A common model of the range expansion of modern humans out of Africa ~50,000 year ago forces us to conceptualize it is as a tree with successive bifurcations. Each of these bifurcations often is accompanied by a bottleneck in one of the daughter populations, with the sum totality of the demographic events producing a “serial bottleneck” model of the origin of modern human lineages around the world. Though this paradigm has been around in various forms for decades, most influential among geneticists has been the 2005 paper, Support from the relationship of genetic and geographic distance in human populations for a serial founder effect originating in Africa. In the paper the authors show persuasively that heterozygosity declines as a function of distance from Addis Ababa, near the likely point-of-departure out of Africa.
But there’s minor problem with this model. Most extant populations may in fact be compounds of highly diverged late Pleistocene lineages. That is, after an initial serial founder expansion ~50,000 years ago there may have been many local extinctions and admixture events overlaying it. That is the model that Joe Pickrell and David Reich argue for in Towards a new history and geography of human genes informed by ancient DNA. While in a serial founder conceptualization of an out of Africa migration modern populations are the tips of the phylogenetic tree, in the Pickrell and Reich framework they’re syntheses of divergent evolutionary histories. As Towards a new history and geography of human genes informed by ancient DNA shows older genetic techniques and data, not informed by ancient DNA, may not have had the power to differentiate between the serial bottleneck model, and one of reticulation and fusion.
It seems likely that the serial founder model has some utility. One can still hold defensibly that everyone outside of Africa, excluding recent admixture from within Sub-Saharan Africa, sits within their own phylogenetic clade. That is, all non-Africans are equally related to Sub-Saharan Africans, because all of them descend from an ancient population of ~1,000 breeding individuals. But, you have populations such as South Asians who are both numerous, and, fusions of two very distinct branches of out of Africa humanity. The stylized model of a tree subject to bifurcations fundamentally misleads in this case.
All this came to mind reading a new preprint that’s on biorxiv, Distance from Sub-Saharan Africa Predicts Mutational Load in Diverse Human Genomes. Theory and intuition should suggest to us that out of Africa populations will have higher genetic load of deleterious mutations than within Africa populations. The reasoning is straightforward: the power of selection to remove deleterious mutations is hampered the smaller the effective population size, as random genetic drift becomes more determinative in generation to generation changes in allele frequency. More formally if 4 N es << 1, where N e is effective population and s is the selection coefficient, then even deleterious alleles behave as if they are neutral. From neutral theory we know that the rate of substitution in this model is simply the rate of mutation. That is, molecular evolution is determined by new mutational input, rather than being constrained or diversified by selection. In small populations which are drift dominated N e can get very small. I’ve seen assertions that the original group of humans who settled North America and South America may have had an effective population in the first generation on the order of ~100. In their history they also had the out of Africa effective population of ~1,000. In addition, there were likely bottlenecks between Berengia and the out of Africa event.
You can see in the figure above from the preprint that the number of deleterious mutations does seem to increase with further distance from Africa. Their genomic coverage was good, ~80x on the exome. That is, when they found variants that differed from the reference sequence they could be confident that it was not in error. On the other hand their population coverage struck me as less than ideal, though I am willing to accept that their result is probably true (and obvious with finite resources they selected their individuals and populations to be informative). They admit for example in the text that the Mozabite population has higher heterozygosity due to a back-to-Africa migration, which has had successive admixtures of Sub-Saharan ancestry. In addition, it is also rather inbred. Its demographic history bears no correspondence to the serial founder bottleneck model which spans 50 to 15 thousand years (i.e., from the out of Africa to the settlement of the NewWorld). The Pathan and Cambodian populations also are actually the product of Holocene fusions between distinct groups with very different histories. The PSMC results in the bottom left panel can be thought of as collapsing distinct population histories, and, from what I am to understand may then inflate the effective population trajectories of individuals who descend from admixture events.
As one might expect from this sort of title the authors refer back to the 2005 paper that I mention above. I don’t want to belabor this point, as I think the authors’ results are probably robust, and, important, population history aside. But, much of the audience will not know that the serial founder bottleneck model is now being challenged. The 2005 paper has 588 citations as of this writing. The Pickrell and Reich paper, which was published in 2014, has 5 citations. I just want to mention this since it’s a preprint and presumably the authors are taking in any critiques.
When it comes to burden of deleterious mutations in human populations there have been some conflicting results, reviewed in the preprint, about mutational load. The authors argue that results which suggested that non-Africans did not exhibit higher load were subject to a bias which did not have power to detect non-common variants, and also modeled mutations as additive, as opposed to across the full range of dominance (h). It turns out that non-Africans, and those populations which are more drifted, exhibit higher recessive deleterious loads. This is what you would expect intuitively, as you need large populations to purify this class of deleterious alleles (since they are only exposed to selection in homozygotes). This matters when it comes to expectations of the number of recessive diseases one might expect in populations which practice consanguinity. I would, though, have liked to see more typical populations in the mix. For example, instead of just African hunter-gatherers it would be nice to see the Yoruba, as well as Han Chinese, and a northern European population. I doubt it would be very surprising, but it would give one a better baseline.
Finally, I want to note that many ancient DNA results, from “archaics” to modern human Mesolithic hunter-gatherer groups, have very low effective population sizes due to inbreeding. Genetic load may be more important in the history of the human species, especially on the edge of the range in the far north, than we may now understand, because of its tendency to reduce fitness of groups due to the drag of recessive disease.
Citation: Distance from Sub-Saharan Africa Predicts Mutational Load in Diverse Human Genomes,, , , , , , , , , , , ,