Table 3.

Selection of topology models (tests of phylogenetic hypotheses) for miRNA datasets based on Bayes factor comparisons of estimated marginal likelihoods

miRNA study	Topology model*	ln P(X|Mi)†	$2\ln B{F}_{ij}$‡	Description	Resulting from§
Amniotes	M0	−114.98 (0.03)	17.52	Lepidosaur hypothesis: turtles + lizards	B and P
	M1	−123.74 (0.14)		Archosaur hypothesis: turtles + archosaurs
Amniotes-corrected¶	M0	−126.21 (0.22)	5.17	Archosaur hypothesis: turtles + archosaurs	B and P
	M1	−128.80 (0.08)		Lepidosaur hypothesis: turtles + lizards
Animals	M0	−574.67 (0.44)	−11.69	(((Acoel 1, Acoel 2), Xenoturbella), (remaining Bilateria))	B
	M1	−568.83 (0.17)		(Acoel 1 (Acoel 2 (Xenoturbella (remaining Bilateria)))	P
Annelids	M0	−414.65 (0.11)	11.97	((Phascolosoma, Nereis), (Lumbricus, Capitella))	B
	M1	−420.64 (0.23)		(Phascolosoma (Nereis (Lumbricus, Capitella)))	P
Bilaterians	M0	−552.63 (0.02)	100.92	Myzostomids sister to annelids	B
	M1	−603.09 (0.26)		Myzostomids nested within annelids	P
Vertebrates	M0	−994.81 (0.14)	0.91	Cyclostome hypothesis: lampreys sister to hagfish	B and P
	M1	−995.26 (0.13)		Jawed vertebrate hypothesis: lampreys sister to jawed vertebrates

^*Topology models refer to various phylogenetic hypotheses corresponding to the description column; see text for details.

^†The marginal log probability of miRNA datasets (and SE) under the stochastic Dollo model and the preferred relaxed-clock model estimated using path sampling as described in the text.

^‡Two times the natural log of the Bayes factor is twice the difference between the natural log marginal likelihoods estimated under the alternative topological models. Our interpretation follows ref. 46.

^§An indicator for which unconstrained analysis type recovered each topological model. B, Bayesian stochastic Dollo; P, Parsimony.

^¶This is a version of the amniote miRNA dataset from the study of Lyson et al. (19) that has been corrected for sampling error; see text for details.