New blogs

Leherensuge was replaced in October 2010 by two new blogs: For what they were... we are and For what we are... they will be. Check them out.

Saturday, June 6, 2009

The Role of Geography in Human Adaptation


This is the title of an interesting new paper published at PLoS Genetics (open access):

G. Coop et al.,The Role of Geography in Human Adaptation. PLoS Genetics, 2009.

It is an extensive paper dwelling on the differences and smilitudes between human populations worldwide.

A major shortcoming is that only Pakistan has been sampled in South Asia and that the NE Africa, another important region in human population history has only one sample. Even if you don't want or can set a joint venture lab in Mumbai to go around the Indian laws on genetic material, you can always sample strategical places like Bangla Desh and Sri Lanka. And undersampling of the Nile valley and the Horn has no excuse whatsoever other than poor planning or intellectual laziness.

But these defects are way to common in global studies like this one so guess it's ok. It is still a worthy study anyhow.

Another possible shortcoming is that they use FST distances to estimate divergence when actually this measure can be very misleading, as far as I know.

One interesting finding is that genic alleles (i.e. those that do express in phenotype) are much more likely than non-genic ones to have extreme differences between populations at global level. This implies that these genes have been under effective selection. (See fig. 1)

Another one is that the populations studied can easily be classified in three groups: Africans (south of the Sahara), West Eurasians (including South Asians) and East Asians. These three groups show clear genetic differences (within human variability). Also East and West Eurasians share some genetics in contrast with Africans. (See fig. 3).

In this sense, figure 4 suggests different pigmentation adaptations (and their corresponding founder effects) for West and East Eurasians. This largely replicates the general structural findings even for some specific genes: Eurasians overall share the derived allele of hair-color related gene KITLG, but the derived allele of SLC24A5 (skin-color related) is almost exclusive of West Eurasians, while the derived allele of MC1R (related also to skin and hair color) is only dominant among East Asians and Native Americans (but not Melanesians). This seems to ratify the already known fact that the pigmentation evolution followed two somewhat different paths in West and East Asia, what in turn ratifies the coastal model of Eurasian expansion and rather rejects the alternative hypothesis of a Central Asian/Siberian corridor early on.

Finally, in the discussion section, they argue that while there is some clear evidence for positive selection of genic traits, this is comparatively weak and that random events such as founder effects and drift in early population history have been much more decisive in determining the differences between populations. That local parallel mutations, such as the ones found in pigmentation strategies, are clearly much more important than transfer of genes between populations.

The article also suggests that there may have been some gene flow between Africa and West Eurasia of which East Asia has been instead excluded. This, they argue, could explain the greater FST distance of East Asians, as they know of no other possible processes behind this phenomenon.

57 comments:

Manjunat said...

Eevn if you don't want or can set a joint venture lab in Mumbai to go around the Indian laws on genetic material, you can always sample strategical places like Bangla Desh and Sri Lanka. And undersampling of the Nile valley and the Horn has no excuse whatsoever other than poor planning or intellectual laziness.

Well said, Maju. There are too many dumb studies that consider only Pakistanis for whole South Asia. I would even venture to say, this is intellectual dishonesty. Many of these studies just want to put a line between East Eurasia and West Eurasia. It's obvious with India sampled they can't achieve this.

Manjunat said...

This seems to ratify the already known fact that the pigmentation evolution followed two somewhat different paths in West and East Asia, what in turn ratifies the coastal model of Eurasian expansion and rather rejects the alternative hypothesis of a Central Asian/Siberian corridor early on.

No, since these mutations arose after the complete colonization of the old world.

Maju said...

I understand that not sampling India is generally caused by legal issues. Guess that setting a joint venture lab in India goes beyond their budgets but sampling only Pakistan in any case is going to bias the results anyhow because Pakistanis are known to be genetically closer to West Eurasians than most other South Asians.

No, since these mutations arose after the complete colonization of the old world.

How do you know?

If you look carefully at fig. 4, you will see that all mutations are pre-existent in the African samples (unless they have back-migrated) but in small ammounts only. It was therefore founder effect plus weak selection what caused them to become dominant either in Eurasia as a whole or in specific regions.

This means, I understand, that the alelles were pre-existent and that they were selected (both by random events and weak selection) after the OoA and, in the regional cases, at the very spread within Asia, probably still in the tropical corridor.

terryt said...

Thanks Manjunat. I was wondering myself how the first comment, 'This seems to ratify the already known fact that the pigmentation evolution followed two somewhat different paths in West and East Asia' necessarily led automatically to the comment, 'what in turn ratifies the coastal model of Eurasian expansion and rather rejects the alternative hypothesis of a Central Asian/Siberian corridor early on'. The first statement does not prove the other.

"they argue that while there is some clear evidence for positive selection of genic traits, this is comparatively weak and that random events such as founder effects and drift in early population history have been much more decisive in determining the differences between populations".

But positive selection operates most effectively, and rapidly, on small, isolated populations. It's only once a particular character becomes fixed in just such a small population that it is able to then spread through the wider population by the formation of hybrids, often helped through 'founder effects and drift'. Selection in human populations is most likely to have occurred in populations at the geographic margins, or otherwise isolated, from the bulk of the human population.

Maju said...

The first statement does not prove the other.

In my opinion it does or at least adds to the evidence in that same direction, strongly suggesting that West and East Eurasians have not exchanged genes intensely as would be the case if the Central Asian corridor would have been very active and even and ancestral homeland.

But positive selection operates most effectively, and rapidly, on small, isolated populations.

Does it? Drift does work that way and it could be enhanced by weak positive selection but could also totally neutralize the effect of selection as well, by randomly supressing the necessary alleles from the limited genentic pool. That a gene gives some fitness is not in itself a guarantee of survival for the individuals carrying it, where drift is intense it can perfectly dominate over fitness-driven selection, specially if the selective pressure is weak.

In practical terms, it's possible that some alleles carried by the Igbo could be fit in the conditions of Yakutia but if these alleles never reach Yakutia to begin with, there's no way selecton could adopt them. Evolution can only work with what it already has.

It's only once a particular character becomes fixed in just such a small population that it is able to then spread through the wider population by the formation of hybrids, often helped through 'founder effects and drift'.

No. This is essentialy wrong. Genes can perfectly flow regardless of their numerical importance in any given population. A gene may be neutral or even weakly unfit in some context but, after it finds its way into a different population or after migration, it may well happen to be adaptative in this new context. It would be then and only then when (ignoring the chaotic effects of drift) it can become dominant via selection.

Mutations do not happen upon request, they just happen by accident. What can happen under selective pressure is that pre-existent mutations can become dominant because they give some additional fitness to the phenotype. But the mutation must be available in the genetic pool to be selected for - otherwise is impossible (like in the example of the four arms).

terryt said...

"if the Central Asian corridor would have been very active".

I doubt if it's been 'very active' for other than very short periods, such as when the Mongols moved west or the Tocharians moved east, or when anybody else managed to negotiate it.

"That a gene gives some fitness is not in itself a guarantee of survival for the individuals carrying it".

Selection can only act on a gene that gives some fitness so if it's canceled by drift we have no way of ever knowing of it's existence. However a gene that provides some level of fitness is far more likely to become fixed in a population of limited size, especially if it's a recessive gene. In a larger popluation it will have far less chance of expression.

"Genes can perfectly flow regardless of their numerical importance in any given population".

And do. But as I've just pointed out they are less likely to become fixed in a larger population. It's only once a group of people are able to pass an advantageous gene on that it can spread.

"Mutations do not happen upon request, they just happen by accident".

Quite. But they happen fairly much continuously. Most are disadvantageous of course. Not all genetic variations in the present human population had occurred before any out of Africa movement.

Manjunat said...

unless they have back-migrated

That is the only plausible explanation. If you read the paper from where these samples have been taken (SLC24A5), there the authors talk about convergent evolution of skin colour among Europeans and East Asians. If it already existed in Africa then we need to prove both East Asian and European mutations were present. We also need to explain why only European specific mutations are observed among West Asians and South Asians. That figure says nothing. At the most you can argue African light skin mutations became extant in Europe, West Asia, North Africa. If you want.

Maju said...

@Manju: they have backmigrated to the Khoisan? I'm sceptic to say the least.

@Terry:

1. I mean in the Paleolithic. Tocharians and Mongols are too recent for our discussion. There is very little or rather no evidence of cultural or genetic transfer through the steppary corridor. At most we have ancient Altaians connected with West Eurasia (but not yet with the East) and some evidence of moderate gene flow in both directions at rather advanced stages of Eurasian colonization (things like Y-DNA Q or mtDNA G, CZ, D and X).

2. I make no sense of your recessive gene claim. Genetics are not for the most part purely mendelian but anyhow a recessive allele will have exactly the same chances of expression in a small or large population, it all depends on the proportion of the allele within the population.

3. There is indeed less chance of fixation in larger populations, at least if only random drift is acting. But genes can flow regardless of being fixed or not. There is absolutely no need for a gene to fixed before it inrogresses in other populations. Of course, the more common it is in general (not in any particular single population), the higher chances it has of propagation but that's just obvious.

