search this blog

Sunday, August 9, 2015

Two Eurasians


If these findings are finally confirmed with ancient DNA, then the human phylogenetic models we've seen recently describing a single Eastern non-African (ENA) lineage won't look too good. But that's a big if. The paper is behind a pay wall, but there's a news story describing the main findings here.

Abstract: The modern human expansion process out of Africa has important implications for understanding the genetic and phenotypic structure of extant populations. While intensely debated, the primary hypotheses focus on either a single dispersal or multiple dispersals out of the continent. Here, we use the human fossil record from Africa and the Levant, as well as an exceptionally large dataset of Holocene human crania sampled from Asia, to model ancestor–descendant relationships along hypothetical dispersal routes. We test the spatial and temporal predictions of competing out-of-Africa models by assessing the correlation of geographical distances between populations and measures of population differentiation derived from quantitative cranial phenotype data. Our results support a model in which extant Australo-Melanesians are descendants of an initial dispersal out of Africa by early anatomically modern humans, while all other populations are descendants of a later migration wave. Our results have implications for understanding the complexity of modern human origins and diversity.

Hugo Reyes-Centeno et al., Testing modern human out-of-Africa dispersal models and implications for modern human origins, Journal of Human Evolution, 8 July 2015, article in press, doi:10.1016/j.jhevol.2015.06.008

34 comments:

Mike Thomas said...

But what about the other study that argued against these findings?

Davidski said...

Do you have a link?

Krefter said...

So, they're saying Australo-Melanesians have distinct cranial to other Asians? That's nothing new. If they also looked at cranial in West and South Asians, I don't see how they could have come to the conclusion they are form the same out of Africa-migration as East Asians. Who is everyone else in Asia?

Australo-Melanesians being distinct in cranial from other Asians doesn't support two seperate out-of Africas? We know from DNA all Eurasians are from the same root, and that Australo-Melanesians share common ancestry with East Asians that they don't with other Eurasians.

I definitely don't think it's smart to include Australo-Melanesians as "East Eurasian" along with Han, Japanese like Laz 2014 and Haak 2015. The two are obviously very divergent populations. Something more complicated is happening.

Mike Thomas said...

I'd seen this

http://www.nature.com/nature/journal/vnfv/ncurrent/full/nature14895.html

And

http://m.sciencemag.org/content/early/2015/07/20/science.aab3884.abstract

Haven't realky read them. I'd just heard that they proposed different explanations for the Melanesian admixture.



Mike Thomas said...

Oh my bad, I was referring to the Amerinidan admixture issue

Kristiina said...

Interesting paper, big questions. However, I am sure that people have been constantly moving back and forth and taking all possible routes that were available, so I agree with the paper when it says that "our results indicate that multiple dispersals out of Africa occurred before ∼37 ka and later than ∼135–80 ka".

Mike, that palaeo-American question is also relevant here. It is really fascinating if there was an earlier migration (120-135 ka) that found its way even in America before 18 ka. Those two papers you referred to and their supplementary material are worth a look. Of course, the differences in the so called Australo-Melanesians could also be caused by admixture between AMH and other hominins.

As I have already commented, it is possible that this earlier wave consisted of mtDNA L3'M and yDNA C and DE. If I am not mistaken, C-M217 (xC-P39) has been found in South America and it is also found in Australia. The later northern wave would then consist of mtDNA N and yDNA F. That would also nicely explain why East-Asians and South Indians look different from other Eurasians. They are a result of admixture between these two waves while north Eurasians result mainly from the later wave. Recently, we have also seen the publication of this mtDNA paper that showed that mtDNA N does not have old coalescence ages in India and the oldest branches are in China.

On the basis of the ancient yDNA that is available to us, my impression is that yDNA gets easily replaced by new lineages that probably had a kind of technological advantage compared to previous males. By contrast, MtDNA remains and also the autosomal trail.

I know that for example Terry has strongly argued for a northern route to yDNA C, but the modern distribution has not shown to be very relevant for the ancient distribution and even less for the earliest migrations 50-100 ka on the basis of the ancient yDNA available to us.

At long last, I would really like to see some ancient DNA analyses from East Asia.

capra internetensis said...

Reich et al in their 2011 paper on Denisovan admixture figured that Australo-Melanesians were descended about half from an Onge-related population and half from a different population that had admixed with Denisovans. This second population is depicted on the graph as springing from the root of ENA (or of all non-Africans, since there were no West Eurasians).

