Domestication affects the composition, diversity, and co-occurrence of the cereal seed microbiota

Abstract

Introduction: The seed-associated microbiome has a strong influence on plant ecology, fitness, and productivity. Plant microbiota could be exploited for a more responsible crop management in sustainable agriculture. However, the relationships between seed microbiota and hosts related to the changes from ancestor species to breeded crops still remain poor understood.

Objectives: Our aims were i) to understand the effect of cereal domestication on seed endophytes in terms of diversity, structure and co-occurrence, by comparing four cereal crops and the respective ancestor species; ii) to test the phylogenetic coherence between cereals and their seed microbiota (clue of co-evolution).

Methods: We investigated the seed microbiota of four cereal crops (Triticum aestivum, Triticum monococcum, Triticum durum, and Hordeum vulgare), along with their respective ancestors (Aegilops tauschii, Triticum baeoticum, Triticum dicoccoides, and Hordeum spontaneum, respectively) using 16S rRNA gene metabarcoding, Randomly Amplified Polymorphic DNA (RAPD) profiling of host plants and co-evolution analysis.

Results: The diversity of seed microbiota was generally higher in cultivated cereals than in wild ancestors, suggesting that domestication lead to a bacterial diversification. On the other hand, more microbe-microbe interactions were detected in wild species, indicating a better-structured, mature community. Typical human-associated taxa, such as Cutibacterium, dominated in cultivated cereals, suggesting an interkingdom transfers of microbes from human to plants during domestication. Co-evolution analysis revealed a significant phylogenetic congruence between seed endophytes and host plants, indicating clues of co-evolution between hosts and seed-associated microbes during domestication.

Conclusion: This study demonstrates a diversification of the seed microbiome as a consequence of domestication, and provides clues of co-evolution between cereals and their seed microbiota. This knowledge is useful to develop effective strategies of microbiome exploitation for sustainable agriculture.

Keywords: 16S metabarcoding; Cereals; Co-evolution; Domestication; Random Amplified Polymorphic DNA - RAPD; Seed microbiome.

Graphical abstract

Fig. 1

The evolutionary history of wheat and barley crops (green boxes) and their wild relatives, both used (orange boxes) and not used (gray boxes) in this work. Dotted arrows show the parental lines of domesticated forms. Crosses indicate cross-breeding events.

Fig. 2

Bacterial taxonomic composition of seed endophytes of wild and cultivated species, at Phylum (A) and Family (B) levels. Triticum aestivum (TA), Triticum durum (TDU), Triticum monococcum (TM), Aegilops tauschii (AT), Triticum diccocoides (TDI), Triticum baeoticum (TB), Hordeum vulgare (HV) and Hordeum spontaneum (HS). Relative abundance of major taxa only (>1% of total reads) according to 16S rRNA gene metabarcoding.

Fig. 3

Relative abundance of the genera Pseudomonas and Cutibacterium in the wild and cultivated cereals, as resulted from metabarcoding analysis, calculated on the rarefied dataset (sequence depth = 1000 reads per sample). ** p = 0.0013; *** p < 0.001.

Fig. 4

Alpha and beta diversity metrics of seed endophyte microbiota. (A) Shannon–Weaver, Simpson, Dominance and Equitability indices of bacterial microbiota (OTU 97%), grouped by cultivation form (t-test, P < 0.05), according to 16S rRNA gene metabarcoding. CS = Cultivated species; WS = Wild species. (B) Non-metric multidimensional scaling plot for bacterial microbiota structure based on weighted Bray-Curtis distances. Samples are colored by plant species and shaped by cultivation form. ADONIS significance test: R² = 0.078, P = 0.003 for the factor “cultivation form”; R² = 0.32, P < 0.001 for the factor “species”; stress value: 0.1495.

Fig. 5

Co-occurrence network of OTUs, calculated for wild and cultivated cereal species separately. Nodes are colored by taxonomy and sized by degree (=n. of connections). Edges are colored by correlation type (blue: positive; red: negative) and the thickness represents merged FDR-corrected P-values (the thicker, the more significant).

Fig. 6

UPGMA dendrogram showing the genetic similarity between cultivars of wheat and barley and their corresponding ancestors, calculated on RAPD data. The dendrogram was constructed using similarity coefficients based on the proportion of different bands.

Fig. 7

Co-evolution analysis. (A) Tanglegram between the cereal phylogeny, based on RAPD analysis, and the bacterial phylogeny (at OTU level), based on the metabarcoding analysis. Blue lines indicate plant–bacterial associations that were more significant than expected by chance, according to the ParaFitGlobal statistic. (B) Visualization of the global goodness-of-fit statistics over 1000 permutation (P = 0.024); the black line indicates the observed global sum of squared residuals (m²_XY) value, while the grey columns indicate the m²_XY value of each of the 1000 randomizations, sorted by ascending values. (C) and (D) Same than (A) and (B), calculated after removal of T. durum and T. dicoccoides samples.