Gastrodia elata tubers and their surrounding soil were collected from the greenhouse (E35.76.85 and N127.37.69) at Medicinal Herb Experiment Station in Jeollabuk-do, South Korea. Gastrodia elata was sown in early April 2021 using a ternary cropping system and cultivated in a rain-sheltering plastic greenhouse for approximately 12 months until harvest at the end of March 2022. Briefly, three lines of holes were made in the ridges, and A. mellea infected chestnut logs (10–12 cm diameter, 30–40 cm length) were placed parallel to the bottom of each hole. After germination of G. elata seeds using the symbiotic fungi Mycena sp. (Kim et al., 2006), juvenile tubers were sown at regular intervals around the logs. The holes were filled with soil (5–10 cm depth), covered with straw, and soil sprinkled over the surface. Tubers, rhizosphere soil, and bulk soils of G. elata were carefully collected at the rooting and juvenile tuber stage (JT; 2 MAS), growing and young tuber stage (YT; 6 MAS), and then at the harvesting and mature tuber stage (MT; 12 MAS; Yuan et al., 2018; Supplementary Figure S1).

Three places (30 × 100 cm) were randomly selected, and intact tubers along with soil 5 cm from the root zone were dug out from the places at each stage and combined to form one biological replicate. The whole rhizocompartment sample was contained in sterile plastic bags and kept at 4°C. Samples were brought to the laboratory and processed within a couple of days to isolate DNA and microbes from the endosphere (root), rhizosphere, and bulk soil samples. The rhizomorphs of A. mellea were removed from each sample. Bulk soil easily detached from the tubers was collected and sieved through a 2-mm sieve. Subsequently, the soils adherent to the tubers were detached using sterilized brushes, and taken as rhizosphere soil. The tubers were washed with tap water, then surface-sterilized with ethanol (70%, v/v) for 1 min and sodium hypochlorite solution (3%, v/v) for 3 min, and finally rinsed with sterile distilled water (DW) five times and used as endosphere samples. Successful surface sterilization was confirmed by spreading 100 μL of the final rinsing water on Luria Bertani (LB) agar plates, and bacterial growth was evaluated after incubation at 28°C for 48 h. Three samples from biological bulk soil, rhizosphere soil, and endosphere were analyzed.

The collected soil samples (5 g) were added to 50 mL sterile DW. The suspensions were thoroughly stirred for 5 min, serially diluted, and 100 μL of each dilution was spread on 0.1% tryptic soy agar (TSA) containing 50 μg/mL cycloheximide for total bacterial count, and 10% potato dextrose agar (PDA) containing 50 μg/mL ampicillin and 20 μg/mL rifampicin for total fungal count. The surface-disinfected tuber tissues (1 g) were ground in a sterilized mortar and pestle and suspended in sterile DW (10 mL) to isolate endophytic bacteria and fungi from the tubers. The suspension was serially diluted 10-fold and plated on TSA and PDA media supplemented with 1% tuber extract (TE) of G. elata. TSA plates (three replicates for each subsample/dilution combination) were incubated at 28°C for 3 days, and PDA plates were incubated at 25°C for 7 days to enumerate viable bacterial and fungal colonies, respectively. Bacterial colonies showing distinct colors and shapes were randomly picked and transferred to fresh Luria-Bertani (LB) agar medium and stored at −80°C in the LB medium supplemented with 15% glycerol. Fungal isolates showing distinct morphology and color were picked and transferred to test tubes containing PDA slant media and stored at 4°C.

DNA was extracted from each collected sample using the GeneAll Exgene Soil DNA isolation kit (GeneAll, Seoul, South Korea), following the manufacturer’s instructions (Dutta et al., 2022). The quality and quantity of the isolated DNA were analyzed using a BioTek, Epoch Spectrometer (Agilent Technologies, Santa Clara, California, United States) and 1% agarose gel electrophoresis.

