What if two bacteria assigned to the same microbial group are not functionally interchangeable? This study shows that hidden genetic differences within leaf-associated species can track differences in growth and in how Arabidopsis responds.

Plant microbiota modulates plant health and appears as a promising lever for innovative, sustainable, and eco-friendly agro-ecosystems. But understanding microbiota functionality is still in its infancy. In Arabidopsis thaliana, collections often include one strain per OTU from limited habitats, neglecting the ecological potential of genetic diversity within microbial species.

This study therefore estimated genetic variation between strains within abundant and prevalent non-pathogenic leaf-associated bacterial species in Arabidopsis thaliana from south-west France. The study isolated thirty-five pure strains from eight non-pathogenic bacterial species and detected intraspecific genetic variation at the genomic level, in growth rate, and in host responses.

Most Arabidopsis thaliana microbial collections used one representative strain per Operational Taxonomic Unit, or OTU, and sampled very limited agricultural and natural sites. That design neglects the ecological potential of genetic diversity within microbial species to strongly impact plant-microbiota interactions.

By contrast, extensive genetic variation has been detected for virulence among strains of major bacterial species in the Arabidopsis pathobiota, alongside large genomic diversity in Pseudomonas syringae and Xanthomonas arboricola. The goal was to establish an informative collection of several isolates from the twelve most abundant and prevalent non-pathogenic leaf-associated bacterial OTUs among one hundred and sixty-three natural Arabidopsis populations in south-west France.

The study focused on the phyllosphere, because leaves provide complex natural habitats with microscale variation and diverse nutrients, while phyllosphere bacteria can benefit hosts through growth-promoting hormones and protection against pathogens. The collection combined a community-based culture approach based on gyrB amplification with specific culture media and primers for each OTU.

Using twenty-two representative strains, the study examined genomic variation, in vitro growth kinetics, and genotype-by-genotype interactions with eight Arabidopsis genotypes at seed and seedling stages. The starting survey characterized bacterial communities across one hundred and sixty-three natural Arabidopsis populations using a metabarcoding approach based on part of the gyrB gene.

The survey included eight hundred and twenty-one rosettes collected in autumn twenty fourteen and spring twenty fifteen, and estimated the relative abundance of six thousand six hundred and twenty-seven abundant OTUs. OTUs had to occur in more than five percent of rosettes and exceed a mean relative abundance of zero point seven percent, producing thirteen leaf OTUs.

OTU8 was removed because it corresponded to Pseudomonas viridiflava, for which seventy-four isolated strains were confirmed pathogenic, leaving twelve OTUs. The culture-based screen produced fourteen million nine hundred fifty-six thousand nine hundred twenty-seven reads from seven thousand sixty-six sequenced colonies.

DADA2 identified six thousand three hundred sixty-two amplicon sequence variants and assigned them to the closest taxon using a curated gyrB database containing thirty-eight thousand nine hundred twenty-nine sequences. To isolate pure strains from five hundred five identified samples, the study combined selective media containing specific antibiotics with genus-specific primers.

For OTUs lacking published specific primers, primers were designed from gyrB sequences. This approach retrieved three pure strains for OTU2, seven for OTU3, eight for OTU5, six for OTU6, and eight for OTU29. Combining the community-based culture and informative-driven approaches produced thirty-five pure strains for seven OTUs: OTU2, OTU3, OTU4, OTU5, OTU6, OTU13, and OTU29.

To estimate Arabidopsis genetic variation in response to each of twenty-two strains, seven accessions were chosen to represent genetic and habitat diversity among the one hundred and sixty-three natural populations. The reference accession Col-0 was also included in the in vitro experiments.

The eight accessions were inoculated at both seed and seedling stages to test whether responses depended on developmental stage. Plant growth was manually scored from images on scales ranging from one to seven for seed-stage inoculation and from one to eight for seedling-stage inoculation.

