Two new papers on population genomics in subSaharan Africa (SSA) are out, and they’re very interesting. The link is to a blog that discusses them a bit. The papers themselves are open access.
I’ve long said that as we get a broader sample of genomes from SSA we’ll gain a better understanding of our ancestral population structure. There is so much diversity here, and our sampling thus far has not been very comprehensive. These papers are helping change that. In other words, the probability that our estimates of effective population size will increase as we survey this region is high.
The effective population sizes for some groups are on the order of ~20,000 over the last 800,000 years. Very, very interesting. My hot take on this is that this will push back even the “error bar Adam” significantly, but I’ll need to spend more time on this - I’m just starting on reading them. Similarly, William Lane Craig’s hope for shoehorning a sole genetic progenitor Adam into the data at 100,000 years ago (not that this was plausible in any case) is even more unrealistic.
Fascinating stuff. It certainly makes sense that as humans migrated to new areas, they would have unique mutations, and as as their populations grew would merge with the other populations, graphically similar to how individual bacterial colonies grow to fill a Petri dish.
I am wondering how the divergence and reintegration of subspecies into the genome would effect those minimum population estimates. From what I am reading in the paper, it hasn’t been taken into account. So it would be interesting to see what assumptions went into the calculation, because I know that the choice of assumptions are often dictated more by what enables calculation than what is most reasonable to believe.
In the papers they account for admixture between all sorts of groups - Neanderthals, Denisovans, and the archaic group that introgressed into the Denisovans. It’s in there, just keep reading.
Matthew is correct. The GA conjecture cannot be ruled out by data. It depends on admixture from Adam’s lineage (coming from Mesopotamia) reaching, and “humanizing”, biologically compatible but nonhuman Homo sapiens dispersed all over the globe. If we find no genetic evidence of admixture, science cannot disprove that contact was made but that Adam’s lineage failed to leave a genetic trace in the contacted populations. So, it all depends when contact was made and every Homo sapiens is now human because they descend from Adam. This undoubtedly can work, but the question is “does it work soon enough to be relevant for Paul when he’s writing Romans.” Getting everyone humanized by that time is possible, but it’s pushing it a bit, and it’s not an idea we can test with genetics.
Of course, what one thinks about “biologically compatible beings becoming human only after they descend from Adam” is another whole kettle of fish. I’ve made my stance on that pretty clear elsewhere. I’m not a fan. It means that when God is talking about Adam and Eve in Genesis as being “one flesh” that this sort of “oneness” about to become very much not the norm. You will soon have many families where one partner is “human” and the other partner is “not human” in some way, but the children are “human”. That doesn’t sound to me like what Genesis is talking about. Adam and Eve are of the same status and type; they are both in the image of God; Eve is made from Adam’s side (“flesh of my flesh, bone of my bone”). Taking the GA model on board means that what is theologically true about Adam and Eve will not apply to the vast majority of Homo sapiens for a very, very long time. But I digress.
I am talking about the actual method used for calculating the minimum populations. That is where we will find what assumptions went into the calculation.
For that you’ll just have to learn more about how the models work. The methods aim to estimate coalescence times for variants we see in the present day, and then make a statistical distribution of coalescence times to infer times when coalescence is more likely (small population size) or less likely (larger population sizes). The main parameters that are varied are mutation rates and generation times. The models also average over multiple generations to give an average Ne. The big discussion we had here last year with Buggs was all about figuring out if PSMC (and other forms of) modelling could, in principle, detect his proposed ultra-rapid bottleneck from 10,000 to 2 in a single generation, followed by exponential growth. These papers use PSMC as well as a variant of this method that uses multiple genomes instead of just one, and they are giving an average Ne. That said, I suspect these data would push the time back for the Buggs conjecture, but that’s not really an issue since he’s comfortable (if memory serves) with going back 2MYA or more (same for the ID folks). So, this doesn’t really shift their thinking.
Of course, at issue is that there has been no mechanism proposed by Buggs (or anyone from the ID group) to explain how all hominins on the planet (except two) were wiped out between 700,000 and 2MY ago and also provided conditions immediately following this event to allow for exponential population growth. If they want scientists to take the conjecture seriously, they have to propose a reasonable mechanism. There isn’t one - at least not one that I (or anyone else, apparently) can think of.
Thanks… though you had me chasing down references to coalescence times, Ne, PSMC, and Dr. Richard Buggs. That at least gives me something of the big picture of what is going on in such calculations. My only conclusion at this point is that this a large and complicated dataset and that estimates are likely to change with the addition of more data.
