Human Genome Diversity in Africa

Two very important papers were published today on the genomic diversity of humans in Africa.
The effective population size is 20,000 to 30,000 going back to the beginnings of our species 300,000 years ago. This should put an end to the Buggs controversy that a bottleneck to a single pair of humans wasn’t proven to be not possible. These papers show that the Buggs Conjecture is false. Humans in great numbers lived as hunter gathers for hundreds of thousands of years all throughout Africa and expanded into Asia Europe in the rest of the world more than 100,000 years ago. Further crushing the Buggs Conjecture, admixure between many species of humans was not a rare occurance but happened early and often although the African Continent.

1 Like

This, nonetheless, is a great paper. Thank you.

I don’t think you know his conjecture. It does not appear this is the case at all.

Quote from the first paper (paragraphs are from two different sections)

Would a bottleneck of 2 after that point wipe out population substructure, or could those substructures be an artifact after a bottleneck of 2?

1 Like

There is strong evidence against a bottleneck of 2 at 200 kya. Right now, Buggs is proposing something farther back in time, at about 500 or 700 kya, or even 2 mya ago. From conversation among @AJRoberts and WLC, I can tell you also that conversation has shifted to how to theologically think about interbreeding in these models. If interbreeding is allowed (and it is for both RTB and WLC), a genetic bottleneck is a red herring.

Instead, focus will be on “how much” interbreeding, and a range of associated theological questions.


Effective population size might not be what you think it is.

1 Like

500,000 years ago, there were tens of thousands of humans in Africa.

1 Like

You are misreading that graph. Sorry Patrick. That is not a point estimate, but a harmonic average over tens of thousands of years, not a minimum.

1 Like

Please educate me on the correct way to read the graph. Thanks.

1 Like

You have to read it as if it was a moving average (Moving average - Wikipedia). Here is the example from wikipedia.


It is as if you are looking only at the yellow line, and can’t see the green one. The wider the window the less of a match you will see between the two. In the case of population genomics in the distant past, the window size is tens of thousands of years wide.

For example, let’s say a stock’s price over the last 50 years is $20. It is possible that the price is lower at times, in fact it is likely so. For example, it is possible the price is $0.01 today, or $1 forty-five years ago, or even $0 on a specific day 10 years ago. The average over a long period of time has little correspondence with the minimum amount on any particular day.

So yes, the average population size was very high, but this has almost nothing to do with the minimum population size.

1 Like

Thanks, I understand a stocks moving average well as I am a fundamentals stock investor that views technical analysis as nonsense, so it is long term green line is what is important to me.

So looking at the Ne graph, I should interpret the curves as moving averages over time with the actuals widely varying +/- 3 sigma at each point? If that it is the case, you can’t see a large short term dip.


Conclusion is right, and so is the analogy, but some differences,

  1. The graph does not specify the 3 sigma line, and the way to do this is not yet published (and would be derived from TMR4A).
  2. The window increases dramatically as we go back in time, making the divergence between average an point estimate larger and larger.
  3. Technically we are using the harmonic mean, not the mean.

What I find interesting about that graph is the dip at about 500,000 years ago. This dip is consistent with either (1) a brief and very sharp bottleneck, (2) a long term dip of population, or (3) some thing in between. We can’t really untangle which one it is. This, however, is not evidence against 1.


Yes, the dip at 500,000 years is interesting. Perhaps climate lead to a dip in human population and rebound over a million years.

1 Like

The dip is better understood as a “shroud of noise” … where it becomes increasingly impossible to give a strong probability to a population of 2 to the maximum estimates (whatever they are).

It’s like listening to a Beatles song that someone says includes cryptic lyrics that say “Paul is Dead”.

So we volunteer to analyze the music and the lyrics using sophisticated sonic analysis. However the process requires turning up the “gain” which unavoidably allows more and more “seemingly random static pulses” to become more frequent and louder.

Your conspiracy-minded friend is absolutely certain that the words “Paul is Dead” is in that song somewhere. And so we dutifully continue to drill down, increasing the “gain”… and systematically looking for any combination of lyrics and noise that sounds like “Paul is Dead”.

At last, we read the point in the analysis where not a single word can be distinguished… only individual noise impulses. And so it is at this point in the analysis (at the 500,000 decible sine wave, or any other acoustic buzz words that fit the circumstances), all the technicians raise their hands in frustration!

All of them say… okay, the analysis is complete, because we can no longer distinguish a single word, let alone a 3 word string. It is at this point where the conspiracy-minded friend says: “At last, that is where the phrase Paul is Dead can be found! Somewhere in that cacophony of a cloud. Thanks for helping to prove my case!”.

What the … ?

Nothing has been proven… except that nobody can say for sure that someone just may have muttered “Paul is Dead” to himself… when nobody else could hear him. This makes for a great story. But it hardly makes for important evidence.

Also, doesn’t have something to do with the genetic diversity observed as opposed to actual population figures. For example Ne in current human populations is in multiples of 10000 (even though actual population is in billions).

I wonder whether its even possible for Ne to be calculated less than in factors of 10000 for humans.It seems to me that the overall genetic variation in ancestral populations will always be calculated higher than in the different branches that diverged. This is also in line with the calculation of Heterozygosity over time -

Ht = (1-1/2Ne) ^t X H0

So the founding population will always have a higher heterozygosity than the current one. (Atleast it will appear so in models based on Drift)

Would this be an accurate understanding?