Continuous cropping of potato changed the metabolic pathway of root exudates to drive rhizosphere microflora

For potato production, continuous cropping (CC) could lead to autotoxicity buildup and microflora imbalance in the field soil, which may result in failure of crops and reduction in yield. In this study, non-targeted metabolomics (via liquid chromatography with tandem mass spectrometry (LC–MS/MS)) combined with metagenomic profiling (via high-throughput amplicon sequencing) were used to evaluate correlations between metabolomics of potato root exudates and communities of bacteria and fungi around potato plants to illustrate the impacts of CC. Potato plants were grown in soil collected from fields with various CC years (0, 1, 4, and 7 years). Metabolomic analysis showed that the contents and types of potential autotoxins in potato root exudates increased significantly in CC4 and CC7 plants (i.e., grown in soils with 4 and 7 years of CC). The differentially expressed metabolites were mainly produced via alpha-linolenic acid metabolism in plant groups CC0 and CC1 (i.e., no CC or 1 year CC). The metabolomics of the groups CC4 and CC7 became dominated by styrene degradation, biosynthesis of siderophore group non-ribosomal peptides, phenylpropanoid biosynthesis, and biosynthesis of various plant secondary metabolites. Continuous cropping beyond 4 years significantly changed the bacterial and fungal communities in the soil around the potato crops, with significant reduction of beneficial bacteria and accumulation of harmful fungi. Correlations between DEMs and microflora biomarkers were established with strong significances. These results suggested that continuous cropping of potato crops changed their metabolism as reflected in the plant root exudates and drove rhizosphere microflora to directions less favorable to plant growth, and it needs to be well managed to assure potato yield.


Introduction
Potato (Solanum tuberosum L.) is an important commodity crop globally as food, feed, and industrial raw material due to its high-yield, strong tolerance against hash environment, and excellent adaptability (Camire et al., 2009).Potato is a good source of proteins, amino acids, minerals, vitamins, and dietary fiber (Camire et al., 2009).Worldwide, acreage of potato planting is increasing every year.High demands of potato create strong motivation for continuous cropping (CC), especially in China where the arable land is limited (Qin et al., 2017a).However, studies showed that the yield of potato from CC fields decreased gradually, by as much as 27, 75, and 85% in years 2, 3, and 4, respectively (Qin et al., 2017a).Such drop in yield calls into question the benefit of continuous cropping for potato.
The deterioration of crop yield during CC is defined as continuous cropping obstacle (CCO), which is most likely resulted from changes in root exudates and soil microflora (Gao et al., 2019(Gao et al., , 2021;;Han et al., 2022).Although most of root exuded metabolites have beneficial effects to the plant, for example, organic acids are known to increase nutrient mobilization (thus absorption by the roots; Khademi et al., 2010), sugars and amino acids feed the microbiota that could play beneficial roles in plant growth (Kang et al., 2015;Topalovic et al., 2020), and as CC continues, root exudates could increase allelopathic autotoxicity that are detrimental to plant growth and recruit harmful microbes to the plants (Han et al., 2022).Accumulation of autotoxic allelochemicals in soil as a mechanism of CCO has been identified in many plants (Qin et al., 2017b;Wu et al., 2018;Huang et al., 2020), which could affect the health of the plants, and the complex interactions between plants and soil microbes (Gao et al., 2019(Gao et al., , 2021)).Changes in types and amounts of allelochemicals secreted by plant roots can have impact on the growth and reproduction of soil microbes and change the ratio of pathogenic to beneficial microorganisms in the soil, leading to soil microflora imbalance (Xia et al., 2015;Bonanomi et al., 2016;Jin et al., 2019).Plant autotoxicity was also reported to cause CCO (Xiang et al., 2022).For instance, autotoxicity has been reported to cause replant failure in the continuous cropping of Angelica sinensis (Xin et al., 2019), Lilium davidii var.unicolor (Wu Z. J. et al., 2015), Nicotiana tabacum L. (Deng et al., 2017), Panax quinquefolium (He et al., 2009), and Panax notoginseng (Yang et al., 2015).Many compounds from different root exudates were shown to be autotoxins.Imperatorin, α-spinasterol, vanillin, dibutyl phthalate, and ferulic acid were discovered to be potential autotoxic allelochemicals in Angelica sinensis (Xin et al., 2019); the accumulation of phthalic acid was reported to be one of the main CCO factors for Lilium davidii var.unicolor replantation (Wu Z. J. et al., 2015); dibutyl phthalate, diisobutyl phthalate, and diisooctyl phthalate in root exudates were shown to play important roles in autotoxicity of Nicotiana tabacum L. (Deng et al., 2017); some phenolic acids in root exudate of Panax quinquefolium and rhizosphere soil have been identified as potential allelochemicals (He et al., 2009); ferulic acid and saponin were shown to inhibit the growth of P. notoginseng (Yang et al., 2015).Studies on potato CCO revealed that high concentrations of allelopathic substances secreted by potato roots could produce allelopathy effects on the plants (Soltys-Kalina et al., 2019;Xin et al., 2022;Szajko et al., 2023), and soil microflora around the roots were changed in response to potato CC (Qin et al., 2017a).However, key questions remain to be answered: What are the main autotoxins secreted by potato roots under continuous cropping?Are they changing over time, and how?What are the relationships between soil microflora changes and allelochemical accumulation around potato plants?Answers to these questions are needed for better understanding CCO for potato and for finding ways to overcome the CCO to promote potato production.
This study aimed to understand how potato CC can alter root exudates and induce changes in rhizosphere microorganisms.Potato plants were grown in pot groups with soil collected from fields with different continuous cropping years (0, 1, 4, and 7, namely, CC0, CC1, CC4, and CC7 groups).Rhizosphere and bulk soil from each pot were collected and analyzed to assess changes of bacterial and fungal communities around the plants.Meanwhile, root exudates from the plants were collected and analyzed for changes and metabolic pathways by non-targeted metabolomics.By identifying biomarkers of root exudates and microbes (i.e., rhizosphere and bulk soil) in samples from different CC groups, the relationships between microbes and root exudates were established to clarify the chemotaxis or avoidance tendencies of microbes toward the root exudates and reveal the mechanisms of CCO for potato.