3. Surely some mutations happened after the OoA but many probably are pre-existent to that epysode, just that were not emphasized by random events or had no fitness effectivity whatsoever in Africa. If we consider the history of Humankind the Eurasian epysode is only about 1/3 of all human history and is also possible that many alleles were drawn from pre-sapiens stages, so there is no particular reason to speculate that these mutations only happened once in Asia or Europe. Though it does seem that they were favored here by a combination of random events and weak selective pressures.

terryt said...

"There is very little or rather no evidence of cultural or genetic transfer through the steppary corridor".

There is actually, but you prefer to adopt other explanations for the evidence. So we'd best leave that for now.

"I make no sense of your recessive gene claim".

Anyone involved in animal, bird or plant breeding knows that in order to fix a characteristic, or particular mutation, in the herd or flock some level of inbreeding is required. If it's done systematically it's usually called line breeding, but it's effectively the same thing.

"a recessive allele will have exactly the same chances of expression in a small or large population, it all depends on the proportion of the allele within the population".

No it won't. A recessive allele is much more likely to become a high proportion of a small population than it is in a large one. Drift alone could increase it in a small population, and give rise to the formation of double recessives, but drift would be unlikely to increase it's proportion in a large population. Of course selection can only operate once double recessives begin to form. To achieve this, individuals with some common ancestry have to form offspring. Again, much more likley in a small population.

"There is absolutely no need for a gene to fixed before it inrogresses in other populations".

It's unlikely to spread in any quantity, or systematically, to any new population unless the gene has already been expressed in a population of double recessives. Otherwise the gene would just randomly introgress and never become fixed.

"Though it does seem that they were favored here by a combination of random events and weak selective pressures".

The fact that, on the large scale, we can identify almost instantly whereabouts in the world a particular individual's ancestors come from indicates that something more than random events and drift are involved. Weak selection pressures are also unable to account for the phenomenon adequately. If an original expanding population had possessed all the gentic variation we see in humans today we would not expect to see the well-defined regional variation we see today. The same gene would emerge randomly through the world.

Maju said...

"There is very little or rather no evidence of cultural or genetic transfer through the steppary corridor".

There is actually...


No, there is not. Not for Paleolithic certainly. And much less Middle Paleolithic, when the human expansion in most of Asia actually took place.

I refuse to discuss anything else until this issue is clear to all.

terryt said...

For a start you have failed utterly to adequately explain how mtDNA A has finished up basically confined to Central Eurasia east of the Altai while its relatives W and IX are found in Central Eurasia west of the Altai and its relatives Y and S are found along the East Asian coastal region down into Australia. A movement starting from India and moving round the Himalayas in a big circle doesn't hold up to any sort of scrutiny.

Maju said...

MtDNA A? It coalesced along East Asia in a S>N migration very apparent in other East Asian haplogroups like M8, M9, G, D or N9. It was then nevertheless a small "private" N* lineage and only expanded once in NE Asia. We have discussed that before.

As for the Western lineages they are part of a differen migration process also very apparent in N1, R0, R2'JT, U and W itself: from South to West Asia.

I have gone through all that before and I think it's fairly clear. That you seem to think that remote shared ancestry means that somehow two "private" lineages must remain attached to each other generation after generation for many many thousand years is something that I just cannot understand nor share.

terryt said...

"It coalesced along East Asia in a S>N migration very apparent in other East Asian haplogroups like M8, M9, G, D or N9 ... As for the Western lineages they are part of a differen migration process also very apparent in N1, R0, R2'JT, U and W itself: from South to West Asia".

You'll have to do better than that. A most unlikely scenario. If that had happened we would expect to see a basic divide in the N line between eastern (N9, Y, S and A) and western branches (X, W, and N1 or I, whatever you like to call that last group), and you were very critical of a study that suggested something like that. Instead, what we find is a secondary expansion involving R-derived haplogroups.

"you seem to think that remote shared ancestry means that somehow two 'private' lineages must remain attached to each other generation after generation for many many thousand years".

You seem to think that ALL haplogroups remained attached to each other generation after generation before they all expanded together from India, in two different directions. That belief is simply the product of two slight alterations of the biblical 'Garden of Eden' scenario. Firstly, your Garden of Eden is in India. And secondly, you realise that the human species must originally have had more than just four variations of many of its genes, therefore you need more than just two individuals.

Your comment, 'I refuse to discuss anything else until this issue is clear to all' contains more than a just a hint of 'I'm taking my bat and ball so you can't play any more' about it. I suggested we leave the subject of Central Asian migration for a while so we could scrutinise your suspect ideas about recessive genes and inbreeding, but obviously you wish to avoid such scrutiny.

Maju said...

I don't see absolutely any need to split N or whatever other REAL haplogroup in order to fit with your preconceptions. We have that data and all we can do is to explain how most likely ended up like that. A southern Asian flow in both directions, specially in the early stages of colonization, seems the most natural explanation. I don't need any other.

You seem to think that ALL haplogroups remained attached to each other generation after generation before they all expanded together from India, in two different directions.

Not necessarily India/South Asia, not necesarily all and not necesarily attached.

What I say is that if a lineage (not yet a haplogroup at the stem stage) shows a long stem, this means that it took some time before expanding (before really becoming a haplogroup), we cannot easily say anything about it before its expansion: it was a private lineage carried by very few people, whose exact movements across the globe are impossible to track directly.

We can nevertheless make some inferences based in the shortest and logical possible paths between place of origin (ancestral node) and place of destination (this haplogroup's node) and, in many cases, also on the patterns show by other lineages (as it's likely and reasonable that different lineages followed the same "highways" more or less).

Hence, as we know nothing between the putative node of N in SE Asia and the more clear node of A in NE Asia, but as we know that other lineages (listed before) show a trend SE>NE Asia along the coast, we can logically infer that pre-A was a private lineage somewhat involved in this general pattern of S>N migration through eastern Asia.

Same for the "western" N lineages: we can trace via South Asia at least R, R2'JT, N2(inc. W) and U, so it's only logical that other lineages like N1, X or R0 did the same, just that their traces have vanished (or maybe even just lay there still undetected).

In general, I don't need to do better: you do. You need to really understand the whole pattern of mtDNA and stop demanding that reality fits your way of thinking and instead try to understand reality as well as possible and to look not for your favorite preconception in it but for which is actually the most reasonable explanation possible considering what we actually know.

Stop calling for long stemmed haplogroups as excuse to argue unlikely hypothesis and focus on the more clear ones: the ones that expanded earlier and are hence more easy to track. The rest just went anonymously with them with all likelihood.

ren said...

Maju wrote,
"In general, I don't need to do better: you do. You need to really understand the whole pattern of mtDNA and stop demanding that reality fits your way of thinking and instead try to understand reality as well as possible and to look not for your favorite preconception in it but for which is actually the most reasonable explanation possible considering what we actually know."

Basically the same thing I said to you when you were vehemently (with insults) arguing for second waves of Ns and Rs coming from the Levant when I said that the Eurasian Eden was India.

terryt said...

"I don't see absolutely any need to split N or whatever other REAL haplogroup in order to fit with your preconceptions".

The situation is that we would need to split it to fit with YOUR preconceptions. I had no preconception until after I'd examined the evidence.

"it was a private lineage carried by very few people, whose exact movements across the globe are impossible to track directly".

But any mutation must first arise in a single individual. In the Paleolithic (a time when we both agree populations were small) the mutation could quite easily come to dominate a single valley, for example. In most cases the 'haplogroup' would probably move very little until after it had come to dominate a particular region, although we have no way of being sure of that.

"we can logically infer that pre-A was a private lineage somewhat involved in this general pattern of S>N migration through eastern Asia"

There's no need to 'infer' any such thing.

"we can trace via South Asia at least R, R2'JT, N2(inc. W) and U, so it's only logical that other lineages like N1, X or R0 did the same".

Again. It's in no way 'only logical'. R's descendant could quite easily have expanded long after other N haplogroups had. In other words N could quite easily be one of 'the ones that expanded earlier'.

Interestingly in this article:

http://www.nature.com/ejhg/journal/v16/n6/full/5201998a.html

The authors write:

"Haplogroups A, C, D*, D5, D5a, G, M7c, M8a, M9, N*, and Z are identified as NDH, with much higher frequencies in north than in south significantly. Haplogroups, M*, B*, B4, B4a, B4b1, B5*, B5a, F*, F1a, F1b, F1c, F2a, M7*, M7a, M7b*, M7b1, M7b2, R*, R9a, R9b, and R9c are identified as SDH, with much higher frequencies in south than in north. Most of the major haplogroups derived from M lineage are NDH except for M7, whereas most of the major haplogroups derived from N lineage are SDH except for N9 and A".

But note that the only N haplogroups identified as southern are actually those derived from R. Now you've got to admit that that's interesting.

Maju said...

But any mutation must first arise in a single individual.

Sure, so what?

In the Paleolithic (a time when we both agree populations were small) the mutation could quite easily come to dominate a single valley, for example.

Or not. The latest studies suggest that actually this putative effect on Y-DNA domination is mostly but not totally unreal. You tend to overemphasize that in order to make the hyper-simplification of one tribe=one haplogroup look real - but it's not (unless it is a very isolated case).

Drift swapping to fixation takes time, and exchange with other groups (something that always happens) weakens drift by expanding very notably the effective population size. There must have been thousands or maybe millions of private lineages jumping around weirdly, along the better estabilished ones. Some of them have survived for long enough to expand at later moments, while others have survived even if they never really experienced any expansion. Of course many many others, probably the vast majority, have vanished - but not necesarily in just few generations.

