Tuesday, August 21, 2012

Why didn't I publish this? Quasi-Repost: Messing around with trees: Ardipithecus ramidus

Hey there, haven't posted anything in a while haven't I? Well this isn't a new post either. I just went through some of my older stuff and stumbled upon this Post I wrote a year ago, which I never published. Probably because I thought it's bad. Well, either I've lost my sense for quality, or I didn't realised this Post was actually quite interesting. Be it as it may, here it is:


A few weeks back when I desperately sought for stuff to waste my time with, I came up with a very nice, but also quite strange Idea: How about ruining some phylogenetic trees?


There are some nice Tools on the Internet which enable you to execute small to middle scale analyses for free and without relying on huge clunky programs. One of them, which I use quite a lot right now, is on the homepage phylogeny.fr (Dereeoer et al. 2008). It’s usually a tool for DNA data, but with some little adjustments you can also use it for basic morphological Datasets.



For my first target, I decided to attack a big fish and therefore the most easiest to attack: Ardipithecus ramidus.

The authors didn’t publish a tree by themselves, but they did publish a nice little table of characters to show how and  good ol’ Ardi is related to other hominids.


White et al. 2009




So my first step was to reconstruct a tree from this table, the results weren’t really surprising:










Now what happens if I change some of the more controversial characters (some characters of the pelvis and some regarding the C/P3 complex) from “derived” to “ancestral” (I refuse to use the word “primitive” in this context)?










Well looks like this didn’t do the trick. Let’s try something more drastic, like deleting all dental characters.






Gotcha!

You see, from the 68 characters that were listed in the initial character table, 34 of them were located on the teeth. And most of the phylogenetic signal which separated Ardi from the other early hominids is located within those 34 characters.

I would’ve loved to deconstruct this tree even further, but I couldn’t find any means without doing completely ridiculous stuff. This is mostly because of the way White et al. (2009) defined their Outgroup Taxa. The Outgroup is one of the most crucial parts in phylogenetic reconstruction because it defines the ancestral state of all characters. Therefore different Outgroups can change the topology of your tree in a huge way. The Outgroup characters in the Table from White et al. consist of inferred character states from a hypothetical last common ancestor. Since we have hardly an Idea how this last common ancestor looked like (I wrote something about this here), there is a huge space for “inference”, others might even call it “speculation” or “guessing”.

Keep in mind that this whole stuff is just the result of me, messing around with this stuff. Nothing of this has any scientific value. However, I still learned some interesting stuff while doing this.

First of all, although discussions about Systematics and phylogenetic trees can become really boring from time to time, they’re sometimes very helpful when you try to get to the core of a certain problem. They also provide nice means to visualise these problems and the effect it has when you change certain parameters.

In the end, when you intend to challenge the initial taxonomic placement of Ardipithecus ramidus there are two things you have to look at.

Try to show that most of the dental characters White et al. (2009) used are redundant and/or false and try to show that their inferred ancestral character states are wrong.




You see, Phylogenetic trees can be good for something and if it only is for stating the obvious...

References:

Dereeper A. et al. (2008) Phylogeny.fr: robust phylogenetic analysis for the non-specialist Nucleic Acids Research. 1; 36


White, T. D. et al. (2009) Ardipithecus ramidus and the Paleobiology of Early Hominids. Science 326, 64. S 64-86.