July 07, 2010

Genetic structure of cattle in Eurasia

The mixed (east-west) affiliation of Mongolian cattle parallels the mixed affiliation of Mongolians themselves. The Caucasoids do not appear to have penetrated east of Lake Baikal, and some Caucasoid-influenced populations that live to the east of that frontier are thought to have originated there. Thus, the evidence from cattle complements that from physical anthropology and human population genetics to fix the border between the Western and Eastern Eurasian spheres of influence.

Anim Sci J. 2010 Jun 1;81(3):281-9.

Genetic diversity and structure in Bos taurus and Bos indicus populations analyzed by SNP markers.

Lin BZ, Sasazaki S, Mannen H.

Abstract

The purpose of this study was to assess genetic diversity, phylogenetic relationship and population structure among nine Eurasian cattle populations using 58 single nucleotide polymorphism (SNP) markers. The calculated distribution of minor allele frequencies and heterozygosities suggested that the genetic diversity of Bos indicus populations was lower than that of Bos taurus populations. Phylogenetic analyses revealed the main divergence between the Bos taurus and Bos indicus populations, and subsequently between Asian and European populations. By principal components analysis, the Bos taurus and Bos indicus populations were clearly distinguished with PC1 (61.1%); however, six Bos taurus populations clustered loosely and the partial separation between European and Asian groups was observed by PC2 (12.5%). The structure analysis was performed using the STRUCTURE program. Distinct separation between Bos taurus and Bos indicus was shown at K = 2, and that between European and Asian populations at K = 3. At K = 4, 5 and 6, Mongolian population showed an admixture pattern with different ancestry of Asian and European cattle. At K = 7, all Bos taurus populations showed each cluster with little proportion of admixture. In conclusion, 58 SNP markers in this study could sufficiently estimate the genetic diversity, relationship and structure for nine Eurasian cattle populations, especially by analyses of principal components and STRUCTURE.

Link

16 comments:

Maju said...

"... parallels the mixed affiliation of Mongolians themselves".

Didn't Hui Li's paper, which you commented here, determined that the apparent West Eurasian component in Mongolians, as well as other more clearly mixed populations of the region (Kazakhs, Uyghurs) is in fact a Central Eurasian specific component and that there is no meaningful trace of specifically European or SW Asian admixture, certainly not in Mongols.

I know you emphasized in the article on Behar's paper that there was an appearance of West Eurasian component in Mongols but the whole thing looks as an artifact of the methods in a context not intended to determine the genetic ancestry of Mongols but of Jews. An artifact that now seems confirmed on light of the Hui Li data.

I think that you are being too one-sided on this issue, really, grabbing the proverbial burning nail instead of facing the facts for what they are.

As for this paper, I can only judge, like you, from the abstract. But instead of choosing to focus on this sentence:

"At K = 4, 5 and 6, Mongolian population showed an admixture pattern with different ancestry of Asian and European cattle".

... I'd focus on the one that follows:

"At K = 7, all Bos taurus populations showed each cluster with little proportion of admixture".

It means that at sufficient depth of analysis the Mongolian cattle (mixed or not, the word itself can become much of a cliché, as only clones are unmixed) shows also a distinctive self-centered genetic pool.

Dienekes said...

Didn't Hui Li's paper, which you commented here, determined that the apparent West Eurasian component in Mongolians, as well as other more clearly mixed populations of the region (Kazakhs, Uyghurs) is in fact a Central Eurasian specific component and that there is no meaningful trace of specifically European or SW Asian admixture, certainly not in Mongols.

You are basically saying nothing here, as there is no magical "Central Eurasian" component, and Mongols are descended partially from Caucasoids.

Caucasoids, however, are divided into several meaningful subgroups, including a Central/South Asian one, which has been known for a long time and is no discovery of the Li et al. (2010) letter

... I'd focus on the one that follows:

"At K = 7, all Bos taurus populations showed each cluster with little proportion of admixture".

It means that at sufficient depth of analysis the Mongolian cattle (mixed or not, the word itself can become much of a cliché, as only clones are unmixed) shows also a distinctive self-centered genetic pool.


How you can arrive at the conclusion that At K = 7, all Bos taurus populations showed each cluster with little proportion of admixture means that there is a distinctive genetic pool is beyond me.

It only means that no meaningful clusters emerge at K=7, and that is why the authors end their analysis there.

Maju said...

"... and Mongols are descended partially from Caucasoids".

Evidence?

The global analysis of Behar did not reach sufficient depth to bring to light the Mongol-Inland China component that is so evident in Hui Li's at k=3. It did detect a "Yakut" component but not the specific-Mongol-Baima one so apparent in Li.

