Introduction

Winter wheat (Triticum aestivum L.) is a nutrient-rich food crop that accounts for 26% of the global agricultural output and feeds > 35% of the global population, and is of great significance for nutrition security1. In China, wheat cultivation accounts for approximately 22% of the total crop area, and the Loess Plateau is one of the most important cultivation areas, with wheat cultivation accounting for 44% of the total cultivated land. As Loess Plateau provides the main source of staple food for residents in this region, it is pivotal to maintain a stable wheat production in the Loess Plateau2,3,4.

Tillage practices have profound effect on soil environment and thereby on crop yield5,6. In dryland wheat field, deep ploughing, subsoiling tillage, or rotary tillage in fallow period (rainy season) has been extensively applied to the recovery of soil moisture7,8. Conservation methods such as subsoiling (SS) and no-tillage (NT) can reduce soil disturbance and increase the water-use efficiency of crops, thereby improving the physical and chemical properties of the soil, water-use efficiency, and crop yield9. Further, SS combined with organic fertilizer not only significantly increases the soil microbial biomass carbon and nitrogen, as well as soil enzyme activity, but also improves total dry matter accumulation and water use efficiency of wheat10,11. However, lower grain yield of winter wheat and water use efficiency were also reported in previous studies1,12. Therefore, understanding the mechanism of soil biochemistry is important for sustainable production of winter wheat under different tillage practices.

Rhizosphere soil is a complex and dynamic system that creates ecological niches for various types of microorganisms and is thus responsible for crop growth and development13. The soil microbiome is a key factor in ecosystem functions and contributes greatly to crop health14. The activity of the soil microbiome is a crucial part of the soil function in organic mineralization, thereby providing phosphorus, nitrogen, and potassium in agroecosystems15. Xia et al.16 illustrated that different tillage methods have substantial impacts on the diversity and composition of bacteria in the rhizosphere soil of wheat and that the less destructive tillage methods (SS-SS and NT-NT) could preserve the integrity of soil bacteria. They reported that SS-SS was the most effective tillage method for accumulating soil water, maintaining the balance of aerobic and anaerobic bacteria, and enhancing the metabolic ability of rhizosphere bacteria. In addition, the rhizosphere microbiome secretes effect or proteins that evade the plant immune system and successfully colonize the plant roots17. Therefore, the interactions between rhizosphere environmental microbes play an essential role in crop growth.

Our previous studies found that soil available nutrients were maintained at higher levels under the conditions of NT and SS than under DT8. SS is a good measure to improve soil nitrogen, carbon, and nutrient availability in dryland wheat field18. However, the effects of different tillage methods on rhizosphere microorganisms and metabolites during dryland wheat growth remain unknown. The present study aimed to discover the key soil microorganisms as well as key metabolites in dryland wheat soil treated with three tillage modes (NT, SS, and DT) using metagenomics and metabolomics. In addition, the relationship between microorganisms and differential metabolites in wheat rhizosphere was analyzed. The present study results can provide a theoretical basis for improving wheat yield in dryland agriculture.

Materials and methods

Research region

The experimental site is located at National Experimental Station of Hongbu, Institute of Wheat Research, Shanxi Agricultural University (Linfen, Shanxi, China, 111° 33′ 07″ E, 36° 13′ 02″ N). The altitude of this site is 457.9 m, and the average annual rainfall is 468.5 mm (with about 65% occurring during fallow period from July to September and only 35% during the growth period from October to June). The average soil water content in 0-20 cm soil layer (W/W) was 12.3% across all growth stages. Annual evaporation is 1829.4 mm. Annual sunshine duration is 2416.5 h with a frost-free period of 184 d. Here, ≥ 0 °C effective accumulative temperature is 4617.5 °C; and ≥ 10 °C effective accumulative temperature is 4151.0 °C. The soil type of the experimental site was calcareous cinnamon soil, and soil properties and nutrient conditions before the experiment were as follows: pH 8.63, organic carbon 11.67 g/kg, total nitrogen 0.93 g/kg, and total phosphorus 0.73 g/kg. No irrigation was applied across all stages in the field.

Experimental design and management

The field experiment has been conducted since 20168. Winter wheat ripens once a year; it is sown in early October and harvested mechanically in early June of the following year, leaving a stubble of 10-15 cm in height. The straw was then crushed, mulched, and returned to the field. Three different tillage modes (NT, no tillage; SS, subsoiling; DT, deep tillage) were used in a random complete block design with three replicates (50 m long× 10 m wide) in mid-July. In the NT treatment, no tillage was performed until subsequent crop planting. SS was initiated at a depth of 30-35 cm using a sub-sowing machine (model number: ISZL-300, Shandong Aolong Agricultural Machinery Manufacturing Co., Ltd., Shandong, China), and DT was started at a depth of 30-35 cm using a plow machine (model number: 1 L-320, Shanxi Jishan Agricultural Machinery Manufacturing Co., Ltd., Shanxi, China).

