Communities in soils, roots and leaves differ significantly
Bacterial communities in soils roots and leaves of axenic beet plants grown in pasteurized soils differed significantly in terms of structure (9.57% of variance explained, Fig. 2A), alpha-diversity, where diversity and richness of ASVs followed a pattern of soils > roots > leaves (Shannon’s H’: robust ANOVA F = 583.60, p < 0.001; Shannon’s E: F = 39.97, p < 0.001; richness: F = 583.60, p < 0.001) and bacterial load, which was lower in leaves than in roots (Wilcoxon test W = 259 411, p < 0.001) (Fig. 2B). Globally, ASV80, classified as Achromobacter, and ASV23, classified as Chryseobacterium, were found to be characteristic for (i.e. significantly more abundant in) plant tissues, while ASV3, ASV21 and ASV27 (Pseudoxanthomonas, Sphingopyxis and Pedobacter, respectively) were characteristic for soils. ASV7 (Cellvibrio) was typical for roots, and ASV13 (Sphigobacterium) for leaves. Generally, differentially abundant ASVs affiliated with Gammaproteobacteria were characteristic for roots and soils, while Bacteroidia-afiliated ones were typical for leaves (Fig. 2E, Table SR3). At the level of genus Bacillus, Brevundimonas, Pedobacter, Pseudoxanthomonas and Stenotrophomonas were characteristic for soils, while Cellvibrio and Flavobacterium were more abundant in roots, and Sphingobacterium in leaves (Fig. 2G, SupplementaryResultsF1). Core microbiome in the three compartments was limited to a few ASVs (8 in soils, 3 in roots and 1 in leaves, Table 1), mainly members of Alpha- and Gammaproteobacteria as well as Bacteroidia.
Table 1
Core microbiomes of soil, roots and leaf samples. Prevalence cutoff of 0.1% and detection of 90% were used.
ASV | Detection1 | Prevalence2 | Taxonomy |
Soil |
ASV3 | 98.84 | 7.04 | Pseudoxanthomonas |
ASV5 | 98.20 | 2.77 | Delftia |
ASV6 | 87.90 | 2.91 | Brevundimonas |
ASV11 | 92.92 | 1.37 | Caulobacter |
ASV14 | 99.87 | 1.80 | Devosia |
ASV27 | 93.05 | 1.18 | Pedobacter |
ASV38 | 90.86 | 0.57 | Hydrogenophaga |
ASV44 | 94.72 | 0.58 | Reyranella |
Root |
ASV4 | 96.26 | 7.26 | Flavobacterium |
ASV5 | 97.01 | 5.30 | Delftia |
ASV14 | 91.29 | 1.18 | Devosia |
Leaf |
ASV4 | 97.77 | 6.80 | Flavobacterium |
1 – percent of the number of samples in which a given ASV was present, 2 – mean abundance of a given ASV in a given compartment |
Table 2
Late and early core microbiomes. Prevalence cutoff of 0.1% and detection of 90% were used.
