Table 4.

Test of alternative phylogenetic hypotheses for continental cichlid groups

Topological constraint	pKH	pSH	2 ln Bayes factor
Best as in Fig. 1	1.000	1.000
Constraint 1: monophyly of Madagascar and Indo/Sri Lanka (Tree 1)	0.006**	0.043*	65.4*
Constraint 2: monophyly of Africa, Madagascar and Indo/Sri Lanka (Tree 2)	0.001**	0.002**	125.1*
Constraint 3: monophyly of Africa and Indo/Sri Lanka (Tree 3)	0.000**	0.000**	297.2*

Probabilities for constrained trees were assessed using the Kishino-Hasegawa (pKH) and Shimodaira-Hasegawa (pSH) tests and the Bayes factor. Single asterisks indicate significant rejection (p < 0.05) and double asterisks indicate highly significant rejection (p < 0.01) of the corresponding hypothesis. We used the traditional criterion of 2 ln Bayes factor over 10 (with an asterisk), indicating very strong evidence against an alternative hypothesis [39]. Constrained trees are the following: Tree 1: ((((Oreochromis sp., (Tropheus duboisi, Neolamprologus brichardi)), Tylochromis polylepis),(Astronotus ocellatus, Hypselecara temporalis)),((Etroplus maculatus, Paretroplus maculatus),(Ptychochromoides katria, Paratilapia polleni))); Tree 2: ((((Oreochromis sp.,(Tropheus duboisi, Neolamprologus brichardi)), Tylochromis polylepis),((Etroplus maculatus, Paretroplus maculatus),(Ptychochromoides katria, Paratilapia polleni))),(Astronotus ocellatus, Hypselecara temporalis)); and Tree 3: (((((Oreochromis sp.,(Tropheus duboisi, Neolamprologus brichardi)), Tylochromis polylepis), Etroplus maculatus),((Astronotus ocellatus, Hypselecara temporalis),(Ptychochromoides katria, Paratilapia polleni))), Paretroplus maculatus).