Fig. 1. A in "A character does not make a genus, but the genus makes the character ": three-taxon statement analysis and intuitive taxonomy

Fig. 1. A. Strict consensus of 61 most parsimonious phylogenetic trees; tree length = 90 steps; CI = 0.6000; RI = 0.8302, recovered from a standard cladistic analysis (MP) (Fitch Parsimony) of the complete conventionalmorphologicalmatrixof Atraphaxis s. l. (TableS1).All 27 unordered (non-additive) characters are parsimony informative. B. Strict consensus of two nested most parsimonious hierarchies of patterns; length = 8328 steps; CI = 0.8521; RI = 0.8264), recovered from a MP analysis of the three-taxon statement representation of the complete conventional 27 characters' morphological matrix of Atraphaxis s. l. (Table S1). The number of 3TSs (characters) is equal to 7096, all are parsimony-informative. C. Strict consensus of six most parsimonious phylogenetic trees; tree length = 56 steps; CI = 0.6071; RI = 0.8370, recovered from a MP analysis of the reduced conventional 18 characters' morphological matrix of Atraphaxis s. l. (Table S1) with the characters one, two, four, eight, 12, 16, 18, 19, and 26 excluded. D. Strict consensus of two nested, most parsimonious hierarchies of patterns; length = 5844 steps; CI = 0.8665; RI = 0.8460), recovered from a MP analysis of the three-taxon statement representation of the reduced conventional 18 characters' morphological matrix of Atraphaxis s. l. (Table S1) with characters 1, 2, 4, 8, 12, 16, 18, 19 and 26 excluded. The number of 3TSs (characters) is equal to 5064, all are parsimony-informative. All MP analyses as in PAUP* 4.0a150 (Swofford 2002) were conducted using either conventional matrices or TAXODIUM's output 3TS NEXUS files with a heuristic search of 1000 random addition replicates (saving no more than 100 trees per replicate), and the TBR branch swapping/MulTrees option into effect. Branches with a minimum length of zero were collapsed. The three-taxon statement analysis (3TA) of the unordered morphological matrix was established after it three-taxon (3TS) Williams-Siebert (WS) representation (Williams & Siebert 2000) using TAXODIUM v. 1.2 (Mavrodiev & Madorsky 2012). The 3TS permutations were performed with the following command: taxodium input_file_name. csv –ium –ob –og –nex The value of the operational outgroup was fixed as a value of Bactria lazkovii. All 3TSs were weighted uniformly and treated as ''ordered'' (Wagner Parsimony). The bootstrap resampling of both conventional and 3TS matrices have been performed as described in Mavrodiev & Madorsky (2012). The diagnostic traits are optimized using Mesquite (Maddison & Maddison 2011).