I have more interest in the scientific hypothesis of a South African surviving remnant of modern humans 100,000-200,000 years ago rather than any Bible based single couple bottleneck. (I not only oppose the idea of Adam and Eve being genetic progenitors, but would tend to place any historical Adam and Eve closer to 10,000 ya and associate them with the beginnings of human civilization). Since the discovery of the genetic input from Neanderthal and Denisovan populations, the South American remnant cannot be a genetic bottleneck in any absolute sense, and I wondered how that would affect estimates of the size of that remnant. The significance of a small population for that remnant would have to do with that being the venue for rapid genetic change.
I am not sure I buy that. Everything I am reading of the PSMC model looks like they are assuming a descent within a single gene pool (with single parameters for mutation and combination rates), when we already know that is not what happened. I think that assumption will lead to overestimating the population Ne when what you really have is descent from separated gene pools with smaller populations with different values for those parameters. This would not have concerned them if they were too focused on evaluating Dr. Buggs conjecture because separated gene pools would already exclude his hypothesis. In other words, the one thing this would not change is how far back in time the possibility of a single extreme population bottleneck would be pushed to… long before they can really be called human.
The methods don’t really “care” where the variation comes from - they simply compare variants in the present day. The idea that different populations had generation times and/or mutation frequencies different enough to matter would need to be supported with evidence (there isn’t any that I know of).
On the contrary, it is a premise in the model already. That is why the values are parameters rather than constants. Of course mutation rates and generation times vary between groups because such things are highly dependent on both the genome and behavior. How much time do they spend in the sun. What kind of materials do they expose their body to both in the food they eat and the environment.
But I think the flaws in the PSMC model go way beyond that, as is well acknowledged in the literature. It is after all a simplistic model for making estimates with a large degree of approximation. I am particularly concerned with the assumption it is making about constant mutation rates and no account for for natural selection greatly reducing population sizes and thereby intensifying the selection process. If there is one thing we have learned is that you don’t have to go very far in the past to discover that life was rather hard and survival was anything but easy. So life spans and infant mortality was highly variable and dependent on many environmental conditions.
And notice again that I am not challenging the conclusion with regards to the Dr. Buggs hypothesis. I think that conclusion is sound. But I think you are somewhat exaggerating the reliability of these calculations, and I expect that new data and evaluation techniques are going to reveal a more complex account of what happened. For I think the real lesson of this discovery is how the human genome gives us a very rich source of information on what happened in the past – one we are only just beginning to explore.
These are populations of essentially the same organism - you’re proposing that they will have significant differences in mutation rate and generation times? Note that these are long term average values. Again, if you’d like to argue for this, what evidence can you present?
Also, how is mutation rate “dependent on the genome”? Not following you here.
Spending time in the sun affects the germline mutation frequency? I think you’ve misunderstood what is being discussed here.
These are populations of humans and so their behavior isn’t just a biological function. They have traditions which adapt to the areas in which they live. Yes, I am proposing that they MIGHT indeed have significance differences in mutation rate and generation times. I think you can observe such differences in human populations today.
Yes and I am familiar with the history of scientific discovery where the reliance on long term average values have greatly misled us.
That would be relevant if I was arguing for a specific conclusion, but I am not. I am only arguing that the certainty you are attributing to these population estimates are excessive, especially when you say that more data can only increase the population estimates. I think that points to a method of calculation which is ignoring too many details.
Mutations have both external and internal causes, and as we all know the internal conditions and operations of the organisms are highly dependent on the genome.
I agree that UV exposure is not a major source of mutation in complex organisms especially because only germline effects are inherited, not like it is for bacteria. But this is only one example of the environmental causes for mutation that is going to be dependent on the behavior of members of a particular population.
Any differences in mutation rates I’ve seen seen suggested for different human populations have been quite small.
I’m not sure what you mean here. Natural selection is less effective in a small population, not more effective.
On the question of structured populations. . . The effective population size is usually a good deal smaller than the census size, but in a highly structured population, the effective size could be larger than the census size. (Or so I’ve heard – I haven’t seen an exploration of when that’s likely to occur.) I’ve long wondered what effect population structure had on published PSMC inferences about demographic history.
Differences in the age at which people have children is rather culturally dependent and this is also linked to differences in mutation rates. The longer generation times the higher the mutation rates because the gametes have more opportunity for replication errors.
The most universal cause for natural selection is death. Thus you are going to see the most natural selection when a population declines to a small remnant. To be sure, the smaller population is likely to have a smaller number of mutations, and yet the population only remains small because the selection remains high.
Ah yes… Ne is the effective population size… thanks for pointing out that difference. I think I glossed over that on my first reading.
