December 22, 2010

Archaic Denisovans contributed to modern Melanesians

I love articles like this, because they force us to re-evaluate everything we thought we knew. I'll probably have much more to say on this paper once I read it, but for the moment I can't help but notice that the finding that the Denisova specimen belonged to a population that contributed DNA to modern Melanesians puts the description of various early Homo sapiens skulls as "Australoid" by various researchers in the past into a whole new perspective.

Indeed, I myself have used MCLUST to classify Upper Paleolithic skulls, and a number of them such as Markina Gora get the label of "Australoid". This is usually explained as an consequence of their greater robusticity, which links them to modern Australo-Melanesians, but the finding that a pre-modern population that lived in Eurasia did contribute genes to Melanesians, certainly raises all sorts of questions.

UPDATE I

This is not a simple paper to read, if we also add the extensive supplementary material, so I will probably give my impressions and thoughts on it in piecemeal fashion.

The first interesting part comes from Table 1 in the paper. This contains values of the authors' D statistic D(H1,H2,archaic,chimpanzee) which shows how more frequently population H1 matches an archaic group than H2 does. A positive value suggests that H1 is more "archaic"-like.

The authors consider Neandertals (in the form of Vindija and Mezmaiskaya), as well as Denisova as the "archaic" groups of interest.

D values are consistently positive for H1=Papuans/Melanesians and H2 either African or Eurasian. This suggests indeed that Australo-Melanesians have archaic admixture from a population related to Denisova.

UPDATE II

It is interesting that the authors are back-pedalling on the idea that specifically Neandertal admixture is responsible for the "archaic" genes found in Eurasians in the previous paper.

I think there are two reasons behind this: first of all, all Eurasians are closer to Denisova than Africans are. Melanesians are even more closer. But, Eurasians are closer to Neandertals than they are to Denisovans.

What this means is that Eurasians did not admix with Neandertals themselves but with a population that was closer to Neandertals than to Denisovans.

The second reason why I think that the idea of specifically Neandertal admixture is rejected is the fact that it makes no sense: the D statistic is 7.5 for Cambodians and 3.3 for Sardinian with standard errors of 1.2 and 1.5.

This means that Cambodians have more "Neandertal" admixture than Sardinians do, which makes absolutely no sense if specifically Neandertal admixture was the reason, as Neandertals were a West Eurasian-distributed species.

So, while not all is well for the Neandertal admixture theory, they also argue (in the supplementary material) that the alternative theory (which I've argued for, of archaic structure) is weakened by the new evidence. I'll think about their argument in a future update.

UPDATE III

The authors did not consider the possibility of modern human to Denisova gene flow. Here is their reason (from the supplement):
Gene flow from modern humans into the ancestors of Denisovans is not only unsupported by the D-statistics, but is also historically implausible. The Denisova phalanx is more than 30,000 years old, and in our opinion is likely to be more that 50,000 years old (SI 12). The more ancient age estimate is older, and the more recent age estimate is only slightly younger than the age of the oldest confirmed modern human remains outside of Africa and the Levant. It is difficult to envision a plausible scenario in which the Denisovan population could have ancestry from a modern human group that experienced mixture in an area near where Melanesians live now, and then migrated to Siberia in just a few thousand years.

And, yet, we have evidence now of much older modern humans in South China, in the form of the Zhirendong mandible, which dates from ~100ky.

UPDATE IV

Looking at supplementary table S8.2 is quite interesting, because it gives the D statistics for intra-African "Neandertal" gene sharing.

The highest one is with H1=Yoruba and H2=Mbuti at 2%. What this means is, in essence, that Yoruba are more Neandertal-like than Mbuti are. How is this possible if a positive D value is reflective of admixture between Eurasians and Neandertals? There are no Neandertals in Nigeria to Neandertalize Yoruba with respect to Mbuti Pygmies.

This either means that Neandertal admixture is not the cause of the positive D statistics, or alternatively, that Yorubans are not pure Africans but have experienced back-migration of Neandertal-admixed Eurasians that Mbuti Pygmies did not experience to the same extent.

