March 28, 2014

Oldest modern human genome from Siberia ~45 thousand years ago

Just a teaser from Ann Gibbons in Science:
In 2008, Siberian ivory carver Nikolay Peristov was searching for ancient mammoth tusks eroding from the banks of the Irtysh River in western Siberia, when he found fossilized bones instead. Back in his workshop in Omsk, he showed the bones to local paleontologist Aleksey Bondarev, who recognized a human thighbone. Bondarev in turn showed it to an anthropologist friend, and it was passed on up the chain to some of the world's top experts in human evolution. They dated it to 45,000 years ago, making it one of the oldest known modern humans in northern Asia and Europe.

Now, the bone has opened a window on the genetics of our species at a crucial moment: soon after their arrival in northern Eurasia. At a meeting* here last week, paleogeneticist Svante Pääbo of the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany, announced that his team has sequenced the thighbone's entire nuclear genome to high accuracy—an astonishing 42x coverage. "This is by far the oldest sequenced genome of a modern human," he said.

Because all living people in Europe and Asia carry roughly the same amount of Neandertal DNA, Pääbo's team thought that the interbreeding probably took place in the Middle East, as moderns first made their way out of Africa. Middle Eastern Neandertal sites are close to Skhul and Qafzeh, so some researchers suspected that those populations were the ones that mingled. But the team's analysis favors a more recent rendezvous. The femur belonged to an H. sapiens man who had slightly more Neandertal DNA, distributed in different parts of his genome, than do living Europeans and Asians. His Neandertal DNA is also concentrated into longer chunks than in living people, Pääbo reported. That indicates that the sequences were recently introduced: With each passing generation, any new segment of DNA gets broken up into shorter chunks as chromosomes from each parent cross over and exchange DNA. Both features of the Neandertal DNA in the femur suggest that the Ust-Ishim man lived soon after the interbreeding, which Pääbo estimated at 50,000 to 60,000 years ago.
The higher Neandertal DNA in the Ust-Ishim sample might be explainable by the negative selection against Neandertal material recently documented.  At 45kya, this sample is right around the time of the Early Upper Paleolithic at Kara-Bom in Siberia (and indeed anywhere), so this will be a hugely interesting sample when it is finally published.

Here is the program of the symposium on which this was apparently discussed (pdf). There seem to be quite a few interesting titles (but no abstracts).

164 comments:

terryt said...

"Both features of the Neandertal DNA in the femur suggest that the Ust-Ishim man lived soon after the interbreeding, which Pääbo estimated at 50,000 to 60,000 years ago".

Plenty early enough to have supplied the humans that reached Australia some 45-50,000 years ago. That fits the recent update of Y-DNA C phylogeny that has an early split between East Asian C2 and South/Southeast Asian C1. The placement is consistent with a Denisova contribution to Australia/New Guinea from the region where the Denisova population presumably actually lived.

Krefter said...

I bet this 45,000 year old Siberian was an ancient ancestor of MA-1 and Upper paleolithic Europeans. If not possible east Asian or an extinct Eurasian type.

gcochran said...


Negative selection doesn't have much effect on neutral alleles (non-coding, for instance), which must make up most (>90%) of the genome.

So the guy probably really had more Neanderthal DNA than we do, unless the difference is small.

Tobus said...

@barakobama:I bet this 45,000 year old Siberian was an ancient ancestor of MA-1

If so it will be interesting to see if he has the Native American affinity that MA-1 has. Although I suspect we'll see a pre-divergence pattern similar to the Tianyuan sample which is roughly the same age - a fairly general affinity to all non-Africans indicating the East/West split had not, or had only just, begun. Hopefully we won't have to wait too long to find out.

Raimo Kangasniemi said...

One thing was not mentioned and it caught my interest: Lack of mention of Denisovan DNA.

Unknown said...

Hmm 42x what coverage?

I wonder if we will get a Y chromosome out of this. Anyone want to play guess the Y chromosome just in case?

I am going to guess D. Found nearby today and old enough. But it is certainly not the only candidate.

eurologist said...

I remember people even quite recently still questioning that there was an "Aurignacian" (more generally, UP) Siberia.

Will indeed be interesting to see relations to Mal'ta and extant Europeans, W Siberians, and other people between there and the NW of the subcontinent.

The y-DNA haplogroup will also be telling. Did P already exist? Or is it earlier, like, e.g., F?

aeolius said...

You seem to suggest that Paablo considers human neanderthal interbreeding a unique event.
I wonder if this is just parsimony.If not, is there the problem of genetics? Of sexual disinterest?should

andrew said...

Wow. It wasn't entirely clear that there were even anatomically modern humans in Siberia until about 30kya until now. And, this is by far the most important data to date for calibrating the timing of Neanderthal admixture.

A 45kya date makes it a close think to know if this was an Aurignacian Cro-Magnon from Europe who spilled over into Siberia, or was involved in an earlier wave of migration from West Eurasia, or from an East Eurasian migration.

It is also very close in time to the also Siberian Denisovan remains although not all that close geographically (and apparently, despite the relative proximity in time and place, there is no Denisovan admixture).

It will be exciting to see the analysis of the contents and affinities of the genome itself. It sounds like the genotyping must have been done already, along with some early analysis, so there should be published analysis soon.

terryt said...

"I wonder if we will get a Y chromosome out of this".

Has anyone seen the whole article? I would have thought mt-DNA would be easier to isolate than would the nuclear genome. Any ideas?

"Anyone want to play guess the Y chromosome just in case?"

Could be almost anything.

"Or is it earlier, like, e.g., F?"

Taking into consideration the weaknesses of molecular-clockology (as Maju calls it) it is claimed F's diversification is younger than the date proposed for the individual. C's diversification is much earlier, although I wouldn't rule out D (or DE).

Rokus said...

What triggered me:
'The femur belonged to an H. sapiens man who had slightly more Neandertal DNA, distributed in different parts of his genome, than do living Europeans and Asians. His Neandertal DNA is also concentrated into longer chunks than in living people, Pääbo reported.'

In different parts of the genome??? What about overlap, or did full integration happen for the older chunks at allele level? If so, isn't this method to look for Neanderthal chunks heavily outdated and insufficient?
Apparently there is something in the dynamics of introgression that still needs to be figured out. I don't have the feeling the current investigation will have this focus.

Unknown said...
This comment has been removed by a blog administrator.
Anonymous said...

I remember some of the people here mentioning something about MA-1 resembling South Indians.

Do we have any genomic information on this individual to see if the same is true?

Sorry if I am getting some things confused here.

Unknown said...

This may be foolish but is it possible this individual came from the EAST, Berengia or even North America? If Neanderthals and other hominids lived to the east and interbred 60,000 years ago, then came west, would the genetic signals be different? And why isn't it possible that the interbreeding of these types is what enabled us to become modern?

terryt said...

"In different parts of the genome??? What about overlap, or did full integration happen for the older chunks at allele level?"

It may represent a totally different hybridisation event. Although the individual 'looks' modern perhaps it is an early arrival who failed to leave descendants. The mt- and Y- DNA will be interesting.

Krefter said...

Dienekes, do you have any idea when a paper on this will be published or there will be abstracts?

Rokus said...

'Anyone want to play guess the Y chromosome just in case?'

Anybody serious in his or her assumptions should feel obliged to come up with a prediction, for this is the perfect occasion to falsify any convential or outlandish claim.

I guess the Ust-Ishim individual was ancestral to Malta Boy. Since in my opinion Malta Boy's correlation with Central/South Asians, West Eurasians, Amerindians, and Oceanians was due to an early worldwide human expansion having MA-1 near the epi-center, it follows this same pattern should already be recognizable in Ust-Ishim. However, since I attributed a lower correlation with Amerindians and Papuans to an early split-off, and the higher correlation of Central/South Eurasians and West Eurasians (in this order) to a longer period of shared drift, I guess the quantification of ancestry components in Ust-Ishim should be reversed, thus resulting in a higher Amerindian/Papuan contribution and a lower West/Central/South Asian contribution.

since I proposed a "continuous process in the period between the earliest Eurasian AMH and Gravettian expansions, that spawned new CF-derived YDNA lineages (C1, G, H, IJ, L, T, MP, NO) all the time from a central or northern position all over Eurasia", I guess Ust-Ishim may be indeed C1 or F*. I don't think YDNA DE will be found here, even though these haplogroups may have been prominent in Asia before the expansion of YDNA F haplogroups.

I dare German, Terry and Hector to do the same, since they all have very strong opinions but so far only in a reactive sense, rearranging evidence afterwards according to the needs of their theories. According to the Criterion of Falsifiability of Popper all these theories can be dismissed as unscientific if they refuse to come up with their own falsification criteria right now. Please make your stances credible!

Grey said...

Quickest route to Siberia from Africa is via the Atlantic coast then across the top.

Krefter said...

"Since in my opinion Malta Boy's correlation with Central/South Asians, West Eurasians, Amerindians, and Oceanians was due to an early worldwide human expansion having MA-1 near the epi-center, it follows this same pattern should already be recognizable in Ust-Ishim."

MA-1 boy was a west Eurasian and also from the same source as Mesolithic/Upper Palaeolithic Europeans. You should not expect admixtures with ancestral components that are certainly much younger than MA-1 to accurately tell his ancestry.

Here is some evidence that MA-1 boy was from the same source as Upper Palaeolithic/Mesolithic Europeans.

Look how close he clusters with La brana-1 in these PCA's.

http://eurogenes.blogspot.com/2014/02/pca-of-five-ancient-genomes.html

I have not been able to read Olalde 2014, but i have heard from multiple people online that he is more related to MA-1 than any modern people are.

Check out the distribution of ANE(based on MA-1) in asia.

http://eurogenes.blogspot.com/2014/03/ancient-north-eurasian-ane-levels.html

It is nearly identical to the distribution of west Asian components in admixtures. West asian components also happen to be most related to north European components, which are distributed nearly identically to WHG(based on Loschbour).

Finally when MA-1 and la Brana-1 are set as the mammouth steppe reference in these admixtures they are distributed almost exactly the same. Even when La brana-1 is the reference native americans score the same as they do with MA-1 as reference.

http://eurogenes.blogspot.com/2014/03/ancient-north-eurasian-ane-levels.html

The reason MA-1 seems so mixed in admixtures is because his people contributed ancestry to so many modern people.

terryt said...

"I dare German, Terry and Hector to do the same, since they all have very strong opinions but so far only in a reactive sense, rearranging evidence afterwards according to the needs of their theories".

Of course theories always need to be adjusted with the discovery of new evidence. Perhaps the fact the Neanderthal admixture is in different parts of the genome from that of modern humans that the individual may represent a totally different hybridisation event. In which case he may not carry a 'modern' Y-DNA or mt-DNA at all. But at 45,000 years it is probably too early to be C1 specifically, or even any specific F-derived Y-DNA such as G, H, IJ, L, T, MP or NO. It could be basal D, C or F as the DE and CF splits are both 'dated' at 54,000 years.

sidoroffs said...

Grey,

when?

Grey said...

"I guess Ust-Ishim may be indeed C1 or F*."

That would get my vote also.

German Dziebel said...

@Rokus

"I guess the Ust-Ishim individual was ancestral to Malta Boy. Since in my opinion Malta Boy's correlation with Central/South Asians, West Eurasians, Amerindians, and Oceanians was due to an early worldwide human expansion having MA-1 near the epi-center, it follows this same pattern should already be recognizable in Ust-Ishim. However, since I attributed a lower correlation with Amerindians and Papuans to an early split-off, and the higher correlation of Central/South Eurasians and West Eurasians (in this order) to a longer period of shared drift, I guess the quantification of ancestry components in Ust-Ishim should be reversed, thus resulting in a higher Amerindian/Papuan contribution and a lower West/Central/South Asian contribution."

Out-of-America is a very simple model. My prediction is no different than yours. Ust-Ishim should be more Amerindian than MA-1. Your "early split-off" = "early migration out of the Americas."

Hector said...

"I dare German, Terry and Hector to do the same,..."

Really? Whatever is his Y haplogroup, you will try to argue that it is Caucasoid. I just had such a good laugh when I saw you guys suddenly make C1 "Caucasoid".

If the guy turns out to have Neanderthal Y DNA you guys will argue that it is a long lost Caucasoid Y haplogroup.

The comic value rivals your being humiliated by Vincent and co. back in the days of DNA-forum.

Rokus said...

@Obama,
'You should not expect admixtures with ancestral components that are certainly much younger than MA-1 to accurately tell his ancestry.'

Certainly not. However, in my opinion various components in MA-1 are not caused by admixture but instead by a combination of ancestry and shared drift. The Papuan component naturally has few shared drift or none, so the ancestry component should make the total Papuan component bigger. The same applies to the Amerindian component. Instread, the total Central/South Asian component would include less shared drift and more ancestry and thus be "smaller". And so on.

Thanks! Your stance that the various components in MA-1 are exclusively due to admixture may be falsified in case of an utter absence of "modern" components for Ust-Ishim.

Rokus said...

German,
'Out-of-America is a very simple model. My prediction is no different than yours. Ust-Ishim should be more Amerindian than MA-1. Your "early split-off" = "early migration out of the Americas."'

Or maybe Out of America is also a littlebit scary? What about the prominence of the various other "modern" components in Ust-Ishim? If Ust-Ishim was ancestral and predating virtually all non-African populations that can be recognized in the modern components of MA-1, but not the Amerindian population, those non-Amerindian components would all be overruled by the Amerindian component. In other words, virtually no other modern components for Ust-Ishmi. Agreed?

Rokus said...

TerryT,
'Of course theories always need to be adjusted with the discovery of new evidence.'

No, according to Karl Popper this would be a serious offence. A theory or hypothesis is only "scientific" if it meets certain criterions, one of hose being the falsification criteria. Hypothesis that don't meet those criteria are unscientific. Pseudoscience!

But what about your stance that YDNA MP and C are SE Asian? wouldn't this require all CF to be SE Asian? An extinct YDNA profile for Ust-Ishmi would then be most adequate. Or did you mean back and forth migrations to and from SE Asia? Anyway, basal D, C or F is about all there is, what means your hypothesis (whatever it is) can't be falsified and thus is not scientific.

Rokus said...

"I1 x I2 split (Europe) : ~25 kya
J1 x J2 split (West Asia) : ~25 kya
R1b x R1a split (Cental Asia ?) : ~ 25 kya
O1 x O2/3 split (East Asia) : ~25 kya"

There is another hypothesis that R1a-SRY10831, for having this "ape-like" mutation (Hammer 1998), may have been the oldest subclade - together with YDNA A.
This would require quite some reforms on the OSIGG-tree, though this hardly can be taken as a serious argument against such a possibility.

The main problem for this "prediction" is that R1a for the "Siberian man" Ust-Ushmi would for sure be dismissed as a contamination.

Grey said...

sideroffs

"when?"

Most (all?) of pre-neolithic.

(so possibly multiple times and possibly in reverse also i.e. Siberia to France by a polar or steppe route > Siberia to France by a southern coastal route)

Northern Eurasia was an autobahn relatively speaking imo.

Grey said...

The map used in this post illustrates how much of an autobahn northern Eurasia would have been relatively speaking for HGs on foot especially when you consider the actual distances are compressed by the effect of the earth being a sphere.

Grey said...

"I just had such a good laugh when I saw you guys suddenly make C1 "Caucasoid"."

If evolution (mutation) is driven by effective population size then to me that implies a succession of "out of" events as humans adapted to successive latitude layers based on which regions in that latitude layer could support the largest effective population size.

Say the sequence went 1) Out of the tropics 2) Out of Africa (spreading around most of the globe at least near coasts and rivers) then looking at the map the most likely region for the next "out of" event in the next latitude band up (if you assume fertile surface area ~ effective population size) is India which would make C* basal everywhere but Africa.

If you assume DE mostly replaced C in certain regions as a result of subsequent "Out of" events centered somewhere around Egypt and South China and the other subclades of C as regional variants then you'd get C1-F* being basal in the original home region of C1 and in the latitude band that goes: Spain - Anatolia - Central Asia - North China and all points north of that latitude band i.e. more or less cut in two into a southern and northern branch.

Grey said...

@myself

"i.e. more or less cut in two into a southern and northern branch."

Alternatively perhaps more of T shape.

terryt said...

"But what about your stance that YDNA MP and C are SE Asian? wouldn't this require all CF to be SE Asian?"

Wrong. It requires no such thing.

"Or did you mean back and forth migrations to and from SE Asia?"

Yes. In fact when you claim I am guilty of 'rearranging evidence afterwards according to the needs of their theories' can you provide any examples of my having done so? You might like to check this essay from 2009:

http://humanevolutionontrial.blogspot.co.nz/2009/06/human-evolution-on-trial-mitochondrial.html

Remember this was written before direct evidence of Neanderthal admixture had become available, before a 'Denisovan' population was even imagined and well before Y-DNA A00 had been discovered. There have also been several adjustments to the haploid phylogeny. In spite of all this there is very little indeed in the essay I would now change. I have therefore not altered my interpretation of the evidence since 2009 (before that in fact).

"This would require quite some reforms on the OSIGG-tree, though this hardly can be taken as a serious argument against such a possibility".

Exactly. Unless you want to join German.

"An extinct YDNA profile for Ust-Ishmi would then be most adequate".

Possibly, but check this out:

http://dienekes.blogspot.co.nz/2012/08/dates-of-major-clades-of-y-chromosome.html

And the comments here:

http://dienekes.blogspot.co.nz/2012/07/estimating-age-of-y-chromosome-adam_30.html

Taking into consideration all the uncertainty of 'molecular-clockology' it is obvious at a glance that the Ust-Ishim man is too early to belong to any haplogroup downstream of IJK. That means there is no chance, with P at 33,000 years, that he is R. However R1's MRCA at 24,000 years is entirely compatable with MA-1 being R.

Although the CF split is given as 56,000 years it seems from other evidence that F's expansion is considerably after that split. That would eliminate any F-derived haplogroup. On the
other hand it is extremely likely humans had reached Australia some time before 46,000 years ago. Therefore it is completely possible the Ust-Ishim man had a haplogroup basal to both C1 and C2. Of course D cannot be ruled out.

"Northern Eurasia was an autobahn relatively speaking imo".

yes. Unlike South Asia. If the individual turns out to be basal C it would support a Central Asian route east and then to Australia, and tend to make a 'southern', or 'coastal', route unlikely. C1 and C2 would have split in East Eurasia somewhere.

German Dziebel said...

@Rokus

". In other words, virtually no other modern components for Ust-Ishmi. Agreed?"

Out-of-America predicts that the closer the sample is to the Americas geographically and the older the sample is, the more likely it's going to be closer to modern Amerindians than to other modern human populations. Depending when the migration out of the Americas occurred, Ust-Ishim may be either all-Amerindian or just exhibit a greater affinity to Amerindians than MA-1.

Also, note that your Popperian criteria of falsification eliminate out-of-Africa as a scientific theory because there aren't (and there cannot be considering the lack of ancient DNA preservation in Africa) any ancient samples from Sub-Saharan Africa to test it.

Rokus said...

@Hector,
'Really?'

Sorry for asking, I thought you might have some ideas of your own. Apparently you pretend to spend a good time laughing at ideas you don't like, though maybe it's rather frustration for you not being able to come up with suitable alternatives? However, such alternatives still need to pass this exam.

C1 Caucasoid? Since when? Why coming up with a Middle East straw man?

Talking about V., I remember somebody reviling my evidence for an expansion wave for YDNA R, and subsequently alerting the team of Myres et al. (2010) about evidence that I put forward. Most curious of all, they did not accept his dating that would have rejected their LBK scenario beforehand, and still he choose to ally with them. Winning a battle and losing the war, what a waste!

eurologist said...

I consider DE as a pre-Toba remnant that got overrun by CF subgroups - so I don't think that's it.

I do think every indication is that K(xLT) went to western SE Asia, and split there, with the K's and M and S moving south, NO east and northeast, and P moving north and then back west to the extreme northwest of the subcontinent. Evidently is was very successful, which might have also driven G and IJ into West Eurasia and Europe, and marginalized H and LT to the south.

I find it most appealing that Siberia was populated in waves from the Upper Indus River Valley moving north, west of the Himalayas, during climatically advantageous time periods.

So, I find P the most plausible haplogroup candidate. I find it possible that Q actually derived from P when it was already present in Siberia, while R derived from P in the Upper Indus River Valley or immediate surrounding regions to the W and NW.

F* is a possibility, because it almost certainly was around in the subcontinent at the time, but I think it is much less likely.

Rokus said...

'Out-of-America predicts that the closer the sample is to the Americas geographically and the older the sample is, the more likely it's going to be closer to modern Amerindians than to other modern human populations. Depending when the migration out of the Americas occurred, Ust-Ishim may be either all-Amerindian or just exhibit a greater affinity to Amerindians than MA-1.'

However, if Native Americans are indeed a mixture between West Eurasians (one third) and East Asians while Ust-Ishim, like MA-1, is not, then Ust-Ishim may never fully represent the Amerindian component. One third will be the maximum, and the rest of the Ust-Ishim components will still find a better representation in other modern populations, even though any shared drift could be excluded because of its age! This should be the criteria where Out of America must be rejected in favour of Raghavan's dual ancestry.

'Also, note that your Popperian criteria of falsification eliminate out-of-Africa as a scientific theory because there aren't (and there cannot be considering the lack of ancient DNA preservation in Africa) any ancient samples from Sub-Saharan Africa to test it.'

The number of violations in modern publications is countless. Few genetic discoveries ever confirmed existing hypotheses or theoretical constructs, unless after being stripped of significance. Pseudoscience is typical for closed systems that may be religious, political or overly comprehensive such as Kurganism or Chomsky.

Shayan said...

I have a question for all the contributors here--I'm highly interested in finding a layperson's history of genetic changes in our species over time, especially after reading this article. Basically, where we come from genetically, how and why we evolved (and when of course), and what we have learned about prehistoric population movements from genetic discoveries. Any books I can read, or shall I just stick to reading this wonderful blog (which I've been enjoying for years but haven't been fully comprehending as a result of my lack of background in genetics).

Rokus said...

'you claim I am guilty of 'rearranging evidence afterwards according to the needs of their theories' can you provide any examples of my having done so?'

New evidence always has to be fit into existing theories, this is not an "accusation" but a fact of life when new evidence was not already accounted for by predictions. Theories need a solid empirical foundation, though beyond that you need the theories to be robust enough to support certain predictions. The falsification criteria is a prediction that can't be true for the theory to be accepted. So, my simple question was just to come up with a possible outcome of the genetic research on Ust-Ishim that could damage your current stance on autonomous developments in Oceania. Something, that could be tested and would mean something. For instance, I think the worse that could happen for your views would be YDNA C being measured in Ust-Ushim? Here you'ld have your falsification criteria. But you say it isn't, so I'm a bit lost about the nature or consistency of your stance. Well, never mind.

German Dziebel said...

@Rokus

"This should be the criteria where Out of America must be rejected in favour of Raghavan's dual ancestry."

The data at hand already rejects Raghavan's model. Let's just wait for Ust-Ishim results.

klevius said...

Continue reading this and other blogs cause there are no good unbiased books - here you get at least some occasional fights. Moreover, our genetic understanding isn't even close yet to what would be needed to write such a book. This opens up for hypothetical scenarios. As a previous photographer I myself use the classic method of turning the picture upside down which sometimes really pays off...

Grey said...

@Shayan

"I'm highly interested in finding a layperson's history of genetic changes in our species over time"

I doubt there's a modern one. There's a lot of 19th century ones which probably have lots of interesting data although I assume the interpretations of that data will reflect the times.

terryt said...

"I do think every indication is that K(xLT) went to western SE Asia, and split there, with the K's and M and S moving south, NO east and northeast, and P moving north and then back west to the extreme northwest of the subcontinent".

Yes. Although to me it is more than just a little possible that M and S had to move very littler from where K(xLT) splintered. That leaves just NO and P with the greatest distance to move from that split.

"Evidently is was very successful, which might have also driven G and IJ into West Eurasia and Europe"

I am reasonably certain those two haplogroups are 'remnants' left behind as haplogroups F and K moved east from somewhere near Africa (The Caucasus/Zagros?).

"and marginalized H and LT to the south".

Two haplogroups that broke off further along the journey would be my interpretation of the phylogeny.

"I find it most appealing that Siberia was populated in waves from the Upper Indus River Valley moving north, west of the Himalayas, during climatically advantageous time periods".

One thing we can be reasonably sure of is that the region where the individual was found would have become uninhabitable at the height of the climate cooling. The low-lying region of the Ob Valley was possibly filled with a glacier or a very large lake. Any population there would have been pushed far to the south (or east).

"F* is a possibility, because it almost certainly was around in the subcontinent at the time, but I think it is much less likely".

I agree, much less likely. Especially when we consider that all the F-derived haplogroups are basically confined to south of the Caucasus/Himalaya axis. The only exceptions are the MNOPS-derived haplogroups where NO has apparently moved north along the China/Tibet border region and P has emerged through the gap between the Caucasus and the Hindu Kush. Neither of these movements can confidently be dated at anywhere near 45,000 years ago.

terryt said...

"my simple question was just to come up with a possible outcome of the genetic research on Ust-Ishim that could damage your current stance on autonomous developments in Oceania. Something, that could be tested and would mean something".