"we can logically infer that pre-A was a private lineage somewhat involved in this general pattern of S>N migration through eastern Asia"

There's no need to 'infer' any such thing
.

There is because you are all the time asking and crying about where pre-A was before expanding as A.

"we can trace via South Asia at least R, R2'JT, N2(inc. W) and U, so it's only logical that other lineages like N1, X or R0 did the same".

Again. It's in no way 'only logical'. R's descendant could quite easily have expanded long after other N haplogroups had. In other words N could quite easily be one of 'the ones that expanded earlier'
.

Erm... R is N, as much as W or A or whatever.

But note that the only N haplogroups identified as southern are actually those derived from R. Now you've got to admit that that's interesting.

Actually 3 out of 4 N(xR) known Eastern Asian lineages are present in SE Asia: N9 (N9a mostly), N21 and N22. You were actually the one who called my attention on these last two. This makes SE Asia having 3 N(xR) lineages, only surpassed by Australia in such number (3+1) and comparable to West Eurasia (2+1). As N2 (N2a+W) is shared between Australia and South/West Asia, I guess we can agree that it necesarily had to go through SE Asia as well in its still untracked migrations. All these led me to concede that SE Asia was probably the ancestral homeland of N expansion.

Maju said...

Basically the same thing I said to you when you were vehemently (with insults) arguing for second waves of Ns and Rs coming from the Levant when I said that the Eurasian Eden was India.

Now explain that to Terry, pls.

terryt said...

"The latest studies suggest that actually this putative effect on Y-DNA domination is mostly but not totally unreal".

That study deals with a series of populations that have been interacting for thousands of years, so it's hardly surprising genes have become thoroughly mixed. What about the situation where the group concerned is the first into a particular region?

"exchange with other groups (something that always happens) weakens drift by expanding very notably the effective population size".

But drift, bottlenecks and founder effect are likely to be very pronounced in populations at an expanding margin of a population, where opportunity for exchange with other groups is minimal. We can see that drift, bottlenecks and founder effect have been very active even in the case of the relatively recent Polynesian migration into the Pacific.

"A southern Asian flow in both directions, specially in the early stages of colonization".

Surely in the early stages of colonisation flow would only be in one direction: into the previously uninhabited region?

"This makes SE Asia having 3 N(xR) lineages, only surpassed by Australia in such number (3+1) and comparable to West Eurasia (2+1)".

And India has how many N(xR) lineages?

Maju said...

That study deals with a series of populations that have been interacting for thousands of years

More or less what seems to me that the peoples of the Indo-Pacific arch were at the first stages of Eurasian expansion.

But drift, bottlenecks and founder effect are likely to be very pronounced in populations at an expanding margin of a population...

Only founder effect.

An expanding population has low drift. A bottleneck by definition implies severe contraction of the population like the death of 90% or more. I really do not believe in bottlenecks as such, at least not as anything common, though founder effect looks similar in its effects.

Surely in the early stages of colonisation flow would only be in one direction: into the previously uninhabited region? .

In early stage #1 yes but in early stage #2 not necearily.

Even if the pressure is dominant in one of the vases, as you may know, the movement of molecules is stochastic and they flow in both directions. Only the dominant trend is unidirectional, until pressure varies and assuming a single connection pipe (i.e. in the simplest possible case, but humans and geography are not simple anyhow).

And India has how many N(xR) lineages? .

Almost none, AFAIK. W.

terryt said...

"More or less what seems to me that the peoples of the Indo-Pacific arch were at the first stages of Eurasian expansion".

By convention there was nobody in the Indo-Pacific arch at the first stages of Eurasian expansion. Eurasian expansion presumably started when just two mtDNA haplogroups and no more than four (possibly one) Y-haps emerged from Africa. Now, that's some bottleneck, or founder effect, perhaps drift. As the population containing those few haplogroups expanded further they almost certainly began to diverge. It's most unlikely that the whole population stuck totally together as they spread out. Small groups would have moved ahead of the bulk and come to dominate regions they passed through.

"An expanding population has low drift. A bottleneck by definition implies severe contraction of the population like the death of 90% or more. I really do not believe in bottlenecks as such, at least not as anything common, though founder effect looks similar in its effects".

You might find this link relevant:

http://anthropology.net/2009/06/09/on-becoming-modern-science-june-5th-edition/

Quote:

"The point to make is that innovative technologies and behaviours could have been invented, forgotten and discontinued many times over during the course of the entire Palaeolithic, but because populations were sparse, largely isolated from one another, constantly on the move, and prone to dying out, passing down innovations – and even language – through many generations simply wouldn’t have been possible".

And passing down haplogroups.

"Almost none, AFAIK. W".

And W most likely came in from the Iranian Plateau long after India had been already colonised. Doesn't look promising as evidence for an original movement of mtDNA N through India.

Maju said...

You are counfusing two things:

1. The OoA epysode, shrouded on mystery, is not the same as the Eurasian expansion: the demographic boom once in Eurasia, so clear in haplogroups structure.

2. We don't know how many lineages started the OoA migration, we know that at least those but were probably many more. Just that the rest never really made it to the bom phase: only a handful did.

passing down innovations – and even language – through many generations simply wouldn’t have been possible...

That's plainly put an exaggeration that contradicts real data: obviously UP peoples innovated and transmitted these innovations to the new geenrations (and language too, of course).

What that study really means is that the largest the communicative networks, the fastest the innovation process. What is surely true in principle.

Doesn't look promising as evidence for an original movement of mtDNA N through India.

Doesn't really matter because W, N1 and X are minor in West Eurasia too. West Eurasia is dominated by one single N subclade: R. And that clade is clearly original from India.

The rest probably migrated with R, so to say: were part of the flow that brought R (also an N clade) to the West. But if you prefer to make them go between Australia and Europe via Manchuria and the Takla Maklan... your choice. Makes absolutely no sense to me and you also hve zero evidence to support it anyhow.

terryt said...

"The OoA epysode, shrouded on mystery, is not the same as the Eurasian expansion".

Neither of them is a 'single' expansion. Population expansions are complicated affairs, and although it is useful to simplify things you like to simplify things far more than is at all reasonable.

"Just that the rest never really made it to the bom phase: only a handful did".

OK. So what do you think the effect of bottleneck, drift or founder effect would have been on the haplogroups that left Africa?

"And that clade is clearly original from India".

The western R-derived haplogroups clearly came through India, but in no way can you claim they definitely 'originated' in India. And just 'because W, N1 and X are minor in West Eurasia' doesn't mean at all that they came from India, let alone accompanied by several R-derived haplogroups in a single migration. All haplogroups must have moved from Africa to India by some route. Surely it is most reasonable to suppose they dropped off some haplogroups along the way.

Maju said...

Neither of them is a 'single' expansion.

IMO, the Eurasian expansion was a single fast episode: the huge starlike structure of M, and to a lesser extent of N as well, is witness to it.

OoA instead is not a expansion, unless you consider it within the African starlike expansion of L3.

OK. So what do you think the effect of bottleneck, drift or founder effect would have been on the haplogroups that left Africa?.

Founder effect and drift seem to have reduced the migrating L3 to M and N. Probably there was something more than just that at the beginning. The reduction of many L3 (and maybe other haplogroups) to just two lineages is already a significative effect.

Right now I am intrigued because I have found that one of the places of the world where L6 is more common if not the one where it's most concentrated is not in Africa but it is Yemen (14%). This may have to do or not with the OoA but I think worth considering as a possibility. In that case, it might have been what the OoA lineages drifted into in Southern Arabia before backmigrations diluted it to the just 14% of today. Of course it may be the product of a more recent migration too.

The western R-derived haplogroups clearly came through India, but in no way can you claim they definitely 'originated' in India.

Out of 13 haplogroups directly derived from R, 8 are present in South Asia and 5 are exclusive from that region. IMNSHO R is clearly original of South Asia.

And just 'because W, N1 and X are minor in West Eurasia' doesn't mean at all that they came from India, let alone accompanied by several R-derived haplogroups in a single migration.

I don't think it was a single migration but a more or less continuous flow: a process of migration that lasted surely many milennia, favored by whichever conditions (of which I think most important were certain saturation of the eastern route and maybe some newly acquired competitive advantage over Neanderthals as well).

All haplogroups must have moved from Africa to India by some route.

By "all haplogroups" you mean just M and N right? Or more properly pre-M and pre-N (would be L3* at that stage in fact). M- and N-derived haplogroups coalesced already in Asia, there's no question about this last.

terryt said...

"This may have to do or not with the OoA but I think worth considering as a possibility".

Most unlikely to be connected at all. L6's arrival in Southern Arabia is much more likely to have been a separate event. The fact that M and N are both L3 tends to suggest that just members of L3 emerged to give rise to all mtDNA haplogroups beyond Arabia. There are no L6-derived haplogroups beyond Arabia.

"By 'all haplogroups' you mean just M and N right?"

I should have specified 'all haplogroups beyond Arabia'.

"Out of 13 haplogroups directly derived from R, 8 are present in South Asia and 5 are exclusive from that region".

It does not at all necessarily follow that it 'is clearly original of South Asia'. It is quite possible its expansion there is a secondary expansion.

"M- and N-derived haplogroups coalesced already in Asia, there's no question about this last".

