Cinnamomum migao H. W. Li belongs to the family Lauraceae (Cinnamomum), which is an aiphyllium. It is a small population species, concentrated in the narrow area at the junction of Guizhou, Guangxi and Yunnan in China, its interpopulation gene differentiation coefficient (Gst) is 0.2636, the gene flow at the population level (Nm = 1.3968) > 1, and the gene flow between the populations of C. migao is confirmed to be large and the genetic differentiation is small (Li et al., 2018). The materials of C. migao in this experiment are wild-type.

Rhizosphere soil and fruit samples were collected in October, 2018, from 5 healthy, wild C. migao trees (diameter 32–38 cm) at each of 9 sampling sites in Guizhou Province, Guangxi Province, and Yunnan Province, China. During the sampling process, the basic information such as sampling location, longitude, latitude and altitude were indicated. Information of the sampling sites are shown in Figure 1 and Table 1.

To collect soil samples, after removing the humus from the soil surface, samples of healthy plant roots were collected at a depth of 0–20 cm along the base of each tree by using a drill. Soil samples were packed into the sterilized plastic bags respectively, and sampling records (time, place, number, etc.) were made on. Then they were quickly stored in ice boxes and brought back to the laboratory. Samples were processed as described by Beckers, the remaining soil was first manually removed by shaking the roots until approximately 0–2 mm soil was left attached to the roots, and then rhizosphere soils (100 g) were collected by shaking on a platform (20 min, 120 rpm; Beckers et al., 2017). The rhizosphere soil samples from 5 trees per site were mixed in equal proportions to form a single conglomerate sample (500 g; Shen et al., 2013). Homogenized soil samples were divided into two sub-samples using a 2-mm sieve: one subsample was dried at 45°C in a blast drying oven to a constant weight for determination of soil chemical properties; the other sample was stored in a − 80°C ultra-low temperature freezer for extraction of microbial DNA.

The C. migao fruits were harvested when it turns from green to brown or black (October). The collection of fruit and soil samples shall be consistent, and the corresponding fruits are taken separately and brought back to the laboratory. Fruits were dried at 45°C in a blast drying oven to a constant weight. The dried fruits was crushed into powder with multifunctional crusher, screened with 40-mesh sieve, and storaged at 4°C for determination of bioactive compounds (Sun et al., 2022).

Soil chemical properties were determined according to the methods of Bao (2000). The potentiometric method was used to measure pH, water-soil ratio of 2.5:1(w/v); Organic matter (OM), Potassium dichromate oxidation volumetric method; Total nitrogen (TN), Alkaline hydrolysis diffusion method + semi micro Kjeldahl method; Total phosphorus (TP), Heating digestion method + Vanadium molybdate blue colorimetric method; Total potassium (TK), Heating digestion method + flame photometric method; Alkali hydrolyzed nitrogen (AN), Alkaline hydrolysis diffusion method + semi micro Kjeldahl method; Available phosphorus (AP), Sodium bicarbonate injection + Vanadium molybdate blue colorimetric method; Available potassium (AK), Ammonium acetate extraction + flame photometric method. The measured indicators of each soil sample (1–2 g) were measured five times in parallel, and the results were averaged.

Gas chromatography (GC) was used to quantify the bioactive compounds in C. migao fruit, including 1,8-Cineole, sabinene, limonene, and α-terpineol (Chen et al., 2021). The powder of C. migao samples was precisely weighed (1.00 g). Chromatographic conditions: the column was HP-5 capillary column (30 m × 320 mm, 0.25 μm); Carrier gas: high-purity nitrogen N₂; Flow rate: 1 ml⋅min⁻¹; Injection volume: 1 μl; Split ratio: 10:1, injector temperature: 200°C; temperature-rising program: the initial oven temperature was 50°C, and maintained for 3 min, raised the temperature to 85°C at 5°C⋅min⁻¹ and held for 2 min, raised the temperature to 90°C at 5°C⋅min⁻¹ and held for 2 min, raised the temperature to 160°C at 5°C⋅min⁻¹ and held for 0 min, raised the temperature to 220°C at 20°C⋅min⁻¹ and held for 0 min. Detector (FID) temperature: 235°C (Dai et al., 2013).

Analysis of variance (ANOVA) and Tukey T-test were employed to detect statistically significant differences (p < 0.05, p < 0.01) in these properties and compounds between samples using SPSS 22.0, Microsoft Excel 2010 and Origin 9.0 system software were used to analyze and map the data of soil chemical properties and fruit bioactive ingredients.

