Updated Erdman reveals tandem repeat copy number is phase-variable and impacts M. tuberculosis adaptation across evolutionary timescales

High-quality reference genomes are essential for comparative genomics and accurate genotype-phenotype mapping. Here, we corrected the Mycobacterium tuberculosis Erdman strain reference genome (Erdman TI ) using ultra-deep HiFi sequencing. Among the small variants ( n = 275) between Erdman TI and the current Erdman reference NC_020559.1 (Erdman STJ ), numerous are likely errors in Erdman STJ . We identified a novel bias toward in-frame structural variations (SVs) in pe/ppe genes and 28 SVs between Erdman TI and Erdman STJ , half representing likely errors in Erdman STJ . Other SVs were consistent with in vitro evolution, including copy number variation (CNV) of promoter tandem repeats (PTRs). PTR CNVs were polyphyletic and within isogenic populations (10 −2 –10 −3 CNVs/chromosome), demonstrating the impact of phase-variable CNV across evolutionary timescales. These hypervariable PTRs pinpoint a genomic basis for rapidly switching nitric oxide resistance (Dop), biofilm formation (LpdA), drug tolerance (EfpA), and glycerol utilization (GlpD2) phenotypes. This work uncovers a common phase variation mechanism obscured by short-read sequencing limitations and provides an improved reference for comparative studies. IMPORTANCE Mycobacterium tuberculosis ( Mtb ), the pathogen responsible for tuberculosis, is often described as genetically stable. Our findings reveal an overlooked evolutionary adaptation mechanism: phase variation driven by tandem repeat copy number changes in gene promoters. Enabled by ultra-deep, long-read sequencing, we corrected errors in the Erdman reference genome and uncovered frequent, spontaneous expansions and contractions of promoter repeats upstream of genes linked to nitric oxide resistance, drug efflux, and biofilm formation. Through altering promoter strength, these dynamic promoter variants may generate phenotypic diversity within subpopulations and across diverse clinical lineages, suggesting a conserved evolutionary advantage for navigating host-imposed stress. This reframes Mtb ’s evolutionary potential, highlighting how adaptive flexibility has been underestimated due to reliance on short-read sequencing and limited resolution of subpopulations at standard genomic depths. Our findings underscore the need to integrate structural variation-aware approaches into studies of Mtb pathogenesis, evolution, and drug response.