High-resolution species delimitation in Acinetobacter baumannii using a novel Core-Gene Consensus Delimitation approach

Abstract

Acinetobacter baumannii is a highly adaptable nosocomial pathogen with extensive antibiotic resistance, a disproportionately large accessory genome, and high genomic plasticity. Owing to these features, the World Health Organisation (WHO) classifies A.baumannii as a critical-priority pathogen. In this study, we analyzed 47 isolates from our VUB (Vrije Universiteit Brussel) collection and applied distance-based species-delimitation algorithms - Automatic Barcode Gap Discovery (ABGD) and Assemble Species by Automatic Partitioning (ASAP) - for the first time at the bacterial core-genome scale. By integrating conspecificity matrices, we extended these traditionally single-locus methods into a multi-locus framework, which we term Core-Gene Consensus Delimitation (CGCD). Across a range of gene-level co-occurrence thresholds, CGCD consistently recovered 11 stable groups using both ABGD and ASAP. Larger-scale validation using 856 A. baumannii genomes recovered the same 11 well-separated groups were recovered, demonstrating the robustness and reliability of our clustering approach. Mapping these groups onto a core-genome phylogeny revealed that each group forms a distinct clade, indicating that they represent evolutionarily independent lineages rather than arbitrary clusters. We further constructed a clustering tree based on accessory gene presence-absence patterns. In this tree, only one strain (AB231-VUB) clustered within group 11; otherwise, the groups remained tightly cohesive, sharing characteristic sets of accessory genes. Together, these results show that the groups defined by CGCD are genomically, evolutionarily, and functionally distinct, supporting their interpretation as separate species. Our findings highlight CGCD as a powerful, high-resolution framework for species delimitation. CGCD is threshold-free, gene-based, and universally applicable--the first species-delimitation approach that can be applied across all domains of life, from bacteria to animals and plants.