July 22, 2010

More Uniform Sampling of Human Genetic Diversity (Xing et al. 2010)

Some observations on the paper:
New World populations (Totonac and Bolivian) are placed between Nepalese and Kyrgyzstanis, indicating higher affinity of these American samples to central Asians than to eastern Asians.

This is more likely an artifact of the mixed (Caucasoid-Amerindian) ancestry of these American samples, rather than an indication of their Central Asian origin, as the authors seem to believe. This is an important caveat, as American Indians and Central Asians are "pulled together" by their shared West Eurasian ancestry of post-Columbus and Neolithic/Chalcolithic Age origin respectively, and correspondingly "pulled away" by Mongoloids proper from East Asia who lack that admixture.

Polynesians also deviate from East Asians towards Europeans, less strongly than Central Asians, and this reflects low-level admixture between ancestral Polynesians and colonial-era Europeans.


The Eurasian PCA is interesting:
At the sub-continental level, we focus first on Eurasia, where most of our samples have been selected (Figure 4A). Overall, PC1 and PC2 mainly reflect the geographic distribution of the populations, with the majority of genetic variation accounted for by their locations. PC1 (accounting for 62.7% of the variance) reflects an east-west gradient, while PC2 (3.3% of the variance) reflects a north-south gradient.
There is absolutely no reason (based on geographical distance) for PC1 to account for twenty times more variance than PC2. PC1 reflects the racial contrast between Caucasoids and Mongoloids, while PC2 reflects the much weaker latitudinal adaptation and south-to-north spread of humans into the higher latitudes.

Another thing to notice is how tightly clustered Caucasoids are, from the Atlantic to Iraq (a distance of about 4,000km), which is -conservatively- about half the distance between Pakistan and South India. This is due to the fact that South Asians were formed by admixture of two elements: an extraneous Caucasoid one and an indigenous Paleo-Indian one. Notice also that this variable admixture (highest Caucasoid component in Brahmins) is not really compatible with an indigenous origin of the caste system, as has been proposed on non-scientific grounds.

More spread (given geographical distance) is also observed in Central Asia and Southeast Asia, and this is explained by relatively recent admixture between Caucasoids and Mongoloids (in the former) and Paleo-Indian-like morphological "Australoids" and Mongoloids (in the latter).


The results of ADMIXTURE analysis (for Eurasian individuals) are presented in graphical format in the paper itself, for (K=7).

Not much to comment on this that hasn't been seen before:

One observation is the existence of some "red" West Asian component in the N. European sample, which is not found in Slovenians. This may parallel the peculiarity of the Caucasoid components observed for Russians and Lithuanians recently, although the several Caucasoid components detected in that study are folded into 2 in the current one.

Notice also, how "red" is the main extraneous component in Indian Brahmins. As expected, even Brahmins are predominantly of "indigenous" origin, as these Brahmins are from Tamil Nadu and Andhya Pradesh, and not from North India. The West Asian affiliations of the main Caucasoid component are evident, and agree with Behar et al. (2010) where South Asians had a major overlap with West Asians (light green) and a minor one with Europeans (dark blue). In this paper, with a lower K the different European and West Asian subclusters are not visible.


The most interesting part of the study -for me- was the inclusion of three novel African samples, the Luhya, Alur, and Hema. Notice the blue component in these people, which resolves partially to orange at K=12. This is an indication of Eurasian affinities that are mostly lacking in other black Africans.

The Luhya are Bantu speakers from Kenya, so they are not indigenous to East Africa, but have probably picked up some native East African ancestry from their non-Bantu neighbors.

The Hema are from the Democratic Republic of Congo, but they are Nilo-Saharan pastoralists. Their fairly noticeable West Eurasian component may reflect origins outside the Congo. Are these another member of the non-Bantu pastoralists expanding from East Africa to the south? It would be interesting to take a look at these people's Y chromosomes.

All in all, a very interesting paper which adds important new populations to the discussion of human origins. Also of note, the free availability of the paper's genotype data and supplementary material at the Jorde Lab.

Genomics doi:10.1016/j.ygeno.2010.07.004

Toward a more uniform sampling of human genetic diversity: A survey of worldwide populations by high-density genotyping

Jinchuan Xing et al.

High-throughput genotyping data are useful for making inferences about human evolutionary history. However, the populations sampled to date are unevenly distributed, and some areas (e.g., South and Central Asia) have rarely been sampled in large-scale studies. To assess human genetic variation more evenly, we sampled 296 individuals from 13 worldwide populations that are not covered by previous studies. By combining these samples with a data set from our laboratory and the HapMap II samples, we assembled a final dataset of ~ 250,000 SNPs in 850 individuals from 40 populations. With more uniform sampling, the estimate of global genetic differentiation (FST) substantially decreases from ~ 16% with the HapMap II samples to ~ 11%. A panel of copy number variations typed in the same populations shows patterns of diversity similar to the SNP data, with highest diversity in African populations. This unique sample collection also permits new inferences about human evolutionary history. The comparison of haplotype variation among populations supports a single out-of-Africa migration event and suggests that the founding population of Eurasia may have been relatively large but isolated from Africans for a period of time. We also found a substantial affinity between populations from central Asia (Kyrgyzstani and Mongolian Buryat) and America, suggesting a central Asian contribution to New World founder populations.

Link

62 comments:

Maju said...

"This is more likely an artifact of the mixed (Caucasoid-Amerindian) ancestry of these American samples"...

No in the supplementary material it is clear that the PC defines an specific Native American dimension after exclusion of Africans (fig. 3D). I am neither too persuaded of their "Central Asian" reasoning (though they may be right but only accidentally, not because it's well reasoned) but it is not because Amerindians are not clearly specific: they are.

Their flaw is not understanding that mixed populations such as Kyrgyzes and Nepalis appear in the NJ tree as higher in the tree but that's an artifact of admixture and nothing more. Cavalli-Sforza, to whom they aim to imitate with these autosomal phylogenies, already acknowledged that such artifacts are common.

