Table 2.

The Kishino-Hasegawa topological test results

	Single gene tree	Concatenated 4 genes tree
	lnL	lnL	pKH
asl	-3561.406	-3642.361	< 0.001*
clpc	-7718.852	-7728.157	0.2696
coI	-10062.16	-10074.33	0.2336
cytb	-6856.617	-6859.401	0.4323

For each gene the likelihood of the phylogeny of that gene was compared to the phylogenetic reconstruction based on all four genes. The log-likelihood values and p-values are shown per gene alignment. Only the asl alignment gives a significantly worse likelihood value for the tree based on the combined data, which indicates conflicting topologies.

For each of the three hypotheses on the Polychromophilus origin a corresponding topology was constructed. This was done by restricting the placing of Polychromophilus during tree reconstruction in RAxML, forcing it either with the mammal-infecting Plasmodium/Hepatocystis clade (hypothesis 1), the Haemoproteus (Haemoproteus) clade (hypothesis 2) or with the sauropsid Plasmodium clade (hypothesis 3). These restricted topologies were then tested together with the topology produced by the maximum likelihood analysis using a Shimodaira-Hasegawa test [32] as implemented in PAML 4 [33].