Saturday, June 06, 2009

Origins of Indians : Version 8.2

Coastal Migration Theory and I:
I have been a skeptic of the 'Coastal Migration Theory'. In my previous post on the same topic I found the theory unreliable because of Y-haplogroup distribution pattern in India.

Evidence from mtDNA:
A new study says (Via my discussion forum) not only mtDNA N is older than M but also M in East Asia is older than South Asia. At least one study before this also found that M in East Asia and Oceania is older than South Asia. These facts coupled with our stereotypical knowledge of India at the receiving end of most of the migrations from East or West goes against the legend of the Coastal Migration.

Sanity is restored:
The argument put forth for the Coastal Migration which I found rather bizarre is;
"It is easy to move along the coast and colonize the whole world than it is possible in any other way"
I don't understand why early man should have the explorer's zeal. The coastal areas of India are generally abundant with food supplies. Why would any population leave that region?

Another argument is northern East Asia, an alternative root to colonize SE Asia and Oceania( and has been the stereotypical case in the known past), because of inhospitable mountainous and cold environment early humans might not have moved to those regions.

But I would think man moved when he found imposing mountain but not the other way round. I would propose another hypothesis for the rapid movement of early humans.

Quest for warmth:
Since Homo Sapiens Migrated from hot African regions, they might have always preferred to live in tropical regions than temperate regions. The reason we don't find many older haplogroups in Europe and Central Asia because those were cooler regions unlike Africa. Early humans were in constant motion in search of summers. Most likely the early movement that gave rise to early Australians is;

Europe-> Central Asia->northern East Asia->western East Asia->SE Asia->Australia.

Note it's all thro' inhospitable regions never touching the coastal regions. Some of them(Y-haplogroup C and D) did adjuste to northern East Asia. Some continued south until they reached mostly hotter SE Asia and Oceania.

Another wave that moved South Asia took West Asia (where you find mtDNA M1) route. South Asia gave warmth and hospitable region leading to stagnation.

22 comments:

Maju said...

Unless you have an AJHG subscirption the real data won't be available before December (if I'm correct that's the policy of AJHG for opening their archives: 6 months).

What you see is an apparently arbitrary charter where the same genealogical node (N or M) is replicated at different time depths with no explanation whatsoever. You may choose to believe it because it fits your pre-existent belief but I choose not to because:

1. I make no sense of such multplicated nodes

2. I see that all other genetic data adds up in favor of the coastal migration model.

3. I make no sense for N being older than M. If anything it would be the opposite because N is separated from L3 by 5 SNPs while M is only by 4 SNPs. And also because M had a starlike exploxion not replicated in all genetic history, not by N, not by R and not even by H (which is the closest in number of derived haplogroups and "asterisk" private lineages to M but still behind). This massive explosion of M can only mean, IMO, a colonization of a virgin land where almost every single descendant had huge opportunities for resproductive success. N is slightly younger so it had less of that boom (though still got a good deal maybe because it expanded in a new frontier first of all: SE Asia).

Since Homo Sapiens Migrated from hot African regions, they might have always preferred to live in tropical regions than temperate regions. The reason we don't find many older haplogroups in Europe and Central Asia because those were cooler regions unlike Africa.

I fully agree with this and this is still another argument for human migration to have gone via the tropical belt of Asia first of all.

Haplogroup D is clearly (Hong Shi 2008) to be original from SE Asia, even if we can't trace their route there. Haplogroup C is that even more clearly so (though some have argued South Asia). Haplogroup F is clearly centered in South Asia. This in regard to Y-DNA. At lower levels haplorgoup K appears to be centered in SE Asia or maybe South Asia and might be related to the expansion of mtDNA R. We don't find any such traces in the north cold and arid part of the continent. I really can't see how people remains insisting on that.

Manjunat said...

pre-existent belief

Never had any until I came across Y-Haplogroup C distribution India.

All my conclusions follow the genetic data I see. Not based on speculations.

2. I see that all other genetic data adds up in favor of the coastal migration model.

No. There is none.

If anything it would be the opposite because N is separated from L3 by 5 SNPs while M is only by 4 SNPs.

What is the basis for your claim? I don't want Ian Logan's database as a reference. I want a proper study that support this seemingly straight idea.

I fully agree with this and this is still another argument for human migration to have gone via the tropical belt of Asia first of all.