Each experiment phenotyped two thousand three hundred four plants, and no disease symptoms were observed in the growing conditions. Figure one links the most prevalent and abundant leaf bacterial OTUs to their evolutionary relationships and colony appearance. Panel A shows a gyrB-based phylogenetic tree for thirty abundant OTUs, with twelve candidates underlined in blue and OTU8, corresponding to Pseudomonas viridiflava, marked in red because seventy-four isolated strains were confirmed pathogenic.

Panel B shows the colony morphologies of seven OTUs with representative strains, illustrating the diversity the authors then investigated experimentally. PacBio technology generated de novo genome sequences for twenty-two strains selected to maximize geographic diversity within each OTU.

Strong genomic differences occurred among the seven OTUs, including chromosome sizes from approximately three point three megabases for OTU4 to approximately seven point seven megabases for one OTU3 strain. Plasmid number and size differed among OTUs and, in some OTUs, among strains, so total genome size was not strictly related to chromosome size.

Gene numbers ranged from approximately three thousand seven hundred twenty-three for OTU29 to approximately eight thousand four hundred twenty-six for OTU2, and gene number was strictly correlated with total genome size. Strong variation in gene content also appeared among strains within a single OTU.

Among six OTU6 Pseudomonas siliginis strains, five thousand one hundred eighty-five orthogroups were identified; eighty-three point one percent were shared among all six strains, while eight point eight percent were specific to one or two strains. Figure two compares orthogroup content across six strains of OTU6 Pseudomonas siliginis.

It identifies five thousand one hundred eighty-five orthogroups in total, including four thousand three hundred ten shared by all six strains, while the stacked bar summarizes groups shared by subsets of strains or unique to one strain. The authors use this visual to show that substantial gene-content variation exists even within a single bacterial OTU.

Growth kinetics were monitored for twenty-two representative strains in R2A minimal medium and TSB rich medium, which provided contrasting nutrient availability. Across forty hours, growth kinetics were more diverse among the eight OTUs in rich medium than in minimal medium.

In rich medium, OTUs 2, 5, and 6 reached larger population sizes on average, while the two OTU13 strains grew very slowly. Within-OTU variation depended strongly on the combination of medium and OTU identity: little variation appeared for OTUs 6 and 5, while variation appeared for OTUs 3a, 3b, and 13.

Figure three tracks absorbance at six hundred nanometers over forty hours, comparing growth kinetics across eight OTUs on minimal R2A medium and rich TSB medium. The lower panels then show replicate strains within OTUs six, three-a, three-b, and twenty-nine, revealing how much growth trajectories can vary even within the same OTU.

This matters because the authors use these contrasting nutrient conditions to characterize both between-OTU diversity and within-OTU genetic variation in bacterial growth. Vegetative-growth scoring revealed significant genetic variation among the eight Arabidopsis accessions in response to twelve of the twenty-two bacterial strains tested.

More specifically, significant variation occurred for four strains when plants were inoculated at the seed stage, five strains at the seedling stage, and three strains at both developmental stages. Figure four shows how eight Arabidopsis accessions responded to twenty-two bacterial strains after inoculation at the seed stage.

Panels A and B document the forty-eight-well assay and the plant-development scoring scale from one to seven, while panel C uses a clustered heatmap to display accession-by-strain variation in genotypic values. This matters because significant genetic variation appeared for twelve of the twenty-two strains overall, including accession-specific effects such as the negative response of MONTG-D-2 and MONTB-A-10 to OTU6 Psi_1.

Figure five shows how eight Arabidopsis accessions responded to twenty-two bacterial strains twenty-one days after seedling inoculation. Panel A shows the plants in twenty-four-well plates, while panel B translates visible development into a score from one, very small or unhealthy, to eight, well-grown or healthy.

Panel C uses a heatmap with hierarchical clustering to display accession–strain response patterns, and its histogram summarizes genotypic values from four to seven point five, highlighting genetic variation in these interactions. The key message is that one representative bacterial strain can miss important ecological variation: strains within the same OTU differed genetically and in growth, while plant responses varied with both bacterial strain and inoculation stage.