How interesting that Mbuti Pygmies (the least "Neandertal") and Papuans (the most "Denisovan") are the two most divergent living humans with an Fst of 0.377. Is it the case the different patterns of archaic admixture are contributing to this?

UPDATE V

It is unfortunate that the authors did not sample East Africans yet again. The authors do not consider Z-scores below 3 (in absolute value) significant, so, the Yoruba/Mbuti score (for Neandertals) of 2.4 does not achieve significance, but it is almost there.

If the authors' theory is correct, then we expect to see no variation (or only variation attributed to noise) within native African populations with respect to their inferred "Neandertal admixture", as Neandertals were a Eurasian species.

It is perplexing why yet again East Africans were not sampled: if you are claiming Eurasian-African differences in archaic admixture, sampling East Africans who live in-between Eurasians and Africans is the natural thing to do!

Nature 468, 1053–1060 (23 December 2010) doi:10.1038/nature09710

Genetic history of an archaic hominin group from Denisova Cave in Siberia

David Reich et al.

Using DNA extracted from a finger bone found in Denisova Cave in southern Siberia, we have sequenced the genome of an archaic hominin to about 1.9-fold coverage. This individual is from a group that shares a common origin with Neanderthals. This population was not involved in the putative gene flow from Neanderthals into Eurasians; however, the data suggest that it contributed 4–6% of its genetic material to the genomes of present-day Melanesians. We designate this hominin population ‘Denisovans’ and suggest that it may have been widespread in Asia during the Late Pleistocene epoch. A tooth found in Denisova Cave carries a mitochondrial genome highly similar to that of the finger bone. This tooth shares no derived morphological features with Neanderthals or modern humans, further indicating that Denisovans have an evolutionary history distinct from Neanderthals and modern humans.

11 comments:

German Dziebel said...

"This means that Cambodians have more "Neandertal" admixture than Sardinians do, which makes absolutely no sense if specifically Neandertal admixture was the reason, as Neandertals were a West Eurasian-distributed species."

This arguments also kills the idea that Neanderthals injected their genes into Homo sapiens in West Asia before our species differentiated into Europeans, Asians and Papuans.
But the study still mentions two admixture events: one for all of Eurasians and one for Papuans.

"So, while not all is well for the Neandertal admixture theory, they also argue (in the supplementary material) that the alternative theory (which I've argued for, of archaic structure) is weakened by the new evidence. I'll think about their argument in a future update."

You could also think about the possibility that the "archaic admixture" in our species is not admixture or substructure but the evidence of the greater antiquity of non-Africans (especially Papuans) than Africans. San and Yoruba divergence may not be the sign of their antiquity but, say, of higher mutation rate in the African lineage.

German Dziebel said...

"How interesting that Mbuti Pygmies (the least "Neandertal") and Papuans (the most "Denisovan") are the two most divergent living humans with an Fst of 0.377. Is it the case the different patterns of archaic admixture are contributing to this?"

In most studies, the greatest Fst is between Africans and Amerindians (Melanesians are slightly less divergent than Amerindians). See http://blogs.discovermagazine.com/gnxp/2010/08/which-population-is-most-genetically-distant-from-africans-amerindians/

Amerindians can't be admixed with pre-existing hominins as there were none of those in the New World. So, this another reason why the story that the data tells us is not that of admixture.

clusteredmaps said...

“Yorubans are not pure Africans but have experienced back-migration of Neandertal-admixed Eurasians that Mbuti Pygmies did not experience to the same extent.” Perhaps macrohaplogroup E is the signal of Eurasian ancestry?

Anonymous said...

"This either means that Neandertal admixture is not the cause of the positive D statistics, or alternatively, that Yorubans are not pure Africans but have experienced back-migration of Neandertal-admixed Eurasians that Mbuti Pygmies did not experience to the same extent."

There are no "pure Africans" and gene flow between West Africa where the Yoruba live and Eurasia has been going on for probably since the last migration from Africa, and has only increased in the last half a thousand years with the increase of maritime contacts between West Africa and Europe since the 15th century.

terryt said...