If you had followed my train of thoughts they you would appreciate that being tropical and food abundant region of South Asia would have made the need for migration of early humans non-existent. It's the cold regions that would have driven the early humans for continuous migration in search of summers. Only that would have set nomadic lifestyle of those population. But South Asia is hardly the region that instill nomadic tendency.

Maju said...

What is the basis for your claim? I don't want Ian Logan's database as a reference. I want a proper study that support this seemingly straight idea.

Check N. Maca-Meyer 2001, for instance.

But a more easy to use compendium of all mtDNA is at PhyloTree that looks like it's going to become the YSOGG of mtDNA (something much needed since long ago).

M has 4 basal mutations at loci 489 (control region), 10400, 14783 and 15043 (coding region). N has 5 basal mutations at loci (all in coding region) 8701, 9540, 10398, 10873 and 15301.

If you had followed my train of thoughts they you would appreciate that being tropical and food abundant region of South Asia would have made the need for migration of early humans non-existent.

True... until they became too numerous. The main drive for migration (apart of human innate curiosity and whatever social conflicts that may exist) is overpopulation. Overpopulation in relation to the economic possibilities of the culture and technology at each space-time, of course.

The other reason is comparatively low pressure in the direction of migration. You can compare to the movement of fluids in fact. South Asia may have been a big cistern or balloon but at some moment the pressure inside became higher than the pressure outside and migration began.

In fact, considering that South and SE Asia had similar climates, makes extremely logic that even a tiny excess of pressure in one of the two cisterns would cause movement towards the other. In other cases, like Neanderthal "pressurized" Central and West Asia, they had to wait till the overall pressure in the southern Asian strip became really enough to overflow in that direction (and/or till the comparative counter-pressure exerted by Neanderthals and aridity was efefctively weaker).

Only that would have set nomadic lifestyle of those population. But South Asia is hardly the region that instill nomadic tendency.

Certain nomadism is intrinsecal to hunter-gatherer way of life. And anyhow migrations always happened in the course of many generations, each of which maybe only moved a few dozen or maybe hundred kilometers. Lineage migration can even actually happen without any population migration as such like via marriage exchange.

And if you don't accept all that, just look at the Roma... I can't think of any other more nomadic people in modern Eurasia and, yes, they are original from India by all accounts. And of course there's a huge number of haploid lineages that appear to have originated in South Asia. So it's not like SA has been always in the recieving side: often it has been the other way around.

terryt said...

"I have been a skeptic of the 'Coastal Migration Theory'".

So have I, as Maju is well aware. I agree with your assessment of the route to Australia. It's the only route that makes sense for that continent. However I agree with Maju regarding the expansion of Y-hap F and mtDNA M. I believe they both arrived in SE Asia, and so East Asia, from India. But not by following the coast. The comment, "This massive explosion of M can only mean, IMO, a colonization of a virgin land where almost every single descendant had huge opportunities for resproductive success" is relevant. I agree. They moved into the largely uninhabited forested mountains of East and Se Asia. But I cannot decide if mtDNA M reached SE Asia before or after mtDNA M, but I'm pretty sure it crossed Wallacea after N.

"N is slightly younger so it had less of that boom".

The 'less of that boom' could easily be the result of extreme selection, a product of its route.

"I make no sense for N being older than M. If anything it would be the opposite because N is separated from L3 by 5 SNPs while M is only by 4 SNPs".

Wouldn't 5 SNPs take longer to accumulate than 4 SNPs? Of course the number of SNPs is not simply a function of time. It's a matter of what survived.

"You can compare to the movement of fluids in fact".

A very useful comparison, and I use it all the time. The land surface of the earth is like a very irregularly shape pond, and mountains, deserts and dense jungle interrupt the flow.

terryt said...

From your previous post on the subject:

"But C* is not a haplogroup properly speaking but a paragroup. A pragroup that apparently is only observed in South Asia and parts of SE Asia".

I wouldn't say 'parts of SE Asia'. It's actually quite common all around the South China Sea, especially in the Philippines. Therefore the statement, "A distribution that fits well with the coastal migration model" is quite true. But the coastal migration was from east to west.

Maju said...

Wouldn't 5 SNPs take longer to accumulate than 4 SNPs? Of course the number of SNPs is not simply a function of time. It's a matter of what survived.

