Morphological phylogenetics

After weeks of travelling back and forth from another campus for computing and ecology, I'm now back at the NHM. An hour's commute each way never seemed so gentle.

This week is phylogenetics, starting in the same way the field did by completely ignoring DNA. We've been learning how to turn the morphological variation you see in organisms into hard facts that a computer can use to compare them, and through comparing many many features, suggest their evolutionary relationships. With molecular data you typically have a fixed number of characters: the base at this point in the sequence can only be A, C, T or G, for example. But there's a real art to choosing and defining morphological characters, as I found out yesterday.

In the practical session we were given a set of taxa to describe. My group came up with our set of characters and we ran the analysis and got a fairly nice cladogram, for something that took an hour. Most taxa had a clear set of relatives and you could trace the evolution of some of our characters back to common ancestors.

When I got back, I decided to ramp it up a notch, adding 16 taxa and changing some of the characters. I experimented with lots of different character arrangements, mostly unsuccessful. This not-brilliant cladogram was the result:

Did I mention the taxa were biscuits?

The methods used by morphological phylogenetics just assume there's an evolutionary relationship between the items you describe. One of my classmates apparently did one on screws.

So, to the science:

It seems most of the finger biscuits are closely related (yellow), and the two chocolate chip cookies have come out as sister taxa (green) as expected. Interestingly, the sandwich condition seems to have evolved multiple times, as it crops up all over the cladogram. The single-layer sandwich fingers (red) probably evolved the finger shape independently of the other fingers.

But, the tree has failed to recognise the close relationships between hobnobs and coated and uncoated digestives respectively; I feel these are unlikely to be convergences but represent true homology. It also failed to recover many deeper-level relationships, leaving a very large polytomy of most round biscuits. The single exception is the Viennese whirl as the basalmost biscuit (the jaffa cake is the outgroup). This is interesting, as the Viennese finger has come out as one of the most derived, so here the Viennese resemblance is likely to be convergent.

In conclusion, this analysis has revealed only a few useful characters with which biscuits can be classified, and these are still apparently subject to homoplasy. It is recommended that molecular methods be applied to offer a different perspective. Sadly, DNA extraction from biscuits is not an active field of research.