Polygonatum sibiricum was planted in walnut (cultivar Xiangling), chestnut (cultivar Taiguo 5) and persimmon (cultivar Yangfeng) orchards at Xiancaogu Base, Tai’an City, Shandong Province, China (36°20′46″N, 116°96′20″E) in 2017, and field cultivation was used as a control. The climate of the test site is a continental subhumid monsoon climate. The average annual precipitation is 697 mm, the average annual temperature is 12.9° and the average annual sunshine duration is 2627.1 h. The soil was sandy soil, and the chemical composition of the soil was as follows: P, 132 mg/kg; N, 324 mg/kg; and K, 1024 mg/kg; with a pH of 6.87. To ensure that the soil factors at the four sites were as similar as possible during the experiment, the agricultural practices used, such as fertilization, irrigation and weeding, were consistent.

Healthy and uniform rhizomes of P. sibiricum were selected and planted in the above experimental sites in March 2020, and rhizomes and rhizosphere soil samples were collected in March 2022. Rhizosphere soil was collected according to methods described by Bulgarelli (Bulgarelli et al., 2015). For each collected rhizome, the excess soil was removed by vigorous shaking, and then the soil closely attached to the rhizome was rinsed away with phosphate-buffered saline (PBS) solution and collected. Subsequently, the rinsing solution for each of the 5 rhizomes in the same treatment was concentrated in a 50 mL centrifuge tube and repeated as a rhizosphere soil sample, with 3 replicates for each treatment. After the rhizosphere soil was collected, the rhizomes were placed into a centrifuge tube with sterilized PBS, washed repeatedly with continuous vortexing until clean, and then transferred to a new centrifuge tube for ultrasonic surface cleaning (Xiao et al., 2017a,b). The obtained rhizomes were frozen in liquid nitrogen and used for the detection of endophytic microorganisms. Another 12 rhizome samples were prepared for metabolite detection. Thus, a total of 36 samples (3 replicates × 4 understory planting modes; 12 rhizome samples for metabolites +12 rhizosphere soil samples for microorganisms +12 endosphere samples for microorganisms) were stored at −80°C until DNA was extracted.

A FastDNA SPIN Kit for Soil and DNA Secure Plant Kit were used to extract DNA from rhizosphere and endosphere microorganisms, respectively.

We used the primers F515 (5′ - GTGCCAGCMGCCGCGGTAA - 3′) and R907 (5′ CCGTCAATTCCTTTGAGTTT - 3′) to amplify the V4 + V5 region of 16S rRNA. The primers 2024F (5′- GCATCGA TGAAGAACGCAGC-3′) and 2409R (5′- TCCTCCGCTTA TTGATATGC-3′) were used to amplify the ITS2 gene. The PCR system volume was 30 μL in total, including 15 μL of PCR Mix (Thermo Fisher Scientific, Waltham, MA, USA), 0.2 μM primers and 10 ng of DNA template. The PCR procedure was as follows: 98°C for 1 min; 30 cycles of 98°C for 10 s, 50°C for 30 s, and 72°C for 30 s; and 72°C for 5 min. PCR products were detected by 2% agarose gel electrophoresis, followed by DNA purification using the GeneJET Gel Extraction Kit (Thermo Fisher Scientific, Waltham, MA, USA). The quality of the DNA library was evaluated by an Agilent Bioanalyzer 2100 system. Sequencing was performed using the Illumina HiSeq 2500 sequencing platform.

The frozen rhizomes were quickly ground to powder in liquid nitrogen, and then 5 mL of precooled solution (acetonitrile:isopropyl alcohol:water at a ratio of 3:3:2) was added to 100 mg of powder. The solution was centrifuged at 12,000× g and 4°C for 10 min, and the supernatant was concentrated with rotary vacuum drying. Then, 0.6 mL of 70% methanol was added to dissolve the substance. The resulting solution was centrifuged, and the supernatant was absorbed and filtered through a 0.22 μm microporous filter membrane into a sample collection bottle. Ultra-performance liquid chromatography–tandem mass spectrometry (UPLC–MS) analysis was used to obtain metabolome data. The metabolites were identified based on the MWDB and multiple reaction monitoring (MRM) model of two-stage spectrum and triple quadrupole mass spectrometry (Xiao et al., 2017a,b).