Fascinating stuff Dienekes.

tt9j said...

Dear Dienekes;
Maybe the Zhirendong mandible is Denisovan ... from the report it is a 'complex mosaic of distinctly derived,
modern human features' (the anterior mandibular symphysis) 'combined with archaic features of the lingual symphysis and
overall mandibular robustness'
How do the isolated teeth also found there compare with the Denisovan tooth?

Tuuli Lappalainen said...

"Amerindians can't be admixed with pre-existing hominins as there were none of those in the New World. So, this another reason why the story that the data tells us is not that of admixture."

They descend from Eurasian populations that have archaic admixture. And experienced subsequent drift.

This is absolutely fascinating stuff. Will think more of the different admixture hypotheses.

Anonymous said...

In my ancestry breakups using my SNP results, a consistent amount of Oceanian has been found. It is small, about 0.7%, but it is interesting that I have SNP results which link me to Melanesians and Papuans, a group of humans that I have absolutely no contact with in time.

Anonymous said...

According to the phylogenetic tree in the paper, Denisovans separate from Neandertals at about the same time with the separation of Sans from the rest of modern humans. So Denisovans may actually be an eastern race of Neandertals that didn't participate in the admixture between modern humans and Neandertals immediately following the modern human migration out of Africa but only mixed with Proto-Australo-Melanesians when Proto-Australo-Melanesians were still somewhere in Asia.

terryt said...

"Will think more of the different admixture hypotheses".

To me it has always been ridiculous that anyone can assume that 'modern humans' emerged from Africa as a relatively small group, spread rapidly round the world and then remained isolated, each in their own particular region, until history began. And over the whole of that period they maintained a balancing act within their environment so that their numbers just slowly and continually increased. Surely thay moved around and formed hybrids with other humans they met along the way.

We know from haplogroup distribution that modern human haplogroups must have moved vast distances during the Paleolithic. Take the Native American Y-hap Q for example. It is related to Y-hap R, so presumably it coalesced somewhere in Central Asia. But its ancestor haplogroup, P, must have moved through India from SE Asia. It formed part of the haplogroup MNOPS. And before then it must have come from Africa, probably in the form of CDEF. So it zigzagged across the world several times before it reached South America.

Surely it's no great step to see that pre-modern humans had also been capable of huge movements.

eurologist said...

While Krause himself indirectly speculates admixture of Asian erectus:

http://www.faz.net/s/Rub268AB64801534CF288DF93BB89F2D797/Doc~EEFF7851E3E6B461F9E75BB829AB775AD~ATpl~Ecommon~Scontent.html

I have no problem associating Denisova mainly with eastern heidelbergensis.

...I commented this in another blog: outside of Spain, antecessor/ heidelbergensis are not really distinguished (their separation time is minuscule; their main difference is growing and larger brain size in heidelbergensis), which means the earliest heidelbergensis or its immediate "antecessor" are about 0.8 to one million years old, whereas Neanderthals only start to diverge ~400,000 years ago, with full-blown features by ~150,000 to 250,000 years ago. So, if you consequently date the autosomal separation time ("split" - not last common ancestor - to both humans and heidelbergensis) at 400,000 years ago, some heidelbergensis mtDNA lines would be expected to be much older (twice as old or older), since originally, Neanderthals were geographically quite restricted, with severe bottlenecks. This is all in agreement with the Reich et al. paper (~1 million year mtDNA divergence time for Denisova).

Non-neanderthalized heidelbergensis arrived in China ~200,000 years ago (sometimes classified as early Homo sapiens to get around the issue of stating relation). So, outside of some tropical pockets left with erectus, this population IMO makes a much better candidate for mating with newly-arriving AMHs than Asian erectus does.

Human occupation at Denisova started at ~280,000 years ago, again in agreement with the suggested wave of heidelbergensis eastward (starting in Europe around 400,000 years ago, and arriving in China by 200,000 years ago).