Table 1.

Log-likelihood scores (l) and parameter estimates under several models for ω estimates in multiple PDE1Cb genes of stickleback

Models	la	Average ωb	ML-estimated parameters
Site-specific model
LRT-1
One ω ratio model (M0)	-5866.90	0.1395	P0 = 1.0000, ω0 = 0.1395
Discrete model (M3)	-5723.25	0.3937	P0 = 0.7798, P1 = 0.2063, P2 = 0.0139, ω0 = 0.0543, ω1 = 0.5031, ω2 = 17.8270
-2Δl [d.f. = 4]	287.29 *
LRT-2
Neutral model (M1)	-5906.08	0.4735	P0 = 0.5265, P1 = 0.4735, ω0 = 0.0000, ω1 = 1.0000
Selection model (M2)	-5743.66	0.1743	P0 = 0.3524, P1 = 0.0911, P2 = 0.5565, ω0 = 0.0000, ω1 = 1.0000, ω2 = 0.1495
-2Δl [d.f. = 2]	324.84 *
LRT-3
Beta model (M7)	-5763.28	0.1814	p = 0.3278, q = 1.3121, P0 = P1 = P2 = 0.3333, ω0 = 0.0030, ω1 = 0.0866, ω2 = 0.4548
Beta and ω model (M8)	-5763.28	0.1814	p = 0.3278, q = 1.3121, P0 = P1 = P2 = 0.3333, P3 = 0.0000, ω0 = 0.0028, ω1 = 0.0866, ω2 = 0.4548, ω3 = 0.3544
-2Δl[d.f. = 2]	0.00
Branch-site model
Selection model (M2)	-5833.54	N.A.	P0 = 0.0939, P1 = 0.0652, P2 = 0.8409, ω0 = 0.0000, ω1 = 1.0000, ω2 = 0.0695
Discrete model (M3)	-5741.70	N.A.	P0 = 0.3084, P1 = 0.0562, P2 = 0.6354, ω0 = 0.0634, ω1 = 0.8749, ω2 = 0.0796

^a Log-likelihood scores.

^b Nonsynonymous–synonymous substitution ratio (ω = d_N/d_S) averaged over sites.

* p < 0.001