Bacterial libraries were generated by amplifying the V3–V4 region of the 16S rRNA gene using the universal primers including 341F and 805R (Dutta et al., 2022). Fungal libraries were produced by amplifying the ITS region using the forward primer ITS3-F and reverse primer ITS4-R (White et al., 1990). These primers appended with the nextera concensus and sequencing adaptor at the 5′ end (Fadrosh et al., 2014). The PCR products were cleaned up and used for further amplification with primers containing Illumina dual indices and sequencing adapter (Dutta et al., 2022). The PCR cycling conditions were a denaturation step at 95°C for 3 min; 25 cycles (1st PCR) and 8 cycles (2nd PCR) of 30 s denaturation at 95°C, 30 s annealing at 55°C, 30 s extension at 72°C; and a 5 min final extension at 72°C. The amount of the PCR product was measured using a Quant-iT PicoGreen dsDNA Assay Kit (Thermo Fisher Scientific, Waltham, Massachusetts, United States), and the quality was assessed using an Agilent Bioanalyzer 2,100 system (Agilent Technologies, Santa Clara, California, United States). Purified amplicon libraries were pooled and sequenced at CJ Bioscience Inc. (South Korea) using an Illumina MiSeq system with MiSeq Reagent Kit v2 (Illumina Inc., San Diego, California, United States). Sequence data were deposited in the GenBank SRA database under BioProject accession numbers PRJNA1074006 and PRJNA1074502 for bacteria and fungi, respectively.

Sequenced data were analyzed using EzBioCloud,¹ a public pipeline that provides quality control, merging paired-end reads, sequence processing, taxonomic classification, and diversity analysis for operational taxonomic units (OTUs), operated by CJ Bioscience, Inc. (Seoul, South Korea), as described by Dutta et al. (2022). Taxonomic assignments were implemented by searching reference databases using the USEARCH program for the 16S rRNA of bacteria and the UNITE² database for the ITS2 of fungi. Query sequences that are matched to reference sequences in EzBioCloud by ≥97% similarity considered to be at the species level. Lower sequence similarity cutoffs are used for genus or higher taxonomic levels as follows (x = similarity), genus (97 > x ≥ 94.5%), family (94.5 > x ≥ 86.5%), order (86.5 > x ≥ 82%), class (82 > x ≥ 78.5%), phylum (78.5 > x ≥ 75%; Yarza et al., 2014). The suffix (_p, _c, _o, _f, _g, _s) of the taxonomic name is used when the sequence of the database that is most similar to the base sequence to be analyzed does not have a formal taxonomic name (phylum, class, order, family, genus, and species, respectively). Non-cultured species are used to the existence only in the nucleotide sequence data. The suffix “_uc” of the taxonomic name is an abbreviation of unclassified. It is a collection of sequences that can be regarded as new because there is no similar base sequence to the species level in the database. In this case, “_uc” is added to the parent system name to create arbitrary matrix and express the result.

The alpha diversity indices of the bacterial and fungal communities, including ACE, Chao1, Shannon, Simpson, Pielou’s evenness, and phylogenetic diversity indices, as well as rarefaction curves, were calculated. Discrepancies in alpha diversity, such as diversity, number of OTUs, and richness, were analyzed and compared between the samples collected at various growth stages of G. elata, and taxonomic compositions were compared from the phylum to the species level. Beta diversity, including principal coordinate analysis (PCoA) and unweighted pair-group method with arithmetic mean (UPGMA) clustering, was analyzed based on the Bray–Curtis dissimilarity index (Bray and Curtis, 1957) at the species level. A heatmap showing the relative abundance of fungi and bacteria at the genus level was analyzed and compared using CL community program v3.43 (CJ Bioscience, Seoul, South Korea; Dutta et al., 2021).

