Table 2.

Incongruence between phylogeny and classifications derived from morphological studies and the molecular evidence.

	# of taxa in constrained clade	SH (p value)	Templeton (p value)
Kailola (2004)
Phylogeny	48 (34 constrained nodes)	0.000	< 0.0001
Ariopsis	20	0.000	< 0.0001
Arius	7	0.411	0.72–0.88
Aspistor	11	0.000	< 0.0001
Cephalocassis	2	0.000	< 0.0001
Cinetodus	3	0.000	< 0.0001
Hemiarius	4	0.000	< 0.0001
Hexanematichthys	2	0.027	< 0.01
Nemapteryx	5	0.000	< 0.001
Netuma	5	0.018	< 0.01
Sciades	8	0.000	< 0.0001
Marceniuk and Menezes (2007)
Arius	10	0.004	< 0.04
Arius excluding A. madagascariensis	9	0.341	0.59–0.77
Brustiarius	3	0.000	< 0.0001
Cephalocassis	2	0.000	< 0.0001
Cinetodus	2	0.153	< 0.08*
Cochlefelis	3	0.355	0.31–0.41
Neoarius	6	0.132	0.18–0.25
Notarius	13	0.013	< 0.03
Potamosilurus	4	0.014	0.09–0.13
Sciades	18	0.000	< 0.0001

Results obtained with Templeton and Shimodaira-Hasegawa (SH) tests of topology congruence between trees constrained under the morphological hypotheses and unconstrained trees (significant p values in bold). Hypothesis testing was performed on Kailola's phylogeny [15] (see Figure 2A) and those genera defined by Kailola [15] and Marceniuk and Menezes [21] that were recovered as non-monophyletic (see also Figure 3). For generic comparisons, only one node was constrained. Taxa in constrained clade include common species only; however, for some comparisons the number of taxa with enforced monophyly is greater than the number of taxa assigned to a particular genus in previous studies due to the greater number of species recognized here (e.g., affinis entities).

*In 52 out of 56 comparisons p value < 0.05