6S rRNA phylogeny outperforms similarity thresholds for species identification and reveals genus-dependent underestimation of Lactobacillales diversity

Species identification underpins biological research, yet bacterial identification remains challenging because commonly used approaches do not fully reflect evolutionary relationships. While the 16S rRNA gene is the gold standard, its limited resolution restricts species-level discrimination among closely related taxa. Here, we show that phylogenetic approaches based on model-based algorithms, such as Maximum Likelihood, provide a stronger framework for bacterial identification. By analysing 479 strains of the industrially and clinically relevant order Lactobacillales isolated from Colombian-native fermented products, we successfully identified more than 50% of the strains at the species level under a phylogenetic framework, corresponding to 41 species across 15 genera. Our results demonstrate that while analysis of the 16S rRNA gene can lead to a lineage-dependent underestimation of species richness, phylogenetic reconstruction based on explicit evolutionary modelling outperforms database approaches such as SILVA and BLASTn. Furthermore, we show that Oxford Nanopore Technologies provides a cost-efficient and accessible solution for in-house laboratory settings. These findings establish a high-resolution, affordable phylogenetic framework that enhances species-level identification in underexplored microbial habitats.