MetaboAnalyst 2.0 was used for principal component analysis (PCA) of identified metabolites with Pearson correlation coefficients set to default values. Variable Importance in the Projection (VIP) > 1.0, fold change (FC) > 1.2 and p value <0.05 were set as thresholds to screen differentially abundant metabolites. Hierarchical clustering analysis of differentially abundant metabolites was performed using MultiExperiment Viewer-Tm4 (MeV) software. Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of differentially abundant metabolites was performed using the Nohe Source Data Analysis Platform, and the threshold was set as a p value ≤0.05.

Quantitative Insights Into Microbial Ecology (QIIME) was used to filter the original microbial reads. The remaining reads of the original DNA fragments were merged using the FLASH tool. Paired-end reads were matched to each sample based on QIIME unique barcodes. Operational taxonomic unit (OTU) identification was performed with the sequence similarity threshold set at 97%, and the RDP classifier was used to annotate representative sequences of each OTU. The alpha-and beta-diversity of the obtained OTUs were analyzed using Perl scripts. PCA based on weighted UniFrac (WUF) distances was performed using the “ape 3.4” package in R software. Permutational multivariate analysis of variance (PERMANOVA) was carried out using “vegan 2.3–0.” Canonical correspondence analyses (CCAs) were conducted using ade 1.7–4. Cointertia analysis (CoIA) was performed on the Tutools platform,¹ a free online data analysis website. The correlation coefficients between differentially abundant microorganisms and metabolites were obtained using R software, and the coexpression network was visualized using Cytoscape software.

The PCA of metabolism (Figure 1A) showed significant differences in metabolic profiles among different groups (p < 0.05), and 12 samples could be significantly divided into 3 categories. The field cultivation group and the other three groups were significantly different in the axis 1 direction. The chestnut-interplanted groups and persimmon-interplanted groups were clustered together, indicating their similarity. A total of 862 metabolites were identified (Supplementary Table S1) in the 12 samples. Supplementary Table S1 shows that the differentially abundant metabolites in the chestnut group vs. field cultivation, walnut group vs. field cultivation, persimmon group vs. field cultivation, chestnut group vs. walnut group, chestnut group vs. persimmon group and walnut group vs. persimmon group comparisons numbered 14, 58, 14, 9, 5, and 17, respectively. A total of 95 differentially abundant metabolites (Supplementary Table S1) were observed in the 6 comparisons. We found that the number of differentially abundant metabolites between the walnut group and field cultivation was the largest and that between the chestnut group and walnut group was the smallest (Supplementary Table S1). Hierarchical clustering analysis (HCA) was conducted for the 95 differentially abundant metabolites obtained. Figure 1B shows that the 95 differentially abundant metabolites could be divided into 4 groups based on their expression patterns. Pattern 1 characterizes 53 metabolites that were highly expressed in P. kingianum rhizomes planted under walnut trees. Pattern 2 described 22 metabolites that were highly expressed in the P. kingianum rhizome in the field cultivation group. Pattern 3 characterized 12 metabolites, which were least expressed in the walnut group. Pattern 4 represented only 8 metabolites, which were highly expressed in the persimmon group. Pathway analysis (Figure 1C) showed that the 95 differentially abundant metabolites were mainly involved in signaling pathways such as purine metabolism, amino acid synthesis, tryptophan metabolism, phenylalanine metabolism, lysine degradation, phenylpropyl synthesis, and tryptophan synthesis. Polysaccharides were the main effective component in P. sibiricum, but the polysaccharide level was not significantly different among the groups. The polysaccharide components included mannose, galactose, arabinose and glucose (Wang, 2020). Figure 1D shows that the mannose content significantly differed among the groups, and it was the lowest in the field cultivation group and highest in the walnut group. Mannotriose is involved in galactose metabolism, and its decomposition yields galactose, which is an important link in the galactose metabolism pathway. Figure 1B and Supplementary Table S1 show that mannotriose was highly expressed in the persimmon group. The above results showed that interplanting mode had significant effects on the metabolites in the P. sibiricum rhizome.

