Header image

john hawks weblog

paleoanthropology, genetics and evolution

Photo Credit: Excavation at Denisova Cave, Russia. John Hawks CC-BY-NC-ND

The Altai Neandertal

It is notable that we now have evidence for interbreeding among every kind of hominin we have DNA from, and some we don’t.

Neandertals and humans. Denisovans and humans. Neandertals and Denisovans. Some living sub-Saharan Africans and one or more unknown ancient populations. Denisovans and one or more unknown, even more ancient populations. They were all mixing.

The picture of Pleistocene human evolution has come rapidly into focus during the last two years. Before the last 30,000 years, the world was full of human populations that were around twice as different from each other as the most diverse recent human groups. Some of these ancient groups grew at the expense of others, but the “losers” over the long term still survived within the genomes of the “winners”.

The process of selection on human genes spanned these different human populations, as genes of adaptive value were exchanged between them, long surviving their progenitor populations. Each of the groups shared genetic variation from their common ancestors, but some local populations were markedly restricted in variation by inbreeding. All in all, humans of the past had a population structure rather like today’s chimpanzees, although ancient humans were slightly more alike across a much larger geographic range.

Last spring we heard that the Max Planck Institute for Evolutionary Anthropology had successfully sequenced a high-coverage genome from a hominin toe bone found at Denisova Cave, Russia. In the announcement they made clear that this genome was substantially different from the existing high-coverage genome from Denisova. That first genome, from the bone of a pinky finger, represented a previously-unknown human population, quite different from Neandertals. This new genome, from the Denisova toe bone, is much more similar to the genomes of other Neandertals already known from Vindija, El Sidrón, Feldhofer and Mezmaiskaya. That seems like enough to call the toe a Neandertal.

Yet this week’s paper makes clear that this genome is not just another Neandertal. Kay Prüfer and colleagues (2013) describe several kinds of analyses on the high-coverage data. The most important of these analyses establish the pattern of similarities and differences between this genome and others, allowing us to test some hypotheses about the relationships of Neandertals, Denisovans and modern humans.

At the moment, I am just going to present and explain a few of the major conclusions of the study. The paper itself is only 7 pages long, but the supplementary data stretch across 248 dense pages of text and figures. It’s much more than a dissertation’s worth of information, and it is going to take some time for me to completely digest. There will be much more to discuss over the next few weeks. Further papers that use these data are in the pipeline, with some interesting additional results. This is good work and I am excited by it, but I am going to present some notes of caution as well. I think some interpretations are likely to shift as we learn more.

Here are the major insights of the present study:

  • The new genome appears to represent an individual that has fewer new derived mutations than the Denisovan high-coverage genome. The research suggests this as a means of “molecular dating” of the specimens, proposing that the Denisovans lived in Denisova cave after this Neandertal population.

  • The Denisovan high-coverage genome includes portions that reflect ancestry from Neandertals.

  • The new genome groups with previously known Neandertals in a genome-wide cluster analysis, but represents a more divergent population of Neandertals than those yet described. Under a model where genetic differences reflect a branching population history, the “Altai Neandertal” population seems to have diverged from other Neandertals sometime between 77,000 and 114,000 years ago.

  • The high-coverage Neandertal genome shares many derived mutations with sub-Saharan Africans, while the high-coverage Denisova genome shares fewer. If these archaic populations were equally related to Africans, they would have the same number of shared derived mutations with Africans. Prüfer and colleagues infer that the Denisovan genome had ancestors who belonged to a yet more ancient hominin population. They suggest this population represents around 4 percent of the ancestry of Denisovans, and that it diverged from the common ancestors of Neandertals and sub-Saharan Africans sometime around a million years ago. The confidence intervals on both estimates are large.

  • The new genome has many extended runs of homozygosity, consistent with inbreeding. The study concludes that the parents of this individual were likely 1/4 degree relatives – such as uncle/niece or half-sibling mating.

  • A comparison of the archaic human genomes with the 1000 Genomes Project samples shows only 96 amino-acid-coding changes shared by nearly all of the 1094 recent humans but absent from Denisovan and Neandertal genomes. A larger number (over 3000) of mutations that “possibly affect gene regulation” are also near fixed in recent humans. These are potentially interesting because they may be related to recent behavioral or anatomical evolution of modern humans.

  • The paper reports on new sequencing of the Mezmaiskaya Neandertal to 0.5x coverage. This genome is substantially closer to recent humans than are the other Neandertal genomes. Presumably the population of Neandertals that accounts for present-day Neandertal genes in living people was closer to the Mezmaiskaya Neandertal than others.

  • The high-coverage Neandertal and Denisova sequences allow a new estimate of the amount of Neandertal and Denisovan ancestry in human populations. Neandertal ancestry of living non-Africans is now estimated between 1.5 and 2.1 percent. This is lower than previous estimates, a discrepancy that the paper does not explain.

  • The paper finds significant evidence for Denisovan ancestry of mainland Asian and Native American populations. The Denisovan fraction in these populations is small, only around two tenths of a percent on average, but the ancestry is spread throughout these populations into the New World.

  • The Denisovan ancestry of living populations of New Guinea represents a substantially different genetic background than the Denisova high-coverage genome. The divergence between the Siberian Denisovan high-coverage genome and the Denisovan intermixture with humans is greater than the divergence between any living groups of humans with each other.

We can now see that the original description of the Denisovan genome in 2010 and follow-up analyses in 2011 were based on a number of inaccurate assumptions. The current high-coverage data have added a lot of precision to some analyses, but several of the changes in this new research have actually come from the adoption of new assumptions and more refined models.

Some of the conclusions in this paper will not last long as more ancient genomes are sequenced. We have recently seen with the publication of the Sima de los Huesos mtDNA that many assumptions about the Denisova population are questionable (“The Denisova-Sima de los Huesos connection”).

Some examples:

  • Why should we assume that the Denisovan ancestry includes only a single “mystery population”? The Sima de los Huesos result shows that several populations may have been in a position to mix with the ancestors of Denisovans.

  • Why should we assume that the Denisovans were a single population? The genetic differences among “Denisovan” groups by our current definition were greater than those between any two human groups today.

  • This current paper is noncommittal about the rate of mutations that should be applied to the ancient genomes, which leads to an uncertainty of more than a factor of two in the date estimates presented. This is unfortunate because the uncertainty prevents the DNA from shedding light on the relationships of pre-Neandertal, Neandertal and modern human fossil remains. But the uncertainty is real, as the relevant mutation rates remain a matter of debate (“A longer timescale for human evolution”, “What is the human mutation rate?”).

At any rate, the new genome has tremendous value for the further study of how we evolved. As I continue to study the supplements of the paper, I will be updating on several areas of interest.


Prüfer, K. et al. (2013). The complete genome sequence of a Neanderthal from the Altai Mountains. Nature (in press) doi:10.1038/nature12886

Reich, D., Green, R. E., Kircher, M., Krause, J., Patterson, N., Durand, E. Y., ... & Pääbo, S. (2010). Genetic history of an archaic hominin group from Denisova Cave in Siberia. Nature, 468(7327), 1053-1060.

Reich, D., Patterson, N., Kircher, M., Delfin, F., Nandineni, M. R., Pugach, I., ... & Stoneking, M. (2011). Denisova admixture and the first modern human dispersals into Southeast Asia and Oceania. The American Journal of Human Genetics, 89(4), 516-528.