Due to its high efficiency, chemical control has been widely used to prevent rhizosphere-mediated disease (Panth et al. 2020). However, some chemicals are harmful to humans or the environment, thus limiting their application potential to the extent (Ristaino and Thomas 1997; Pesonen and Vähäkangas 2020).
Chlorine dioxide (ClO2) is a strong oxidant with broad antimicrobial spectrum, non-toxic to human body, and does not pollute the environment (Gómez-López et al. 2009; Zhong et al. 2019; Jefri et al. 2022). Therefore, it plays a vital role in tap water and sewage treatment (Aieta and Berg 1986; Katz et al. 1994), medical apparatus and environment disinfection (Lowe et al. 2013; Meyers et al. 2020), and food preservation (Park and Kang 2015; Sun et al. 2017). After 12 hours of exposure to chlorine dioxide gas (4 mg/l, 0.16 mg/g), the number of yeasts, molds,
The application of gaseous chlorine dioxide to soil disinfection has been well studied. A field study applied to the control of
There are complex interactions between soil microbes and plants (Trivedi et al. 2020). Plants alter their root exudates to recruit rhizosphere microorganisms as the first barrier against soil-borne pathogens (Mendes et al. 2011; Liu et al. 2021; Song et al. 2021). Soil pathogens can also evade or suppress the plant’s immune system to infect the host (De Jonge et al. 2010; Sánchez-Vallet et al. 2013; Gao et al. 2019). Therefore, rhizosphere microorganisms deserve attention. Since tobacco leaves are rich in volatile compounds (Zhang et al. 2013) and become even more abundant after flue-curing (Wahlberg et al. 1977), it is suitable for non-target metabolome analysis of the effects of chlorine dioxide on plant quality. Hence, tobacco was chosen as an indicator crop.
With these ideas in mind, we designed experiments combining amplicon sequencing and gas chromatography-mass spectrometry with studying the effects of chlorine dioxide soil disinfection on rhizosphere microbial communities and plants. We ask the following questions: 1) whether and how does chlorine dioxide solution affects rhizosphere microbial communities? 2) whether and how chlorine dioxide solution affects crops?
Summary of experimental design.
Ten ng of DNA was used for PCR amplification (20 μl reaction, 30 cycles) in triplicate. In detail, primer sets 338F/806R, and ITS1F/ITS2R were used to amplify the 16S rRNA V3-V4 region gene and the ITS1 region gene. After that, the amplified products were sequenced with the 2 × 300 bp kit using the Illumina MiSeq platform at the Majorbio Corporation (China).
GC conditions: A Hp-5 (Agilent 19091J-413) column (30 m × 0.25 mm × 0.25 μm) (Agilent, USA) was used; the injection port temperature was 250°C. The column temperature was programmed as follows: the initial temperature at 80°C (held for 5 min), increased to 100°C at 2°C/min, increased to 180°C at 5°C/min (held for 1 min), increased to 200°C at 2°C/min, increased to 280°C at 10°C/min (held for 15 min). The injection volume of filter liquor was 0.5 μl. High-purity helium was utilized as the carrier gas. The gas flow rate was set as 1.0 ml/min.
MS conditions: An electron ionization (EI) ion source was used with 70 EV ionization energy. The mass spectra scanning range was set as 30 ~ 500 m/z. The ion source temperature was 230°C, and the GC/MS interface temperature was 280°C.
Data processing: NIST 2014 was used in the matching comparison. The screening criteria was that the matching rate should be more than 600. The relative content of each component was presented by peak area.
Subsequent analyses of taxon abundances were principally conducted in R software (version 4.1.3) (R Core Team 2013). Bacterial and fungal ASVs abundance tables were resampled to a median of 4,910 and 10,940 reads per sample using the phyloseq R package (McMurdie and Holmes 2013). The Newick trees were visualized using table2itol (freely available in
Due to the ability of PICRUSt2 software (Douglas et al. 2020) to predict metabolic pathways according to taxonomy annotation of ASVs, the calculation of metabolic pathways and orthologs abundances was executed in default settings. We wrote a script of Python 3 to analyze predicted MetaCyc pathways (Caspi et al. 2020), and its visualization was accomplished using the R package networkD3 (freely available at
To unveil microbe-plant links possibly caused by chlorine dioxide, we developed a co-occurrence network analysis R script. All networks were constructed based on Spearman correlations between normalized family abundances of rhizosphere soil microorganisms, and metabolite abundances of ligero leaves measured by GC-MS (absolute correlation ≥ 0.8,
During the flood irrigation, plastic tanks used as water sources were placed on the side of the road. To analyze the relationship between the inlet distance of flood irrigation and tobacco phenotype, we selected 16 tobacco plants from four plots for the
ASVs found in a) bacterial and b) fungal microbiomes of different groups.
From inside to outside: taxonomic dendrogram showing each group’s bacterial and fungal microbiome. The first color ring identifies microbial phyla within the rhizosphere soil. Other color rings represent the relative abundance of ASVs in different groups.
Taxonomy of a) bacterial and b) fungal microbiome in different groups.
