The paper reporte
d in the news (on the BBC website, e.g.) appears in Science's new issue. Make no mistake, it's a good and important piece of work, long promised and finally arrived. It is a sequence of roughly the entire Neandertal genome compared to five available whole-genome sequences from modern humans. Getting and assembling anything close to a whole genome sequence from fragmentary bits in fossils, contaminated with DNA from other things such as bacteria in the earth where the individual fell thousands of years ago, is no easy task and Paabo's group has been one of the global leaders. Studies of ancient DNA are important because they provide direct evidence of the past, so where DNA is preserved it will remain valuable to sequence and interpret it.One thing to note, that seems like double-think, but is not relevant to the points we want to make here, is that this Neandertal whole genome sequence is not the whole genome sequence of a Neandertal. This sequence is a composite assembled from ancient DNA extracted from three different individual Neandertals' remains. But 'the' human genome sequence online at GenBank is also a composite. Some technical issues are affected by this, but they aren't relevant here.
Whatever the details of the assembly, or whether variation among Neandertals was observed, the issue here is the origin of modern human sequences: did any of it descend directly from Neandertals, or were they an entirely separate group (or species, even) that separated from the common human stock and had no subsequent inter-breeding. That is, we today would have no descent directly from the Neandertals. Or was there some inter-group hanky-panky?
The new paper suggests that there was, but there are two major problems with that. The 1-4% are in segments that seem to have a different ancestry from the rest of the Neandertal genome, less divergent from us. The rest diverges from us by about the amount you'd expect given our joint time of separation from our common ancestry with chimpanzees.
The first problem is one we harp on regularly, the playing to the media and exaggeration of the results. In this case, the exaggeration was the definitive way the admixture issue was made melodramatic and definitive. It suggests that interbreeding was something exotic or immoral, like a human mating with a chimp, rather than what at the time would have been considered routine mate choice among individuals from neighboring groups.
They would probably have coexisted together in times when nobody moved very far, and would have differed from each other far less than, say, Africans and Europeans do today, and between whom mating is thankfully no longer a big deal in our society. In fact, the evidence reported is that this interbreeding occurred after both groups were part of the Eurasian population after its expansion out of Africa. In that sense, the groups may have diverged somewhat, and come in contact again later, and became good neighbors for a while. Whatever happened way back then involves our ancestry which is certainly interesting and worth knowing. But when the evidence is tentative so should be the claims.
But there is a second and much more important problem. It is a subtle issue, that in essence is that whether or not any direct human genetic ancestry traces back through Neanderthals basically doesn't matter related to how 'different' we are from them. In round numbers, here's why:
A copy of your genome and a copy of a chimp's (our nearest living relative) differ by about 2 to 5% in terms of DNA sequence. Two copies of the human genome today differ by about 0.1 to 2% depending on the comparison one makes.
We've been separated from our common ancestor by 7-10 million years. Corresponding to that, the paper shows that the Neandertals differ from modern humans by about 7% which is about what you'd expect given that (regardless of admixture issues) the Neandertal split happened only after about 90-95% of the time had passed since we and chimps split.
By that time, basically everybody was human, and in turn that means that overall we are essentially as similar to Neandertals as we are to each other (crudely speaking, we're 95% closer to them than to chimps). And of course the vast majority of sequence differences generally, and hence in this case, will have little if any function. If humans are virtually identical to each other then we are virtually identical to Neanderthals whether there was any inter-mixing or not.
But consider how much functionally meaningful (as opposed to evolutionary clock-meaningful) variation there is in modern humans around the world. Within our single species, there's plenty of room for differences, and they can be important. They can protect you in very important ways from the environment (as skin pigmentation does in the tropics), they can protect you from disease (as immunological differences among us do), and there is a lot of variation in behavioral abilities of all sorts. As many diseases show, even just one single DNA change can be lethal.
The point is that whatever important functional differences or similarities there were between us and Neanderthals need have nothing to do with whether there was any admixture between their populations and populations of our other direct ancestors. Natural selection will purge bad variation, favor sterling advantages, and ignore most of the rest wherever it comes from.
If there is major functional difference between us and our burly cousins, it is to be found in the relevant genes, not in the score card (or dance card) of our sequence differences. And they could have existed in them then, but not us now, even if there was inter-breeding.
This means that Dr Paabo is right to treat this as a story for publicity. Its scientific impact is far less than its human interest value. To portray the inter-mixing question as an important one about human function is to misrepresent (or misunderstand?) how genes and evolution work. But to understand that takes more than a sound byte, and of course that means not many people will be interested.
At the same time, there's nothing wrong with trying to find out, especially from direct genetic data when it's available, what we can about our closest, if dearly departed, ancestors.
9 comments:
A correction and thanks to Holly for pointing it out. We differ from Neandertals by about 7% of the amount by which we differ from chimps. That's also roughly the amount of time in which we've been (mainly) separated from Neandertals, compared to the time we've been separated from our common ancestor with chimps.
did they look at any Homo sapiens DNA that was contemporaneous with the Neanderthal DNA, or did they use only modern human DNA as comparison? I would imagine they used the modern human genome as comparison—however wouldn’t you expect a stronger signal of recent admixture, in two populations that recently experienced gene flow? I know this would not necessarily tell us how much Neanderthal we have in our modern populations, so it would be altogether less ‘sexy’, but interesting nonetheless.
