The Denisova-Sima de los Huesos connection

7 minute read

I direct your attention to a new paper by Mattias Meyer and colleagues describing a mitochondrial DNA sequence from Sima de los Huesos, Spain (Meyer et al. 2013). It is super awesomely cool work, and I can’t wait for the further development as they attempt to get more DNA sequence data from the Sima sample. The recovery of cave bear DNA earlier this year from Sima presaged the current paper, and it seems we are now in a time where we can expect more results from Middle Pleistocene human remains. Very, very good.

Still, there seems to be a widespread confusion about the current result, which shows the Sima mtDNA sequence to be on the same clade as the mtDNA sequences from Denisova, Russia.

I mean, take a look at the quotes from these news articles:

In the New York Times story by Carl Zimmer (“At 400,000 Years, Oldest Human DNA Yet Found Raises New Mysteries”):

Dr. Meyer is hopeful that he and his colleagues will be able to get more DNA from the Spanish fossil, as well as other fossils from the site, to help solve the puzzle they have now stumbled across. Its extremely hard to make sense of, Dr. Meyer said. We still are a bit lost here.

From Ewen Callaway’s Nature news article (“Hominin DNA baffles experts”):

Even Pääbo admits that he was befuddled by his teams latest discovery. My hope is, of course, eventually we will not bring turmoil but clarity to this world, he says.

I sort of understand the confusion.

For more than a hundred years, scientists have been drawing straight lines connecting different fossils, to try to understand the human family tree. Those straight lines always diverged over time, leading toward increasing specialization and extinction of fossil groups. And for more than twenty-five years, geneticists have been assuming that the lines connecting the genealogy of mtDNA should be the same as the lines connecting the fossils. When those lines were different, geneticists have been happy to toss the fossils out of the human family tree, content to accept the story that the fossil people had become too specialized, too peripheral to be ancestors of today’s people.

But the last five years have made clear that both groups – the fossil scientists drawing straight lines of diverging fossil populations, and the geneticists drawing straight lines of diverging – were wrong.

Just look at the evidence. Humans today descend in part from Neandertals, even though Neandertal mtDNA is gone. Europeans today are largely different from the Europeans of 10,000 years ago, with a massive mtDNA replacement along with the introduction of Neolithic culture, and at least a second later large-scale replacement of genetic diversity. Earlier Neandertals in Europe have different mtDNA diversity than later Neandertals in Europe. Denisova cave was home to an earlier population of hominins with different mtDNA than the later Neandertals who lived there. Mitochondrial DNA has never been a straight line linking earlier and later populations within a single location. Whenever we look at ancient DNA in hominins, the earlier populations have different mtDNA diversity than the later ones. Moreover, wherever we have ancient mtDNA from other species – bison, mammoths, cave bears, and others – we find that later mtDNA sequences do not represent the earlier diversity. The Sima cave bear mtDNA is a direct example of this, but the same phenomenon has happened again and again.

The fossil evidence, we now know, is no different. Paleoanthropologists have widely assumed that the Sima de los Huesos hominins are ancestors of Neandertals. That’s a straight line.

There are essentially two reasons for this assumption. One is that Neandertals need ancestors, and the Sima sample seems to be in the right place at the right time—300,000 years ago or more, in western Europe.

The other reason is a bit more substantial: the Sima sample exhibits a number of features that are shared with Neandertals but not African fossil humans, and are rare in recent humans. So the sample is not only at the right place and the right time, it sort of looks the part of incipient Neandertals. Jean-Jacques Hublin and others have described this idea as an “accretion” of Neandertal features in European populations over time. Go back far enough in Europe—say, to the Gran Dolina sample—and you don’t see fossils with Neandertal features. As you proceed forward through the Middle Pleistocene, you start seeing more similarity to Neandertals. Scientists fitted this data to a straight line, projecting a gradual divergence of the European population away from other human populations, eventually becoming Neandertals.