At least we agree on that. But did they both coalesce within just the same small region?

Maju said...

This is a good example on how you want things coming in neatly packed boxes:

Most unlikely to be connected at all. L6's arrival in Southern Arabia is much more likely to have been a separate event. The fact that M and N are both L3 tends to suggest that just members of L3 emerged to give rise to all mtDNA haplogroups beyond Arabia. There are no L6-derived haplogroups beyond Arabia.

But now imagine that the emigrant group of the OoA had both L3 and L6 (and maybe other lineages, who knows?) and that while this group was "distilled" into M and N while arriving to South Asia, a parallel process of drift in the areas behind may have instead "distilled" L6. It's all random.

Of course, it could be a different event but the important thing is that you can come to understand how the same population can "evolve" into two groups with very different lineages by just isolation and numbers low enough to allow drift to happen.

"By 'all haplogroups' you mean just M and N right?"

I should have specified 'all haplogroups beyond Arabia'
.

That makes thing even more confuse. You still seem to be speaking of downstream haplogroups, which did never arrive from Africa but coalesced already in Asia.

I think you just want to be tricky and/or sound mysterious here.

It is quite possible its expansion there is a secondary expansion.

There's no evidence of such thing: a starlike structure indicates a rapid expansion from a single localized core, what you claim would need another type of structure with a clear distinct subbranch excluding the Eastern lineages.

You are therefore wrong in this and you are just being stubbornly stuck in your preconceptions again.

But did they both coalesce within just the same small region?.

That we can't know. Actually I'd say they coalesced on the march. We may be able to pinpoint with some security the area of expansion of a haplogroup but where it was in the stem phase is impossible to tell. In the case of these two major Eurasian macro-lineages, we can more or less guess that they coalesced between the Nile and the Ganges (and maybe as far as the Mekong in the case of N).

terryt said...

"but the important thing is that you can come to understand how the same population can 'evolve' into two groups with very different lineages by just isolation and numbers low enough to allow drift to happen".

That's an interesting comment, coming from you. After all it is I who have always maintained that haplogroups M and N evolved 'into two groups with very different lineages by just isolation and numbers low enough to allow drift to happen' whereas you have always maintained the M and N haplogroups remained associated together long after they'd emerged from Africa. By the way, the 14% of L6 in Yemen is just as likely to be the product of women carried from East Africa much more recently than any Great Southern Migration.

"You still seem to be speaking of downstream haplogroups, which did never arrive from Africa but coalesced already in Asia".

True, they coalesced in Asia. But to me it looks as though haplogroups M and N must have separated pretty much as soon as they had emerged from Africa. After all they are just two of the seven or so branches of a rapidly expanding clade, L3. Most of this expansion took place within Africa. (Refer your earlier comment, 'OoA instead is not a expansion, unless you consider it within the African starlike expansion of L3'. It presumably was connected in some way to that expansion).

"a starlike structure indicates a rapid expansion from a single localized core".

Presumably M has a starlike structure because it was the first modern mtDNA haplogroup to enter Northwest India, probably into the savannah region. Consequently it could expand rapidly. N's expansion was more subdued, for some reason or other. I'll leave you to consider possible reasons for that.

Maju said...

I think that M and N are the product of drift at the early stages of the Eurasian adventure - but most likely together. They do not represent distinct populations IMO.

The 12% L6 of Yemen (I said 14% before but was an error) is much higher than the percentage of L6 found in any African population anywhere. It might be of course a more recent founder effect of some sort but it doesn't look like it can be explained by mere normal recent import of East African women, like in the slave trade. The study where I got that from (Kivisild 2004) sampled the Horn and Mozambique and, while all the other Yemeni L(xM,N) could be traced easily to East Africa, L6 is extremely rare in it (and elsewhere for what I know).

If you actually know of a possible source for this lineage anywhere in Africa, please tell me. So far I know of 14 Yemenis and two Ethiopians (one Amhara and one Gurage).

True, they coalesced in Asia. But to me it looks as though haplogroups M and N must have separated pretty much as soon as they had emerged from Africa. After all they are just two of the seven or so branches of a rapidly expanding clade, L3. Most of this expansion took place within Africa.

But I don't see how this could possibly tell that M and N followed different routes. We find them as L3 in East Africa and as M and N in the South/SE Asian node, with N and M subclades expanding again together in many (most) cases. For me it rather argues for a joint journey together rather than for any separate migration.

It would be different if we found them "exploding" at different spots (like SE Asia and Europe, for example) or very different time frames but what we see is rather the opposite. If they did not migrate together, they did almost that.

terryt said...

"If you actually know of a possible source for this lineage anywhere in Africa, please tell me".

I don't, sorry. The only reference I could find merely implied (note) that the Yemen L6 wasn't closely related to Ethiopian L6 and that it had come from somewhere else. The nearest I could gather was further south, presumably Mozambique, although that is the region you mention that Kivisild et al sampled. It's possible the L6 in any 'original' homeland has died out as a result of later migration into that region.

"with N and M subclades expanding again together in many (most) cases".

Not really. The pattern of the two haplogroups doesn't really overlap at all except in the case of N-derived R and some M-derived haplogroups. But the R expansion seems to be largely independent of any original N expansion. And L3 is by no means just confined to just East Africa. Each main branch of the haplogroup is reasonably well geographically defined so it is reasonable to suppose the two lines of the haplogroup found outside Africa would also be so defined.

Maju said...

Kivisld seems to place himself as the discoverer of L6 with that paper. Seems that talking of L6 as "African" is only out of ignorance on such light. On the other hand he mentions that 12 of the 14 yemenis sampled had the same subhaplogroup what means they are product of some founder effect. But still the diversity found in Yemen for L6 is greater than the one found in Ethiopia (3 vs 2 subclades) and is not clear from the phylogeny that either region is closer to the root. Also L6 was only found among Semitic speakers in Africa, what is very suggestive of a migration from Arabia in ancient but post-Neolithic times.

Btw, the correct link for Kivisild 07 is this one. Found zat Anthroforum.

Searching for further L6 info (in vain), I found this recent paper on Indian mtDNA N, which I am sure you'll find interesting. A notable finding is that B and F do exist in India (even in the west), as well as R0 and even HV. Special attention is given to haplogroup U.

Not really. The pattern of the two haplogroups doesn't really overlap at all except in the case of N-derived R and some M-derived haplogroups. But the R expansion seems to be largely independent of any original N expansion.

I think you're missing the overall process, the proverbial trees don't let you see the forest.

The process is something that happened in South AND SE Asia primarily and that kept both areas (and their periphery) exchanging genes for a long while. And the second part of the process is the exhaustion of this route and the opening of the western route. R is most important in the Western phase but it did not move alone (at least in the long run): it was accompanied by several N(xR) and M subclades.

Maju said...

And L3 is by no means just confined to just East Africa. Each main branch of the haplogroup is reasonably well geographically defined so it is reasonable to suppose the two lines of the haplogroup found outside Africa would also be so defined.

Following Salas'02 this is indeed the case: L3b'd is most concentrated in West Africa, while L3e is most concentrated in the Bantu expansion area. But most of the L3* (many different lineages) is in East (NE) Africa, making up a 25% of the gene pool there. If I understand correctly, this means that, out of 5 L3(xM,N) sublineages, all them are well represented in the NE, maybe with the exception of "Western
" L3b'c'd'j (just maybe because we don't know the exact composition of L3*).

L3e, most common in the Bantu area, is not a direct L3 sublineage but part of L3e'i'k'x. This seems to suggest that, even if Niger-Kongo by language, the Bantus are genetically closer to East/NE Africans in fact, at least matrilineally (I read something about that somewhere in the past - can't remember where).

Whatever the case, there is a process, a flow, in Africa too: a process involving primarily East and West Africa and, secondarily, Middle Africa too. All lineage flows are part of that process. L3 probably flowed from East to West, while L2 surely did in the opposite direction. And L1 did it in a manner that is difficult to read now, as it is quite widespread, probably implying many different migrations in all directions, since older times.

It's just neighbouring peoples among whom genes flow, we imagine that via migrations, with relative ease. But not as easily as to erase all trace of structure.

We are interested in the structure because it gives us important clues on how many people (proportionally) and when (roughly) moved from here to there and vice versa. But the structure was always fluid and imprecise. Trying to go extreme in disentagling the structural components as you do, imagining that every lineage represents a single and unique clan or tribe (or even individual maybe) making a prehistorical event is going too far.

terryt said...

Thanks for the link to Indian mtDNA. Quotes from the article:

"In India most of the sub-haplogroups of N are the region (Indian) specific variants of haplogroup R".

No, or very little, N(xR) (and what there is looks as though it is almost certainly a later immigrant) and the Indian-specific nature of the haplogroups in no way argues definitively for an Indian origin.

"A notable finding is that B and F do exist in India (even in the west)"

I see them mentioned in the maps but the authors say that Cordaux et al (2003) found "F being resricted only to the northeast Indian tribes". Looks suspiciously as though it has come in from the east, probably with eastern Y-haps.

Regarding your second post:

"Whatever the case, there is a process, a flow, in Africa too ... implying many different migrations in all directions, since older times".

Yet you seem to be arguing against such a back and forth process operating outside Africa in older times, excepting the R expansion. Surely we should expect just such a flow in Eurasia as well.

