Comments on: On the limits of feathery trees

By: Christopher@BorderWars

Christopher@BorderWars — Sun, 12 Aug 2012 05:01:52 +0000

I’ve written on how these trees can be deceptive especially when they are sparsely populated with data that has a strong selection bias. While most scientists are likely interested in how closely populations are related, I’ve documented a poor study which uses these trees to suggest that there’s a significant genetic split within the Border Collie breed.

http://www.astraean.com/borderwars/2008/09/breed-apart-ii.html

The researcher who added the Border Collie spin has a particular political bias in favor of such a split (if it exists).

By: Razib Khan

Razib Khan — Sat, 11 Aug 2012 23:15:16 +0000

#1, the issue i have is that these two dimensional representations remove some information…which is fine i think when you have a small number of branches. but it seems that too many clades are being thrown down here, and misleading people as to relationships. i’ve been mislead in the past by unrooted trees. i think people who look at these all the time don’t have a problem, but for anyone who is outside the core of the field it can give a false sense of precision.

By: S.J. Esposito

S.J. Esposito — Sat, 11 Aug 2012 13:24:58 +0000

I don’t mind the tree–it’s just as useful as any other network or neighbor diagram. But you asked if it offers anything extra, and the answer is no. In fact, it offers less than a table in terms of information (as Dienekes said); but it’s a decent visual representation of a result, I think.

By: Brazen Normalcy

Brazen Normalcy — Sat, 11 Aug 2012 12:52:33 +0000

Agreed. Pretty. And pretty useless. I could see such a tree having utility in a digital setting, but mainly because then you might zoom in on a branch, revealing actual data there.

By: Dienekes

Dienekes — Sat, 11 Aug 2012 11:44:03 +0000

Figures have their uses, it’s easier to spot a little red outlier in a background of blue, rather than going through a long list of numbers.

But, I think there should definitely be more “numbers” to accompany the figures. Numbers can be re-analyzed, but figures usually can’t.

By: Mike

Mike — Sat, 11 Aug 2012 08:38:48 +0000

I would MUCH rather have the tree. Of course, I am constantly looking at gene trees. To try to reflect the topology and the branch lengths for so much data would be a mess of a table.