Pot experiment design
Soil for the potato pot experiments was collected in 2021 from the positioning testing site located in the National Soil Quality Stability Observation and Experiment Station of Gansu Academy of Agricultural Sciences, Gansu Province, North-west China (104 ° 36′E, 35 ° 35ʹN).The average altitude of the site is 1970 m, the average annual temperature is 6.2°C, the average annual precipitation is 415 mm, the average annual evaporation is 1,531 mm, and the frostfree period is 146-149 day/year.The physical and chemical properties of the soil in 2015 were organic carbon 10.62 g•kg −1 , total nitrogen 0.78 g•kg −1 , available phosphorus 10.12 mg•kg −1 , available potassium 163.8 mg•kg −1 , and pH 8.32.The characteristics of the soil changed over 7 years of CC, as shown in Table 1.Topsoil (0-20 cm) from fields of rotation (maize/potato, with maize planted the previous season) and continuous cropping of 1, 4, and 7 years (potato alone) were collected in 2021 and sieved through a 10.0 mm sieve to remove large root chips and stones after air-drying, four groups were set up and denoted by CC0 (rotation control), CC1 (continuous cropping for 1 year), CC4 (continuous cropping for 4 years), and CC7 (continuous cropping for 7 years), and each group included six pots for replication.Each pot had 2.0 kɡ air-dried soil, with fertilizers applied at 0.1 gN/kg soil, N/P = 2:1 (urea for N, and potassium dihydrogen phosphate for P).In each pot, two plants (Longshu no.3) were seeded at 9 cm apart and 4 cm from the surface.Pots were placed in the greenhouse on campus of Gansu Agricultural University from 10 May to 14 July of 2021.

