Phylogenetic congruence between cranial and postcranial characters in archosaur systematics

To infer phylogeny, palaeomorphological character data are usually analyzed by maximum parsimony. Moreover, cladograms are usually inferred from a simultaneous analysis of all of the character data (‘total evidence’), even when there may be some theoretical or operational reasons for comparing cladograms from subsets of it. For example, the cranial characters of vertebrates are often claimed to be more phylogenetically informative than those of the postcranium. We test this assumption for non-avian archosaurian taxa, re-analyzing 50 data matrices, most of which were published between 2010 and 2011, all in peer-reviewed journals. For cranial and postcranial characters, we compare Consistency Index (CI) and Homoplasy Excess Ratio (HER) statistics, showing that cranial characters do appear to be significantly less homoplastic than postcranial characters. One possible reason for these differences is the greater number of cranial characters compared with postcranial characters in most published datasets, a factor not controlled in previous studies. We investigate the
effect of numbers of characters here using data rarefaction methods. These demonstrate that the strength of signal is more comparable once partition size is controlled. We also confirm here that the HER is a better indicator of homoplasy between datasets than the CI, because it is less biased by data set parameters including the number of characters and taxa. We next implement two randomized partition homogeneity tests: the Incongruence Length Difference (ILD) test and the new Incongruence Relationship Difference (IRD) test, the first of which measures difference in cladogram-length (steps) and the second of which is based on tree-to-tree distance metrics. In many cases, partition homogeneity tests return significant results (p-values < 0.05). Finally, we present a novel method for visualising taxonomic overlap between our tested datasets in the form of a network diagram, enabling us to understand their similarity in terms of taxon sampling.