So if there really was an earlier wave, it might have been the guys who hooked up with the Denisovans.

Australo-Melanesians are already pulled away from other Asians by their Denisovan component, so it would be hard to distinguish a different component basal to their Onge-like ancestry.

Kristiina said...

Correction! I meant to say that C-M217 (xC-P39) has been detected in America while other C lineages are found in Australia (C-M347) and Oceania (C-M38). However, C is found in every corner of the world outside Africa and in particular in peripheral areas (Europe - NB! La Braña, Spain; Southwest Asia; India; Australia-Indonesia-Oceania; East Asia, e.g. Japan has several C lineages, Siberia; America). Its old age and distribution fits for an early offshoot of modern ancient humans.

Mike Thomas said...

Kristiina
What do you make of K14 and La Brana being probably different subclades of C ?

Kristiina said...

As many of you probably observed, things are usually complex. I took a look at the haplotree here p. 3 (http://genome.cshlp.org/content/suppl/2015/02/18/gr.186684.114.DC1/Supplemental_Figures.pdf)

Coalescences ages may not ultimately be an obstacle as it is said that age estimates of younger branches tend to be too old and age estimates of old branches too young. If this two-dispersals model should be worked out from the current Y haplotree, the southern branch could not be anything else but DE and the northern branch CT.

However, do not know … Usually results have outdone my imagination, and then all seems so natural.

Kristiina said...

Mike, I think that it says that C is very old and therefore very diverse and was once found everywhere in Eurasia.

Maju said...

Can someone with access synthesize what is the paper actually about, i.e. what is the "evidence" in a nutshell? The abstract says absolutely nothing. Thanks in advance.

In preliminary assessment: the claim seems to clash with the reality of uniparental markers, right? Neither Y-DNA nor mtDNA support anything of the like in any way I can discern. So most perplex, honestly.

Karl_K said...

This study does not account for the effects that admixture between AMH and other hominins may have had on skeletal morphology. Such hybridization could potentially influence the structure of phenotypic variation in extant human groups, biasing the results.

Karl_K said...

@Maju

Yes. The study does not take into account DNA at all. It seems a bit absurd that 100% of the original mtDNA and Y-haplogroups were eliminated by later groups.

fz said...

This paper is based on statistical analysis of cranial parameters of pleistocene and holocene cranial samples. The statistics are impressive, but authors admitted that their results are in conflict with genetic data.

Karl_K said...

@Karl_K

"It seems a bit absurd that 100% of the original mtDNA and Y-haplogroups were eliminated by later groups."

Whatever, Karl! This is obviously what happened with both the Neanderthals and Denisovans. Your argument is what is ridiculous!

Awale Abdi said...

They don't seem to use any DNA based data so this is seriously pretty shaky to be honest.

While it is possible that the whole "Eastern Non-African" grouping that puts Onges, East Asians & Australo-Polynesians together into one subgroup will get more complex with ancient DNA; this idea that Australo-Melanesians or some such are the source of a different Out of Africa migration makes little genetic sense to me-> they seem too autosomally tied to other Out of Africa populations for such a thing to be true.

Chris Davies said...

"while all other populations are descendants of a later migration wave."

Should be descendants of a number of later migration waves, not just the one (at least IMHO).

Mike Thomas said...

Craniometric studies = waftam in today's age.

Awale Abdi said...

"Craniometric studies = waftam in today's age."

Pretty much...

Karl_K said...

""Craniometric studies = waftam in today's age."

Pretty much..."
Pretty much...

Matt said...

Davidski: If these findings are finally confirmed with ancient DNA, then the human phylogenetic models we've seen recently describing a single Eastern non-African (ENA) lineage won't look too good.

Although in particular this would not affect methods which are used to find the Basal Eurasian / WHG+ANE clade structure necessarily. Those really depend on the idea that Basal Eurasian and WHG+ANE clades differ in relatedness to ENA, not so much on the different relatedness of ENA to one another.... Particularly unless it were the idea that extant Australian-Melanesians are totally undescended from a shared later wave of "ENA" (shared with East Asians).

