Jeanson's Method of Inferring Past Population Sizes

[swamidass (S. Joshua Swamidass)]

Continuing the discussion from Testing Jeanson's Model: Y Chromosome Mutation Rates:

Jeanson uses Y-Chromosomes to infer population sizes in the past.

His method is…“unique”…@evograd provides a link to the paper:

Testing Predictions for a Human Y Chromosome Molecular Clock

@swamidass

@thoughtful

Then he’s saying he used his y-chromosome mutation rate to create a family tree, and then takes all of the males from the family tree to estimate population growth throughout history.

His curve matches the historians’ curve.

That is not a valid way to estimate population size, and it is frankly stunning if this is in fact what he did.

@thoughtful

I watched the entire series of videos, and he compares his data to what evolution says about population growth. He is saying population growth proves him right, and he’s also making testable claims about history.

@John_Harshman

Perhaps most importantly, his initial assumption is wrong: the Y-chromosome tree doesn’t track population size. You would get a similar tree even if the population hadn’t changed at all in thousands of years. Jeanson seems completely unaware of coalescence theory, which you should look up.

@swamidass

It doesn’t even make sense from within YEC. There is no way even other YEC scientists (e.g. Sanford, Wood, Carter) are going to agree with this analysis. It is just too way out there.

@John_Harshman

Further, he claims that it’s universally agreed that the population curve fits what he shows, but he shows a world population of only a few million in 1000 BC, and nobody agrees with that.

@evograd

Here is a simple diagram:

It’s about mtDNA, but works just as well for Y chromosome DNA if you swap the sexes.

Each row is a different generation, so the top row represents the first generation of this population, and the bottom row represents the 5th generation. As you can, this hypothetical population has a constant population size - there are 8 women (men) in each generation. The lines connect ancestors with descendants.

In the first generation (bottom row), you can see that there are only women (men) coloured in black, indicating (along with the connecting lines) that they descended from the woman (man) coloured in black in the first generation (top row).

If we construct a phylogeny using the 8 modern-day (latest generation) women (men) and then apply Jeanson’s method to it (counting the number of branches through time), we would conclude that the population grew from 1 in the first generation, 2 in the second, 3 in the third, 5 in the fourth, and 8 in the fifth. But we know that the population size was constant, so Jeanson’s method leads to a completely incorrect conclusion and is therefore invalid.

[swamidass (S. Joshua Swamidass)]

Fell free to quote any salient points made on the other thread…

I want to understand this one:

@thoughtful

@John_Harshman

That was intensely painful. Jeanson is making up facts and fitting other made-up facts to them. It also has nothing to do with mutation rates. You understand that Jeanson’s force-fitting of the Y-chromosome tree to his model means that many human populations diverged from each other after 1500, right?

Yes, he does talk about that in the second video. He said his team almost stopped completely after they uncovered that fact, because they didn’t believe it either.

What timestamp is that in the video?

[John_Harshman]

Let’s wait to see if your audience of one follows you here.

[thoughtful (Valerie (non-scientist))]

@swamidass

What timestamp is that in the video?

I would love to have tell you I found it, as the idea of those populations being theoretically indistinct is covered in video 2. But I must have remembered incorrectly when they were talking about it because I went through a lot of the first five videos and didn’t find it. It was in a short back and forth, but I remember it distinctly, as it made sense to me that I would make the same mistake. I am ethnically Dutch, and the state church was Reformed. Dutch immigrants brought their church tradition to America with them. But Calvin was in Geneva, Switzerland. And in the back of my mind I always thought that was weird growing up. How did the Dutch state and churches adopt that when he was far away and wasn’t even a citizen of their country? Then it occurred to me after watching this that that part of Europe was really a whole that mixed, and national identity developed after the Reformation. After watching the series, I can now see that certain things happened because of population sizes and weren’t random. That the Renaissance and Reformation happened because population grew large enough, but people were still mixing enough to spread new ideas throughout Europe. And we can do the science we do today because our population is massive and any scientific knowledge is shared around the world.

