The Stained Glass Window of Life

Cephalopoda, originally uploaded by Arenamontanus.

Sometimes a good thing is hard to improve upon.

Hierarchical classification schemes for living things go way back in Western throught, evolving from Aristotle’s Scala Naturae to Carolus Linnaeus’ Systema Naturae in the 1700s. Later, phylogenetic trees were used to represent the evolutionary relationships among organisms and like many concepts in evolution, their use dates to Darwin’s time. In fact, Darwin scrawled something of a phylogenetic tree in his notebooks and published a tree in the Origin of Species.

Darwin’s Notebook Tree.

Darwin’s Figure in the “Origin“

There are a variety of ways that phylogenetic trees are represented: rooted or unrooted, vertically, horizontally, or diagonally branched. There are many sources of data that can be used to construct phylogenetic trees, with morphological and molecular characteristics being those most commonly used. There are also many methods by which the trees can be constructed; the major frameworks favored for constructing these trees stem from the principles of parsimony and likelihood. Parsimony trees run on economy, minimizing the number of evolutionary changes necessary. Likelihood trees are generated by maximizing the “likelihood” of the tree given a set of data. There’s been lots of argument in the systematics crowds about which ways are the best to infer the relationships among organisms, and it continues to be a place to make work for yourself if you have a really sharp mind.

Sifting through that debate is not the point of this article. This is more about convention, utility, and aesthetics. I just ran into Arenamontanus‘ Flickr stream the other day and found these images of “Voronoi Treemaps” that he has used to represent the evolutionary relationships of different organisms. Instead of the “tree of life”, these look like stained glass windows of life.

I haven’t yet finished reading the rules on how these are constructed, but like other methods they are a way of organizing information hierarchically. There is more information here and a wiki here. I also know that they were created using MatLab, a very flexible statistical, mathematical, and graphical software system (if you can get your head wrapped around it).

So what’s the point of the Voronoi Treemaps? They are unconventional and I don’t think that they are as useful as “normal” trees, but it just may be that I am used to interpreting the traditional representation. The aesthetic is what attracted me to these images (the one that I posted at the top is gaudier than some of the other’s that Arenamontanus has created). But compare the one at the top’o'the post with the one below. Both represent relationships of the Cephalopods, the latter being published by: A. R. Lindgren, G. Giribet, M. K. Nishiguchi (2004) A combined approach to the phylogeny of Cephalopoda (Mollusca) Cladistics 20 (5), 454–486.

Is the Voronoi Treemap an improvement on this tree? Maybe not, but I appreciate a new way to represent a hierarchical scheme. There may be more that can be done with it, and there may be more to it than what I can currently appreciate. But then again, it is hard to improve on something so simple that has worked for so long.