Friday, November 7, 2008
Misinterpreting Phylogenetic Trees
Last night, Marta told us about the history of phylogenetic trees and touched on their modern incarnations. In yet another serendipity (we seem to be finding them everywhere), I just saw something this morning that expands on the modern part of the history, as well as supports Bob's comments last night about how misleading circular phylogenetic tree diagrams can be. I think it was Brad who first told us about the UC Berkeley evolution site, at http://evolution.berkeley.edu. There is a newly posted article that talks about how to read and interpret phylogenetic trees. It's written at an accessible level, and is completely understandable to those of us (guilty) who are not expert in biology.
One of the most interesting sections of the article is towards the end, where the author spells out the ten most common mistakes people make in interpreting phylogenetic trees.
I won't go through all ten, but here are a couple of the more interesting ones:
Misconception Number 1: Higher and Lower. Of particular interest to me since my first paper is on the idea of hierarchy of the natural world, the first misconception is that further up the tree is "higher" or "better" and further down the tree is "lower" or "less good." As the author puts it, "there is no scientifically defensible basis on which to rank living species in this way."
Misconception Number 2: Mainline vs. Sidetrack. Just by virtue of the use of the tree diagram, you visually see what looks like a main line of evolution, in the case of figure A from the root to a human, and all the other lines look like sidetracks. It's important not to take that seriously, as one could (as in figure B) simply reconfigure the exact same diagram to make the fish look like the main line and the human a sidetrack. Even T.H. Huxley made this conceptual error when he wrote that certain fish "appear to me to be off the main line of evolution—to represent, as it were, side tracks starting from certain points of that line."
Misconception Number 6: Long Branch Implies No Change. Visually, it can appear that the result of a long, unbranching line is more related to, or closer to, the root ancestor than something that is at the end of lots of branches. How many branches there are between the initial and terminal nodes is in no sense a measure of the how much evolutionary change there has been. Referring back to the figures A and B above, figure A could be misinterpreted to suggest that humans are very similar to the ancestral form, and figure B could be misinterpreted to mean that we are quite different. Of course, the diagrams are logically identical, and to draw either conclusion from the diagrams themselves would be a fallacy. (This is not to say, of course, that there are not some forms more like the ancestral forms than others, just that reading the long line of the diagram to be a measure of that similarity is a logical error).
Misconception Number 9: More Intervening Nodes Equals More Distantly Related. The author doesn't actually use this illustration to make this point, but it makes it quite clearly. If you interpret figure a naively, you might guess that frogs are more related to fish than to humans, being "closer" in a left to right sense. Figure B shows you another rendering of the identical phylogenetic tree that makes frogs look closer to humans. Of course, the right way to think about it is to look for the common ancestor. Frogs are more closely related to humans than to fish because the most recent common ancestor of the frog and the human is more recent than the most recent common ancestor between frogs and fish, a relationship that is apparent from either diagram.
Overall, none of this is rocket science, and we should be able to figure these things out on our own, but as the T.H. Huxley example shows, even the experts can think carelessly, and this article is a useful corrective. The url is http://www.springerlink.com/content/v41w288751r82653/fulltext.html
P.S. I am apparently a moron, unable to perform the simple tax of putting a live link that actually works into my posts. The links are correct, but the way I have inserted them is not working. Robbie has given me careful instructions, but I still am incapable of getting it right. I will get him to show me live, and meantime, you can copy the links into your browser.