A finding of any Y-DNA F-derived haplogroup would tend to damage my stance on Oceanian migration. F should not have spread so early, especially not so far north, whereas C should have.

"For instance, I think the worse that could happen for your views would be YDNA C being measured in Ust-Ushim?"

Y-DNA C would be absolutely what I would expect. To me C looks to have moved first across Central Asia into East Asia, and then south to Wallacea and Australia. Later extreme climate cooling eliminated C and associated mt-DNA from much of Central Eurasia leaving the habitat open for the later expansion of P. In the meantime C had diversified in Eastern Eurasia into C1 and C2, with C1 reaching Wallacea and Australia, then South Asia and ultimately back into Central Eurasia from there.

"I'm a bit lost about the nature or consistency of your stance".

The first Y-DNA to cross Wallace's Line was almost certainly C1. And it had done so by 46,000 years ago, and probably earlier. In Australia it formed C1d and in Wallacea C1c. Some time around 40,000 years ago K(xLT) arrived in Wallacea and K1, K2, M and S crossed the line, perhaps 35,000 years ago, and entered Melanesia/New Guinea especially, but K1 also entered Australia. This population had a better boating technology than the first arrivals and spread themselves through Near Melanesia with just a relatively minor presence on the New Guinea mainland. A third movement of C1c from Wallacea to the Bird's Head and to Near Melanesia around 10,000 years ago eventually merged into the movement of Y-DNAs O1 and O3 from the Asian mainland and gave rise to the Austronesian eastward expansion around 4000 years ago.

I am aware of no evidence that conflicts with the above, and most corroborates it absolutely.

pconroy said...

My guess is:
CF, C, F or P

Grey said...

If recent strong selection reduces diversity then couldn't you get a situation where you have two sub-populations of a main population with two different haplotypes with the same level of diversity and one of them moves to a new region where they experience stronger selection and become **less** diverse through selection leading to a false impression that the two haplotypes have very different relative ages?

Rokus said...

@German,
'"This should be the criteria where Out of America must be rejected in favour of Raghavan's dual ancestry."'

'The data at hand already rejects Raghavan's model. Let's just wait for Ust-Ishim results.'

It's OK that you've already convinced yourself, though it worries me that apparently you don't accept any falsification criteria beforehand. You'll talk yourself out of that one also, wouldn't you? However, this is precisely what defines your theory as unscientific.

eurologist said...

Neither of these movements can confidently be dated at anywhere near 45,000 years ago.

But P could be now - if it is confirmed in Ust-Ishim. But then, life is full of surprises and disappointments - sometimes a single haplogroup can open the door, sometimes it tells us almost nothing.

However, let's not forget that many ancient haplogroup finds have pushed their previously "established" time estimates back, often by a factor of two or more.

I don't see a big problem with the timing of P since the extreme eastward expansion of early Siberians has been dated to ~35 kya, and almost certainly was related to Q, but pre-dated Mal'ta (R* with a few mutations) by 10 ky. It is not uncommon for splits like that to occur over a huge time span, when populations at times were rather small.

German Dziebel said...

@Rokus

"It's OK that you've already convinced yourself, though it worries me that apparently you don't accept any falsification criteria beforehand. You'll talk yourself out of that one also, wouldn't you? However, this is precisely what defines your theory as unscientific."

You are confusing theory falsification criteria with data misinterpretation. Raghavan have misinterpreted the facts. MA-1 is a mix of Amerindian and West Eurasian (plus other components) genes. There's nothing in MA-1 that suggests that Amerindians are a mix of West Eurasians and East Asians. MA-1 is not even located in the Americas.

Raghavan's study is a good illustration of how mainstream ideas of the peopling of the Americas are unscientific by your own definition. Out-of-America works, on the contrary, within very strict criteria of falsifiability.

terryt said...

"But P could be now - if it is confirmed in Ust-Ishim".

I'd be very surprised. I am not aware of any 'dating' that places P as early as that.

"However, let's not forget that many ancient haplogroup finds have pushed their previously 'established' time estimates back, often by a factor of two or more".

Yes. For me the big surprise was mt-DNA B in northern China by 45,000 years ago. But in that case the date didn't actually conflict with projected dates because I don't think there were any.

"I don't see a big problem with the timing of P since the extreme eastward expansion of early Siberians has been dated to ~35 kya, and almost certainly was related to Q, but pre-dated Mal'ta (R* with a few mutations) by 10 ky."

That expansion is likely to have been extremely rapid though. And a date of 35,000 years fits closely the date of 33,000 usually suggested for P's TMRCA.

"If recent strong selection reduces diversity then couldn't you get a situation where you have two sub-populations of a main population with two different haplotypes with the same level of diversity and one of them moves to a new region where they experience stronger selection and become **less** diverse through selection leading to a false impression that the two haplotypes have very different relative ages?"

Yes. Although a population moving to a new region will usually be able to expand and so diversify. It is more likely to be the 'resident' population that suffers reduced diversity. On the other hand the formation of new subhaplogroups is often assumed to be the result of selection at an expanding margin of a haplogroup's geographic range.

Rokus said...

@German,
'MA-1 is a mix of Amerindian and West Eurasian (plus other components) genes.'

No, because of the age of MA-1 this is impossible. Instead, the components in MA-1 reflect their shared ancestry plus a period of shared drift.

'There's nothing in MA-1 that suggests that Amerindians are a mix of West Eurasians and East Asians.'

Only when you miss the wider picture. MA-1 represents close affinity to only one component of Amerindians.

'MA-1 is not even located in the Americas.'

This is only a problem for your hypothesis.

'Raghavan's study is a good illustration of how mainstream ideas of the peopling of the Americas are unscientific by your own definition.'

It's all about applying logic to arrive at a hypothesis. Falsification criteria are one step beyond this. However, you could still test the dual origin by applying falsification criteria on MA-1, such as the availability of just one component! The result is affirmative.

'Out-of-America works, on the contrary, within very strict criteria of falsifiability.'

Such as...?

German Dziebel said...

@Rokus

"No, because of the age of MA-1 this is impossible. "

What does age have to do with this?

"Only when you miss the wider picture. MA-1 represents close affinity to only one component of Amerindians."

No, both components are actually one Amerindian component at K=2. Take a look at Rosenberg et al. 2002. Plus if you re-read Raghavan, you'll see that MA-1 is equally close to virtually all Amerindians with the exception of the northernmost groups.

"This is only a problem for your hypothesis."

Not at all. This supports my hypothesis: an Amerindian component in Asia, not an Asian component in America.

"It's all about applying logic to arrive at a hypothesis. Falsification criteria are one step beyond this. However, you could still test the dual origin by applying falsification criteria on MA-1, such as the availability of just one component! The result is affirmative."

Where's that single component in MA-1?

"Such as...?"

Read above. The frequency of the Amerindian component increases with time depth and geographic proximity to the Americas. All the current data supports this observation but we need more. The African component is nowhere to be found in the ancient remains in Eurasia. At least so far...

Grey said...

@terryt

"On the other hand the formation of new subhaplogroups is often assumed to be the result of selection at an expanding margin of a haplogroup's geographic range."

Yes, that's the other reason you might expect less diversity in a sub-population that moves compared to the one that stays behind.

If you have population A with diversity x and a sub-population of A moves to a new region and expands rapidly then the derived population B only took a sub-set of the original diversity with them.

e.g. if population A has six sub tribes and one sub tribe moves to region B and expands wouldn't you expect the derived population B to have less diversity than A even though they originally came from the same population?

This is before any stronger selection applied by moving to a new environment which would also tend to reduce diversity while it lasted.

terryt said...

"Raghavan have misinterpreted the facts".

Of course he has. He has not the slightest clue waht he is talking about. Only God (otherwise known as German Dziebel) knows anything at all about human origins.

"There's nothing in MA-1 that suggests that Amerindians are a mix of West Eurasians and East Asians. MA-1 is not even located in the Americas".

There are none so blind as those who will not see.

"Only when you miss the wider picture. MA-1 represents close affinity to only one component of Amerindians".

You're wasting your time, Rokus. German is in no way interested in 'the wider picture'.

By the way, I've come to the conclusion that the Ust-Ishim Y-DNA is most likely D. Although haplogroup C* may turn out to belong to the new 'C!' it is possible they are actually C(xC1,C2). When we consider C* and D* we find D in Mongolia/Tibet and C in East/Southeast Asia. That means at around 50,000 years ago we have the four CT haplogroups spread in an arc thus: E - Africa, F - Anatolia/Iran, D - Central/East Asia and C - East/Southeast Asia.

Rokus said...

'What does age have to do with this?'

MA-1 is older than the modern components. per definition MA-1 can't be the result of modern admixture, unless your hypothesis includes time travelling. Instead, the modern components in MA-1 should exclusively be interpreted as a combination of shared ancestry and shared drift in the period between split-off and MA-1.

'"Only when you miss the wider picture. MA-1 represents close affinity to only one component of Amerindians."'

'No, both components are actually one Amerindian component at K=2. Take a look at Rosenberg et al. 2002. Plus if you re-read Raghavan, you'll see that MA-1 is equally close to virtually all Amerindians with the exception of the northernmost groups.'

Components "aren't", they were defined. The investigators tested various K values and chose to distinguish the East Asian component and a global Amerindian component from the rest. This does not mean that Amerindians don't have an East Asian component, only that this was deliberately excluded from the definition. This is just a matter of mathematics. Say you have two pure populations and one mixed population, but choose to consider the mixed population as pure (eg. Amerindian). If then you define one pure population as "undefined" and the other pure population as "East Asian", then the mixed but defined pure "Amerindian" population will never appear as a mixture of East Asian and the undefined other pure population. Instead, the undefined population (say MA1) may be described as a mixture of the mixed (Amerindian) population and "something else". This "something else" may subsequently be decomposed into other components, but never the pure East Asian population.

That is also why I don't exclude the possibility that East Asian is actually a mixture between ancestral East Asian (some local hominine) and Amerindian-like AMH. The only way to find out is to consider Ust-Ishim as ancestral AMH and set this out against East Asian as a mixed population. Then most likely Amerindians appear as a mixture between ancestral Ust-Ishim and ancestral East Asian, while MA-1 does not.

'both components are actually one Amerindian component at K=2'

Of course, that's the consequence of chosing K=2. Please try to understand the basics.

'The African component is nowhere to be found in the ancient remains in Eurasia. At least so far...'

This could have been a valuable falsification criterion for Out of Africa, though for Out of America its nothing but a strawman.

epoch said...

"In 2008, Siberian ivory carver Nikolay Peristov was searching for ancient mammoth tusks eroding from the banks of the Irtysh River in western Siberia, when he found fossilized bones instead. "

Uhm. If the bone really was fossilized it wouldn't be able to yield a genome, would it?

terryt said...

@ German:

"This supports my hypothesis: an Amerindian component in Asia, not an Asian component in America".

Let's get this straight at last. I understand German to believe the following:

At some unspecified time an unidentified H. erectus-type population entered America, presumably via Beringia. A problem immediately arises in that the minute ecological footprint shows numbers were certainly not huge. Therefore a reasonable genetic diversity is impossible. The American population numbers didn't become substantial until Clovis.

But before that time German further believes that at about 45,000 years ago members of this very sparsely distributed, genetically restricted, population entered Eurasia, ultimately replaceing resident H. erectus-derived populations there. He provides no driving force to expalin this expansion. It was obvioulsy not population pressure.

The next step, so German apparently believes, is that the genetically restricted American population diverged into a 'northern' genetic group and 'the rest'. Around 10,000 years ago this northern group in turn emerged and came to dominate Northern Eurasia.

The problem with his whole idea is that he can offer no explanation for the apparent genetic diversity noted in America. The generally accepted scenario, on the other hand, easily explains that diversity: two Eurasian populations formed a hybrid population in northeast Eurasia, thus making the combined population more genetically diverse than either source population. The hybrid population was able to maintain this diversity as it expanded into and through America without suffering either selection or drift. The haploid phylogeny also supports the concept of a hybrid population. That is why German is so hell bent on dismissing the whole presently accepted phylogeny.

agiering said...

In regards to Mal'ta, here is the way I see it: The admixture plot in Raghavan et al's study shows [Southern] amerindians are pure and Mal'ta as admixed (page 56 --- here. In addition, Mal'ta showed no East Asian admixture. Lazaridis et al found precisely the same thing here. Indeed, southern Amerindians are rivaled only by southern Africans in their degree of purity at every k level. Instead of making the most natural interpretation of the data at hand, however, Lazaridis and Raghavan assume Mal'ta was pure and Amerindians (and West Eurasians) were mixed. They assumed Mal'ta was mixed because they were so shocked that there was no East Asian admixture in Mal'ta. That's why they invented the "ANE" population that is supposedly similar to Mal'ta and contributed to the ancestry of West Eurasians, South Asians and Amerindians (but not East Asians). Of course, this hypothesis has its own problem, namely, how ANE managed to completely avoid intermixing with East Asians, despite intermixing with so many other non-African populations.

Basically, I see the Admixture plots in the Mal'ta papers as supporting "Out of America" for 4 related reasons:
A) Establishing the antiquity of the Amerindian genetic signature, at least as far back as 24K years ago.
B) Failing to provide evidence for the east Asian genetic signature -24K years ago in the part of the world that you would expect to find it.
C) showing that Amerindians (southern ones at least) are pure at all K levels.
D) showing trace amounts of Amerindian admixture in Europeans, which are higher for the older foragers than the more recent neolithic farmer.

Really, A and B combined form the most important part of the finding IMHO, because it's so hard to reconcile with the conventional hypothesis that Amerindians derive from an East Asian population.

In other words, as far as I can tell German is spot on with his analysis of the data. Either that or German and I are smoking the same stuff.

agiering said...

@ terryt

"The problem with his whole idea is that he can offer no explanation for the apparent genetic diversity noted in America. The generally accepted scenario, on the other hand, easily explains that diversity: two Eurasian populations formed a hybrid population in northeast Eurasia, thus making the combined population more genetically diverse than either source population. The hybrid population was able to maintain this diversity as it expanded into and through America without suffering either selection or drift. The haploid phylogeny also supports the concept of a hybrid population. That is why German is so hell bent on dismissing the whole presently accepted phylogeny."

Now I'm just downright confused. I've heard it so many times that the LACK of genetic diversity amongst Amerindians is proof of their recency and the high amount of genetic diversity amongst Africans is proof of their status as an archaic ancestral population. Out-of-Africa critics like German have been saying for some time now that genetic diversity is not a good marker for age; things like admixture can at least as well explain genetic diversity.

Now you are turning things on their head by stating that Amerindians are TOO diverse to be ancestral? And that the only explanation for their genetic diversity is that they are recent and admixed? You do realize that the exact same argument could be used against the out of Africa model, don't you?

German Dziebel said...

@Rokus

"MA-1 is older than the modern components. per definition MA-1 can't be the result of modern admixture, unless your hypothesis includes time travelling. Instead, the modern components in MA-1 should exclusively be interpreted as a combination of shared ancestry and shared drift in the period between split-off and MA-1."

You're making a typical logical mistake for the students of prehistory confusing "derivation" with "attestation". MA-1 attests to an Amerindian component in Asia at 24,000 years ago. It existed in the Americas all the way down. Only if you have an ancient sample in the Americas that's East Asian and West Eurasian, but not Amerindian that you can say that modern Amerindians are derived from it.

"Components "aren't", they were defined."

Components first are. And then they are denied.

"Say you have two pure populations and one mixed population, but choose to consider the mixed population as pure (eg. Amerindian). "

Amerindians are unadmixed in all the ADMIXTURE runs.

"That is also why I don't exclude the possibility that East Asian is actually a mixture between ancestral East Asian (some local hominine) and Amerindian-like AMH."

Why not between Papuans and Amerindians? Just curious.

"Of course, that's the consequence of chosing K=2. Please try to understand the basics."

You are beginning to sound like Tobus. Of course, I understand the basics. You are starting to sound like you're denying them.

"This could have been a valuable falsification criterion for Out of Africa, though for Out of America its nothing but a strawman."

You are referring to just half of my falsification criteria.

terryt said...

@ agiering:

"Of course, this hypothesis has its own problem, namely, how ANE managed to completely avoid intermixing with East Asians, despite intermixing with so many other non-African populations".

The simple, and obvious, explanation is that the MA-1 population had made it further north that had the 'East Asian' population at that time. Once MA-1 had moved east past the East Asian population they were able to expand further north as well. Elements of that East asian population came to more prominence in northern America than it did further south, thus making southern American populations more like MA-1 and the more northern American populations more 'East Asian'. That is what the haplogroup evidence tells us.

"I see the Admixture plots in the Mal'ta papers as supporting 'Out of America' for 4 related reasons:
A) Establishing the antiquity of the Amerindian genetic signature, at least as far back as 24K years ago".

You're forgetting MA-1 was not found in America, and we don't know anything like it was present in America 24k years ago .

"B) Failing to provide evidence for the east Asian genetic signature -24K years ago in the part of the world that you would expect to find it".

Actually the EDAR 370A mutation is estimated to be around 35,000 years old and originated to the south of where the MA-1 individual was found. Thus supporting the idea that the East Asian population was yet to venture so far north as MA-1.

"C) showing that Amerindians (southern ones at least) are pure at all K levels".

Only because, as Tobus has so consistently tried to explain, they have been taken as such to enable more recent admixture to be traced. It is a matter of definition.

"D) showing trace amounts of Amerindian admixture in Europeans, which are higher for the older foragers than the more recent neolithic farmer".

Because the MA-1 population contributed to much of the northern Eurasian population, not just to Amerindians.

"I've heard it so many times that the LACK of genetic diversity amongst Amerindians is proof of their recency and the high amount of genetic diversity amongst Africans is proof of their status as an archaic ancestral population".

Amerindians have far less diversity than do Africans.

"Out-of-Africa critics like German have been saying for some time now that genetic diversity is not a good marker for age"

German has actually been claiming the supposed great diversity within Amerindians 'proves' they are an ancirent population.

"things like admixture can at least as well explain genetic diversity".

That is what Rokus, Tobus and I have been trying to point out to German.

"You do realize that the exact same argument could be used against the out of Africa model, don't you?"

Possibly, but all sorts of other evidence shows that the Amerindians are a hybrid population, as is every other population on earth, including Africans. I agre that diversity alone, except haplogroup diversity, tells us nothing about ancestry. With my earlier comment I was trying to point out German's faulty logic over his claimed Amerindian genetic diversity.

@ Grey:

"If you have population A with diversity x and a sub-population of A moves to a new region and expands rapidly then the derived population B only took a sub-set of the original diversity with them".

True. But once that subgroup had entered the new region they would usually be able to expand once more leading to diversification within the haplogrop(s) that had made it out.

eurologist said...

"Of course, this hypothesis has its own problem, namely, how ANE managed to completely avoid intermixing with East Asians, despite intermixing with so many other non-African populations. "

agiering,

That is not a problem at all.

Firstly, it is wrong: "ANE" or Ancient Siberian populations did admix with East Asians - and they still live there, all over Siberia and Central Asia. Except, most of that admixture apparently took place after LGM, for reasons.

Secondly, they did not admix with South Asians. Their origin is just old enough that both carry related elements. After all, it is established that almost all non-African maternal haplogroups originated in S and SE Asia, and the paternal haplogroup of importance here (P --> Q, R) as well. So, no surprise - in fact, this was predicted.

Thirdly, Ancient Siberians existed as far back as 45,000 years ago, long before East Asians ventured even remotely as far north. By archaeological finds, Ancient Siberians expanded into the far East of Siberia and into Beringia ~35,000 ya, and occupied that far northerly area almost unchallenged - except for a coastal expansion of Ancient E Asian northward, with which they apparently admixed. During climatically challenging times, the Beringian population became an isolate in an island of relatively mild climate that did not exist in E Siberia, and E Siberia got depopulated until after LGM. The Beringians expanded into the Americas, while the (now almost modern) E Asians expanded into Siberia to the W, N, and NE, where in two successive waves two more and more typical Mongolian-like populations (Na-Dene and then Inuit) expanded into the islands and into the Americas, with subsequent admixture.

This "model" explains everything and is 100% consistent with the archaeological, climatic, linguistic, and genetic record.

Rokus said...

'Only if you have an ancient sample in the Americas that's East Asian and West Eurasian, but not Amerindian that you can say that modern Amerindians are derived from it.'

That's not true, why in the Americas if admixture most probably already occurred in Asia? Moreover, the Asian component of Amerindians was already clear from previous investigation.

'You are referring to just half of my falsification criteria.

The other half was not very clear yet.

'Amerindians are unadmixed in all the ADMIXTURE runs.'

'"Of course, that's the consequence of chosing K=2. Please try to understand the basics."'

'You are beginning to sound like Tobus. Of course, I understand the basics. You are starting to sound like you're denying them.'

Anyway, reading some technical documentation won't hurt, eg.: Alexander et al. - Fast Model-Based Estimation of Ancestry in Unrelated Individuals (2009)

'We present a new algorithm and a program, ADMIXTURE, for model-based estimation of ancestry in unrelated individuals.'
[...]
'Choice of an appropriate value for K is a notoriously difficult statistical problem. It seems to us that this choice should be guided by knowledge of a population’s history. Because experimentation with different values of K is advisable [...]'

You can see that the choice to have Amerindians as a reference population by Raghavan et al. doesn't have anything to do with ADMIXTURE or Amerindians being an unadmixed population. The ADMIXTURE runs of this publication were exclusively designed to show that Amerindians had a second non-East Asian component that could already be recognized in MA-1. There is no limit whatsoever in running ADMIXTURE using different reference populations at varying K.
However, Amerindians may only show up as an admixed population by running ADMIXTURE without Amerindians as a reference population!

Please, try to be objective and let it go. Maybe ADMIXTURE could help you in another way, but at least Raghavan et al. are not helpful in your quest. Somebody has to run ADMIXTURE in the way I told you in my previous comment, such that East Asians could be recognized as a mixed population having an Amerindian component. Something similar might indeed be the case.

eurologist said...

Two quick corrections:

"Mongolian-like" may be better expressed as "Mongoloid."

At the end of the second-to-last paragraph, this should be restricted to N America, not "the Americas."

batman said...

@rokus

You seem to be on to it.

Recent archeological works, such as Terberger 2011, have tried to place the origin of the Hamburg/Ahrensburg culture, that disappeared 12.700 yBP, to resurface 12.300 yBP. In between we had Dryas III, the third and coldest dip in the longer cold period of northern Eurasia; 23-12.000 yrs BP.

The last known mammuts i northern Eurasia survived at the shores of the old curve of Gulf-stream - that seem to have passed todays Friesland/Slesien, behind the (old) islands of Julland and Skane - turning at the bay of Riga to leave the Baltic across the present lowlands of middle Sweden.

This may explain why the oldest traces of arctic humans seem to have survived the coldest dip (Dryas III) at the former islands of the danish Sealand and the swedish Skane.

The funny fact of the matter is that at 12.500 yBP, when Dryas III was at its worst, there have been people at a site called Bromme in Sealand and Hasselberga in Skane. 300 yrs later - as the cold dip started to pass - they were back in the Hamburg/Ahrensburg area, as well.

Looking at the later "spread-sheet" of mesolithic sites we see them spreading north to the very North cape as the Fosna-culture 11.500 BP - at the same time as they spread eastwards to make the Swidrien-Kunda-Volga cultures - as well as south along the rivers to the Black Sea, the Med and the Atlantic facade.

The entire Eurasian mesolitic seems to be a result of this old, arctic refugiants. In that case you should have a pretty clear point of origin concerning the first caucasians - none the less...



Consequently

German Dziebel said...

@Rokus

"why in the Americas if admixture most probably already occurred in Asia?"

because MA-1 has no East Asian component. MA-1 is also far removed from East Asians in PCA plots and in the f3 plots anchored in Sardinian vs. Han (see Olalde EDF 5). MA-1 is a mix of West Eurasians and Amerindians and it's located in Asia, so Amerindians clearly expanded out of the Americas into Asia.

"The other half was not very clear yet."

As clear as the first half: Out-of-America predicts that the closer the sample is to the Americas geographically and the older the sample is, the more likely it's going to be closer to modern Amerindians than to other modern human populations.

"The ADMIXTURE runs of this publication were exclusively designed to show that Amerindians had a second non-East Asian component that could already be recognized in MA-1. "

It's Raghavan's misinterpretation of his own data. What he found was an Amerindian component in MA-1. It's clearly non-modern East Asian and it's clearly non-modern West Eurasian. So, it's Amerindian. As Olalde EDF 5 shows, East Asians, MA-1 and West Eurasians are close to each other when the plot is anchored in Karitiana. It means that modern West Eurasians are Amerindian + "Caucasoids" and modern East Asians are Amerindians + "Mongoloid." Since MA-1 doesn't have affinity to modern East Asians, the "Mongoloid" component was added to what was essentially an Amerindian-derived population in East Asia much later (in the Holocene).

"Please, try to be objective and let it go."

I'm the one who's being objective.

" Somebody has to run ADMIXTURE in the way I told you in my previous comment, such that East Asians could be recognized as a mixed population having an Amerindian component. Something similar might indeed be the case."

It was done more than 10 years ago. See Rosenberg et al. 2002. East Asians are a mix of mostly Amerindian with some African, West Eurasians are a mix of mostly African with some Amerindian. All the STRUCTURE and ADMIXTURE paint essentially the same picture.

