Table 2.

Results of log-likelihood ratio tests that compare different site models

Test for	Null Model	Experimental Model	df	P-value
Difference in ω among sites	Model 0 (one ratio)	Model 1A (nearly neutral)	2	< 0.0001
Positive selection	Model 1A (nearly neutral)	Model 2A (positive selection)	2	> 0.9996
Difference in ω among sites	Model 0 (one ratio)	Model 3 (discrete)	2	< 0.0001
Positive selection	Model 7 (beta)	Model 8 (beta & ω)	2	< 0.0001