A total of 3,267,438 high-quality reads were obtained from 48 samples, with an average of 68,071 reads per sample. The sequence similarity threshold was set to 97%. A total of 9,672 bacterial OTUs and 2,204 fungal OTUs were obtained from the rhizosphere soil. A total of 1812 bacterial OTUs and 995 fungal OTUs were obtained for endosphere microorganisms (Supplementary Table S2).

Understory intercropping can provide shade conditions for P. sibiricum and prevent the damage caused by high summer temperatures, and the physical and chemical properties of soil could also be improved (Yong et al., 2019); therefore, tree type is one of the main factors affecting P. sibiricum growth. In this study, the mannose content of P. sibiricum in the interplanting groups was significantly higher than that in the farmland group, and the mannose content in the walnut group was the highest. It has been reported that some plant growth-promoting rhizobacteria, such as Acinetobacter and Streptomyces, increased the concentrations of mannose, glucose and galactose in maize (Lopes et al., 2022); meanwhile, the relative abundance of Verrucomicrobiae and Proteobacteria was strongly associated with a change in the proportion of mannose in Acropora muricata, Vibrio coralliilyticus and Vibrio natriegens, which may potentially be able to utilize a large amount of mannose (Sonny et al., 2016). In our study, the increase in mannose content in P. sibiricum planted under walnut may be related to specific bacteria.

In addition, compared to differences within the three interplanting groups, the number of different metabolites between the farmland group and the three groups was larger (Supplementary Table S1). Similarly, P. sibiricum planted under oak trees had better growth, less disease and higher yield than that planted under citrus trees, Platycladus trees, pine trees, peach trees and laurel trees (Chen et al., 2021). This may be related to ecological factors such as scattered light, air temperature and air humidity (Huang et al., 2016) or differences in rhizosphere microorganisms caused by tree interplanting (Fang et al., 2022). Specifically, rhizospheric microorganisms could affect the nutrient availability, resistance, and antagonism of pathogenic microorganisms (Zhu et al., 2021a,b). Habitat conditions are one of the factors that affect the growth and accumulation of secondary metabolites in P. sibiricum. In this study, 53 secondary metabolites were highly expressed in the walnut group, possibly because shading under walnut trees is more suitable for P. sibiricum growth because excessive shading or no shading could inhibit growth (Ni et al., 2020). Although there were significant differences in secondary metabolism among the groups, no significant differences were observed in morphology or polysaccharide content, which may be because the experimental period was only 2 years and the accumulation effect had not yet occurred.

In this study, the metabolome and microbiome were analyzed for different understory interplanting groups. The PCA of the metabolome and microbiome separated the farmland group and understory interplanting groups along axis 1 and axis 2, respectively, suggesting differences in metabolism and microorganisms between the farmland group and understory interplanting groups. Similarly, in a previous study, the bacterial structure of ginseng in the understory interplanting group was more similar to that in the wild group than that in the farmland cultivation group (Fang et al., 2022). Compared with that in the farmland cultivation group, the alpha diversity of rhizosphere bacteria and endosphere fungi in the three understory interplanting groups was higher; however, the alpha diversity of rhizosphere fungi and endosphere bacteria was significantly decreased in the three understory interplanting groups. A series of farmland management measures, such as artificial shade and fertilization, may have affected the soil characteristics experienced by the farmland cultivation group (Huang, 2019), which then affected the intrinsic characteristics of P. sibiricum. In addition, P. sibiricum is susceptible to diseases and pests in the process of artificial cultivation of rhizomes in farmland (Rahman and Punja, 2006); therefore, pesticide spraying can indirectly affect rhizosphere secretions or directly inhibit the propagation of some rhizosphere microorganisms (Arango et al., 2014; Chen et al., 2017), thus affecting rhizosphere microbial diversity. Cultivated crops have the characteristics of fast growth and high yield, which may have led to differences in the amount and type of organic compounds secreted by the P. sibiricum rhizome and thus differences in rhizosphere microorganisms between the farmland group and the understory interplanting groups.