"imagining that every lineage represents a single and unique clan or tribe (or even individual maybe) making a prehistorical event is going too far".

Of course we can't link each haplogroup with a specific tribe but what we do know is that each mutation must have first occurred in a single individual. That individual may have moved greatly during his or her lifetime but generally I would think only a little way. So from haplogroup distributions we can gain some idea of where a mutation is most likely to have first occurred. Of course it can only have occurred somewhere within the geographic distribution of its ancestor hapolgroup. So if it now appears in a region its ancestor haplogroup is unlikely to have occupied the mutated haplogroup is very likely to be an immigrant.

What we can tell from the mtDNA L branches and distribution is that humans were already widespread throughout sub-Saharan Africa by the time just two subgroups of of one haplogroup, L3, made it out. In fact the mutation separating L3 from L4 occurs relatively late in the process. By which time L1 alone had spread itself right through Central, South and West Africa, not to mention L2.

And if we go back to L0 and L1-6 we can safely bet that the 'original' population consisted of more than just two women each carrying a different haplotype. The haplogroups presumably introgressed into a largely pre-existing population.

Maju said...

No, or very little, N(xR) (and what there is looks as though it is almost certainly a later immigrant) and the Indian-specific nature of the haplogroups in no way argues definitively for an Indian origin.

Yah, it looks like N would be original from further east but R seems South Asian by origin instead (the main Indian branch of N). I think that W could have South Asian origin anyhow because it's most concentrated in the Afghan-Pakistani border anywhere in the World. We'd have to check diversity anyhow to be reasonably sure but the link to Australian N2a of this clade also suggests a southern connection.

N1 is also derived from N1'5, whose N5 branch is South Asian. The only WEA case the would need any explanation is X, if anything.

Notice that the author mentions that he thinks that only c. 2.5% of the Indian mtDNA diversity has been sampled so far and that he still expects important findings.

As for B and F, they are located in Gujarat and other western locations. Not sure what to do with that but I feel it's meaningful of the expansion of R within South Asia first and foremost.

Yet you seem to be arguing against such a back and forth process operating outside Africa in older times, excepting the R expansion. Surely we should expect just such a flow in Eurasia as well.

Through India (logically because of climatic conditionants first of all). You are the one who reapeatedly rejects migrations between South and SE Asia (unless massive evidence is brought forward and even then...)

Of course we can't link each haplogroup with a specific tribe but what we do know is that each mutation must have first occurred in a single individual. That individual may have moved greatly during his or her lifetime but generally I would think only a little way.

But it was then inherited by other individuals and many generations may have happened between one mutation and the rest. My rough estimate for mtDNA SNPs is (from memory, would have to re-analyze to be safe) of some 4 thousand years each on average, what means some 200 generations. If this is correct, the lineage could have been carried by an immense number of individuals before any downstream mutation happened.

In cases like your pet lineage of mtDNA A that would mean something like 32 thousand years and 1600 generations. Enough time for anything to happen. Refine these estimates all you wish but even if you go radical and divide them by four you still would get some 8000 years and the corresponding 400 generations.

So from haplogroup distributions we can gain some idea of where a mutation is most likely to have first occurred. Of course it can only have occurred somewhere within the geographic distribution of its ancestor hapolgroup. So if it now appears in a region its ancestor haplogroup is unlikely to have occupied the mutated haplogroup is very likely to be an immigrant.

I don't think I agree with this. There is no such thing as "the geographic distribution of its ancestor hapolgroup" that is distinct from the the geographic distribution of the derived haplogroups.

Anyhow, even if we can pinpoint with great accuracy the original area of a ancestral node as N, between it and the closest derived lineage (say: S, N9 or N1'5) there's a lapse of some 4000 years (from the early Hittites or Bell Beaker apogee to us, if you need a comparison) and the corresponding 200 or so generations. Even in such closely derived lineages, we cannot trace the events better one node and the next one, between underived N and S for example.

Ibra said...

"We'd have to check diversity anyhow to be reasonably sure but the link to Australian N2a of this clade also suggests a southern connection."

Hi Maju, nice blog :-) N2a in Australia? Is there a reference to that?

Maju said...

Excellent observation, Ibra, thanks. And welcome, btw.

I was assuming that N2b was the same as the Asutralian N* reported by Ian Logan at this graph, which allegedly shares a basal SNP at locus 5460 with W.

But seems I was wrong. N2a is actually referenced at PhyloTree to Metspalu'06, which only deals with South Asia. So it must be a South Asian lineage and actually reinforces the likely origin of N2 and possibly W too in South Asia.

Nevertheless I noticed that Wikipedia mentions an N2a proposed clade that would include W. I wonder if this connects with the Australian N* or is just another newly discovered pre-W South Asian lineage.

terryt said...

"You are the one who reapeatedly rejects migrations between South and SE Asia".

I'm more than happy to accept migrations from SE Asia to South Asia. All the migrations we can actually find evidence for have been in that direction. Migrations the other way are at best hypothetical, although I'd concede mtDNA M and Y-haps F and K probably moved in that direction.

"If this is correct, the lineage could have been carried by an immense number of individuals before any downstream mutation happened".

And that 'immense number of individuals' may not have moved far from where their ancestor with the mutation first appeared. Just a slow expansion, especially so in the case of mtDNA.

"There is no such thing as 'the geographic distribution of its ancestor hapolgroup' that is distinct from the the geographic distribution of the derived haplogroups".

A downstream mutation must happen to an individual with the ancestral haplogroup, and this individual must live within the region the ancestral haplogroup occupies. In many cases the derived haplogroup seems to have then spread much further than its ancestor haplogroup. Of course often members having the ancestral haplogroup would take part in the derived haplogroup's expansion. But it is obvious that in many cases the derived haplogroup has completely replaced the ancestral haplogroup. Survivors with the ancestral haplogroup cannot be found.

"Wikipedia mentions an N2a proposed clade that would include W. I wonder if this connects with the Australian N*".

Possibly. But any connection tells us nothing about which direction it moved.

terryt said...

Thanks for the Metspalu '06 link. Their map of R1a is interesting. Spreads particularly along the Ganges and Indus valleys.

This part is also interesting:

"The O lineages fall out with C* and D (the latter tending to derive from Sino-Tibetan speakers)".

The authors later mention that the O lineages too are associated with both Sino-Tibetan and Austro-Asiatic languages, and came from the east. The authors neglect to mention that Y-hap C* is relatively common around the South China Sea, as well as being found in India.

Maju said...

I'm more than happy to accept migrations from SE Asia to South Asia. All the migrations we can actually find evidence for have been in that direction. Migrations the other way are at best hypothetical, although I'd concede mtDNA M and Y-haps F and K probably moved in that direction

You are entrenched in your "Siberian" pre-conceptions. If mt-M and ychr-F, the two most important Eurasian haplogroups, moved in that direction the rest could perfectly do too.

And you have R as well, as the East Asian R lineages also did that W-E migration.

You argue: if there is strong evidence that they went through South Asia, I concede reluctantly, but if the evidence is even slightly weak, then they used a UFO instead (oops, I mean the "Siberian route").

Well, I contend that the Siberian route is not acceptable unless you can prove it. So, please: start your own blog and prove it if you can.

This arguing in circles can be very annoying, really.

And that 'immense number of individuals' may not have moved far from where their ancestor with the mutation first appeared. Just a slow expansion, especially so in the case of mtDNA

Pure convenient speculation. The more the number of individuals involved the greater the chances of improbable things to happen.

And we're talking of Paleolithic seminomads anyhow.

A downstream mutation must happen to an individual with the ancestral haplogroup, and this individual must live within the region the ancestral haplogroup occupies. In many cases the derived haplogroup seems to have then spread much further than its ancestor haplogroup. Of course often members having the ancestral haplogroup would take part in the derived haplogroup's expansion. But it is obvious that in many cases the derived haplogroup has completely replaced the ancestral haplogroup. Survivors with the ancestral haplogroup cannot be found.

The ancestor haplogroup is dead! Eventually all lines mutate. Nobody carries pure ancestral N or even H. The may carry private lineaages that do not relate with any of the major sublineages but it is distinct from "the ancestral lineage" which is as dead as Jesus.

Possibly. But any connection tells us nothing about which direction it moved.

Like R and like everything else: from India to West Asia.

Ibra said...
This comment has been removed by the author.
Ibra said...

“I wonder if this connects with the Australian N* or is just another newly discovered pre-W South Asian lineage.”

Speaking of that, there are some abstracts online that modify our understanding of haplogroups:

1. The connection between some Indian M types and Austrian M42 and possibly East Asian M10.

2. There are at least 13 new SA M haplogroups (M51-63)

3. South Asian M haplogroups tend to be 10kyr younger than in most regions (abstract explains why)

http://s6.zetaboards.com/man/topic/527761/1/

http://hgm2008.hugo-international.org/Abstracts/Publish/WorkshopPosters/#3

I tend to agree that H and M (SA) are very similar in terms of the number of lineages produced and population growth. The same can be said for D4. 39,33,16 respectively.

Maju said...

Hmmm...

I can't access the first link (not even logged in) and the second one is a long list that looks too complicated to browse.

