Bradyrhizobium japonicum has two types of flagella. One has thin filaments consisting of the 33-kDa flagellins FliCI and FliCII (FliCI-II) and the other has thick filaments consisting of the 65-kDa flagellins FliC1, FliC2, FliC3, and FliC4 (FliC1-4). To investigate the roles of each flagellum in competition for nodulation, we obtained mutants deleted in fliCI-II and/or fliC1-4 in the genomic backgrounds of two derivatives from the reference strain USDA 110: the streptomycin-resistant derivative LP 3004 and its more motile derivative LP 3008. All mutations diminished swimming motility. When each mutant was co-inoculated with the parental strain on soybean plants cultivated in vermiculite either at field capacity or flooded, their competitiveness differed according to the flagellin altered. ΔfliCI-II mutants were more competitive, occupying 64-80% of the nodules, while ΔfliC1-4 mutants occupied 45-49% of the nodules. Occupation by the nonmotile double mutant decreased from 55% to 11% as the water content of the vermiculite increased from 85% to 95% field capacity to flooding. These results indicate that the influence of motility on competitiveness depended on the water status of the rooting substrate.

A Bradyrhizobium japonicum USDA 110-derived strain able to produce wider halos in soft-agar medium than its parental strain was obtained by recurrent selection. It was more chemotactic than the wild type towards mannitol and three amino acids. When cultured in minimal medium with mannitol as a single carbon-source, it had one thick subpolar flagellum as the wild type, plus several other flagella that were thinner and sinusoidal. Root adsorption and infectivity in liquid media were 50-100% higher for the selected strain, but root colonization in water-unsaturated vermiculite was similar to the wild type. A field experiment was then carried out in a soil with a naturalized population of 1.8 x 10(5) soybean-nodulating rhizobia g of soil(-1). Bradyrhizobium japonicum strains were inoculated either on the soybean seeds or in the sowing furrows. Nodule occupation was doubled when the strains were inoculated in the sowing furrows with respect to seed inoculation (significant with P<0.05). On comparing strains, nodule occupation with seed inoculation was 6% or 10% for the wild type or selected strains, respectively, without a statistically significant difference, while when inoculated in the sowing furrows, nodule occupation increased to 12% and 22%, respectively (differences significant with P<0.05).