Tuesday, September 11, 2012

Updated Mutation Rate Dates Make More Sense

Slowly but more or less surely, the human archeogenetics community is coming to the realization that the mutation rate used to estimate the age of various human and primate lineages was miscalibrated at a rate roughly twice as fast as it is, as we are coming to realize from a variety of new measurements of it. The dates calculated using the new calibration methods make much more sense in light of the likely archaeological corrolates of the human lineages to which dates have been assigned using mutation rate dating.

The four key points may be summarized as follows.

First, the divergence between modern humans and both Neanderthals and Denisovans, which was originally estimated to be 272,000–435,000 years ago, is revised to 400,000–600,000 years ago. This is in better agreement with the range of estimated split times from mtDNA and also with the idea that the ancestral population of these groups may have been H. heidelbergensis.

Second, for the split between the Khoe–San and other modern humans, revised estimates from nuclear genomic data suggest a divergence 250,000–300,000 years ago, older than single locus estimates for the root of the human tree.

Third, revised estimates of the separation time between Africans and non-Africans suggest that this predates the appearance of modern humans in Europe and Asia by up to 60,000 years. We have suggested a scenario of exodus from Africa via an intermediate population in East Africa and the Middle East, which may fit better with growing evidence for modern human occupation of the latter region before the wider colonization of Eurasia and may provide a longer interval for Neanderthal admixture with non-African populations.

Finally, revised split times of 40,000–80,000 years ago for Europeans and Asians agree better with the palaeoanthropological record and with estimates from mtDNA.

From here quoting Aylwyn Scally & Richard Durbin, "Revising the human mutation rate: implications for understanding human evolution," Nature Reviews Genetics, advance online publication, 11 September 2012. doi:10.1038/nrg3295.

General Analysis

The difference between the old faster mutation rate estimate and the new slower mutation rate estimate in sufficiently great to make essentially every historically interesting hypothesis conclusion reached with the old dates invalid. The implausibility of the old dates is one of the reasons that people who have looked at the old estimates from an interdisciplinary perspective, myself included, have long been skeptical of them, even before direct evidence that the slower mutation rate estimate was more accurate emerged.

The new dates for Neanderthal divergence from modern humans, for the appearance of anatomically modern humans, for Out of Africa, and for a staged migration scenario that causes the West Eurasian-East Eurasian divide within non-Africans to occur much later than the African-non-African divide all make much more sense with the slower automsomal mutation rate than the old dates. With the new calibrations, other dates estimated via mutation rates for which we don't have archaeological data to calibrate against may also become more credible.

The fact that is possible to come up with a calibration of mutation rate dates that coroborates the archaeological data and has internal coherence is also a relief. It had looked possible in the last year or two that meaningful absolute dating based on mutation rates might simply be impossible, because the assumption that there was a stable genetic mutation rate in humans that could be applied across the nuclear genome might have been fundamentallly flawed. If this was the case, the population genetic contribution to discerning the course of prehistory in a scientifically reliable way could have been virtually destroyed. This fear increasingly looks unfounded, at least outside areas of the genome where neutral genetic drift rather than strong natural selection. Some parts of the non-recombining Y-DNA genome may remain particularly problematic due to strong selective effects, and cofounding instances of introgression of multiple populations that make populations look older than they actually are may also continue to be an issue, however.

Analysis of Discrepency In The New Dates

The archaeological date typically cited for the appearance of anatomically modern humans in Africa is around 230,000 to 250,000 years ago, both the genetic and the archaeological dates have a margin of error, and the archaeological dates are always subject to significant adjustment if a new, older archaeological find is found and reliably dated. The archaeological date is the latest possible date at which anatomically modern humans could have arisen. The archaeological record of hominin remains generally that are that old is quite thin in the period from, for example, 200,000 years ago until 350,000 years ago, and the total number of anatomically modern humans alive at first would have been very small, so it wouldn't be very surprising at all for an accurate date for the origin of modern humans to have a genetic date to be a bit older than the oldest discovered anatomoically modern human remains. So, the Khoe-San divergence date cited isn't far off from the likely point of origin of modern humans in the archaeological record.

