It News: Mating with Neanderthals is off-again, on-again

Mating with Neanderthals is off-again, on-again

By JOHN TIMMER
2012/8/14

One of the odd aspects of scientific peer review is that, because it's handled anonymously, you sometimes have no idea someone else is working on the same problem, preparing a paper at the same time. That seems to have been the case with a couple of papers that have appeared online over the last couple of days on the subject of what our ancestors may or may not have done with Neanderthals.

Today, a paper was released by PNAS that indicates the evidence produced in favor of interbreeding with Neanderthals is perfectly consistent with a structured population within Africa that would mean our ancestors never mated with them. Perhaps knowing this paper was in the works, however, the group that brought us the Neanderthal genome released a draft of their own work, scheduled for publication in PLoS Genetics, that argues strongly for interbreeding.

And if that wasn't enough, some of the authors of that work went and revised all the dates for our ancestors' splits with gorillas and chimps.

Off again

Neanderthals first. The initial finding that suggested interbreeding was that Europeans and Asians share a small set of features with the Neanderthal genome that are absent in present-day Africans. This could easily be explained by the first modern humans mating with the Neanderthals they found in the Middle East after exiting Africa and before they spread through Europe and Asia.

But, as the authors of the PNAS paper point out, this model treats all of Africa as a single, giant pool of genes. We know that most populations that cover a large piece of geography aren't like that. Instead, they contain what's technically termed "structure"—different populations may be more or less geographically isolated, and the ones at the extreme ends of the range won't directly exchange genetic variants very often. So, in the case of Africa, you might expect that a population in Ethiopia would have some variants that don't show up at all in a South African population, simply because they're relatively isolated.

The paper raises the possibility, recognized in the original Neanderthal genome work, that the pre-modern African population was structured. So, the group that gave rise to Neanderthals could have a genetic signature that was rare elsewhere in Africa. And, if that same population gave rise to the modern humans that left Africa, it could leave them with that same genetic signature. Thus, Neanderthals and non-Africans would end up looking more similar than we'd otherwise expect.

The authors build a model that incorporates a structure into African populations, with different pools arranged north-to-south, and each pool only allowed to breed with its nearest neighbors. They then modeled the Neanderthals splitting off, and a later population of modern humans migrating out of Africa. Their model produced some results that were completely consistent with the Neanderthal genome data, leading them to caution that we shouldn't be too hasty about concluding there was interbreeding.

On again

With that, we can turn to the preprint that was placed on the arXiv on Friday. In it, some of the people involved in the Neanderthal genome work try to clarify when the shared human/Neanderthal sequences appeared in the human genome. To do that, they look at the shared genetic variants in more detail.

When two genetic variants are close together on a chromosome, they'll tend to be inherited together. Thus, if you got a small chunk of (for example) chromosome five from Neanderthals, all the variants in that section would be the Neanderthal type. Over many generations, however, recombination would start to separate them, leaving the area a mix of Neanderthal and human pieces. Thus, you can look at how mixed up a given genetic area is, and make an estimate of how much time has passed since it was uniform.

In practical terms, you can simply look at variants that have Neanderthal versions, and see how often their nearest neighbors are a Neanderthal version as well. The more often that occurs, the more recent the split must be. The draft paper performed that analysis, and came up with the introduction of Neanderthal DNA occurring between 37,000 and 86,000 years ago. That's after modern humans left Africa, and much, much more recent than the human-Neanderthal split. So, it's clear that the DNA got there through interbreeding, rather than a structured population in Africa.

And older

In the same edition of PNAS, there's another paper that shares a key author: Svante Pääbo, one of the key players in the Neanderthal genome work. This paper speaks to a key date: the split between Neanderthals and modern humans. But the work that was done also speaks to the split between humans and chimps, and our common ancestor with gorillas.

These dates are typically generated by a mix of DNA and fossil evidence. So, on the DNA level, we can determine the relative differences among humans, Neanderthals, chimps, gorillas, and a distant relative like macaques. And, if we have a good fossil date for one of these splits, we can turn these relative differences into actual ages. The fossil date anchors the tree for us, and allows us to turn DNA differences into ages.

The problem is that both the dates of some fossils and their relative relationships are often the subject of some argument (as ably demonstrated by last week's science news). That means we can rarely assign an absolute date to a given split based on fossils, only a range of possible ages. And that range adds to the uncertainty about dates all through the tree.

But the authors noticed that we now have data that wasn't available when molecular clocks were first developed. By sequencing trios of two parents and one of their children, researchers have calculated the number of mutations that occur within a single generation of humans. If we assume that other large primates have similar mutation rates, we can just figure out what their generation time is, and work backwards from there to when the different lineages last shared a common ancestor.

So, the authors obtained new estimates for the average generation time of chimps and gorillas. These included the ages of first and last fertility, as well as the relative fecundity of different ages, so the numbers are fairly exact.

Based on these numbers, most of the molecular clocks that are based on fossil data are too young. Rather than 4-6 million years ago, the human-chimp split is pushed back to over 7 million years, and possibly as late as 13 million. Neanderthals split off at least 400,000 years ago, making interbreeding even more likely, based on the results described just above. Our break with gorillas goes up from 7 million years to between eight and 19.

The dates may be derived without fossils, but they may make the fossil hunters very happy. A couple of species that were suggested to be distinctly on the human lineage date back to around 6 million years ago. This places them a bit awkwardly, given that the molecular clock dates had suggested there was no distinct human lineage at that point.

That said, the new dates shouldn't be viewed as definitive, either. We don't yet know whether the chimp mutation rate is that same as that of humans, or whether average generation times have changed over the course of evolution. And there's always the chance that someone else has had a good idea that will shed light on the topic, but it hasn't cleared peer review yet.