Metagenomic and machine learning-aided identification of biomarkers driving distinctive Cd accumulation features in the root-associated microbiome of two rice cultivars

Abstract

Developing low-cadmium (Cd) rice cultivars has emerged as a promising avenue for food safety in Cd-contaminated farmlands. The root-associated microbiomes of rice have been shown to enhance rice growth and alleviate Cd stress. However, the microbial taxon-specific Cd resistance mechanisms underlying different Cd accumulation characteristics between different rice cultivars remain largely unknown. This study compared low-Cd cultivar XS14 and hybrid rice cultivar YY17 for Cd accumulation with five soil amendments. The results showed that XS14 was characterized by more variable community structures and stable co-occurrence networks in the soil-root continuum compared to YY17. The stronger stochastic processes in assembly of the XS14 (~25%) rhizosphere community than that of YY17 (~12%) suggested XS14 may have higher resistance to changes in soil properties. Microbial co-occurrence networks and machine learning models jointly identified keystone indicator microbiota, such as Desulfobacteria in XS14 and Nitrospiraceae in YY17. Meanwhile, genes involved in sulfur cycling and nitrogen cycling were observed among the root-associated microbiome of these two cultivars, respectively. Microbiomes in the rhizosphere and root of XS14 showed a higher diversity in functioning, with the significant enrichment of functional genes related to amino acid and carbohydrate transport and metabolism, and sulfur cycling. Our findings revealed differences and similarities in the microbial communities associated with two rice cultivars, as well as bacterial biomarkers predictive of Cd-accumulation capacity. Thus, we provide new insights into taxon-specific recruitment strategies of two rice cultivars under Cd stress and highlight the utility of biomarkers in offering clues for enhancing crop resilience to Cd stresses in the future.