Last week, Science ran a couple of items by Ann Gibbons that give further perspective on the discoveries last year that Neandertals and Denisovans both contributed to the ancestry of recent human populations. The sidebar piece, titled, "The species problem," raises the taxonomic question:
Our ancestors had sex with at least two kinds of archaic humans at two different times and places—and those liaisons produced surviving children, according to the latest ancient DNA research (see main text, p. 392). But were the participants in these prehistoric encounters members of separate species? Doesn't a species, by definition, breed only with others of that species?
I get the most awesome quote in the short article, because I get to defend the Biological Species Concept!
Gibbons describes this as a "minority view among paleoanthropologists." I can't disagree: Get a dozen paleoanthropologists in a room, and I bet only 1 or 2 will seriously propose that we could apply BSC to hominins.
This would be sort of understandable, if we were limited to the evidence of 15 years ago, with no genetics. It just wasn't really possible to test the hypothesis of interbreeding among populations, not at the scale at which we can today. There was always skeletal evidence that suggested Neandertals had contributed to later populations, as many of us pointed out repeatedly. But it was hard to quantify the phenomenon, and without quantification, it was possible for people to argue that interbreeding had been "evolutionarily insignificant".
Paleoanthropologists have instead held species concepts that did not use interbreeding as a primary criterion. Many adopted Cracraft's Phylogenetic Species Concept, or Wiley's treatment of Simpson's Evolutionary Species Concept. Both of these define species in terms of morphological characters visible to the systematist, although they differ with respect to the pattern that justifies recognizing a species.
Much of this is just ridiculous now. Genetic evidence shows a substantial amount of interbreeding between these Pleistocene groups. Lots of living humans trace more than a couple of percent of their genomes from some ancient non-African population; some may derive more than 8 percent of their ancestry from such populations. That's not rare hybridization. With this kind of evidence, we can apply the BSC.
One hangup: maybe we can't prove that our Neandertal ancestors contributed genes in proportion to their population numbers. Maybe a large fraction of their genes had a fitness disadvantage in the later population. But this is a hypothesis, not a fact. Any estimate of the fitness of Neandertals in mating with their non-Neandertal contemporaries has to take account of the demographic growth of later populations, including selection on new gene variants. We know for a fact that some Neandertal genes are today very common -- for example, one 100-kilobase region occurs at a frequency of 28 percent outside of Africa. Any assignment to a species is a hypothesis, provisional on finding new facts to refute it. For the moment the facts point to them being the same species as us.
What do we do with a population like the Neandertals, or the Denisovans? Each was more genetically distant from the average living human than members of living populations are from each other. Each evolved during a long period of isolation or strongly restricted gene flow from each other and from sub-Saharan Africans. Still, the level of genetic difference among these populations was comparable or less than that separating populations of great apes that historically have been recognized as subspecies. So that's what I would call them. Subspecies of Homo sapiens.
As a postscript, I think that whoever came up with the idea of "Denisovans" as a population name has done us a tremendous favor. The great benefit of the name, "Neandertals", is that we could talk about them without trotting out a taxonomic name. Now we have something similar in eastern Eurasia.
Also Jerry Coyne, coauthor of Speciation, discusses the issue.