"'Choice of an appropriate value for K is a notoriously difficult statistical problem. It seems to us that this choice should be guided by knowledge of a population’s history."

yes, scholars tend to take phylogenies as a starting point and then overlay them with admixture estimates. This is because geneticists are used to think in terms of phylogenies, with admixture studies being a recent theoretical development. But I believe one could start with admixture studies and then deduce a phylogeny on their basis, then test it. E.g., it's clear that all Eurasians have Amerindian affinity, consequently this should be reflected in the way Y-DNA and mtDNA lineages in Eurasia are constructed.

Tobus said...

@agiering:I see the Admixture plots in the Mal'ta papers as supporting "Out of America" for 4 related reasons

You are forgetting that East Asians and Amerindians have a high affinity to each other. If MA-1 received "Amerindian" DNA then it would be highly "East Asian" as well, and MA-1 would be pulled towards both modern Amerindians and East Asians. That fact the MA-1 shows very little East Asian affinity but high Amerindian affinity means that the DNA flow could only have gone the other way - MA-1 into what eventually became Amerindians.

@German: As Olalde EDF 5 shows, East Asians, MA-1 and West Eurasians are close to each other when the plot is anchored in Karitiana.

This is nonsensical interpretation that shows you have no idea how f3 stats are formulated.


agiering said...

@eurologist and @TerryT

"The simple, and obvious, explanation is that the MA-1 population had made it further north that had the 'East Asian' population at that time. "

"Thirdly, Ancient Siberians existed as far back as 45,000 years ago, long before East Asians ventured even remotely as far north. "

Even if East Asian ancestors were more "southern" than Amerindians' ancestors at 24K years ago, explaining Mal'ta's affinities to the latter over the former, I would still expect Mal'ta to be significantly closer to East Asians than the "extreme" southerly Papuans. Yet that does not appear to be the case.

@eurologist
"Secondly, they did not admix with South Asians. Their origin is just old enough that both carry related elements. After all, it is established that almost all non-African maternal haplogroups originated in S and SE Asia, and the paternal haplogroup of importance here (P --> Q, R) as well. So, no surprise - in fact, this was predicted."

Under that explanation, I would expect ALL non-Africans to have more of the "South Asian" component the older they are. Yet that does not appear to be the case with Europeans --- the Lazaridis ADMIXTURE plot shows modern europeans have more of a South Asian component than the Mesolithic/Neolithic samples.

"By archaeological finds, Ancient Siberians expanded into the far East of Siberia and into Beringia ~35,000 ya, and occupied that far northerly area almost unchallenged - except for a coastal expansion of Ancient E Asian northward, with which they apparently admixed. "

And I'm arguing that, while technically not impossible, it's implausible that ANE would have traveled far and wide to contribute DNA to Europeans and Native Americans while failing to contribute to East Asians over the course of the millennia. Especially given that we know that East Asians and ANE did intermix, as shown by Amerindians (according to the scenario you support). East Asia is geographically closer to Mal'ta than Europe or America, after all.

@TerryT

"You're forgetting MA-1 was not found in America, and we don't know anything like it was present in America 24k years ago . "

True, but we do have a 12.5K year old Amerindian skeleton (Anzick) that is very similar to modern Amerindians (especially central/southern) but is no more Han (a modern Asian population) than modern day Amerindians are. Indeed, one of the PCA plots seems to indicate that Anzick boy was a bit less Han-shifted compared to modern Mayans.

agiering said...

@Rokus

"That's not true, why in the Americas if admixture most probably already occurred in Asia? "

Signals of said admixture aren't going to suddenly disappear as soon as the proto-amerindians migrate to America. They are going to gradually diverge from Asians (and ancient Siberians, aka ANE) into a separate population over the course of the millennia via natural selection/genetic drift. That's why older populations of Amerindians (e.g. Anzick) should show stronger signals of the ancestral components (i.e. East Asian and ancient Siberian).

"Moreover, the Asian component of Amerindians was already clear from previous investigation."

I'm confused what "previous investigation" you are referring to. What everyone agrees about is that Amerindians show an affinity to East Asians. What we don't know which direction the gene flow went (East Asians to Amerindians or Amerindians to East Asians).

Basically, here's my point: Pretty much everybody agrees than modern Amerindians are genetically and geographically closer to modern East Asians than any other continental wide population ("race", if you will). It is generally assumed that this is because Amerindians come from East Asia and are descended from the same population in Asia that gave rise to modern East Asians. If this were the case, I would expect two things: A) evidence that East Asians are an older population than Amerindians. This could be established by finding an ancient skeleton in Asia that is genetically more like modern East Asians than modern Amerindians. and B) evidence that modern Amerindians have diverged from an East Asian source; in other words, dna findings from an ancient population in the Americas that appear to be ancestral to modern Amerindians, but with more East Asian affinities.

So far, we have no evidence for either. Mal'ta presented a golden opportunity to validate point A, but failed to, which surprised even the researchers. Anzick failed to support B.

To all those who disagree with me, I ask this question: what hypothetical piece of evidence, if discovered, would falsify the idea that Amerindians derived from East Asians (rather than vice versa)?

terryt said...

"This 'model' explains everything and is 100% consistent with the archaeological, climatic, linguistic, and genetic record".

Absolutely. In fact nothing else makes any sense at all. I notice German has made no attempt to explain the problems with his belief I brought up on the 6th April.

terryt said...

"Out-of-America predicts that the closer the sample is to the Americas geographically and the older the sample is, the more likely it's going to be closer to modern Amerindians than to other modern human populations".

I look forward with interest to you comments on this:

http://dienekes.blogspot.co.nz/2014/04/svante-paabo-talk-at-nih.html

terryt said...

"what hypothetical piece of evidence, if discovered, would falsify the idea that Amerindians derived from East Asians (rather than vice versa)?"

A wholesale revision of the haplogroup phylogeny.

"It is generally assumed that this is because Amerindians come from East Asia and are descended from the same population in Asia that gave rise to modern East Asians".

No. The evidence indicates Amerindians are 'partly' descended from East Asians and partly from a Central Eurasian population as represented by MA-1.

" A) evidence that East Asians are an older population than Amerindians".

We do know that to be so. We know that a significant genetic element of modern East Asians had become dominant in an East Asian population by 35,000 years ago (EDAR370A). That is certainly older than MA-1 and older still than Anzick.

" B) evidence that modern Amerindians have diverged from an East Asian source"

We do have such evidence. The vast majority of Amerindian mt-DNA haplogroups are of East Asian origin.

"Even if East Asian ancestors were more 'southern' than Amerindians' ancestors at 24K years ago, explaining Mal'ta's affinities to the latter over the former, I would still expect Mal'ta to be significantly closer to East Asians than the 'extreme' southerly Papuans. Yet that does not appear to be the case".

It 'does not appear to be the case' because your supposition is wrong. If Mal'ta has no 'East Asian' affinity how on earth would it be 'significantly closer to East Asians' than would any other population?

"it's implausible that ANE would have traveled far and wide to contribute DNA to Europeans and Native Americans while failing to contribute to East Asians"

Why implausible? Basically there is an ecological contiuum from Europe to North America. Most other animals have had no trouble spreading right across the region, so what makes you believe humans not have done so? It seems extremely likely that East Asians developed somewhere near the China/Tibet border region and began to expand from there only around some 10,000 years ago. Early elements of that expansion contributed to the Amerindian genotype of course.

"we know that East Asians and ANE did intermix, as shown by Amerindians"

Yes.

"but we do have a 12.5K year old Amerindian skeleton (Anzick) that is very similar to modern Amerindians (especially central/southern) but is no more Han (a modern Asian population) than modern day Amerindians are".

Showing that Amerindians had become isolated from Eurasian populations by 12,500 years ago (apart from some later, more 'East Asian' intrusions into the far north). But that date is just half as long ago as MA-1.

"findings from an ancient population in the Americas that appear to be ancestral to modern Amerindians, but with more East Asian affinities".

Surely earlier American populations would have far fewer Asian affinities than later ones. In fact:

"Indeed, one of the PCA plots seems to indicate that Anzick boy was a bit less Han-shifted compared to modern Mayans".

Indicating that elements of the East Asian intrusion were not confined just to the far north. In fact don't we know that Apaches, for example, are fairly recent arrivals from further north?

"What everyone agrees about is that Amerindians show an affinity to East Asians. What we don't know which direction the gene flow went"

Very few people are in any doubt as to which direction the flow went. I am yet to understand the motives of the very few who propose an alternative. Perhaps you can enlighten me?

German Dziebel said...

@Tobus

"This is nonsensical interpretation that shows you have no idea how f3 stats are formulated."

With your superior pseudomathematical expertise you should be measuring distance to Heaven. Instead you are wasting your time denying straightforward scientific data that have very simple implications. Why?

@ agiering

"Basically, here's my point: Pretty much everybody agrees than modern Amerindians are genetically and geographically closer to modern East Asians than any other continental wide population ("race", if you will). It is generally assumed that this is because Amerindians come from East Asia and are descended from the same population in Asia that gave rise to modern East Asians. If this were the case, I would expect two things: A) evidence that East Asians are an older population than Amerindians. This could be established by finding an ancient skeleton in Asia that is genetically more like modern East Asians than modern Amerindians. and B) evidence that modern Amerindians have diverged from an East Asian source; in other words, dna findings from an ancient population in the Americas that appear to be ancestral to modern Amerindians, but with more East Asian affinities.

So far, we have no evidence for either. Mal'ta presented a golden opportunity to validate point A, but failed to, which surprised even the researchers. Anzick failed to support B."

This is exactly right. East Asians derive from Amerindians, not the other way around.

eurologist said...

"Under that explanation, I would expect ALL non-Africans to have more of the "South Asian" component the older they are. Yet that does not appear to be the case with Europeans --- the Lazaridis ADMIXTURE plot shows modern europeans have more of a South Asian component than the Mesolithic/Neolithic samples.

And I'm arguing that, while technically not impossible, it's implausible that ANE would have traveled far and wide to contribute DNA to Europeans and Native Americans while failing to contribute to East Asians over the course of the millennia. Especially given that we know that East Asians and ANE did intermix, as shown by Amerindians (according to the scenario you support). East Asia is geographically closer to Mal'ta than Europe or America, after all. "


agiering,

There were numerous W Asian (i.e., S Asian-influenced) migrations and admixtures into/ in Europe, including well-documented ones during the neolithic and Metal Ages, and even later. Their late timing has been very well documented by genetics. Your expectations are not based on but rather contradict widely accessible evidence.

Pretty much every one I talk to finds it most plausible that "ANE" originated somewhere close to the extreme NW of the subcontinent and traveled N from there, W of the Himalayas (for reasons I explained above). From there, it is pretty logical that some people traveled W, others E, when the time and climate was opportune with lush grasslands inhabited by huge herds of migrating mammals. For perspective, do you know that parts of Kazakhstan are in Europe?

Rokus said...

@ageering,
'older populations of Amerindians (e.g. Anzick) should show stronger signals of the ancestral components (i.e. East Asian and ancient Siberian).'

They didn't, what suggests ongoing East Asian influence in the Americas, especially North America.

'Signals of said admixture aren't going to suddenly disappear'
Amerindians certainly aren't homozygous. Their heterozygosity may be reminiscent of their arrival.

'What we don't know which direction the gene flow went'
Since you guys refuse to consider two separate Amerindian components, one MA1-derived and one East Asian derived, you are begging the question.

'what hypothetical piece of evidence, if discovered, would falsify the idea that Amerindians derived from East Asians (rather than vice versa)?'

Since for every answer you could push the hypothetic existence of Amerindians in the Americas back in time, I think the main problem with your idea is the lack of dating. It was already impossible to find Amerindian remains older than 15,000 years, while mongolic features in East Asia are much older.

Grey said...

@agiering

"I would still expect Mal'ta to be significantly closer to East Asians than the "extreme" southerly Papuans. Yet that does not appear to be the case."

Wouldn't layers explain it easily enough i.e. an expansion out of India (or somewhere) that covers Papua to Siberia and then East Asians develop somewhere in between e.g. South China, and expand outwards mostly displacing the older layer in the central region and separating the remainder into a northern part (Siberia) and a southern part (Papua).

Like painting a wall brown and then afterwards painting the middle section blue leaving the top and bottom brown.

(If so you might expect autosomal remnants of the older layer to survive in higher concentrations in the remoter parts of the central region e.g. among some of the non-Han minority populations of China).

Tobus said...

@agiering:
would falsify the idea that Amerindians derived from East Asians (rather than vice versa)

Neither modern East Asians nor modern Amerindians derived from the other - their close affinity indicates that they both derive from a common ancient ancestor, not one from the other. Mal-ta's lopsided affinity at 24kya indicates that they had already diverged (or were just beginning to) at this point in time. We have solid evidence that modern humans were in Eurasia for at least the last 40-50ky, and the archaelogical and genetic evidence both point to the arrival of modern human in America no earlier than 15-20kya. This means that this common ancestor likely lived in Asia, not in America.

German Dziebel said...

@Rokus

"Since you guys refuse to consider two separate Amerindian components, one MA1-derived and one East Asian derived, you are begging the question."

Show me the data that Amerindians have two components. Instead, you've so far provided a conspiracy argument explaining why there's only one and why MA-1 is a mix of Amerindians and West Eurasians.

" It was already impossible to find Amerindian remains older than 15,000 years, while mongolic features in East Asia are much older"

No. East Asian skulls with "Mongoloid' features are younger than Amerindian skulls with similar features.

@Tobus

"We have solid evidence that modern humans were in Eurasia for at least the last 40-50ky, and the archaelogical and genetic evidence both point to the arrival of modern human in America no earlier than 15-20kya. This means that this common ancestor likely lived in Asia, not in America."

This is non sequitur. The fact that the "Amerindian component" found in MA-1 in East Asia in best represented in modern Amerindians, not in modern Siberians or East Asians is direct evidence that the common ancestor of modern Amerindians and modern Eurasians lived in America. Now we know where to dig.

"Mal-ta's lopsided affinity at 24kya indicates that they had already diverged (or were just beginning to) at this point in time."

No, what it shows is that future"Mongoloids" haven't admixed yet with New Word-derived ancient East Asians.

German Dziebel said...

@Tobus

"Mal-ta's lopsided affinity at 24kya indicates that they had already diverged (or were just beginning to) at this point in time."

One other correction: MA-1 shows that Amerindians, but not East Asians, had indeed diverged into a separate population by that time. This is consistent with Tianyuan admixture results that show that modern Amerindians are not as close to Tianyuan as West Eurasians and East Asians (Rasmussen EDF 5f).

http://www.nature.com/nature/journal/v506/n7487/full/nature13025.html.

Olalde EDF 5a is consistent with Rasmussen EDF 5f: East Asians and West Eurasians are closer to each other when the plot is anchored in Karitiana. So, Karitiana and Tianyuan behave similarly in f3 statistics.

Tobus said...

@German:you are wasting your time denying straightforward scientific data that have very simple implications. Why?

You need to understand that there's a difference between data and your personal interpretation of that data. However "simple" these implications might seem to you, if they are inconsistent with the measurements made and/or the mathematics applied to them, then they are not valid. The idea that you can use two f3(X, A; O) stats to measure the distance between the two "X" populations is logically invalid and mathematically impossible. As you yourself acknowledge on another thread, X populations that appear "close" on the f3 plots may in fact be similar in distance from A but in totally opposite directions, and hence not "close" to each other at all.

So contrary to your original statement, we simply can't tell how close East Asians, MA-1 and West Eurasians are when the plot in anchored in Karitiana.

German Dziebel said...

@Tobus

"You need to understand that there's a difference between data and your personal interpretation of that data."

The data is objective, my interpretations are scientific. Your counterarguments haven't passed the test for validity. You need to show how an individual population ends up falling into a wrong continental cluster using f3 statistics.

"The idea that you can use two f3(X, A; O) stats to measure the distance between the two "X" populations is logically invalid and mathematically impossible."

It's perfectly valid and mathematically proven with dozens of populations.

"As you yourself acknowledge on another thread, X populations that appear "close" on the f3 plots may in fact be similar in distance from A but in totally opposite directions, and hence not "close" to each other at all."

As a cryptocreationist, you are just incapable of comprehending and interpreting the data. All the data across multiple measurements is consistent with each other.

"So contrary to your original statement, we simply can't tell how close East Asians, MA-1 and West Eurasians are when the plot in anchored in Karitiana."

I knew you would hate that Tianyuan admixture plot and will try to suppress it.

Unknown said...

Hurumph

Forget the Han in any discussion about ancient East Asians. The Han may be East Asian NOW. But genetically they are South East Asians not East Asians. They were not in East Asia when the Amerinds made the crossing to the Americas, they may not even have been in East Asia when the Amerinds came back.

Forget the Han.

Tobus said...

@German:
It's perfectly valid and mathematically proven with dozens of populations.

Please show me this mathematical proof.



terryt said...

"The Han may be East Asian NOW. But genetically they are South East Asians not East Asians".

Not really. Their Y-DNA is primarily O3 which almost certainly spread (mainly south) from the middle Yellow/Yangtze River basins although NO presumably originated in Se Asia. As a result we can say the male element of the Han probably was in 'East Asia' when humans arrived in America. However it seems obvious those pre-American people had no contact with the people that were to become Amerindians. Those people must have passed by further north than any N or O haplogroups had reached by that time.

On the other hand the Han mt-DNA haplogroups are largely indigenous to the region 'over-run' by the Neolithic Han expansion.

Rokus said...

@German,
'No. East Asian skulls with "Mongoloid' features are younger than Amerindian skulls with similar features.'

The Minatogawa skeleton no I (male) of Okinawa was dated 16,600-18,250 BP. Most probably Liujiang, southern China, is older dan 100,000 years. These fossils may easily be recognized as Mongoloid since their zygomatic bones are situated anterolaterally and face anteriorly. These characteristics have also been found in some reamins from Upper Cave 101, though direct contact from this direction can be excluded: Minatogawa exhibits 'morphological features slightly closer to the Wadjak 1 skull in Java than to the Zhoukoudian Upper Cave and Liujiang skulls in China. This suggests that there were no major human migrations between China and Japanese Islands in Late Pleistocene' (Baba e al., 1998). There is no shred of evidence these mongoloid features came from the north, or from time-traveling Amerindians.

'Show me the data that Amerindians have two components.'

(Sigh). Raghavan arrived at the estimation 'that 14% to 38% of Native American ancestry may originate through gene flow' from populations related to contemporary western Eurasians, well before the diversification of Native American populations in the New World. In the supplement is explained where the remaining part is from: '496 bootstrap replicates supported Karitiana as having 26.1% (7.7-44.4%) ancestry from the MA-1 lineage and the remainder from Han, consistent with the previous analysis.'

It didn't? It did. But you still say it didn't. It did according to scientific research.

'What we don't know which direction the gene flow went'

We do, unless Amerindians split up neatly in East Asians and West Eurasians. But they didn't since both East Asians and West Eurasians have archaic components shared by Amerindians, but that can't derive from Amerindians.

Kristiina said...

At the moment we have three highly interesting ancient genomes: Anzick, Tianyuan and MA-1. Soon we may even have Paabo’s Ust-Ishim genome. Only a South Chinese specimen is missing. Comparison of these ancient genomes would certainly give insights into population movements in Northeast Asia. We might even overcome this never ending battle over the Amerindian origins.

Anyway, in the recent Anzick paper, they provide an admixture analysis (Extended Data Figure 3):
At K3, In Eurasia, Anzick orange is seen in the following populations in the descending order (approximately): Chukchi, Koryak, Yukaghir, Selkup, Khanty, Tundra Nenets, Altaians, Nganassan, Shors, Tuvinian, Mari, Mordovian, Russian, Burusho, Kalash, Buryat, Evenki, Even, Nivkh, Mongolian, Yakut , Dolgan, Ukrainian, Lithuanian, Ukrainian, Estonian, Hungarian, Uygur, Kazakh, Kyrghyz, Uzbekh, Tajik, Bulgarian, Lezgin, Turkmen, Adyghe, Orcadian, French, Pathan, Hazara, Dharkar, Balochi, Sindhi, Brahui, Makrani + other Indian groups, Italian, Basque, Tuscan, Iranian, Armenian, Japanese, Druze.

The least affected are the Bedouins, Palestinians, Papuans, Melanesians and East Asian populations, such as Han, Tujia, She, Dai.

At K11, the most Native American like populations in Eurasia are Selkups and Khantys. I think that this is due to some kind of backflow.

The Papuan component is quite high in India and China, so this should be an indication of a southern movement of people.

Terry, I agree with you on many points, but when you say that “the vast majority of Amerindian mt-DNA haplogroups are of East Asian origin”, you may not be entirely correct. Native Americans lack all southern haplogroups, such as F, M7, M9, M10, E, R(xB). It is possible that haplogroups A, C, D and B, at least in part, derive from the ancient Northeast Asians that did not come from South China. It is interesting that a pure East Asian population, Atayal of Taiwan (only yDNA O) lack all northern haplogroups such as A, C and D and share with Native Americans only haplogroup B.

Tobus, I like your comment ”neither modern East Asians nor modern Amerindians are derived from the other - their close affinity indicates that they both derive from a common ancient ancestor, not one from the other”.

terryt said...

"It is possible that haplogroups A, C, D and B, at least in part, derive from the ancient Northeast Asians that did not come from South China".

I agree totally in the case of A. Haplogroup A looks almost certainly to have arisen somewhere immediately north of Tibet. B, on the other hand, looks to have arisen somewhere in SE Asia or South China, from R. To me that demonstrates that mt-DNA N did not move east from 'Africa' through any southern route, as you point out. But the two haplogroups are still 'East Asian'. Turning to C and D:

"Native Americans lack all southern haplogroups, such as F, M7, M9, M10, E, R(xB)".

Yes, but both C and D also derive from M, which is most definitely a 'southern' haplogroup. I agree that most C haplogroups are northeast Asian but C is part of CZ, and Z is most probably South Asian in origin, or at least somewhere near the India/China border region. This becomes especially likely when we consider that CZ is part of M8, definitely a South China/SE Asian haplogroup. At some time members of the haplogroup must have moved north through East Asia, and so again a specifically 'East Asian' haplogroup.

D is actually just one branch of M80, a Palawan haplogroup. That implies strongly that D's origin too is further south than much of the region where it is now present. It is certainly not a 'West Eurasian' haplogroup in any case.

"It is interesting that a pure East Asian population, Atayal of Taiwan (only yDNA O) lack all northern haplogroups such as A, C and D and share with Native Americans only haplogroup B".

But by no means surprising. In fact the B they share with Native Americans is a very distant relative within the B4 clade. Basically American B can be thought of as B4b2, part of B4b'd'e'j, and Taiwanese B is B4a1, a South Chinese/SE Asian haplogroup.

German Dziebel said...

@Tobus

"Please show me this mathematical proof."

Just take a look at the plots and you'll see that all X populations fall into their respective continental clusters. This is the mathematical proof. what you've provided is pseudomathematical possibilities that all those plots are wrong. Don't set up this fallacy as a standard of proof.

@Rokus

"The Minatogawa skeleton no I (male) of Okinawa was dated 16,600-18,250 BP..."

Those are all "pre-Mongoloid skulls." Out of all the skulls (New World and Old World) Zhoukoudian is the closest to Lagoa Santa, which is a famous pre-Mongoloid skull in Brazil. For "Mongoloid" traits attestation in America vs. East Asia see Peter Brown (1999). "The First Modern East Asians? another Look at Upper Cave 101, Liujiang, and Minatogawa". K. Omoto (ed.) Interdisciplinary Perspectives on the Origins of the Japanese, International Research Center for Japanese Studies.

"Raghavan arrived at the estimation 'that 14% to 38% of Native American ancestry may originate through gene flow' from populations related to contemporary western Eurasians, well before the diversification of Native American populations in the New World. In the supplement is explained where the remaining part is from: '496 bootstrap replicates supported Karitiana as having 26.1% (7.7-44.4%) ancestry from the MA-1 lineage and the remainder from Han, consistent with the previous analysis.'

It didn't? It did. But you still say it didn't. It did according to scientific research."

No, Raghavan misinterpreted the data and provided estimates based on this mistake. It's clear from all the ADMIXTURE runs that MA-1 is an admixed population - minority Amerindian and majority West Eurasian. You don't have a single K level in which Amerindians are a mix of West Eurasians and East Asians. Haploid DNA confirms that there was no gene flow from MA-1 to Amerindians, as Amerindians don't have mtDNA U and Y-DNA R. Raghavan's only argument for the MA-1 to Amerindians gene flow, namely that MA-1 are equally removed from East Asians and Papuans, is irrelevant because Amerindians are divergent from both West Eurasians and East Asians, hence they could contribute genes to MA-1 without East Asians becoming close to MA-1.

"We do, unless Amerindians split up neatly in East Asians and West Eurasians. But they didn't since both East Asians and West Eurasians have archaic components shared by Amerindians, but that can't derive from Amerindians."

What are you talking about? West Eurasians and East Asians share the Amerindian component but they diverged into their respective geographic clusters. This is what a migration in two different directions looks like.