What it seems we'd be talking about here is really a "post African divergence yet pre-Basal Eurasian" edge into Oceanians that is equally related to Basal Eurasian, WHG-ANE and ENA, and also less related to Basal Eurasian, WHG-ANE and ENA than they are to one another. I think that's possible to fit in the post Laziridis AdmixtureGraph models (without any alteration to the known relationships), but I'm not totally sure,

With the crania specifically, it would be kind of interesting to know what the actual shape form variable is that defines the continuity, as I thought there was often a conclusion along the lines of what we find in Dan Lieberman's Evolution of the Human Head - "Note that Holocene crania tend to cluster goether in regional groups and almost all the (Homo Sapiens) fossils 10,000 years old or older cluster together, regardless of their geographic location." (http://tinyurl.com/njeerjt) Although perhaps this depends on what is actually analysed, as (for a non-academic example) apparently Dienekes for instance has done some shape analysis where this is not so.

Davidski said...

Matt,

Structure within ENA populations is going to be relevant to working out what happened in prehistoric Europe. For example.

f4(Yoruba, Scandinavian hunter-gatherers, Han, Onge Andaman Islanders) Z=-3.9

Motala SHGs were supposed to be WHG/ANE, but now we find that they're probably WHG/ANE plus some type of ENA.

It'll be interesting to see how EHG come out when some more ancient samples from Siberia and the Arctic roll in.

andrew said...

I'm not quite so dismissive of cranial studies as Mike Thomas, but do agree that this is very low quality evidence relative to the available genetic evidence and that cranial studies are often at odds with DNA evidence, most recently in the comparison of DNA and unusually shaped skulls from South America.

Another huge problem is summed up in this sentence from the abstract: "[W]e use the human fossil record from Africa and the Levant, as well as an exceptionally large dataset of Holocene human crania sampled from Asia, to model ancestor–descendant relationships along hypothetical dispersal routes."

I'm sorry. There is only so much information you can derive from Holocene era (i.e. less than 10,000 year old) skulls about events that took place in the Upper Paleolithic era or even the late Middle Paleolithic era (i.e. 125,000 to 10,000 years ago), from skulls this young, particularly when we have non-Levantine, non-African archaeology and genetic evidence that tells a story of multiple waves of migration during that time period, including a few much older ancient DNA samples.

It is basically irresponsible to even try to use cranial evidence at all at this point without integrating that data with data from the recent archaeological finds from Arabia and India (pre- and post-Toba), a recent set of Upper Paleolithic human remains from mainland SE Asia ca. 65kya, Mungo Man, the phylogenies of Y-DNA and mtDNA clades found in Aboriginal Australian and Melanesian populations, in Paleo-Siberians, in the Jomon, in the Andamanese, in the New World, and in other Asians, and the evidence from Neanderthal and Denisovan admixture. A frequentist analysis of the cranial data alone, uninformed by Baysean priors from other data, simply cannot offer any accurate insights.

CroMagnon said...

Well, if craniometric conclusions have already been proven wrong, what grounds are there for not being dismissive ?

The only left use for them are studies of local, culturalcontexts, such as the great study performed by Hackenbeck on the appearance of artificial cranial deformation in the late Roman west.

terryt said...

"It seems a bit absurd that 100% of the original mtDNA and Y-haplogroups were eliminated by later groups".

I agree. However it is possible to fit the surviving haploid evidence to the two routes theory. Allowing for some extinction through Central Asia during extremely cold periods.

"The later northern wave would then consist of mtDNA N and yDNA F".

Difficult to make an at all convincing case that Y-DNA F moved west via a northern route. The only F-derived haplotypes found anywhere north of Iran/China/South Asia are the two very downstream haplogroups K2b2 (Q and R) and one branch within K2a (O). All other F-derived groups are spread through the three regions mentioned, one branch in all three: H2 in SW Asia, H1 in South Asia and H3 in SE Asia. Apart from that we have F1 and F2 in South Asia and F3 in SE Asia. It is remotely possible that K2 moved via a northern route but I think it is far more likely it accompanied its 'cousins' east through South Asia. However I agree the mt-DNA N must have moved east via a northern route, but that was part of the earlier movement as N is almost certainly the first mt-DNA to enter Australia.

"we have also seen the publication of this mtDNA paper that showed that mtDNA N does not have old coalescence ages in India and the oldest branches are in China".

Have you seen this paper:

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4460043/

Shows that N probably followed a northern route. There are no basal N haplogroups in South Asia.