3. South Asian M haplogroups tend to be 10kyr younger than in most regions (abstract explains why)

Why? Cannot access link but anyhow I think I read the abstract already and explained nothing. Maybe the haplogroups are younger on average but the depiction on the graph was made to look like if there was a different M node for South Asia, what is plainly wrong and misleading (IMO). All M first degree sublineages must be about as old as the others at the stem stage (and that applies to any other haplogroup) the difference may be at their respective subhaplogroup defining nodes, at the moment of their expansion and formation as separate haplogroups (and not anymore just private lineages).

I tend to agree that H and M (SA) are very similar in terms of the sheer number of lineages produced. The same can be said for D4. 39,33,16 respectively.

D4 has 16 sublineages (including D1). This is more in the line of N (12) or R (15), I believe: a major starlike explosion of lineages indeed but not as brutal as M (35 sublineages plus the new ones) or H (31 plus all those private H* sublineages still around).It is still a lot for a Siberian-American lineage. Is it the main D subclade among Native Americans? That would explain it (explosion at the expansion into a virgin land).

Ibra said...

“Cannot access link but anyhow I think I read the abstract already and explained nothing. Maybe the haplogroups are younger on average but the depiction on the graph was made to look like if there was a different M node for South Asia, what is plainly wrong and misleading (IMO).”

I know which paper you are talking about (Soares et al. 2009), however the younger age of SA M haplogroups was calculated in that paper too. The multiple nodes of N, M and was reflective of the pooled variance of each macro-haplogroup for its particular region. The lower age of M in South Asia is equivalent to lower variance (rho value) of SA M brought about by later expansion times. The same goes for other macro-haplogroups in different regions.

“I can't access the first link (not even logged in) and the second one is a long list that looks too complicated to browse.”

Ok I’ll give the abstracts directly:

http://tiny.cc/k76Do
http://tiny.cc/SBBq9
http://tiny.cc/UjXcK

terryt said...

"The ancestor haplogroup is dead!"

Quite. So you won't disagree with me in what follows, although I remember you disagreeing earlier. Back to Metspalu '06:

"The frequency and subgroup variation of J in West Asia, in the context of the complete absence of J1 and most J2 subgroups within the Indian sample, is consistent with an influx of a subset of J2 lineages to India from the Near East, followed by their subsequent diffusion from India's northwest toward the south and east".

Don't you claim that all F-derived haplogroups originated in India? The authors certainly appear to be definite about J in India being the result of immigration into the subcontinent. But as you've often pointed out: its ancestor, some sort of F*, is commonest in India. So what's J doing outside India? It must have come from, or at least through, Iran and perhaps Afghanistan.

Perhaps a few of the basal F-derived haplogroups descend from an ancient F* that had dropped by the wayside during the advance from Africa. G, J, I, and perhaps H may have replaced the dead ancestral F. Maybe from that we can discern the pattern of F's, and so modern human, route.

In the diagram I've already put up at remotecentral I have J1 in Arabia, J2 in the Zagros, G in the Caucasus and I in Central Eurasia. Therefore F's route seems to have been across the Arabian Peninsular through the region of savannah (perhaps just the Levant coast) and then through the Zagros into similar habitat to their north. Not along the coast.

Only after that was F able to enter India. In fact the only Y-haplogroups that we can be fairly certain actually arose in India are F1, F4, L, and perhaps R, H and K1. The presence of all other Y-haps in India is most parsimoniously explained as by immigration from elsewhere. And, of course, even the ones that arose in India arose from haplogroups who had themselves been immigrants, ultimately from Africa.

Maju said...

I know which paper you are talking about (Soares et al. 2009), however the younger age of SA M haplogroups was calculated in that paper too. The multiple nodes of N, M and was reflective of the pooled variance of each macro-haplogroup for its particular region. The lower age of M in South Asia is equivalent to lower variance (rho value) of SA M brought about by later expansion times. The same goes for other macro-haplogroups in different regions.

I see now what it means and thanks for the clarification.

I think it is misleading. Any place that has the largest number of subclades for any haplogroup should give a pooled more recent dates for the set of all those subclades taken togeter.

You should not pool the haplogroups. You should consider them on individual basis one by one, as I have tried to do (with whatever shortcomings) here and here.

When you do that you get a sufficient number of M subclades that appear to expand at the very first moment after the "M Bang", namely: M4'18'30'37'38'43'45, M6'16'17, M22'25, M33 and M31'32 (which is not Andamnese-only anymore).

There are also 4 M sub lineages that seem to expand in Eastern Asia in this first moment.

When you go to the second moment (2 mutations downsteam of M, though I haven't yet refined it for the coding/control mutation distinction), you get 3 new secondary M expansions in SA, 3 in EA and 1 in Melanesia.

Etc.

So while the average SA M sublineages may appear to be younger than those elsewhere, this is surely only because at later moments new expansive processes elsewhere were starred by other lineages (N and specially R derived), while in SA it was largely (even if not exclusively) the M pool still expnanding instead (this also applies to West Eurasian and African M lineages that appear rather "young").

Thanks for the links anyhow. When I read the word "pooled" I understood the matter instantly.

Maju said...

@Terry:

Don't you claim that all F-derived haplogroups originated in India?.

Not at all. I think that F originated in India. The individual derived haplogroups must be considred on individual basis. In the case of IJ and J I do suspect it coalesced in West Asia, of course. Another thng would be JK, which connects too many lineages from West and East Eurasia to be thought as original from anywhere but South Asia (again).

Perhaps a few of the basal F-derived haplogroups descend from an ancient F* that had dropped by the wayside during the advance from Africa. G, J, I, and perhaps H may have replaced the dead ancestral F.

First, J is not basal F-derived. JK is instead (and from JK: IJ).

Second, unless some new evidence comes to light (what may perfectly happen within the Y-DNA context, so badly explored out of some privileged ethnicities), the state of the art is that F1, F2, F3, F4, G, H and IJK (the basal derivates from F) are overall concentrated in South Asia of all places.

Third, I am basing my work hypothesis almost only on mtDNA, whose real structure is much better known for all Humankind and would not be subject to sudden sweeps of the male-only lineages. I'll try to fit Y-DNA to the mtDNA structure and not viceversa.

In the diagram I've already put up at remotecentral I have J1 in Arabia, J2 in the Zagros, G in the Caucasus and I in Central Eurasia.

Your diagram is hypersimplistic.

And here you are mixing apples (F 1st tier sublineages like G) with oranges (2nd and 3rd tier sublineages: all the others).

G is comparable with F1, F2, F3, F4, H and IJK. Those are the known F basal sublineages. We can argue about IJK origins all night, but, excepting G (WEA) and F2 (EA), all the rest are SA. So F expanded in SA with all likelihood.

terryt said...

"Your diagram is hypersimplistic".

Hang on. Your concept is even simpler. I'll try to sum it up:

'Several haplogroups emerged together from Africa and arrived in India by UFO. Next, from there suddenly large groups containing a representative sample of multiple haplogroups all expanded, in all directions'.

Surely we'd still expect to find some evidence for the route F or its ancestors took between Africa and India, somewhere. Covered by rising sea level? Simply an excuse for the lack of evidence. And ancient coastal sites are relatively common in Southern Africa. It's difficult to argue differing degrees of sea level change in different regions without invoking tectonic movements. For which we have evidence in many places, but not for the region in question.

"I think that F originated in India".

But surely you can see that it or its ancestor was almost certainly an immigrant there.

"So F expanded in SA with all likelihood".

I'm prepared to agree that several branches of F expanded in SA, but not necessarily all of them.

"In the case of IJ and J I do suspect it coalesced in West Asia, of course".

That means that someone carrying the IJ mutation must have emerged from India, and some time later members of haplogroup J returned. But IJ's ancestors need never have entered India in the first place.

"G is comparable with F1, F2, F3, F4, H and IJK. Those are the known F basal sublineages".

Yes. We've already agreed on just such a widespread geographical Y-hap diversification elsewhere. Hang on. That was in relation to K1, K2, K3, K4, T, L, NOP, S, M1, M2 and M3. Is there any reason why F would automatically be different?

"G, H and IJK (the basal derivates from F) are overall concentrated in South Asia of all places".

G? Concentrated in South Asia? Only if we include everything between Africa and India south of 45 degrees north. I'd agree with you then.

"Another thng would be JK, which connects too many lineages from West and East Eurasia to be thought as original from anywhere but South Asia (again)".

There is actually no reason at all why South Asia should be the only option.

"J is not basal F-derived. JK is instead (and from JK: IJ)".

And K need not have entered India until after the IJ/K separation.

"I'll try to fit Y-DNA to the mtDNA structure and not viceversa".

Surely both are needed to reveal the overall structure. But there is actually no reason why they should be all that closely connected anyway. That's what so surprised me when I mixed them in my diagram. They often are connected. But just as often they're not, unless they're moving into previously unoccupied territory.

"I am basing my work hypothesis almost only on mtDNA, whose real structure is much better known for all Humankind and would not be subject to sudden sweeps of the male-only lineages".

And therefore mtDNA probably reveals a deeper, slower, and possibly older, pattern of population movements.

Maju said...

Lack of archaeological findings (partial anyhow) only means (IMO) that we should make more archaeology, focusing in that time window and those relevant areas: NE Africa, Arabia and India.

It is the default route anyhow: the simplest one, the most parsimonious one.

And ancient coastal sites are relatively common in Southern Africa.

Not sure what you mean here: AFAIK all archaic human findings from SA are from Transvaal. But I may be wrong.