In very abstract statistical terms, more SNPs should mean longer coalescence times. Of course there is a very slight chance that 100 SNPs accumulate in a single individual (and that such highly mutated individual survives) but is negligible. Inversely there is a very slight chance that no mutation whatsoever accumulates over the course of tens of thousands years but again it is negligible.

If statistical probablity is going to give us any information on all this (with whatever margin of error), we must necesarily assume that 5 mutations probably took longer than 4 to accumulate. The opposite is also possible but much less likely.

In any case there is no apaprent reason to claim N as older than M. Much less to claim separate regional subnodes for N and M: that has abslutely no phylogenetic grounds.

A very useful comparison, and I use it all the time. The land surface of the earth is like a very irregularly shape pond, and mountains, deserts and dense jungle interrupt the flow.

My approach is less merely physical. Maybe the Ice Age Hymalayas or the immense width of Pacific Ocean are absolute barriers (for Paleolithic tech levels) but most of the "obstacles" you usually mention are not such thing.

If a "barrier" is not absolute: it will anyhow allow the flow. At most it can slow it down.

Hence the main issue is the opposite pressure at the possible destination, which can act as a much more effecive barrier than physical obstacles in between. This opposed pressure can be a physical element like cold, aridity or whatever but most often will be population pressure. That's why Beringians expanded easily into America but hardly into Siberia, that's why West Asians expanded easily into Europe (once overcome the Neanderthal obstacle) but only limitedly into South Asia, etc. And that's probably why meaningful gene flow between South and SE Asia stopped at some time: because the respective pressures (already pretty big surely) went in equilibrium (and whatever crossed since then was just like a drop in the sea).

But the coastal migration was from east to west.

For C only. But how did pre-C reach SE Asia? UFO or fast coastal migration?

Manjunat said...

This massive explosion of M can only mean, IMO, a colonization of a virgin land where almost every single descendant had huge opportunities for resproductive success

This should be matched by archeological findings of late Paleolithic period. I'm not so sure about it. South Asia, I believe is one of the most pathogen infected region along with Africa and Western Europe. Wouldn't this check population explosion. I believe many communities that continued in tribal life didn't explode until recent times in India(some of them probably shrunk). Can't this be taken as a model?

Does geography affect the mutation rate?

Maju said...

Against popular belief in this age of institutionalized asepsia, pathogens have never really been any major obstacle to populaton growth (though indeed too large population growth can magnify their propagation and effects). Even the Black Death only killed a fraction of the whole population and cannot be considered even close to a real bottleneck (and the expansion few decades after it took place compensated its damage rapidly).

What is indeed a major barrier to population growth is population growth itself, or, in other words: resources and our ability to exploit them.

Also 60,000 years ago, India was mostly savannah, not jungle and swamp as would be in later times. Tropical Savannah is the most natural enviroment for humans: we evolved there.

I agree with the need for an archaeological backing. Sadly what I have researched on Asian MP and UP archaelogy shows that the area has only been researched so much. Still c. 75,000 years ago there is a sudden aboundance of findings, many of which even show a clear trend towards the most archetypical UP style of stone tech (blades) - though of course not all UP is made up of stone blades, the East specially is aboundant on flake and cobble techs instead.

Also the actual patterns of known settlement fit well with GIS models(excepting that we lack of much evindece for the strict coastal route because of higher sea levels).

Does geography affect the mutation rate? .

Should not in principle.

Population size instead may alter the whole estimates a lot. It's not the same to have X chance of mutation per capita among 100 people than among 10,000 people. On the other side, rare mutations have less chances to become dominant in large populations but also to survive "dormant" (as small "private" lineages) among them. Instead small populations sizes emphasize drift and eventual fixation (but also extinction of other less fortunate lineages).

It is a complex matter because we really have not the slightest idea on which were the actual population sizes beyond some very generalistic guesses, and these sizes, apart of a very general trend of growth, may have fluctuated locally in ways we cannot really understand.

So it's only natural to ignore that part (but by ignoring it we itroduce a huge potential for error).

What is clear is that in a large or growing population, minor lineages will probably survive for long enough as to be there when a opportunity to cause a founder effect arises. This is surely the case with all those lineages that appear "hidden" until they cause founder effects, sometimes in areas far away from their relatives.

Maju said...

Erratum:

On the other side, rare mutations have less chances to become dominant in large populations but also to survive "dormant" (as small "private" lineages) among them.