@Kristiina

'Native Americans lack all southern haplogroups, such as F, M7, M9, M10, E, R(xB). It is possible that haplogroups A, C, D and B, at least in part, derive from the ancient Northeast Asians that did not come from South China. It is interesting that a pure East Asian population, Atayal of Taiwan (only yDNA O) lack all northern haplogroups such as A, C and D and share with Native Americans only haplogroup B."

Good observation. Also note that Amerindians are linked to West Asia with hg X, which seems to be the western counterpart of hg B. The Amerindian component is correspondingly attested in West Asia, while Tianyuan was shown to be the closest to Amerindians in Fu et al. and it is hg B mitochondrially.

agiering said...

I don't have a whole lot more to say, since it sounds like we're arguing in circles and nobody looks likely to change their mind anytime soon. But just to clarify/elaborate on a few points:

@ Rokus
"They didn't, what suggests ongoing East Asian influence in the Americas, especially North America."

Well, it sounds like you are agreeing that the Anzick evidence contradicts the idea of a single migration out of Asia into the Americas.

@ Kristiina

"Tobus, I like your comment ”neither modern East Asians nor modern Amerindians are derived from the other - their close affinity indicates that they both derive from a common ancient ancestor, not one from the other”. "

I agree, which is why I summarized prevailing opinion as, "It is generally assumed that this is because Amerindians come from East Asia and are descended from the same population in Asia that gave rise to modern East Asians. "

@ Tobus

"There were numerous W Asian (i.e., S Asian-influenced) migrations and admixtures into/ in Europe, including well-documented ones during the neolithic and Metal Ages, and even later. Their late timing has been very well documented by genetics. Your expectations are not based on but rather contradict widely accessible evidence."

I agree that there were south/west asian influxes of genes during the neolithic. That's exactly what the evidence shows. However, you were arguing that Ma'lta's South Asian affinity was not due to admixture but, in your own words, "almost all non-African maternal haplogroups originated in S and SE Asia, and the paternal haplogroup of importance here (P --> Q, R)", while the Amerindian component in Mal'ta was due to gene flow from the latter to the former. In other words, characteristically "South Asian" genes are old, while Amerindian ones are derivative and did not flow to Old World populations. If that's the case, why do Mesolithic Europeans have more of an Amerindian signal than a South Asian one?

@ German

I'm curious how you'd interpret Kristiina's "good observation." Would you argue that the X/B haplotypes originate from the first Out-of-America migration and the "northern Asian" haplotypes (A, C, and D) from the second?

terryt said...

@ Kristiina:

"It is interesting that a pure East Asian population, Atayal of Taiwan (only yDNA O) lack all northern haplogroups such as A, C and D and share with Native Americans only haplogroup B."

All four haplogroups are present in both Laos and Hainan though:

http://www.biomedcentral.com/1471-2148/11/49/

And:

http://www.biomedcentral.com/1471-2148/11/46

A in both cases is presumably A5, the only A haplogroup with 'southern' representatives. B5a, B5b, B4a, B4b and B4c are present in both regions while Laos also has B6, B4e and B4g. That's quite a collection. Both regions have both M8a and C (probably C7 in both cases) while Hainan has Z, but in a non-Li population. Each region tends to have differing D haplogroups although both have D5. Hainan has D4a while Laos (on the mainland) has D4b, D4e and D4g. I agree several of these haplogroups probably owe their presence in the region to the Mongoloid phenotype's southward movenment during the Neolithic.

@ German:

"Amerindians are linked to West Asia with hg X, which seems to be the western counterpart of hg B".

On what grounds do you claim that to be the case?

"Haploid DNA confirms that there was no gene flow from MA-1 to Amerindians, as Amerindians don't have mtDNA U and Y-DNA R".

You've just shot yourself in the foot and destroyed your own argument. If Amerindians don't have 'mtDNA U and Y-DNA R' how on earth would those haplogroups have suddenly appeared in MA-1? Therefore your position has to be that 'there was no gene flow from Amerindians to MA-1' either.

Tobus said...

@German: Just take a look at the plots and you'll see that all X populations fall into their respective continental clusters. This is the mathematical proof.

I don't know what they taught you in Humanities school, but a "mathematical proof" has to be a) mathematical and b) proof. Simply restating your assumptions doesn't constitute mathematical proof. You need to show me the mathematical process that turns the results of f3(X1, A; O) and f3(X2, A; O) into a measurement of genetic distance between X1 and X2.

Also your statement is wrong, X populations don't always fall into neat continental clusters - Papuans on EDF 5a's X-axis for example cluster with both Europeans and South Asians, yet on the Y-axis cluster with neither. Your "proof" is rife with inconsistencies like this - proving that this kind of interpretation is inherently flawed.

what you've provided is pseudomathematical possibilities that all those plots are wrong.

This is the 3rd time you've called my my proof "pseudomathematical" with no refutation to back it up. If I have made an error in the proof can you please point it out to me - otherwise stop spouting unfounded marketing nonsense.

Also note that these plots are 100% correct, what is wrong is your subjective interpretation of them: eg saying "East Asians and West Eurasians are closer to each other when the plot is anchored in Karitiana" when the Karitiana axis doesn't measure affinity between East Asians and West Eurasians.

Kristiina said...

Terry, haplogroups D and C that stem from M* might just as well have arrived from India, as M is highly frequent and old in India. The coalescence age of haplogroup M has been estimated as 50ky BP, and because it is so old, it allows multiple routes across Asia.

Similarly, that tiny M80 may be an early offshoot of ”pre-D” that just found its way from north to south. It is not at all so clear at the moment.

When you suggest that CZ is most probably from somewhere near the India/China border region, you are making a good observation. All these northern haplogroups that are found in Amerindians, i.e. A, C and D, peak in India/China border region, where in particular Tibeto-Burman languages are spoken, compared to South China. For a reason or other, also ANE(=MA-1) component peaks in this region compared to South China.
In ANE_K4_ADMIXTURE-TEST (https://docs.google.com/spreadsheet/ccc?key=0Ato3EYTdM8lQdC01RENtMGpRc3FHTGZpSUFKR3hHY1E#gid=0), you find the following ANE(=MA-1) percentages:
She (a South Chinese ethnic group speaking a Hmong–Mien She language) ANE 0.0001
Dai (an ethnic group speaking a Daic language) ANE 0.0001
Naxi (an ethnic group speaking a Tibeto-Burman Lolo-Burmese language) ANE 0.02-0.04
Tibeto-Burman Burma ANE 0.03-0.09
Austro Asiatic Santhal (an ethmic group in India speaking an Austroasiatic language) ANE 0.24
Pulliyar (an ethnic group in South India) ANE 0.27

According to Eurogenes K15 population averages 22.3.2014, the following Amerindian (AI) and Siberian (SI) percentages are found in Southeast Asia:
She 0.35 (AI) 12.87 (SI)
Cambodian 0.43 (AI) 1.08 (SI)
Dai 0.57 (AI) 2 (SI)
Lahu 0.83 (AI) 9.22 (SI)
Naxi 0.36 (AI) 24.27 (SI)
T-B Burma 0.9 (AI) 13.19 (SI)
We have again the highest values of Amerindian in Lahu (Lolo-Burmese) speaking Tibeto-Burmans and Tibeto-Burmans in Burma. Instead, the Siberian component has a somewhat different distribution.

In this recent paper (http://www.biomedcentral.com/1471-2148/14/17) they found out that there is an over-representation of haplogroups A4 (8.28%) and C4b1 (12.41%) in the Karen population compared to other Southeast Asian populations in Burma. Karen speak again a Karenic Tibeto-Burman language. Wikipedia tells us that according to Karen legends their ancestors crossed a 'river of running sand. Many Karen think this refers to the Gobi Desert, although they have lived in Burma for centuries. A4 and C4b1 are young haplogroups in Burma, while the oldest are M90, M91 and M49e, followed by B6a1 and M20a.

At the moment, I think that Vajda’s Dené–Caucasian hypothesis is definitely worth a more thoroughly analysis.

Rokus said...

'K. Omoto (ed.) Interdisciplinary Perspectives'

Your answer on evidence that archaic Mongoloid features PREDATE any Amerindian fossil to date is nothing but a lot of gibberish. UC 101 (Zhoukoudian) was not ancestral to Minatogawa, the latter being older and closer related to southern archaic fossils.

'Raghavan misinterpreted the data and provided estimates based on this mistake. It's clear from all the ADMIXTURE runs that MA-1 is an admixed population'

I already explained to you that decomposing MA-1 into modern components is not the same as admixture. Per definition admixture only happens between contemporanous populations, not between future populations. What ADMIXTURE really says is that several modern populations in West, Central and South Eurasia share ancestry and drift.

'This is what a migration in two different directions looks like.'

Then how do you explain for example mongoloid features in East Asia from an Amerindian source, that also featured those same traits, if these same features were already present in archaic East Asians? You never did.

I'm sorry your radical version of Out of America left the pretention of being a thought experiment. Now it didn't survive genetic evidence and applies rhetoric against the most basic principles, it should stop.

German Dziebel said...

@Tobus

"I don't know what they taught you in Humanities school, but a "mathematical proof" has to be a) mathematical and b) proof. Simply restating your assumptions doesn't constitute mathematical proof. You need to show me the mathematical process that turns the results of f3(X1, A; O) and f3(X2, A; O) into a measurement of genetic distance between X1 and X2."

As I already asked you, if you don't trust my judgment, you're most welcome to write to a geneticist you trust and ask for a detailed explanation of why the data points plotted together shouldn't be interpreted as measuring genetic affinity between X populations. My inference from f3 plots that LB has Amerindian affinity was confirmed by f4 measurements. If you want your mathematical skills to be put to a good cause, devise a f10 formula and report the results.

"Papuans on EDF 5a's X-axis for example cluster with both Europeans and South Asians, yet on the Y-axis cluster with neither."

Well, that's kind of the point. Papuans, Europeans and South Asians have the same amount of Amerindian ancestry, but they occupy different geographic clusters, which are rather neatly ranked from north (European) to the very south (Papuans). That's evolution, my creationist friend - same descent, different modifications.

"Your "proof" is rife with inconsistencies like this - proving that this kind of interpretation is inherently flawed."

You can keep doing your bad publicity stunts, but the reality is harsh.

"This is the 3rd time you've called my my proof "pseudomathematical" with no refutation to back it up."

You haven't made mathematics relevant to the problem beyond what has already been done with f3 and 4 statistics. Instead, you're trying to impose a standard of mathematical proof where it doesn't belong. You need to create a better mathematical formula to run more complex f-stats and test smaller-variable calculations against larger-variable calculations. If you show me that a f10 calculation generates a different picture from a f3 calculation, I may give you a "God is Math" bumper sticker.

"East Asians and West Eurasians are closer to each other when the plot is anchored in Karitiana" when the Karitiana axis doesn't measure affinity between East Asians and West Eurasians."

You are misguided on the level of theory. But this aside, my inference that East Asians and Europeans are close to each other when compared with Amerindians is supported by a Tianyuan-anchored plot, which shows that East Asians and West Eurasians behave exactly in the same way against a 40,000 year old sample. All of this supports the notion that Amerindians have been isolated from Old World populations for at least 40,000 years, while Europeans and East Asians have been separated from each for much less time. (This is consistent with paleobiology that shows that Amerindians predate the European-East Asian split - http://onlinelibrary.wiley.com/doi/10.1002/ajpa.21425/abstract - sorry, if your "computer" freezes from the amount of information I'm asking it to process.) Coupled with the fact that all West Eurasians are Amerindian shifted, gives you out-of-America plain and simple.

"otherwise stop spouting unfounded marketing nonsense."

FYI: there's a difference between sales and marketing. If what I'm doing is marketing, what you are doing is sales. But the whole metaphor seems forced, so drop it.

German Dziebel said...

@Rokus

"Your answer on evidence that archaic Mongoloid features PREDATE any Amerindian fossil to date is nothing but a lot of gibberish."

Just read the source I cited first.

"UC 101 (Zhoukoudian) was not ancestral to Minatogawa, the latter being older and closer related to southern archaic fossils."

Did I say it was? I said Zhoukoudian's closest morphological affinities are with Lagoa Santa in Brazil. Lagoa Santa is a different skull from Minatogawa, if you care to know about it.

"Then how do you explain for example mongoloid features in East Asia from an Amerindian source, that also featured those same traits, if these same features were already present in archaic East Asians? You never did."

"Mongoloid" crania are very recent in East Asia, less than 10,000 years. This is general consensus, which I don't see as flawed in any way. There are some features such as dental shoveling that are spread widely around Eurasia and tend to occur at higher frequencies in some West Eurasian samples (Catalhoyuk, Gravettian, etc.) but those are precisely the traits that are at their highest in Amerindians. They may be plesiomorphic, although Neandertal shoveling is different from modern human shoveling. The progressive dilution of shoveling as one goes into the Sahul and West Eurasia is consistent with a Late Pleistocene migration out of the Americas that replenished this trait in the areas that are geographically the closest to the Americas.

"I'm sorry your radical version of Out of America left the pretention of being a thought experiment. Now it didn't survive genetic evidence and applies rhetoric against the most basic principles, it should stop."

On the contrary, all the most recent genome-wide and ancient DNA data consistent with/supports out-of-America. We both agree that it also disproves out of Africa. You want to have the cake and eat it, too, it seems.

German Dziebel said...

@agierung

"I'm curious how you'd interpret Kristiina's "good observation." Would you argue that the X/B haplotypes originate from the first Out-of-America migration and the "northern Asian" haplotypes (A, C, and D) from the second?"

This would be a useful heuristic. I have issues with the way mtDNA haplotypes are defined on the sequence level (hot spots shouldn't be used as branch-defining mutations, ancestral states should be determined by comparison with Neandertal and Denisova sequences) but there does seem to be a division between more broadly distributed Amerindian-like mtDNA lineages that are NOT found in northeast Asia (hgs B and X), northeast Asia-attested lineages (A and D) and a lineage that's found across Northern Eurasia from Iceland through Karelia to the Amur Basin (C1). Hgs X and B are also not common in northern North America, while A and D are. So, yes, considering that modern (north) East Asians gravitate toward northern Amerindians in genomewide studies and hgs A, C and D are geographically restricted to northeast Asia and peter away as one goes south in Asia and west in Eurasia, it's likely that these are the mtDNA signs of a recent back migration from the Americas.

Rokus said...

@German,
'"UC 101 (Zhoukoudian) was not ancestral to Minatogawa, the latter being older and closer related to southern archaic fossils."'

'Did I say it was?'

I would say the origin of Mongoloid features is essential if you want to prove they are Amerindian. But obviously your sense of logic does not work the same way.

'"Mongoloid" crania are very recent in East Asia, less than 10,000 years.'

No, they are not. Only the contemporaneous mongoloid "packages" if this is what you mean. But the whole package of mongoloid features could never have originated in the Americas now at least some of those features already existed in East Asia long before. Moreover, this makes the neat split that you propose of Amerindians into West Eurasians, that lack these mongoloid features, and East Eurasians, that are continuous to mongolid features, completely impossible.

'On the contrary, all the most recent genome-wide and ancient DNA data consistent with/supports out-of-America. We both agree that it also disproves out of Africa.'

Disproving Out of Africa supports all non-African scenarios, not just one extremist cause. Specific requirements for Out of America are not met, and especially failed regarding evidence on a dual Native American origin. Indeed, probably you just want to have too much of the cake, while actually you should settle with the crumbs: some peripheric role of Native American gene flow at most.

agiering said...

@German

"On the contrary, all the most recent genome-wide and ancient DNA data consistent with/supports out-of-America. "

It seems to me that ancient DNA is the most valuable tool in understanding human evolution, more so than hypothetical family trees constructed from modern DNA or early human anatomy. I think people are really struggling to comes to terms with the fact that what ancient dna we have so far doesn't seem to lend much support for our pre-conceived theories.

"a lineage that's found across Northern Eurasia from Iceland through Karelia to the Amur Basin (C1)"

mtDNA C seems like a weird case here. It's prevalence in Iceland reminds me of the relatively high Amerindian signal in Northern Europeans and Scandinavian hunter-gathers compared to southern Europeans and non-scandinavian mesolithic europeans. I wonder if this is all part of the same pattern.

Tobus said...

@German:
As I already asked you, if you don't trust my judgment, you're most welcome to write to a geneticist you trust and ask for a detailed explanation of why the data points plotted together shouldn't be interpreted as measuring genetic affinity between X populations.

In other words, you have no proof - it's just something you made up.

I don't need to ask a geneticist, I have a solid understanding of the maths and can see for myself that there's no measurement between the X populations here. If *your* education is lacking in this area and *you* are having trouble understanding the maths, then maybe *you* should ask someone who knows what they're talking about - didn't you say you've got friends in the Reich lab or something once?

Papuans, Europeans and South Asians have the same amount of Amerindian ancestry, but they occupy different geographic clusters, which are rather neatly ranked from north (European) to the very south (Papuans).

Then the same goes with EDF 5d - LB and Amerindians have the same amount of MA-1 ancestry, but they occupy different geographic clusters, LB with Europeans at the top and Amerindians near East Asians and Papuans at the bottom.

I can see you are finally getting some grasp on what these plots represent - now you just need to apply your understanding to ALL of the charts, not just the ones that suit you.

Instead, you're trying to impose a standard of mathematical proof where it doesn't belong.

You misunderstand. I'm simply looking under the hood at how the f3 stats are calculated: There is no comparison of X1 vs X2 there, and it's possible (as you just confirmed with the Papuan/European sample) that two X populations can get the same score without necessarily having any direct affinity with each other.

Tianyuan

As usual your conclusions are totally wrong, but this is an attempt to change the subject - the point here is that you can't assume genetic affinity between the different X populations on an f3(X, A; O) plot... let's just get that sorted first.

But the whole metaphor seems forced, so drop it.

I'll drop it when you start using actual proof (science!) instead of unsupported aspersions (marketing!) - like when you show me the fault in my maths instead of just calling it "pseudomathematical".


German Dziebel said...

@agierung

"Now I'm just downright confused. I've heard it so many times that the LACK of genetic diversity amongst Amerindians is proof of their recency and the high amount of genetic diversity amongst Africans is proof of their status as an archaic ancestral population. Out-of-Africa critics like German have been saying for some time now that genetic diversity is not a good marker for age; things like admixture can at least as well explain genetic diversity.

Now you are turning things on their head by stating that Amerindians are TOO diverse to be ancestral? And that the only explanation for their genetic diversity is that they are recent and admixed? You do realize that the exact same argument could be used against the out of Africa model, don't you?"

It's an excellent observation and check out Terry's response:

"Amerindians have far less diversity than do Africans."

"German has actually been claiming the supposed great diversity within Amerindians 'proves' they are an ancirent population."

"things like admixture can at least as well explain genetic diversity".

That is what Rokus, Tobus and I have been trying to point out to German."

Clearly, Terry is completely confused.

terryt said...

@ German:

"I have issues with the way mtDNA haplotypes are defined on the sequence level"

Of course you do. Haplogroups provide the most convincing evidence against your belief. You have to 'have issues with' them or you would be forced to admit you are totally wrong, and certifiably mad as well.

@ Kristiina:

"haplogroups D and C that stem from M* might just as well have arrived from India, as M is highly frequent and old in India. The coalescence age of haplogroup M has been estimated as 50ky BP, and because it is so old, it allows multiple routes across Asia".

The 'age' would do, but the 'distribution' of M haplogroups argues against any route but via East Asia, presumably form the India/China border region. West Eurasia is remarkably free of any M haplogroups except for some M1 and several northern haplogroups that are very prominent in East Eurasia.

"that tiny M80 may be an early offshoot of 'pre-D' that just found its way from north to south. It is not at all so clear at the moment".

Once more what you say is possible, and I agree that M80 is a 'pre-D'. But it is pretty much confined to Palawan and so presumably arrived there very early in the piece. There is no obvious route to Palwan from the north. On the other hand many D haplogroups are very common in South China and SE Asia, and that includes D5 and D6, not just D4.

"When you suggest that CZ is most probably from somewhere near the India/China border region, you are making a good observation".

A few M haplogroups look specifically 'South Asian' but almost all the others fit an origin somewhere near the India/China border region.

"All these northern haplogroups that are found in Amerindians, i.e. A, C and D, peak in India/China border region, where in particular Tibeto-Burman languages are spoken, compared to South China".

Which argues against any significant ancient 'coastal migration' from, or to, anywhere.

"In this recent paper (http://www.biomedcentral.com/1471-2148/14/17) they found out that there is an over-representation of haplogroups A4 (8.28%) and C4b1 (12.41%) in the Karen population"

An extremely interesting paper, and one that German should read. Some of those oldest haplogroups you mention have been found only in Myanmar, and recently at that.

Kristiina said...

Terry, in that Laos paper they say that ”Haplogroup C, in contrast, showed an unexpectedly high frequency of 6%, but very limited diversity: all haplotypes belonged to haplogroup C7. Interestingly, 12 of the 13 samples derive from Northern provinces. Together with the singular presence of haplogroup M8, this is indicative of a founder effect - possibly the immigration of a small group of females carrying the C7 haplogroup from its Northern pool.”

In that Hainan mtDNA paper, the oldest coalescence ages are given to R9b and M7e, followed by M12, B5a, B4b1, M7b1 and F1a1. All northern mtDNA lines such as C, Z, M8a and different D and A haplotypes should therefore be relatively young haplogroups in Hainan.

By contrast, In ”Genetic evidence of Paleolithic colonization and Neolithic expansion of modern humans on the Tibetan Plateau”, 2013, A, C and D are given old coalescence ages in Tibet:
A11 and A4 Last Glacial Period
D4, D5a2a, D6a Upper Palaeolithic
C4a1 Upper Palaeolithic
C4a2’3’4 and Z, Last Glacial Period
Interestingly, the oldest coalescence age in Tibet is given to D5a2a.

In China, the oldest coalescence ages are given to R9, B, F1 and N9 (over 60kya ± 20kya). In the next group we have G2, CZ, B4, M8, M7 and D5 (50-60 kya ± c. 18kya). Haplogroup A’s age is estimated only as 29 kya ± 14kya.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC384943/

We know that there were people moving in Northern Eurasia in the Upper Palaeolithic and they reached China. It remains open to debate what haplogroups they were carrying.

German Dziebel said...

@Tobus

"In other words, you have no proof - it's just something you made up."

No, you made up your objection, so I want you to bolster it with someone with better credentials than yours. I consider them plain nonsense but you're welcome to try.

"I don't need to ask a geneticist, I have a solid understanding of the maths and can see for myself that there's no measurement between the X populations here."

The measurements are perfectly solid and direct. But you can work on further improving them.

"I can see you are finally getting some grasp on what these plots represent - now you just need to apply your understanding to ALL of the charts, not just the ones that suit you."

More baloney. I'm educating you on how to make historical inferences from the charts but your cryptocreationist bias prevents you from accepting scientific evidence.

"Then the same goes with EDF 5d - LB and Amerindians have the same amount of MA-1 ancestry, but they occupy different geographic clusters, LB with Europeans at the top and Amerindians near East Asians and Papuans at the bottom."

No. MA-1 already has Amerindian affinity as EDF 5a and e attest. So, it's not that Amerindians have MA-1 ancestry. It's MA-1 and LB have Amerindian ancestry.

"You misunderstand. I'm simply looking under the hood at how the f3 stats are calculated: There is no comparison of X1 vs X2 there, and it's possible (as you just confirmed with the Papuan/European sample) that two X populations can get the same score without necessarily having any direct affinity with each other."

There's direct affinity. All those populations are related, all of them have Amerindian ancestry, they just live in different places. But like I said you can devise a better formula to challenge it. Your verbal assault is not grounded in any better maths.

"As usual your conclusions are totally wrong, but this is an attempt to change the subject - the point here is that you can't assume genetic affinity between the different X populations on an f3(X, A; O) plot... let's just get that sorted first."

It'e been sorted already. You are either in or your out. I assume you are out.

"I'll drop it when you start using actual proof (science!) instead of unsupported aspersions (marketing!) - like when you show me the fault in my maths instead of just calling it "pseudomathematical"."

You haven't provided any maths. You are trying to undermine the existing maths and its logical implications. Hence, it's "pseudomathematics." And a sales pitch, not a smart marketing of your ideas.

terryt said...

@ agiering:

"I think people are really struggling to comes to terms with the fact that what ancient dna we have so far doesn't seem to lend much support for our pre-conceived theories".

I don't 'think', I know, that you are completely wrong with that assertion. Ancient DNA has consistently supported what I and many others have been suggesting for many years to be the case. What's more ancient DNA has consistently supported the 'hypothetical family trees constructed from modern DNA'.

@ German:

"Clearly, Terry is completely confused".

These days you are consistently confirming my impression of you as a complete idiot. Check those quotes again and see where I've been inconsistent. The only remark I made concerning any great diversity in Amerindians is where I quoted you. The single aspect you have actually been consistent over is ignoring any comment pointing out the vast amount of evidence you are having to ignore in order to maintain your belief.

terryt said...

@ Kristiina:

"Haplogroup C, in contrast, showed an unexpectedly high frequency of 6% [in Laos], but very limited diversity: all haplotypes belonged to haplogroup C7. Interestingly, 12 of the 13 samples derive from Northern provinces. Together with the singular presence of haplogroup M8, this is indicative of a founder effect - possibly the immigration of a small group of females carrying the C7 haplogroup from its Northern pool".