Phyla that accounted for less than 0.6% of total abundance within the four groups were classified as “Others”.
Alteration of predicted metagenomic pathways in a) bacterial and b) fungal microbiome.
UP –
Further analysis of bacteria was conducted with the KEGG Pathway database. 29.54% of level 2 pathways were significantly changed (Fig. 5). A comparison of KEGG ortholog (KO) abundance led to the identification of 1612 KOs with different abundance. Of these, 1,377 genes increased, and 235 genes decreased in abundance (Table SVIII). Many KOs were missing as few in the control group, which resulted many asymmetric points with high fold-change values (Fig. 6a). The KEGG analysis showed that many movement-related, reproduction-related, and metabolism-related pathways were significantly enriched, which was similar to the results of bacterial MetaCyc metabolic pathways (Fig. 6b and Table SIX).
Relative abundance of bacterial KEGG pathways in levels 1 and 2.
CK – H2O treatment; CD – chlorine dioxide treatment.
* 0.01 <
One-sample
KEGG enrichment analysis of bacterial communities.
a) Volcano plots of DEGs between the control and the treatment group. The horizontal and vertical lines indicate a significance threshold (
b) Bubble diagram of the upregulated and downregulated DEGs in the KEGG database. The size of a bubble represents the number of DEGs. The color of a bubble represents the enrichment value of DEGs. One-sample
Visualization of the a) bacteria-metabolite and the b) fungi-metabolite network.
Network construction is based on Spearman correlation calculation results. Blue dots represent volatile compounds.
The other dots represent different family-level microbial taxonomy, respectively. Bacterial dots with different colors represent different bacterial phyla. Fungal dots with different colors represent different fungal phyla. Red lines represent positive correlations, and green lines represent negative correlations (absolute correlation ≥ 0.8,
In addition to distribution regularity, metabolite and microbial nodes they also had different topological properties. The degree of metabolite nodes was usually lower than microbe nodes. Among-module connectivity of metabolite nodes was often greater, while within-module connectivity showed an opposite trend. Thus, metabolite nodes always were connectors, and module hubs always were microbe nodes (Fig. 8). PCoA results of node properties showed distribution difference between microbe nodes and metabolite nodes (Table SXIII and SXIV, Fig. 9a and 9b). Natural connectivity analysis suggested that the robustness of microbe-metabolite networks was fragile, although their trends slightly differed (Fig. 10a). PCoA results of volatile metabolite showed no significant difference between the groups (Fig. 10b). PCoA results of tobacco phenotype demonstrated no significant differences among the groups (Fig. 10c). According to the
Relationships among partial network node attributes.
a), d), Relationships between degree and within-module connectivity (Zi) in a) bacteria-metabolite and d) fungi-metabolite network;
b), e), relationships between degree and among module-connectivity (Pi) in b) bacteria-metabolite and e) fungi-metabolite network;
c), f), keystone taxa were speculated based on their topological node features in c) bacteria-metabolite and f) fungi-metabolite network.
Blue dots represent metabolite nodes, and red dots represent microbe nodes. A node was identified as a module hub if its Zi ≥ 2.5, as a connector if its Pi ≥ 0.62, and as a network hub if it had Zi ≥ 2.5 and Pi ≥ 0.62.
Node attributes analysis.
Seven topological node parameters in a) bacteria-metabolite and b) fungi-metabolite network (listed in Table SXII and SXIII) were used for a pharmacy curriculum outcomes assessment (PCoA) analysis. Blue dots represent metabolite nodes, and red dots represent microbe nodes. In order to remove the influence of point overlap on observation, a random offset of 0.05 in the horizontal and vertical directions was added to each point. Analyses of similarity (ANOSIM) were performed on the Bray-Curtis distance matrix to evaluate whether there are differences between the metabolite node group and the microbe node group.
Network robustness, volatile metabolite, and tobacco phenotype analysis.
a) The natural connectivity of microbe-metabolite networks. Blue dots represent the bacteria-metabolite network, and red dots represent the fungi-metabolite network. b) Abundance of 207 volatile components from 12 plots was used for a pharmacy curriculum outcomes assessment (PCoA) analysis. c) Data on seven phenotypes of 60 tobacco plants were used for a pharmacy curriculum outcomes assessment (PCoA) analysis. The analysis of similarity (ANOSIM) was performed on the Bray-Curtis distance matrix to evaluate whether there are differences between groups.
If we define phyla that appeared in the taxa bar after chlorine dioxide treatment as a “new-phylum”, bacterial new-phyla (Firmicutes, Bacteroidota, Myxococcota, Patescibacteria, and Verrucomicroboata) accounted for 11.3% of the bacterial community. Fungal new-phylum (Basidomycota) accounted for 5.8% of the fungal community (Table SII and SIII). However, according to the importance of nodes in the network, four of ten nodes (40%) removed from the bacteria-metabolite network belonged to bacterial new-phylum, and three of ten nodes (30%) removed from the fungi-metabolite network belonged to fungal new-phylum (Table SXV).