The above (my) sentence is wrong for two reasons:

1. The second half should read as "they have much better chances to survive as dormant or small private lineages" in large populations (not less!)

2. The first half is actually confuse: all clades have less chances to becoem dominant in a large population, as drif is reduced. But all clades also have much better chances of survival for the same reason.

Sorry about this (can't think well when congested). Hope that now it's clear.

terryt said...

"If a 'barrier' is not absolute: it will anyhow allow the flow. At most it can slow it down".

Do I read you correctly? What have we been arguing about then.

"But how did pre-C reach SE Asia? UFO or fast coastal migration?"

There is another option but you, like some religious fundamentalist, refuse to see it.

Maju said...

There is another option but you, like some religious fundamentalist, refuse to see it.

Man, you're trying to sell that the shortest path between two points is not just the curved line but a very peripherical one via habitats that are clearly hostile to human (AMH) inhabitation and that were at the time probably controlled by (better adapted) Neanderthals anyhow.

If you would have at least some evidence, some indication... but nothing. That's why I call that the UFO model, because it's as likely as colonization by means of UFO abduction.

Manjunat said...

My post proposes a model where early human migrated because of inhospitable regions. There wouldn't have been such rapid migrations if the regions were hospitable like South Asia. I think the present genetic evidences (right or wrong) only point to that direction. No need for alien intervention in my model.

If number of mutations make M older than N, I wonder why I haven't seen any present studies making use of that information. Is it considered unreliable? Also, M has both control region and coding region mutations whereas N has only coding region mutation differentiating from L3. So, is it possible to compare the number of mutations of these two for coalescence period? I mean the mutation rate could be different for control and coding regions.

Maju said...

My post proposes a model where early human migrated because of inhospitable regions. There wouldn't have been such rapid migrations if the regions were hospitable like South Asia

That's what I cannot agree with. I think that the opposite is much more logical: good region -> fast expansion -> migration -> bad region -> expansion stopped. It's the law of diminishing returns.

If the area would have been arid an unhospitable, there would have never been a massive expansion to begin with because the country could not support it to begin with. This may have been the case in southern Arabia before crossing at Hormuz: such situation shows a no growth scenario not a fast expansion at all.

If number of mutations make M older than N, I wonder why I haven't seen any present studies making use of that information.

MCH uses short tandem repeats (STRs) and not SNPs. Guess that's the reason.

In any case there is huge uncertainty about age estimates based on mere genentic data: we can know that some mutation is upstream (older) or downstream (younger) but cannot know for sure which is exactly their age. There are some controversial models that make a lot of arbitrary assumptions even before starting the calculations and are therefore essentially unreliable. TRMCA is not at all comparable with C14 or other archaeological datation methods, that have proven their reliability once and again.

As M and N are sisters we just cannot say much but that they are younger than L3 and older than their descendants. Assuming that they expanded at about the same time is a safe assumption, though, if anything, I'd say that N is the "younger sister" and never the other way around.

Also, M has both control region and coding region mutations whereas N has only coding region mutation differentiating from L3. So, is it possible to compare the number of mutations of these two for coalescence period? I mean the mutation rate could be different for control and coding regions.

That's a good observation but I cannot really provide a good answer. If we'd take only coding region, then N would look even younger. If we'd take instead only control region then N would be not different from L3.

Seems that in some species the control region, aka hypervariable region (HVR 1/2), evolves very slowly in contrast to the coding region. For what I've seen, that doesn't seem to be the case in humans, where most lineages have almost as many control region mutations as they have coding region ones - this in spite that the coding region is almost all the length of the mtDNA (like 95%).

I'd take the coding region as more relevant if anything.

Manjunat said...

good region -> fast expansion -> migration -> bad region -> expansion stopped

Very soon you'll appreciate the fact that how counterintuitive it sounds.

terryt said...

I'm inclined to agree with Manjunat on that. Any good region is likley to be already occupied. Expansion usually occurs into less desirable, or at least less accessible, regions, often through the development of improved technology. With a change in technology expansion into the new ecological unit may be rapid, but often only involves just a subsection of the available population, i.e. just a few haplogroups. If a population is suddenly able to expand it implies that for some reason they had not been able to expand until then. Something changed.

Maju said...

I'm inclined to agree with Manjunat on that. (...) Expansion usually occurs into less desirable, or at least less accessible, regions...

