PAST - Case study 6 - Systematics of Cambrian trilobites and the Eutheria

Case study 6 - Systematics of Cambrian trilobites and the Eutheria

Data files: paradoxididae.nex, and eutheria.nex

Babcock (1994) produced character matrices for several groups of Middle Cambrian trilobites. We will look at a matrix for the Paradoxididae family, with 24 characters and 9 genera.

Read in the Nexus file paradoxididae.nex. Note the presence of unknown character states, indicated by '?'. Select all, and choose 'Parsimony analysis' in the Cladistics menu. Use the 'Branch-and-bound' algorithm with Fitch (unordered character states) optimization. You should get the following tree:

Elrathia is here used as the outgroup. Can you write down a possible reconstruction of how character H evolved along the branches of the tree? Also try Wagner optimization (assuming ordered characters). Does the tree change?

Distribution of cladogram lengths

(If you have a slow computer, you might want to skip this exercise. It takes about 4 minutes on a medium-speed Pentium machine).

Choose Exhaustive search, with Fitch optimization. Around 135000 trees will be searched.

Does the distribution of cladogram lengths indicate that you have a 'good' or a 'bad' tree?

Bootstrap analysis

Now choose the heuristic (NNI) algorithm, and set the number of bootstrap replicates to 100. You should get a result somewhat similar to this:

Discuss the bootstrap values. Which groups are well supported, and which are poorly supported?

Systematics of the Eutheria

We will now work with a larger matrix, containing 64 characters of 20 Eutherian (mammalian) orders (Novacek et al. 1988).

Read in the Nexus file eutheria.nex. There are no unknown character states. Select all, and choose 'Parsimony analysis' in the Cladistics menu. Use the 'Heuristic (NNI)' algorithm with Fitch (unordered character states) optimization. You will get a large number of most parsimonious trees. Look at a few of them, and then click on 'Consensus' in the 'Most parsimoniuos trees' window. The result may vary a little from run to run because of the random reordering procedure, but here is a typical result:

Look at the position of the primates. They seem to form a monophyletic group with the Scandentia (tree shrews), the Chiroptera (bats), and the Dermoptera (flying lemurs). Is that reasonable? What do you think of the position of the Perissodactyla?

Cluster Analysis

Just for fun, we can try a cluster analysis of the Eutheria matrix (the old 'numerical taxonomy' approach). Select the whole matrix, and choose 'Cluster analysis' in the Multivar menu. Use the Euclidean similarity measure. Compare with the cladogram above, and comment on some of the differences. Which method seems to perform 'best', that is, gives results in best correspondence with traditional classification?

More information about cladistic analysis can be found in the manual.