Sample collection
Potato plants in the pot were taken out as a whole at the blossoming stage, 65 days after seeding, and paired rhizosphere soils (R) and bulk soils (B) were collected with the "soil adhering to fine roots after shaking" method (Huo et al., 2022); in brief, soil was first collected by shaking the roots gently till dropping of large soil aggregates ceased, and the soil was defined as bulk soil (BS); then, soil adhering to the fine roots was collected by brushing the roots with sterile brushes, which was defined as rhizosphere soil (RS).Soil samples were then split into two parts: One part was air-dried, and another was frozen at −80°C.After RS collection, the plants were put into a 250 mL conical flask, and ultrapure water was added until the roots were completely submerged.After 24 h (16 h under light and 8 h in the dark), the solution was collected and centrifuged at 1200 rpm, at 4°C for 15 min; then, the supernatant was filtrated by a 0.22 μm microfilter, and the filtrate was freeze-dried to get the root exudates, which was frozen-stored at −80°C.The root exudates alongside with soil samples from three pots (randomly selected) of each group were sent to Biomarker Technologies Co., Ltd.(Beijing, China) for metabolomics analysis and gene sequencing.

Non-targeted metabolomic analysis
Untargeted metabolites in the root exudates were determined using a liquid chromatography with tandem mass spectrometry (LC-MS/MS) platform (Biomarker Technologies Co., Ltd.).In brief,1 mL extraction liquid (methanol: acetonitrile: water = 2:2:1) was added into 50 mg of the samples and vortexed for 30 s, and steel balls were added and ground in a 45 Hz grinder for 10 min and ultrasonicated in an ice water bath for 10 min (Dunn et al., 2011).The sample was then sat at 20°C for 1 h to precipitate the proteins and then centrifugated at 4°C, 12000 rpm for 15 min.500 μL of supernatant were then moved into an EP tube and dried in a vacuum concentrator; then, 160 μL of an extraction liquid (acetonitrile: water = 1:1) was added to redissolve the sample.It was then vortexed for 30 s, ultrasonicated for 10 min (in ice water bath), and centrifugated for 15 min (4°C, 12,000 rpm).Finally, 120 μL of the supernatant were used for Ultra-High Performance Liquid Chromatography-Q Exactive (UHPLC-QE) orbital trap/mass spectrometry analysis.The LC/MS system for metabolomics analysis is composed of Waters Acquity I-Class PLUS ultra-high performance liquid tandem Waters Xevo G2-XS QT of high-resolution mass spectrometer, and the column used was from Waters Acquity UPLC HSS T3 column (Waters Corp., Milford, CT, United States).

DNA extraction and sequencing
The total genomic DNA from 0.5 g of both BS and RS samples of each group was extracted with the TGuide S96 Magnetic Soil/Stool DNA Kit (Tiangen Biotech (Beijing) Co., Ltd.) according to manufacturer instructions, and the DNA concentration of the samples was measured with the Qubit dsDNA HS Assay Kit and Qubit 4.0 Fluorometer (Invitrogen, Thermo Fisher Scientific, Oregon, United States).PCR amplification was performed for each soil DNA extract in triplicate and combined into a single composite sample.Fungal ITS region was amplified using the primer pair ITS1F (5'-CTTGGTCATTTAGAGGAAGTAA-3′) and ITS2 (5'-GCTGCGTTCTTCATCGATGC-3′) (Yang et al., 2017), and the thermal cycling conditions were 95°C for 5 min, 15 cycles of 95°C for 1 min, 50°C for 1 min, and 72°C for 1 min, followed by 72°C for 7 min.Bacterial 16S rRNA gene V3-V4 region of 16S rRNA was amplified using the primer pair 338F (5'-ACTCCTACGGGAGGCAGCA-3′) and 806R (5'-GGACTACHVGGGTWTCTAAT-3′) (Wang et al., 2020), and the thermal cycling conditions were pre-denaturation at 98°C for 2 min, denaturation at 98°C for 15 s, annealing at 55°C for 30 s, extension at 72°C for 30 s, and final extension at 72°C for 5 min (30 cycles).The PCR amplicons were gel-purified with Agencourt AMPure XP Beads (Beckman Coulter, Indianapolis, IN).The resultant PCR products were combined at equimolar concentrations and use Illumina NovaSeq 6000 (Illumina, Santiago CA, United States) for sequencing (250 × 250 bp) in Beijing Biomarker Technologies Co., Ltd.(Beijing, China).
Paired-end (PE) reads obtained from previous steps were assembled by USEARCH (version 10) (Segata et al., 2011) and followed by chimera removal using UCHIME (version 8.1) (Quast et al., 2013).The high-quality reads generated by the above steps will be used for subsequent analysis.