ASV | Detection1 | Prevalence2 | Taxonomy |
Soil early |
ASV3 | 97.94 | 4.22 | Pseudoxanthomonas |
ASV4 | 96.11 | 3.74 | Flavobacterium |
ASV5 | 96.80 | 3.14 | Delftia |
ASV6 | 90.16 | 3.77 | Brevundimonas |
ASV14 | 99.77 | 1.58 | Devosia |
ASV44 | 91.76 | 0.31 | Reyranella |
Soil late |
ASV3 | 100 | 10.66 | Pseudoxanthomonas |
ASV5 | 100 | 2.28 | Delftia |
ASV10 | 93.53 | 1.83 | Pseudomonas |
ASV11 | 98.24 | 1.27 | Caulobacter |
ASV14 | 100 | 2.09 | Devosia |
ASV21 | 90.88 | 2.37 | Sphingopyxis |
ASV24 | 95.88 | 1.59 | Pseudohangiellaceae, BIyi10 |
ASV27 | 98.53 | 2.05 | Pedobacter |
ASV33 | 99.12 | 1.54 | Micropepsaceae |
ASV38 | 94.71 | 0.78 | Hydrogenophaga |
ASV44 | 98.53 | 0.93 | Reyranella |
ASV54 | 92.35 | 0.73 | Shinella |
ASV58 | 96.47 | 0.37 | Sphingobacteriaceae |
ASV59 | 96.76 | 0.68 | Bosea |
ASV74 | 92.65 | 0.66 | Gemmatirosa |
Root early |
ASV4 | 98.50 | 7.92 | Flavobacterium |
ASV5 | 98.80 | 5.75 | Delftia |
ASV14 | 94.31 | 1.21 | Devosia |
Root late |
ASV4 | 91.18 | 4.03 | Flavobacterium |
Leaf early |
ASV4 | 97.30 | 8.72 | Flavobacterium |
ASV23 | 93.69 | 8.24 | Chryseobacterium |
Leaf late |
ASV3 | 96.74 | 3.63 | Pseudoxanthomonas |
ASV4 | 98.91 | 2.19 | Flavobacterium |
ASV6 | 92.39 | 2.45 | Brevundimonas |
ASV11 | 91.30 | 1.70 | Caulobacter |
ASV13 | 94.57 | 2.31 | Sphingobacterium |
ASV15 | 97.83 | 4.18 | Flavobacterium |
ASV17 | 96.74 | 2.43 | Thermomonas |
1 – percent of the number of samples in a given set in which a given ASV was present, 2 – mean abundance of a given ASV in a given set of samples |
Table 3
Inoculated and non-inoculated samples core microbiomes. Prevalence cutoff of 0.1% and detection of 90% were used.
ASV | Detection1 | Prevalence2 | Taxonomy |
Soil early inoculated |
ASV5 | 96.80 | 5.81 | Delftia |
ASV6 | 90.16 | 6.97 | Brevundimonas |
ASV14 | 99.77 | 2.92 | Devosia |
ASV44 | 91.76 | 0.57 | Reyranella |
Soil early non-inoculated |
ASV3 | 97.94 | 9.18 | Pseudoxanthomonas |
ASV4 | 96.11 | 8.14 | Flavobacterium |
Soil late inoculated |
ASV5 | 100 | 4.44 | Delftia |
ASV10 | 93.53 | 3.57 | Pseudomonas |
ASV11 | 98.24 | 2.47 | Caulobacter |
ASV14 | 100 | 4.07 | Devosia |
ASV21 | 90.88 | 4.60 | Sphingopyxis |
ASV44 | 98.53 | 1.81 | Reyranella |
ASV54 | 92.35 | 1.43 | Shinella |
ASV58 | 96.47 | 0.73 | Sphingobacteriaceae |
ASV59 | 96.76 | 1.32 | Bosea |
ASV74 | 92.65 | 1.28 | Gemmatirosa |
Soil late non-inoculated |
ASV3 | 100 | 21.95 | Pseudoxanthomonas |
ASV24 | 95.88 | 3.27 | Pseudohallangiaceae, BIyi10 |
ASV27 | 98.53 | 4.22 | Pedobacter |
ASV33 | 99.12 | 3.17 | Micropepsaceae |
ASV38 | 94.71 | 1.60 | Hydrogenophaga |
Root early inoculated |
ASV5 | 98.80 | 10.48 | Delftia |
ASV14 | 94.31 | 2.20 | Devosia |
Root early non-inoculated |
ASV4 | 98.50 | 17.56 | Flavobacterium |
Root late inoculated |
No core ASVs |
Root late non-inoculated |
ASV4 | 91.18 | 7.85 | Flavobacterium |
Leaf early inoculated |
No core ASVs |
Leaf early non-inoculated |
ASV4 | 97.30 | 16.95 | Flavobacterium |
ASV23 | 93.69 | 16.03 | Chryseobacterium |
Leaf late inoculated |
ASV6 | 92.39 | 4.63 | Brevundimonas |
ASV11 | 91.30 | 3.21 | Caulobacter |
ASV13 | 94.57 | 4.37 | Sphingobacterium |
ASV15 | 97.83 | 7.90 | Flavobacterium |
Leaf late non-inoculated |
ASV3 | 96.74 | 7.71 | Pseudoxanthomonas |
ASV4 | 98.91 | 4.64 | Flavobacterium |
ASV17 | 96.74 | 5.17 | Thermomonas |
1 – percent of the number of samples in which a given ASV was present, 2 – mean abundance of a given ASV in a given compartment |
The differences, albeit smaller, were also visible at the level of PICRUSt2-predicted metabolic capabilities, which also grouped according to material (Fig. 2C, 6.70% variance explained) and alpha-diversity of functions followed the pattern observed for ASVs (Fig. 2D, Shannon’s H’: robust ANOVA F = 104.44, p < 0.001; Shannon’s E: F = 173.20, p < 0.001; richness: F = 851.62, p < 0.001). Functions related to competition between microorganisms (antibiotic resistance and biosynthesis, quorum sensing) appeared to be characteristic for soils and roots, while carbohydrate metabolism-related ones were predicted to be more frequent in genomes of soil and leaves-dwelling bacteria (Fig. 2F; Table SR4).