The antagonistic activity of Mycena sp. Jinan-1 (MJ-1) and A. mellea KACC52941 (Am52941) were assayed against soil-inhabiting or plant-pathogenic fungal strains following the methods of Yu et al. (2012) with minor modifications. The fungal strains, such as Botrytis cinerea, Chaetomium novozelandicum, Fusarium graminearum KACC41040, Penicillium digitatum, and Rhizopus stolonifer KACC41364 were maintained on PDA slants at 4°C were cultured using fresh PDA plates. Mycelial disk (8 mm) from the active culture of each fungal strain were placed on PDA plate 2.0 cm from the edge of the plate. Similarly, a mycelial disk (8 mm in diameter) of MJ-1, Am52941, or a PDA plug (control) was placed on the opposite side of the plate. The plates were incubated at 25°C for 7 days, and the radial growth of the tested fungi was measured. The radial inhibition rate (RI) was calculated according to the following formula (Nuangmek et al., 2021): RI = (Ro – Rt) /Ro × 100, where Ro represents the radius of tested fungi in the control plates, and Rt represents the radius of tested fungi in treatment plates. The experiment was repeated thrice with three replicates. The KACC strains were originally obtained from Korean Agriculture Culture Collection; Mycena sp. Jinan-1 (MJ-1) was donated from the Medicinal Herb Experiment Station; B. cinerea and P. digitatum were isolated and identified by plant pathology lab of Jeonbuk National University (Dutta and Lee, 2022); Chaetomium novozelandicum was isolated and identified from the rhizosphere of G. elata.

Culture filtrates (CFs) from MJ-1 and Am52941 cells were obtained using method described by Yu et al. (2012). Briefly, the actively growing mycelium plug from freshly cultured plate was inoculated into 100 mL of potato dextrose broth (PDB) in a 250 mL Erlenmeyer flask and incubated at 27°C for 4 weeks with slow shaking. The culture broth was collected by filtering through Whatman No. 4 filter paper and then centrifuged at 10,000 rpm for 20 min to remove cellular debris, followed by filtering the supernatant through a 0.22 μm filter. The CFs were stored at −20°C until further use. The antagonistic activity of CF against the growth of fungal strains, such as B. cinerea, C. novozelandicum, F. graminearum, P. digitatum, and R. stolonifer, was assessed by mixing a gradual two-fold CF dilution with autoclaved, molten PDA at 45°C to eventual concentrations of 50%, 25%, 12.5%, and 6.25%. Fresh PDA was used as a control. Mycelial disks (8 mm in diameter) were removed from the fringe of a 5- to 7-day-old colony of each fungal strain grown on PDA and placed on the center of the PDA plates. The plates were maintained at 25°C in darkness, and the diameters of fungal colonies were measured after 10 days of growth. The percentage of inhibition was calculated using the following formula (Kong et al., 2018): [(D_o – D_t) /D_o] × 100, where D_o represents the average colony diameter of the control, and D_t represents the average colony diameter of the treatment. All treatments were effectuated independently thrice in triplicate.

The antagonistic activity of the CF against Gram-positive (Bacillus subtilis KACC10854, B. velezensis, and Cytobacillus firmus JBRS159) and Gram-negative (Agrobacterium tumefaciens, Pseudomonas putida JBC17, and Variovorax paradoxus JBCE486) bacteria was assessed by spreading a 100 μL of cell suspension of each strain on LB agar plates amended with various concentrations (6.25%, 12.5%, 25%, 50%) of CF of MJ-1 or Am52941. The number and size of the bacterial colonies were determined after incubation at 28°C for 48 h. The experiment was replicated thrice with three replicates. The KACC strains were originally obtained from Korean Agriculture Culture Collection; Other strains were stock culture of plant pathology lab of Jeanbuk National University (Dutta et al., 2022; Dutta and Lee, 2022; Kang et al., 2023).

Mature tubers of G. elata were peeled, surface-sterilized with ethanol and sodium hypochlorite, and washed with sterile DW as described above. Tuber tissues (200 g) were sliced using a sterilized knife and ground with a mortar and pestle. The homogenates were resuspended in sterile DW (200 mL) and shaken for 2 h at 200 rpm. The slurry was filtered through four layers of cheesecloth, and then the suspension was filtered sequentially through Whatman filter paper and a sterile syringe filter (0.2 μm pore size) to create tuber extract (TE, 100%). The LB agar and PDA medium containing TE at various concentrations (6.25%, 12.5%, 25%, and 50%) were used to evaluate the inhibitory activity of TE against bacterial and fungal strains, respectively, as described above.

All the experiments were performed using a completely randomized design. The statistical significance of OTU numbers, alpha diversity of microbial communities, and relative abundance at taxonomic ranks among different growth stages of G. elata, as well as the size of the tested fungal colonies in the antagonism experiments, was analyzed by one-way analysis of variance (ANOVA) and Tukey’s test at the p < 0.05 level of significance using Minitab version 16.2.0 software (Bower, 2000).

