Bacterial richness within samples is higher for fields under organic management
To assess the effect of field management on Rhizobium diversity, we analysed the genospecies composition of each individual sample. The amplicons from all four genes were assigned to genospecies A-E (Figure S2 ) . nodA and nodD amplicons were assigned as X if they were within an introgressing clade of plasmid-borne sym genes (Figure S2C-D ) . The genospecies composition was plotted for each individual sample of 100 nodules (Figure 5 ). At two of the trial sites (F and UK), only core genes from a single genospecies were detected, gsC and gsB, respectively (Figure 5A and B ), though several alleles of the same genospecies were present (Figures 2 and 5 ). The DK trial site had low levels of gsE, gsD, and gsB, while the dominant clade was gsC. For the DKO samples, core genes belonging to all genospecies clades were found, and most samples contained several genospecies at intermediate frequencies. The raw UMI count was not higher for the DKO samples, indicating that the increase in observed genospecies is not due to sequencing bias (Figure S3 ). The diversity of core genes,rpoB and recA , for the F and UK trial sites matched the genospecies distribution of isolates from these fields. For the DK site, low levels of core and accessory gsB amplicons were found, although no gsB strains were isolated from here. The local gs distribution matches the observations from our previous study of isolates from these sites , indicating that the sampling of nodule isolates from fields can yield a good approximation of the actual rhizobium population present in the nodules (Figure 5 ).
Overall, more genetic diversity was observed within the samples for the accessory genes than for core genes for most sampling sites, predominantly due to the introgressing clade, X. Furthermore, most samples from the three trial sites contained low levels of nodgene amplicons that were not assigned to the core gene genospecies observed in the samples (Figure 5E and G ). Most of these belong to the introgressing clade, X, which is consistent with previous observations on large Rlt populations .
Some nod gene amplicons do not match the core gene composition. The French samples are exclusively gsC for the core genes, rpoB andrecA , whereas the UK samples are exclusively gsB (Figure 5A and 5C ). The accessory genes, nodA and nodD , include sequences designated as gsE and gsC for the French and UK samples, respectively, (Figure 5E and 5G ), based on the genospecies designation from the 196 Rlt isolates. This indicates that these sym gene alleles introgress between genospecies, but this introgression event was not captured by the isolates .
To investigate whether the microbial diversity was higher for DKO than for the field trial samples, we calculated the nucleotide diversities (π). Comparing the nucleotide diversity of the individual DKO groupings to the three separate field trial sites, most DKO groupings had a significantly higher 𝛑 than the DK, F, and UK trial sites (Figure 6 ). Grouping the samples by field management revealed that overall the organic DKO samples have significantly more diversity than the field trial samples for all four amplicons (rpoB : p-value=3.007e-06, recA : p-value=8.601e-07, nodA : p-value=1.916e-09, nodD : p-value=2.875e-08) (Figure S5 ). For our samples, this increase in nucleotide diversity within nodule populations indicates that the microbial diversity of theRhizobium population is higher in fields under organic management than in field trial sites under conventional management.