It's difficult to argue differing degrees of sea level change in different regions without invoking tectonic movements.

Not sure if it's relevant but different coasts have different continental platforms. For instance here you are at 100m depth in no time, while in other places the continental platform has a length of several thousand kilometers.

But surely you can see that it or its ancestor was almost certainly an immigrant there.

Obviously: pre-F diverged in Africa surely.

I'm prepared to agree that several branches of F expanded in SA, but not necessarily all of them.

The F node!!! This has nothing to do with derived branches that belong to more recent time necessarily.

That means that someone carrying the IJ mutation must have emerged from India, and some time later members of haplogroup J returned. But IJ's ancestors need never have entered India in the first place.

No. IJ coalesced in West Asia with all likelihood. It is pre-IJ (IJK*, leading to IJ) which was probably at some time in South Asia, at the F expansion at least and the IJK secondary expansion probably too.

Hang on. That was in relation to K1, K2, K3, K4, T, L, NOP, S, M1, M2 and M3. Is there any reason why F would automatically be different?-

F and K are similar in structure. And that is precisely what I say.

Caveat to your statement: M1, M2 and M3 are not 1st tier K-derivates, the same that P or NO aren't. M is instead.

"G, H and IJK (the basal derivates from F) are overall concentrated in South Asia of all places".

G? Concentrated in South Asia?


Don't misinterpretate me, please. I mean that the majority of 1st tier F sublineages are concentrated in South Asia. Please...

There is actually no reason at all why South Asia should be the only option.

Yes: basal diversity: I repeat:

F derived lineages:

In South Asia: F1, F3, F4, H (total=4)
In East Asia: F2 (total=1)
In West Asia: G (total=1)
Scattered through all Eurasia: IJK (total=1)

Anyone (but you) will agree that SA is the clear homeland of F on this light.

And K need not have entered India until after the IJ/K separation.

Yes, there's no evidence of the contrary. But no evidence in favor of this explanation either.

Maju said...

(continuation)

But there is actually no reason why they should be all that closely connected anyway. That's what so surprised me when I mixed them in my diagram. They often are connected. But just as often they're not, unless they're moving into previously unoccupied territory.

I'll be happy if I can conclude that the Y-DNA F/D explosion is the same as the mtDNA M/N one. And even more if I can find any connection between Y-DNA K and mtDNA R, which so far is quite elusive.

But what I meant is that I first look at mtDNA and then look at the Y-DNA and not vice versa. That's because I think that mtDNA is a lot more stable and gives much more clear signals. Poeple emphasizes Y-DNA because we live in a Patriarchal culture but mtDNA actually speaks much more clear.

And therefore mtDNA probably reveals a deeper, slower, and possibly older, pattern of population movements.

MtDNA reveals the bulk of the movements. Y-DNA can sweep on "random" events easily according to many. When you look at phenotype, for example, it often has no relation with Y-DNA but has instead with mtDNA. MtDNA is distributed much more like autosomal DNA and reflects much better the bulk of the ancestry than Y-DNA.

For example, Greeks and Albanians are high in E1b1b1b but this is because of some founder effect, not because it reflects some migration from Egypt or whatever. Unless you want to call "migration" to a randomly transmitted haplogroup at individual level... what I find ridiculous.

This kind of random accidents are much less likely to happen with mtDNA, and when they happened, they needed a lot more time and very low population levels for all that time.

terryt said...

"And even more if I can find any connection between Y-DNA K and mtDNA R, which so far is quite elusive".

To me the connection between Y-hap NOP (just a subset of K) and mtDNA R seems pretty clear.

"Not sure what you mean here: AFAIK all archaic human findings from SA are from Transvaal. But I may be wrong".

Lombos cave and a few other places in S Africa are considered evidence for early modern human coastal economy.

"different coasts have different continental platforms. For instance here you are at 100m depth in no time".

Same along the southern coast of Arabia and Baluchistan. You can't argue submerged continental platform along that coast.

"It is pre-IJ (IJK*, leading to IJ) which was probably at some time in South Asia".

There's no reason why it has to have ever been in South Asia.

"M1, M2 and M3 are not 1st tier K-derivates, the same that P or NO aren't".

And K is not a first tier F-derivative.

"Yes: basal diversity".

But the basal diversity includes G.

"I mean that the majority of 1st tier F sublineages are concentrated in South Asia".

But that doesn't mean at all that F originated there, just that it had a major expansion there. Probably because it was the first modern human Y-chromosome to enter India, but that doesn't mean it magically appeared there.

Maju said...

To me the connection between Y-hap NOP (just a subset of K) and mtDNA R seems pretty clear.

I don't see it. In North Africa and West Asia mtR is dominant and not just yNOP but also yK is minor. In South and East Asia, as well as Native America, yNOP is largely dominant and mtR is minor.

The only places where such equation works is in Europe. We can add Papua here if we consider all yK and not just NOP and we could add West Asia too if we considered all yIJK and not just K.

But even stretching it so much, the case remains weak for North Africa and East Asia/Native America. In the first case we can conclude easily that some ancestral IJK was largely replaced by E1b1b in a second migration (Capsian) but there's no explanation in the latter, unless you replace mtR by some other clades, at least partly.

In fact this part of the discussion is quite interesting because I'm starting to think that it was IJK and not K (nor NOP) which participated in the mtR explosion. Though then we would still need of some complex male-only mechanics to explain K and NOP.

This, if real, would imply some qualitative change in UP "demographic politics", so to say: from a first phase in which males and females migrated together (let's say egalitarian demographics) to a second phase in which males migrated at last partly alone (machista demographics).

Of course, the other possibility is that males and females just mated with some deal of randomness in general, so each micro-phase must be considered on its own light and we cannot say much on that at continental levels. This is the generally accepted approach.

Now, what Y-DNA lineages would I expect having migrated westward with the earliest mtDNA lineages: N1, R0, U and JT? I'd say that G and IJ are the most likely candidates. The case for R1 or R1b participating in early West Eurasian colonization is weak.

Maju said...

Lombos cave and a few other places in S Africa are considered evidence for early modern human coastal economy.

Blombos cave. Wikipedia mentions that shells increase with time and climatic warming, being rare in the early phase, increased in the second (when the coast was at some 5 km only) and highest in the third one (when sea levels were similar to those of today).

Anyhow, there's plenty of evidence in India for all the modelled migration routes, except in the purely coastal route by the southern tip. Even if you want to argue that this precludes the "coastal migration" you are wrong, because the "coastal migration" may well have followed interior routes as well, such as the very well documented Narmada-Son-Ganges one (which approaches pretty well the straight line between Mumbay and Calcutta). There is also evidence for coastal settlement in the West Indian coast as far south as Kerala and in the East coast too at the Andrah Pradesh/Tamilnadu border.

Same along the southern coast of Arabia and Baluchistan. You can't argue submerged continental platform along that coast.

You can argue lack of archaeological research certainly. There are very few Paleolithic sites known in the Muslim World and most of them were found in the colonial period. Until recently there was also lack of genetic research in Arabia peninsula as well.

There's no reason why it has to have ever been in South Asia.

Well two reasons I can think of:

1. K had to migrate eastward somehow and SA is not just the shortest route but the only one with some archaeology for the MP.

2. It makes no sense to me that IJ (and G) males in West Asia had no women with them or that they replaced them altogether. And all West Eurasian mtDNA, notably the one we can think of as of early arrival (N1, R0, U and even JT) shows clear signs of South Asian origin. So for me the ancestors of IJ and G migrated from South Asia together with the female ancestors of N1, R0 several U lineages (if not all U) and JT.

And K is not a first tier F-derivative.

Who said it is?

But the basal diversity includes G.

And?

1/7 < 4/7; ergo: the greatest basal for F diversity is in SA by far; ergo: SA is the most likely candidate for the homeland of Y-DNA F.

I hate arguing in circles. I can be patient but your insistence in already discussed points without adding anything new is exasperating.

But that doesn't mean at all that F originated there, just that it had a major expansion there. Probably because it was the first modern human Y-chromosome to enter India, but that doesn't mean it magically appeared there.

F appears when it branches out, not before (that could be called pre-F or whatever but has left no traces I know of). And all basal sublineages must have spawned in a short period of time. Most of them are in SA (and the rest around it), so SA is where F expanded. A couple of F sublineages (namely F2 and G) migrated out of India before their own expansion at a later moment but that says nothing against F having expanded in SA.

Please!

terryt said...

"I don't see it".

So you believe, 'once together, always together'. No divorce. A product of you Catholic upbringing?

"the case remains weak for North Africa and East Asia/Native America".

In the latter case surely the A, C and D mtDNAs were simply picked up along the way, as probably was B because it's an eastern haplogroup.

"we would still need of some complex male-only mechanics to explain K and NOP".

Male-only mechanics are quite easy to explain, if through regions where humans are already present. Female-only is more difficult to explain but not impossible.

"K had to migrate eastward somehow and SA is not just the shortest route but the only one with some archaeology for the MP".

But K's migration eastward could well have taken place after it had separated from IJ. And perhaps it left (what later became) T behind. After all T is an African/Southwest Asian haplogroup, not Indian.

"It makes no sense to me that IJ (and G) males in West Asia had no women with them:.

Obviously they had women with them. How about N2/W, X and N1? I totally agree that R0, U and even JT 'shows clear signs of South Asian origin', but I wouldn't be so sure of N1.