Cultural management and soil properties for crop cultivation induce changes in the quality and quantity of microbiota. In this study, G. elata tubers and surrounding soils following the growth stages were used to investigate the characteristics of the microbial communities associated with the various compartments of G. elata cultivated in greenhouses. The soils were loosely attached to the tubers, making it difficult to distinguish between the rhizosphere and the bulk soil around the tubers (Supplementary Figure S1). There were no significant differences in the total culturable bacterial and fungal population densities in the bulk and rhizosphere soils of G. elata plants (Supplementary Figure S2). Almost no bacterial and fungal isolates were obtained from the tubers, except for a few strains from the skin of the tubers. Our findings indicated no notable distinctions in the quantities of culturable bacteria and fungi between bulk and rhizosphere soils, and no culturable microbes were present as endophytes.

We analyzed the bacterial and fungal community structures of G. elata at various growth stages. High-throughput sequencing generated 570,245 (median: 63,360) bacterial 16S rDNA reads and 617,010 (median: 68,557) fungal ITS region reads. After quality trimming, merging, and removal of chimeric reads, a total of 451,059 and 355,536 reads with median read lengths of 416.74 and 339.55 for bacteria and fungi, respectively, were obtained. The rarefaction curve at 3% dissimilarity in rhizosphere showed an average of 4,713, 3,201, and 2,683 OTUs for bacteria and an average of 696, 573, and 484 OTUs for fungi in JT, YT, and MT stages, respectively (Supplementary Figure S3; Supplementary Tables S1, S2), indicating that the sequencing depth adequately covered detectable species in rhizosphere soil samples. The Good’s coverage analysis revealed more than 98% and 99% of the bacterial and fungal taxonomic richness, respectively, which were covered by sequencing endeavors in all the samples of growth stages of G. elata, indicating that they represented the actual rhizosphere bacterial and fungal populations in the microbial community of each sample (Supplementary Tables S1, S2).

The highest alpha diversity (Shannon, Chao1, ACE, Simpson, Jackknife, OTUs, NPShannon, and phylogenetic diversity) of the rhizosphere bacterial and fungal communities was observed in the JT samples, and the indices were not significantly different between the YT and MT samples. Pielou’s evenness index showed that the microbial communities were evenly distributed in the JT samples (Supplementary Tables S1, S2). The indices indicated that the richness and diversity of bacterial and fungal communities in YT and MT were lower than in JT.

A PCoA based on Bray–Curtis analysis was performed, and a UPGMA clustering tree was built using 16S rDNA and ITS sequences to compare the bacterial and fungal community structures in rhizosphere soils at various growth stages of G. elata. The PCoA analysis of bacterial communities showed a coefficient of variation of 49.67%, indicating that they could characterize the bacterial community composition in rhizosphere. The beta diversity indices of the PCoA revealed that the bacterial taxonomic structure and composition at each growth stage were distinctly clustered, indicating changes in the bacterial community structure over time. The JT stage samples were closely grouped in comparison to the YT and MT stages, indicating that the bacterial community composition in the YT and MT stages was more diverse between the locations (Figure 1A). The UPGMA clustering also showed that the species-level bacterial community structures of the JT, YT, and MT stages clustered differently (Figure 1B).

The coefficients of variation of PCoA for fungal community in G. elata rhizosphere were 27.28% and 25.31% for PC1 and PC2, respectively. The fungal communities in JT, YT, and MT were distinctly separated, indicating that their fungal community composition changed during the growth stages. The fungal community at the JT stage was more closely grouped than that at the YT and MT stages, indicating less variation in the fungal community composition at the JT stage than at the YT and MT stages (Figure 2A). The UPGMA clustering showed that the species-level fungal community structures of MT were widely scattered among the locations (Figure 2B). The fungal communities were distinctly separated at JT and YT. However, there were similarities between the fungal communities at MT with that of JT and YT, respectively. Taken together, the structures of bacterial and fungal communities in the rhizosphere of G. elata showed distinct clusters over time. These results indicated a unique bacterial and fungal composition at each growth stage of G. elata.

