Calibrating public genome archives for AMR surveillance: Four decades of redundancy-aware phylogenomics of pandemic Escherichia coli ST131

AIMS: Escherichia coli sequence type (ST)131 is a globally disseminated multidrug-resistant pathogen, yet public genome collections are skewed by uneven sampling and repeated sequencing of genetically near-identical isolates. Rather than another reconstruction of ST131 evolution, we quantified redundancy, reassessed population structure, and tested how sampling bias affects estimates of phylogenetic diversity, clade-specific dating, and fluoroquinolone-resistance-associated diversification. METHODS: EnteroBase core-genome multilocus sequence typing (cgMLST) hierarchical clustering was used to reduce redundancy among 16,713 public ST131 complex genomes. Representative public genomes were combined with 390 unpublished New Zealand ST131 surveillance genomes from 1999 to 2019. Clade-specific phylogenetic and temporal analyses were performed for Clades A, B, and C, with sensitivity analyses using stricter cgMLST clustering and phylogenetic-diversity-guided subsampling. RESULTS: Public genomes were highly redundant: just 2/409 cgMLST groups contained 85.2% of public genomes. After redundancy reduction, dated Clades A, B, and C phylogenies spanning 1999-2023 (n = 113), 1982-2025 (n = 240), and 1999-2025 (n = 614), respectively, were constructed. Clades A and B showed distinct temporal histories and evolutionary rates. Date-randomization supported temporal information in Clade C, although incomplete convergence of the observed-date relaxed-clock and Bayesian Skyline analyses means that exact Clade C molecular-date and demographic estimates should be interpreted cautiously. The broad Clade C reconstruction nevertheless indicated expansion of the dominant fluoroquinolone-resistant lineage in the late 1990s/early 2000s and again in the mid-2000s. CONCLUSIONS: Public ST131 datasets reflect sampling distribution as well as number of available genomes. Redundancy in dominant cgMLST groups can obscure uncommon lineages and blur emergence with later expansion. In Clade C, redundancy-aware analysis was broadly consistent with previous estimates placing major fluoroquinolone-resistant diversification in the early 1990s, while indicating that early resistance diversification preceded later amplification of resistant sublineages.