TABLE 2.

Estimates of substitution reconstructions used in Maximum-likelihood and Bayesian inference analyses.

	Maximum likelihood parameter estimates	95% credible interval of Bayesian parameter samples
			Total ESS
Refinement	GTR + Γ + I	codon + GTR	NI
Llk	−179,358,746	NI	NI
AIC	373,717,492	NI	NI
TL	NI	0.20176–0.27790	4511
r(A<->C)	1.68181	0.15769–0.28965	2379
r(A<->G)	1.94162	0.20164–0.33773	1810
r(A<->T)	0.77428	0.04246–0.13191	2588
r(C<->G)	0.81229	0.07187–0.17619	2573
r(C<->T)	1.21419	0.09647–0.23001	1760
r(G<->T)	1.00000	0.07639–0.19743	2630
pi(A)	0.28822	NI	NI
pi(C)	0.26051	NI	NI
pi(G)	0.25992	NI	NI
pi(T)	0.19135	NI	NI
Alpha	0.72500	NI	NI
Pinvar	0.53100	NI	NI

Maximum-likelihood estimates for the selected reconstruction were obtained using the ML tree. Credible intervals were obtained from the stationary distribution. Abreviations: Llk, likelihood scores; AIC, Akaike Information Criterion; TL, tree length; r, reversible substitution rate between indicated bases; pi, stationary frequency of indicated base; alpha, shape parameter of the gamma distribution of rate variation; pinvar, proportion of invariable sites; ESS, estimated sample size; NI, parameter not included in selected refinement. For a complete description of the total stationary frequency of codons and the Effective Sample Size (ESS) of Bayesian analyses see Supplementary Table S7.