Ghost populations and ancient DNA

The rapidly changing field of ancient DNA has settled into a kind of normal science, as several teams of researchers have coalesced around a set of approaches to discover the genetic relationships among ancient peoples. Ewen Callaway this week in Nature profiles some of the key investigators and their recent work: "Human evolution: The Neanderthal in the family".

The headline is driven by Neandertals and the successful sequencing of even more ancient DNA from Sima de los Huesos. But the quest for the most ancient DNA is maybe the less interesting of the two developments discussed in the article. The other is the theoretical paradigm that attempts to break down the genomes of living and ancient people into parts that come from different original populations:

A few years ago, David Reich discovered a ghost. Reich, a population geneticist at Harvard Medical School in Boston, Massachusetts, and his team were reconstructing the history of Europe using genomes from modern people, when they found a connection between northern Europeans and Native Americans. They proposed that a now-extinct population in northern Eurasia had interbred with both the ancestors of Europeans and a Siberian group that later migrated to the Americas6. Reich calls such groups ghost populations, because they are identified by the echoes that they leave in genomes — not by bones or ancient DNA.
Ghost populations are the product of statistical models, and as such should be handled with care when genetic data from fossils are lacking, says Carlos Bustamante, a population geneticist at Stanford University in California. “When are we reifying something that's a statistical artefact, versus when are we understanding something that's a true biological event?”

In the case of the putative ancestral connection between European and Native American groups, the ancient Mal'ta specimen from near Lake Baikal appears to confirm the hypothesis that an ancient group really did exist that contributed to both present-day groups. The advantage of the "ghost population" approach is that it does make clear predictions that can be tested with ancient DNA. Probably the most famous at this moment is the hypothesis that a very ancient "ghost population" must account for some fraction of the ancestry of the Denisova genome.

I think in many cases that "ghost population" approach is too simplistic. It is always possible to split a population into some number of dissimilar parts, but it's not obvious what the most parsimonious scenario should be. One possibility is two "pure" ancestral populations that mixed together, but there are many other possibilities -- including a single geographically dispersed population that coalesced across its range. Mathematically, the two "pure" population scenario is simpler, and it does capture some parts of the evolutionary divergence of populations. But simpler math doesn't necessarily make a more parsimonious hypothesis. When we ignore the archaeological and skeletal records in favor of math, we miss lots of information that might help shape these hypotheses.

But then, that's where anthropologists are important to understanding the past. It is interesting to see the jockeying among geneticists for new results, but you start to notice how they are describing models that don't describe any archaeological reality!