Microbial dysbiosis in roots and rhizosphere of grapevines experiencing decline is associated with active metabolic functions

Abstract

When grapevine decline, characterized by a premature decrease in vigor and yield and sometimes plant death, cannot be explained by pathological or physiological diseases, one may inquire whether the microbiological status of the soil is responsible. Previous studies have shown that the composition and structure of bacterial and fungal microbial communities in inter-row soil are affected in areas displaying vine decline, compared to areas with non-declining vines within the same plot. A more comprehensive analysis was conducted in one such plot. Although soil chemical parameters could not directly explain these differences, the declining vines presented lower vigor, yield, berry quality, and petiole mineral content than those in non-declining vines. The bacterial and fungal microbiome of the root endosphere, rhizosphere, and different horizons of the bulk soil were explored through enzymatic, metabolic diversity, and metabarcoding analysis in both areas. Despite the lower microbial diversity and richness in symptomatic roots and soil, higher microbial activity and enrichment of potentially both beneficial bacteria and pathogenic fungi were found in the declining area. Path modeling analysis linked the root microbial activity to berry quality, suggesting a determinant role of root microbiome in the berry mineral content. Furthermore, certain fungal and bacterial taxa were correlated with predicted metabolic pathways and metabolic processes assessed with Eco-Plates. These results unexpectedly revealed active microbial profiles in the belowground compartments associated with stressed vines, highlighting the interest of exploring the functional microbiota of plants, and more specifically roots and rhizosphere, under stressed conditions.

Keywords: Vitis vinifera growth; belowground microbiome; grapevine fitness; metabarcoding-based predicted functionality; root endophytes; soil quality.

Figure 1

Ordination biplot PCA for (A) petiole, (B) leaf blade, and (C) must compositions across symptomatic (orange, S) and asymptomatic (green, AS) areas. The size of the arrows indicates the contribution strength of the variables. Standard error ellipses show 95% confidence areas. (D) Yield is presented as the mean mass of primary and secondary bunches per plant (n = 28) and the average berry mass over 100 berries (n = 3). p-values were calculated using t-tests or Wilcoxon tests depending on the parametric assumption.

Figure 2

Bulk soil profile in symptomatic (orange, S) and asymptomatic (green, AS) areas. Biplot PCA of (A) physicochemical parameters (n = 3) and (B) Eco-Plates measurements represented by Simpson’s index, AWCD, AUC, functional richness, and family compounds consumed (i.e., amines, amino acids, carbohydrates, carboxylic acids, phenolic compounds, and polymers) coupled to bacterial and fungal level of cultivable populations. Standard error ellipses show 95% confidence areas. (C) Enzymatic activities represented by arylamidase, β-glucosidase, and phosphatase alkaline, and (D) q-PCR measurements for archaeal and bacterial 16S and 18S rRNA genes (n = 5). (E) Shared OTUs represented by Venn diagram with significant overlaps detected using hyper-geometric tests, and (F) α-diversity metrics (richness = Chao1, diversity = Simpson), as well as (G) relative abundance of phyla. Significant differences, except for shared OTUs, were calculated with t or Wilcoxon tests, depending on the normality hypothesis.

Figure 3

Enriched bacterial and fungal genera in symptomatic and asymptomatic root and rhizosphere compartments. Enriched bacterial and fungal genera using Limma-Voom differential analysis (p < 0.05) corrected with FDR are shown.

Figure 4

Compartmentalization effect on microbial communities. NMDS plot ordination of (A) bacterial and (B) fungal communities among the bulk (cross), rhizosphere (triangle), and root endosphere (square) across symptomatic S (orange) and asymptomatic AS (light green) soils. Dashed lines represent 95% confidence ellipses. The stress value and the results from betadisper analyses are included. LEfSe analysis (p < 0.05, FDR, LDA > 4) of enriched genera among the soil status × compartment conditions for (C) bacterial and (D) fungal communities with p_: phylum, c_: class, o_: order, f_: family.

Figure 5

Functional inference of (A) fungal community using the FUNGuild database represented by the relative abundances of the trophic modes (i.e., multi-function, pathotroph, saprotroph, and symbiotroph) and the total abundances of associated guilds. Predicted (B) pathways and (C) functions of bacterial communities using PICRUSt2. Barplots are represented with means ± standard error. p-values were calculated using Student t or Wilcoxon tests. Biplot analysis represents only the functions in which pathways were significantly different between S and AS roots, represented by 1 to 6 in (B).

Figure 6

Correlation analyses between consumption of the grouped carbon sources from Eco-Plates measurements (gray) and the dominant bacterial (green) and fungal (yellow) genera, and the PICRUSt2 inferences from the microbial communities (pink) present in symptomatic (left) and asymptomatic (right) bulk (top), rhizosphere (middle), and root (bottom) compartments. For root analyses, the Eco-Plates measurements were the ones from the rhizosphere samples since no Eco-Plates assays were performed on root samples. The color intensity shows the R-value of correlation in each panel, and the asterisk represents significant correlations (* indicate p<0.05; ** p<0.01; *** p<0.001). Significantly correlated variables were presented in bold.

Figure 7

Partial Least Squares Path Modeling (PLS-PM) showing the effects of soil status on berry must, leaf composition, and microbial communities and functioning in roots, bulk soil, and rhizosphere. (A) PLS-PM diagram showing the positive and negative relationships, respectively represented with blue and red arrows. Asterisks indicate significant paths, *p < 0.05 and ***p < 0.001. Numbers associated with the arrows indicate standardized path coefficients. R² values indicate the variance of latent variables explained by the model. (B) Direct and indirect effects between the different latent variables tested.