The possibility that the most recent common ancestor of modern humans as determined by non-recombining Y-DNA genetics and mitochondrial genetics could be younger than the date determined by autosomal genetics isn't necessarily very troubling either.

As the presence of Neanderthal and Denisovian autosomal genetic contributions in the human genome despite the absence of any Y-DNA or mtDNA traces of this admixture (or archaic African admixture that indirect methods suggest is found in Paleo-African populations at a fraction lower than that of Neanderthal admixture in non-Africas), all uniparental genetic traces of an introgressing population can be lost yet still leave autosomal genetic traces.

There are some very specific considerations that apply to the purging of uniparental mutations in the case of archaic hominin admixture that I have discussed in previous posts and that may not apply to early anatomically modern humans pre-Out of Africa. In general, low frequency uniparental gene in a population are unlikely to drop out of the gene pool during periods when a population is expanding, but do drop out of the gene pool with some frequency when the size of the population is small and stable, fluctuates up and down multiple times, or when it is getting smaller due to a population bottleneck. Some of these conditions were probably present for some of the period betweeen the evolution of the earliest modern humans and the Out of Africa date, so it wouldn't be surprising if age of the most recent common Y-DNA or mtDNA ancestor of modern humans is at least somewhat less than that of the same date determined looking at the richer set of automomal genetic data.

Could Population Structure And Admixture Explain Automsomal Genetic Ages Much Older Than Uniparental Genetic Ages?

If I recall correctly, uniparental mutation rate dating has pointed to a Paleo-African divergence from other modern humans more or less around the time of the Out of Africa event when consistent dating methods are used to fix both dates (ca. 60,000-70,000 years ago in the old calibration and closer to 100,000-140,000 years ago with the newly calibrated dates, if my memory served me). Uniparental phylogenies of non-recombining Y-DNA and mtDNA have also implied that the Khoe-San and Pygmies may share a common ancestor distinct from other modern humans at this point.

This is a big discrepency with the 250,000-300,000 year old date above, which may be hard to fit to a model of uniparental haplogroup purging solely due to random genetic drift. This is no genetic evidence so far for a modern human population bottleneck in Africa itself in early human prehistory, slow population size growth in Africa over these many tens of thousands of years is a plausible demographic scenario used in most population genetic models with success, and estimates of African effective population sizes are generally greater than estimates of non-African founding population effective population sizes.

But, admixture of diverse populations can make the automsomal genetic of the population look older. The same effect, by the way, is also seen when one estimates the time of emergence of two languages from each other based on lexical data, that is quite visible in recent efforts to date Indo-European linguistic origins on this basis.

It also seems unlikely, however that the modest amount of archaic admixture inferred to be present in modern Paleo-African populations is enough to give rise to this discrepency. There simply isn't enough of inferred archaic admixture in these populations (we have no ancient DNA from Africa which we can use to test the hypothesis directly as we do with Neanderthals and Denisovans). While Dienekes' has suspected that admixture with archaic hominins may be producing an artificially old autosomal estimate for Paleo-African population divergence, my own intuition is that admixture of diverse modern human African populations, some of which no longer have unadmixed survivors, seems like a more likely variant of this hypothesis.

A Plausible Admixture Scenario That Could Explain The Discrepency

Consider the following scenario that could explain why Paleo-African populations look older from an autosomal genetic perpsective, dating to around the time that modern human arise, than the apparent date of their divergence from other modern humans based on mutations in their uniparental genetics.

Suppose that anatomically modern humans began to fracture into diverse and geographically structured populations within Africa, with limited admixture with populations other than ones immediately bordering them, almost immediately upon evolving from one of the Homo heidelbergensis hominin species branches around 250,000-300,000 years ago.

