Did Africans join in on archaic interbreeding?

The evidence for interbreeding between modern humans and archaic variants has involved a bit of asymmetry. Humans met the Neanderthals and Denisovans only after they left Africa, and so the DNA from these archaic humans can be identified by comparing European and Asian populations with those whose ancestors never left Africa. But that leaves the converse question—whether Africans interbred with some of the archaic populations that were presumably present there—difficult to answer. But a study released by PNAS argues that the sort of interbreeding we've seen elsewhere did, in fact, take place in Africa as well.

If anything, we might expect interbreeding to be more likely in Africa; the authors of the new article note that "the fossil record indicates that a variety of transitional forms with a mosaic of archaic and modern features lived over an extensive geographic area from Morocco to South Africa between 200 and 35 kya [thousands of years ago]." However, there's none of the sort of evidence that made the case for interbreeding with Neanderthals and Denisovans an inescapable conclusion: ancient DNA. Many areas of the continent aren't congenial to DNA preservation, meaning that we might never get that sort of evidence.

In the absence of this sort of conclusive evidence, the authors have made a statistical argument. DNA that has come from a population that's otherwise reproductively isolated (the technical term for this process is "introgression") should have certain properties. In terms of DNA sequence changes, it should appear to be much more distant from other human variants than those variants are to each other. And it should only be present in a limited subset of human populations, the descendants of the one where the interbreeding took place.

This evidence isn't conclusive; these sorts of patterns are unlikely to occur by chance, but it's still possible. And rare variants may end up being common in a population due to positive selection for a favorable mutation. Still, multiple areas that fit this description can provide an indication of archaic interbreeding.

(These sorts of statistical arguments based on DNA differences in populations were being made immediately prior to the publication of the Neanderthal genome, which ended all debate.)

The authors did their analysis using two of the oldest and most diverse populations in Africa (hunter-gatherers) and an agricultural group from West Africa. They then examined the genetic diversity within these populations to estimate its population dynamics. These showed a large recent expansion of the human population, consistent with other studies. But that wasn't all they showed. "The estimate of a divergence time that predates the origin of modern humans based on fossil data (450 kya, Biaka-Mandenka comparison) was unexpected." In other words, it looked like the two populations had split before modern humans existed as a distinct group. The probability of this being explained by chance is only two percent.

The authors built a model for a population in which there had been a relatively ancient separation, followed by a more recent case of introgression. They found that it fit the data from the two hunter-gatherer groups very well, and suggested a pattern similar to that seen with Neanderthal DNA: a source that had separated from the modern lineage about 700,000 years ago, followed by inbreeding about 35,000 years ago. They even identify a single area of chromosome 4 as likely to be the product of the introgression.

All of it is very suggestive, but there are a few aspects of the work that make it very tentative. A lot of the statistical tests of significance are pretty borderline, enough so that you'd expect them to occur by chance if you did a reasonable number of similar studies. And the amount of the genome that the authors' model assumes came from archaic sources—one percent—and it's a fairly borderline effect to start with. The DNA evidence is a bit more solid, statistically, but also shows an anomalous pattern of inheritance across the region of interest.

So the results don't seem like a slam-dunk. It would make sense that the sort of interbreeding that went on outside of Africa would also occur there, and the evidence in the paper is certainly a good fit for that occurrence. But a study that involved more populations or used whole genomes would probably provide a more compelling case.

Or there's always the chance that someone will manage to pull DNA out of one of the skeletons the authors mention.