Soil redox status governs within-field spatial variation in microbial arsenic methylation and rice straighthead disease

Abstract Microbial arsenic (As) methylation in paddy soil produces mainly dimethylarsenate (DMA), which can cause physiological straighthead disease in rice. The disease is often highly patchy in the field, but the reasons remain unknown. We investigated within-field spatial variations in straighthead disease severity, As species in rice husks and in soil porewater, microbial composition and abundance of arsM gene encoding arsenite S-adenosylmethionine methyltransferase in two paddy fields. The spatial pattern of disease severity matched those of soil redox potential, arsM gene abundance, porewater DMA concentration, and husk DMA concentration in both fields. Structural equation modelling identified soil redox potential as the key factor affecting arsM gene abundance, consequently impacting porewater DMA and husk DMA concentrations. Core amplicon variants that correlated positively with husk DMA concentration belonged mainly to the phyla of Chloroflexi, Bacillota, Acidobacteriota, Actinobacteriota, and Myxococcota. Meta-omics analyses of soil samples from the disease and non-disease patches identified 5129 arsM gene sequences, with 71% being transcribed. The arsM-carrying hosts were diverse and dominated by anaerobic bacteria. Between 96 and 115 arsM sequences were significantly more expressed in the soil samples from the disease than from the non-disease patch, which were distributed across 18 phyla, especially Acidobacteriota, Bacteroidota, Verrucomicrobiota, Chloroflexota, Pseudomonadota, and Actinomycetota. This study demonstrates that even a small variation in soil redox potential within the anoxic range can cause a large variation in the abundance of As-methylating microorganisms, thus resulting in within-field variation in rice straighthead disease. Raising soil redox potential could be an effective way to prevent straighthead disease.


Figure S1 .
Figure S1.Layout of field experiments in TC and SY paddy fields.Locations of the field experiments (A).Sampling points of rice (n = 153) and paddy soil and porewater (n = 81) in TC and SY fields (B).(C) A diagram showing the slope plot experiments in TC and SY fields.The slope was created from the irrigation inlet to the outlet.Red circles represent the sampling sites of soil and porewater samples, black crosses the sampling sites of rice panicles.

Figure S2 .
Figure S2.Soil samples from the straighthead disease and non-disease patches selected for metagenomic and metatranscriptomic analysis in TC (A) and SY (B) fields.Black and yellow circles represent the samples from the non-disease and disease patches, respectively.

Figure S3 .
Figure S3.Spatial variations in As species in rice husk: concentrations of iAs (A, B) and MMA (C, D) in TC (A, C) and SY (B, D) fields.Relationships between seed setting rate and iAs (E) and MMA (F) concentration in rice husk.

Figure S4 .
Figure S4.Spatial variations in soil total As concentration (A, B), soil pH (C, D) and soil organic matter content (E, F) in TC (A, C, E) and SY (B, D, F) fields.Particle size distribution of soils collected from non-disease and disease patches in TC and SY fields (G). in (A) and (B) denotes the sampling sites of soils used for particle size analysis.

Figure S5 .
Figure S5.Spatial variations in soil Eh at the rice growth stages of late tillering and heading in TC (A) and SY (B) paddy fields.

Figure S6 .
Figure S6.Spatial variations in As species in porewater.Spatial variation in iAs concentration (A, B) in porewater collected from TC (A) and SY (B) fields, and MMA concentration (C) in porewater from TC field.

Figure S7 .
Figure S7.Diversity of bacterial compositions in TC and SY soils.Shannon index (A) and Principal coordinate analysis based on Bray-curtis distance (B) of bacterial compositions.Relationship between soil Eh at the booting stage and Shannon index of TC (C) and SY (D) soils.Redundancy analysis showing the effect of environmental variables on bacterial compositions in TC (E) and SY (F) soils.

Figure S8 .
Figure S8.Co-abundant network analysis of core ASVs that were negatively related to soil Eh in TC (A) and SY (B) soils.

Figure S9 .
Figure S9.Relationship between co-abundant groups (CAGs) of ASVs and soil Eh in TC soil.

Figure S10 .
Figure S10.Relationship between co-abundant groups (CAGs) of ASVs and soil Eh in SY soil.

Figure S12 .
Figure S12.Important ASVs identified by Randomforest for husk DMA concentration in TC (A) and SY (B) fields.

Figure S13 .
Figure S13.Distribution of arsM genes in bacteria and archaea based on metagenomic analysis (A), and distribution of arsM genes in different phyla (B).

Figure S14 .
Figure S14.Metagenomic analysis reveals significant differences in the relative abundance of arsM genes between the straighthead disease and non-disease patches in TC (A) and SY (B) fields.Number (C) and species (D) of arsM genes with a significantly higher abundance in the disease patch and shared between TC and SY fields.Data in A, B and D are means ± SD (n = 3).

Figure S15 .
Figure S15.Metagenome-assembled genomes (MAGs) of archaea in TC and SY soils.arsM and mcrA genes were found to coexist in MAGs via blast against the database.Star and triangle represent the presence of arsM and mcrA genes in MAGs.Relative abundance of arsM gene and transcript in the straighthead disease and non-disease patches are displayed using pie charts.The scale (0.08) indicates sequence divergence.

Figure S16 .
Figure S16.Field slope experiments: spatial variations in soil Eh (A, B), soil total As concentration (C, D), porewater iAs concentration (E, F) and porewater DMA concentration (G, H), and arsM gene abundance (I, J) in TC (A, C, E, G, I) and SY (B, D, F, H, J) slope plots.

Figure S17 .
Figure S17.Field slope experiments: spatial variations in husk DMA concentration (A, B), seed setting rate (C, D) and husk iAs concentration (E, F) in TC (A, C, E) and SY (B, D, F) field slopes.