Data processing and analysis
The MS raw data were collected using MassLynx software (version 4.2, Waters Corp., Milford, CT, United States) and processed by the Progenesis QI software (Waters Corp., Milford, CT, United States).The METLIN database (Waters Corp., Milford, CT, United States) and an in-house database (Biomarker Technologies Co., LTD.) were used for peak annotation and identification of various compounds (Zhang et al., 2022).The projection (VIP > 1), Student's t-test (p < 0.05), and |log 2 FC| ≥ 0.58 were used as criteria to screen for differentially expressed metabolites (DEMs), and the threshold (VIP > 1, p < 0.05, |log2FC| ≥2.32) was used to screen the DEMS that varied greatly, defined as greatly differentially expressed metabolites (GDEMs).
Operational taxonomic units (OTUs) were clustered with 97% similarity using Usearch (version 10.0) software.Beta diversity analysis of samples was evaluated by QIIME2 software (version 2020.6), and non-metric multidimensional scaling (NMDS) analysis of Gower distance and sperman-approx distance was used to show the divergence of the rhizosphere microbial communities for bacterial and fungal, respectively.NMDS analysis of Hellinger distance and binarychord distance was used to show the divergence of bacterial and fungal communities in both rhizosphere and bulk soil, respectively.The linear discriminant analysis (LDA) effect size (LEfSe) was used to identify biomarker with statistical difference using R (version3.1.1,stats package) and python (version1.0.0, scipy package; Segata et al., 2011).Correlation network analysis was used to identify key species using Cytoscape (version 3.7.1)and R (version3.1.1,igraph package; Shannon et al., 2003;Zheng et al., 2018), genera with average relative abundances higher than 0.1% were subject to Spearman's correlation analysis, and bacterial and fungal genera with Spearman's correlation >0.8 or < −0.8 and significance p < 0.01 were used to establish microbial networks.
The related network between metabolites and genus was established with Pearson's correlation coefficient (PCC) > |0.8| and a p < 0.05 and visualized with Cytoscape (version 3.7.1;Pyo et al., 2022).The high-level correlation network analysis (Spearman's correlation (r > 0.8 and p < 0.05)) between top 20 microbe and DEMs was involved in metabolic pathways, the correlation heatmap between DEMs, and the significantly differentiated microbes from rhizosphere and bulk soil with t-test (p < 0.05).

Physiological indicators of plants grown in soils with different CC years
Dry weight, stem thickness, and height of plants all decreased in groups of increasing CC years (Table 2), while the amount of root secretion increased significantly, indicating CC depressed the potato growth, while promoted root secretion, which consumed large amount of photosynthates.Biomass of the plant (i.e., dry weight) and physiological indicators (stem thickness and height) suggested that plants grown with soil under the growing impact of CCO were getting less healthy.Meanwhile, the plants were producing more root secretion, especially in soil with longer CC, which were signs of plants responding to the changing soil environment.These signs were further clarified by the metabolomic analysis of the root exudates.

Metabolomic analysis of root exudates of potato with CCO
A total of 27,975 peaks were detected in 12 samples by metabolome characterization and quantitative analysis, of which 1,722 metabolites were annotated (Supplementary Table S1), and these compounds mainly included fatty acyls, prenol lipids, carboxylic acids and derivatives, and organooxygen compounds (Supplementary Figure S1).PCA analysis and OPLS-DA model showed that metabolite data from different groups (CC0, CC1, CC4, and CC7) formed clusters that are completely separated (Figures 1A,C), indicating that the metabolites from root exudates of potato grown in pots with soil from different CC fields (0, 1, 4, and 7 years, respectively) were significantly different.Analysis of the root exudates showed that CC significantly changed the chemical composition of the root exudates, and longer the continuous cropping years, greater the differences shown in the differentially expressed metabolites (DEMs).Upregulated DEMs in CC1, CC4, and CC7 were 54,142,and 184,and downregulated DEMs were 169,114,and 149 (Figures 1B,D), respectively, and the entire lists are given in Supplementary Tables S2-S4.It is worth noting that more upregulated DEMs are produced with the increase of CC years.