Which fits completely my comment elsewhere that several haplogroups had later moved south with the Mongoloid phenotype's Neolithic expansion.

"In that Hainan mtDNA paper, the oldest coalescence ages are given to R9b and M7e"

Yes. Hainan was settled relatively late. But it gives an indication of the haplogroups available on the mainland for that settlement.

"By contrast, In 'Genetic evidence of Paleolithic colonization and Neolithic expansion of modern humans on the Tibetan Plateau', 2013, A, C and D are given old coalescence ages in Tibet"

Another very good paper. However it doesn't really show the deeper origin for those three haplogroups. And some may have moved north into Tibet from near the India/China border region. For example:

"Interestingly, the oldest coalescence age in Tibet is given to D5a2a".

Very much a downstream haplogroup indicating various D haplogroups had moved around considerably before any surviving member of the haplogroup was able to enter Tibet. In my notes I have written that both D5a'b and D5c are, 'especially common in North Asia' but also present in Malaya and spread thinly from Vietnam to Japan and Mongolia.

"In China, the oldest coalescence ages are given to R9, B, F1 and N9 (over 60kya ± 20kya)".

Makes complete sense. Some sort of N must have reached SE Asia by that time because several basal N haplogroups along with R-derived P are present in Australia, and are almost certainly early there, probably first there.

"Haplogroup A’s age is estimated only as 29 kya ± 14kya."

That is when it coalesced as the A we know today. Presumably it didn't suddenly coalesce from nothing. Almost certainly its ancestral form was present in northern Tibet/NW China long before its actual coalescence time.

"We know that there were people moving in Northern Eurasia in the Upper Palaeolithic and they reached China. It remains open to debate what haplogroups they were carrying".

MA-1 has largely solved that problem. We know that the Upper Paleolithic was introduced into Northern China from further west. Consequently it has become reasonable to assume that the first haplogroups into Northern Eurasia were Central/West Eurasian haplogroups lacking what has come to be considered the 'typical' East Asian phenotype. That suggests Y-DNAs P and Q along with mt-DNAs U and X. The East Asian phenotype's movement further north followed that West Eurasian population's establishment, and almost certainly consisted of Y-DNAs C2 and N along with mt-DNAs A, B, C and D. Through most of the region these East Asian haplogroups have subsequently largely replaced the West Eurasian haplogroups in Northeast Asia.

Kristiina said...

Agiering, when you say that you ”wonder if this is all part of the same pattern”, I must say that I have been wondering the same since I received a year ago my DNA and admixture results which said that I have 2% of Native American. I am 100% Finnish, and 23andme keeps reporting that I have several 4th cousins among people from Northern America with Native American ancestry. My Native American sequence must therefore be quite long and with all probability not a result of an admixture as far back as 15000 years ago but instead a more recent event.

Tobus said...

@agierung:
Now I'm just downright confused.

It's because there are two measures of "diversity" being used. When comparing overall genetic diversity of global populations (Amerindians vs Africans for instance), Amerindians score the lowest, but when comparing sub-groups within each population (Maya vs Apache, Zulu vs Igbo) Amerindians score the highest. This means the diversity in Africa is large and widespead (lots of old populations interacting with each other), while the diversity in America is small and scattered (recent bottleneck and subsequent expansion).

When people talk about "high diversity" in America, they mean the relatively large differences between individual American sub-populations, and when they talk about Amerindian "low diversity" they mean the relatively small amount of overall genetic variation.

agiering said...

@ Kristiina

As I said before, Mesolithic foragers have an Amerindian genetic signal, especially Scandinavian ones, so there must be something ancient going on. Of course that doesn't preclude the possibility of a recent admixture event as well, but Finland isn't a country that would likely have received much recent American gene flow. Do you know if these North American cousins of yours know they have recent Native American ancestors?

@Tobus

You are giving TerryT too much credit for clarity/consistency. He has gone on the record as saying: "Amerindians have far less diversity than do Africans"
AND
"The problem with his whole idea is that he [German] can offer no explanation for the apparent genetic diversity noted in America. The generally accepted scenario, on the other hand, easily explains that diversity: two Eurasian populations formed a hybrid population in northeast Eurasia, thus making the combined population more genetically diverse than either source population. "

@ TerryT

"I don't 'think', I know, that you are completely wrong with that assertion. Ancient DNA has consistently supported what I and many others have been suggesting for many years to be the case. "

At least the researchers behind the Mal'ta findings were honest enough to admit they were surprised by the results.

German Dziebel said...

@agierung via @Tobus

"This means the diversity in Africa is large and widespead (lots of old populations interacting with each other), while the diversity in America is small and scattered (recent bottleneck and subsequent expansion)."

Tobus is right in his definition of the two kinds of diversity measures being used. This is because I explained it to him. He's wrong in his interpretation of what it means. African populations are not old, some of their genetic lineages are. They've been interacting with African archaics, with each other and with Eurasians a lot, and gene flow drives INTRA-group diversity up and INTER-group diversity down. Low heterozygosity (intragroup diversity) in Amerindians doesn't mean they went through a recent bottleneck. The bottleneck could just have been long time ago but their demographic structure has maintained low heterozygosity for a long time. Denisovan and Neandertal diversity profile is more similar to Amerindians than to Africans suggesting that the African profile is derived from recent admixture, while the Amerindian profile is derived from continuity with Mid-Pleistocene populations.

But since 1492 America has arguably the highest intragroup diversity values because of post-Columbian admixture between several source populations. So, if a geneticist of the future looked at the New World, he would've concluded that the Old World was peopled from the New World. He would've been right but for a wrong reason.

Tobus said...

@German:
No, you made up your objection

So you *DO* have proof? ...where is it?

The measurements are perfectly solid and direct.

That they are - but all are measurements between X and A... there's been no measurement between the X populations, they're plotted solely on their measurement to A on each axis.

No. MA-1 already has Amerindian affinity as EDF 5a and e attest.

Sorry German, that's a non sequitur. If Papuans and Europeans get the same score on the Karitiana axis but occupy different continental clusters on the Sardinian axis, why can that not apply to LB and Amerindians on the MA-1/Sardinian plot?

Your verbal assault is not grounded in any better maths.

It's grounded in a much better understanding of the maths. You are claiming that the maths is measuring something that is simply isn't. The f3 formula (O-X)(A-A) counts allele frequency patterns between the X, A and O populations - there's no way to measure the genetic distance between the two X populations used in two different f3 calcs. If you believe otherwise then please show me how it's done.

You haven't provided any maths.

What? You must have missed that part:

f3 is essentially Sum((O-X)(O-A)). If we take two alleles both at 0% in O and fixed in A and run the f3 with two populations both equally related to A but in different parts of the genome - X1 fixed in allele 1 but missing allele 2, and X2 fixed in allele 2 but missing allele 1, we get:

f3(X1, A; 0) = (0-1)(0-1) + (0-0)(0-1) = 1
f3(X2, A; 0) = (0-0)(0-1) + (0-1)(0-1) = 1

To summarise: In the panel used X1 and X2 both share 50% of A's DNA, but are totally unrelated and share 0% of each other's DNA, yet the X1 and X2 f3 stats both come up with the same answer. This proves that comparison of the f3 scores of the various X populations is not necessarily indicative of genetic affinity between those X populations.

Europeans and East Asians being "close to each other" as X populations on a particular axis do not equate to them having increased genetic affinity. If you don't understand why, ask an expert to work through it with you, but don't just label it as "pseudo" because it's beyond your comprehension level.

Kristiina said...

Terry, I somewhat disagree with you on your end remark about U's or X's role in Northeast Asia. Now Pääbo is about to show that they were modern humans in Eurasia before the Upper Paleolithic revolution and well before MA-1. Tianyuan is clearly older than MA-1, and when you read the Anzick paper you see that Tianyuan was equally close to Europeans as he was to Southeast Asians. Skeletal finds in China also report European affinities in ancient specimens. On the basis of this, it seems that people reached China ALSO from the north, and probably on multiple occasions. The first arrivals may have carried pre-N9, if you like, or even B, as B is Tianyuan’s haplogroup and B could be related to U on a very deep level (someone who has more thorough understanding, could comment this possibility!). It is a pity that we do not know Tianyuan’s yDNA. Perhaps it was C*.

I agree with you that CZ probably originated somewhere in East India/Tibet/Sichuan areas. In another Tibetan mtDNA paper ”A mitochondrial revelation of early human migrations to the Tibetan Plateau before and after the last glacial maximum”, 2010, haplogroup D is not given such an old date in Tibet, but C4d and A10 (probably the same as A11 in the other paper) are considered as c. 22kya and 26 kya old. As hg A is derived directly from N, it could be related to hg X.

agiering said...

"But since 1492 America has arguably the highest intragroup diversity values because of post-Columbian admixture between several source populations"

How has intergroup diversity in America changed since 1492?

German Dziebel said...

@Tobus

"So you *DO* have proof? ...where is it?"

Dozens of populations correctly plotted into their respective continental clusters.

"That they are - but all are measurements between X and A... there's been no measurement between the X populations, they're plotted solely on their measurement to A on each axis."

And what we can infer from this is that X1 has more or less of the A ancestry than X2 and by how much. This measures the distance between them. Plus we can see how far each one of them are from a population cluster from which the anchor is derived. That's how it was established that MA-1 is the closest to Amerindians out of all modern human populations. An f4 confirmed an inference from f3 when it came to LB's Amerindian affinity. You can invent a better formula and test it against dozens

"If Papuans and Europeans get the same score on the Karitiana axis but occupy different continental clusters on the Sardinian axis, why can that not apply to LB and Amerindians on the MA-1/Sardinian plot?"

Amerindians show much greater divergence on the X axis than any other population, including those that have high values on the Y axis. MA-1 falls neatly between Amerindians and Europeans (in close proximity to East Asians consistently with the fact that the sample comes from east Asia) in EDF 5a. We know that MA-1 didn't descend from heaven, but is closest to Amerindians out of all the modern human populations. Correspondingly, Amerindians maintain their rightmost position on a plot/axis anchored in MA-1. So, for all practical purposes MA-1 is an ancient Amerindian-derived population shifted toward Europeans. It's more parsimonious to think this way and not invent a ghost one-population-strong ANE cluster represented by MA-1 only that gave rise to both West Eurasians and Amerindians. The inference is consistent with f4 runs showing that ancient West Eurasians have Amerindian affinity (with or without MA-1). This wouldn't be the case if MA-1 was ancestral to both Amerindians and West Eurasians.

"It's grounded in a much better understanding of the maths. You are claiming that the maths is measuring something that is simply isn't. The f3 formula (O-X)(A-A) counts allele frequency patterns between the X, A and O populations - there's no way to measure the genetic distance between the two X populations used in two different f3 calcs. If you believe otherwise then please show me how it's done."

Like I said, you can't have a single formula that measures relationships between all populations at once. But you can try and invent it. In the meantime, every f3 run generates values that can be directly compared to the values generated in other runs.

"What? You must have missed that part: Europeans and East Asians being "close to each other" as X populations on a particular axis do not equate to them having increased genetic affinity. If you don't understand why, ask an expert to work through it with you, but don't just label it as "pseudo" because it's beyond your comprehension level."

I'm happy you are proud of your high school math expertise, but all of your calculations are irrelevant until you provide evidence that this creates noise in the data. You need to prove that your possibility is actually not a possibility but a reality for some populations with known geographic clusters. You remain a "pseudomathematician" because you continue to misapply maths instead of evolving it to better capture the complexity of the data. In the meantime, the fact remains that Europeans and East Asians are close to each other when they are compared to Amerindians, and Amerindians show great divergence from them. This fits perfectly with the rest of the data (including Tiyanyuan runs) at hand.

terryt said...

"Now Pääbo is about to show that they were modern humans in Eurasia before the Upper Paleolithic revolution and well before MA-1".

I have long accepted that to be the case. In fact my major disagreements with Maju were over that matter. We are yet to see what mt- and Y- DNA lines the individual has.

"Tianyuan was equally close to Europeans as he was to Southeast Asians".

The mt-DNA was B which is not a European haplogroup, but even at that time the individual could have been of mixed East and West Eurasian populations.

"it seems that people reached China ALSO from the north, and probably on multiple occasions".

Yes, multiple occasions.

"As hg A is derived directly from N, it could be related to hg X".

Absolutely. I am sure the northern N haplogroups are remnant survivals from an early west to east expansion north of the Tibetan Plateau.

"The first arrivals may have carried pre-N9, if you like, or even B, as B is Tianyuan’s haplogroup and B could be related to U on a very deep level"

They are both members of R, but any particularly close relationship would indicate R too had moved east to the north of the Tibetan Plateau. Possible.

@ agiering:

"At least the researchers behind the Mal'ta findings were honest enough to admit they were surprised by the results".

If I remember correctly they were mainly surprised at the presence of Y-DNA R as it is not very common in the region today. They were probably expecting Q, a close relation, or an eastern haplogroup. As far as I'm aware they were not at all surprised at its American or West Eurasian affinity.

"You are giving TerryT too much credit for clarity/consistency. He has gone on the record as saying: 'Amerindians have far less diversity than do Africans'"

I see that you couldn't understand Tobus' explanation.

Tobus said...

@aigerung via @German:
He's wrong in his interpretation of what it means.

It's not my interpretation, it's from Tishkoff et. al. 2009 and is widely accepted as fact by the scientific community.

@German:
Dozens of populations correctly plotted into their respective continental clusters.

That's not proof because these same dozens of populations get *incorrectly* plotted together in often conflicting arrangements with different continental clusters on different axes (eg Papuans and Europeans as pointed out numerous times). What you need to show is that there is a consistent relationship between the f3 scores of various X populations and their genetic affinity.

And what we can infer from this is that X1 has more or less of the A ancestry than X2 and by how much. This measures the distance between them.

Then please show me the formula/working of how you calculate this genetic distance between X1 and X2 based on their relative distances from A, and please tell me what the resulting figure represents - is it f3(X1, X2; O)... or what exactly?

Amerindians show much greater divergence on the X axis than any other population....

Another non-sequitur. Perhaps you didn't understand the question - you've proposed that the contradictory position of Papuans on the 5a axes is because they occupy a different "continental cluster" on the Y axis to the X axis (same net difference, but different direction). Why can this same logic not also be used to explain the contradictory position of LB on the EDF 5d axes, since on that chart he too occupies a different "continental cluster" on that Y axis to the X axis?

We know that MA-1 didn't descend from heaven, but is closest to Amerindians out of all the modern human populations.

This is not correct. East Asians are closer to Amerindians than MA-1 is. I believe I've had to correct you on this before.

You need to prove that your possibility is actually not a possibility but a reality for some populations with known geographic clusters

You're going round in circles German - I've already shown this with Papuans and Europeans plotting as a single cluster on the Karitiana axis (plus a host of other examples on various axes). Thus we have both mathematical proof and real world examples showing that the positions of X populations relative to each other is not a consistent indicator of affinity between those X populations... ready to accept it yet, or do you have more hoops for me to jump through in lieu of providing some proof of your own?

In the meantime, the fact remains that Europeans and East Asians are close to each other when they are compared to Amerindians

The fact is that Europeans and East Asians are the same distance from each other whatever other population you compare them with - their genetic relationship doesn't change no matter how the pretty coloured dots on the f3 plots move. Something I think you'll see if you try to define this measurement of "closeness" in consistent, concrete and meaningful terms.

German Dziebel said...

@Tobus

"That's not proof because these same dozens of populations get *incorrectly* plotted together in often conflicting arrangements with different continental clusters on different axes (eg Papuans and Europeans as pointed out numerous times). What you need to show is that there is a consistent relationship between the f3 scores of various X populations and their genetic affinity."

There's nothing incorrect about their placements. As I pointed out to you many times, your example of papuans and Europeans has a perfect explanation: both groups share the same amount of Amerindian ancestry. In fact it's a good example of how the plots "function" outside of the immediate continental clusters. You interpret them as "incorrect" placement because you can't imagine that continental clusters share genetic ancestry. Since there're no individual populations that cluster outside of their known continental affiliations proves that continental clusters follow the same logic. Individual populations are obviously closer to their respective continental neighbors than continental clusters to each other because more time has elapsed in the latter case.

"Then please show me the formula/working of how you calculate this genetic distance between X1 and X2 based on their relative distances from A, and please tell me what the resulting figure represents - is it f3(X1, X2; O)... or what exactly?"

It's called f4. You can calculate it yourself. Maybe this would make you busy. You haven't brought any value (pun intended) so far. All that matters right now is that East Asians, MA-1 and West Eurasians are all close to each other genetically in terms of their shared Amerindian ancestry when Yoruba is used as an outgroup.

"Another non-sequitur."

Amerindians are the most divergent population in those plots. Consistently, all Eurasians have a certain degree of Amerindian ancestry.

" Perhaps you didn't understand the question."

yes, I did. Re-read my answer. But maybe in addition to your cryptocreationist beliefs you also believes in UFOs who landed in South Siberia and contributed genes to LB and Amerindians.

"Why can this same logic not also be used to explain the contradictory position of LB on the EDF 5d axes, since on that chart he too occupies a different "continental cluster" on that Y axis to the X axis?"

It is the same logic. LB shares non-West Eurasian affinity with MA-1 and Amerindians.

German Dziebel said...

@Tobus (contd.)

"This is not correct. East Asians are closer to Amerindians than MA-1 is. I believe I've had to correct you on this before."

You are confused. I said that Amerindians are the closest to MA-1 among modern human populations. This is the direct read from f3 statistics in Raghavan, Fig. 1c. This figure also shows you how genetically closest populations (e.g., Amerindians and Eskimo-Aleuts) also have closest f3 values related to MA-1 (in this case). Raghavan Fig. 1c parallels Olalde EDF 5d. The only difference is that LB falls outside of its continental West Eurasian cluster and shows the same amount of MA=Amerindian ancestry as - likely - northern Amerindians.

BTW, proximity between East Asians and Amerindians is irrelevant to the present discussion because it's likely inflated compared to West Eurasians and Pauans due to recent, post-24,000 YA gene flow from North America into East Asia,

"I've already shown this with Papuans and Europeans plotting as a single cluster on the Karitiana axis (plus a host of other examples on various axes). Thus we have both mathematical proof and real world examples showing that the positions of X populations relative to each other is not a consistent indicator of affinity between those X populations... ready to accept it yet, or do you have more hoops for me to jump through in lieu of providing some proof of your own? "

Nonsense.

"The fact is that Europeans and East Asians are the same distance from each other whatever other population you compare them with."

No, they are far from each other when the plots are anchored in Sardinians vs. Han, but they are close when the plot is anchored in Karitiana.

German Dziebel said...

@agiering

"How has intergroup diversity in America changed since 1492?"

Sorry, missed that one. Because of TerryT's constant spamming of the thread. Honestly, I don't have figures in front of me. I would assume it's very different (lower) than among unadmixed Amerindians because of massive gene flow.

Kristiina said...

According to this new paper "Insights from high-resolution Y chromosome and mtDNA sequences" http://biorxiv.org/content/biorxiv/suppl/2014/01/13/001792.DC1/001792-1.pdf,
haplogroup X seems to be closer to N1 than W, and X and W together form a sister clade to East Asian A + N9/Y. As for U, it stands apart from all other R clades in the same way as R9/F, but instead P, B and R0/HV seem to form a group. This may suggest that U is the westernmost branch and F the easternmost branch while the rest of the R group has a more central origin. Anyway, it is interesting that the Papuan/Australian P seems to cluster with the West Eurasian R0/HV.

terryt said...

@ Kristiina:

"haplogroup X seems to be closer to N1 than W, and X and W together form a sister clade to East Asian A + N9/Y. As for U, it stands apart from all other R clades in the same way as R9/F, but instead P, B and R0/HV seem to form a group. This may suggest that U is the westernmost branch and F the easternmost branch while the rest of the R group has a more central origin".

Dienekes posted concerning this paper and I made cooments there similar to those I made above. I notice the conclusions haven't been accpeted by Phylotree yet but if we accept the data we can say goodbye to any R expansion via South Asia. South Asian R haplogroups entered that region from either side.

@ German:

"missed that one. Because of TerryT's constant spamming of the thread".

You have posted many more comments here than I have. Besides which you have carefully avoided considering any of the points I've raised. I realise that in order to maintain your belief you are forced to do that. And to use illogical arguments against Tobus. As a result I cannot take anything you say seriously.

agiering said...

@ terryT

"If I remember correctly they were mainly surprised at the presence of Y-DNA R as it is not very common in the region today. They were probably expecting Q, a close relation, or an eastern haplogroup. As far as I'm aware they were not at all surprised at its American or West Eurasian affinity. "

It was Mal'ta's affinity to south/west Eurasians and Amerindians OVER east Asians that was so surprising. It forced them to accept a link between Amerindians and west Eurasians, in addition to the already-established Amerindian-East Asian link.

I'm quoting one of the researchers, Eske Willerslev, directly: “The result came as a complete surprise to us. Who would have thought that present-day Native Americans, who we learned in school derive from East Asians, share recent evolutionary history with contemporary western Eurasians? Even more intriguingly, this happened by gene flow from an ancient population that is so far represented only by the MA-1 individual living some 24,000 years ago."

@ German

"'No, they [East Asians and Europeans] are far from each other when the plots are anchored in Sardinians vs. Han, but they are close when the plot is anchored in Karitiana."

I'm looking at the Anzick Paper's f3 plots and I'm getting an impression similar to yours regarding Olade and Raghavan --- namely that Europeans and East Asians are most similar when the plots are anchored at MA-1, Mayan, Karitiana and Anzick. Indeed on those graphs East Asians are noticably closer to Europeans than to Amerindians. I suppose this would be problematic for the idea of an ancient [pre-Tianyuan] East-West Eurasian split w/ the Amerindian split-off happening later.

"Sorry, missed that one. Because of TerryT's constant spamming of the thread. Honestly, I don't have figures in front of me. I would assume it's very different (lower) than among unadmixed Amerindians because of massive gene flow."

I'm genuinely curious how "groups" would be defined in the modern day Americas.

Tobus said...

@German:
There's nothing incorrect about their placements

There is if you are positing that f3 scores indicate genetic affinity between the X populations. When a Papuan scores in between a French and a Russian that breaks up your idea of "continental clustering". It confirms that f3 doesn't necessarily reflect genetic distance between the X populations used, only their relative distances to the A population.

It's called f4

That is absolutely laughable German! Are you honestly trying to tell me that the difference between f3(X1, A; O) and f3(X2, A; O) is equivalent to f4(X1, X2; A, O)? Do you even know how these calculations work?

Please, if you are able, present a serious method of using two f3 scores to get the genetic distance between the X populations used. If you are not able, then please accept that what you are proposing is not possible.

All that matters right now is that East Asians, MA-1 and West Eurasians are all close to each other genetically in terms of their shared Amerindian ancestry

This is where you are completely wrong! The f4 stats show clearly that the "Amerindian ancestry" shared by MA-1 and Eurasians is completely *NOT* shared by East Asians... that's why MA-1 can not have recieved Amerindian DNA, but must have given West Eurasian DNA to Amerindians instead.

Amerindians are the most divergent population in those plots.
yes, I did. Re-read my answer

The question is why are you applying one interpretation to the Papuan/European scores and a different one to LB/Amerindians... neither this answer nor your previous two answered this.

It is the same logic. LB shares non-West Eurasian affinity with MA-1 and Amerindians

No, that's different logic - the same logic would be "LB and Amerindians have the same amount of MA-1 ancestry but are in different continental clusters". You are mixing up the X and A populations when moving from graph to graph - what applies to Papuans/Europeans on 5a also applies to LB/Amerindians on 5d.

I said that Amerindians are the closest to MA-1 among modern human populations.

So you agree that East Asians are closer to Amerindians than MA-1? ... and thus that any association you make between MA-1 and Amerindians based on this is even more applicable to East Asians and Amerindians?

Nonsense

Nonsense what? The maths is wrong? (show me!) or Papuans and Europeans are the same distance apart in every f3 axis? (again, show me!).

We have a mathematical formule that allows genetically diverse X populations to get the same score and we have examples of genetically diverse X populations getting the same score. I conclude that this means we can't rely on that score to determine genetic distance between X populations using these scores and you say "Nonsense" with no explanation. Nice work genius.

No, they are far from each other when the plots are anchored in Sardinians vs. Han, but they are close when the plot is anchored in Karitiana.

Ta da! So how can we possibly rely on this information to tell us how close they *actually* are?

terryt said...

"I'm quoting one of the researchers, Eske Willerslev, directly"

More than a little hyperbole and promotion in that comment.

"It was Mal'ta's affinity to south/west Eurasians and Amerindians OVER east Asians that was so surprising. It forced them to accept a link between Amerindians and west Eurasians, in addition to the already-established Amerindian-East Asian link".

Not so. Several other papers have claimed much the same thing, and even this 1999 paper (before the haplogroup nomenclature was standardised) claims it as the most logical conclusion:

http://www.ncbi.nlm.nih.gov/pubmed/10053017

Most woul;d not have been surprised by MA-1's affinity with West Eurasians. By the way, I doubt you will get German to agree with, 'already-established Amerindian-East Asian link'.

"The question is why are you applying one interpretation to the Papuan/European scores and a different one to LB/Amerindians..."

Of course we all know the answer to that.

German Dziebel said...

