10 files

Classification of mosasaurs - using networks

posted on 2017-10-13, 11:04 authored by Guido GrimmGuido Grimm
The file set contains figures (orginal and additional) and data relating to my post on mosasaur matrix used by Madzia & Cau (2017: Inferring `weak spots' in phylogenetic trees: application to mosasauroid nomenclature. PeerJ 5:e3782). The post outlines why networks and not trees should be used to identify weak spots in the phylogenies based on morphological data sets.

Method: Neighbour-nets were computed with SplitsTree 4 based on simple (Hamming) mean pairwise morphological distances; consensus networks are either based on a sample of equally (most) parsimonious trees (MPT) or 10,000 non-parametric bootstrap pseudoreplicates (generated with PAUP* under the least-squares and parsimony optimality criteria; and RAxML 8 under maximum likelihood.

Main result: The networks allow straightforward identification of unambiguous or unchallengend phylogenetic hypothesis and general signal issues in the data. Thus, providing a much more comprehensive analytical framework for defining systematic groups using 'cladistic' (clade-, node-based), 'phylogenetic' (monophyly) or 'evolutionary' (overall similarity + monophyla and paraphyla) classification philosophies. See provided links for further details on the data and the graphs.

Content: Figures 1–5 refer to the according figures in the blogpost. Fig. 2 shows several consensus graphs of the MPT sample, Figs 3 and 4 neighbour-nets, Fig. 5 parsimony bootstrap (BS) support networks (consensus networks in which the edge-lengths reflect the frequency of a split in the BS replicate sample)
Mosauroidea.nex is the NEXUS-formatted character matrix including the code-lines for parsimony and distance analysis.
PairwiseDistanceHeatMap.xlsx is an EXCEL-formatted spread-sheet file with the distance matrix re-ordered according to proposed systematic groups, with the cells gradually (auto-)coloured green (no difference, morphological distance, MD, of 0) to red (differing in all characters covered in both taxa, MD = 1)

The complete analysis files are included in an archive hosted at (see ReadMe.txt in the archive for file naming conventions).