I don't know why aren't you posting the NJ tree because that's really the crux of the paper.

As for East Africans, they just seem to be most directly related to Eurasians in general, as shown in this NJ tree. This is totally consistent with all we know of the coalescence in Africa before the OoA via haploid genetics. The Hema by the way speak either Hema, a Bantu language, or Lendu a Nilo-Saharan language borrowed from a neighbor people of the same name. Their language is closely related with Bantu-speaker pastoralists from Nkore (Uganda). The only true Nilo-Saharan speakers in that East African sample are the Alur, closely related to Kenyan Luos.

Dienekes said...

No in the supplementary material it is clear that the PC defines an specific Native American dimension after exclusion of Africans (fig. 3D). I am neither too persuaded of their "Central Asian" reasoning (though they may be right but only accidentally, not because it's well reasoned) but it is not because Amerindians are not clearly specific: they are.

As usual, you are not making any sense, because figure 3D does nothing to support the idea that the unadmixed component of Native Americans is closer to Central rather than East Asians.

Indeed, Totonac appear to be about 20% Caucasoid, so the unadmixed Native Americans can be inferred to lie about 20% to the left-bottom on the European-Totonac axis which places them exactly in the same position as East Asians on PC1 which reflects East-West Eurasian differentiation.
I don't know why aren't you posting the NJ tree because that's really the crux of the paper.

Because these populations did not evolve tree-like from a common ancestor. NJ is useful to determine phylogeny when there is no significant lateral gene flow, and this is clearly not the case here. They shouldn't even have put an NJ tree in the paper, or they should' ve followed a strategy similar to the Indian paper where they determined separately the origins of the ancestral components of admixed populations.

Polak said...

Haha.

The European population with the least amount of red are the CEU (Utah Americans of Northern and Western European descent), who always cluster in the middle of Northwestern Europe.

Then come the Slovenians, then, with about the same amount of red, the North Europeans of unknown origin (falsely described by the authors as of Utah American descent)...and then the Tuscans.

Scandinavians would have even less of the red than the CEU.

Andrew Oh-Willeke said...

"The Luhya are Bantu speakers from Kenya, so they are not indigenous to East Africa, but have probably picked up some native East African ancestry from their non-Bantu neighbors.

The Hema are from the Democratic Republic of Congo, but they are Nilo-Saharan pastoralists. Their fairly noticeable West Eurasian component may reflect origins outside the Congo. Are these another member of the non-Bantu pastoralists expanding from East Africa to the south? It would be interesting to take a look at these people's Y chromosomes."

Two African populations in Northern Cameroon (in two different language families) are also outlier populations genetically, presumably from ancestors who followed the Nile to its Westernmost sources and then made a short hop to the Lake Chad Basin.

These particularly new outlier groups appear to have affinities that would be consistent with ancestry from the same or similar waves of migration into Africa via the Nile.

Onur Dincer said...

It seems Mongolians have genetically much more in common with Central Asian Turkics than with East Asian peoples.

Anonymous said...

I don't have any criticisms of this paper. Usually with genetic studies I see the hidden agenda behind them. And it is good the mtDNA and Y chromosome haplogroups have not been emphasized. Haplogroups are just so hackneyed, old hat, and frankly misleading when you really want to know the ancestral populations that made up various ethnic groups existing today.

I being European have a Euro bias, not Eurocentric just more interested in European genetics than the rest of humanity. It is very good to see some other groups involved in the studies other than the typical cast of ethnic groups. It is interesting to see that the Iraqi Kurds have a definite "Indian" admixture which I suppose confirms the linking of Elamite to Dravidian languages.

In a sense there is no surprises. What we can see with our eyes is being confirmed by genetic results. Thais for instance have a visible South Asian component to their phenotypes. What the old style physical anthropologists said about various NE Africans is being confirmed also.

I am hoping for another study with more players from Europe and West Asia. I have had enough of the regular cast.

Maju said...

"As usual, you are not making any sense"

As of late you are becoming very nasty with your readers Dienekes. And I'm every day less interested in reading/commenting here. You may have noticed already.

"... because figure 3D does nothing to support the idea that the unadmixed component of Native Americans is closer to Central rather than East Asians".

I am not saying that. I am saying that Native Americans do not show West Eurasian admixture but clearly look a totally different third group in the Greater Eurasian context.

"Indeed, Totonac appear to be about 20% Caucasoid".

Where do you see that? Totonac appear very "pure" at K=14. Some Bolivians do show some admixture but others do not.

You cannot measure admixture in the PC graph as such, you need some support from something else like cluster analysis, which in this case says: no meaningful admixture in Totonac.

"Because these populations did not evolve tree-like from a common ancestor. NJ is useful to determine phylogeny when there is no significant lateral gene flow, and this is clearly not the case here".

Fair enough. However the authors emphasize that the production of these trees and the information they provide (which is nicely consistent in the great patterns with what haploid genetics seem to say) is a central part of their effort, in which they emulate Cavalli-Sforza's quite productive work of the 1990s.

However they do seem to fail to understand that, as you say, admixture can cause major distortions in the tree and that's why I do not agree with them when they claim a Central Asian affinity for NAs. But what I think are misplaced by admixture are not Amerindians, certainly not Totonac, but Kyrgyzes and Nepalis.

princenuadha said...

Its interesting that the middle eastern component is smaller in Slovenians than in the Northern Europeans. I suppose Slovenia was somewhat shielded from the neolithic migrations.

I also find it interesting that the Thai have a smaller east Asia component than the Cambodians. Language/culture and geography had me thinking it would have been the other way around. What I notice about about Thailand, Cambodia, and Vietnam is that their affinity towards the Chinese increases as you move towards China along the coast, but not necessarily absolute distance. maybe that is suggestive that the migrations mostly took place along the coast. (however I don't know what the coastline was like when the migrations took place)

Onur Dincer said...

Indeed, Totonac appear to be about 20% Caucasoid

How do you know that?

