Question relates to the signal issues of this data matrix (see title) – all morphological data matrices struggle with incompatible split patterns — well reflected by the overall low bootstrap support in this preferred tree and the large soft polytomies in Fig. 6. Inferring trees on such data can be non-comprehensive. One should a) use methods designed to deal with in incompatible signals (e.g. the much-underused neighbour-net); b) explore and visualise the signal conflict behind the low branch supports (e.g. using bootstrap consensus networks) We had last year some posts on Genealogical World of Phylogenetic Networks regarding dinosaur matrices: Networks, not trees, identify "weak spots" More non-treelike data forced into trees