Convergent Architecture of the Acinetobacter baumannii Resistome: A Co-Occurrence Network Analysis of 20,739 Genomes Under the One Health Framework

Acinetobacter baumannii is a critical priority ESKAPE pathogen whose resistome co-occurrence architecture, temporal dynamics, and One Health distribution remain poorly characterized. A total of 20,739 genomes (2000–2025; NCBI, PubMLST, BV-BRC) were annotated using CARD-RGI v6.0.5, AMRFinderPlus v4.2.7, and ResFinder v4.7.2. A co-occurrence network was constructed using Jaccard filtering (≥10%; >0.30) with Bayesian bootstrap and Louvain. Temporal trends were assessed by linear regression with Benjamini–Hochberg (2003–2024), and One Health compartments were assessed using balanced PERMANOVA (Bray–Curtis, 999 permutations). The network comprised 41 nodes and 82 edges with a small-world topology; five communities and nine hub genes were identified. Eighteen ARGs showed significant trends (FDR); six last-resort determinants (blaOXA-23-like, blaNDM, armA, ftsI, msr(E), mph(E)) increased in prevalence. The compartment explained significant variance in the resistome (R2 = 0.210; p < 0.001). abaF showed the highest One Health convergence (0.814), and blaNDM was detected in 52 countries. The A. baumannii resistome exhibits a convergent architecture of co-resistance: a small-world network preserved in One Health compartments for more than two decades. The rise in six last-resort determinants signals a population-based transition toward broader profiles, establishing a framework for integrated antimicrobial resistance surveillance.