Winter wheat (Triticum aestivum cv. Jinmai 92) was sown at a seeding rate of 150 kg/ha in early October. Sowing and fertilization were performed simultaneously, and approximately 750 kg/ha Apollo compound fertilizer (N: P2O5: K2O= 22:16:5, total nutrient content ≥ 40%) was applied as the basal fertilizer for all treatments. No topdressing was applied in the later stage. The later management (Prevention and control of diseases, pests and grasses) of all tillage treatments was consistent.

Soil sampling

Three replicates for each treatment were collected randomly in April 2021. At the grain filling stage, the rhizosphere soil samples were collected from the roots of wheat. Wheat roots were manually excavated along with the surrounding soil. Impurities and loose soil were picked and shaken off gently, and the rhizosphere soil attached to the root surface was collected using a brush into sterile tube (50 mL). Then, one part of the samples (1 g) was stored in a -80℃ refrigerator as soon as possible for subsequent analysis of microbial community structure. The other portion (1 g) was used for metabolomic analysis.

Sequencing and microbiome analysis

Soil samples (250 mg) collected from wheat rhizosphere of NT, SS and DT treatment groups were sent to Personal biotechnology Co., Ltd. (Shanghai, China) for 16S rRNA gene sequencing and ITS sequencing. Briefly, an OMEGA soil DNA kit (M5635-02, Omega Bio-Tek, USA) was used to isolate total genomic DNA from each rhizosphere sample. The quantity and quality of the extracted DNA were assessed using a NanoDrop NC2000 spectrophotometer (Thermo Fisher Scientific, USA) and 1.5% agarose gel electrophoresis, respectively. For bacterial community analysis, the V3-V4 region of the 16S rRNA gene was amplified using the forward primer (V3V4-F: ACTCCTACGGGAGGCAGCA) and the reverse primer (V3V4-R: GGACTACHVGGGTWTCTAAT). For fungal community analysis, the ITS sequence was amplified with the forward primer (ITS1-F: GGAAGTAAAAGTCGTAACAAGG) and the reverse primer (ITS1-R: GCTGCGTTCTTCATCGATGC). PCR amplification products were purified using Vazyme VAHTSTM DNA Clean Beads (Vazyme, Nanjing, China) and quantified using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen, USA). Afterwards, the amplification products were pooled at equal concentrations, followed by 16S rRNA genes and ITS sequencing using the Illumina NovaSeq 6000 SP Reagent Kit (500 cycles).

QIIME 2 software (version 2019.4, https://qiime2.org) was used to analyze the sequencing data according to Bolyen et al.19. Briefly, raw sequencing data were quality-filtered, denoised, assembled, and chimeras removed using the DADA2 method to obtain unique amplicon sequence variants (ASVs). Subsequently, based on the classify-sklearn algorithm, the Greengenes reference database classifier20 (release 13.8) and UNITE reference database classifier (release 8.0) were used to annotate operational taxonomic units (OTUs) from 16S rRNA gene sequencing and ASVs from ITS sequencing, respectively. Alpha and beta diversities were determined using the QIIME 2 software. Chao1 and Observed species indices indicated species richness, Shannon and Simpson indices reflected species diversity, Faith’s PD index represented species diversity based on evolution, and Pielou’s evenness indicated species evenness. The heatmap package of R software (version 3.3.1, https://cran.R-project.org) was employed to analyze the different microbiota taxa in the different groups, and the Python LEfSe package in R software (version 3.3.1) was used to identify crucial biomarkers among the groups21. Finally, the functions of the identified soil microbiota were predicted using the PICRUSt2 software and analyzed using the metagenome Seq package in R software (version 3.3.1).

Isolation of metabolites and metabolomics analysis in rhizosphere

The rhizosphere samples (200 mg) were added with 600 μL pre-cooled 2-chloro-L-phenylalanine (4 ppm) in methanol. After vortexing for 60 s, 100 mg of glass beads was added, and the samples were ground at 60 Hz in a tissue grinder for 90 s. After sonication for 30 min at room temperature, as well as on ice for 30 min, the mixture was centrifuged at 12000 rpm for 10 min at 4 °C, and then the supernatant was filtered through a 0.22-μm membrane. The filtered samples were subjected to liquid chromatography-mass spectrometry (LC-MS) detection.