Unknown said...

Th most interesting thing for me is that this Figure A looks like the traces of the original Eastern Migration from India. Fascinating.

Dienekes said...

Indeed, Totonac appear to be about 20% Caucasoid

How do you know that?


Look at the supplementary table.

I am not saying that. I am saying that Native Americans do not show West Eurasian admixture but clearly look a totally different third group in the Greater Eurasian context.

As usual you are wrong, as is proven by the supplementary table which can be found at the Jorde Lab, which clearly show European admixture in Totonac in the ~20% range, and even greater (and more variable) admixture in Bolivians.

Onur Dincer said...

as is proven by the supplementary table which can be found at the Jorde Lab, which clearly show European admixture in Totonac in the ~20% range, and even greater (and more variable) admixture in Bolivians

But is that ~20% really completely or near completely from Europeans? I have doubts about that.

Onur Dincer said...

I think a mtDNA and Y-chromosome haplogroup analysis (maybe only that) on Totonac would resolve the origin of ~20% issue.

Maju said...

"Look at the supplementary table".

That beats me: why do you choose to look in a K=3 when you have K=4 and K=12 available?!

At K=3 the Amerindian specific cluster is not yet defined, so they cluster with other gorups. But that's just an artifact.

It may be useful to discern the overall deep ancestry in the context of Paleolithic Eurasia, assuming the simplified scenario of two branches (the third one represents the African genetic pool or more precisely the African genetic pool that is sufficiently distant from the proto-Eurasian one). But that's about it.

princenuadha said...

@ponto, What do you mean by hidden agenda?

princenuadha said...

Also, I wonder if the absence of a southeast Asian component in the Japanese is suggestive that the ancestors of the modern Japanese came to the islands almost entirely before the Neolithic (basically they are Paleolithic). Most the rest of Asia has an SE component and SE Asia is thought to be where rice cultivation started and often associated with Neolithic expansions in the far east.

Its too bad I don't get to see Korea

Dienekes said...

But is that ~20% really completely or near completely from Europeans? I have doubts about that.

What other likely source do you envision for Americans?


That beats me: why do you choose to look in a K=3 when you have K=4 and K=12 available?!

Because at K=3 it is revealed that the Native American cluster is the product of admixture between Caucasoids and Mongoloids.

aargiedude said...

So how do we explain R1a's presence in Europe and India? R1a, even after exluding the homegrown M458+ Slavic lineages, makes up 15% of European y-dna. It makes up 20% of India's y-dna, and 20% to 30% in Pakistan. But Europe has virtually 0% of the Indo-Pakistani dark blue cluster, and Indo-Pakistanis have 0% to 3% of the European dark orange cluster.


Totonac appear to be about 20% Caucasoid

America and East Asia may have shared an ancestral origin, but afterwards they were separated for 15,000+ years, they each formed their own genetic bloc with their own genetic histories. The closeness between them in PC1 shows their ancestral common origin, but the significant separation belies their subsequent 15,000 years of separation, in which each region drifted to its own tune and developed its own set of AIMs independently of the other region. This is confirmed by the cluster analysis, in which we can easily see that Totonac have virtually 0% European ancestry.


Dienekes: Indeed, Totonac appear to be about 20% Caucasoid
Honest question: How do you know that?
Dienekes snappy answer: Look at the supplementary table.

Oh, brother, let me answer you on Dienekes' behalf. Supplementary pdf, page 6, figure 3D. Look at PC1. The Totonac are 20% of the distance between Chinese/JPT and Europeans.


and even greater (and more variable) admixture in Bolivians

From yhrd. Bolivians have 20% European y-dna, and of course you can imagine they must have virtually 0% European mtdna, resulting in 10% European ancestry, much less than we would conclude from your interpretation of the PC1 graph. On the other hand, the cluster chart of the study, at K=12, shows Bolivians to have 8% autosomal European ancestry, a perfect fit with the sex-chromosome estimate.

why do you choose to look in a K=3

Maju, are you talking about a PC graph? Because I don't see a cluster chart with K=3 (just K=4 and K=12).

Unrelated: what's that 3D image on page 8 of the supplementary pdf?

PS: Just in case, I'm going to repeat the first thing I wrote, which is what I'm really interested in:

So how do we explain R1a's presence in Europe and India? R1a, even after exluding the homegrown M458+ Slavic lineages, makes up 15% of European y-dna. It makes up 20% of India's y-dna, and 20% to 30% in Pakistan. But Europe has virtually 0% of the Indo-Pakistani dark blue cluster, and Indo-Pakistanis have 0% to 3% of the European dark orange cluster.

Maju said...

"Because at K=3 it is revealed that the Native American cluster is the product of admixture between Caucasoids and Mongoloids".

You don't seem to understand how cluster analysis work. It's just an illusion that vanishes as soon the Amerindian specific cluster is created. There is still some minor WEA admixture, specially among some Bolivians, but that's it.

The shallower the K value is not better but, normally, much worse.

"What other likely source do you envision for Americans?"

There are no sources at K=3, if anything it might (doubts included) of the ultimate UP genesis of the proto-Amerindian genetic pool, whose Y-DNA (Q) is probably of West Eurasian origin for the most part (but whose mtDNA is almost exclusively East Asian). So maybe proto-Amerindians in Siberia, some 20 or 30,000 years ago had that "admixture" but then they clearly coalesced apart onto their own exclusive unique genetic pool, which shows up as early as K=4 (which is very shallow, considering that American Natives are not oversampled at all).

Would it be recent European admixture, it'd show up at K=14, as does in fact in some cases. But that's totally different from what happens at K=3.

Onur Dincer said...

What other likely source do you envision for Americans?