It's a well known phenomenon in cluster analysis that, before the appropiate depth for each specific cluster is reached, the population appears as "admixed" between other components, which then turn into just minor stuff.

In Li's paper, Mongols do appear as somewhat mixed at k=2 but, as the analysis is specifically designed to deal with Asian genetics, as soon as k=3 we can see how this "Western" component nearly vanishes.

"Caucasoids, however, are divided into several meaningful subgroups, including a Central/South Asian one, which has been known for a long time and is no discovery of the Li et al. (2010) letter".

I am not aware of this "Central/South Asian" component (though I'm aware of a bad habit of creating such a sub-tag beginning with Rosenberg's work, where Uyghurs and Hazaras, Central Asian peoples are dumped in the same category as Pakistanis). Can you point me to your source?

In Li's paper South Asian specificity is not found, not because it does not exist but because they are not the main target of the research and are comparatively undersampled. Still, they do not cluster well with Central Asians nor any other West Eurasian population and appear as an illusory mix of all three components.

Polak said...

The Europid component in Western East Asians (inc. Mongols) is at least part North and East European. IBD segment analysis shows this clearly...

http://eurogenes.blogspot.com/2010/03/north-european-admixture-in-east-asia.html

The reason Mongols appear more "Central Asian" than European in Structure analyses, is because they have a more similar mix of East and West Eurasian haplotypes to the former, thus making it easier for them to be confused for other Asians than Europeans, who have very little Asian admix. Structure analyses are, after all, only probability scores of belonging to other test groups, and don't actually show admixture.

Dienekes said...

There are multiple lines of evidence for the Caucasoid ancestry of Mongols.

For example, or here, or here

Maju said...

Dienekes and Polak: your references are about haploid lineages (Y-DNA, mtDNA), which in my understanding can (and probably should in most cases) be very old, from before the formation of regional differences we see in the clusters.

We must remember that R1a, for instance, has been reasonably demonstrated to have originated in South Asia and cannot be just considered a mere signature of any Indo-European flow anymore but something that in many cases should be pre-IE and possibly pre-Neolithic as well.

If Mongolians in the properly designed analysis such as Li's, do not show any European component nor SW Asian nor Indian component, that means that the minor R1a and such among them carried the Central Asian specific component and that's it.

Same for the other "Western" lineages, which actually encompass all lineages involving in the Out-of-India migration by the Western route. The lack of a South Asian specific detected cluster, which we know it does exist, is a weakness of Li's cluster analysis because it would have served for comparison. And it should be quite comparable, but distinct too, to the Khanty one in the overall genetic distance tree, as it would stand roughly between West and East too in a different, surely longer, sub-branch.

So we have a group of "Western" haploid lineages that are nothing but a subset of the "Southern" ones and we also have at least three neatly different genetic clusters between Africa and East Asia: West Eurasian (West Asian and European are close enough to be considered one in this context), South Asian (detected repeatedly in other papers in spite of the low attention this key area gets) and Central Eurasian (finally found by Li and throwing a jar of cold water to the oversimplistic fantasies of Eurasian bilateralism).

Dienekes said...

We must remember that R1a, for instance, has been reasonably demonstrated to have originated in South Asia and cannot be just considered a mere signature of any Indo-European flow anymore but something that in many cases should be pre-IE and possibly pre-Neolithic as well.

The light pigmentation of some of the prehistoric R1a-bearing populations precludes any notion of a "South Asian" origin.

This also agrees with their mtDNA composition which shows the presence in Mongolians of West Eurasian mtDNA haplogroups.

You have taken your Paleolithicism to new levels of absurdity.

Maju said...

From what I can see at Jean Manco's aDNA page, all the relevant claims of light pigmentation in South Siberia/Central Asia east of the Aral Sea are from Keyser 2009, which is not publicly available.

However it's clear from the abstract that the claim is limited to eye color. Browsing SNPedia, and after discarding all unclear SNPs, we are left with four OCA2-related alleles (rs1800407, rs1800401, rs12913832, rs7495174).

None of them says anything too conclusive (raises/lowers probability somewhat) except rs12913832, which seems to be decisive in forming blue eyes (which it causes in 99% of cases, it seems). Furthermore, this SNP is the only one that can be clearly associated with Europeans, specially CEU but also TSI, but have not been studied in living Central Eurasian populations anyhow, where it is also probably abundant.

So maybe means something... or nothing. After all Central Eurasia has been culturally connected with West Eurasia all the time since the early UP, which is my whole point after all, including also connections with East Eurasia resulting in a unique regional cluster of its own.

There are rare East Asians with blue eyes and even red hair (both surely traits arrived from the West at some point) but that says not much about the overall impact of such flows, which seem minor and more likely originated/mediated by Central Asia rather than Europe.

