Strong Host Modulation of Rhizosphere-to-Endosphere Microbial Colonisation in Natural Populations of the Pan-Palaeotropical Keystone Grass Species, Themeda triandra

Abstract

Soil microbiota can colonise plant roots through a two-step selection process, involving recruitment of microbiota first from bulk soil into plant rhizospheres, then into root endospheres. This process is poorly understood in all but a few model species (e.g., Arabidopsis), which is surprising given its fundamental role in plant and soil ecology. Here, we examined the microbial community assembly processes across the rhizospheres and root endospheres in eight natural populations of the pan-palaeotropical C4 grass, Themeda triandra, in southern Australia. Using a space-for-time substitution approach, we assessed whether bacterial root colonisation patterns conformed to the two-step model and tested whether community assembly was driven more by deterministic or stochastic processes. Our results show that the two-step selection process shaped bacterial recruitment dynamics across these natural T. triandra populations, and we provide clear evidence that host plants influence microbial assembly via deterministic pressures that produce strong community convergence within endospheres. These findings highlight the central role of host filtering in shaping a conserved 'core' endosphere microbiome. However, limited understanding of these endosphere communities constrains efforts to harness these important relationships to, for example, improve plant propagation and revegetation practices.

Keywords: Themeda triandra; endosphere; microbial ecology; neutral theory model; rhizosphere; two‐step selection process.

FIGURE 1

(a) Map showing Australia and the sampling locations of Themeda triandra populations (blue points) across a strong aridity gradient in southern Australia. (b) Bacterial alpha diversity as effective number of ASVs in T. triandra rhizospheres. (c) NMDS ordination showing the differences in bacterial community composition between rhizospheres (blue) and endospheres (red). (d) Distance to centroid of samples comparing rhizosphere (blue) and endosphere (red) samples, calculated from Bray–Curtis dissimilarity.

FIGURE 2

Bacterial ASV relative abundances visualised at the phylum level in endospheres and rhizospheres. (a) Stacked bars represent samples, grouped by aridity index of their sampling site. The bar colours represent the bacterial phylum. Chord diagrams for (b) rhizospheres and (c) endospheres, showing the relative proportion of each bacterial ASV within each phylum (groupings: A–K) found within bacterial abundance categories (AT—abundant taxa, MT—moderate taxa, RT—rare taxa, CAT—categorically abundant taxa, CRT—categorically rare taxa, and CRAT—categorically rare and abundant taxa). (d) Differential abundance analysis of major and minor bacterial phyla across the rhizospheres and endospheres.

FIGURE 3

(a) Venn diagram of unique ASVs across T. triandra endospheres and rhizospheres showing the number of unique ASVs and the percentage of reads within each grouping, and (b) plot summarising relative abundance of phyla for the unique and shared ASVs in the endospheres and rhizospheres. (c) Partial Venn diagram showing unique T. triandra rhizosphere ASVs across each sampling site and shared across all sites, and the percentage of reads within each grouping; and (d) partial Venn diagram of unique ASVs across T. triandra endospheres in each site and shared across all sites, showing the number of unique ASVs and the percentage of reads within each grouping.

FIGURE 4

(a) Heatmap showing 218 differentially abundant ASVs across T. triandra rhizospheres and endospheres, with clustering of ASVs with high and low log‐fold changes represented by the dendrogram, and (b) the number of differentially abundant ASVs calculated within each sampling site. The negative grouping includes those ASVs favoured in endospheres (negative log‐fold change), whereas the positive grouping includes ASVs favoured in rhizospheres (positive log‐fold change). Sites are ordered from most to least arid (top to bottom, respectively). (c) Upset plot showing the number of shared and unique bacterial ASVs across each site. This plot shows the first 30 most populated ASV intersections between sites (see Figure S14 for all site intersections).

FIGURE 5

Abundance‐occupancy curves fitted with the Sloan neutral model in T. triandra (a) rhizospheres and (b) endospheres. Each point represents a bacterial ASV that occurs above (blue), below (pink), or within (white) neutral model predictions. ASVs that occur at greater occupancies than predicted by the neutral model (blue) are hypothesised to be positively selected by the environment, and those occurring with lower occupancies than predicted by the neutral model (pink) are hypothesised to be negatively selected by the environment. (c) βNTI values across the rhizospheres and endospheres of each sampling site in order of aridity index, and (d) stacked bar plot illustrating the relative contribution of ecological assembly processes across rhizospheres and endospheres in each sampling site in order of aridity index. Heterogeneous and homogeneous selection is attributed to βNTI values of > +2 or < −2, respectively. Communities without significant βNTI deviations (|βNTI | < 2) were investigated for homogenising dispersal or limiting dispersal with RCbray values of < −0.95 or > 0.95, respectively.

FIGURE 6

Network analysis of bacterial ASVs in (a) rhizospheres and (b) endospheres, sampled across T. triandra aridity gradient. Vertex colour indicates taxonomic groups at the ASV level. Positive associations are represented by blue edges, and negative associations are represented by red edges. The average degree and average edge weight are shown below for each network with their respective standard error.