Native Americans have been isolated from other Mongoloids (including proto-Mongoloids) for quite a long time. This long isolation + bottlenecks in either side of the Pacific (especially in the American side) + adaptation to different environments have led to significant racial differences between East Asian and American Mongoloids that are apparent from phenotypes. So the traditional three-race system may be insufficient for explaining the situation at American Mongoloids. Look at K=4, Native Americans have a different component from the major Asian Mongoloid component, while Pygmies and Khoisans are virtually identical to proper Blacks, and Polynesians are virtually identical to proper Asian Mongoloids. Long-term zero contact between Asian and American Mongoloids may be the biggest reason of this strange situation at Ks.

Polak said...

"Its interesting that the middle eastern component is smaller in Slovenians than in the Northern Europeans. I suppose Slovenia was somewhat shielded from the neolithic migrations."

The only true North Europeans in this study are the CEU (Utah Americans of Northern and Western Euro descent). They carry the least amount of red, followed by the Slovenians.

The set labeled North European by the authors here looks like the HGDP-CEPH French set from Lyon. In the study they made a mistake of identifying it as "Utah American", which is actually the CEU. But clearly it's not North European, but from somewhere with South Euro admix, like France.

aargiedude said...

Link to abstract

Population structure and genome-wide patterns of variation in Ireland and Britain (O'Dushlaine, 2010)

Anyone have access to this study? I'd like to know if they calculated FST distances between the samples, which include novel locations: Portugal and Bulgaria, and also Sweden, England, Scotland, and Ireland. The study has supplementary info, so if anyone has access make sure to check in that part, too.

Ebizur said...

Dienekes said,

"The Hema are from the Democratic Republic of Congo, but they are Nilo-Saharan pastoralists. Their fairly noticeable West Eurasian component may reflect origins outside the Congo. Are these another member of the non-Bantu pastoralists expanding from East Africa to the south? It would be interesting to take a look at these people's Y chromosomes."

Wood et al. 2005 have tested the Y-DNA of 18 Hema and 9 Alur individuals from the Democratic Republic of the Congo:

Hema (DRC; Linguistic affiliation is listed as Niger-Congo > Bantu)
7/18 = 38.9% E2a-M41
5/18 = 27.8% E1b1b1-M35(xE1b1b1a-M78, E1b1b1b-M81)
3/18 = 16.7% E1b1a-P1(xE1b1a7-M191)
2/18 = 11.1% E1b1a7-M191
1/18 = 5.6% A3b2-M13

Alur (DRC; Linguistic affiliation is listed as Nilo-Saharan > Nilotic)
6/9 = 66.7% E2a-M41
2/9 = 22.2% A3b2-M13
1/9 = 11.1% E1b1a7-M191

The small sample sizes are unfortunate.

RJ said...

Funny. The plot corresponds to almost the exact geographical location of the groups involved in the study (if superimposed on a map).
So - Chinese are from China and Europeans are from Europe, Pakistanis are from Pakistan and South Indians are from South India. Very revealing. Wouldn't have known. :D

Dienekes said...

So how do we explain R1a's presence in Europe and India? R1a, even after exluding the homegrown M458+ Slavic lineages, makes up 15% of European y-dna. It makes up 20% of India's y-dna, and 20% to 30% in Pakistan. But Europe has virtually 0% of the Indo-Pakistani dark blue cluster, and Indo-Pakistanis have 0% to 3% of the European dark orange cluster.

R1a originated among Caucasoids, so there is no problem with the absence of the Indo-Pakistani cluster in Europeans.

>> Totonac appear to be about 20% Caucasoid

America and East Asia may have shared an ancestral origin, but afterwards they were separated for 15,000+ years, they each formed their own genetic bloc with their own genetic histories. The closeness between them in PC1 shows their ancestral common origin, but the significant separation belies their subsequent 15,000 years of separation, in which each region drifted to its own tune and developed its own set of AIMs independently of the other region. This is confirmed by the cluster analysis, in which we can easily see that Totonac have virtually 0% European ancestry.


If Amerindians separated from Eurasians 15,000 years ago and there was no European admixture since, then we do not expect them to be closer to Europeans than they are to East Asians (as they are).


Oh, brother, let me answer you on Dienekes' behalf. Supplementary pdf, page 6, figure 3D. Look at PC1. The Totonac are 20% of the distance between Chinese/JPT and Europeans.

Not surprising, as they are ~20% Caucasoid at K=3.

Maju said:

You don't seem to understand how cluster analysis work.


This is the joke of the century.

I want to see your explanation of why they are shown to have 20% admixture, and they are 20% closer to Europeans in the PCA.

Onur Dincer said...

I want to see your explanation of why they are shown to have 20% admixture, and they are 20% closer to Europeans in the PCA.

OK, but how do you explain the very early (at K=4) and almost complete separation of East Asians and Native Americans from each other in the ADMIXTURE analysis while even Blacks proper, Khoisans and Pygmies separate from each other later at Ks (and even at that Khoisans and Blacks proper never completely or almost completely separate from each other and the Alur appear to have significant genetic relationship with Pygmies)?

Because of the above factors, I think the "Caucasoid affinity" in the Totonac and genetically pure-looking Bolivians may not have anything to do with the European colonialism (of course maternal and paternal haplogroup tests would clarify the picture more, maybe Ebizur can help here.:D), rather it may be due to a Paleolithic connection between the ancestors of today's Caucasoids and the still Asia-dwelling ancestors of Native Americans. Also the thousands of years of isolation between East Asians and Native Americans led to completely separate evolution of these two originally closely-related groups and in the meantime East Asians may have diverged from Caucasoids more than Native Americans did (not so unbelievable given the very limited - both by quantity and territory - contact between Caucasoids and East Asians up until today).

In conclusion, there are many possible ways to explain the 20% issue. The correct answer may be a combination of these possibilities.

Polak said...

"So how do we explain R1a's presence in Europe and India? R1a, even after exluding the homegrown M458+ Slavic lineages, makes up 15% of European y-dna. It makes up 20% of India's y-dna, and 20% to 30% in Pakistan. But Europe has virtually 0% of the Indo-Pakistani dark blue cluster, and Indo-Pakistanis have 0% to 3% of the European dark orange cluster."

