Siliceous and Delicious

After a week of relatively low efficiency, exhaustion, and feeling sorry for myself, kicked in the help of SelfControl this morning to get myself back on track. At this point I’ve reached the “morphospace vs. phylogeny” section, which requires plotting a phylogeny side-by-side with a PCO plot, with matching color codes. This requires quite a bit of R learning, since it isn’t something I’ve done before.

I first wanted to visualize what the plot should look like. I printed out tree plots of both trees I’d gotten, the one from Sörhannus and the one from Medlin. The latter I had not actually even downloaded from my emails yet. I did this, and had a good look at it, but was not impressed by the data she had sent me—the tree plotted up OK, but all the species names were abbreviated, and the tree had far fewer taxa in it than the one from Sörhannus. So, I decided to go with the Sörhannus one, whatever the repercussions might be.

I printed out his tree and the list of genus names found in the morphospace as well as on the tree (there are 41). Then I went through and manually highlighted all species on the tree belonging to those genera. It’s a pretty good, broad spread. What I want to do is to produce a pruned version of the tree with only one node for each of those genera. Had to spend a lot of time in the ape() package (which stands for “analysis of phylogenetics and evolution”) to get it to go from looking like this (the original phylogeny):

To this (the version I ended up with last night at about 7:30):

This involved removing a lot of species, shortening the names, and rotating a good many nodes to put them in an order at least broadly comparable to the Sörhannus paper (also the order, roughly, of the Medlin and Kooistra phylogenies). The motivation-sapping thing was to see how different this phylogeny is topologically quite different from the one Sörhannus published. I had just assumed that the tree he sent me was the same one that was published in the paper, and it was impossible to tell (given the image above) with the dense tangle of branches, that it was actually different. Now that I’ve spent a whole day monkeying with it, though, I don’t feel like it makes sense to abandon it and try all over. I suppose I will just have to acknowledge that it’s topologically different, and maybe email Sörhannus and ask him what the deal is. Maybe add a line or two in the paper about why, if I can figure it out.

The alternatives are to use Kooistra’s phylogeny or Medlin’s. But Kooistra was spectacularly unhelpful when I emailed him with my original character list for review, so I don’t really feel like engaging with him, and Medlin sent me her phylogeny, but it’s even messier than the one from Sörhannus because all the names are abbreviated and it would probably take me at least a day just to decipher what species the abbreviations actually stand for. So no really good alternatives.

Best to press ahead with this. I’ve got to get this thing done, after all, whether it’s perfect or not. And it won’t be.