Ancient DNA research has transformed our understanding of human evolution in recent years. In particular, the recovery of genomes of a previously unknown group of hominins—the Denisovans—from scraps of bone and teeth in a Siberian cave has shown us how many mysteries about human evolution are still unsolved. We didn’t have the faintest idea that Denisovans existed until we stumbled upon their DNA. How many other archaic hominins are out there to be discovered?
The answer to this question depends on not only obtaining more ancient genomes, but also on having robust computational tools for modeling ancient population histories from those genomes. These tools can identify which genetic variants come from having sex with which ancient hominins (otherwise known as ‘admixture’), give us estimations of past population sizes, as well as an estimate of when two groups last stopped swapping genes. Knowing these details is critical to understanding our histories, but their accuracy is entirely dependent upon how good our models are. The wrong assumptions can seriously bias the results.
Last week a team of researchers at the 2016 American Society of Human Genetics meeting described a new tool for modeling archaic admixture. Their approach differed from previous approaches by allowing multiple archaic hominins to be sources of admixture simultaneously in each population. In other words, it let researchers model the scenario of our ancestors having sex with Neandertals and Denisovans, rather than just one or the other. (This seems plausible, knowing humans).
This new approach yielded some pretty interesting results when applied to human genomes, including:
- A more recent date for the divergence of archaic hominins and H. sapiens, at about 440,000 years before present.
- Variation in admixture between archaic hominins (related to Neandertals and Denisovans) within Africa.
- Bias in estimates of archaic ancestry proportions for non-Africans, especially in Melenesia, which had lower levels of Deniosovan ancestry (1.11%) than previously estimated (3-6%). This supports the finding of previous researchers (Sankararamen et al. 2016), but many authors are still using a ~5% estimate of Denisovan admixture in Melanesian. It’s perhaps important to revisit this question.
One explanation for why present-day Melanesian might look different than expected could be due to admixture from a population not included in their models (i.e. not Denisovan or Neandertal). But that’s not the only explanation. Instead, what appear to be genes from a new hominin might be from genetically distinct populations of Denisovans. We simply can’t know until more research is done.
Unfortunately, media reports about the conference have veered wildly from what the researchers actually discussed. Reporters have focused exclusively on the potential presence of ancestry from previously unknown hominin in Melanesian—which co-author Alan Rogers said is a completely wrong interpretation of the talk. “This was sheer speculation. We have no evidence that it is so. Nonetheless, it has become the headline.”
The fact that this story has gone viral is a testament to how interested the public is in our genetic histories. One the one hand, I find this fascination with human evolution tremendously encouraging. (I live in Kansas, after all.) On the other, it is incredibly frustrating to see this sort of nonsense distort the work of careful scientists. There are marvelous things to learn about ourselves from our genomes, but this kind of uncritical storytelling is how genetic myths–not histories–are spread.