Genomic codon usage is structurally consistent with First-Classness across the tree of life

This is the Author Accepted Manuscript (AAM) of an article published by Springer in the Journal of Molecular Evolution on September 19, 2026. The final authenticated publisher's version is available online at: https://doi.org/10.1007/s00239-026-10343-2 This record contains the main manuscript, Genomic codon usage is structurally consistent with First-Classness across the tree of life. This version is the final revised manuscript prepared in response to editorial and peer-review comments on the submitted paper. The study tests whether genome-scale codon-usage data are structurally consistent with the 21-family genetic-code partition derived independently from the First-Classness (FC) framework. The analysis uses codon counts from 70,950 RefSeq species spanning the tree of life. Three principal empirical results are reported. First, within-family codon-usage dispersion follows a reproducible non-monotonic ordering across the FC-derived family classes: 2-fold < 4-fold < Ser 6-fold < 3-fold < Leu/Arg 6-fold. Second, the three split six-fold families—serine, leucine and arginine—show persistent four-codon sub-block dominance in bacteria, archaea and eukarya, although the magnitude of that dominance varies by lineage. Third, a family-balanced permutation test detects no partition-specific global signal ((Z=0.06,\ p=0.44)); the paper therefore treats this result cautiously and discusses both limited statistical sensitivity and the FC non-separability hypothesis as possible interpretations. The revised manuscript places these structural findings explicitly within the established biology of synonymous codon usage. It discusses translational selection for both efficiency and accuracy, tRNA abundance and adaptation, wobble pairing, anticodon modification, and inosine at anticodon position 34. Particular attention is given to the TAPSLIVR amino-acid group and its relation to the FC partition. The eight eukaryotic I34-bearing tRNAs map onto eight of the nine FC spacelet boxes, with glycine as the sole exception; in the split six-fold families, I34 decoding occurs within the spacelet component rather than the complementary half-spacelet component. Domain-resolved analysis shows that four-codon sub-block dominance persists even in archaea, which lack I34 tRNAs. The manuscript therefore does not attribute the dominance generally to inosine-mediated decoding; rather, the observed concordance is interpreted as consistent with decoding mechanisms operating within boundaries already present in the genetic-code partition. The paper also compares the FC derivation with recent biochemical and prebiotic accounts of genetic-code architecture, including the codon-redundancy rules of Harrison et al. (2022) and the anticodon-base interaction evidence of Halpern et al. (2023). These approaches are presented as complementary: biochemical accounts address how coding relationships may arise physically, whereas FC addresses the formal architecture of the resulting codon-family partition. The manuscript does not claim that codon-usage statistics prove FC uniquely correct or establish a causal biological mechanism. Its claim is narrower: the large-scale genomic data display structural regularities that are consistent with the independently derived FC partition, while known biological mechanisms determine and modify codon usage within that architecture. The associated Supplementary Information and reproducibility archive provide the extended robustness analyses, domain-stratified results, computational verification of the FC-derived family partition, and executable reproduction workflow. The present record is intended primarily as the archival record of the biological paper itself.