You are emphasizing that Ob Ugrians are mostly Y-DNA N but a reader just pointed me to this paper where it's clear that 63% of their mtDNA is West Eurasian (specifically Western R, as neither I, X or W are found), with the rest looking old NE Asian (A, C, D, F, G and M*). So they do look like a melting pot that coalesced into their own unique autosomal genetic pool.

Would it be only the Ob Ugrians, we could think of a very specific isolation/inbreeding cluster, as happens in other small populations (Druzes, Mlabri, etc.), but the same component is found at quite high levels in Central Asians, so it's not likely to be such a case.

Polak said...

Maju, I didn't make any references to haploid markers. Did you check out my link?

I'm talking about North and East European specific Identity by Descent (IBD) genome wide segments being present in modern western Chinese and Siberian populations. That's got nothing to do with R1a1a, except that they both got there at the same time and with the same people.

Read my posts more carefully next time.

Maju said...

"Maju, I didn't make any references to haploid markers. Did you check out my link?"

My bad. I read it too fast and was confused by your first line reference to R1a. Sorry.

"I'm talking about North and East European specific Identity by Descent (IBD) genome wide segments being present in modern western Chinese and Siberian populations".

I don't understand why a population is considered "donor" or "receiver" in such model. I remember having read and discussed Auton's paper back upon its time and my memory impression is that essentially everybody considered it quite worthless and more confussionist than enlightening.

I would have to delve in it again in order to provide an actualized assessment and I don't know if it's so important after all. The selection of populations was also quite unfortunate You know, the usual: no South Asians, marginal Orcadians as only representative of NW Europe, admixed Mozabites as supposed "missing link" between Africa and Eurasia, etc. It may have some validity but one has to take it with a whole cart of salt considering the huge blanks of the sample specially in key regions, including South Asia and what affects us most in this discussion Central Eurasia (NW Asia).

It may perfectly, it is probably, yet another case of erroneous sampling strategy in which key populations are ignored, totally distorting the results. I'm not too persuaded of the method either but well...

Polak said...

So those North/Central/East European IBD segments aren't really North/Central/East European, because Maju says so.

And the North/Central/East European Y-DNA and mtDNA haplotypes aren't really North/Central/East European either, because, again, Maju says so. These maps are lying, because only Maju knows the truth.

http://i129.photobucket.com/albums/p217/dpwes/Ystr.jpg

http://i129.photobucket.com/albums/p217/dpwes/mtDNA.jpg

And the skulls of the early Kurgan tribes aren't North/Central/East European either. Not because they're not, but because Maju says they can't be. These are lies too, apparently.

http://eurogenes.blogspot.com/2009/12/more-on-scythians-and-their-origins.html

Maju said...

I am not saying what you put in my mouth, Polak. What I am considering is that the Central Eurasian population cluster formed AFTER most of those haploid lineages (and others from East Asia and a few also from the South, like U7, etc.) spread, which may well have happened in the Upper Paleolithic for what I can tell.

Anyhow, I am somewhat surprised at your belligerance because I remember discussing with you about the fraction of Altai-related autosomal DNA (some 15%) found in Finnish and Vologda Russians (Bauchet 2007 and possibly Rosenberg 2005), which you argued it was not strictly East Asian but Siberian. Now that an specifically "Siberian" ('NW Asian', 'Central Eurasian', whatever) component has been found to actually exist, potentially giving you the reason (or some reason at least) after all, do you have also to bounce heads against it?

As for the anthropometric data, I cannot the content of pay-per-view papers I have no access to. I have already discussed the issues regarding pigmentation, which seem to be rather hyped in the Nordicist sense, something that seems to sell better but is often unwarranted.

Maju said...

Btw, Polak, I'm re-reading Hellenthal's paper and it's interesting that in what refers to Yakut and it says:

"We tested the robustness of this inference by putting Orcadians last in the ordering. The Yakut replaced the Orcadians with Sardinians, who are a major donor to the Orcadians. The Hezhen and the Han from Northern China did not acquire new European donors, consistent with the gene flow from Europe being less quantitatively important to these two populations than to the more Northerly Yakut".

Hmm... the Sardinians and not the French nor the Tuscans nor more Russians? Interesting, don't you think, specially because it does suggest that, assuming the model is ok, whatever connection "Orcadians" have with Yakuts may not go through the steppes north of the Black Sea and Caspian but most likely through the south of those seas.

Curiously, in this model (table S2) one of the major "donors" of Sardinians are Uyghurs, ranked only after Palestinians, Tuscans and Druze.

I think that, if you are going to give so much credit to Hellensthal's method and results (not independently replicated so far AFAIK), you should take this faith to its ultimate consequences, which seem to be that the connection is through West Asia and the Mediterranean and not the steppes in the case you argue.

Anonymous said...