As material explained far greater fraction of variance than any other variable (Table SR5), to make the influence of other variables visible, further analyses were carried out on data divided into soils, roots and leaves sets.
First weeks of axenic beets growth in soil can be divided into two stages differing in community structure, diversity, bacterial load, predicted metabolic capabilities and nestedness
Timepoint was the second most important grouping variable, regardless of material (5.82% of variance explained in whole data). Three ‘early’ timepoints (T0, T1 and T2) clustered together and were significantly different from the ‘late’ ones (T3 and T4). Percent of variance explained by this grouping was 15.64% in leaves, 3.05% in roots and 9.67% in soils (Fig. 3A). Henceforth, the belonging of a sample to the ‘early’ or ‘late’ cluster will be called its ‘status’.
The difference between the early and late clusters was observed regardless of material, soil and genotype (Figs. SR4,5,6) and was also visible in alpha diversity measures, which were higher, and, in the case of plant tissue samples, in bacterial load, which was lower in late samples (Fig. 3B). The former effect was most pronounced in leaves, and least conspicuous in soils, while the drop of the number of bacterial 16S rRNA gene sequences was greater in roots than in leaves. Different organisms were characteristic for early and late samples in soils, roots and leaves. The organisms characteristic for soils and roots were rare ones (Fig. SR8ACE; SupplementaryResultsF1).
Traits characteristic for genomes of organisms thriving in late and early samples differed in soils, roots and leaves (Fig. 3C; SupplementaryResultsF2). Early soils harbored organisms whose genomes were enriched in genes involved in diverse functional arrays, among which methane metabolism, protein and nucleotide rescue from glyoxal glycation, as well as heavy metals resistance were most prominent. On the other hand, metabolism of aromatic compounds was characteristic for genomes of organisms dwelling in late soil samples (Fig. SR9F). In roots biofilm formation, exopolysaccharide synthesis and regulation of aminoacids pool was characteristic for early samples, and metabolism of aromatic compounds was characteristic for late ones (Fig. SR9D). Toxin/antitoxin systems were characteristic for leaves in general, and polysaccharide (chitin, pectin) utilization was characteristic for early leaf samples, while aromatic compounds metabolism was typical for late ones (Fig. SR9B). Functional diversity was significantly higher in late samples coming from leaves and soils, but not in roots (Fig. 3D).
The degree of nestedness calculated for soil – root – leaf matrices (for each plant (technical replicate) separately) was very low, essentially did not deviate from expected values derived from a null model (SupplementaryFile3) and decreased with time. The difference between early and late samples was significant (Fig. 3F; Table SR6).
Dispersal limitation (DL) dominated mechanisms governing entering bacteria to roots and their transfer to leaves in early samples, while other stochastic processes (drift) were more pronounced in late ones. DL’s share was greater in case of soil → roots transfer than in root → leaf one. Interestingly, levels of selection, albeit generally low, were higher in early samples than in late ones (Fig. 3G). Similar picture was seen while maintaining of soil, root and leaf communities was assessed (i.e. samples from the same biological replicate were compared), however in case of leaf communities DL was replaced with homogenizing dispersal (Fig. 3E).