Roughly contemporary DNA from several 'modern' looking fossils in Europe has been sequenced and showed closer affinities with modern DNA. There was some skepticism about whether there had been contamination with real modern humans (e.g., archeologists or lab personnel), and I don't know if there has been any resolution of that.
What I would love to find somewhere is a stellar explanation (for non-geneticists) for how you test these three different hypotheses:
1. We share a common ancestor and then diverged and stopped sharing genes.
2. We share a common ancestor and then went our separate ways and then came back together and experienced gene flow.
3. There was never a divergence (there was always a bit of gene flow) and a subpopulation (a small part of the overall variation) just died out.
These questions are spot-on. We've developed a simulation program to try to answer them, comparing different models to data. Jeff Long at the University of New Mexico is an ace at this kind of analysis and has work soon to be submitted (and presented at the recent AAPA meetings) that finds admixture signatures in modern global genotype data.
Essentially, with some separation and coming together, and then the demise of the Neandertals (if this is what the fossil record actually says happened), recombination over subsequent generations will remove the Neandertal DNA component in chunks, so that today only some of them would remain. Or, even with admixture, they might all have gone away, but apparently didn't.
Continual admixture but with a minor component from the N population in the past in Europe could leave similar 'signatures'. For example, had two populations emigrated to different areas out of Africa, then come back together in some subsequent demographic expansion, but been basically 'races' of the same species (which is likely what they'd have been), we could see what we see.
In the US there is about 15% admixture of European DNA in the African American population. Each person has segments that are identifiable from the two parental sources.
At least, this is the general idea, I think, of what the issues are.
Returning to Holly's #1 query, Neandertals and "modern" humans shared W.Europe for a considerable time a good deal later than the time posited for initial introgression (somewhere in the middle East?), and it's here in Europe late in the game that Trinkaus and colleagues have argued for anatomical evidence of "hybrids". But there is no evidence of E. Asian Neandertals anywhere at this time. So why then doesn't the European genome exhibit more introgression beyond the initial OoA episode(s) and a higher % of shared DNA than do Asians? Doesn't sound like a committed relationship ;)
I don't know the data well, and don't yet know what the details of Jeff Long's findings are. But if the Neandertals were a minority population, what might today be called a 'race', then the probability of survival of segments of their ancestry might be lower than that of the majority population.
Different relative sizes of the groups, if they were separate groups, and cultural isolation between them (e.g., 'tribal' boundaries etc.) could perhaps have resulted in what we see.
So maybe not committed on a population basis, but it doesn't indicate any sort of speciation barrier of the sort that most people (including Svante) traditionally argued. No?
I want to add a comment to our post, which is that the Green et al. paper has looked for, and found a few, signals of natural selection in specific genes. As we noted, if those are true reflections of selection, it will be interesting to see what the explanations might be. But it won't have any particular relevance to the overall amount of sharing.
Thank you, Ken. I counted on you commenting on this.:)
First, I'm in a position where I don't have the time or resources to scrutinize the data and analysis. And I conjecture that if Paabo fesses up to admixture, then it most likely happened. But I will follow up on the literature over the next couple of years to examine any opposing voices while assuming that the analysis is correct, regardless if I would have bet against these results ten years ago or ten days ago.
Second, for those who hold to the biological species concept, this lends more weight to classifying Neandertals as Homo sapiens neanderthalensis instead of H. Neandertalensis, an ongoing debate for more than a couple of decades. But assuming the validity of the evidence of admixture, the H. sapiens neanderthalensis classification isn't a shoo-in. For example, significant mating boundaries may have existed between Neandertals and anatomically modern humans living in the same neighborhood, which doesn't fit everybody's definition of subspecies. And the apparent lack of morphological traits in contemporary Eurasians might suggest that there was no selection for Neandertal morphological traits. And all Neandertal SNP's and indels in the Eurasians may be neutral and fixed by drift, but future research will tell us more about that. And the small amount of Neandertal DNA may suggest some selection against Neandertal-modern human offspring.
Significant morphological differences include bone structure. For example, I know it's not hands down that Neandertals couldn't throw a spear, but this might have been the case. And Neandertal oral speech capability appears limited. And a big difference is the Neandertal enlarged sloping forehead compared the modern human nearly vertical forehead, which might have some correlation to the development of the neocortex. And this might have something to do with the modern human level of inventiveness and communication.
I've read enough about the genetics and morphology of subpopulations (so-called races) in modern humans that differences between modern subpopulations such as the San and Papuan are a drop in the bucket compared to the differences between modern humans and Neandertals.
This fascinating genome research still doesn't say if Neandertals could develop attachable arrowheads and pigment artwork or if they learned to produce attachable arrowheads and pigment artwork from their literal anatomically modern human relatives, who still may be the standout inventors.
I look forward to the future plans of the Neandertal genome project that hope to tell us more about brain development in Neandertals. This and more field study of traditional paleontologists will help to answer these questions.
Ken, per your last comment, I think Green et al. implies that the comparison of the Neandertal genome helps us to see selection in modern humans since the divergence from Neandertals while it's not saying that Neandertal DNA had selective advantage in Eurasians, if I'm reading the article correctly.
Post a Comment