Suppose also that anatomically modern humans did not admix with peoples other than anatomically modern humans after the emergence of this new species, except for a few clearly defined episodes of admixture with archaic hominins. Specifically, Neanderthal and Denisovian admixture outside Africa, and one or two smaller episodes of admixture with archaic hominins with Paleo-Africans within Africa where larger effective population sizes of anatomically modern humans may have reduced the impact of archaic hominin introgression.

Suppose that some of the "first wave" Paleo-African populations that began to diverge immediately from a core single modern human population near where anatomically modern humans evolved in the first place, probably somewhere in East Africa, which was more or less homogeneous genetically. There would have been many smaller first wave Paleo-African populations at diverse locations elsewhere in Africa, most of which were less successful because they had not evolved as specifically to be adapted to the conditions elsewhere in Africa.

Suppose that the core of extant Paleo-African populations in Africa are traceable not to this first wave of Paleo-Africans, but to a second wave of hunter-gatherers that split off from the core East African hunter-gatherer population much later. This second wave of Paleo-Africans would have evolved genetically and culturally in a way that made them more fit than first wave Paleo-Africans in environments less like the one that modern humans had originally evolved in, in other words, generalist advantages. These adaptations, present in non-Africans and second wave Paleo-Africans, but not in archaic hominins and first wave Paleo-Africans, gave the second wave Paleo-Africans a selective advantage over the existing "fringe" hominin populations of Africa and over archaic hominin populations outside Africa.

The remnants of first wave Paleo-African may have eventually been absorbed as minor components of other African populations. In extant Paleo-African populations, genetic drift could have eliminated uniparental markers of the existence other Paleo-African populations at the fringe of the pre-Out of Africa modern human world, while still leaving enough of an autosomal genetic impact on populations of extant mostly "second wave" Paleo-African populations to make the autosomal genomes of extant Paleo-Africans look older than the point in time at which the bulk of their ancestors diverged from a modern human core population in Africa. But, any automsomal contribution from introgression of "first wave" Paleo-African populations into the core modern human population in Africa, if there was one, may have been too slight to be detectable in the genomes of populations derived from this core modern human African population because it was larger relative to the introgressing first wave Paleo-Africans.


terryt said...

"Suppose also that anatomically modern humans did not admix with peoples other than anatomically modern humans after the emergence of this new species"

That's fairly impossible. For it to be so the suite of genes differentiating 'anatomically modern humans' from their ancestors would have to arise instantaneously across a population. Surely that is a most unlikely scenario. The problem of inbreeding depression arises immediately also. Anatomically modern humans almost certainly arose, and evolved, from the interaction of different populations over quite a period of time. In fact it would involve more than just African populations as the process seems to have continued with the admixture of neanderthal and Denisova populations (as well as others we are yet to discover).

andrew said...

You are being overly literal. In the sense relevant here, a new species does not arise until some separate population genetic population has arisen rather than at the moment the defining mutations present in the new species take place. A new species does not exist until this has happened.

terryt said...

"In the sense relevant here, a new species does not arise until some separate population genetic population has arisen"

But that separate genetic population cannot arise suddenly. The genes must be able to spread through members of the 'parent' species. And we know from the study of dairy cows that most mutations that lead to genetic change start off as giving rise to recessive genes. Recessive genes cannot be expressed until present as double recessives. The change to a new species is therefore obviously a protracted process. And a long period of the ability to form hybrids with the parent or other related populations continues. There is no point when the 'child' species is suddenly unable to form such hybrids. So your idea that 'anatomically modern humans did not admix with peoples other than anatomically modern humans after the emergence of this new species' is unlikely to be the case. I would actually guess that all Homo species have always been able to form hybrids whenever geography has allowed them to do so. After all they have been separated for no more than two million years and most pairs of species separated for just that long are freely able to form fertile hybrids.

The process of speciation is long and complicated, not sudden.