Enrichment of differentially expressed metabolites as affected by CC
Analysis of metabolic pathways can reveal the metabolic process leading to the production of the DEMs among groups.KEGG map of metabolic pathways (Figure 2) showed that DEMs were mainly enriched via alpha-linolenic acid metabolism, biosynthesis of plant secondary metabolites, benzoate degradation, Notably, as shown in Figure 3, biosynthesis of siderophore group non-ribosomal peptides, styrene degradation, butanoate metabolism, glycolysis/gluconeogenesis, naphthalene degradation, and polycyclic aromatic hydrocarbon degradation was all boosted in groups CC4 and CC7.Further analysis showed that alpha-linolenic acid pathway was downregulated in group CC1, while phenolic acids produced from L-phenylalanine were upregulated in groups CC4 and CC7.This indicated that the metabolic pathways of CC1 plants were different from that of CC4 and CC7 plants, and the metabolic pathways of CC4 and CC7 plants were dominated by the synthesis of autotoxins (e.g., phenols and coumarins).Fungal LEfSe analysis (Figure 4D) showed that Cladosporium in group CC4, Plectosphaerella and Metarhizium in group CC7,  Analysis of microbial correlation networks was conducted to reveal the key microbial species (e.g., bacteria and fungi) under different continuous cropping conditions.As shown in Figure 5, Unculturedbacterium-c-Subgroup-6 was identified to be a key bacterium in all CC groups, Sphingomonas was a key bacterium in groups CC0, CC1, and CC7, while its relative abundance reduced to minimum in group CC4.Key fungal species under different continuous cropping conditions were Iodophanus, Funneliformis, and Mortierella in group CC1, Aspergillus, Mortierella, and Chalara in groups CC1 and CC7, Kazachstania, Fusarium and Chaetomium in group CC4, respectively.Notably, CC4 showed different characteristics comparing to CC1 and CC7.It should also be noted that the relative abundance of Chalara and Aspergillus in groups CC4 and CC7 was all significantly higher than those in group CC0, while the opposite was observed for Mortierella, whereas continuous cropping seemed to reduce its relative abundance, which went down from that of CC0, the relative abundance of microorganisms was given in Supplementary Table S7.

Analysis of rhizosphere via bulk soil microbiota
As shown in Figures 6A,B NMDS analyses showed that the rhizosphere and bulk soil microbial communities of bacteria and fungi in each group were separated clearly, indicating that root exudates significantly influenced the formation of unique rhizosphere microbiota different from that of the bulk soil.
Bacterial LEfSe analysis between rhizosphere and bulk soil of different CC groups (Figure 6C   Fungal LEfSe analysis between rhizosphere and bulk soil of different CC groups (Figure 6D) showed that for CC0, Powellomyces, Pseudomassariosphaeria, and Entoloma were predominant in the rhizosphere, while Pyrenochaeta was predominant in the bulk soil; For CC1, Mortierella, Brunneochlamydosporium, Tetracladium, Efibulobasidium, and Plectosphaerella were predominant in the rhizosphere, while Monascus, Pseudocercospora, Sarocladium, and Neodidymella were predominant in the bulk soil; for CC4, Zygosaccharomyces was predominant in the bulk soil; for CC7, Paecilomyces, Botryotrichum, Russula, Lasiobolidium, and Dactylella were predominant in the rhizosphere, while Symmetrospora, Pseudocercospora, Phialophora, Bipolaris, Vishniacozyma, Byssochlamys, Penicillium, Zasmidium, Wickerhamomyces, Erythrobasidium, Fusarium, and Kazachstania were predominant in the bulk soil.As a general observation, it was discovered that in CC1 (1 year of continuous cropping), beneficial bacteria were enriched in the rhizosphere, while in CC7, pathogens and yeasts such as Fusarium, Pseudocercospora, and Wickerhamomyces became predominant in the bulk soil, which could be a significant contributor to the reduction in potato yield.