A Vanquish UHPLC system (Thermo Fisher Scientific) equipped with an ACQUITY UPLC® HSS T3 column (1.8 µm, 2.1 × 150 mm; Waters, Milford, MA, USA) and a mass spectrometer (Orbitrap Exploris 120; Thermo Fisher Scientific) was used for LC-MS. The temperature of column was maintained at 40 °C, the flow rate was 0.25 mL/min, and the injection volume was 2 μL. Additionally, the mobile phases for positive were 0.1% formic acid in acetonitrile (v/v, B2) and 0.1% formic acid in water (A2); and for negative were 5 mM ammonium formic water (A3) and acetonitrile (B3). The elution program was set as follows: 0-1 min, 2% B2/B3; 1- 9 min, 2%-50% B2/B3; 9-12 min, 50%-98% B2/B3; 12-13.5 min, 98% B2/B3; 13.5-14 min, 98%-2% B2/B3; 14-20 min, 2% B2 (positive) or 14-27 min, 2% B3 (negative). The spray voltage of MS for positive and negative modes was respectively 3.5 kV and 2.5 kV; the capillary temperature was 325 °C; as well as sheath gas and auxiliary gas were set at 30 and 10 arbitrary units. Full scanning was performed at a resolution of 60000, and a scanning range of 100-1000 m/z.

The original data obtained from LC-MS were converted to a mzXML format using Proteowizard software (v3.0.8789), and then peaks detection, peaks filtration, and peaks alignment were performed using the “xcms” package of R with the parameters of bw = 5, ppm = 15, peakwidth = c (5, 30), mzwid = 0.015, mzdiff = 0.01 and method = “centWave.” Then, the metabolites were identified according to the public databases HMDB (http://www.hmdb.ca/), LipidMaps (http://www.lipidmaps.org/), MassBank (https://www.massbank.jp/), mzcloud (https://www.mzcloud.org/), and KEGG (http://www.genome.jp/kegg/), and the parameter was set as ppm < 30 ppm. The “ropls” package in R was used for all multivariate data analyses and modelling, and significantly differential metabolites were screened based on the thresholds of VIP > 1 and P < 0.05. Finally, the screened differential metabolites were subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. And the metabonomic data in the form of the relative quantification (peak area) were used for all subsequent analysis.

Conjoint analysis of the microbiome and metabolomics data

Spearman’s correlation coefficient and p-values between soil microbiota (including bacterial and fungal communities) at the genus level and differential metabolites were analyzed using Mothur software, and a correlation heatmap was drawn using R software. Additionally, the relationships based on |rho| > 0.8 and P < 0.01 were selected for the construction of the correlation network, which was visualized using Cytoscape software.

Results

Structure of soil bacterial communities

According to the 16S rRNA gene sequencing results, there were 12566 OTUs, 13376 OTUs and 12024 OTUs in the DT, NT, and SS groups, respectively, and 3009 OTUs overlapped in the three groups (Fig. 1A). Principal Coordinate Analysis (PCoA) revealed obvious clustering of soil bacterial communities in the DT, NT, and SS groups (Fig. 1B). These results suggest a high depth and reliability of the sequencing results, which can be used for further analyses.

Fig. 1
figure 1

Soil bacterial communities under different tillage practices. (A) Venn diagram (OTUs); (B) Principal coordinate analysis (PCoA); (C) Alpha-diversity difference; (D) Distribution of five most dominant phyla. Lowercase indicates significant difference at 0.05 level; (E) Distribution of top 30 dominant genera. NT: no tillage; SS: subsoiling; DT: deep tillage.

Full size image

Good’s coverage index of each group was approximately 0.97 (Fig. 1C), which implied that the sequencing results of each sample contained soil bacterial communities. There were no significant differences in the Chao1, Faith’s PD, Shannon, Pielou’s evenness, Simpson, and observed species values among the DT, NT, and SS groups (P > 0.05, Fig. 1C). Therefore, it can be inferred that three tillage modes (DT, NT, and SS) had no significant effects on the α diversity of soil bacterial communities on wheat roots.