There's no reason why Y-DNA should correlate with ADMIXTURE clusters, especially if the men who migrated took local wives of a very different genetic character.

But if you look at the Eurasian ADMIXTURE map, R1a1a originated during the Neolithic in the Near East (Red cluster). It then moved into Europe and South Asia.

However, there was then a back migration of R1a1a from Central Europe to Asia, north of the Black Sea. That migration took the Gray cluster from Europe to the Kyrgyzstani and Nepalese there, and also to a much lesser extent to the South Asians.

BTW, That's why we're now seeing R1a1a mummies in burial Kurgans across Central Asia, with North/Central European skulls and pigmentation SNPs.

German Dziebel said...

"The average haplotype heterozygosity is significantly higher in African populations than non-African populations (Table 1, Wilcoxon rank test p = 1.2x10-4), and haplotype diversity decreases as geographic distance to east Africa increases (Figure 2A, r = -0.76, p = 4.3x10-6). Despite the overall significant correlation, there appears to be little correlation within Africa between haplotype diversity and distance to east Africa (r = -0.13, p = 0.78). Indeed, when African populations were excluded from the analysis, a stronger correlation is obtained (r = -0.94, p = 4.3x10-10, Figure 2A upper panel)."

This means that it's not that homozygosity increases with increasing distance from East Africa but that heterozygosity increases with increasing distance from America (especially, South America, I presume). So, a cline that's supposed to represent an expansion out of Africa is purely artificial. The data pretty much describes a very simple evolutionary process: from a small population in America to larger and larger populations further and further away from the new World. As populations expanded to colonize the Old World, heterozygosity progressively increased. Amerindians have been isolated from before the earliest expansion in the Old World.

So, onur is essentially right in saying that "Native Americans have been isolated from other Mongoloids (including proto-Mongoloids) for quite a long time," and evidence points to even longer period of isolation.

Matt said...

"I am not saying that. I am saying that Native Americans do not show West Eurasian admixture but clearly look a totally different third group in the Greater Eurasian context."

Yes (caveat: a third group that is far closer to the East Eurasian group than the West Eurasian group), BUT the Native Americans in this sample are Euro (mainly) admixed. Why? Even with additional Native American specific PC dimensions they are way closer to West Eurasians than they should be based on what we've seen from prior Native American samples (like the HGDP) and, crucially, closer to West Eurasians than East Eurasians are. Admixture is really the only way to explain this.

"What other likely source do you envision for Americans?"

Africans? But obviously not independently from Europeans.

Onur Dincer said...

and even at that Khoisans and Blacks proper never completely or almost completely separate from each other

On a second thought, those Blacks proper who have Khoisan component at K=12 (Sotho/Tswana, Nguni and Pedi) are all the ones that live in southern Africa and thus probably have significant Khoisan admixture. Besides, the rest of the Blacks proper analyzed in this study (all of them are away from southern Africa) do not have any Khoisan component.

Onur Dincer said...

On a second thought, those Blacks proper who have Khoisan component at K=12 (Sotho/Tswana, Nguni and Pedi) are all the ones that live in southern Africa and thus probably have significant Khoisan admixture. Besides, the rest of the Blacks proper analyzed in this study (all of them are away from southern Africa) do not have any Khoisan component.

Of course, that doesn't diminish my main argument about the origin of the Totonac (who look genetically almost pure natives according to K=4) and pure Bolivians and pure Native Americans in general.

Also the thousands of years of isolation between East Asians and Native Americans led to completely separate evolution of these two originally closely-related groups

On the other hand, East Asians, or at least most of them, may already have been quite divergent from the ancestors of Native Americans even before their migration to the Americas.

Onur Dincer said...

Yes (caveat: a third group that is far closer to the East Eurasian group than the West Eurasian group), BUT the Native Americans in this sample are Euro (mainly) admixed. Why? Even with additional Native American specific PC dimensions they are way closer to West Eurasians than they should be based on what we've seen from prior Native American samples (like the HGDP) and, crucially, closer to West Eurasians than East Eurasians are.

You may be totally right here...

Admixture is really the only way to explain this.

...but I don't think it (admixture with Europeans) is the only way to explain the %20 issue, as I explained in my previous posts.

Marnie said...

"Of course, that doesn't diminish my main argument about the origin of the Totonac (who look genetically almost pure natives according to K=4) and pure Bolivians and pure Native Americans in general."

The idea of indigenous genetic "purity" in Bolivians or the Totonac belies the written historical record of the last five hundred years.

In the case of the Totonac, it is in direct contradiction to another study, published on this site in May, which shows significant European/Indigenous admixture at varying levels across the Mexican population.

My overall visual and cultural impression of various Native American people is that there must have been two or more Beringia migrations into the Americas. There is significant difference even in British Columbia between various native groups, let alone between Eastern and Western Canada or between Canada, the US, Mexico and South America.

It is strange and very glib to see sweeping generalisations about native groups in the Americans, who themselves must be admixtures of various Beringia migrations over the last 20 thousand years.
(Ignoring European and African admixture, which complicates the picture even more.)

Certainly, the genetic identity of American natives (Northern and Southern) is at least as complex as that of Europe, over the same timeframe.

In the end, it may vary well be the case that Bolivians originate from a specific locality in Eurasia, but at this point, I do not see that the authors of this paper have sufficient basis for their assertion that Bolivians originate in Central Asia.

It is very tempting to make these kind of guesses, but I do think that a higher standard of statistical scrutiny is on order.

aargiedude said...

I found some mtdna studies of Bolivians, and as expected they have either 0% or 1% non-American haplogroups. I stated previously that according to the yhrd samples of Bolivians, they have 20% West Asian y-dna. So Bolivian (Indians) are 10% West Asian according to their sex-chromosome results. That's what the autosomal cluster chart shows, also (8% to be precise).

terryt said...

"it may be due to a Paleolithic connection between the ancestors of today's Caucasoids and the still Asia-dwelling ancestors of Native Americans".