We compared the bacterial taxonomic profiles of the rhizosphere soils collected at various growth stages of G. elata. A total of 44, 44, and 37 bacterial phyla were detected in the JT, YT, and MT stages, respectively. The dominant bacterial phyla in the rhizosphere of G. elata were consistently abundant throughout the growth stages of G. elata. Proteobacteria was the most abundant phylum in all stages (40.61%–45.17%), followed by Acidobacteria (21.15%–24.61%), and Bacteroidetes (6.098%–13.21%; Figure 3; Supplementary Table S3).

In total, 616, 550, and 476 genera with more than 0.01% relative abundance were identified at the JT, YT, and MT stages, respectively. Among these, 353 genera were consistently detected in all growth stages of G. elata, comprising 57%, 64%, and 74% of the total population at the JT, YT, and MT stages, respectively. The numbers of unique bacterial genera at the JT, YT, and MT stages were 139, 89, and 49, respectively (Supplementary Figure S4). The genera such as Paraburkholderia (3.38%–5.97%), Rhizomicrobium (2.35%–4.44%), Mucilaginibacter (1.72%–3.33%), Sphingomonas (1.73%–3.99%), and Edaphobacter (1.58%–2.98%) were constantly detected with high abundance. The average relative abundances of Pseudomonas and Rahnella increased with the growth of G. elata, whereas Solibacter, Acidibacter, Bradyrhizobium, and Rhodanobacter decreased (Figure 4; Supplementary Table S4). These results indicated that the number of bacterial genera gradually decreased with the growth of G. elata, whereas some strains remained constant throughout the growth period.

At the bacterial species level, Edaphobacter dinghuensis, Sphingomonas lutea, and Bradyrhizobium japonicum were highly abundant at all growth stages of G. elata. The relative abundance of Rhodanobacter glycinis at the JT stage was higher than that at the YT and MT stages. In contrast, increases in relative abundance were recorded for Pseudomonas species such as Pseudomonas_uc and Pseudomonas putida group (Supplementary Figure S5). Overall, the results indicated that the bacterial community in the rhizosphere changed as the tubers grew, suggesting an active selection of bacterial communities by G. elata tubers.

A total of 14, 12, and 12 fungal phyla were detected in the JT, YT, and MT stages, respectively. The dominant fungal phyla across the growth stages of G. elata were Ascomycota (22.23%–49.48%), Basidiomycota (19.90%–33.10%), and Mortierellomycota (13.59%–19.30%; Figure 5; Supplementary Table S5).

As the tubers grew, the number of fungal genera in JT, YT, and MT gradually decreased to 335, 266, and 265, respectively. Among the detected fungal genera, 168 were consistently present at all growth stages. A total of 209, 203, and 184 genera were shared between the JT and YT, JT and MT, and YT and MT stages, respectively. By contrast, 91, 41, and 46 genera were unique to the JT, YT, and MT stages, respectively (Supplementary Figure S6). The abundance of fungal genera such as Capnodiales_g, Conlarium, Oidiodendron, Sordariomycetes_g, Apiotrichum, Mortierellaceae_g, and Umbelopsis decreased as G elata tubers grew. The relative abundances of Geomyces, Guehomyces, Hymenoscyphus, Leptodontidium, Mucor, Pseudogymnoascus, Sistotrema, and Solicoccozyma increased with the growth of G. elata. The fungal genera such as Mortierella (13.17%–19.09%), Chaetomium (1.43%–9.03%), Trichoderma (1.04%–3.30%), Fusarium (1.90%–2.79%), Saitozyma (1.03%–1.44%) were highly abundant throughout the growth of G. elata, which indicate possible involvement of these genera in the growth of G. elata (Figure 6; Supplementary Table S6). The Mycena genus, which plays an important role in the germination of G. elata seeds (Xu and Guo, 1989; Chen et al., 2019) was detected with low abundance (none at JT stage and 0.013% and 0.008% at YT and MT stages, respectively).