The relationships between microbiota and DEMs in different CC groups
Correlation network between top 20 microbes (based on relative abundance) and DEMs involved in metabolic pathways was generated.As shown in Figures 7A,B connections between rhizosphere microflora (both bacteria and fungi) to DEMs could be identified to suggest that the changes of potential autotoxins could be originated from changes of rhizosphere microflora due to continuous cropping.

Discussion
Continuous cropping obstacles (CCOs) widely exist in crops, and the mechanisms are very complex.More and more studies indicated that autotoxins secreted by roots combine with rhizosphere microbial imbalance induced by root exudates were the main reasons of CCOs (Jin et al., 2019).Autotoxin secretion is controlled by plant metabolism, and microbiota changes in rhizosphere and bulk soil are responses to it.Both factors were investigated in this study for potato CCOs.It is reasonable to believe that the analytical methods developed in this study would be equally effective for assessment of CCOs for crops other than potato.

Metabolic pathways in potato controlling root exudates affected by continuous cropping
Root exudates often change when plants are under duress and responding to environmental stressors.The resulted autotoxicity of root exudates was regarded as one of the main reasons of CCOs.As in the cases of Angelica sinensis, Lilium davidii var.unicolor, Panax quinquefolium, Nicotiana tabacum L., and P. notoginseng, the increase of autotoxic allelochemicals in root exudates due to CC was identified as a key reason for failure of replanting (He et al., 2009;Wu Z. J. et al., 2015;Yang et al., 2015;Deng et al., 2017;Xin et al., 2019).In this study, it was discovered that potato root exudates were significantly changed as a function of CC years.Compared with rotation control (CC0), the changes of root exudates of potato from group CC1 could be attributed to α-linolenic acid metabolism, benzoate degradation, biosynthesis of plant secondary metabolites, and biosynthesis of plant hormones.Especially, α-linolenic acid synthesis was downregulated.Linolenic acid was the precursor of jasmonic acid (JA), and JA is a lipogenic plant hormone that regulates the defensive responses of plants to biological and abiotic stresses (Zhao et al., 2014;Balfagon et al., 2019;Zhu et al., 2021;Pyo et al., 2022).Downregulated linolenic acid may suggest that stress levels in potato plants of group CC1 were not increased very much compared to that of group CC0.Meanwhile, for groups CC4 and CC7, metabolic pathways of styrene degradation and biosynthesis of siderophore group non-ribosomal peptides were significantly altered, and the production of phenolic acids (e.g., homogentisic acid and salicylic acid) went up significantly.Homogentisic acid is the precursor for the biosynthesis of α-tocopherol, which regulates the concentration of reactive oxygen and plant hormones in response to stress (Munné-Bosch and Alegre, 2002).Its upregulation was in sharp contrast to the downregulation of linolenic acid-JA synthesis in plants of group CC1.The effects of salicylic acid were concentration-dependent; when the concentration exceeded 0.5 mM, it was shown to significantly affect the mineral absorption of plants (Harper and Balke, 1981), stomatal movement, and chlorophyll content (Manthe et al., 1992;Pancheva et al., 1996).It was also shown to reduce protein content and photosynthetic rate in barley plants (Pancheva et al., 1996).In the range of 3-5 mM, it was shown to completely inhibit the germination of maize embryo (Guan and Scandalios, 1995).In group CC7, metabolic pathway of phenylpropanoid biosynthesis and biosynthesis of various plant secondary metabolites were also significantly changed, and the production of phenolic acids (e.g., sinapic acid, ferulic acid, 4-hydroxycinnamic acid, 4-hydroxybenzoic acid, and vanillin) and coumarin (e.g., psoralen and fraxetin) went up.Vanillin was shown to be connected to changes in microbial communities of the cucumber rhizosphere and the replant failure of eggplant due to CCOs (Chen et al., 2011;Jia et al., 2018;Zhang