I suspect that is exactly the case. The Native Americans were already a mixture by the time they arrived.

Onur Dincer said...

In the case of the Totonac, it is in direct contradiction to another study, published on this site in May, which shows significant European/Indigenous admixture at varying levels across the Mexican population.

Does that study say anything about the Totonac? Also which study are you talking about exactly?

Onur Dincer said...

The idea of indigenous genetic "purity" in Bolivians or the Totonac belies the written historical record of the last five hundred years.

And what is written in the historical record about the genetic purity or impurity of the Totonac?

Marnie said...

Onur,

We've had many interesting discussions and I respect your thoughtfulness. You and I, we've run across this before, this discussion about race. I've mentioned that the notion of "racial purity" or "impurity", to me, is somewhat ridiculous in a scientific discussion. We only need to look to the latest paper on the Malana to see how unscientific such a term can be. The Malana, a genetically isolated population, is comprised of at least three y-dna haplogroups, likely originating from different parts of Western Eurasia. That doesn't even get to their autosomal DNA.

The people of the Americas, or the people of Europe, for that matter, are also not comprised of a single haplogroup.

That's not to say that there are not genetic isolates and identities. However, the term "racial purity" is not a precise or meaningful scientific term.

I would respectfully say that I am uncomfortable engaging in any discuss about genetics when that term is used.

Regarding the paper on population structure in Hispanics:

http://dienekes.blogspot.com/2010/05/population-structure-in-hispanics-bryc.html

It's instructive to read the histories of Cortes, Pizarro, and Bolivar, for starters, to piece together a picture of the genetic complexity of Central and South America.

I'm sure that La Malinche was only one of many thousands of women who where mothers of children with European fathers:

La Malinche:

http://en.wikipedia.org/wiki/La_Malinche

Maju said...

Marnie: you're talking of mainstream Mexicans, usually called Mestizos (essentially in Mexico anyone who is not Indio is Mestizo, either by blood or culture) but there are still rather unmixed Native Americans in Mexico, of which this Totonac are a sample. Not all Mexicans are the same.

The cluster graph and data is very clear about that and, at least in this case, Onur is correct IMNSHO. When Native Americans are admixed, that does happen, it's very evident at all levels. Cluster algorithms are ideal for that but first you have to detect the two mixing clusters and the Native American cluster here only arises at K=4.

Onur Dincer said...

Marnie,

You have answered none of my questions and are instead resorting to discussions of unrelated subjects. That purity and impurity I asked was purity and impurity from outside genetic admixture (mainly European) within the last 500 years (i.e., beginning from Columbus), nothing else. BTW, the Native Americans (not Hispanics/Latinos) tested in the study you linked are fairly pure in this sense according to the frappe analysis (except, to an extent, the Maya and Quechua). But that study doesn't include the Totonac or Bolivians or even any East or Central Asian population and is thus irrelevant to our discussion. My whole point is that a similar purity can be proposed for the Totonac and, to a lesser extent, Bolivians tested in this study based on the ADMIXTURE analysis results. I don't understand what you disagree with on this issue (terminology or arguments themselves?).

Onur Dincer said...

You have answered none of my questions

except giving link to that study of course

Onur Dincer said...

That purity and impurity I asked was...

also the purity and impurity I mentioned throughout this thread

Dienekes said...

onur, you continue to disrespect the rules of this blog, by quadruple posting, submitting the same post twice, etc.

This is the last warning.

Dienekes said...

Maju, your irrational opinions in the face of overwhelming evidence to the contrary are becoming tiresome. It is a very bad trait to stick to an opinion after you have been shown evidence that you are wrong.

Amerindian groups that lack European admixture do exist, and the lack of such admixture is evident at K=3.

For example, see Figure 1 in Rosenberg et al. (2002)

http://rosenberglab.bioinformatics.med.umich.edu/papers/popstruct.pdf

Where most Amerindian groups lack any West Eurasian admixture at K=3, while one Amerindian group (the Maya) possess it.

Case closed.

Onur Dincer said...

Marnie: you're talking of mainstream Mexicans, usually called Mestizos (essentially in Mexico anyone who is not Indio is Mestizo, either by blood or culture) but there are still rather unmixed Native Americans in Mexico, of which this Totonac are a sample. Not all Mexicans are the same.

The funny thing is, the paper Marnie linked to clearly differentiates between these two groups:

1) Hispanics/Latinos/Mestizos/Mulattos

2) Native Americans

Off the topic, Marnie, what does being comprised of a single haplogroup have to do with race and racial purity?

onur, you continue to disrespect the rules of this blog, by quadruple posting, submitting the same post twice, etc.

This is the last warning.


Again I am dreadfully sorry, Dieneke, I promise there won't be a next time.

Maju said...

Different algorithms and sampling strategies seem to produce the low level sets at different orders, that's pretty clear. Here we only get the Amerindian sample at K=4 but they are only two populations, in Rosenberg 2002 the Amerindian cluster arises at K=3 (and the East Asian one at K=5) and in Behar 2010, where there are no Amerindians, Africans and West Eurasians only diverge after East Asians do contradicting somewhat the common genetic wisdom that Africans must diverge first (and also Rosenberg's pattern).

In Rosenberg'02, Amerindians are more numerous overall and anyhow only an order of the production of the clusters is found, as East Asians (who could appear also "mixed with Caucasoids" at K=3) reveal themselves as a distinct cluster at K=4.

The comparison is perfectly valid but just in order to understand that low level Ks are meaningless and that sampling strategies are important. There's an equation to estimate the optimal number of clusters but in general it needs to be quite high, specially when we try to understand hugely diverse populations such as Humankind as a whole.

A totally different case is when you want to estimate admixture between two discretely distinct populations such as in African Americans, then K=2 (or K=3 if you also mean to estimate Amerindian input) is enough and further clusters should reveal nothing but subclusters of the two/three matrix populations (IDK stuff like Mandinga and Mozabican).

I'm pretty much disappointed of the low level of the debate anyhow. So bye (unsubscribing).

