Pan-SSR architecture in Acrossocheilus fasciatus links cross-assembly repeat conservation, repeat-unit count...

Microsatellites, or simple sequence repeats (SSRs), are abundant genomic features, but their evolutionary interpretation is often limited by detector-dependent discovery and single-reference surveys that count repeats without resolving homologous repeat states. We characterized pan-SSR architecture in the cyprinid Acrossocheilus fasciatus by integrating four genome assemblies within the AF_xajfe coordinate system. Synthetic ground-truth calibration identified Tandem Repeats Finder as the most accurate and parameter-stable algorithm, and coordinate-anchored merging resolved 499,190 pan-SSR entries comprising 39,150 core, 235,720 dispensable, and 224,320 private components. In homologous shared components, repeat-unit count variation dominated polymorphism (26.2% of core and 22.4% of dispensable SSRs), whereas motif mutations and mixed polymorphisms each remained near 1%. Genomic context differed among categories: core SSRs were preferentially associated with genes and promoters, private SSRs showed the highest classified transposable element overlap, and DNA transposons showed the strongest local association. At 1-Mb genomic resolution, SSR density was inversely associated with GC content (Pearson r = -0.485; Spearman ρ = -0.547), indicating scale-dependent constraints on repeat accumulation. Independent resequencing of 130 libraries from eight river populations showed that neutral/core and gene-associated/core sets could be genotyped in low-depth short-read data and retained population-informative repeat-unit dosage variation, including separation of the QingYiJiang group in weighted principal-component summaries. Together, these results provide a detector-calibrated, coordinate-anchored strategy for repeat evolutionary analysis in non-model vertebrates by comparing homologous SSRs across heterogeneous public assemblies and show that repeat-unit count variation, gene- and transposable-element-associated context, GC background, and short-read genotypability jointly organize the pan-SSR landscape.