Stingless bees, turtle ants and tea plants are rich sources of undescribed Lactobacillaceae

Knowledge of the family Lactobacillaceae rests largely on isolates from foods and a few repeatedly sampled hosts. Reference databases give an unrecognised sequence the name of its nearest relative, so a lineage with no entry of its own is renamed rather than flagged. Here we classify the family across the public plant and invertebrate amplicon record to locate the hosts carrying undescribed lineages. Across 3,344 independent 16S rRNA gene amplicon studies, every sequence cluster was tested against a type-strain reference and placed at the deepest rank it supports. Of 204,813 classified clusters, 20,516 were named to species and 127,906 to genus, while 4,741 matched no described species. Six published datasets whose authors could name their Lactobacillaceae only as "Lactobacillus", or not at all, are reclassified here. Genera described from one habitat occur far beyond it, three bee-associated genera occurring on Rosaceae and Brassicaceae at several times their rate on wind-pollinated grasses, and we found no published Bombilactobacillus record from a plant. Undescribed lineages concentrate in the least-cultured hosts, reaching 86.1% and 77.4% of studies in the stingless bees Melipona and Tetragonula, and are most divergent in the turtle ant Cephalotes and, among plants, in tea, Camellia. An independent genome-resolved survey of pot honey converges on the same two genera. The primary descriptions of forty-eight species from these hosts specify a supplemented medium, so the hosts carrying undescribed lineages also indicate how to culture them.