@Tobus

"When a Papuan scores in between a French and a Russian that breaks up your idea of "continental clustering". It confirms that f3 doesn't necessarily reflect genetic distance between the X populations used, only their relative distances to the A population."

It doesn't break anything. But, as I told you before, f3 conflates ancient common descent with recent admixture. Some European populations such as Russians may be pulled slightly away from the rest of Europeans due to recent flow from Asia. And "relative distance" to the A population is precisely what we want when it comes to understanding admixture patterns. The traditional measures of genetic distance by themselves won't tell you the directionality of gene flow.

"That is absolutely laughable German! Are you honestly trying to tell me that the difference between f3(X1, A; O) and f3(X2, A; O) is equivalent to f4(X1, X2; A, O)? Do you even know how these calculations work?"

You are laughing at yourself. Both stats measure affinity between 3 or 4 populations. If you doubt that East Asians are close to West Eurasians when it comes to Amerindian affinity seen on f3 plots, you can include all of them into one formula and get what you are looking for. Lazaridis used f4 to ascertain that LB, just like all other ancient European samples, has Amerindian affinities.

"Please, if you are able, present a serious method of using two f3 scores to get the genetic distance between the X populations used. If you are not able, then please accept that what you are proposing is not possible."

You simply don't know how to apply mathematics to population genetics. And why do you need "absolute" genetic distances when you already have genetic distances between two populations derived from a third one, or admixed with a third one?

"The f4 stats show clearly that the "Amerindian ancestry" shared by MA-1 and Eurasians is completely *NOT* shared by East Asians... that's why MA-1 can not have recieved Amerindian DNA, but must have given West Eurasian DNA to Amerindians instead."

Nonsense. Raghavan didn't consider the possibility that East Asians and West Eurasians derive from the same Amerindian source in the first place. He assumed that Amerindians are a recent offshoot of East Asians and since West Eurasians are not closer to East Asians than to Papuans, he assumed that the flow went to America. But Amerindians are divergent from all of Eurasians, hence contributing genes to all three populations (plus later additionally to East Asians after West Eurasians and Papuans had become separate populations) wouldn't make one of them closer to the other. Modern East Asians show greater proximity to Amerindians compared to East Asians because of later Holocene gene flow from North America. f3 stats show this nicely.

"No, that's different logic - the same logic would be "LB and Amerindians have the same amount of MA-1 ancestry but are in different continental clusters". You are mixing up the X and A populations when moving from graph to graph - what applies to Papuans/Europeans on 5a also applies to LB/Amerindians on 5d."

No, it's the exactly the same logic. MA-1 is an ancient Amerindian-derived population, LB and modern Amerindians occupy different continental clusters but they are related to each other through an ancient Amerindian population. Amerindians stayed behind, while West Eurasians moved to Europe. It's very simple, but not for Big Brain Tobus.


Of course we can: they diverged from a common source and didn't mix much ever since, hence they are distinct from each other but related to a third population, which is Amerindians. You really are a creationist: since the very idea of Old World populations descending from Amerindians is abhorrent to you, you refuse to accept scientific evidence for it, no matter how simple and straightforward the evolutionary logic is.

German Dziebel said...

@Tobus (contd.)

"So you agree that East Asians are closer to Amerindians than MA-1? ... and thus that any association you make between MA-1 and Amerindians based on this is even more applicable to East Asians and Amerindians?"

Modern East Asians are closer to Amerindians (especially northern Amerindians) because of additional Holocene gene flow from North America. But Amerindians are the closest relatives of MA-1 among modern human populations.

"We have a mathematical formule that allows genetically diverse X populations to get the same score and we have examples of genetically diverse X populations getting the same score. I conclude that this means we can't rely on that score to determine genetic distance between X populations using these scores and you say "Nonsense" with no explanation."

I've given you loads of explanations already. Unless you believe in polygenesis, all modern human populations are related to each other, they come from the same source population, then they diverged. What you in f3 charts is divergence between West Eurasians and Papuans but also the traces of their common descent from an Amerindian-like/proto-Amerindian population.

"So how can we possibly rely on this information to tell us how close they *actually* are?"

terryt said...

" You really are a creationist: since the very idea of Old World populations descending from Amerindians is abhorrent to you"

No, German. You are the 'creationist'. You have an American population suddenly appearing from nowhere (specially created) and then spreading through much of Eurasia. Where an early movement out mixes with populations that also suddenly appear from nowhere (multiple creations). Then a second movement out of America with a completely different genetic makeup to the first (yet another creation) moves out and becomes East Asian. You can provide no evidence for any of those migrations and no mechanism by which they may have developed. The ultimate creationist.

Secondly, I'm sure that Tobus, just like me, is in no way emotionally against an American origin. It is just that you have provided not the slightest amount of evidence for such a belief. If you ever manage to provide such evidence both of us will readily accept 'out of America'.

"you refuse to accept scientific evidence for it, no matter how simple and straightforward the evolutionary logic is".

There is absolutely no evidence for it in spite of what you wish, and your claims consistently fail to demonstrate any logic.

"Modern East Asians are closer to Amerindians (especially northern Amerindians) because of additional Holocene gene flow from North America".

Obviously involving yet another specially created American population.

Tobus said...

@German:
It doesn't break anything

So Papuans live in Europe do they?

You are laughing at yourself.

Nope, squarely at you - you're like Curly trying to hammer in a screw, getting told "you can't use a hammer for that" so he gleefully pulls out a monkey wrench!

Both stats measure affinity between 3 or 4 populations

If you knew anything about these stats you'd know that neither of them measures affinity between 3 or 4 populations. Outgroup f3 measures shared drift between 2 populations, and f4 measures which of 2 populations is the closest to a third. Read the sources - it's pretty easy to understand and the maths for both f3 and f4 are well within the reach of anyone with basic high-school level maths.

You simply don't know how to apply mathematics to population genetics.

No offense, but you just claimed that f4 is the difference between 2 f3 stats... I obviously know a great deal more about it than you do!

But Amerindians are divergent from all of Eurasians

West Eurasians are more divergent from both Amerindians and East Asians than they are from each other.... but surely you know that already.

MA-1 is an ancient Amerindian-derived population

Science-denying belief-seller!! MA-1 is ancestral to modern Amerindians, not derived from them! Time goes forwards remember!

LB and modern Amerindians occupy different continental clusters but they are related to each other through an ancient Amerindian population.

Almost right, we're getting somewhere! But you forgot that MA-1 only shares 30% (at best!) of his DNA with modern Amerindians, and that he lived in Eurasia not America... so there's no (honest) way he can be considered an "ancient Amerindian" genetically. If your whole theory that LB has more Amerindian affinity than modern Europeans is based on this false belief that MA-1 is somehow an Amerindian, then it's no wonder that this supposed Amerindian affinity doesn't show up in EDF 5a - it's all just based on hot air!

But Amerindians are the closest relatives of MA-1 among modern human populations.

Well they certainly get the highest f3 (shared drift) and f4 (shared derived allele) scores when the panel is masked for European admixture, but they don't plot closest on the PCA, nor do they represent the highest component in ADMIXTURE... so it's a debatable point.

I've given you loads of explanations already.

That you have, but zero proof to back these explanations up. It's patently obvious to anyone looking objectively at the data that the f3 scores between two given X populations can vary dramatically from axis to axis, so attempting to ascertain a consistent and meaningful measure of affinity from these data is sheer folly. (It's also obvious from the maths, but that's way over your head). What we *can* actually ascertain from these data is what the f3 actually measures - relative distance of each X population to the A population. In the case we are discussing, this means we can use direct measurement against Karitiana to ascertain LB's affinity to Amerindians relative to other populations... and as everyone not blinded by prior belief can see, LB's affinity to Amerindians is within the range shown by modern Europeans.

German Dziebel said...

@Tobus

"So Papuans live in Europe do they?"

If you believe that Europe is "two different directions" from Amerindians, then yes. They occupy two different geographic areas peopled from a third one, America. I didn't expect your geography to be on the same level as your mathematics.

"If you knew anything about these stats you'd know that neither of them measures affinity between 3 or 4 populations. Outgroup f3 measures shared drift between 2 populations, and f4 measures which of 2 populations is the closest to a third. Read the sources - it's pretty easy to understand and the maths for both f3 and f4 are well within the reach of anyone with basic high-school level maths."

I know everything about these maths. They measure everything one needs to make inferences about prehistory. Your "absolute" genetic distance is a made up requirement intended to prove some kind of "polygenist" agenda.

"No offense, but you just claimed that f4 is the difference between 2 f3 stats... I obviously know a great deal more about it than you do!"

You complained that f3 doesn't "directly" calculate the distance between two X populations. f4 includes them into a singe measurement against a third population, which admixed into populations X1 and X2 to a varying degree. That's all we need.

"West Eurasians are more divergent from both Amerindians and East Asians than they are from each other.... but surely you know that already."

Nonsense.

"Science-denying belief-seller!! MA-1 is ancestral to modern Amerindians, not derived from them! Time goes forwards remember!"

I remember it very well. MA-1 is ancestral to modern East Asians, as it's located in East Asia and shows no separate East Asian component (because it emerged later), and derived from ancient Amerindians.

"Well they certainly get the highest f3 (shared drift) and f4 (shared derived allele) scores when the panel is masked for European admixture, but they don't plot closest on the PCA, nor do they represent the highest component in ADMIXTURE... so it's a debatable point."

PCA is an outdated method. A component doesn't need to have the highest ADMIXTURE rank to be ancestral.

"That you have, but zero proof to back these explanations up. It's patently obvious to anyone looking objectively at the data that the f3 scores between two given X populations can vary dramatically from axis to axis, so attempting to ascertain a consistent and meaningful measure of affinity from these data is sheer folly. (It's also obvious from the maths, but that's way over your head). What we *can* actually ascertain from these data is what the f3 actually measures - relative distance of each X population to the A population. In the case we are discussing, this means we can use direct measurement against Karitiana to ascertain LB's affinity to Amerindians relative to other populations... and as everyone not blinded by prior belief can see, LB's affinity to Amerindians is within the range shown by modern Europeans.

More baloney from Science Denier Tobus. f3 evidence is compelling and f4 confirmed that LB has Amerindian affinity.

Tobus said...

@German:
They occupy two different geographic areas peopled from a third one, America.

... and so applying that same logic to EDF 5d means LB and Amerindians occupy two different geopgraphic areas peopled from a third one, MA-1. Whatever explanation you give for Papuans and Europeans on EDF 5a must logically also be a possible explanation for LB and Amerindians on EDF 5b.

The main point here is that their relative distance on a single axis on the f3 plot doesn't represent their actual genetic distance. Papuans getting an identical score as Europeans on the Karitiana axis *doesn't* mean that they are genetically identical. Agreed?

I know everything about these maths.

Yet you think the difference between two f3's equals f4?!?! How about you show me everything you know about the maths behind this proposition.

You complained that f3 doesn't "directly" calculate the distance between two X populations. f4 includes them into a singe measurement against a third population, which admixed into populations X1 and X2 to a varying degree. That's all we need.

Didn't you just say you know everything about that maths? f4 doesn't measure two populations against a 3rd - it measures by what degree one of the two (if any) has more derived alleles in common with the 3rd - there's no way to determine the genetic distance between X1 and X2 in f4, just like there's no way to measure the difference between X populations in two f3 stats.

Nonsense.

Nonsense what? Look at Olalde EDF 5b - Amerindians are much closer to East Asians than Europeans are, hence Europeans are the more divergent of the 3.

MA-1 is ancestral to modern East Asians, as it's located in East Asia and shows no separate East Asian component (because it emerged later)....

You're getting more ridiculous every second. MA-1 shows no affinity to East Asians, so is certainly not ancestral to them, they'd already diverged by MA-1's time and have not had any subsequent MA-1-like admixture.

...and derived from ancient Amerindians.

No... MA-1 is ancestral to modern Amerindians, not derived from them! Time goes forwards remember!

More baloney

Are you saying that f3 scores between two given X populations *DON'T* vary dramatically from axis to axis? How about you tell me the distance between Papuans and Europeans on each of EDF 5a's X and Y axes.

German Dziebel said...

@Tobus

"and so applying that same logic to EDF 5d means LB and Amerindians occupy two different geopgraphic areas peopled from a third one, MA-1. Whatever explanation you give for Papuans and Europeans on EDF 5a must logically also be a possible explanation for LB and Amerindians on EDF 5b."

Talking in circles. MA-1 is an ancient Amerindian population just like Sardinians are a West Eurasian population. MA-1 has additional West Eurasian affinities but they don't affect (or predate) its Amerindianness, since Amerindians and not West Eurasians are rightmost on the MA-1-anchored axis.

"The main point here is that their relative distance on a single axis on the f3 plot doesn't represent their actual genetic distance. "

Your "actual genetic distance" is a product of your imagination.

"Papuans getting an identical score as Europeans on the Karitiana axis *doesn't* mean that they are genetically identical. Agreed? "

They are genetically related via a (New World) population ancestral to modern Amerindians. Of course not identical. Which populations are genetically "identical"?

"it measures by what degree one of the two (if any) has more derived alleles in common with the 3rd - there's no way to determine the genetic distance between X1 and X2 in f4, just like there's no way to measure the difference between X populations in two f3 stats."

Nonsense. This is all we need. What other measurements of genetic affinity can we hope for? What's the difference between genetic affinity and genetic distance? What do you mean by genetic distance if not the amount of shared alleles? I begin thinking you're just a robot making spam comments.

"Yet you think the difference between two f3's equals f4?"

I don't know if they "equal" each other. But if you're not happy that the genetic distance between two populations in two f3 runs are not measured "directly," you can include them in a f4 run. in the case of LB and Amerindians f4 confirmed the results of f3 - they share Amerindian affinity.

German Dziebel said...

@Tobus (contd.)

"MA-1 shows no affinity to East Asians, so is certainly not ancestral to them."

Oh, an ancient sample should show as a discrete component for a population that hadn't yet diverged? Very funny, Tobus.

"they'd already diverged by MA-1's time and have not had any subsequent MA-1-like admixture."

Yes, sure, and they lived on Atlantis. MA-1 has West Eurasian, Amerindian, Papuan and South Asian components. It has all the components that bookend East Asians geographically. This suggests that East Asians didn't exist yet, otherwise they would've been captured by a rather comprehensive MA-1 "coverage.". But Amerindians have surely diverged by that time. Later East Asians emerged but they didn't admix with West Eurasians. It's all very simple.

"No... MA-1 is ancestral to modern Amerindians, not derived from them!"

By your ahistorical logic, MA-1 is an unadmixed East Asian population at 24,000 YBP that's ancestral to Amerindians, West Eurasians, Papuans and South Asians... but not East Asians. Again, confusion between derivation and attestation. You need to have Anzick 30% West Eurasian and 70% East Asian for your to work, not MA-1. But Anzick is already Amerindian and so is MA-1 (plus West Eurasian admixture unattested in the New World). Amerindians, West Eurasians, Papuans and South Asians had already diverged by then, but East Asians had not.

"Are you saying that f3 scores between two given X populations *DON'T* vary dramatically from axis to axis?"

So, what's wrong with them varying within a certain range? We have a perfect structural alignment between West Eurasians and South Asians, on the one hand, and Papuans and East Asians, on the other, in their relationship to Amerindians. This is key. In tree-like terms, the first split was between Amerindians and all of Eurasians, then between Papuans and West Eurasians, then between West Eurasians and South Asians and between Papuans and East Asians.

Tobus said...

@German:
MA-1 is an ancient Amerindian population just like Sardinians are a West Eurasian population

No, MA-1 is an ancient Eurasian population, and his relationship to Amerindians is small (less than 30% of his genome) and distant in time and space (24,000 years and some 15,000km). This is not at all like how Sardinians are a West Eurasian population. I know you'd like to think of him as an "Amerindian", but look at the data again, he's clearly very different.

They are genetically related via a (New World) population ancestral to modern Amerindians

MA-1 wasn't in the New World.

This is all we need. What other measurements of genetic affinity can we hope for?

A direct measurement against the two populations, obviously. We have these and they disgree with what you are saying - if the measurements don't say you want, you're not allowed to just ignore them and make up some arbitrary interpretation.

What do you mean by genetic distance if not the amount of shared alleles?

f3(X1, A; O) measures shared alleles between X1 and A. f3(X2, A; O) measured shared alleles between X2 and A. What you are saying is that the difference between these two is representative of shared alleles between X1 and X2. This is not necessarily true, since X1 and X2 can have non-identical relationships to A and hence the X1/A alleles can be completely different to the X2/A alleles... understand?

in the case of LB and Amerindians f4 confirmed the results of f3 - they share Amerindian affinity.

We know LB has Amerindian affinity - what we're discussing is whether this is more, less or about the same as what modern Europeans show, and EDF 5a is direct measurement of that.


This suggests that East Asians didn't exist yet, otherwise they would've been captured by a rather comprehensive MA-1 "coverage.". But Amerindians have surely diverged by that time. Later East Asians emerged but they didn't admix with West Eurasians.

If East Asians hadn't already diverged then they'd show the exact same affinity to MA-1 has modern Amerindians, since Amerindians and East Asians would have still been a single population at that time. For a population *not* to show affinity, it means it (or it's ancestors) *had* already diverged... you seem to have it back to front.

MA-1 is an unadmixed East Asian population at 24,000 YBP that's ancestral to Amerindians, West Eurasians, Papuans and South Asians

Where are you getting Papuans from? Raghavan has MA-1 with equal affinity to Papuans and East Asians, hence the rejection of Amerindian->MA-1 possibility.

But Anzick is already Amerindian and so is MA-1

MA-1 is not Amerindian, not in ADMIXTURE, not in PCA, not in f3, not in D-stats, not nowhere but in your head. I'm not sure where you originally got the idea that he is, but please revisit it after rereading the data.

So, what's wrong with them varying within a certain range?

It means we can't use it as reliable measure of distance. If a ruler gives varying measurements between the same two points then it means you're not using it right.

We have a perfect structural alignment between West Eurasians and South Asians, on the one hand, and Papuans and East Asians, on the other, in their relationship to Amerindians.

And a totally different picture of each of the other axes... shall we make up a different theory for each one?

terryt said...

"No, MA-1 is an ancient Eurasian population, and his relationship to Amerindians is small (less than 30% of his genome) and distant in time and space (24,000 years and some 15,000km)".

German seems to have no idea that MA-1 was actually found in Eurasia, not in America. Perhaps that is where his misunderstanding arises. German, take note: MA-1 was found in Eurasia!

"By your ahistorical logic, MA-1 is an unadmixed East Asian population at 24,000 YBP that's ancestral to Amerindians, West Eurasians, Papuans and South Asians... but not East Asians. Again, confusion between derivation and attestation".

But hang on German. You now seem agree that MA-1 is has no East Asian EDAR370A mutation. But within America it gave rise to a population with the mutation which population gave rise to East Asians. The diversification within a single population into two completely different populations then subsequent hybridism between these populations within America is removed from the scenario Tobus and I have been proposing only in a matter of timing. Simply place the EDAR- population in Central Eurasia and the EDAR+ population in East Asia and you have it exactly.

German Dziebel said...

@Tobus

"No, MA-1 is an ancient Eurasian population, and his relationship to Amerindians is small (less than 30% of his genome) and distant in time and space (24,000 years and some 15,000km). This is not at all like how Sardinians are a West Eurasian population. I know you'd like to think of him as an "Amerindian", but look at the data again, he's clearly very different... MA-1 is not Amerindian, not in ADMIXTURE, not in PCA, not in f3, not in D-stats, not nowhere but in your head. I'm not sure where you originally got the idea that he is, but please revisit it after rereading the data. "

MA-1 is the closest to Amerindians out of all modern populations. This means out of all Eurasian populations. This was established by Raghavan and reconfirmed by everybody else including latest Lazaridis. This makes MA-1 more Amerindian than Eurasian. But of course why accept scientific data if you can reject it?!

"f3(X1, A; O) measures shared alleles between X1 and A. f3(X2, A; O) measured shared alleles between X2 and A. What you are saying is that the difference between these two is representative of shared alleles between X1 and X2. This is not necessarily true, since X1 and X2 can have non-identical relationships to A and hence the X1/A alleles can be completely different to the X2/A alleles... understand?"

Well, prove it - that those alleles are "completely different." Don't just say that they may be different. This is a special explanation that requires a special proof. And also provide evidence that those "completely different" alleles are shared with another population. In this case, with which other population? UFOs are off the table

"We know LB has Amerindian affinity - what we're discussing is whether this is more, less or about the same as what modern Europeans show, and EDF 5a is direct measurement of that."

Nonsense. I've already explained to you how to read this data.

"If East Asians hadn't already diverged then they'd show the exact same affinity to MA-1 has modern Amerindians, since Amerindians and East Asians would have still been a single population at that time. For a population *not* to show affinity, it means it (or it's ancestors) *had* already diverged... you seem to have it back to front."

No, East Asians show enough affinity with Amerindians just like MA-1 shows enough (up to 50%) affinity with Amerindians. East Asians just diverged from Amerindians in post-MA-1 times and didn't admix with West Eurasians as a separate Amerindian spinoff ever since, hence no proximity to MA-1 via its West Eurasian side. But as Olalde EDF 5a and 5e show, East Asians are closer to West Eurasians than either of them to Amerindians.

"It means we can't use it as reliable measure of distance. If a ruler gives varying measurements between the same two points then it means you're not using it right."

Your whole idea of genetic distance is product of your imagination.

Tobus said...

@German:
MA-1 is the closest to Amerindians out of all modern populations.

No, East Asians are the closest to Amerindians out of all modern populations. At least the 3rd time I've had to correct you on this, please remember it this time.

This makes MA-1 more Amerindian than Eurasian

You are forgetting that there is more than one "Eurasian" population in the samples. Only 30% (max, possibly only 14%) of MA-1 DNA is Amerindian-like, the rest is Eurasian-like, but spread between various Eurasian groups. Amerindians as a distinct group might represent the largest apportioned slice of MA-1 DNA, but they do not represent the majority of his DNA, not by a long way.

Well, prove it - that those alleles are "completely different."

I've already proved it multiple times: (0.5 - 1)(0.5 - 0) = (0.5 - 0)(0.5 - 1) remember? Opposite alleles, identical f3 scores - two f3 scores with the same result *does not* prove that the two X populations share alleles with each other. What it proves is that the two X populations share the *same number* of alleles with the A population, not necessarily the exact *same alleles*.

Nonsense

Nonsense what? Are you saying that EDF 5a is *NOT* a direct measurement of shared drift with Karitiana?!?!

East Asians just diverged from Amerindians in post-MA-1 times

Then at MA-1's time Amerindians and East Asians were still a single population no? Meaning MA-1's "Amerindian" affinity should be as equally "East Asian" as "Amerindian".

But as Olalde EDF 5a and 5e show, East Asians are closer to West Eurasians than either of them to Amerindians.

*Sigh*, we've been through this before. EDF 5b clearly shows that Amerindians are much closer to East Asians than Europeans are, as does every other set of data ever published on the subject.

Your whole idea of genetic distance is product of your imagination.

Maybe, but at least it gives the same answer each time, and an answer that is consistent with all other ways of measuring it.






Rokus said...

'MA-1 is the closest to Amerindians out of all modern populations. This means out of all Eurasian populations. This was established by Raghavan and reconfirmed by everybody else including latest Lazaridis. This makes MA-1 more Amerindian than Eurasian. But of course why accept scientific data if you can reject it?!'

Some logical error and misinterpretation is going on and on. Raghavan only intended to look at a single Amerindian component, and he also explained about the other component. In the supplement is explained where the remaining part is from: '496 bootstrap replicates supported Karitiana as having 26.1% (7.7-44.4%) ancestry from the MA-1 lineage and the remainder from Han, consistent with the previous analysis.'

Hence, not MA-1 but "Han" is closest to Amerindians "somehow", what can't be revealed by looking at graphs meant for showing dual ancestry.

German Dziebel said...

@Tobus

"No, East Asians are the closest to Amerindians out of all modern populations. '

Re-read Raghavan. Out of all modern populations MA-1 is the closest to Amerindians. Just forget about East Asians. We're talking about the genetic affinities of MA-1. (As a side note, even if East Asians are closer to Amerindians on PCAs, this can easily be explained as more recent admixture from the Americas.)

"You are forgetting that there is more than one "Eurasian" population in the samples. Only 30% (max, possibly only 14%) of MA-1 DNA is Amerindian-like, the rest is Eurasian-like, but spread between various Eurasian groups. Amerindians as a distinct group might represent the largest apportioned slice of MA-1 DNA, but they do not represent the majority of his DNA, not by a long way."

MA-1 is closer to Amerindians than to Eurasian populations.

"I've already proved it multiple times: (0.5 - 1)(0.5 - 0) = (0.5 - 0)(0.5 - 1) remember? "

You have a perverted idea of what a proof is. You may have outlined a possibility but in the absence of actual calculations involving real populations, it's no proof of anything.

"Are you saying that EDF 5a is *NOT* a direct measurement of shared drift with Karitiana?!?!"

yes, all Eurasian populations share drift with Amerindians.

"EDF 5b clearly shows that Amerindians are much closer to East Asians than Europeans are, as does every other set of data ever published on the subject."