"I know that for example Terry has strongly argued for a northern route to yDNA C, but the modern distribution has not shown to be very relevant for the ancient distribution"

I agree, but something must have accompanied mt-DNA N. And that is unlikely to have been any F-derived group.

"C is found in every corner of the world outside Africa and in particular in peripheral areas (Europe - NB! La Braña, Spain; Southwest Asia; India; Australia-Indonesia-Oceania; East Asia, e.g. Japan has several C lineages, Siberia; America). Its old age and distribution fits for an early offshoot of modern ancient humans".

Quite.

"I meant to say that C-M217 (xC-P39) has been detected in America while other C lineages are found in Australia (C-M347) and Oceania (C-M38)".

But C-M217 is called C2 while the Australian Melanesian branches are C1 (C1b2a and C1b2b respectively). Japan has a branch of both C1 and C2 (C1a and C2a), otherwise the branches are fairly regionally specific: C2 in the north and East and C1 in the south. But notably South Asia has only the relatively downstream clade of C1b1a1. Its 'brother' clade, C1b1a2, is found in Borneo. Quite near the Australian/Melanesian C1b2.

Tobus said...

Isn't this similar to what was reported when they sequenced the first Aboriginal genome? http://www.abc.net.au/news/2011-09-23/aboriginal-dna-dates-australian-arrival/2913010

Kristiina said...

I agree that haplogroups may come and go, and it is not impossible that haplogroups that belonged to the first wave have been lost. It is also possible that one day we will find a lost yDNA branch for example in the Near East or Australia that will change all age estimates. A lost branch could belong to DE or CT or their predecessor.

Terry, I checked the age estimates in that recent mtDNA N paper:
L3, Khor Angar, Djibouti 70.8 kya
N11, Kunming China, 75,9 kya
S, Darwin, Australia, 46.8 kya.

I am quite sure that obtaining really old palaeo DNA data will change our current mtDNA and Y DNA haplotrees.

As for yDNA C, I think it was the winner of its time, in the same way as yDNA K has been in later times. Perhaps, there was another yDNA branch that conquered the world before C.

Awale Abdi said...

@ David

"Structure within ENA populations is going to be relevant to working out what happened in prehistoric Europe. For example.

f4(Yoruba, Scandinavian hunter-gatherers, Han, Onge Andaman Islanders) Z=-3.9

Motala SHGs were supposed to be WHG/ANE, but now we find that they're probably WHG/ANE plus some type of ENA.

It'll be interesting to see how EHG come out when some more ancient samples from Siberia and the Arctic roll in."

But do you think the actual grouping of "ENA" which brings together groups like the Andamanese and Han will "collapse" or that it'll remain and just get more complex?

Davidski said...

I don't know. But whatever happens on this front, the ENA branch will have to get more complex sooner or later, because using the Onge for everything probably doesn't work.

Grey said...

I had a pet theory once that given how important iodine is in child brain development that early out of Africa might have been constrained to within x distance from the coast with the interior beyond that as a population sink.

So, out of Africa around all the coast lines east and west until some event somewhere along that range that compensated for the lack of iodine?

(Also why do pictures of OoA usually leave out the western strand along the Atlantic coast to Scandinavia? Seems odd especially as the shortest and fishiest route to Siberia would be from the Baltic region.)

terryt said...

"Perhaps, there was another yDNA branch that conquered the world before C".

C may well have been preceded by some other Y-DNA branch, but C does seem to have been the first to reach Australia as, remarkably, it is not found in New Guinea except in the form of derived Wallacean C1b2a1. Basal mt-DNA is not present in New Guinea either suggesting Australia was settled before New Guinea was. That makes Y-DNA K and mt-DNA M later arrivals in the region.

I wrote yesterday:

"Japan has a branch of both C1 and C2 (C1a and C2a)"

I managed to miss another Japanese specific C2: C2f. And Japanese C1a is specifically C1a1. Even though Japan has both C1 and C2 representatives I don't think for a moment that C originated there. C can only have reached Japan some 30,000 years ago, long after C1b2 ha reached Australia.

terryt said...

Correction: basal mt-DNA N is not present in New Guinea either. Sorry, left out the relevant bit.

CroMagnon said...

One can only wish that Mungo man had been properly sequenced, or at least some dnA had been stored for later (ie now) times when aDNA technology had improved.
Now with the ridiculous left-wing, anti-science red tape, we might never know