That, or I really need to place more blind trust in my mathematical intuition. Thus spake Red Team Grok: The ~1,580–1,590 generations per fixation (pooled, strict whole-population definition) is the appropriate figure for “normal” LTEE populations. That is the rate observed in populations that still have intact DNA-repair systems.
The all-population rate of ~132 generations per fixation is real, but it is dominated by the seven hypermutator populations. Those populations have broken mismatch repair (or oxidative-damage repair), are accumulating mutations at ~100× the ancestral rate, show the genomic decay documented by Couce et al., and in several cases exhibit intervals of net loss of high-frequency alleles. They are not healthy, steady-state evolving populations; they are populations that have traded genome maintenance for short-term mutational supply. Calling their throughput “the fastest experimental rate ever measured” without strong qualification is therefore misleading if the intended comparison is to ordinary evolutionary processes.
This is why the source papers (and your own earlier summaries) routinely separated or excluded the hypermutators when stating a representative rate. The corrected data do not change that logic; they simply make the separation cleaner.
Recommended reporting:
- Primary: Non-mutator pooled rate ≈ 1,587 generations per whole-population fixation (5 populations, 189 fixations).
Secondary / completeness: When hypermutators are included, the overall rate falls to ~132 generations per fixation (12 populations, 5,496 fixations). This figure is driven by populations with defective DNA repair and should not be treated as a general benchmark for intact systems. - That framing is accurate, consistent with prior literature practice, and avoids inflating the apparent speed of ordinary fixation by averaging in genomic-cancer lineages.
The data tables and the analysis are both available at Zenodo: Fixation in the LTEE: Whole-Population Fixation Counts at 500-Generation Resolution