Functional traits dominate the diversity-related selection of bacterial communities in the rhizosphere

Abstract

We studied the impact of community diversity on the selection of bacterial communities in the rhizosphere by comparing the composition and the functional traits of these communities in soil and rhizosphere. Differences in diversity were established by inoculating into sterilized soils diluted suspensions of the same soil. We used 16S ribosomal RNA amplicon sequencing to determine the taxonomical structure of the bacterial communities and a shotgun metagenomics approach to investigate the potential functional diversity of the communities. By comparing the bacterial communities in soil and rhizosphere, the selective power of the plant was observed both at the taxonomic and functional level, although the diversity indices of soil and rhizosphere samples showed a highly variable, irregular pattern. Lesser variation, that is, more homogenization, was found for both the taxonomic structure and the functional profile of the rhizosphere communities as compared to the communities of the bulk soil. Network analysis revealed stronger interactions among bacterial operational taxonomic units in the rhizosphere than in the soil. The enrichment processes in the rhizosphere selected microbes with particular functional genes related to transporters, the Embden-Meyerhof-Parnas pathway and hydrogen metabolism. This selection was not random across bacteria with these functional traits, but it was species specific. Overall, this suggests that functional traits are a key to the assembly of bacterial rhizosphere communities.

Figure 1

PCoA of the soil and rhizosphere bacteria community compositions and functional traits. (a) Variation between samples of soil and rhizosphere based on the Bray–Curtis similarity for taxonomical data and (b) functional traits using relative abundances based on FOAM level 2. Variation between dilutions of soil samples based on the Bray–Curtis similarity for taxonomical data (c) and functional traits (d). Variation between dilutions of rhizosphere samples based on the Bray–Curtis similarity for taxonomical data (e) and functional traits (f). Similarity values (analysis of similarity) are shown in the upper left of each plot. Similarities between replicates of each dilution are shown in (g) Dark gray bars indicate taxonomical data and light gray bars indicate functional traits. The error bars show standard errors of six replicates.

Figure 2

Co-occurrence patterns of bacteria in soil and rhizosphere. Correlations were presented in the soil samples (a) and in the rhizosphere samples of each dilution (b). Nodes indicate taxonomic affiliation at genus level. Red lines indicate positive correlations, and blue lines indicate negative correlations. The color of each node indicates the phylum shown below of the figures. The size of each node is proportional to the BC (c). The box-and-whiskers graphics show the median of BC as a line, the 25th and 27th percentiles of the data as the top and bottom of the box, and outlier dots to indicate the most extreme data point within 1.5*(75th–25th percentile) of the median. The size of outlier data points corresponds to the value of the BC.

Figure 3

Profiles of soil and rhizosphere bacterial community functional traits. (a) The relative abundance of groups of functional genes in soil and rhizosphere for three dilutions. Relative abundance of functional genes (FOAM Database level 1) based on normalized shotgun metagenomics data in dilutions of 10⁻¹, 10⁻⁶ and 10⁻⁹. The percentage of the total sequence reads in samples from soil and rhizosphere is presented for each dilution. The error bars show standard errors of six replicates and orange asterisks (*) indicate categories that are significantly more abundant in rhizosphere samples (P value<0.05) and blue asterisks (*) indicate categories that are significantly more abundant in soil samples. (b) PCoA plots of species with particular functional genes that were more abundant in the soil than in the rhizosphere (cellular response to stress and carbohydrate activity enzymes) and plots of species with particular functional genes that were more abundant in the rhizosphere than in the soil (transporter genes, EMP pathway and hydrogen metabolism). Similarity values are shown in the upper right corner of each plot. The circles represent the clustering of the soil and rhizosphere samples, respectively. (c) Differences in abundance of families with transporters between soil and rhizosphere samples (Welch's t-test; P<0.05).