In my last post I drilled down on just a few of the results in the paper The genetic history of Ice Age Europe (ungated). There are many results which I didn’t really explore, in particular, the finding that there seems to be a gradual decline in Neanderthal ancestry within European populations over time. That’s for a follow up.
In any case, it’s an interesting time to be alive and be interested in these topics.
The epoch we are situated is between an age of ignorance, and one in which we will be overwhelmed by interpretations based on a surfeit of data. The whole genome of the Neanderthal was published in 2010. Today we have many more whole genomes, and probably on the order of 1,000 ancient genomes of varying quality in the pipeline (i.e., some of it is unpublished). Reconstructing the history of humanity from genetic data has transformed from inference from the tips of the phylogenetic tree, to the examining of points deep in the nodes.
This reminds me of an argument that was highlighted in The Monkey’s Voyage between cladists with a background in systematics and paleontologists. The way paleontologists understood evolutionary relationships was to examine the fossil record, and reconstruct trees with putative ancestral forms and their descendants. The cladists asserted that this method relied upon the incomplete and unreliable fossil record, and so was not nearly as powerful as simply looking at extant variation in a more rigorous manner. Though the added rigor of the cladists arguably transformed the field of phylogenetics, as I have suggested before the extremism of the cladists in dismissing whole domains of knowledge and alternative methods has not swept the field.
Personally, I think that is a good thing. But, some of the warnings of the cladists probably need to be considered when taking into account the new results from paleogenetics. The reality is that in many ways there is little difference in terms of the raw data which paleogenetics and paleontology based on fossils provides. For various technical reasons phylogenetic inference from whole genomes of DNA sequence can be much more powerful than analyzing, for example, the teeth of an ancient hominin. Those teeth would give you phylogenetic and functional information. But reconstruction is more robust when you have tens of thousands (or perhaps millions) of variations, which is what DNA gives you. Second, those markers can tell you a great deal more about a variety of functions than simply teeth (I am not denigrating teeth here, as they are very informative!).
Extended data table 5
One thing which is more and more clear as more data comes in this that the genetic architecture of pigmentation in modern Europeans is a product of the Holocene, and perhaps even the last 4,000 years. A more sensational way to state this is that the Nordic phenotype may not have been present in appreciable amounts in any population when the pyramids of Giza were being constructed! Of course, there is a major caveat here in that we know that light skin emerges with different genetic architectures, so ancient Pleistocene Europeans may simply have had a different arrangement of functional SNPs. The main caution on this caveat is that pigmentation is a trait that is very well characterized across mammals as a whole, so prediction is much less dodgy than in other traits. If we eventually get enough high quality genomes from Gravettian period Europeans, and they lack derived SNPs across ten major pigmentation genes, then we can be pretty certain that they were in the ancestral state.
Researchers are then literally putting flesh on ancient bones. And yet we still see what we see. Paleogenetics suffers from the same issue as paleontology: skewed sampling. Especially when sample sizes for certain periods and regions are small, our illumination can’t give us a sense of what we don’t know. The genetic history of Ice Age Europe gives us a picture of genetic turnover, but one in which the Goyet sample representative of early Aurignacians turns out be the ancestor of a population which pops back up into prehistory (the Magdalenians) after a long 10,000 year Gravettian interlude. Unless they traveled through a wormhole, it seems clear that this lacunae is a consequence of patchy and biased sampling. As the number of DNA samples increases we’ll get a better sense of how patchy and biased the methods turn out to be, but we’ll never totally abolish this problem. I believe that the same fact explains why many papers see a resurgence of Mesolithic hunter-gatherer ancestry among Neolithic farmers; the former were always there, but they are not being sampled across much of the temporal transect due to spatial patchiness.
As we traverse this period between ignorance and the potential for knowledge, we can start forming conjectures as to the shape of the future. In the comments below Andrew Oh-Wilke suggests that ancient people traveled much further than we might have guessed. I think this is right. In The Monkey’s Voyage the author suggests that in biogeography there has been a move away from vicariance to a somewhat stochastic long distance dispersal model to explain variation. The vicariance model emphasized the emergence of geographical barriers due to geological processes, and the subsequent divergence between two populations due to reduced gene flow. The idea behind stochastic long distance dispersal is basically that a lot of the patterns are due to random freak events, such as a small group of Old World monkeys somehow making it across the Atlantic from Africa, and becoming the ancestors of the whole family of New World monkeys.
The vicariance model has less relevance for human prehistory, because in most cases we’re not talking about geological time scales. There are exceptions, after a fashion. Berengia and Sahul both harbored populations, and after the sea levels rose groups were isolated on opposite sites of the water barrier. But there was always some contact even after this, because humans can traverse water barriers. There is an analog to the vicariance model in historical population genetics, and that is the isolation-by-distance model of human genetic variation and diversity. The major example is in the 2005 paper Support from the relationship of genetic and geographic distance in human populations for a serial founder effect originating in Africa. This model’s logic is sound. One imagines that humans start at a point in space, and expand outward through a demographic wave of advance, as groups disperse into territory inhabited by archaic hominins, or not inhabited by hominins at all (e.g., the Americas and Oceania). Because this results in a serial founder effect, you see a pattern where populations further away from Africa exhibit less genetic diversity. Additionally, genetic structure in humans can be conceived as as dominated by geographic distance and exhibiting clinal variation.