EDF 5d clearly shows that West Eurasians are closer to Amerindians than East Asians are. East Asians and West Eurasians split from a proto-Amerindian population, hence they are variably related to modern Amerindians. Like I said multiple times, it's irrelevant if East Asians sometimes look closer to Amerindians - Holocene gene flow from North America would explain this easily. But we're talking about 24,000 years ago.

"Maybe, but at least it gives the same answer each time, and an answer that is consistent with all other ways of measuring it."

No answer whatsoever.

"*Sigh*, "

You still naively assume that you are explaining to me how data works. Don't flatter your self - you are the lost cause, not me.

"Then at MA-1's time Amerindians and East Asians were still a single population no? Meaning MA-1's "Amerindian" affinity should be as equally "East Asian" as "Amerindian".

No. MA-1's Amerindian affinity "should be equally" Amerindian as that of modern Amerindians and it is! East Asians diverged from Amerindians since then and haven't mixed with West Eurasians. How can they be equally related to MA-1 as Amerindians? You are using lack of evidence (no affiliation with modern East Asians of a geographically East Asian ancient population) as positive evidence that this population (East Asians) existed at that time and was more divergent from the actually attested and clearly differentiated populations (West Eurasians, Amerindians, Papuans, South Asians). So, with your framework in mind, MA-1 is a ghost ANE population and East Asians is a ghost ancient East Asian population.

German Dziebel said...

@Rokus

We're talking about Raghavan's Fig. 1 C. "Heat map of the statistic f3(Yoruba; MA-1, X) where X is one of 147 worldwide non-African populations (standard errors shown in Supplementary Fig. 21). The graded heat key represents the magnitude of the computed f3 statistics." MA-1 shares strongest affinity with Amerindians and not with Eurasians.

If Amerindians were 50% Han, MA-1 would've been part Han, too, but it clearly isn't. So the only alternative is that Amerindians are ancestral to both West Eurasians and East Asians.

terryt said...

"If Amerindians were 50% Han, MA-1 would've been part Han, too, but it clearly isn't. So the only alternative is that Amerindians are ancestral to both West Eurasians and East Asians".

Your 'conclusion' is by no means 'the only alternative'. In fact it cannot be correct at all. I'm sure we have established that MA-1 has no East Asian (Han) element. Therefore the fact that Amerindians have some East Asian element while MA-1 doesn't demonstrates conclusively that Amerindians cannot be ancestral to MA-1. And the fact that East Asians have minimal MA-1 element demonstrates conclusively that Amerindians cannot be ancestral to them either. So the only alternative is actually that Amerindians are descended from both West Eurasians and East Asians, not the other way round.

"Like I said multiple times, it's irrelevant if East Asians sometimes look closer to Amerindians - Holocene gene flow from North America would explain this easily"

No. The distribution of the EDAR370A mutation alone shows that cannot be so. The mutation is scattered almost randomly through America but concentrated in East Asia in Mongolia, the north of China Korea and Japan. It is impossible to explain that distribution with out of America.

"(no affiliation with modern East Asians of a geographically East Asian ancient population"

For some weird reason (although I actually know exactly why) you are deliberately confusing 'East Asian' as a geographic concept with 'East Asian' as a human genetic concept. Are you really claiming no population in Eurasia has changed over the last 24,000 years? I'm sure I remember you claiming very rapid evolution in America, for example, yet here you are denying any possible change in genetic structure in northeast Asia over a prolonged period. Yet another example of you consistent inconsistency.

Tobus said...

@German:
MA-1 is closer to Amerindians than to Eurasian populations

MA-1 is something like 30% Amerindian, 25% South Asian, 25% Central Asians and 20% European... that makes him more Eurasian than Amerindian.

If you want to divide up the continental sub-populations then he's something like 20% South Amerindian and 15% North Amerindian - so closer to individual Eurasian populations than to individual American populations.

You may have outlined a possibility but in the absence of actual calculations involving real populations, it's no proof of anything.

We *do* have actual calculations involving real populations! 5a tells us that MA-1 shares more alleles with Amerindians than LB does. 5d says that LB and Amerindians share the same number of alleles with MA-1. Since LB has less Amerindian alleles than MA-1 does (5a), the only way he can have the same number of alleles in common with MA-1 as Amerindians do (5d) is if he has some extra ones that Amerindians don't have.

East Asians diverged from Amerindians since then and haven't mixed with West Eurasians. How can they be equally related to MA-1 as Amerindians?

Because if they hadn't yet diverged at the time of the MA-1 admixture then they were still a single population, and so your "Amerindian" admixture was just as much "East Asian" at the time - it would have been common Amerindian/East Asian DNA that MA-1 would have received, not just Amerindian. This scenario is something we can detect with the null result D-stat... if Amerindians and East Asians are a clade relative to MA-1, then D(Chimp, MA-1, Han, Karitiana) should return 0, not significantly non-zero. There is no doubt that East Asians and Amerindians diverged sometime before the Amerindian/MA-1 admixture event.

If Amerindians were 50% Han, MA-1 would've been part Han, too

Only if the gene flow came from Amerindians into MA-1. If the gene flow went the other way, then MA-1 wouldn't have received any Han at all. The fact the MA-1 and East Asians don't show the affinity with each other that either shows with Amerindians is clear evidence that East Asians were a already a distinct lineage at the time of the admixture.

German Dziebel said...

@Tobus

"MA-1 is something like 30% Amerindian, 25% South Asian, 25% Central Asians and 20% European... that makes him more Eurasian than Amerindian."

Raghavan is clear that MA-1's closest relatives are Amerindians. Latest Lazaridis writes: "Ancient Eurasians (Europeans and MA1) are genetically closer to Karitiana than to North Asians, intermediately related to Onge and East Asians, and least related to Papuans."

But you can keep denying science. Which part of Genesis did you read in the morning?

"We *do* have actual calculations involving real populations! 5a tells us that MA-1 shares more alleles with Amerindians than LB does."

Yes, that's correct.

"LB and Amerindians share the same number of alleles with MA-1."

No, there're some Amerindian populations that may fit this, but not all. Karitiana and especially Anzick clearly don't.

"if he has some extra ones that Amerindians don't have."

Sure, those are the West Eurasian alleles that Amerindians don't have. This is consistent withe everything else and shows that Amerindians did not experience West Eurasian gene flow.

"There is no doubt that East Asians and Amerindians diverged sometime before the Amerindian/MA-1 admixture event."

In this case, MA-1 would've been as close to East Asians as to Amerindians (precisely for your own reason that "it would have been common Amerindian/East Asian DNA that MA-1 would have received, not just Amerindian") and a distinct East Asian component would have been visible in MA-1. This is clearly not the case. Lazaridis, latest, p. 97, puts it nicely: " Ancient Eurasians (Europeans and MA1) are genetically closer to Karitiana than to North
Asians, intermediately related to Onge and East Asians, and least related to Papuans." (This is exactly what Olalde 5d shows us.) West Eurasians are still related to East Asians but not as strongly as West Eurasians are related to Amerindians and East Asians related to Amerindians. This suggests that all three (four including Papuans) share an Amerindian substrate from which they diverged in their own respective directions and didn't intermix much ever since.

"If the gene flow went the other way, then MA-1 wouldn't have received any Han at all."

But MA-1 is 30-50% Amerindian (as you yourself admit above saying "MA-1 is something like 30% Amerindian.") if Amerindians diverged from East Asians prior to absorbing West Eurasian gene flow, MA-1 which has Amerindian admixture would've been closer to East Asians than it is. East Asians may have diverged from Amerindians prior to 24,000 years but in any case it was a Amerindian>East Asian process, not East Asian>Amerindian process. This would mean that East Asians were still a small population at 24,000 years, consequently didn't stretch into such a wide territory as they presently occupy. So, the separation of East Asians into a distinct population from a wider Amerindian base may have happened not too long prior to 24,000 years. And this whole scenario is possible not for the reason you identified (if East Asians are not part of MA-1, hence they must have diverged earlier) but because Amerindians are related to East Asians independently of West Eurasians.

terryt said...

" those are the West Eurasian alleles that Amerindians don't have".

You mean 'the non-existent West Eurasian alleles that Amerindians don't have'. How on earth do you propose that there were any West Eurasians at all before Amerindians entered the region? Which part of Genesis are you quoting?

"In this case, MA-1 would've been as close to East Asians as to Amerindians"

Are you able to provide one good reason why that would be so? Or are you relying on your creationist 'logic' here?

"Ancient Eurasians (Europeans and MA1) are genetically closer to Karitiana than to North
Asians, intermediately related to Onge and East Asians, and least related to Papuans."

Which absolutely rules out an Amerindian origin for those later groups if you're going to claim MA-1 as 'Amerindian'.

"This suggests that all three (four including Papuans) share an Amerindian substrate from which they diverged in their own respective directions and didn't intermix much ever since".

This 'diversification you're proposing, how did it happen? Separate creations in each region? Your lack of understanding of biological evolution is still preventing you from seeing that most mutations are neutral of actually harmful and useful mutations need inbreeding and selection to become established in a population.

"East Asians were still a small population at 24,000 years, consequently didn't stretch into such a wide territory as they presently occupy".

At last. I hope you can now avoid making the stupid claims regarding MA-1 and East Asians you have consistently made in the past. Of course that 'small population' didn't prevent them contributing genes to Amerindians.

Tobus said...

@German:
Raghavan is clear that MA-1's closest relatives are Amerindians.

Rhagavan explicitly states that "MA-1 is composed of five genetic components of which the two major ones make up ca. 70% of the total. The most prominent component is shown in green and is otherwise prevalent in South Asia but does also appear in the Caucasus, Near East or even Europe. The other major genetic component (dark blue) in MA-1 is the one dominant in contemporary European populations, especially among northern and northeastern Europeans." I'm not sure what you are reading, but this clearly puts MA-1 as ~70% Eurasian.

Sure, those are the West Eurasian alleles that Amerindians don't have. This is consistent withe everything else

Most importantly it's actual calculations involving real populations where f3 is counting different alleles and reaching the same score. This proves that we can't rely two f3 scores to infer the distance between the X populations used.

In this case, MA-1 would've been as close to East Asians as to Amerindians (precisely for your own reason that "it would have been common Amerindian/East Asian DNA that MA-1 would have received, not just Amerindian")

Ummm... if East Asians had already diverged the it *wouldn't* have been common Amerindian/East Asian DNA - East Asians and Amerindian would have been different populations already! Your logic seems to be back to front - "diverged" means different populations, "not diverged" means still combined as a single population.

and a distinct East Asian component would have been visible in MA-1.

No, because a) East Asians and Amerindians were already separate populations, and more importantly, b) MA-1 didn't receive *any* DNA - he gave it to an north-east Asian tribe/branch/population that went on to become Amerindians (and don't get confused again, it only shows up as "Amerindian" in ADMIXTURE because they're the best modern representation of it, not because it necessarily came from them).

But MA-1 is 30-50% Amerindian (as you yourself admit above saying "MA-1 is something like 30% Amerindian.")

More like 14 to 30%, but hey, fudge the data in your favour.

If Amerindians diverged from East Asians prior to absorbing West Eurasian gene flow, MA-1 which has Amerindian admixture would've been closer to East Asians than it is

MA-1 doesn't have Amerindian admixture - that's the whole point... if he did he'd be closer to East Asians.

Amerindians are related to East Asians independently of West Eurasians.

EXACTLY! Amerindians are related to both West and East Eurasians, but these two are not (particularly) related to each other - this perfectly describes a situation where Amerindians have a mix of West and East Eurasian ancestors.

German Dziebel said...

@Tobus

"I'm not sure what you are reading.."

Look at Fig. 1C in Raghavan. All of your percentages are post-Amerindian, as Amerindians don't have any of the Eurasian components found in MA-1. What matters, using your own words, is that "Amerindians as a distinct group might represent the largest apportioned slice of MA-1 DNA."

"This proves that we can't rely two f3 scores to infer the distance between the X populations used."

As good as it gets.

"Ummm... if East Asians had already diverged the it *wouldn't* have been common Amerindian/East Asian DNA - East Asians and Amerindian would have been different populations already!"

So, East Asians diverged from Amerindians in East Asia and that's how they became East Asians? I think you intuitively gravitated to the right solution: East Asians diverged from Amerindians, hence MA-1 which has Amerindian ancestry, is not particularly close to East Asians. That's exactly what out-of-America postulates but only that East Asians are a group of Amerindians that diverged on their way to East Asia. The key is that MA-1 has an "Amerindian" component, not an "East Asian" component. The idea that Amerindians are West Eurasians + East Asians would've worked if MA-1 was half West Eurasian, half East Asian and then Anzick was half-Amerindian, 1/4 East Asian, 1/4 West Eurasian. This is demonstrably not the case.

"More like 14 to 30%, but hey, fudge the data in your favour."

The percentages don't matter (Lazaridis gives 50% p. 134) but you're trying to squirm out of the fact that MA-1 has an Amerindian component. Just to re-post: "But MA-1 is 30-50% Amerindian (as you yourself admit above saying "MA-1 is something like 30% Amerindian.") if Amerindians diverged from East Asians prior to absorbing West Eurasian gene flow, MA-1 which has Amerindian admixture would've been closer to East Asians than it is."

"MA-1 doesn't have Amerindian admixture - that's the whole point... if he did he'd be closer to East Asians."

No, only under the idea that Amerindians derived from East Asians. If East Asians derived from Amerindians, then MA-1 shouldn't be expected to be close to East Asians.

"EXACTLY! Amerindians are related to both West and East Eurasians, but these two are not (particularly) related to each other - this perfectly describes a situation where Amerindians have a mix of West and East Eurasian ancestors."

EXACTLY THE OPPOSITE! West and East Eurasians are both related to Amerindians. All of West Eurasians and all of East Asians. They are intermediately related to each other because they share a more recent common Amerindian ancestor than, say, Papuans. Going back to Lazaridis: "Ancient Eurasians (Europeans and MA1) are genetically closer to Karitiana than to North
Asians, intermediately related to Onge and East Asians, and least related to Papuans."

terryt said...

"Amerindians are related to both West and East Eurasians, but these two are not (particularly) related to each other - this perfectly describes a situation where Amerindians have a mix of West and East Eurasian ancestors".

Amen.

terryt said...

"So, East Asians diverged from Amerindians in East Asia and that's how they became East Asians?"

That is not what Tobus is saying, and I'm sure you know it. Unless your creationist belief is clouding your judgement completely. East Asians did not diverge from Amerindians anywhere. The evidence is overwhelming that Amerindians diverged from East Asians through admixture with an MA-1 like population.

"East Asians diverged from Amerindians, hence MA-1 which has Amerindian ancestry, is not particularly close to East Asians".

As Tobus has attempted to explain multiple times if the picture you proclaim is be correct MA-1 should have a substantial genetic connection to East Asians. MA-1 shows no such connection therefore your belief is incorrect.

"That's exactly what out-of-America postulates"

Out of America actually postulates a whole mish-mash of improbable scenarios. As demonstrated by:

"If East Asians derived from Amerindians, then MA-1 shouldn't be expected to be close to East Asians".

You are conveniently ignoring the fact that if MA-1 too is derived from Amerindians he should have a significant East Asian genetic component.

"you're trying to squirm out of the fact that MA-1 has an Amerindian component".

Absolutely everyone agrees that MA-1 and Amerindians share a genetic component. Tobus is not 'trying to squirm out of' anything. It is you who is doing the squirming. Trying to make the data fit your creationist belief.

"The idea that Amerindians are West Eurasians + East Asians would've worked if MA-1 was half West Eurasian, half East Asian and then Anzick was half-Amerindian, 1/4 East Asian, 1/4 West Eurasian. This is demonstrably not the case".

Largely correct except for the but 'MA-1 was half West Eurasian, half East Asian'. Why do you believe that is necessary? Why does MA-1 need any East Asian genetic connection if Amerindians are 1/2 MA-1 and 1/2 East Asian? (as you say, 'The percentages don't matter' but actual percentages are more like 1/3 MA-1 and 2/3 East Asian). The mistake you are continually making is that you refuse to consider, even for a moment, that your belief might be wrong. This prevents you from looking at the evidence objectively and enables you to continue with your creationist belief. To maintain that belief you are forced to use completely conflicting methods of analysing similar data. As demonstrated by the way you use MA-1's genetic connections. And this completely stupid comment:

"EXACTLY THE OPPOSITE! West and East Eurasians are both related to Amerindians".

But West and East Eurasians are not related to each other, which is absolutely impossible if both descent from Amerindians. And you call yourself 'intelligent'?

Tobus said...

@German:
What matters, using your own words, is that "Amerindians as a distinct group might represent the largest apportioned slice of MA-1 DNA."

What matters in this discussion is whether MA-1 is "Amerindian" in the same way that Sardinians are "European" (since you are trying to use that proposition to backup your point). It doesn't matter what way you slice it - they're not the same, not even close.

So, East Asians diverged from Amerindians in East Asia and that's how they became East Asians?

Technically, both East Asians and Amerindians diverged from a combined proto-"AmerAsian" (Tianyuan-like?) population, and yes this almost certainly happened in geographic East Asia.

The idea that Amerindians are West Eurasians + East Asians would've worked if MA-1 was half West Eurasian, half East Asian and then Anzick was half-Amerindian, 1/4 East Asian, 1/4 West Eurasian.

I genuinely don't understand your logic German, I suspect you are missing something - why would MA-1 have to be half-East Asian if he represents the non-East Asian part of Amerindians?

To repost... "MA-1 which has Amerindian admixture"

To repost "MA-1 doesn't have Amerindian admixture - that's the whole point... if he did he'd be closer to East Asians."

Lazaridis gives 50% p. 134

That's the "the MA1 admixture into Karitiana" figure, not the amount of "Amerindian" in MA-1's DNA.

If East Asians derived from Amerindians, then MA-1 shouldn't be expected to be close to East Asians.

Yes he would, unless you are now saying that East Asians diverged from Amerindians long before MA-1.

EXACTLY THE OPPOSITE!

What then *would* you expect to see if Amerindians are in fact the descendants an East Asian population that received West Eurasian admixture? The demonstrated East/West affinity of Amerindians but lack of East/West affinity in Eurasia is precisely what we'd expect.

German Dziebel said...

@Tobus

"What matters in this discussion is whether MA-1 is "Amerindian" in the same way that Sardinians are "European" (since you are trying to use that proposition to backup your point). It doesn't matter what way you slice it - they're not the same, not even close."

It's the same principle applied to modern populations vs. a cluster that involves a 24,000 year old population. In the same way as Sardinians diverged from Europeans (or absorbed a SSAfrican-like admixture), MA-1 diverged from Amerindians into the West Eurasian territory. But ancestrally both clusters are directly comparable. Another option is of course your preferred one, namely that MA-1 is a UFO-derived population.

"I genuinely don't understand your logic German, I suspect you are missing something - why would MA-1 have to be half-East Asian if he represents the non-East Asian part of Amerindians? "

Your conclusion obviously precedes the evidence, hence you write "he represents the non-East Asian part of Amerindians" as if it was the idea we need to find evidence for. Your logic is pre-scientific.

One needs to have an ancient Siberian population with affinities to modern East Asians and modern West Eurasians to postulate Amerindians as an admixed population. Note that the understanding is that the admixture event that supposedly "produced" modern Amerindians happened in Asia. But sadly there's no evidence for that admixed West Eurasian-east Asian population in Asia. MA-1 has an Amerindian component but not an East Asian component. So out-of-America is more consistent with facts.

"To repost "MA-1 doesn't have Amerindian admixture - that's the whole point... if he did he'd be closer to East Asians."

Only if Amerindians had derived from from East Asians, but if East Asians and West Eurasians split from Amerindians they would be close to Amerindians (which is a fact) and distant from each other. That's what divergence without subsequent admixture is by definition. Ancient West Eurasians are closer to East Asians than to Papuans (per latest Lazaridis), so this falsifies Raghavan's claim that MA-1 is as removed from East Asians as from Papuans.

"Technically, both East Asians and Amerindians diverged from a combined proto-"AmerAsian" (Tianyuan-like?) population, and yes this almost certainly happened in geographic East Asia."

This is inconsistent with your other belief in the admixed, West Eurasian-East Asian origin of Amerindians. If Amerindians are part of an Amer-Asian population, they already precede the divergence of modern East Asians from this Amerindian-Tianyuan-like population. Using your terminology, we could say that Amerindians derive from an Amer-European population. And then it's the "admixture" between Amer-Asian and Amer-European population that supposedly generated modern Amerindians. Or, using normal logic, East Asians and West Eurasians split off from a single ancestral population for which modern Amerindians is the best surviving proxy.

German Dziebel said...

@Tobus (contd.)

"What then *would* you expect to see if Amerindians are in fact the descendants an East Asian population that received West Eurasian admixture?"

I described the expectations multiple times. We shouldn't have seen the Amerindian component differentiated at 24,000 years in the first place. A 24,000 year old Siberian sample should be 1/2 West Eurasian, 1/2 East Asian. ADMIXTURE plots should've shown BLUE (West Eurasian) and ORANGE (East Asian) mixed in individual American populations at k=3, K=4 levels. A 12,000 year old Amerindian sample should be 1/4 East Asian, 1/4 West Eurasian, 1/2 Amerindian. All of West Eurasians and all of East Asians shouldn't have been pulled closer to Amerindians, but just a subset of both in the areas closest to America. So, basically we would expect only Siberians (some of whom, such as Mansi and Khanty, known to be West Eurasian-East Asian admixed) to show an Amerindian pull. Amerindian heterozygosity levels should've been higher than East Asian.

Y-DNA hgs N, O and R should've been found in America. f3 should've showed greater drift sharing between MA-1 and modern East Asians (especially Siberians) followed by Europeans and only then Amerindians. There wouldn't have been an "Amerindian shift" across Eurasia, with Amerindians occupying the rightmost corner of the plots. Rather, they would've been in the middle of a plot, with West Eurasians being rightmost in MA-1 anchored plots.

"The demonstrated East/West affinity of Amerindians but lack of East/West affinity in Eurasia is precisely what we'd expect."

There's an Amerindian affinity of Amerindians. Then there's an East Asian pull in North America and West Eurasian pull in South America (roughly). And then there's Amerindian affinity in East Asia and Amerindian affinity in West Eurasia but no East Asian-West Eurasian affinity without Amerindians. This is precisely what we would expect to see if out-of-America was true.






























































terryt said...

"Another option is of course your preferred one, namely that MA-1 is a UFO-derived population".

From considering all the evidence you have mustered so far it is difficult to avoid the conclusion that Amerindians are a UFO-derived population. You have offered no alternative that I can see. What's more that Amerindian UFO-derived population has then mixed with several different populations in Eurasia that also appear to be UFO-derived. Are you able to explain to us exactly where you believe these different populations originated. And, considering you believe an Amerindian population mixed with non-Amerindian populations in Eurasia why you can still maintain modern humans descend from Amerindians. Nothing you say makes evolutionary sense. I suppose that comes from your having a PhD in 'dressing up as an Indian'.

"A 24,000 year old Siberian sample should be 1/2 West Eurasian, 1/2 East Asian".

Why? That need only be so if you believe Amerindians are as ancient as 24,000 years. The fact that MA-1 shows Amerindian affinity, but not completely, indicates Amerindians had not formed by 24,000 years ago. Or, if they had formed, the relationship was not directly with MA-1 but with an MA-1-like population.

"Y-DNA hgs N, O and R should've been found in America".

The fact they are not is alone enough to destroy your out of America belief.

"Your conclusion obviously precedes the evidence"

WHAT? Your circular arguments demonstrate clearly that you have already formed your conclusion and then have consistently proceeded to ignore any evidence that demolishes that conclusion, concentrating instead on evidence that could be considered ambiguous.

"So out-of-America is more consistent with facts".

Out of America is completely inconsistent with almost all the facts. Even the 'facts' it could be considered consistent with can best be described as 'ambiguous'. They just as easily fit an out of Eurasia origin for Amerindians.

" but if East Asians and West Eurasians split from Amerindians they would be close to Amerindians (which is a fact) and distant from each other".

No they wouldn't. They would be virtually indistinguishable. Once more you demonstrate your complete ignorance of evolutionary biology in spite of your claim to having a degree in dressing up as an Indian.

Tobus said...

@German:
One needs to have an ancient Siberian population with affinities to modern East Asians and modern West Eurasians to postulate Amerindians as an admixed population

No one doesn't, one just needs one of each and a modern population that is clearly a mixture of them. Do we have need an ancient sample that's a mix of WHG and EEF to deduce that modern Europeans are a mixture of the two?

You're like a Creationist insisting of physical proof for every step of the evolutionary tree and claiming there's a "missing link" if there's a logically inferable piece missing from the display case...

But sadly there's no evidence for that admixed West Eurasian-east Asian population in Asia

... and yet there's no evidence for *ANY* population in America pre 15,000kya but that's never stopped you positing the existence of such.

Ancient West Eurasians are closer to East Asians than to Papuans (per latest Lazaridis), so this falsifies Raghavan's claim that MA-1 is as removed from East Asians as from Papuans.

Lazaridis is measuring relative to Yoruba (consistent with Papuans diverging from the Eurasian populations first), while Raghavan is relative to Sardinians. MA-1 is no more East Asian than Sardinians, but much more Amerindian than Sardinians. Since Amerindians are very East-Asian-like, this lack of East Asian affinity is inconsistent with a theory of Amerindian gene flow into MA-1.