The diversity of rhizosphere fungi and bacteria showed opposite differences between the farmland cultivation group and understory interplanting groups; that is, the alpha diversity of rhizosphere bacteria (Shannon–Wiener index) significantly increased while that of rhizosphere fungi decreased in the understory interplanting group. A similar phenomenon has also been found in soybean studies (Shi et al., 2019). Modern selective breeding of crops may promote the propagation of certain crop-associated microorganisms, leading to changes in the diversity of fungi in cultivated crops (Wissuwa et al., 2009). Second, previous studies have shown that ginseng in understory interplanting groups contains a higher content of saponins than that in farmland cultivation groups (Choi et al., 2007), which may affect the diversity of crop rhizosphere fungi (Zhang et al., 2016).

This study showed that Proteobacteria, Acidobacteria and Pseudomonas were the dominant bacteria in both the rhizosphere and endosphere of P. sibiricum, and Acidobacteria was one of the marker bacteria in the farmland cultivation group. Previous studies have also shown that Proteobacteria, Acidobacteria and Pseudomonas are dominant among the rhizosphere microorganisms of ginseng and P. sibiricum (Ying et al., 2012; Cai et al., 2021). In fungal communities, Fusarium is a potential plant pathogen that can cause root rot in a variety of plants, such as ginseng, American ginseng, soybean, and sunflower (Punja et al., 2007; Ying et al., 2012; Okello et al., 2020; Wei et al., 2020). In this study, Simper test analysis showed that the relative abundance of Fusarium in the P. sibiricum cultivated in farmland was significantly higher than that in P. sibiricum in the other three interplanting groups. Similarly, cultivated rice has a higher pathogen abundance than wild rice (Shi et al., 2019). Our results showed that the dominant microbial genera were different under different planting modes (Supplementary Figure S4), which should be closely related to planting vegetation. In fact, different species accumulate specific fungal pathogens in their roots, showing plant species-specific effects (Hannula et al., 2021). In our study, Talaromyces, Setophoma and Rhizobium were the unique dominant genera in the chestnut, persimmon and walnut interplanting groups; interestingly, mannose in root secretions can be used by 100% of Rhizobium species, thus attracting more Rhizobium (He et al., 2011). This is consistent with the highest mannose content being observed in the walnut interplanting group. A potential caveat of our study was that we did not measure soil chemical-mediated legacy effects, which have been shown to play a role in mediating plant legacy effects in some studies (Petermann et al., 2008). However, we tried to keep agronomic measures consistent so that there were no significant differences in soil chemistry.

The plant microbiome is closely related to the synthesis and accumulation of secondary metabolites in medicinal plants. The rhizosphere is an important site for plant–soil–microbe information and material exchange, and the interaction between plant roots and rhizosphere microorganisms is crucial for plant growth and quality. At the same time, many plant endophytes also have important biological value and establish special relationships with host plants. They play an important role in the formation of active metabolites in medicinal plants and then affect the quality and yield of medicinal materials. The metabolism of active components in different plants is affected differently by rhizosphere microorganisms and endophytic microorganisms (Zhou et al., 2022). This study showed that the differences in P. sibiricum metabolites were more strongly affected by rhizosphere microorganisms than by the endophytic microorganisms. In addition, LEfSe analysis and Metastat analysis showed that the differences in endophytic bacteria among the four groups were minimal, and the alpha diversity of rhizosphere microbes was higher than that of endophytic bacteria. The results suggested that the influence of interplanting under different trees on the rhizosphere microbiota was greater than that on the endophytic bacteria and the stability of plant endophytic bacteria (Edwards et al., 2015). The shared bacteria and fungi in the rhizosphere and endosphere accounted for 78.6 and 69.3% of the endophytic microorganisms, respectively, indicating that most of the microorganisms that did not colonize the surface of the rhizosphere also did not colonize the root (Edwards et al., 2015) since the barrier effect of the root surface selects for certain endophytic microorganisms (Zhou et al., 2022).