Interesting a study on bovines could start a discussion on the genesis of Caucasoids.

The whole crux of the problem is the minor physical differences between Europeans, the lighter Northerners, and not so light Southerners. It would be simpler for all if both European Caucasoid groups became extinct. It would save a lot of paper, computer time and money.

Most of the ideas of Europe's peopling are reasonable and imaginative. I just want someone, anyone actually, to find reasonable well preserved skeletal remains of Paleolithic, Mesolithic and Neolithic Europeans and do definitive DNA tests, so we can all know, and put the fanciful ideas of European's human colonization to rest.

I know Polako believes that the originators of Proto I.E languages were from Eastern Europe, an in situ European genesis for that language group, including those light pigmentation SNPs Maju referred to and the introduction of lactase retention in adults.

Maju believes in some sort of Paleolithic continuation with those primitives having Y chromosome haplogroup R1b and I, and mtDNA haplogroups U and H. It is part of the idea that Europeans were really White until interlopers came from the east with their dark skins and disgusting habits to blacken the European race. Those darkies were haplogroups J, E and G.

All I am saying is whatever idea you believe in, it cannot be conclusively proven, and the way to scotch all argument is to provide the skeletal and dna proof. At the moment, all is just hot air, and slightly racist in tone.

There are a lot of variation among humans that SNPs don't catch. Indels are instance. The differences in dna like gene deletions are totally missed by the SNP chips used today. We need whole genomic scans, not just for disease causation but to finally show the real cause for race differences in humans.

Maju said...

"Maju believes in some sort of Paleolithic continuation with those primitives having Y chromosome haplogroup R1b and I, and mtDNA haplogroups U and H. It is part of the idea that Europeans were really White until interlopers came from the east with their dark skins and disgusting habits to blacken the European race. Those darkies were haplogroups J, E and G".

I don't think, much less "believe", that there's a clear pigmentation correlation in this matter. I have at times speculated with blondisms being at least in some important fraction from East Europe and maybe even the Balcans (and therefore spread maybe with Indoeuropeans and Neolithic colonists). It is in any case anecdotal, as I suspect that most pigmentation traits are old enough to have been around well before Neolithic (and of course Metal Ages' IE expansion, whose genetic impact was surely very limited except in Northern Europe).

I think I can detect some diffuse anthropometric pattern that has a West-East cline and therefore is probably correlated with these migrations or lack of them, but it is not really related to pigmentation (unless it's a pheomelanin) but to cranial angulosity, with Western and particularly SW types having greater frequencies of "angular" (rectangular instead of oval) facial structure and specially in the forehead area. But it's a clinal trait, like everything.

But I agree that Y-DNA lineages J (specially J2), E (specially E1b1b1a2) and G (specially G2a), as well as T and surely some I, spread with Neolithic and post-Neolithic flows, these also including some R1a1, notably R1a1a7. Not necessarily from West Asia, some probably only began expansion from the Balcans and surrounding areas.

Whatever the case, I don't have all things totally clear. I understand that it is important to keep an open mind and I have changed my understanding or relativized my convictions on many of these matters. However I can just take sides for what seems at least somewhat likely (preferably 'most likely') and not what is most unlikely, specially when considered from different angles. I also give much more importance to coherence with archaeological knowledge where available and to SNP-based phylogeny than to STR-based molecular clock speculations, which taken alone have absolutely no value whatsoever.

But let's them come. I agree that testing of ancient DNA may be helpful but so far it's providing contradictory evidence such as apparent population replacement in Neolithic Central Europe and continuity instead in Iberia and North Africa. However the samples we have are anything but optimal with marginal regions (Baltic area) and cultural groups oversampled and central ones (Rhine, France, etc.) almost without any relevant data.

...

Maju said...

...

"Indels are instance. The differences in dna like gene deletions are totally missed by the SNP chips used today".

I am aware but I'm not persuaded that indels will give us a more clear picture.

"We need whole genomic scans, not just for disease causation but to finally show the real cause for race differences in humans".

I don't think we will have the power to do such whole genome analysis across wide samples while also providing understandable results (i.e. more than mere statistical simplifications) in many decades.

It's the "promise" of genetics: to unravel the causes of everything we are by looking at the "manual" (DNA) but (1) the genome is so huge and complex that analyzing it is most challenging (an information management problem), (2) genes are not isolated but interact with each other (multiplying the complexity once and again to almost infinite magnitudes) and (3) not all is in the genes: a good deal of what we are is epigenetic, environmental modifications of the code.

Metaphorically a gene can be like a formula where X=2Y³ or whatever. Y being the environmental variable that changes the result of the equation dramatically. And this is a highly simplified example... go figure what is actually going on in our cells, specially in the early stages of development, when we are most plastic!