Bacterial richness and biodiversity
In soil and root samples, 1455 and 915 bacterial ASVs remained after filtering out sparse ASVs and outliers, respectively. These AVSs were affiliated with 20 phyla, 39 classes, 87 orders, 130 families and 184 genera in root and 27 phyla, 51 classes, 114 orders, 170 Families and 238 Genera in soil.
Both soil and root samples were dominated by Proteobacteria, which made up, on average, 89% of classified reads in root samples and 53% in soil samples (Figure 4A). Taking a more detailed look at the distribution of orders within the Proteobacteria (Figure 4B), pronounced different patterns between the compartments (root, soil) as well as the treatments (C, H, H+D, D). In soil samples, a large fraction of Micropepsales (average of 11.5 % of reads) was detected, which is substantially less frequent in root samples (1.2% of reads). In the roots, treatment differences also refer to proteobacteria, mainly classified as Rhizobiales and Burkolderiales. A large fraction of Pseudomonadales occurs in both the C and D treatments. The combined H+D treatment only was characterised by a substantial increase in Enterobacteriales (26.4% of reads), which were also present in the heat treatment to a lesser extent (mean 1.56%).
A decrease in bacterial alpha diversity was observed from the rhizosphere to the root. Considering the different stress, differences in alpha-diversity were evident under drought stress (with lower values than the not stressed control). Analysis of variance (ANOVA) followed by Tukey’s honestly significant difference (HSD) post hoc test revealed no effect of the treatments on the richness in either compartment. However, a significant interaction between treatment and Inverse Simpson index for the root samples was identified (p≥0.05).
The beta-diversity was determined using principal coordinate analysis (PCoA) with Bray-Curtis dissimilarities. The first two axes accounted for over 72% and 58 % of the variance between treatments of root and soil samples, respectively. PERMANOVA analysis on the root and soil datasets indicated a significant effect of all treatments, where drought alone explains 23.2% of the variance, heat 38.7% and the combination 22.2% in root samples. In soil samples, D explains 29.1% of the variance, heat only 18.9 % and combined H+D only 13.6% of the variance (p≥0.001) ( Supplementary Table S4). Then, linear discriminant analysis effect size (LEfSe) was carried out to identify the taxa driving the differences between treatments (Supplementary Figure S2) and the ANCOM-BC differential analysis was used to identify the taxa differentially abundant between the treatments. Both methods perform a similar analysis but look at the problem differently. As expected, the marker taxa established as significant were largely shared between the methods, being 131 in roots and 138 in soil (Supplementary Figure S3). For interpretation, we focused then on LefSE results to identify keystone taxa that explain differences between sample groups.
In root samples, LefSe analysis revealed 21 microbial biomarkers in C, 11 in D, 71 in H and 14 in H+D.The most discriminant taxa in H belonged to the phyla of Actinobacteriota and Proteobacteria , in particular, Rhizobiales , Xanthomonadales andStreptomycetales, while Burkholderiales characterized the D samples. The combined stress (H+D) was characterized by the presence of Gammaproteobacteria, being the Enterobacterales and Azospirillaleles the most discriminating taxa. However, non-stressed roots (C) were characterized by Pseudomanadales andCaulobacterales . (Supplementary Figure S4A)
In Soil samples there were 12 microbial biomarkers detected in C, 9 in D, 20 in H and 25 in H+D. Here the most discriminant taxa for H were theAlphaproteobacterium Azospirillum and the PhylumFirmicutes member of Bacillales and Clostridiales . At the same time, H+D was most defined by a taxon fromPolyangiales , followed by Rhizobiales, Fibrobacterales, Sphingobacteriales and Chitinophagales, and member from the phylum Elusimicrobiota. In D differences were driven byDevosiaceae and Cryseobacterium and C byGeobacterales and Ktedonobacterales (Supplementary Figure S4B).