I will keep looking for that particular section; it’s a bit tough as there is 25 in the series.

[thoughtful (Valerie (non-scientist))]

@evograd

But we know that the population size was constant, so Jeanson’s method leads to a completely incorrect conclusion and is therefore invalid.

I’ll have to read the technical papers to see if this is addressed.

[thoughtful (Valerie (non-scientist))]

@evograd Testing Jeanson's Model: Y Chromosome Mutation Rates I think here you’re still thinking in terms of the entire genome, not the y-chromosome, if I understand this correctly. I’m not familiar with these.

[evograd (Blogging Graduate Student)]

I believe @Joe_Felsenstein has published extensively on this kind of topic (phylogenetics + estimating population size), if he wants to add to this discussion.

[thoughtful (Valerie (non-scientist))]

@evograd

f we construct a phylogeny using the 8 modern-day (latest generation) women (men) and then apply Jeanson’s method to it (counting the number of branches through time), we would conclude that the population grew from 1 in the first generation, 2 in the second, 3 in the third, 5 in the fourth, and 8 in the fifth. But we know that the population size was constant, so Jeanson’s method leads to a completely incorrect conclusion and is therefore invalid.

OK - I took a look at the paper. What I understand from my non-science background is that he said, “I’ll try it out and see if it works. And it works pretty deep in the the tree.” Your diagram shows five generation, so it doesn’t work well on a small scale, I see that.

Here’s a quote from the paper. (My bold for emphasis if I’m understanding the different between shallow branches and deep branches.)
"This formula predicts the frequency with which deep-rooting Y chromosome lineages will be discovered in the future, and it derives from the multiplicative relationships among the known historical population sizes. As figs. 3–5 show, the multiplicative relationships among this historical population sizes match the multiplicative relationships among deep and shallow Y chromosome lineages. Thus, historical population sizes can be used to predict the discovery of deep Y chromosome lineages.

And this:
“As another arena of testable predictions, my results predict that the entire tree should reflect the known history of civilization post-1000 B.C.—not just the global history as a whole but also the history for specific subregions. Prior to 1000 B.C., human history is model-dependent, and, therefore, cannot be used as an independent test of my results. However, post- 1000 B.C., YEC and evolution largely agree on the specifics of human history. My results indicate that the Y chromosome tree should be consistent with this.”

Basically, it’s…“you may not like this, but if it keeps coming up with results, then it’s harder to argue it’s wrong.”

Let me know if I’m understanding that wrong.

(I also saw he got male population by dividing the population from sources consulted by 2.)

[John_Harshman]

@thoughtful

Let me know if I’m understanding that wrong.

Yes, you are. The Y-chromosome tree does not represent population size. Have you looked up coalescent theory yet? Remember the part about the men who have only daughters? The tree must narrow as you go back in time regardless of whether the population is increasing, decreasing, or staying the same.

Further, Jeanson’s population curve is bogus, and there’s nothing special about 1000 BC. His mutation rate puts many ancient divergences within historical times, and that can be checked. His mutation rate is cherry-picked and doesn’t match the great majority of measurements. And of course the archeology to which he appeals shows our species to be at least 150,000 years old, to have originated in Africa, and to have inhabited the entire world at least 15,000 years ago, the Americas being the last-settled continents.

[evograd (Blogging Graduate Student)]

@thoughtful

Let me know if I’m understanding that wrong.

I’m afraid you are understanding those passages incorrectly. They have nothing to do with the problem I pointed out. The first quote is talking about some potential future predictions of the number of as-yet-undiscovered very deep-branching Y chromosomal lineages that might be able to be made from his model. The second is simply about his model being able to infer region-specific patterns, not just about humanity as a whole.

Neither address the central problem. You pointed out that my diagram is a small-scale scenario, with just 5 generations, but the principle remains even in larger populations and over many more generations. Imagine the same diagram but 100 times wider and 100 times taller, encompassing 500 generations of a population with a constant population size of 800. The branching pattern within would be identical.