Onur Dincer said...

Case closed.

In order for the case to be closed, we should see more genetic studies of the Totonac (especially haplogroup studies as European and Native American haplogroups are almost totally different from each other). Rosenberg's study is irrelevant as it doesn't include the Totonac and Bolivians and uses different and outdated methods.

Marnie said...

Returning to the topic at hand, I will state that the final statement of the authors of this paper seems premature.

The authors were careful to use the term "substantial affinity."

It is highly improbable that Kyrgyzstanis walked straight from central Asia and plunked themselves down in the Bolivian Andes.

Living in a Latino part of San Francisco, I often see people from Central America who look to be various flavors of Asian and Central Asian, even Persian. However, I always stop myself from extrapolating to any easy conclusion.

It's interesting, but something that will take many careful studies to unravel.

The real story is probably far more interesting than any easy conclusion.

princenuadha said...

now that native American discussion has been settled, can someone tell me the significance of Japan being almost entirely blue. If Japan were to have a small amount of admixture from Southeast Asia because of Neolithic Migrations (rice agriculturists) would Japan necessarily have green?

I don't really understand these maps. For example even though Northwestern Europe has very little red I know that Europeans at one point or another mostly came from middle eastern populations. So I assume grey and red are distinguished by more recent evolution. But do grey and red have similar variability (are they are the same "level"). If grey exists must there necessarily be a population with all grey.

In the case of Japan does being nearly all blue mean that they just lie on one end of the spectrum in Asia or does it mean they lack a "green people" migration which the rest of Asia got?

Dienekes said...

Different algorithms and sampling strategies seem to produce the low level sets at different orders, that's pretty clear. Here we only get the Amerindian sample at K=4 but they are only two populations, in Rosenberg 2002 the Amerindian cluster arises at K=3 (and the East Asian one at K=5)

That is incorrect. In Rosenberg et al. (2002) there is no "Amerindian" cluster at K=3, but rather a Mongoloid one encompassing both East Asians and Amerindians. This Mongoloid cluster is split off at K=4, and that is when the East Asian cluster emerges, and not at K=5.

onur said:

Rosenberg's study is irrelevant as it doesn't include the Totonac and Bolivians and uses different and outdated methods.

he had previously said:

it may be due to a Paleolithic connection between the ancestors of today's Caucasoids and the still Asia-dwelling ancestors of Native Americans.

Get your theories straightened out. You suggest that Native Americans had ancestors in Asia that were related to Caucasoids. Now that I've proven that many Native Americans do not show this alleged relationship, you are backpedalling without offering an explanation.

If the ancestors of Native Americans were related to Caucasoids, then we might expect to find spurious low-level "Caucasoid" admixture across Amerindian groups. However, that is not the case, and we find it only in _some_ groups and not in others.

Moreover, the great variability of this component in Bolivians proves beyond any doubt that it is not any "Paleolithic" element, in which case it would be uniformly distributed across Bolivians, but rather the product of recent admixture with Europeans.

I also direct interested readers to Auton et al.

http://www.pnas.org/content/early/2010/05/04/0914618107.full.pdf+html

which show once again that some Amerindian groups have Caucasoid admixture while others do not (K=3)

Onur Dincer said...

Get your theories straightened out. You suggest that Native Americans had ancestors in Asia that were related to Caucasoids. Now that I've proven that many Native Americans do not show this alleged relationship, you are backpedalling without offering an explanation.

As I said, that paper neither includes the Totonac nor Bolivians, so the issue is open to speculation. Your quotation from me is just one of the possible scenarios to explain the 20% issue, many other scenarios are possible. So there doesn't need to be a connection with Caucasoids.

Moreover, the great variability of this component in Bolivians proves beyond any doubt that it is not any "Paleolithic" element, in which case it would be uniformly distributed across Bolivians, but rather the product of recent admixture with Europeans.

The Caucasoid component in Bolivians and the Totonac is almost certainly from recent European admixture, I have no objection to that. But it is so insignificant (especially in the Totonac) that it doesn't confirm your "20% European admixture in the Totonac" argument. In the ADMIXTURE analysis many Totonac and Bolivian individuals appear pure Native Americans if the components are to be believed. So the Totonac and Bolivian individuals remotest to Caucasoids on the PCA plot may really be pure Native Americans.

I also direct interested readers to Auton et al.

http://www.pnas.org/content/early/2010/05/04/0914618107.full.pdf+html

which show once again that some Amerindian groups have Caucasoid admixture while others do not (K=3)


I already gave answers to these.

princenuadha said...

replace green with yellow to correct my previous post

Onur Dincer said...

The Caucasoid component in Bolivians and the Totonac is almost certainly...

Here by component I mean components in the ADMIXTURE analysis.

German Dziebel said...

"That is incorrect. In Rosenberg et al. (2002) there is no "Amerindian" cluster at K=3, but rather a Mongoloid one encompassing both East Asians and Amerindians. This Mongoloid cluster is split off at K=4, and that is when the East Asian cluster emerges, and not at K=5."

I see it differently. In Rosenberg, there's an Amerindian cluster at K=2 in which Eurasian and Oceanic clusters are nested. Even Africa has traces of the Amerindian cluster. Then the Amerindian "substrate" slowly recedes at K=3 in Africa and Europe and at K=4 in East Asia and Oceania, then, finally, Oceania breaks off from the East Asian cluster. Once again, this is the real evolutionary history of human populations: from small isolated demes to large aggregated populations. At first, as John Locke once put it, all the world was America.

"in Behar 2010, where there are no Amerindians, Africans and West Eurasians only diverge after East Asians do contradicting somewhat the common genetic wisdom that Africans must diverge first (and also Rosenberg's pattern)."

Consistent with the non-African origin of modern humans. Usually, Amerindians don't make the published sample if the data shows that they diverged first.

Onur Dincer said...