At the phylum level, the dominant phyla of the annotated OTUs were Proteobacteria, Acidobacteria, Gemmatimonadetes, Actinobacteria, and Firmicutes (Fig. 1D). Among them, the abundance of Proteobacteria in the NT, SS, and DT groups was respectively about 30.74%, 31.04%, and 32.40%; and the Actinobacteria abundance was 18.86%, 17.3% and 20.15% in the NT, SS, and DT groups. Furthermore, the abundances of Acidobacteria/Gemmatimonadetes/Firmicutes in the NT, SS, and DT groups were approximately 17.45%, 7.63%, 10.97%, 20.54%, 9.86%, and 3.44% and 16.06%, 10.94%, and 3.42%, respectively. These results indicate that compared with the DT group, SS decreased the Actinobacteria abundance significantly (P < 0.05), while increasing Acidobacteria (P < 0.05), Firmicutes abundance, whereas reduce the abundance of Gemmatimonadetes (Fig. 1D).

The top 30 soil bacterial genera were analyzed, including Subgroup-6, Bacillus, Rokubacteriales, MND1, Sphingomonas, Nitrospira, Longimicrobiaceae, Solirubrobacter, Nocardioides, JG30-KF-CM45, and Gaiella (Fig. 1E). The abundance of Subgroup-6 in the NT, SS, and DT groups was 11.21%, 12.22%, and 8.74%, respectively, and the Bacillus abundance was 9.16%, 2.65%, and 2.35% in the NT, SS, and DT groups, respectively. Additionally, the Rokubacteriales (3.98%) and MND1 (3.04%) abundance were the highest in the SS group, the Sphingomonas abundance (3.33%) was the highest in the DT group, and Nitrospira (0.98%), Solirubrobacter (0.87%), and Gaiella (0.70%) were the lowest in the NT group (Fig. 1E).

LDA effect size (LEFSe) analysis was used to identify biomarkers among the groups at different levels. At the phylum level, Gemmatimonadetes were especially abundant in the DT group, and Acidobacteria, Nitrospirae, Planctomycetes, and Rokubacteria were the primary phyla in the SS group. Conversely, Firmicutes, and Mortierellomycota was more abundant in the NT group (Fig. 2A). At the genus level, we found that Cryptosporangium, Crossiella, Umezawaea, AKYH767, OPB56, Rhodothermaceae, OLB13, Candidatus-Pacebacteria, Bdellovibrio, Leptothrix, and CCD24 were signature bacterial communities in the NT group; and Nocardia, Aeromicrobium, Chthonomonas, Phenylobacterium, Neorhizobium, Sphinhopyxis, Silvanigrella, and Hydrogenophaga were important species in the DT group; as well as the crucial genera in the SS group were RB41, MB-A2-108, Rubrobacter, Lachnospiraceae-NK4A136-group, Latescibacteraceae, Nitrospira, bacteriap25, Aquicella, JTB255-marine-benthic-group, Rokubacteriales, and ADurb-Bin063-1 (Fig. 2A).

Fig. 2
figure 2

Bacterial biomarkers and differential pathways among different tillage practices. (A) LDA effect Size (LEFSe) analysis; (B) Differential pathways enriched between the NT vs. DT, the SS vs. DT, and the SS vs. NT. NT: no tillage; SS: subsoiling; DT: deep tillage.

Full size image

The annotated soil bacterial communities were subjected to functional analysis, and it was found that the soil bacterial communities were related to “amino acid biosynthesis”, “carbohydrate biosynthesis”, “carboxylate degradation”, “fatty acid and lipid biosynthesis”, “fermentation”, “TCA cycle”, “cofactor, prosthetic group, and vitamin biosynthesis”, and “electron transfer” (Fig. S1). We used the metagenomeSeq method to identify significantly different metabolic pathways. PWY-7377 (cob(II) yrinatea, c-diamide biosynthesis I (early cobalt insertion)), PWY-6654 (phosphopantothenate biosynthesis III), PWY-6349 (CDP-archaeol biosynthesis), PWY-6350 (archaetidylinositol biosynthesis), PWY-5532 (adenosine nucleotides degradation IV), PWY-5198 (factor 420 biosynthesis), PWY-7286 (7-(3-amino-3-carboxypropyl)-wyosine biosynthesis), and PWY-5507 (adenosylcobalamin biosynthesis I (early cobalt insertion)) was the differential functional pathways between NT vs. DT group (Fig. 2B). LACTOSECAT-PWY (lactose and galactose degradation I) was the differentially expressed pathway between the SS vs. DT group (Fig. 2B). Moreover, PWY-5392 (reductive TCA cycle II), PWY-6915 (pentalenolactone biosynthesis), ORNDEG-PWY (superpathway of ornithine degradation), PWY-5507, PWY-7377, PWY-7644 (heparin degradation), and THREOCAT-PWY (superpathway of L-threonine metabolism) were the differential pathways between NT vs. SS group (Fig. 2B).

