Neanderthals, Denisovans and everything else

A recent analysis of the nuclear DNA of a Neanderthal toe from Altai has caused widespread interest.

Kay Prüffer et al., The complete genome sequence of a Neanderthal from the Altai Mountains. Nature 2013. Pay per view → LINK [doi:10.1038/nature12886]

The story of a finger and a toe

Both the Denisovan and Neanderthal DNA sequences discussed in this paper come from small bones found at the same location: Denisova cave, Altai Republic. The Denisovan sequence that revolutionized human paleogenetics a few years ago corresponds to a finger phalanx bone of some 50,000 years ago. The less notorious Neanderthal sequence discussed in this study corresponds to a toe imal phalanx, which was found in a lower layer in the same gallery of the same cave, and hence should be older.

This is very interesting to underscore because it seems to imply that Neanderthals were in Altai and specifically in Denisova cave very early, at dates similar to those we find in West Asia (Tabun excepted) and they may even be older than Denisovans in the very cave that gave them their name.

The toe sequence was found in a previous study to have Neanderthal mtDNA, closely related to the lineages of European Neanderthals of various dates and sites. Instead the finger mtDNA (Denisovan) was derived from a more ancient branch of humankind than the very point of split between Neanderthals and modern humans (H. sapiens) and has been recently shown to be related to European H. heidelbergensis from Atapuerca. 

Notes in red are mine.

This study focuses on the autosomal DNA of both Neanderthals and Denisovans. Unlike mtDNA, whose phylogenetic position is simple and quite straightforward, autosomal or nuclear DNA (nDNA) is extremely much more complex to understand because of its recombining nature, requiring of statistical approaches, which may get extremely complex and potentially subject to premise biases. When comparing two individuals this gets largely simplified but it is a lot more complex when doing the same with larger samples.

And that is precisely what this study does: comparing one Denisovan, several Neanderthals and also several modern humans. Therefore it is a very complex paper and the authors necessarily assume some evaluation risks, which nevertheless are discussed in depth in the supplemental material, a methodology of the Pääbo team that we can't but greatly appreciate.

Age estimates

The study makes two age estimates, one based on a very conservative and truly unbelievable Pan-Homo split date of 6.5 Ma BP and the other based on observed per generation mutation rates, which happens to be perfectly coincident with a Pan-Homo split of 13 Ma BP, the oldest extreme of Langergraber's estimate. This coincidence alone is of enough relevance for all molecular clock approaches, because it effectively demands the doubling of all age estimates based on the ridiculously short 6.5 Ma Pan-Homo split supposition. 

Red outlines are mine. Click to enlarge.

It also produces a semi-reasonable San-West African age estimate of c. 86-130 Ka, although I would think it a bit older in fact or at the very least at the top end. This highlights the severe difficulties of such molecular clock estimates, because a 4 Ma divergence between the alleged introgressing mystery archaic in the Denisovan genome, seems out of the question according on the archaeological and paleontological record, which only documents Homo species since c. 2 Ma ago, half that time (within the estimate but clearly very far from the top end).

Altai Neanderthal inbreeding

An important finding of this study is that the studied individual was extremely inbred, with parents in effective relationship comparable to that of grandparent and grandchild or half siblings. This inbreeding tendency, even if extreme, is not so strange in populations that have experienced founder effect bottlenecks and small population sizes. The Denisovan and the modern human Karitiana people are not so extreme but range in the lower end of double first cousins level of genetic relationship between the parents. Other Native Americans like the Mixe are close to that range, while the other compared populations, Papuans and Sardinians, show much lower levels of inbreeding.

Whatever we may think of Altai Neanderthal inbreeding, their drift parameter is still very low when compared with European Neanderthals. This is not discussed in the paper but such extreme drift also seems to imply extreme inbreeding issues in European Neanderthals, even if these may have other causes such as an extremely strong founder effect or whatever.

Bonobo-specific segments were removed, so the bonobo position is not realistic.

Inferred population history

Both populations leading to the Altai Neanderthal and Denisovans, but not modern humans, appear to have gone through a strong decline in population size since hundreds of millennia ago. The Denisovan decline seems to begin c. 800 Ka ago while the Neanderthal one may have begun c. 500 Ka ago. While this is coincident with a general expansion of the H. sapiens branch (still undifferentiated in Africa), peaking around c. 250 Ka ago before differentiation and relative decline. In their words:

All genomes analysed show evidence of a reduction in population size that occurred sometime before 1.0 million years ago. Subsequently, the population ancestral to present-day humans increased in size,whereas the Altai and Denisovan ancestral populations decreased further in size. It is thus clear that the demographic histories of both archaic populations differ substantially from that of present-day humans.

Neanderthal and Denisovan admixture in modern humans

The new tests confirm in essence the previous findings: there is significant Neanderthal introgression in modern humans descending from the migrants out of Africa and there is also significant Denisovan one among Australasian populations.

Additionally and with some caution, the authors think that much lesser Denisovan introgression (of around 0.2%) is found among East Asians and that these, as well as Native Americans, show slightly more Neanderthal admixture than West Eurasians. In my understanding this may be caused by minor African flow to West Eurasia after the admixture event (and/or residual "First Arabian" persistence) and I would think that measuring South Asians would help to clarify this issue (because African admixture is negligible in the subcontinent but they are also distinct from East Asians).