Aldamiz: Different algorithms and sampling strategies seem to produce the low level sets at different orders, that's pretty clear. Here we only get the Amerindian sample at K=4 but they are only two populations, in Rosenberg 2002 the Amerindian cluster arises at K=3 (and the East Asian one at K=5)

Pontikos: That is incorrect. In Rosenberg et al. (2002) there is no "Amerindian" cluster at K=3, but rather a Mongoloid one encompassing both East Asians and Amerindians. This Mongoloid cluster is split off at K=4, and that is when the East Asian cluster emerges, and not at K=5.

You are both partially wrong. In Rosenberg et al. 2002 the Amerindian cluster emerges at K=4 and the East Asian (Asian Mongoloid) cluster emerges at K=5 (look at when Melanesians and Papuans separate from East Asians). Next time look at the populations more carefully.

BTW, Amerindians' earlier separation from East Asians than Melanesians/Papuans at Ks in Rosenberg et al. 2002 despite Amerindians being closer to East Asians by origin than Melanesians/Papuans are is in line with my contention that complete isolation of Amerindians from the rest of the world led to their genetic divergence from all other humans in a very marked manner.

In this last study (Xing et al. 2010) we again see Amerindians' very early separation from East Asians and the rest of the world at Ks (I think with increases in precision in methods such distinctions and separations will be more clear in the future studies).

Marnie: It is highly improbable that Kyrgyzstanis walked straight from central Asia and plunked themselves down in the Bolivian Andes.

No, what they propose must be that both Central Asians (including Mongolians here) and Amerindians have in their DNA remnants of an ancient population that over time largely disappeared elsewhere in Eurasia (note that yellow component at K=4 can also be found - albeit in much smaller amounts - in many East Asian and even in some European populations). I agree that it is premature to make such a quick conclusion from just one study.

Onur Dincer said...

I see it differently. In Rosenberg, there's an Amerindian cluster at K=2 in which Eurasian and Oceanic clusters are nested. Even Africa has traces of the Amerindian cluster. Then the Amerindian "substrate" slowly recedes at K=3 in Africa and Europe and at K=4 in East Asia and Oceania, then, finally, Oceania breaks off from the East Asian cluster. Once again, this is the real evolutionary history of human populations: from small isolated demes to large aggregated populations. At first, as John Locke once put it, all the world was America.

The Amerindian cluster emerges at K=4, not before, and the extremely small (except one Japanese individual and to a much lesser extent one Oroqen individual and one Pathan individual) traces of the Amerindian cluster (beginning from K=4) in non-Amerindian populations can easily be explained as a result of limited pecision in discenring Amerindian-originated and non-Amerindian-originated sections of the DNA, so those "traces" may not be Amerindian traces, but a result of limited precision in methodology thus a human error. Another explanation is that they may be traces of an ancient population that contributed both to Amerindians and to some Eurasian populations (especially eastern ones) and maybe to a lesser extent also some Africans (see the end of my previous post).

Consistent with the non-African origin of modern humans.

It may instead be a result of the sampling strategy.

German Dziebel said...

"In Rosenberg et al. 2002 the Amerindian cluster emerges at K=4 and the East Asian (Asian Mongoloid) cluster emerges at K=5 (look at when Melanesians and Papuans separate from East Asians)."

The Amerindian cluster didn't "emerge" at K=4. It's been there from K=2, but it's just shrank to a continent-specific one at K=4 because populations outside of America diverged into their own regional clusters, with no contact with America.

"...are is in line with my contention that complete isolation of Amerindians from the rest of the world led to their genetic divergence from all other humans in a very marked manner."

Because Amerindians diverged first and stayed in isolation ever since. The primary split is therefore between the New World and the Old World, not between Africans and non-Africans. It's quite clear in Rosenberg's K=2 that, while the brown cluster (highest frequencies in Africans) is only locally represented in the Americas (Maya), the purple cluster (highest frequencies in America) extends, at low frequencies, through all of the African populations. The frequencies of the purple Amerindian cluster are higher the closer the region is to America geographically.

Onur Dincer said...

German, in a global study nothing can be clear at K=2 as all of humanity can relatively accurately be represented at least with a three-racial system, thus at least at K=3 (pecision increases as K increases). But the most realistic representaion in Rosenberg et al. 2002 is at K=5.

BTW, racially completely pure East Asians (like all of Japanese and overwhelming majority of Han Chinese) in Rosenberg et al. 2002 have western components at K=2 and K=3 and, to a much lesser extent, later at Ks almost certainly due to poor and outdated methodology. That error almost doesn't not occur any more in the latest studies (global or regional) and racially completely pure East Asians appear completely comprised of eastern components at all Ks (see, for instance, Li et al. 2008 - I'll put its link below). So your contention that racially completely pure East Asians have little western components at Ks (especially at early Ks) while racially completely pure Amerindians have none at all Ks is wrong, as both racially completely pure East Asians and racially completely pure Amerindians have no western component at all Ks according to the latest studies, so they are equal in their easternness (Mongoloidness).

Here is the link to the supplement of Li et al. (look at the Fig. S1):

http://www.sciencemag.org/cgi/data/319/5866/1100/DC1/1

Onur Dincer said...

Correction: That error almost doesn't not occur any more

I accidentally made a typo here, I must have written so:

That error almost doesn't occur any more

AWood said...

Realistically I don't foresee the red component being so low in the UK. If that is true representation, then it shows you how quickly an island community could change. I suspect it's closer to the representation given to Belgium/Netherlands. I also question that the Iraqi Kurd is completely red. I guess the results all depend on which SNPs they are using for the study.I'm taking the finer details with a grain of salt.

jes-r said...

The Karitiana and Surui who are 100% pure Amerindians from the Brazilian Amazons show the 'European' peak component at K=3 in the Harappa admix runs (they have one of the most cosmopolitan datasets). So the K=3 'European' in the Totonac is in actually mostly Amerindian.

http://www.harappadna.org/2011/04/reference-3-admixture-k3/

Dienekes said...

http://dienekes.blogspot.com/2010/12/human-genetic-variation-first.html