Skip to main content
. 2016 Feb 17;116(6):506–515. doi: 10.1038/hdy.2016.8

Table 2. CODEML model comparisons and associated hypothesis that is being tested are described.

Model comparison Log ratio test Testing for
Model 0 vs Model 1a χ2=40.04, P<0.05 Non-neutrality
Model 1a vs Model 2a χ2=0, NS Positive selection
Model M7 vs Model M8 χ2=1.43, NS Positive selection
Model 0 vs branch model 2 χ2=20.55, P<0.001 Higher ω ratios in pre-specified foreground branches (coastal branches here)
MAnull vs branch-sites-model-A-modified χ2=2327.9, P<0.001 Positive selection or relaxation of selective constraint affecting certain sites in pre-specified foreground branches (coastal branches here)
Model 1a vs branch-sites-model-A-modified χ2=18.99, P<0.001 Positive selection (if ω 2a and/or ω2b>1 in foreground), or relaxation of selective constraint (if ω 2a and/or ω2b⩽1 in foreground) affecting certain sites in pre-specified foreground branches (coastal branches here)

Bold font indicates a significant (P<0.05) log ratio test and support for the stated hypothesis.