This is inconsistent with your other belief in the admixed, West Eurasian-East Asian origin of Amerindians. If Amerindians are part of an Amer-Asian population, they already precede the divergence of modern East Asians from this Amerindian-Tianyuan-like population

It's a moot point if Amerindians diverged from the East Asians line or whether East Asians diverged from the Amerindian line - the point is they were once the same population and they split from each other. The West Eurasian admixture occurred after this split.

Using your terminology, we could say that Amerindians derive from an Amer-European population.

No we couldn't, because the European lineage diverged before the "AmerAsian" one, so it would need to be "AmerAsianEuro" - if you're going back to before the West/East Eurasian split then just use "Eurasians" or "non-Africans".

Tobus said...

@German (cont):
We shouldn't have seen the Amerindian component differentiated at 24,000 years in the first place.

The components in that PCA are all modern - they are differentiated *now* - and retrofitted back to the 24kya sample. Considering that the relationship that most modern Eurasians have with MA-1 is via a basal lineage, not direct ancestry, is it really a surprise that a population *with* direct ancestry would best represent a portion of the ancestral lineage?

A 24,000 year old Siberian sample should be 1/2 West Eurasian, 1/2 East Asian

Why? If there was no admixture with Amerindians it would be 100% West Eurasian, the Amerindian component only comes into play because some 30% of Amerindian DNA is inherited from this 24kya West Eurasian sample.

A 12,000 year old Amerindian sample should be 1/4 East Asian, 1/4 West Eurasian, 1/2 Amerindian.

At 12,000 years we are at least half way from the divergence point. It makes perfect sense that the modern populations that best represent Anzick are his modern Amerindian descendants. He's only 12kya away from them, but at least 36kya from any other modern population (12kya back up the branch to MA-1/East Asians plus 24kya down the branch of any other modern lineage).

It seems you might be thinking that the components in the ancient PCAs actually existed at the time of the ancient sample - they didn't. They exist now and are being projected back in time to the ancient sample on a "best fit" basis.

Amerindian heterozygosity levels should've been higher than East Asian.

At the time of the admixture 24kya they probably were, the variation we see today is more a reflection of migrations, demographics and admixture since that time. It would be overly simplistic to expect a sum of modern heterozygosity levels to reflect an admixture event that long ago on a different continent.

Y-DNA hgs N, O and R should've been found in America.

Haplogroups are notoriously bad at representing ancestry - before the Lazaridis paper, theories based on haplogroup distribution posited a nearly complete replacement of HG Europeans by Neolithic farmers, when in fact HGs represent roughly 50% of modern European DNA. If you expect haplogroups to follow ancestry, you're going to be disappointed, misled or confused fairly often.

f3 should've showed greater drift sharing between MA-1 and modern East Asians (especially Siberians) followed by Europeans and only then Amerindians

Amerindians received more MA-1 specific drift at 24kya than earlier branches off the basal lineage, so their high f3 stats, while somewhat unexpected due to them only sharing 1/3 of his DNA, is not completely improbable... particularly if we consider their isolated demographic history relative to Europeans. We'd only expect shared drift with populations that have shared ancestry or admixture with MA-1, and I'm not sure why you think East Asians would have this.

This is precisely what we would expect to see if out-of-America was true.

It's not at all what I'd expect to see - I'd expect Amerindians to be an outgroup to all non-American populations, or, if some populations are "late divergences", that the other non-American populations would have identical affinity to Amerindians and these late-comers. The varied level of affinity between Amerindians and non-Amerindians (and within non-Amerindians) makes it very hard to reconcile America as the origin of all modern humans.

German Dziebel said...

@Tobus

"No one doesn't, one just needs one of each and a modern population that is clearly a mixture of them."

And this population exists. It's Uygurs! Not Amerindians.

" and yet there's no evidence for *ANY* population in America pre 15,000kya but that's never stopped you positing the existence of such."

19th century creationists, too, demanded from Darwin fossils to illustrate his theory. Since then the fossils keep coming and creationists continue to debate them 150 years later. The archaeological signs of Amerindian antiquity are growing (slower than on other continents but nevertheless) but cryptocreationists such as Tobus are denying their existence as well as the existence of millions of years ahead of us to keep finding them. But we shouldn't switch to archaeology when we're talking about genetics. MA-1 proves that Amerindians are at least 24,000 years old.

"Raghavan is relative to Sardinians."

Wrong. MA-1 is slightly more East Asian than Sardinians.

"(consistent with Papuans diverging from the Eurasian populations first)"

What Eurasians? Consistent with Papuans diverging from Amerindians first.

"Since Amerindians are very East-Asian-like."

Wrong. It's East Asians who are Amerindian-like. East Asians are more Amerindian than West Eurasians (at some sites) and West Eurasians are more Amerindian than East Asians (at other sites). That's a sure sign of their divergence from a common Amerindian ancestor.

"Why? If there was no admixture with Amerindians it would be 100% West Eurasian, the Amerindian component only comes into play because some 30% of Amerindian DNA is inherited from this 24kya West Eurasian sample."

Oh, so there was an admixture with Amerindians? I thought your whole idea was that MA-1 is not Amerindian-admixed.

"At 12,000 years we are at least half way from the divergence point. It makes perfect sense that the modern populations that best represent Anzick are his modern Amerindian descendants. He's only 12kya away from them, but at least 36kya from any other modern population (12kya back up the branch to MA-1/East Asians plus 24kya down the branch of any other modern lineage)."

Well, you need to have some proof from a New World sample that the admixture occurred. Since you don't believe there were any Amerindians in the New World prior to 15,000 years ago and by 12,000 years ago they are already Amerindian, you've maxed out on supplying conditions of falsifiability for your beliefs.

"It seems you might be thinking that the components in the ancient PCAs actually existed at the time of the ancient sample - they didn't."

ANE is a population that went extinct. Ancestors of modern populations didn't exist at the time of ANE. Back to the UFO idea?

"At the time of the admixture 24kya they probably were, the variation we see today is more a reflection of migrations, demographics and admixture since that time. It would be overly simplistic to expect a sum of modern heterozygosity levels to reflect an admixture event that long ago on a different continent."

Pseudoscience! Amerindians went through a bottleneck when they diverged from East Asians, so their heterozygosity went down. Then they admixed with West Eurasians, their heterozygosity went up. Then they went through a new bottleneck that resulted in their heterozygosity again going down. All this hassle just to please one 21st centiry cryptocreationist by the name of Tobus.

German Dziebel said...

@Tobus (contd.)

"Haplogroups are notoriously bad at representing ancestry - before the Lazaridis paper, theories based on haplogroup distribution posited a nearly complete replacement of HG Europeans by Neolithic farmers, when in fact HGs represent roughly 50% of modern European DNA. If you expect haplogroups to follow ancestry, you're going to be disappointed, misled or confused fairly often."

I hope TerryT is reading it. He's a big fan of haplogroups. What do you think of blood groups instead, Tobus? Amerindians don't have a blood group commonly found in Asia, namely bg B. One Amerindian tribe, Blackfoot, has world highest frequencies of bg A, which is otherwise most common in Europe. But the rest of Amerindians are nearly fixed at bg O. You gotta learn how to think across datasets and furnish some support for your ideas. Well, maybe archaeology can help you? But, alas, the earliest New World technologies are similar to West Eurasian ones, not East Asian.

"Amerindians received more MA-1 specific drift at 24kya than earlier branches off the basal lineage, so their high f3 stats, while somewhat unexpected due to them only sharing 1/3 of his DNA, is not completely improbable... particularly if we consider their isolated demographic history relative to Europeans."

No. MA-1 received Amerindian drift prior to 24,000 years ago. Amerindians is that basal lineage and it behaves as such in the stats. No need to misinterpret it to fit your beliefs.

"We'd only expect shared drift with populations that have shared ancestry or admixture with MA-1, and I'm not sure why you think East Asians would have this."

Because you believe Amerindians are "very East Asian-like."

"It's not at all what I'd expect to see - I'd expect Amerindians to be an outgroup to all non-American populations, or, if some populations are "late divergences", that the other non-American populations would have identical affinity to Amerindians and these late-comers. The varied level of affinity between Amerindians and non-Amerindians (and within non-Amerindians) makes it very hard to reconcile America as the origin of all modern humans."

You need to get to a true outgroup by peeling off the onion on thousands of years of continent-specific history. What makes modern Africans an "outgroup" to modern populations is archaic admixture in Africa some 40,000 years. What makes Amerindians share variable affinity with different populations is the varying degree of gene flow post separation between a) Papuans and Denisovans (some 40,000 years); c) northern Amerindians and East Asians (12-10,000 years); b) northern Amerindians and southern Amerindians.

terryt said...

"I hope TerryT is reading it. He's a big fan of haplogroups. What do you think of blood groups instead"

Yes, I am 'a big fan of haplogroups'. They tell us a great deal about the often separate male and female migration patterns. However I agree that 'Haplogroups are notoriously bad at representing ancestry'. They are only part of the story and are reasonably easily replaced by later arriving ones. Blood groups, like haplogroups, also show human migration patterns. But are also 'notoriously bad at representing ancestry'. But your PhD in Dancing-with-Indians doesn't let you understand genetics.

"Amerindians is that basal lineage and it behaves as such in the stats. No need to misinterpret it to fit your beliefs".

German, there is absolutely no doubt as to who here is fitting misinterpretations to their pre-existing beliefs. A particularly great example:

"MA-1 proves that Amerindians are at least 24,000 years old".

It does no such thing. All it proves is that a population that forms part of the Amerindian gene pool existed in Siberia 24,000 years ago. Your statement is a perfect example of fitting misinterpretations to pre-existing beliefs.

"East Asians are more Amerindian than West Eurasians (at some sites) and West Eurasians are more Amerindian than East Asians (at other sites). That's a sure sign of their divergence from a common Amerindian ancestor".

Rubbish. In fact it makes their divergence from a common Amerindian ancestor completely impossible. If they had each derived from a common Amerindian ancestor they would be far more similar to each other than they actually are.

Tobus said...

@German:
19th century creationists, too, demanded from Darwin fossils to illustrate his theory.

Nice try, but it's you who's demanding an "ANE" fossil - still think it's a creationist position?

MA-1 proves that Amerindians are at least 24,000 years old.

And were in East Asia, not America.

Wrong. MA-1 is slightly more East Asian than Sardinians.

The Rhagavan D-stat that shows MA-1 equidistant to Han as Papuan is D(Sardinian, MA-1; Han, Papuan)... the Lazaridis stat showing MA-1 is closer to Han is D(Yoruba, MA-1, Han, Papuan). You claimed the two were contradictory, but they are measuring different things. Rhagavan's assertion of no East Asian affinity in MA-1 (and hence, no gene flow from Amerindians) remains valid.

Well, you need to have some proof from a New World sample that the admixture occurred

All Native Amerindians have West Eurasian affinity... the only way they could have got it, as Raghavan clearly demonstrates in a number of different measurements, is via admixture from an MA-1-like population.

All this hassle just to please one 21st centiry cryptocreationist

Yes German, *I'm* the one positing a theory that has no support and every expert in the world agrees with *you*!!

Amerindians is that basal lineage and it behaves as such in the stats

Funny, the latest Schiffels paper disagrees with you, as do. Raghavan and Lazaridis... I guess they're creationists too right?

Because you believe Amerindians are "very East Asian-like."

Of course - that's what the data says.

What makes Amerindians share variable affinity with different populations is the varying degree of gene flow post separation

Except that f3 stats (and f4 to some degree) ignore non-related gene flow, and yet still show the discrepancy. You're running out of excuses.

German Dziebel said...

@Tobus

"Nice try, but it's you who's demanding an "ANE" fossil - still think it's a creationist position?"

Wake up, Tobus, Mal'ta is a fossil. And it's transitional between Amerindians and West Eurasians. From one of early Amerindian populations to the ancestor of West Eurasians, that is.

"And were in East Asia, not America."

So, Amerindians were in East Asia 24,000 years ago. But, wait, may be we can think about it differently? Every time you run into a Pakistani person in the U.K, you must be thinking that Pakistanis in Pakistan came from the U.K.

"The Rhagavan D-stat that shows MA-1 equidistant to Han as Papuan is D(Sardinian, MA-1; Han, Papuan)... the Lazaridis stat showing MA-1 is closer to Han is D(Yoruba, MA-1, Han, Papuan). You claimed the two were contradictory, but they are measuring different things>"

They don't measure different things. That's your fantasy. First of all, it was Raghavan that put forth minor East Asian contribution to MA-1 on the basis of Han being closer to MA-1 than to Sardinians. It was in fact Lazaridis who used a different set of populations (Loschbour instead of Sardinians) and observed that MA-1 and Loschbour were equidistant. They decided to explain the Sardinian shift as influenced by another fictitious population, namely Basal Eurasian, which happened after Loschbour times. Lazaridis admitted that both Loschbour and MA-1 may have experienced East Asian gene flow. What they overlooked is that both MA-1 and Loschbour are known to have Amerindian admixture and Han is "very much like Amerindians."

"Funny, the latest Schiffels paper disagrees with you, as do. Raghavan and Lazaridis... I guess they're creationists too right?"

They haven't considered my alternative. (And Schiffels used admixed MXL for both Amerindians and West Eurasians.) But you have been exposed to it for the past months on a day-to-day basis. And you still refuse to adopt science. You must be a creationist. Not sure about Schiffels, Raghavan and Lazaridis.

"Yes German, *I'm* the one positing a theory that has no support and every expert in the world agrees with *you*!!"

Your mistake is that you misclassify my theory as "baseless" instead of "new." New and baseless theories sometimes do look similar but you need to be able to differentiate between them.

"Of course - that's what the data says."

Where is the data that says that Amerindians are East Asian like vs. East Asians are Amerindian like?

"Except that f3 stats (and f4 to some degree) ignore non-related gene flow, and yet still show the discrepancy. You're running out of excuses."

What discrepancy?

terryt said...

"Your mistake is that you misclassify my theory as 'baseless' instead of 'new.' New and baseless theories sometimes do look similar but you need to be able to differentiate between them".

Agreed. Your theory is both 'new' and 'baseless'. Actually it's not particularly new. I remember it was seriously considered some years ago, but abandoned when it became obvious it could not be correct.

Tobus said...

@German:
Mal'ta is a fossil.

So ANE is not "fictional"?

So, Amerindians were in East Asia 24,000 years ago. But, wait, may be we can think about it differently? Every time you run into a Pakistani person in the U.K, you must be thinking that Pakistanis in Pakistan came from the U.K.

I might if there was nobody at all in Pakistan - we have "Amerindians" in Asia at 24kya but not in America until 15kya... You claim that ANE is "fictional" yet have no problem positing an entirely fictitious population (and sometimes multiple ones!) that there is even less evidence for... hypocrite much?

They don't measure different things. That's your fantasy.

They certainly don't contradict each other - MA-1 has the same East Asian affinity that Sardinians have, and yet has significantly more Amerindian affinity than Sardinians have, something we would not expect to see if MA-1's received Amerindian gene-flow.


What they overlooked is that both MA-1 and Loschbour are known to have Amerindian admixture

They didn't "overlook" that because it's not a fact - it's your fantasy. MA-1 and Loschbour (and all West Eurasians) have some degree of Amerindian affinity, but as as Raghavans's tree mix runs prove, this is due to Amerindians having West Eurasian adxmiture, not the other way round. The f4 stats confirm this by showing a lack of corresponding East Asian affinity in those populations with higher Amerindian affinity.

And Schiffels used admixed MXL for both Amerindians and West Eurasians

And plotted both with and without European alleles masked. In neither case is there support for the Amerindian lineage predating all the others. You are making up excuses to ignore the data - you should be embracing it and finding a solution that works with it.

Your mistake is that you misclassify my theory as "baseless" instead of "new."

No, your theory is old and has been made obsolete by recent data.

What discrepancy?

The discrepancy between how the X populations plot on the MA-1 axis compared to the Karitiana axis... do try to keep up.


Where is the data that says that Amerindians are East Asian like vs. East Asians are Amerindian like?

Six of one, half a dozen of the other. If MA-1 received DNA from Amerindians we'd expect to see a corresponding increased affinity to the very "Amerindian-like" East Asians.

German Dziebel said...

@Tobus

"So ANE is not "fictional"?"

As a pure population, it's a fiction. As a mix of Amerindians and West Eurasians it's pretty real.

"we have "Amerindians" in Asia at 24kya but not in America until 15kya.."

We haven't found Amerindian DNA in America at 25,000 years. Just like we haven't found myriads of other things. But the fact that it's attested in Asia at 24,000 proves that Amerindians were in the New World for at least that long.

"MA-1 has the same East Asian affinity that Sardinians have."

Sardinians have less east Asian affinity than MA-1, at least according to Raghavan.

"MA-1 and Loschbour (and all West Eurasians) have some degree of Amerindian affinity, but as as Raghavans's tree mix runs prove, this is due to Amerindians having West Eurasian adxmiture, not the other way round. The f4 stats confirm this by showing a lack of corresponding East Asian affinity in those populations with higher Amerindian affinity."

Raghavan's TreeMix is base don a flawed assumption of Amerindians falling into the East Asian cluster. The fact that West Eurasians have Amerindian affinity but not East Asian affinity proves that Amerindians are not of East Asian extraction.

"In neither case is there support for the Amerindian lineage predating all the others. You are making up excuses to ignore the data - you should be embracing it and finding a solution that works with it."

You take every academic paper as divine revelation instead of critically interpreting its results. The very fact that Schiffels makes Amerindians as the closest to the Chinese puts its results in conflict with all the Amerindian admixture in West Eurasian literature we've been discussing. BTW, look at Fig. 3 at Schiffels and the Amerindian dotted line separating from the rest at 100,000+ years ago.

"No, your theory is old and has been made obsolete by recent data."

All the recent genomic and ancient DNA data supports it. But you're welcome to believe that Father d'Acosta is still fresh and illuminating. That's what makes you a cryptocreationist.

"The discrepancy between how the X populations plot on the MA-1 axis compared to the Karitiana axis... ."

I explained to you how to read those charts. There's no discrepancy.

" If MA-1 received DNA from Amerindians we'd expect to see a corresponding increased affinity to the very "Amerindian-like" East Asians."

Nonsense. Its the other way around. East Asians diverged from Amerindians. West Eurasians diverged from Amerindians. They were isolated from each other since the split. They are different from each other but equally descended from Amerindians. This is basic evolution, my creationist friend.

terryt said...

"All the recent genomic and ancient DNA data supports it".

Such as? Please supply at least one example of a scientist who supports your belief. Haplogroup phylogenies alone certainly make your creationist belief untenable.

"But the fact that it's attested in Asia at 24,000 [what you like to call 'Amerindian DNA'] proves that Amerindians were in the New World for at least that long".

Would you mind explaining just how it proves that to be so.

"The fact that West Eurasians have Amerindian affinity but not East Asian affinity proves that Amerindians are not of East Asian extraction".

Again would you mind explaining exactly how you come to that conclusion. It doesn't make sense.

"The very fact that Schiffels makes Amerindians as the closest to the Chinese puts its results in conflict with all the Amerindian admixture in West Eurasian literature we've been discussing".

Once more you demonstrate a lack of logic. Surely Schiffels' findings completely match the idea that the Amerindian admixture in West Eurasian is not from Amerindian to West Eurasia but in the opposite direction.

"You take every academic paper as divine revelation instead of critically interpreting its results".

That is a far less biased approach than the one you adopt. It is blindingly obvious that you take every academic paper as divine revelation if it can possibly be twisted enough to remotely fit your belief and absolutely ignore the multitude of papers that make your belief untenable. Any valid hypothesis has to fit ALL the evidence, not just a selected sampling of that data.

"West Eurasians diverged from Amerindians. They were isolated from each other since the split. They are different from each other but equally descended from Amerindians. This is basic evolution, my creationist friend".

No, it's basic bulldust. It's got nothing at all to do with evolution, and everything to do with creationism. If West Eurasians and East Eurasians diverged from Amerindians they would show significant similarity to each other. If their present significant difference from each other was the product of separate drift and evolution since that separation their connection with Amerindians would have become completely obscured. But because you lack any understanding of genetics or evolutionary biology you will be unable to see that.

Tobus said...

@German:
But the fact that it's attested in Asia at 24,000 proves that Amerindians were in the New World for at least that long.

Weird, I woulda thought it being attested in *Asia* would prove they were in *Asia*.

Sardinians have less east Asian affinity than MA-1, at least according to Raghavan.

Raghavan SI14.6 (pg 92) : Dfreq(Papuan, Han; Sardinian, MA-1) = -0.002±0.005 (Z = -0.36)

Raghavan's TreeMix is base don a flawed assumption of Amerindians falling into the East Asian cluster.

No, it's not - this "flawed assumption" is a reality directly dictated by the data. TreeMix first uses the data to build the maximum likelihood tree, then finds the admixture that best fits the data to that tree, the tree is not preconceived or assumed.

The very fact that Schiffels makes Amerindians as the closest to the Chinese puts its results in conflict with all the Amerindian admixture in West Eurasian literature we've been discussing.

Why? Amerindians have 70% East Eurasian and 30% West Eurasian ancestry... if they *weren't* closer to Chinese that would be weird.

All the recent genomic and ancient DNA data supports it

Show me some published papers?

I explained to you how to read those charts. There's no discrepancy.

German Dziebel on another thread, in response to "East Asians, C/S Asians, Europeans, LB etc. all get different MA-1 scores (EDF 5d) to their Amerindian scores (EDF 5a). The two are clearly not the same.": "Sure, MA-1 is West Eurasian-admixed." - there's a major discrepancy and you know it.

East Asians diverged from Amerindians. West Eurasians diverged from Amerindians. They were isolated from each other since the split. They are different from each other but equally descended from Amerindians.

Except that they *aren't* equally "descended" from Amerindians at all - East Asians are much closer, meaning West Eurasians had already diverged long before Amerindians and East Asians did... and hence the problem with Amerindian admixture into MA-1, it should have given MA-1 higher East Asian affinity as well.

German Dziebel said...

@Tobus

"Weird, I woulda thought it being attested in *Asia* would prove they were in *Asia*."

Of course you would. And they got to Asia from Mars, right?

"Raghavan SI14.6 (pg 92) : Dfreq(Papuan, Han; Sardinian, MA-1) = -0.002±0.005 (Z = -0.36)"

Raghavan, p. 93: "Under a model where MA-1 is from the same lineage as the Sardinians,
the ratio of these two statistics for unrelated populations is expected to be 1.0, but
East Asians and Oceanians were both observed as being closer to MA-1 than to the
Sardinian (Figure SI 27)."

"No, it's not - this "flawed assumption" is a reality directly dictated by the data. TreeMix first uses the data to build the maximum likelihood tree, then finds the admixture that best fits the data to that tree, the tree is not preconceived or assumed. "

It is, though, as any admixture is incompatible with prior clustering by definition. if you start with a cluster and add admixture, you just keep generating artificially admixed populations. It's self-fulfilling prophecy.

"Why? Amerindians have 70% East Eurasian and 30% West Eurasian ancestry... if they *weren't* closer to Chinese that would be weird."

In Schiffels, Amerindians and Chinese share a node. According to Raghavan, admixture with West Eurasians happened after the split from East Asians. Nothing of this sort is depicted in Schiffels. Schiffels shows that West Eurasians, East Asians and Amerindians all share a deeper node, which would make them all equidistant from each other in Raghavan. But they are not. In Raghavan, West Eurasians are closer to Amerindians than to East Asians. In Schiffels, West Eurasians should share more drift with East Asians than with Amerindians as Amerindians are shown to have an additional bottleneck, while East Asians don't. The two studies are completely inconsistent with each other. Very typical for geneticists!

"German Dziebel on another thread, in response to "East Asians, C/S Asians, Europeans, LB etc. all get different MA-1 scores (EDF 5d) to their Amerindian scores (EDF 5a). The two are clearly not the same.": "Sure, MA-1 is West Eurasian-admixed." - there's a major discrepancy and you know it."

It's not called "discrepancy." It's meaningful difference. Of course, you wouldn't know it.

Tobus said...

@German:
Of course you would. And they got to Asia from Mars, right?

No, from Africa (duh!)

but East Asians and Oceanians were both observed as being closer to MA-1 than to the
Sardinian


C'mon German! We're talking about MA-1 not being closer to East Asians than to Papuans, nothing to do with being closer to both of them than Sardinians. MA-1 has no discernable East Asian affinity over Papuan, something at odds with your theory of him receiving Amerindian admixture.

It is, though, as any admixture is incompatible with prior clustering by definition. if you start with a cluster and add admixture, you just keep generating artificially admixed populations. It's self-fulfilling prophecy.

"By definition"? Really? Which definitions are you using then? Can you define the difference between 70% "ancestry" and 70% "admixture"?

Clustering first maps the primary genetic associations ("ancestry"), adding "admixture" then allows inclusion of secondary genetic associations. The algorithms and methods in TreeMix have been rigorously tested and repeatedly confirmed as accurate in both human and non-human contexts, if you genuinely think you've found a flaw I suggest you write it up and get it published.

It's not called "discrepancy." It's meaningful difference

I see. My bad. f3 stats (and f4 to some degree) ignore non-related gene flow, and yet still show "meaningful differences" in Amerindian affinities to various other populations - something we wouldn't expect from genuine outgroup.