These measurements are so weak that the authors agree to all kind of cautions about them in any case.

In addition to all this, the supplemental material (section 13) also detects tiny, almost homeopathic, amounts of Neanderthal gene flow to Yorubas (~0.02%), obviously mediated by H. sapiens backflow from Asia and Europe into parts of Africa, which eventually influenced other African populations. An even more diluted amount may also be present among the Mbuti Pygmies.



Altai Neanderthal admixture in Denisovans

This issue is not really explained in the paper as such, and we have to reach out to the Supplemental Information chapter 15 in order to grasp it.

It is clear that the Altai Neanderthals are closer to Denisovans than other Neanderthals are by approx. the following fractions (directly deduced from the raw affinities listed in fig. S6a.2):

• 2% more than Mezhmaiskaya
• 7% more than Vindija (avg.)
• 9% more than El Sidrón

Feldhofer appears closer instead but this sequence was not used by the authors in most tests because it has too dubious quality.

In section 15 of the supplementary material, using complex methodology and lamenting the lack of a second Denisovan sample which would be most useful, they estimate a minimal 0.5% (Altai) Neanderthal introgression in Denisovans, with strong warnings that this could well be quite higher. I don't know why they are not even considering a more direct approach, but I would dare to guesstimate the introgression to be close to 8% from the above raw data, assuming that there are no further complexities at play, such as other Heidelbergensis introgression in European Neanderthals, etc. The drift parameter (see above) does not seem to be one such complexity because Mezhmaiskaya is almost as drifted as Vindija yet it is consistently much closer, as it seems to correspond to its specific relatedness to Altai Neanderthals in mtDNA (and possibly also in nDNA if it is admixture what causes their pseudo-tree positioning closer to the root, what would be typical).

Note in blue is mine.

Mystery archaic genetic flow into Denisovans

The authors find that some 0.5-8% of the Denisovan genome appears to come from another hominin, which split from the human trunk even earlier.

We caution that these analyses make several simplifying assumptions. Despite these limitations, we show that the Denisova genome harbors a component that derives from a population that lived before the separation of Neanderthals, Denisovans and modern humans. This component may be present due to gene flow, or to a more complex population history such as ancient population structure maintaining a larger proportion of ancestral alleles in the ancestors of Denisovans over hundreds of thousands of years.

Later in the discussion section they ponder further the implications of this finding:

The evidence suggestive of gene flow into Denisovans from an unknown hominin is interesting. The estimated age of 0.9 to 4 million years for the population split of this unknown hominin from the modern human lineage is compatible with a model where this unknown hominin contributed its mtDNA to Denisovans since the Denisovan mtDNA diverged from the mtDNA of the other hominins about 0.7–1.3 million years ago⁴¹. The estimated population split time is also compatible with the possibility that this unknown hominin was what is known from the fossil record as Homo erectus. This group started to spread out of Africa around 1.8 million years ago⁴², but Asian and African H. erectus populations may have become finally separated only about one million years ago⁴³. However, further work is necessary to establish if and how this gene flow event occurred.

Going to the detail of the matter (i.e. supplemental material sections 16a and 16b), one of the key details is that present-day Africans share more derived alleles with Neanderthals than with Denisovans. This can only be explained because Denisovans have other archaic ancestry prior to their apparent divergence from Neanderthals or (what is about the same) because Denisovans diverged themselves prior to the Neanderthal-Sapiens split, what is what the mtDNA (unlike the nDNA) suggests. However the difference, even if consistent across comparisons, is too small (a few percentage points) to be attributed to the later scenario.

This means that Denisovans appear to be at nDNA level some sort of an independent branch of proto-Neanderthals with some other but minor archaic admixture. Instead at mtDNA level they appear to be unrelated to Neanderthals and related instead to H. heidelbergensis (a detail not discussed in this paper because it is a too recent independent discovery).

There are still many details to explore but, in principle, it would seem that the Denisovan branch appears to be a divergent proto-Neanderthal one (maybe related to the Hathnora hominin, which looks very much Neanderthal) with lesser other archaic (H. heidelbergensis?) admixture, which nevertheless remained prominent in their mtDNA for whatever accidental reason.

Whether the H. heidelbergensis population of Atapuerca responds to this same profile (i.e. they were Denisovans too) or belongs instead to the "other archaic" population which introgressed in the Denisovan genome remains to be solved. So far we only know the mitochondrial lineage and this one may be misleading, as seems to be the case with the Denisova hominin.

Note in red is mine

Modern human genetic evolution

Benefiting from the high quality of the archaic genomes of Altai, the authors cataloged a long list of simple mutations exclusive to our species: 31,389 single nucleotide substitutions and 4,113 short insertions and deletions (indels). Additionally they found other 105,757 substitutions and 3,900 indels shared by 90% of their modern human sample of 1094 individuals.

They suggest some lines for future research in this regard, maybe focusing on genes known to influence brain development or regions that could show signs of positive selection. These preliminary lines of research are explored in SI-20, noticing potential selection in genes that affect the ventricular zone of the brain and cell proliferation in fetal brain development.