This study showed that Mortierella, Aspergillus, Ellin6067 and Ruminococcus could significantly distinguish the differentiated metabolites mediated by understory interplanting. Meanwhile, CCA suggested that Mortierella, Aspergillus and RB41 had the strongest influence on the metabolites in the walnut group, farmland group and chestnut group, respectively. RB41 is thought to play an important role in the maintenance of plant rhizosphere soil ecology (Shi et al., 2021). In addition, coexpression network analysis showed that menadione, dUMP, mogroside and L-alanyl-L-leucine had the most interactions with microorganisms. At the same time, Myrmecridium, as a common microbe in the rhizosphere and endosphere, also had a high degree of interaction in this coexpression network. Myrmecridium is believed to be able to antagonize diseases caused by continuous cropping disorders by affecting secondary metabolism in various medicinal plants (Dong et al., 2016; Wu et al., 2021). However, RB41 and Myrmecridium are rarely studied and still need further study. These results suggest that there is a potential mechanism of association between the rhizosphere or rhizosphere microorganisms and secondary metabolism.

Indeed, studies have shown that the microbiomes of several plants can synthesize secondary metabolites, some of which contribute to the quality of medicinal plants. Inoculation with endophytic bacteria isolated from alkaloid-rich rose significantly increased the malsaline content of low-yielding alkaloid-poor rose (Singh et al., 2020; Zhu et al., 2021a,b). Plant growth-promoting rhizobia (PGPRs) in Platycodon (Huang C. M. et al., 2019), mycorrhizal fungi in ginseng (Kim et al., 2017) and Actinomycetes in turmeric all significantly promote plant growth and resistance (Kim et al., 2017). Conversely, amino acids, organic acids, sugars, and other small molecules secreted by roots can also affect the growth of some soil microorganisms (Hartmann et al., 2009). Studies have shown that rhizosphere bacteria preferentially utilize aromatic organic acids secreted by plants (Zhalnina et al., 2018), salicylic acid can regulate the colonization of the rhizosphere by specific bacterial groups (Lebeis et al., 2015), and metabolites such as triterpene can regulate the growth of specific bacterial groups in the rhizosphere of Arabidopsis thaliana (Huang A. C. et al., 2019). Among medicinal plants, Salvia miltiorrhiza serves as a model medicinal plant, and the mechanism of interaction between microorganisms and secondary metabolism has also begun to be studied. Salvia miltiorrhiza may have a unique microbiome that is rich in functions related to secondary metabolism (Huang et al., 2018); for example, Pseudomonas and Pantoea promote the synthesis of salvia phenol acid by stimulating the production of abundant plant hormones (Zhalnina et al., 2018). Microbial interactions with tanshinone may affect tanshinone biomass and metabolic pathways, and the main functional microorganisms were Pantoea, Pseudomonas, Enterobacter, Sphingomonas and Alternaria (Chen et al., 2018). Conversely, the presence of some secondary metabolites in S. miltiorrhiza, such as tanshinine I and dihydrotanshinine I, also has a significant effect on the structure of the endophytic microorganisms. Moreover, the expression patterns of ginseng secondary metabolites and the rhizosphere microbial community were similar (Zhu et al., 2021a,b). Anthracis and Penicillium in the endosphere showed a significant negative correlation with the polysaccharide content, while Fusarium showed an extremely significant positive correlation with 5-HMF in P. sibiricum (Cai et al., 2021).