Adam and Eve: Genealogical Ancestors of What Fraction of Humanity?

[swamidass (S. Joshua Swamidass)]

@sfmatheson

@swamidass

here is the paper for clarity:

I don’t think your paper or the Nature paper did simulations to address the question with an endpoint of zero BCE, but correct me if I’m wrong.

I did not do any new simulations but I did work out if it was valid to extrapolate the 2004 study’s estimates back 2000 years.

@sfmatheson

Rohde et al. show that as little as 2000 years can work with an endpoint of today , and my read of their model is that it’s based in large part on estimates of migration and on demographics during the last 2000 years.

That is what I thought too, till I looked closer. The model is only influenced by differences between today and AD 1 in some focused and limited ways. The most difficult place to reach, in their simulation, is always the Hawaiian and Easter Islands. Common ancestry there contributed most to variability between runs. However, 2000 years ago, these islands are not occupied, so we do not have to deal with this uncertainty (which would stochastically increase convergence times substantially if it was relevant).

Perhaps most importantly, they did not model modern transportation and artificially restricted long-range migration. So the simulation is actually modeling fairly local diffusion (with high population structure), not the high dispersion we’d expect over the last 1000 years.

For reasons like these, I do argue that it is valid to extrapolate this back. Of course, that reasoning is subject to criticism, but to be effective in that criticism we have to get into the weeds.

@sfmatheson

The claim in this thread, which is that “Computer simulations using conservative migration assumption have produced times of about 2000 years,” gives (to me) the strong impression that this estimate of 2000 years is a result from genetics or biology or something like that. But it’s not. It’s based on simulations of how people move, in a model of the world during the first several centuries CE. Again, correct me if I’m wrong here.

It is based on simulations of how people move, but there is good evidence from ancient DNA (and other sources) that shows that people in the real world moved around far more than they did in that simulation. So the genetic evidence does support logic here. The actual simulation modeled less migration and movement than we expect in the time range we care about for AD 1 ancestry.

Now, one idea I’m curious about is running new simulations parameterized in detail with information from population genetics. That might become possible in the coming few years.

@sfmatheson

I think it is important to avoid extrapolating specific results like those of Rohde at al. into blanket statements about genetics, in response to important questions like the one asked by @Paul.B.Rimmer. His question, it seems to me, is about genealogical common ancestry in zero BCE.

I agree very much about having precision in this. The book, as you know, gets into far more detail, but we will sometimes speak in a more general way on the forum. Perhaps the more precise way to put it is:

The GAE is based on work that models the universal ancestry of everyone today (which converges in just about 2 ky), and makes the case that these estimates can be reasonably applied to estimate the universal ancestry of everyone alive in AD 1.