Table 1. Estimated parameters under different codon substitution models for OXTR, AVPR1a, AVPR1b, and AVPR2.

	Model	dN/dS	Estimated parameters	ℓ	p value
AVPR1a	M7: β	0.1504	[p=0.22139, q=1.22306]	-10046.63	M7 vs M8 p < 0.001
	M8: β&ω	0.1399	p0=0.95125, (p2 =0.04875), ω2=1.04782	-10036.96
AVPR1b	M7: β	0.1723	[p=0.33171, q=1.56124]	-10988.96	M7 vs M8 p < 0.001
	M8: β&ω	0.1709	p0=0.96686, (p2 =0.03314), ω2 =1.39536	-10973.98
AVPR2	M7: β	0.1123	[p=0.17525, q=1.33373]	-5496.23	M7 vs M8 p = 0.04
	M8: β&ω	0.1053	p0=0.96649, (p2 =0.03351), ω2 =1.00000	-5493.606
OXTR	M1a: Nearly	0.0896	p0=0.93893, (p1=0.06107)	-4401.07
	Neutral
			(ω0=0.03039), (ω1=1.00000)		M1a vs M2a p > 0.999
	M2a: Selection	0.0896	p0=0.93893, p1=0.06107, (p2=0.00000)	-4401.07
			(ω0=0.03039), (ω1=1.00000), ω2=36.81273

p0 = proportion of sites where ω < 1, p1 = proportion of sites where ω = 1, and p2 = proportion of sites where ω > 1 (selection models only); ω0 < 1 (negative selection), ω1 ≅ 1 (neutral or relaxing selection), and ω2 > 1 (positive selection). ℓ= Log likelihood values. Likelihood ratio tests were performed between neutral models (M1a- nearly neutral, and M7 - beta) and models that identify positive selection (M2a - selection, and M8, β&ω - beta +selection). The comparisons M1 vs M2 and M7 vs M8 had 2 degrees of freedom. Within parentheses: fixed parameters; within brackets: βparameters p and q. dN/dS = non-synonymous/synonymous rate ratio.