Jeanson’s method can only ever show a growing population, regardless of whether the real population remained a constant size, or even went up and down at different points in history.

[thoughtful (Valerie (non-scientist))]

@John_Harshman

Yes, you are. The Y-chromosome tree does not represent population size

I logged in this morning, as I was thinking about this and wondering if I was understanding explaining this wrong: that the y chromosome tree doesn’t represent population size, but it shows its own growth curve that can be matched to population size, since the probability of having a boy or girl is so similar and boys are usually prioritized.

[John_Harshman]

@thoughtful

that the y chromosome tree doesn’t represent population size, but it shows its own growth curve that can be matched to population size, since the probability of having a boy or girl is so similar and boys are usually prioritized.

No, that isn’t true, as I’ve been trying to explain. The Y chromosome tree neither represents the population size nor can be matched to population size. If a man has only daughters, his lineage is truncated permanently. In the present day, it will be as if he never existed. The farther back in time, the more the lineages descending from men of that time will have been eliminated. That’s all the tree shows. It says nothing about how many men were in the population in the past. You really need to look up coalescent theory. Here.

[Joe_Felsenstein]

What John H said. What evograd said.

[swamidass (S. Joshua Swamidass)]

Well I’m thankful to @thoughtful for bringing this to our attention. I had missed this, and may just not have believed upfront he had done something this out there. I’m reminded of Poe’s law…

[thoughtful (Valerie (non-scientist))]

@evograd

The second is simply about his model being able to infer region-specific patterns, not just about humanity as a whole.

It does say the “entire tree” of humanity and “but also” region specific patterns

[thoughtful (Valerie (non-scientist))]

@John_Harshman

And of course the archeology to which he appeals shows our species to be at least 150,000 years old, to have originated in Africa, and to have inhabited the entire world at least 15,000 years ago, the Americas being the last-settled continents.

After I began to be curious about the scientific process behind Genesis 1, I took a detour into looking into Nimrod from Genesis 10:8-11. In just a few sentences, it gave me enough info to compare the first empire in recorded history, Akkadian empire with the one the Bible also lists as the first. I kept following the rabbit hole in Wikipedia and satisfied my own curiosity that every significant civilization left behind ruins or written records and that tools dated later were just from farmers at the same time. The details are fascinating, but you wouldn’t believe me. Anyway satisfied my own curiosity to not worry about the dating but look at details of the archaeological and written records for clues instead.

[John_Harshman]

@thoughtful

Anyway satisfied my own curiosity to not worry about the dating but look at details of the archaeological and written records for clues instead.

You really should worry about the dating, because that destroys Jeanson’s time scale. Aren’t you worried that you ignore everything except the data you find convenient to your preconceptions? That’s not a good way to determine if those preconceptions are correct.

[evograd (Blogging Graduate Student)]

@thoughtful

It does say the “entire tree” of humanity and “but also” region specific patterns

I know, that’s what I said…

[thoughtful (Valerie (non-scientist))]

@evograd

If we construct a phylogeny using the 8 modern-day (latest generation) women (men) and then apply Jeanson’s method to it (counting the number of branches through time), we would conclude that the population grew from 1 in the first generation, 2 in the second, 3 in the third, 5 in the fourth, and 8 in the fifth. But we know that the population size was constant, so Jeanson’s method leads to a completely incorrect conclusion and is therefore invalid.

If I’m understanding it right, this is perhaps just a worldview problem. Jeanson has less than 200 generations for all humanity, so there would be very few branches that get cut off - like I said, most cultures prize having at least one son, probably many more.

Evolution has a constant population size for millennia, and there would be many branches that get cut off.

Would I be correct in framing it this way?

[swamidass (S. Joshua Swamidass)]

@thoughtful

Would I be correct in framing it this way?

No. Even from a YEC worldview, Jeanson’s argument makes no sense. It is like arguing 1+1=3. That isn’t true, even if you are YEC.