Then you are agreeing with me, I think.

Any good region is likley to be already occupied

Not at the begnning of all. Not when our ancestors arrived there for the first time.

If a population is suddenly able to expand it implies that for some reason they had not been able to expand until then. Something changed.

Yah. And is not that what you see in the basic coastal migration model? The deserts of South Arabia were replaced by the aboundance of mainland Asia. Population exploded then.

terryt said...

"Not at the begnning of all".

If humans (or their close relations) had been able to reach any good region they would have already occupied it, even at any 'beginning' you wish to nominate.

"And is not that what you see in the basic coastal migration model?"

There's no evidence at all for a 'basic coastal migration'. On the other hand there is abundant evidence for modern humans in the Levant by soon after 100,000 years ago. Sure, they were later replaced there by Neanderthals as the climate cooled but they could have gone almost anywhere by then, including India. The first moden humans to leave Africa most likely followed the usual human migration route between the Levant and mainland Asia, over land via the Tigris and Euphrates river valleys, and then through the Iranian Plateau to India and Central Asia. There's no need at all for them to pass through 'The deserts of South Arabia'. That's a modern urban myth invented by Spencer Wells.

Maju said...

Let's see:

Humans (mostly L3) leave Africa, live for a time in small numbers in Souther Arabia and eventually arrive to South Asia. Only then there is a real demographic explosion, explosion that is well attested in the huge starlike structure of haplogroup M, the largest of any haplogroup (only H approaches such dimensions somewhat).

Only the optimal conditions of Ice Age South Asia would allow for such explosion. At least that would be within the simplest model of rapid coastal migration.

A more complex model could be speculated with people migrating via North Africa and West Asia earlier and with a bottleneck later on, maybe caused by a combination of Neanderthal pressure, increased aridity and the Toba supervolcano. Nevertheless, at this moment, I think that mtDNA rather supports the first model. The apparent time of coalescence of M and N is not long enough to support an OoA 100 kya probably. Whatever the case, the explosion of modern Eurasians, as attested by mtDNA only happened in South Asia after the Toba event with all likelihood and whatever existed earlier in West Asia and North Africa seems to have left no or nearly no genetic remains.

terryt said...

"live for a time in small numbers in Southern Arabia".

Is there the slightest bit of evidence for that?

Manjunat said...

Only then there is a real demographic explosion, explosion that is well attested in the huge starlike structure of haplogroup M, the largest of any haplogroup (only H approaches such dimensions somewhat).

Here H, I suppose H1, which is also younger in India(Sengupta et al.). The demographic explosion might have happened probably during Neolithic times. Again, if H also exhibits starlike structure, it goes against the Coastal Migration as we don't find H in SE Asia/Oceania.

Maju said...

Is there the slightest bit of evidence for that?.

I s part of the RCM hypothesis. The area has not been archaeologically researched for the most part and local genetics just begin to show up. In any case the relative aridity of the area would perfectly allow for rather fast drift and no expansive process.

What we know that favors such model:

1. Indian MP is close typologically to African MSA (MP), notably to South African one.

2. Eurasian mtDNA is part of L3 and there are not known intermediate lineages (pre-M, pre-N) found anywhere (like, say Palestine or North Africa, where older human remains are known).

The remainders of the migrants in peninsular Arabia could well have gone extinct as did their relatives further north, because of marked increase in aridity at several periods. They could also have been reabsorbed by other migrations (back-migrations from Asia mostly). You still have 14% L6 in Yemen and an aboriginal group in that area (Mahra Arabs) that are more Veddoid/Australoid (i.e. archaic Eurasian) than anything else.

Here H, I suppose H1, which is also younger in India(Sengupta et al.). The demographic explosion might have happened probably during Neolithic times. Again, if H also exhibits starlike structure, it goes against the Coastal Migration as we don't find H in SE Asia/Oceania.

Not sure if we're talking about the same, Manju. I mean mtDNA H, which shows a very large starlike "explosion", IMO centered in Europe (I'd say Central Europe to be more precise). It is the only mtDNA node comparable in its dimensions with that of mtDNA M, though still somewhat smaller. These huge demographic explosions seem to require an unusual window of opportunity: a virgin or nearly virgin land of aboundance where people could be so successful so quickly.

Manjunat said...

Well, I thought you were talking about Y-Haplogroup H! My mistake.