Accounting for Gene Tree Estimation Error in Species Tree Reconstruction: Quantity Versus Quality
summary
Species-tree recovery under the multispecies coalescent has a concrete trade-off: let F be the probability a lineage coalesces inside the short branch. Methods that use topology frequency (R / most frequent) or pairwise averages (STEAC (average)) require about 1/F^2 genes; using the minimum coalescence time (GLASS (min)) needs ≈1/F genes if true times are known but with sequence noise demands per-gene length L ~ 1/F^2. A quantile-based QUEST estimator interpolates and achieves G ≈ F^-2 / sqrt(L), which implies total sequencing effort G·L scales like sqrt(L) and favors many short loci.
AI-written summary · about