Model comparisons and parameter estimates under models of constant (H₀) or variable <i>ω</i> rate ratios across branches (Br), clades (H₁ to H₃) or both clades and sites (A and B) for the θ subunit.

Notes–ℓ: log-likelihood value; HA: alternative hypothesis to the current model; ω0: background rate; ω1–5: branch/clade specific rates as indicated in Figure 4; p0–2: proportions of sites in each rate category. The hat notation indicates parameters that are free to vary. Positively selected sites were identified with BEB and NEB between brackets for model A and NEB only for model B; a 99% cut-off level of posterior probability was used. Sites putatively under positive selection are numbered according to the human θ reference sequence (accession number BC109210/AAI09211). The underlined sites are those common to both model A and model B. Curly braces indicate equally parsimonious ancestral sites.