et al., 2018).Oxidative stress arising from elevated ferulic acid was connected to cellular dysfunction and cell death, and inhibition of the growth of the seedlings (Pergo and Ishii-Iwamoto, 2011;Mandavikia et al., 2017).4-Hydroxybenzoic acid was shown to regulate grapevine secretion and could cause replant disease (Wang et al., 2019).Psoralen and fraxetin are coumarins, whose effects on plant growth were also dosage-dependent.When the concentration of coumarin was at ~680 μM, the root growth of cucumber and maize seedlings was completely inhibited (Pergo et al., 2008).4-Hydroxycinnamic acid (p-coumaric acid) could also significantly inhibit crop growth, such as cucumber (Zhou and Wu, 2012), strawberry (Chen et al., 2020), and asparagus (Kato-Noguchi et al., 2017).The elevation of phenolic acids and coumarin production in groups CC7 clearly was one of the main reasons of yield drop in this group.These observations also confirmed that none of these autotoxic allelochemicals were unique to potato, they belong to a group of compounds that could affect various plants, and monitoring their levels could hold keys to understand CCOs in various crops.
Closer investigation revealed some complex patterns.For example, the relative abundance of Fusarium in the rhizosphere of CC crops, surprisingly, did not peak in group CC7 but in group CC4.Furthermore, Fusarium were significantly enriched in the bulk soil of group CC7.This "puzzle" could be solved from the spatial correlations between DEMs and microbiota.Our study showed that the increase of 4-methylumbelliferon (from bulk to rhizosphere) was negatively correlated with the changes in relative abundance of Fusarium in the bulk soil, which could be attributed to coumarins.Studies have shown that coumarin could have a strong inhibitory effect on Fusarium in a dosage-dependent way.At high concentrations, they could inhibit mycelial growth, sporulation, and pathogenicity-related enzyme activities (Wu et al., 2008).Hence, the accumulation of high coumarins in the rhizosphere of group CC7 could inhibit the growth of Fusarium.However, Fusarium was still highly enriched in the bulk soil, which attributed to the much-reduced yield of potato from the group CC7.Finally, stimulation of psoralen accumulation by biotic elicitors such as yeast extract and chitosan has previously been observed in the cell cultures of plant species, viz., Calendula officinalis (Wu et al., 2008), Sorbus aucuparia (Gaid et al., 2011), and Abrus precatorius (Karwasara and Dixit, 2009;Karwasara et al., 2011).In this study, psoralen was shown to be positively correlated with Kazachstania (p ≤ 0.05), Byssochlamys (p ≤ 0.05), and Wickerhamomyces (p ≤ 0.01), and these fungi were reported to promote plant growth by producing plant hormones and other growth regulators (Nassar et al., 2005;Cloete et al., 2009;Xin et al., 2009;Amprayn et al., 2012;Nutaratat et al., 2014) or interacted indirectly with symbiotic microorganisms such as arbuscular mycorrhizal fungi (AMF; Fracchia et al., 2003;Boby et al., 2008).In general, the knowledge gained on relationships between metabolites and microbial communities could provide guidance for the regulation/manipulation of rhizosphere microflora via biotechnology (e.g., microbial fertilizer) to overcome continuous cropping obstacles and improve both yield and quality of potato crops.

Conclusion
The results showed continuous cropping of potato changed the metabolic pathways in potato that significantly changed alphalinolenic acid metabolism in plants from 1 year CC field, styrene degradation, biosynthesis of siderophore group non-ribosomal peptides, phenylpropanoid biosynthesis, and biosynthesis of various plant secondary metabolites in plants from 4 to 7 years CC field, and increased phenols, flavonoids, coumarins, and alkaloids in root exudates.Continuous cropping of potato beyond 4 years changed their metabolism as reflected in the plant root exudates to drive rhizosphere microflora toward the direction of reducing beneficial bacteria and promoting harmful fungi, which need to be better controlled to reduce the impact of CCO on potato production.