According to microbial community structure and function analysis results, chlorine dioxide had an effect on rhizosphere microorganisms, and the effect on microbial community structure was enhanced with the increase of the dose (Fig. 2, 3, and 4). The disinfection treatment was only carried out once before transplanting, while the effects were not eliminated when the tobacco was harvested. KEGG enrichment analysis of rhizosphere bacterial communities revealed more details of the stress response. Chlorine dioxide might induce chemotaxis in soil bacterial community. Flood irrigation brings excess water, and soil pores are filled with water, while soil particles are less affected (Haghnazari et al. 2015). Because of chemotaxis, bacteria tend to escape chlorine dioxide by entering the interior of soil particles. Bacteria with well-developed flagella and strong motility are more likely to survive. Therefore, pathways of flagellar assembly and bacterial motility proteins are enriched in the bacteria kingdom suffering from chlorine dioxide disinfection (Fig. 6b and S2). The peptidoglycan synthesis pathway was also enriched since the P ring of flagellar is embedded in the peptidoglycan layer (Gupta and Gupta 2021) (Fig. 6b and S3). Simultaneous enrichment of flagellar assembly and peptidoglycan synthesis pathways was reported in the transcriptome analysis of the biocide stress effect on
Chlorine dioxide might have a better inhibitory effect on bacterial soil-borne diseases. We observed that bacteria respond more strongly than fungi in terms of community structure (Fig. 3) and metabolic pathways (Fig. 4). In addition, pathways related to reproduction were also enriched in the bacterial community after disinfection treatment (Fig. 6). These phenomena suggest that bacteria were sensitive to chlorine dioxide and inactivated massively. As a result, the internal competitive pressure decreased. This difference was consistent with the previous observations that gaseous chlorine dioxide disinfection reduced bacterial concentration levels more than fungi (Popa et al. 2007; Hsu and Huang 2013). We hypothesize that a combination of the antioxidant enzyme and endomembrane systems could explain the different responses of bacteria and fungi to chlorine dioxide. Contemporary microbes inherited iron-dependent enzymes from their anoxic ancestors (Khademian and Imlay 2021). Therefore, oxidative stress induced by reactive nitrogen and oxygen species can also be an immune defense strategy against pathogenic microorganisms (Staerck et al. 2017). Microbes evolved the antioxidant enzyme system to fend off the destructive actions of oxidizing reactions (Khademian and Imlay 2021). Since eukaryotic cells have numerous endomembrane structures which improve the efficiency of cellular physiological and biochemical reactions significantly, tolerance of the antioxidant enzyme system could be better than bacterial one in terms of chlorine dioxide disinfection.
There were correlations between the soil microbial community and volatile tobacco metabolites under the influence of chlorine dioxide. We observed that several enriched pathways (flagella assembly, bacterial motility, bacterial secretion) were also associated with endophytic colonization. Once soil microorganisms perceive plant-derived signals, they move to the root through flagella and attach to the surface. To colonize the plant, they need to form biofilms and produce substances such as microbial detoxification enzymes and lytic enzymes subsequently (Trivedi et al. 2020). Microbe-metabolite co-occurrence network analysis showed that microbe nodes and metabolite nodes are not randomly combined (Fig. 8). In terms of bacteria and fungi, the different modes of topology attributes between two nodes types were also diverse (Fig. 9a and 9b), suggesting potential connections between rhizosphere microorganisms and plant metabolites. In addition, robustness results showed that the association is relatively fragile (Fig. 10a). It may be a reason for the lack of significant difference in tobacco quality and phenotype (Fig. 10b and 10c). Since plants interact with functional traits rather than a taxonomy (Trivedi et al. 2020), functional changes of rhizosphere microbial communities caused by chlorine dioxide might not be enough to affect plants significantly (Fig. 5). Therefore, chlorine dioxide soil disinfection is safe for agricultural application.
According to the ecological community theory, the emergence of “new-phylum” in treatment groups might be the result of “disturbance” and “dispersal” (Vellend 2010). Chlorine dioxide disturbs the soil microbial community, reducing the species’ density and making the resources relatively adequate. For the experiment’s convenience, the blocks’ location was selected near the road, river, and tobacco curing room. Therefore, there was a regional species pool around blocks, and alien species could enter plots through dispersal. In the control group, all the ecological niches were occupied, and colonization by foreign microorganisms was challenging. Treatment groups are the opposite due to the diminished intraspecific and interspecific competition. Plant roots recruited rhizosphere microorganisms from the soil; thus, “new-phylum” appeared in treatment groups. Some beneficial microbes also emerged in treatment groups, such as
There have been some critical studies of chlorine dioxide disinfection in agriculture applications (Layman et al. 2020; Ramsey and Mathiason 2020). We aim to extend their achievements by assessing the effects on the rhizosphere microbial community and plants after flood irrigation with chlorine dioxide. The advantage of our field experiments is that effects can be evaluated through rhizosphere microbes and crops at different dose gradients. However, a more persuasive mechanism of the presented generalization requires further research with more samples from various sites, using various crops and elucidating temporal variation in soil microorganisms and crop quality.