However, over the last few years, neither of these straight-line reasons has been looking especially good. First, the mtDNA landscape of Neandertals has shifted our knowledge of their population dynamics. Dalen and colleagues (2012) showed that later Neandertals do not have the same diversity as earlier Neandertals in western Europe, and that central Asian Neandertals have more diversity than European ones. From this perspective, the evolution of Neandertals looks less and less like a European phenomenon. Instead, Europe may have been invaded repeatedly by Neandertal populations that were much more numerous elsewhere, such as western or central Asia. I developed that idea last year (Hawks 2012), but in fact it is an old idea going back to the 1950s or earlier.

Now that we know that the last 100,000 years of Neandertal evolution was complex and not centered in western Europe, I don’t see why we should assume a straight line between Sima de los Huesos at more than 300,000 years ago and later Neandertals.

Second, the Denisova discoveries have made it clear that other populations existed outside the current visibility of our fossil and archaeological evidence. Why should we assume that these populations looked different from Neandertals? The reality is that we know essentially nothing about the morphology of West or Central Asian hominins of 300,000 years ago. South Asia and Southeast Asia were likewise inhabited throughout this period but we have only the barest hints about the morphology of their inhabitants. These peoples existed just inside the range of archaeological visibility but we lack any but the most rudimentary fossil evidence of them.

To be sure, many people have been assuming that the Denisovans were some kind of East Asian population, for example in China or Southeast Asia. In the process, they have projected the characteristics of the Asian fossil record upon them. That idea has been supported by the existence of Neandertals to the west, and also the sharing of some Denisovan similarity in the genomes of living Australians and Melanesians.

But that’s a big assumption. Let’s explore an alternative: that the Denisovans we know are in part descendants of an earlier stratum of the western Eurasian population. Although they are on the same mtDNA clade, the difference between Sima and Denisova sequences is about as large as the difference between Neandertal and living human sequences. It would not be fair to say that Denisova and Sima represent a single population, any more than that Neandertals and living people do. But they could share a heritage within the Middle Pleistocene of western Eurasia, deriving their mtDNA from this earlier population.

We know that the Denisovan nuclear genome is much closer to Neandertals than the Denisovan mtDNA. We are still waiting for the long-rumored publication of the idea that Denisovan genomes have a “mystery hominin” element in their ancestry. They could be a mixture of any number of earlier populations. None of these have to be East Asian, and as yet we have no suggestion that this “earlier” element of Denisovan ancestry could be as ancient as the first known habitation of Eurasia, as much as 1.8 million years ago. Maybe the Sima hominins represent this “mystery hominin” population.

Maybe the Denisovans were west Asian Neandertals. It does seem like known genetics of Neandertals may represent something like an earlier iteration of the origin of modern humans – more African than earlier hominins like the Sima sample, less influenced by Eurasian mixture than the Denisova genome, only a subset of the diversity of surrounding contemporaries. But we have no idea what the Neandertals of the Levant or southwest Asia may have been like genetically – maybe they were more like Denisovans. This is all basically speculation, which indicates how little we still understand about the dynamics of these populations.

They were complicated. Their relationships cannot be described by drawing straight lines between fossil samples. There were multiple lines of influence among them, small degrees of mixture and large-scale migrations. Europe was far from a slowly evolving population “accreting” Neandertal features over time. The Neandertals we know did not lumber into their doom; they expanded rapidly, multiple times, from non-European origins. They were as dynamic as the Middle Stone Age Africans who would later mix with them and expand across the world.

So I don’t find the Sima mtDNA to be the least bit surprising. It’s refreshing!


Dabney, J. et al. Complete mitochondrial genome sequence of a Middle Pleistocene cave bear reconstructed from ultrashort DNA fragments. Proc. Natl Acad. Sci. USA 110, 1575815763 (2013)

Meyer, M. et al. A mitochondrial genome sequence of a hominin from Sima de los Huesos. Nature (2013) doi:10.1038/nature12788.

Hawks, John (2012) Dynamics of genetic and morphological variability within Neandertals. Journal of Anthropological Sciences 90:81-87.