The bacterial genomic DNA was extracted from of rhizosphere soil samples using the FastDNA^® Spin Kit, the method was carried out according to the kit instructions. Added up to 0.5 g soil to Lysing Matrix E tube. Secured tubes in FastPrep Instrument and process for 30 s at speed 5.5. Centrifuged Lysing Matrix E tube at 14,000 ×g for 30 s. Transferred supernatant to a clean tube. Added 250 μl PPS and mixed by shaking the tube by hand 10 times. Centrifuged at 14,000 ×g for 5 min to pellet precipitate. Transferred supernatant to a clean 15 ml tube. Added 1 ml Binding Matrix Suspension to the supernatant. Placed on a rotator or inverted by hand for 2 min to allow binding of DNA to matrix. Placed tube in a rack for 3 min to allow settling of silica matrix. Discarded 500 μl of supernatant. Resuspend Binding Matrix in the remaining amount of supernatant. Transferred approximately 600 μl of the mixture to a SPIN^TМ Filter and Centrifuged at 14,000xg for 1 min. Emptied the catch tube and added the remaining supernatant to SPIN^TМ Filter and spun again. Added 500 uL SEWS-M to the SPIN^TМFilter and Centrifuged at 14,000 ×g for 1 min. Decanted flow-through and replaced SPIN^TМ Filter in catch tube. Centrifuged at 14,000 ×g for 2 min to “dry” the matrix of residual SEWS-M wash solution. Removed SPIN^TМ Filter and placed in fresh kit-supplied catch tube. Air dried the SPIN^TМ Filter for 5 min at room temperature. Added 50 μl DES and gently stirred matrix on filter membrane with a pipette tip or vortex/finger flipped to resuspend the silica for efficient elution of the DNA. Centrifuged at 14,000 ×g for 1 min to transfer eluted DNA to catch tube. DNA was application-ready.

DNA quality was assessed by 1% agarose gel electrophoresis (5 V/cm, 20 min). The V4-V5 region of the bacterial 16S rDNA was used for high-throughput sequencing. Prior to PCR amplification, the 338F forward primer was combined with 0.8 μl ACTCCTACGGGAGGCAGCAG (5 μmol/l) forward primer adapter, and the 806R reverse primer was combined with 0.8 μl GGACTACHVGGGTWTCTAAT (5 μmol/l) reverse primer adapter. PCR amplification was carried out in a 20-μL reaction system: 5× Buffer (4 μl), 2.5 mmol/l dNTPs (2 μl), TransStart Fastpfu DNA Polymerase (0.4 μl), template DNA (10 ng), plus ddH₂O to 20 μl. The PCR reaction was carried out according to the following parameters: pre-denaturation at 95°C for 5 min, denaturation at 95°C for 3 min, annealing at 55°C for 3 s, 29 cycles; extension at 72°C for 10 min, hold at 10°C (ABI GeneAmp 9,700). The PCR products were sequenced reagents on an Illumina HiSeq platform. Data processing and biodiversity analysis were performed on the I-Sanger bioinformatics analysis cloud platform.¹

After the paired-end (PE) sequence data were obtained by HiSeq sequencing, the PE reads were spliced into one coding sequence according to the overlap relationship between PE reads. Low-quality reads and chimera sequences were filtered out, and effective sequences were distinguished according to the barcode and primer sequences (Caporaso et al., 2010). Clean reads were used to perform taxonomic and operational taxonomic unit (OTU) cluster analysis. The QIIME platform was used to cluster sequences with a similarity of ≥0.97 into OTUs, and the longest sequence of each OTU was selected as the representative sequence. The representative OUT sequences were aligned using the Silva database² to obtain taxonomic information for each OTU. Species composition and redundancy analysis (RDA) were performed on the I-Sanger bioinformatics analysis cloud platform. The Mothur v.1.30.1 index analysis package³ was used to calculate alpha diversity indices, including Shannon-Wiener, Simpson, ACE, and Chao1.

PICRUSt software was used for functional prediction of 16S amplicon sequencing results. First, PICRUSt was used to normalize OTU abundance. Then, the Greengene ID of each OTU was used to obtain COG (clusters of orthologs groups) family and KEGG (Kyoto encyclopedia of genes and genomes) Ortholog (KO) information. The eggNOG database was used to obtain the functional abundance spectrum based on the COG database. The KEGG database was used to obtain pathway, KO, and EC information, as well as the abundance of each functional category. Finally, PICRUSt was used to obtain three levels of metabolic pathway information and pathway abundance.