Structure of soil fungal communities

ITS sequencing results showed that 516 OTUs, 500 OTUs, and 417 OTUs were observed in the NT, SS, and DT groups, respectively, which included 200 shared OTUs among the three groups (Fig. 3A). The PCoA results showed that the samples in the NT, DT, and SS groups were well-aggregated (Fig. 3B), indicating that the sequencing was reliable and could be used for further analyses. Good’s coverage values for the NT, SS, and DT groups were all close to 1 (Fig. 3C), indicating that the sequencing results of each sample covered all soil fungal communities. No significant differences in the values of Pielou’s evenness, Chao1, Shannon, Simpson, as well as Observed species indexes were observed among the NT, SS, and DT groups (P > 0.05, Fig. 3C), which implied that the α diversity of soil fungal communities in wheat cultivation could not be significantly altered by the three tillage modes (NT, SS, and DT).

Fig. 3
figure 3

Soil fungal communities under different tillage practices. (A) Venn diagram (OTUs); (B) Principal coordinate analysis (PCoA); (C) Alpha-diversity; (D) Distribution of top5 dominant phyla. Lowercase indicates significant difference at 0.05 level; (E) Distribution of top30 dominant genera. NT: no tillage; SS: subsoiling; DT: deep tillage.

Full size image

We further explored the changes in the composition of specific soil fungal communities at both the phylum and genus levels. As shown in Fig. 3D, Ascomycota, Basidiomycota, Glomeromycota, Mortierellomycota, and Mucoromycota were the five most dominant phyla in the soil fungal communities at the phylum level. Relative to the DT group (77.41%), the Ascomycota abundance was higher in the SS group (85.24%), whereas the Basidiomycota abundance in the NT group (1.41%) was lower than that in the DT group (4.64%), with non-significant difference. The abundances of Mortierellomycota were 2.43%, 0.98%, and 0.80% in the NT, SS, and DT groups, respectively. At the fungal genus level, top 30 fungal genera in NT, SS, and DT were analyzed, such as Staphylotrichum, Coniochaeta, Dichotomopilus, Phaeosphaeria, Immersiella, Fusarium, Mycosphaerella, Trichoderma, Humicola, Cladorrhinum, Cercophora, and Preussia (Fig. 3E). The relative abundance of Staphylotrichum in the SS group reached 11.63%, while the abundances of Coniochaeta, Dichotomopilus, and Immersiella in the DT group were 14.87%, 10.99%, and 7.03%, respectively. Phaeosphaeria abundance was approximately 9.84% in the NT group, which was higher than that in the other two groups (Fig. 3E). For Mycosphaerella and Trichoderma, their abundance in the NT, SS, and DT groups were approximately 5.55%, 6.60%, 3.10%, 1.25%, and 0.63%/1.07%, respectively. Additionally, the relative abundances of Humicola, Cladorrhinum, and Preussia in the SS group were 4.30%, 4.84%, and 4.43%, respectively, which were higher than those in the other two groups. Compared with the SS (0.20%) and DT (0.10%) groups, the Cercophora abundance was highest in the NT group (4.59%).

Based on the LEFSe results, we found that Ctenomyces was a crucial fungal species in the SS group and Pezizella, Cordyceps, Monocillium, and Subulicystidium were the signature fungal genera in the DT group. The biomarker fungal genera in the NT group were Alfaria, Stilbella, Metacordyceps, Diutina, Candida, Pichia, Cyphellophora, and Pyrenochaetopsis (Fig. 4A).

Fig. 4
figure 4

Fungal biomarkers and differential pathways among different tillage practices. (A) LDA effect Size (LEFSe) analysis; (B) Differential pathways enriched between the SS vs. NT. NT: no tillage; SS: subsoiling; DT: deep tillage.

Full size image

Finally, functional analysis was performed on these soil fungal communities, and it was predicted that the annotated soil fungal communities were associated with “carbohydrate biosynthesis”, “amino acid biosynthesis”, “electron transfer”, “respiration”, “nucleoside and nucleotide biosynthesis”, “pentose phosphate pathways”, “carboxylate degradation”, and “TCA cycle” (Fig. S2). Gluconeogenesis I (Gluconeo-PWY) was identified as a differential pathway as a differential pathway between the NT vs. SS (Fig. 4B).

