Comments on Jeanson Accuses Duff Again

The gift of the benefit of the doubt is gone with me when it comes to Jeanson. He’s either completely ignorant of even the most basic underlying concepts or he’s a liar. I’m inclined to think it’s a bit of both.

The fact is there is no reason inherent in an unrooted tree to say that any node is relatively older or younger than any other. It becomes merely a cluster diagram of similarities. Such trees have no direction to them.

That doesn’t sound right… @John_Harshman and @Joe_Felsenstein?

There still is nested clades in an unrooted tree…

That is hilarious. It shows how absurd and ad hoc all this is.

1 Like

Of course. It’s a nested pattern of similarities. They however have no inherent direction. Without an outgroup you could choose to root them everywhere and depending on where you choose to root you change the relative ages of the nodes. That is the whole reason for using an outgroup is to give a tree some temporal polarity. Like John said however there is also using a midroot but Jeanson didn’t do that either.

There are many more possible rooted bifrucating trees than unrooted bifrucating trees for the same number of leaves. This means for the same number of taxa you can have many different possible histories depending on where you choose to root your tree. Jeanson ignores this entirely and just makes up ages for nodes that fit his intuition.

It’s not relevant for the present discussion, but I wanted to note that there are some recent interesting alternatives to midpoint, and outgroup rooting. For example:

and

5 Likes

With the assumption of equal mutation rates on all legs (which isn’t so bad of a simplification) there should be a unique rooted tree consistent with the data. That tree also would be rooted at a place that’s fairly easy to eyeball off the unrooted tree.

This is nearly over my head but that doesn’t sound exactly right. I feel like there could still be pairs of legs with equal length which allowed two equal choices.

And where would the directionality come from?

None of which Jeanson employs.

1 Like

Nope, he certainly did not.

2 Likes

We don’t determine roots by “eyeball”.

Directionality comes from centrality in the tree. The more ancient coalescences are longer wait times too, which substantially reduces noise in the inference of the root.

What you can’t do is resolve which mutations fall on either side of the last coalescent. That’s a different question than what we are asking here.

Of course not. We’d determine it mathematically in a way that happens to be easy to eyeball :slight_smile:.

1 Like

You could say that, but you wouldn’t be able to tell what any of them are. Nesting requires direction. Think of it this way: in a fully resolved tree, three branches enter each node. One of those branches is the ancestral branch and two of them are the descendants, forming a clade. But unless the tree is rooted, there’s no way to tell which is which.

2 Likes

I mean if he were to root his trees and then explicitly cite how he did so and why he chose that method to root his tree over another then we are fine but he explicitly goes out of his way to emphasize his trees have no root which explicitly means he can not make inferences regarding relative ages of the nodes.

2 Likes

Sure, but it would not be “merely similarities”. That’s what I was objecting to.

That is all the information you would have on an unrooted tree, the relative similarities of the leaves on the tree. You have no information about direction as John explained.

Well that’s where I disagree. There are experts here of course. I’m happy to be corrected. Please show me how I am wrong.

That would be the midpoint rooting I mentioned previously. Jeanson doesn’t do it. And if there are short branches near the midpoint, small fluctuations in rate will give you the wrong root.

2 Likes

@John_Harshman just explained it. Each node in a bifrucating tree has three branches. In any unrooted tree you have no information as to which one is ancestral to the others.