FIGURE 1
FIGURE 1 Metabolomic analysis of root exudates.(A) Principal component analysis (PCA) of root exudates, where the x-axis represents the first principal component, the Y-axis represents the second principal component, and the Z-axis represents the third principal component.(B) Number of upregulated and downregulated differentially expressed metabolites (DEMs).(C) Orthogonal partial least-squares discrimination analysis (OPLS-DA) scores plot of metabolites in soil between control and treatments, the parameters of the model, including R 2 X, R 2 Y, Q 2 Y, RMSEE (root mean squared error), pre (predict the number of components), and ort (the number of orthogonal components), where R 2 X and R 2 Y denote the percentage of X and Y matrix information that can be interpreted by the OPLS-DA classification model, respectively, and Q 2 Y is calculated by cross-validation to evaluate the predictive power of the OPLS-DA model, and the closer these metrics are to 1, the better the OPLS-DA model fits the data.(D) Volcano map of differential metabolites, green dots represent downregulated metabolites, red dots represent upregulated metabolites, and gray dots represent nodifferential metabolites.

FIGURE 2 KEGG
FIGURE 2KEGG map of metabolic pathways with significant enrichment of DEMs between groups CC0 and CC1, CC4, and CC7.The x-axis is the enrich factor of the DEMs enriched in the pathway, and the y-axis lists the names of the pathways; the color depth of the dot represents the p-value, the redder the color, the more significant the enrichment, and the size of the dot represents the number of DMEs enriched.

FIGURE 3
FIGURE 3KEGG metabolic pathway map of (A) CC1 and (B) CC4 and CC7.The red marked are the DEMs, with ↑ indicating upregulated and ↓ indicating downregulated, and the blue marked are the enzymes involved in the synthesis of the metabolites.Gray shows the specific metabolic pathway in CC1, green shows the specific metabolic pathway in CC7, orange shows the specific metabolic pathway in CC4, and yellow shows the common metabolic pathways in both CC4 and CC7.

FIGURE 4
FIGURE 4 Differential analysis of soil microbial communities in group CC.(A) Bacterial non-metric multidimensional scaling (NMDS).(B) Fungal NMDS, each dot in the graph represents a sample, different colors represent different groups, and the ellipse circle represents its 95% confidence ellipse.(C) Bacterial linear discriminant analysis (LDA) effect size (LEfSe) analysis.(D) Fungal LEfSe analysis.The figures show the genera with LDA score greater than 3.5.
FIGURE 5 Correlation network diagram of each group [(A) bacteria; (B) fungi].Red lines represent significant positive relationships (Spearman's correlation, r > 0.8 and p < 0.01), and green lines denote negative relationships (Spearman's correlation, r < 0.8 and p < 0.01).The genera shown on these figures are the module hub in each network diagram (ZI > 2.5 and PI < 0.62).

FIGURE 6
FIGURE 6 Differential analysis between rhizosphere and bulk soil microbiota in each group.(A) Bacterial non-metric multidimensional scaling (NMDS).(B) fungal NMDS, each dot in the graph represents a sample, different colors represent different groups, and the ellipse circle represents its 95% confidence.(C) Bacterial linear discriminant analysis (LDA) effect size (LEfSe) analysis.(D) Fungal LEfSe analysis, the genera with LDA score greater than 3.0.

TABLE 1
Soil physical and chemical properties under different treatments (2021).
Values are mean ± SE in triplicate replicates.Different lowercase letters indicate significant differences with a p < 0.05 based on the analysis of variance.

TABLE 2
Plant physiological indicators under different CC years.
Values are mean ± SE in triplicate replicates.Different lowercase letters indicate significant differences with a p < 0.05 based on the analysis of variance.10.3389/fmicb.2023.1318586Frontiers in Microbiology 05 frontiersin.organd biosynthesis of plant hormones in group CC1; pyruvate metabolism, styrene degradation, biosynthesis of siderophore group non-ribosomal peptides, alanine, aspartate, and glutamate metabolism in group CC4; phenylpropanoid biosynthesis, protein digestion and absorption, aminobenzoate degradation, toluene degradation, and biosynthesis of various plant secondary metabolites in group CC7, respectively.