Update to Eller et al.'s "Local Extinction and Recolonization, Species Effective Population Size, and Modern Human Origins" (2004)

Humans have about the level of genetic diversity that you would expect from a Wright-Fisher population model with 10,000 individuals.

That sentence opens a labyrinth of interpretation. Why do we have the genetic variation that we do? Is 10,000 the number of our ancestors, or are we inbred for other reasons? Haigh and Maynard Smith, writing in 1972 about β-globin polymorphism, first proposed a population bottleneck. At the time, few human biologists recognized that such a hypothesis would cut the Neanderthals and other archaic humans from our ancestry. A relationship between population size and protein diversity across a large sample of animal species was documented by Soulé (1976). Nei and Graur (1984) showed low protein polymorphism in dozens of species, from wolves to flies, suggesting that Pleistocene bottlenecks may have affected most species. When human mtDNA sequence data showed low worldwide nucleotide diversity in humans (Cann et al. 1987), a population bottleneck entered the human genetics mainstream.

But long before this, the interpretation of polymorphisms became a focus of the antagonists in the neutralist/selectionist controversy (Gillespie 1991). Neutral evolution in a Wright-Fisher population model is algebraically simple, but natural populations depart from the assumptions of this model in many ways. Some population structures radically increase the chance of inbreeding, thereby lowering genetic variation even in large populations. Motoo Kimura and his students explored the consequences of many such models.

One is the subject of our 2004 paper (Eller et al. 2004). Maruyama and Kimura (1980) initially developed the model of extinction and recolonization, showing that the right combination of parameters could generate strong inbreeding. The model was interesting because of its application to a famous model species: E. coli. This species may be a ponderous swarm of quadrillions of cells, but it behaves genetically like a nimble population of a few billion. When its colonic hosts die, they take whole communities of gut flora with them.

Of course, humans aren't bacteria. We set out to show whether Maruyama and Kimura's model could work in a hunter-gatherer population setting. Takahata (1994) had considered the scenario as a possible explanation for human variation, but it was clear that we needed to consider what parameter values could be realistic in the Pleistocene. Hunter-gatherer groups historically intermarry extensively. Groups sometimes disappear, but most of the individuals don't die—they just move in with their neighbors. These facts could have proven fatal to the model, which depends on a high rate of extinction and a low rate of gene flow to cause inbreeding. Still, we found a range of parameter values that might be realistic and that really cut down the genetic variation in simulation models.

After five years, how does our study hold up? The biggest change in the succeeding interval has been a better appreciation of variation in our close relatives. In 2004, it was evident that the mtDNA variation of apes was many times higher than it was in humans, but autosomal variation was only starting to be understood. Bonobos have roughly the same variation as humans (Yu et al. 2003), chimpanzees and gorillas run around double, and orangutans have three times the human value or higher (Caswell et al. 2008; Fischer et al. 2006; Yu et al. 2003). It now seems possible that humans and apes were subject to similar demographic processes, with inbreeding amped up a bit in our ancestors and regional isolation higher in the ancestral populations of most apes. Recently Neanderthal genetic information, including multiple mtDNA sequences (Briggs et al. 2009) and nuclear genome analysis (Green et al. 2006; Noonan et al. 2006), has yielded empirical observations of genetic variation within ancient humans. Modern humans, Neanderthals, and the human-Neanderthal ancestral population all appear to have had humanlike levels of genetic variation (Premo and Hublin 2009). Explaining human genetic variation is no longer a problem of a single bottleneck; it requires a demographic model that can encompass Pleistocene humans and to a greater or lesser extent other...