Inoculation with lyophilized wild beet roots influences bacterial communities in soils and plants
Inoculant characterization
Reads affiliated with Pseudomonadota (formerly Proteobacteria), Bacteroidota (formerly Bacteroidetes) and Bacillota (formerly Firmicutes) were found in libraries prepared from DNA isolated from inoculant samples. The most abundant genera were Pseudoxanthomonas and Brevundimonas (> 5% each), while Pedobacter, Devosia, Caulobacter, Flavobacterium, Rhizobium, Sphingobacterium, Pseudomonas, Cellvibrio, Thermomonas, and Dyadobacter were less frequent (~ 2–5%; Fig. 4A). Rare genera (< 2% abundance) comprised 55% of reads. Diversity, measured as Shannon’s H’, was 4.40 ± 0.75, evenness 0.93 ± 0.003, and 144 ± 125 ASVs were observed in the inoculant samples (rarefied data, n = 3), while 437 were found in the non-rarefied dataset. Culturable bacterial density was 4.0 ± 0.09 ´ 105cfu/g (n = 6), while bacterial 16S rRNA counts were 2.0 ± 0.5 ´ 104 copies/ng of DNA, which translated into 1.4 ± 0.587 ´ 108 copies/g of inoculant (n = 8). The bacteria were able to grow in presence of up to 900 mM NaCl (Fig. 4B).
Community structure in inoculated and non-inoculated samples differs significantly and the differing ASVs depend on compartment, soil and genotype
Inoculation had no influence on bacterial alpha-diversity (Fig. 5A), and its effect on bacterial community structure was small but significant in each of the studied compartments and greater in soils than in roots or leaves (0.48, 0.17 and 0.09 percent of explained variance, respectively, Fig. 5B). Bacterial load in plant samples did not differ in inoculated and non-inoculated ones, regardless of experimental variant (Fig. 5A and Fig. SR35D). Further analyses showed that inoculation significantly impacted bacterial communities in all experimental variants in soils, regardless of status, but only in certain variants in the case of plant samples (Figs. SR32-34, Table SR7). Mean d05 generalized UniFrac distances between inoculated samples and inoculant was surprisingly slightly greater (0.3810 ± 0.0547) than for non-inoculated ones (0.3699 ± 0.0554), and the difference was significant (Wilcoxon test, W = 3.3482e + 10, p < 0.001).
The effect of inoculation was even smaller for predicted functional potential, both for alpha- (Fig. 5D) and beta-diversity (Fig. 5C), and in plant samples was insignificant. Inoculation influenced only early samples of genotypes C and M and late ones in genotype B in soils (Figs. SR36-38 and Table SR7).
Nestedness level was not changed by inoculation (Fig. 5F) and differences in shares of community assembly processes were visible only in case of late leaves where shares of homogenizing dispersal were lower in inoculated samples (Fig. 5E and 5G).
Taxonomic compositon at the genus level was similar in inoculated and non-inoculated samples, only three non-rare genera were differentially abundant in soils and one in roots (Fig. 5H, SupplementaryResults F1). Sets of ASVs and functions characteristic for inoculated and non-inoculated samples were different in early and late samples as well as in each experimental variant (Figs. SR40 and SR42-47, SR41 and SR48-53, respectively as well as SupplementaryResultsF1 and SupplementaryResultsF2). Out of 437 ASVs detected in inoculant, 268 were found exclusively in inoculated samples, albeit they were rare (low abundant) ones. However, when rarefied data were used, only fifteen such ASVs were found (Table SR9). No ASV present in inoculant was found only in non-inoculated samples. Globally, 29 ASVs were identified with biosigner as signature for either inoculated or non-inoculated samples. The ASVs were classified mainly as Alpha- and Gammaproteobacteria as well as Flavobacteria and Chitinophaga (Table SR8). Characteristic ASVs could be found mainly in soils, in case of plant samples it was possible only in certain variants. The organisms that differentiated inoculated late samples from non-inoculated ones were different in each combination of material, soil and genotype. The influence of inoculation was most visible in soil samples (greatest number of ASVs differentiating inoculated and non-inoculated samples), while in roots and leaves there were only single ASVs in certain variants. They belonged mainly to Proteobacteria and Firmicutes. Differences in functional potential comprised diverse functions, and genes involved in antibiotics biosynthesis and resistance were frequently found to be higher represented in inoculated samples than in non-inoculated ones, potentially pointing at increased level of competition.