Identification of differential metabolites and functional analysis

Metabolomic analysis (Fig. 5A) showed that the proportion of characteristic peaks with a relative standard deviation (RSD) < 30% in the quality control samples was 83.2%, indicating that the metabolomics data were reliable and conducive for the detection of biomarkers. Orthogonal Projections to Latent Structures Discriminant Analysis (OPLS-DA) showed that metabolites in the NT, SS, and DT treatments were significantly separated (Fig. 5B), which implied that the three tillage modes (NT, SS, and DT) had substantial effects on the metabolic profiles of wheat rhizosphere and could be used for subsequent secondary structure analysis.

Fig. 5
figure 5

Differential metabolites among different tillage practices. (A) Quality assessment of the quality control sample; (B) Orthogonal projections to latent structures discriminant analysis (OPLS-DA); (C) Numbers of the differential metabolites; (D) Volcano plots of the differential metabolites. NT: no tillage; SS: subsoiling; DT: deep tillage.

Full size image

According to the thresholds of P < 0.05, VIP > 1, 47, 29, and 40 differential metabolites were respectively identified between the DT vs. SS, the DT vs. NT, and the SS vs. NT (Fig. 5C). Compared with the SS group, 21 down-regulated metabolites and 26 up-regulated metabolites (N-methyl-2-pyrrolidinone, spermidine, pelargonic acid, isovitexin, and 3,4-dihydroxymandelic acid) were found in the DT group. Compared with the NT group, there were 16 down-regulated metabolites (Terephtalate, L-Dopa, and 3-Oxo-5beta-cholanate) and 13 up-regulated metabolites (5’-methylthioadenosine, and 3-methylthiopropionic acid) in the DT groups (Fig. 5C, D). Additionally, compared with the NT group, 27 down-regulated metabolites (isovitexin, terephthalate, spermidine, and N-methyl-2-pyrrolidinone) and 13 up-regulated metabolites (pentadecanoic acid) were identified in the SS group (Fig. 5C, D). The clustering heatmaps of all identified differential metabolites in the different groups are shown in Fig. S3.

The identified differential metabolites were subjected to KEGG pathway enrichment analysis. Specificallt, 57, 96, and 81 KEGG pathways were enriched for the differential metabolites in the DT vs. NT, DT vs. SS, and SS vs. NT comparisons, respectively (Fig. 6). The differential metabolites between the DT vs. NT were mainly involved in the “biosynthesis of plant secondary metabolites”, “catecholamine transferase inhibitors”, “synthesis and degradation of ketone bodies”, “cocaine addiction”, as well as “cysteine and methionine metabolism” (Fig. 6A). The significantly enriched pathways between the DT vs. SS were “galactose metabolism”, “ABC transporters”, “taste transduction”, “beta-alanine metabolism”, and “central carbon metabolism in cancer” (Fig. 6B). The mainly enriched pathway between the SS vs. NT was related to “ABC transporters”, “phosphotransferase system (PTS)”, “beta-alanine metabolism”, “taste transduction”, and “starch and sucrose metabolism” (Fig. 6C).

Fig. 6
figure 6

Kyoto encyclopedia of genes and genomes (KEGG) pathways map of the identified differential metabolites between the NT vs. DT (A), SS vs. DT (B) and SS vs. NT (C). The larger the dot, the greater the impact. Color is related to P value, and the darker the color, the smaller the P value. NT: no tillage; SS: subsoiling; DT: deep tillage.

Full size image

Correlation between the crucial bacterial communities, fungal communities and differential metabolites

After removing the duplicates, 72 differentially expressed metabolites were obtained from the NT, SS, and DT groups, combined with the 30 differential bacterial genera and 13 differential fungal genera. Among the differential metabolites, norselegiline and pentadecanoic acid had fifteen correlations on average with bacterial genera (Fig. 7A). Especially, pentadecanoic acid had the most associations including ten positive correlations and ten negative correlations. Differential metabolites had four correlations on average with fungal genera (Fig. 7B). Examining these fungal genera, Ctenomyces had the most associations including twenty-nine positive correlations and ten negative correlations with differential metabolites.

Fig. 7
figure 7

Spearman’s correlation analysis between bacterial (A) and fungal (B) genus communities and differential metabolites showed in the heatmap. Red represents a positive correlation, while blue represents a negative correlation. The darker the color, the higher the correlation. * and ** indicate a significant correlation at P < 0.05 and P < 0.01, respectively.