What the paper Toward a new history and geography of human genes informed by ancient DNA did was show that the genetic data used to support the isolation-by-distance model of decay of genetic diversity did not have the power to truly show this was the correct model. What David Reich and Eske Willerslev (and others) have shown with ancient DNA (as well as novel methods in Reich’s case) is that 1) population turnover has been relatively common 2) most (all?) modern populations are best thought of as admixtures between ancient lineages, in many cases pulse admixtures that occurred rapidly. Like the vicariance model the isolation-by-distance model was boring and general. It was easy to model, and didn’t engage in special pleadings to historical contingency. In other words, it’s a perfect model to use as null hypothesis. But that doesn’t mean that it’s correct, or, more accurately, captures most of the dynamics.
An example may suffice. Europe is the most well elucidated of the major regions of the world in terms of prehistory. The “standard model”, utilizing simple and generic population genetic demographic processes produces a nice and simple model to fit the data. ~50,000 years ago humans leave Africa, they settle the Middle East/Central Asia. ~40,000 years ago they arrive in Europe. ~10,000 years ago farmers arrive from the Middle East, and expand into Europe from the southeast, with their genetic signal diluting over time to the northwest.
Here is the model, sketchily, informed by ancient DNA. ~50,000 years ago humans leave Africa, and mix with a number of Neanderthals. ~40,000 years ago, they arrive in Europe. ~35-40,000 years ago the first modern Europeans are replaced by another population. This second population is culturally similar to the first, and contributes some (though small proportionally) ancestry to modern Europeans. It is replaced by another population, which does not contribute much to modern Europeans (Gravettians), though populations related to it do. It is replaced by a population related to the first Europeans with descendants (Magdalenians, who are descended in part from Aurignacians, and do not share much drift with Gravettians). Then, the Magdalenians are replaced by Villabruna populations, the very late Paleolithic populations at the tail end of the Ice Age. The Villabruna have mixture from both the Near East, and to a lesser extent East Asia. Or, Villabruna populations were intrusive to the Near East, and possibly East Asia, or there were mediating populations between. It is all somewhat unclear. Then the Villabruna populations, which become Mesolithic hunter-gatherers, are overwhelmed by Near Eastern groups, which have very exotic ancestry unrelated to all other non-Africans (Basal Eurasian). Finally, the Neolithic groups are overwhelmed by populations from the steppe, who are themselves compounds of very distinct elements.
This is a difficult and historically contingent story. It is not neat, tidy, and is a dog of a model. It is not easy to generalize. But, it is probably a model which captures many more of the salient dynamics than the earlier one.
Going forward what generalizations can we take from this? Europe has been well elucidated for historically contingent and biogeographic reasons. But the rest of the world will come into the light of understanding in a similar fashion over the next ten years. One prediction I will make is that inter-group barriers were more powerful earlier in the human past than today, at least in terms of how they were relevant genetically. The emergence of meta-ethnic religions and fictive kinship may have paved the way for gene flow on a massive scale over the past 4,000 years. Additionally, human population density is such that the landscape of habitation is less patchy, and conventional continuous gene flow between adjacent populations is just more feasible. In prehistory human groups thin on the ground may have had organize proactively to exchange mates, perhaps during gatherings which were culturally focused. This might imply that mate exchange was less a function of proximity than cultural affinity.
A pattern of turnovers that we see in Pleistocene Europeans aligns with the idea that socio-cultural boundaries were major fault-lines which were inimical to gene flow. Admixture between two groups in the recent past can occur when one collapses culturally, as occurred in the New World. But it also occurs as a matter of course through proximity, as is the case with the Hui in China. The balance of forces in the hunter-gatherer world may have been toward the former. Patching sampling means we don’t know where the pre-Magdalenian and post-Aurignacian peoples were persisting over the 10,000 years of Gravettian domination, but they were there, biding their time. Any modern understanding of 10,000 years would expect us to lead to massive mixing and gene flow, but that did not seem to occur (some did, but look at the admixture graphs and the Magdalenians are >50% Aurignacian, while the Gravettians are ~0%).
Second, the turnovers probably were partly due to ecological forces. At this stage in history humans were animals whose existence was strongly conditioned on natural vicissitudes. Small numbers of people may easily have gone extinct because of diminished opportunities, and drifted below sustainable levels. Particularly if they weren’t part of a broader network of redundant support, which seems unlikely to have been the case. Agricultural populations still retain a reservoir numerically even after famine. Hunter-gatherers may not have.
Finally, Europe may be a special cases because it is on the frontier of habitation during a phase of glaciation, but it is unlikely to be totally sui generis. The branches of the human phylogenetic tree see to be pruned rather regularly. The genetic history of other parts of the world are likely to exhibit the same pattern of turnover, and relatively recent roots for the demographically dominant group.