The rhizosphere soil chemical properties were analyzed across all nine sample sites (Figure 5). Overall, the soil across all sites was found to be more or less acidic, with significant differences between sites. Across sites, significant differences were found in OM, AN, TP, TK, and AK, with small differences in AP, and no significant difference in TN. Correlation analysis between soil chemical factors showed that soil pH were positively correlated with the 7 soil nutrient indicators, especially with AK, AP and OM (P < 0.01); TK with TP and TP with OM were negatively related (P < 0.01; Table 3). Overall, both the pH and nutrient contents of LB and LY were higher than the other sites, with ZF having the lowest pH and nutrient content. Across all sites, the AP content tended to be low while the AK and AN contents tended to be high. According to the “Nutrient Classification Standard of the Second Soil Census of China,” across sites the AP content would be classified as grade 4 (5–10 mg/kg) or grade 5 (3–5 mg/kg), the AK content would be classified as grade 1 (>200 mg/kg), and the AN content would be classified as grade 1 (>150 mg/kg).

The content of several medicinally important bioactive compounds of C. migao fruit, including 1,8-cineole, sabinene, limonene, and α-terpineol, were determined by GC (Figure 6). The most plentiful compound was 1,8-cineole, followed by α-terpineol, sabinene, and limonene. Fruit from LY contained the highest content of 1,8-cineole, α-terpineol, and limonene, while fruit from FN contained the lowest content of all compounds. The differences in the content of bioactive compounds of C. migao fruit from different regions were considerable, suggesting that differences in environmental and soil characteristics translate into vast differences in bioactive compound accumulation.

RDA was used to examine the relationships between the top 20 bacterial genera and soil chemical and nutritional properties (Figure 7A), with the horizontal axis as the first ranking axis (57.57% contribution) and the vertical axis as the second ranking axis (17.97% contribution). The bacterial genera Nitrospira, Xanthobacteraceae (unidentified), H16, and Gaiellales (unidentified) were positively correlated with all eight soil chemical indicators. Rhizobiales (unidentified) was positively correlated with AN, but not TN. Bryobacter and Acidobacteria (unidentified) were positively correlated with AP, but not TP. Bradyrhizobium was negatively correlated with seven soil chemical indicators. Most of the other bacterial genera, such as Acidibacter, Acidobacteriaceae subgroup 1 (unidentified), Candidatus-Solibacter, Variibacter, Acidothermus, Acidimicrobiales, and Alphaproteobacteria, were negatively correlated with the eight soil chemical indicators to different degrees. Additionally, pH, AN, and TN were strongly correlated, and together had the greatest influence on bacterial community composition at LB, LY, and FN.

RDA was also used to examine the relationships between the top 20 bacterial genera and the content of C. migao fruit bioactive compounds (Figure 7B), with the horizontal axis as the first ranking axis (12.49% contribution) and the vertical axis as the second ranking axis (7.20% contribution). The bacterial genera Acidothermus, Acidibacter, Bryobacter, Candidatus_Solibacter, and Acidimicrobiales (unidentified) were positively correlated with all four bioactive compounds, particularly, Acidothermus and limonene have a strong positive correlation. Acidobacteria (unidentified) was positively correlated with all bioactive compounds except limonene. Nitrospira, Alphaproteobacteria and Xanthobacteraceae (unidentified) were negatively correlated with all four bioactive compounds.

Analysis of the bacteria present in the rhizosphere of C. migao from different regions revealed rich and varied communities. The dominant genera of rhizosphere bacteria from different habitats were basically the same, although there are some differences in abundance. Overall, Proteobacteria (34.81%), Acidobacteria (21.85%), Actinobacteria (16.71%), and Chloroflexi (10.16%) were the dominant bacterial phyla across all field sites; the defined bacterial genera, including Variibacter (5.40%), Acidothermus (4.59%), Alphaproteobacteria (3.99%), Candidatus_Solibacter (3.18%), Bradyrhizobium (3.05%), Nitrospira (2.86%), Acidibacter (1.91%) and Bryobacter (1.88%) were found to be the dominant genera. In comparing with previous studies, there are some species with similar dominant bacterial genera to C. migao, such as: natural mixed forests, apple trees, Abies Fabri, Robinia Pseudoacacia, and Cunninghamia Lanceolata (Jiang et al., 2017; Yang, 2019; Zhang, 2019; Nie et al., 2020; Xiong et al., 2021). However, some species dominant-phylum are as similar as 90% to C. migao, the specific bacterial genera differ from our results, such as: Taxus, which is an arboreal medicinal plant (Hao et al., 2016). Additionally, some plants have different genotypes from C. migao, and their dominant taxa are distinct (Li et al., 2021). