Lack of ongoing evolution in ΔGY-controllers. (A) Linear regression analyzed the pairwise distance (p-distance) as a function of time in controllers. Regressions were fit from peak viral load of each animal to week 66. (B) Segmental linear regression analyzed the p-distance as a function of time in non-controllers. Regressions were fit from day 14 to week 66, with transition point constrained to ART initiation, represented by a dotted red line. Defective proviruses were excluded. p-distance was calculated using the maximum composite likelihood model. p-distance of ∼0.0004 corresponds to a difference of 1 nucleotide. Each point is the mean p-distance for a time point (n = 1–57). An extra-sum-of-squares F test determined whether slope was significantly different from 0. Slopes and P-values are indicated under the respective graph. The gray-shaded area bounded by dotted lines is the 95% CI.