"SA is the most likely candidate for the homeland of Y-DNA F".

Not so. I agree we are arguing in circles here, and I'm sure that I hate is as much as you do. But a number of factors can be responsible for a delayed diversity.

"And all basal sublineages must have spawned in a short period of time".

Yes. But G and IJK could easily have formed during a rapid expansion of some sort of basal F before it had reached India. F2 is simply the eastern end of the resulting distribution, and G the western. F2's ancestor obviously passed through India but G need not have. Just because a particular haplogroup's own expansion comes later than that of its related haplogroups doesn't mean at all that it had just recently arrived in the particular region of its expansion. New technology, either invented or introduced, can easily explain an expansion. As can arrival in a so far unoccupied region as the small founding population spreads out.

"A couple of F sublineages (namely F2 and G) migrated out of India before their own expansion at a later moment but that says nothing against F having expanded in SA".

And yet all these F-derived haplogroups separated at approximately the same time? And neither F2 nor G left any trace behind, or of their passing? Doesn't that raise difficulties for your theory? It is much more likely that the basal F had spread before its subgroups differentiated. You are obsessed with an Indian Garden of Eden for some reason or other.

Maju said...

So you believe, 'once together, always together'. No divorce. A product of you Catholic upbringing?.

Fuck off!

What I say is that where did those go? What I say is that they cannot vanish in the sands like Lilith.

This is about real people, having real children in real life. So where did the women and their offspring go?

What I don't accept is that people vanish without leaving any trace, much less female lineages... just because it is convenient to your theories.

Male-only mechanics are quite easy to explain, if through regions where humans are already present. Female-only is more difficult to explain but not impossible.

Both are very difficult to believe before the metal ages, or at least Neolithic, IMO. And notably I find impossible to believe in female-only layers on btop of male only ones.

Making yIJK and mtR associated solves this problem quite well.

But K's migration eastward could well have taken place after it had separated from IJ. And perhaps it left (what later became) T behind.

No basal haplohroups that explode in many lineages as yK don't work that way: going here and leaving one lineage and then there and leaving another like a butterfly.

They expand at a single location and then the derived lineages, in stem phase, migrate around.

Western T but also L and R need a better expalnation than that shit you claim.

Obviously they had women with them. How about N2/W, X and N1?.

All those but N1 arrived after R sublineages, and W much later in fact. Also, they are minor even in West Asia, they are simply just not comparable to the dominance of IJ in West Asia, North Africa and possibly Europe itself before R1 arrived.

I totally agree that R0, U and even JT 'shows clear signs of South Asian origin', but I wouldn't be so sure of N1.

Look N1'5 up. It's a very clear sign and also, if N expanded in SEA (as you agree as likely) then they must have crossed SA again in order to arrive to WA.

Not so. I agree we are arguing in circles here, and I'm sure that I hate is as much as you do. But a number of factors can be responsible for a delayed diversity.

There are not. And also we would not discuss in circles if you:

1. Would stop repeating claims that have been discussed (and refuted) elsewhere.

2. Would stop rejecting the obvious without even proposing an alternative model of their own.

I really think that people who are not upfront and prefer to try to make some sort of guerrilla against others ideals without really proposing alternatives not even making consistent criticisms, who do that once and again and again and again...

Those are intellectually dishonest: mere cheaters of dialectics. A waste arguing with.

So if you have a model, put it forward and defend it.

But if you have a weak, battered model, model, don't put it forward nor defend it... put it down! Be honest with yourself and admit it's mere junk and based on preconceptions.

Take a short vacation and start all over from zero then.

(Of course, if you prefer intellectual dishonesty, then join Ren's forums).

F2's ancestor obviously passed through India but G need not have.

Or vice versa.

I don't even bother building a theory on Y-DNA alone anymore. I build on mtDNA and then make Y-DNA fit with it.

And neither F2 nor G left any trace behind, or of their passing? Doesn't that raise difficulties for your theory? It is much more likely that the basal F had spread before its subgroups differentiated.-

Well that is it: F expanded and overflew SA. F2 and G are the sole products of that overflow.

How many times do I have to repeat the same thing?, how many times do we have to discuss the obvious?, how many times do you have to understand what happens between a basal node (F) and a derived one (G or whatever)? Is this blog the primary school of genetics?

No. Come here with your homework done, please!

terryt said...

"What I say is that where did those go?"

They didn't go anywhere. Their descendants are still there, you idiot.

"What I don't accept is that people vanish without leaving any trace".

What people have vanished without leaving any trace? Apart from all the members of haplogroup G who moved from India?

"All those but N1 arrived after R sublineages, and W much later in fact".

What evidence do you have for that?

"Would stop repeating claims that have been discussed (and refuted) elsewhere".

You haven't actually successfully refuted any of my claims.

"Or vice versa".

You can obviously see that it's impossible to tell which direction members of a haplogroup have moved by simply looking at its modern distribution but you still refuse to see that the spread doesn't involve an already diversified haplogroup. Diversification happens once the earlier haplogroup has spread.

"Well that is it: F expanded and overflew SA. F2 and G are the sole products of that overflow".

You're obviously going (as we say here) off your rocker.

Maju said...

They didn't go anywhere. Their descendants are still there, you idiot.

Where?

What people have vanished without leaving any trace? -

According to YOU the female companions of the Y-DNA G and IJ guys.

... Apart from all the members of haplogroup G who moved from India? -

No member of haplogroup G did such thing. Pre-G did.

Get that straight, please.

"All those but N1 arrived after R sublineages, and W much later in fact".

What evidence do you have for that?
-

It's my reconstruction and is discussed elsewhere on this blog, as you know.

Whatever the case, all Western mtDNA N(xR) is a tiny fraction in comparison to a huge fraction of Y-DNA of F(xK), which is even clearly dominant in West Asia.

Everywhere else it may happen that Y-DNA new arrivals partly or maybe even totally displace the older layer, leaving the mtDNA more or less intact. But the opposite happens nowhere.

So Y-DNA F(xK), G and IJ, must have arrived with mtDNA R as main companion.

"Would stop repeating claims that have been discussed (and refuted) elsewhere".

You haven't actually successfully refuted any of my claims
.

I believe I have. And I also believe you have not even bothered discussing my refutations in their own terms but just leave the issue hanging without really advancing rationally in the dialectics of that discussion and, then, you come back weeks later on a totally disconnected thread and jump again with the same discussion even if way off topic from the very beginning.

It's tiresome for me.

You can obviously see that it's impossible to tell which direction members of a haplogroup have moved by simply looking at its modern distribution but you still refuse to see that the spread doesn't involve an already diversified haplogroup. Diversification happens once the earlier haplogroup has spread.

I think it is possible within reasonable limits. Everything leaves a trace. The traces are there for us to reconstruct the scenario.

If you think it is impossible to reconstruct why do you even bother dicussing it at all?

"Well that is it: F expanded and overflew SA. F2 and G are the sole products of that overflow".

You're obviously going (as we say here) off your rocker
.

No. It's the only logical explanation. There is just no other possible explanation for F's diversity.

You have some prejudice against South Asia. Well, it's your problem. No need to bring it here, thanks.

terryt said...

Obviously it's time to stop this but I'll make a few comments and then leave.

"According to YOU the female companions of the Y-DNA G and IJ guys".

It's you who claims they've disappeared. I claim the female companions are still there in the form of mtDNA N1 and N2.

"No member of haplogroup G did such thing. Pre-G did".

Isn't pre-G some sort of F*? What evidence do you have that this pre-G had come from India and that Indian F* isn't just part of this pre-G/F* migration into India?

"It's my reconstruction and is discussed elsewhere on this blog, as you know".

You constantly demand evidence from others, but not when it's you making the claim obviously.

"You have some prejudice against South Asia".

I don't have any prejudice against south Asia. I just don't believe it explains everything. The story is not as simple as you portray it.

Maju said...

I claim the female companions are still there in the form of mtDNA N1 and N2.

Ok, it's an opinion. But why? Just because they happen to be N? IMO W (N2) is the latest mtDNA arrival to West Eurasia, not the first one.

Also I've been reviewing the mtDNA tree these days, skipping the control area (because it's hypervariable and its mutations should have evolved in much shorter periods of time) and I get that R0 would be even older than N1 in West Asia. The overall scenario doesn't change substantially but some lineages are more or less altered. But by no means R lineages seem to have arrived later than N ones to the West.

In fact, one of the things that appear slightly changed with this new revision is that M and N are somewhat more separated and instead N and R are more tightly together, becoming a single phenomenon or almost.

But well... I have my reasons to think as I do, which are your reasons apart of Indophobia?

Isn't pre-G some sort of F*? What evidence do you have that this pre-G had come from India and that Indian F* isn't just part of this pre-G/F* migration into India? -

As simple as this: pre-F cannot produce pre-G (nor F*). Only F can.

And F (the F node, the F event, the F mini-explosion) seems to have happened in South Asia by all accounts.

"It's my reconstruction and is discussed elsewhere on this blog, as you know".

You constantly demand evidence from others, but not when it's you making the claim obviously
.

I have explained why often to you. Specifically to you. One gets tired of being asked to repeat once and again.

I don't have any prejudice against south Asia. I just don't believe it explains everything. The story is not as simple as you portray it.

*Shrug*