age estimate for U5a of 22,440 ybp, and for H1 of 9,889 ybp. While there is uncertainty in the age estimates, this seems to fit well with a slow post-glacial expansion of U5a and a rapid Neolithic farming and or herding expansion of H1.

On a different topic, I posted an anlysis of U5 in the Behar et al. 2012 Basque paper at Anthrogenica. 16% of the Basque population is haplogroup U5, but the majority of those (65%) are in a single, very rare subclade U5b1. Of the total population, 11% are U5b1f while only 5.8% were U5-(excluding U5b1f). It will be very interesting to see more FMS analysis of U5b1f. Currently we only have 4 U5b1f FMS samples, plus 1 from the 1000 Genomes Project.

GailT said...

"I think you are too hung up on (the arbitrary) nomenclature. This says little about expansion, except perhaps whether it started softly or briskly."

And that is the problem with the paper - it glosses over the fact that some haplogroups expanded extremely briskly, while other seem to have expand very slowly, or perhaps experienced contraction or bottleneck. If you break down the results by subclades, you can see precisely when certain sublades began to expand briskly.

Using U5a2a as an example, its age estimate is about 13000 ybp, and there are three lineages of U5a2a, of which two are represented by a single FMS sample. The third lineage, U5a2a1 has 61 FMS samples with a very strong star pattern and an age estimate of 6000 ybp. This suggests to me that U5a2a was part of a Mesolithic hunter-gather population (confirmed by an ancient DNA sample in Germany at 8700 ybp) and that U5a2a1 adopted agriculture or herding around 6000 years ago and then began its rapid population growth.

So it's really not just about the nomenclature - ignoring the extremely different expansion patterns in the data raises concerns about the accuracy of the analysis.