Full size image

Based on the network of bacterial genera and metabolites, glutaric acid was positively correlated with Bdellovibrio and Aeromicrobium but negatively correlated with Aquicella (Fig. 8A). Phenylobacterium was negatively correlated with 1-hexadecanol, norselegiline, and azelaic acid; as well as Spingopyxis was negatively correlated with azelaic acid, (R)-salsolinol, pantothenic acid, and pyridoxal-phosphate; and pantothenic acid and pyridoxal-phosphate were positively correlated with Latescibacteraceae and JTB255-marine-benthic-group, respectively (Fig. 8A). 9(S)-HPODE was negatively correlated with RB41, MB-A2-108, Nitrospira, bacteriap25, and Rokubacteriales, whereas Rokubacteriales and RB41 were both positively correlated with D-fructose, and RB41 was also negatively correlated with lithocholic acid and terephthalate (Fig. 8A). Moreover, CCD24 positively correlated with (R)-3-hydroxybutyric-acid and 1,7-dimethyluric-acid (Fig. 8A). From the network of fungal genera and metabolites, we found that Cordyceps was positively correlated with fluvoxamine but negatively correlated with 1-hexadecanol, and Pichia, Candida, and Diutina were positively correlated with methyl jasmonate, p-synephrine, and terephthalate (Fig. 8B).

Fig. 8
figure 8

Correlation networks between bacterial (A) and fungal (B) genus communities and differential metabolites. Red lines indicate positive correlation, and green lines indicate negative correlation. Yellow circles indicate crucial bacterial and fungal genera; and green circles represent differential metabolites.

Full size image

Discussion

The highly diverse composition of rhizosphere microbial communities influences the adaptability, productivity, and growth of plants and plays important roles in maintaining crop physiology, nutrient uptake, abiotic and biological stress resilience, and defense22. Different tillage practices have profound effects on soil and plant-associated microorganisms, thereby affecting soil fertility and crop growth6. Our previous data showed that SS and NT could improve soil aggregate structure and realize the synergistic effect of soil carbon and nitrogen retention compared to DT8. However, the effects of different tillage modes on the soil microbiome and metabolites in the wheat rhizosphere remain unclear. In this study, 30 differential bacterial genera, 13 differential fungal genera, and 72 differential metabolites were identified in the NT, SS, and DT groups using 16S rRNA gene sequencing, ITS sequencing, and metabolomics, respectively.

Our sequencing results showed that the three tillage modes (DT, NT, and SS) had no significant effects on the α diversity of soil bacterial and fungal communities in wheat, which is in accordance with a previous study on maize agroecosystems with the tillage practices of NT, DT, and rotary tillage23 and Mediterranean rainfed conditions with minimum tillage (MT) and no-till (NT) practices24. Similarly, Li et al.25 also reported tillage practices had no significant difference on the alpha diversity of bacterial community. This may due to the intraspecies stability of bacterial species with respect to soil disturbances. However, conservation tillage (NT and SS) increased fungal richness and diversity compared with DT treatment in the present study. It is possible that less physical disturbance to soil created suitable microhabitat for fungal communities than bacterial communities26. Tillage methods significantly altered the compositions of bacterial and fungal communities at the phylum and genus levels. Mainly (about 89.7%) keystone species were derived from Proteobacteria, Bacteroidetes, Acidobacteria, and Gemmatimonadetes27. At the phylum level, Gemmatimonadetes and Actinobacteria were more abundant in the DT group. Acidobacteria and Ascomycota were more abundant in the SS group, which are considered vital decomposers of soil organic matter28. Firmicutes and Mortierellomycota were higher in the NT group. Actinomycetes are enriched in arid soils and root environments in different crops, such as peanuts and other angiosperms29. Besides, the proportion of Gemmatimonadetes was also higher in arid soils30. The decrease in soil water content was accompanied by a significant decrease in the proportion of active Acidobacteria31, and Ascomycetes members play dominant roles in the decomposition process of straw residues in cultivated land and are responsible for the degradation of residues in the soil32. Firmicutes belongs to eutrophic microorganisms33, and Mortierellomycota abundance decreases during the P cycle in conifer forests34. Taken together, we speculate that different tillage methods may change the soil water content and fertility by affecting the abundance of Actinobacteria, Acidobacteria, Gemmatimonadetes, Firmicutes, Mortierellomycota and Actinomycete. These reports, together with our results, suggest that the three tillage methods significantly alter the composition of soil rhizosphere microorganisms in wheat.