Mortierella rishikesha was the most abundant fungal species in the rhizosphere soils of JT, YT and MT stages (11.1%, 9.7%, and 14.5%, respectively), followed by Candida subhashii (1.9%, 4.1%, and 2.1%, respectively). Among the fungal species with relative abundance above 1.0%, Mortierella rishikesha, Candida subhashii, Fusarium oxysporum, Saitozyma podzolica, and Umbelopsis isabellina were highly abundant across all growth stages. The abundance of Oidiodendron rhodogenum, Apiotrichum xylopini, Capnodiales sp._GU721261, Mortierellaceae sp._HQ445985, Conlarium sp._JX489782, and Trichoderma viride decreased with the growth of G. elata (Supplementary Figure S7).

The proliferation of symbiotic fungi around seeds and tubers can influence the bacterial and fungal communities in the rhizosphere. The results showed the symbiotic fungus MJ-1 suppressed the growth of soil-inhabiting fungi such as B. cinerea (55.4% in radial inhibition rate), C. novozelandicum (48.9%), F. graminearum (39.4%), P. digitatum (46.1%) and R. stolonifer (54.7%). Another symbiont Am52941 inhibited the mycelial growth of B. cinerea (34.9%), C. novozelandicum (16.7%), P. digitatum (23.4%), and R. stolonifer (39.5%), but did not inhibit the growth of F. graminearum (Figure 7; Supplementary Table S7).

The CFs of MJ-1 and Am52941 also exhibited different antagonistic activities against different fungal and bacterial strains. In which, the CF of MJ-1 significantly suppressed mycelial growth of all tested fungal strains, including B. cinerea, C. novozelandicum, F. graminearum, P. digitatum, and R. stolonifer. The CF of Am52941 inhibited the growth of B. cinerea, C. novozelandicum and R. stolonifera, but did not inhibit the growth of F. graminearum and P. digitatum (Table 1; Supplementary Figure S8).

The CF of MJ-1 suppressed the growth of B. velezensis and B. subtilis at 25% concentration, and that of C. firmus was significantly inhibited at a concentration of 50%. The growth of Gram-negative P. putida and A. tumefacien was not inhibited by CF of MJ-1. However, the survival of V. paradoxus was significantly suppressed by the CF addition at 6.25% concentration. The CF of Am52941 inhibited the growth of Gram-positive B. subtilis at 25% concentration and Gram-negative V. paradoxus at 12.5% concentration (Table 2; Supplementary Figure S9). Overall, the results showed that both symbiotic fungi of G. elata had different effects on different bacterial and fungal strains, indicating that they play a role in selecting the composition of the rhizosphere microbial community of G. elata.

No fungal or bacterial strains were detected in the tubers in culture-dependent assays for fungal and bacterial populations. In addition, the diversity and relative abundance of microbes decreased as G. elata grew. The tuber extracts were assessed for effect on fungal and bacterial growth. The addition of 25% TE inhibited the growth of B. cinerea (13.1%), C. novozelandicum (7.8%), and P. digitatum (22.1%), and the inhibition zone increased with increasing concentrations (50%) of TE. Intriguingly, the growth of F. graminearum was significantly increased by the addition of TE (from 29.1% to 52.0%) without any inhibition, and the growth of R. stolonifer was inhibited at low concentration (6.25% to 12.5%) but was significantly increased by adding 50% TE (Table 3; Supplementary Figure S10).

The addition of 12.5% TE significantly reduced the survival of the Gram-positive bacteria B. subtilis and B. velezensis. Cytobacillus firmus was completely inhibited by the addition of 25% TE to the medium. The survival of Gram-negative A. tumefaciens decreased with the addition of 6.25% TE and decreased further as the TE concentration increased. However, the survival of P. putida was not affected by adding TE and that of V. paradoxus was inhibited at a concentration of 50% TE (Table 4; Supplementary Figure S11). The differences in the antimicrobial activity of TE against bacterial and fungal species indicated that G. elata tubers greatly contribute to the selection of components in the rhizosphere microbial community, leading to a reduction in the diversity of bacterial and fungal communities with the growth of G. elata. The colony sizes of most tested bacterial strains decreased with increasing of TE concentrations. However, the colony sizes of A. tumefaciens, B. velezensis, and V. paradoxus were increased by the addition of low concentrations (6.25%, 12.5%, and 25%) of TE, and colony colors were changed (Supplementary Figure S11). This indicated that tubers also influence physiology and growth.