Core groups in soil have the ability to predict environmental preferences and redistribute to adapt the ecological environment status35. LEFSe indicated that Cryptosporangium, Crossiella, Rhodothermaceae, Bdellovibrio, Leptothrix, Stilbella, Diutina, Candida, Pichia, Cyphellophora, and Pyrenochaetopsis were considerably more abundant in NT practice. Fungi favoured by no tillage could be root endophytes or species suitable to utilize intact decaying roots6. The essential microbiota for the SS were RB41, Rubrobacter, Latescibacteraceae, Nitrospira, JTB255-marine-benthic-group, Rokubacteriales, and Ctenomyces. We identified Nitrospira as the nitrite-oxidizing bacteria, as also reported by2,3,4. In the study of Chen et al.36, RB41 was detected under N conditions, and confirmed of the important role in mediating the crop N uptake derived from soil. The crucial genera enriched in the DT groups were Nocardia, Aeromicrobium, Chthonomonas, Phenylobacterium, Neorhizobium, Sphinhopyxis, Cordyceps, Monocillium, and Subulicystidium. Nocardia species can produce several kinds of biosurfactants such as lipopeptides and glycolipids to degrade hydrocarbon compounds37. Therefore, soil disturbances affected by tillage would establish new niches and select different microbial composition38.

Metabolites in the rhizosphere are related to the soil microbiota, and their interactions can promote the healthy growth of plants39. Plant secondary metabolites are not only a series of beneficial natural products but also an important part of plant defense systems against pathogenic attacks and environmental stress, which help construct ecological relationships between plants and other organisms40. Amino acids and lipids were most redundant metabolites affected by tillage practices25, which were related to nitrogen cycling. In the present study, functional analysis obtained soil bacterial and fungal communities were associated with the amino acid biosynthesis and carbohydrate biosynthesis. These suggested that amino acids and carbohydrate are the main metabolite influencing the soil bacterial and fungal communities involved in carbon metabolism25. A previous study showed that increased levels of certain amino acids (e.g., alanine, arginine, and ornithine) in Sesuvium portulacastrum L. under metal toxicity could be attributed to high levels of stress tolerance41. Hou et al. reported that hydrogen-rich water promoted the monochromatic bulb formation of lilium davidii var. unicolor by regulating sucrose and starch metabolism42. Glucose plays an important role in metabolism. Sugar is an important assimilative product of plant photosynthesis, and its anabolism and operational distribution directly influence plant growth and development as well as crop yield and quality2,3,4. ABC transporters have essential functions in the transport of biomolecules across membranes and regulate the interactions between the composition of root secretions and rhizosphere microbiota43. Therefore, tillage methods may regulate the related characteristics of wheat rhizosphere soil through the pathways of the biosynthesis of plant secondary metabolites, cysteine and methionine metabolism, galactose metabolism, ABC transporters, beta-alanine metabolism, synthesis and degradation of ketone bodies, and starch and sucrose metabolism enriched by the identified differential metabolites, thus influencing the growth and yield of wheat. However, the specific effects of the different metabolites and pathways involved in wheat growth should be further explored.

Collectively, the present study suggested that soil bacterial communities was highly associated with the changes in wheat metabolites that may affect the wheat metabolome directly or indirectly, as reported in sugarcane rhizosphere by Huang et al.44. Yamazaki et al.45 also reported close relationship between the bacterial communities and mineral properties in the soybean rhizosphere. In the study of Li et al.25, soil bacterial and fungal communities were significantly correlated with soil organic carbon. Therefore, we speculated different stratified structure of nutrients caused by tillage resulted in adaptive changes of bacterial communities and metabolites on wheat roots. Bacteria can produce glucoheptonic acid, gluconic acid and cellobiose to cross-feed this rhizosphere bacteria46. In the present study, glutaric acid was positively correlated with Bdellovibrio and Aeromicrobium but negatively correlated with Aquicella. Phenylobacterium was negatively correlated with 1-hexadecanol, norselegiline, and azelaic acid, whereas Spingopyxis was negatively correlated with azelaic acid, pantothenic acid, and pyridoxal phosphate. Pantothenic acid and d-fructose were positively correlated with Latescibacteraceae and Rokubacteriales, respectively. Taken together, our study implies that the interaction between the soil microbiome and metabolites may affect the soil fertility and wheat growth.

In conclusion, this study investigated the changes of microbial communities and metabolites in wheat rhizosphere in the actual field condition responses to tillage practices. Bacterial, fungal communities and metabolites were distinct between tillage groups. Proteobacteria and Ascomycota were the predominant taxonomic groups among tillage soils. Metabolites were more closely related to rhizosphere bacterial community than that of fungi community. Our findings suggested that varied the composition of microbial communities and metabolites and their interactions in the wheat rhizosphere responses to tillage may affect soil fertility and wheat growth.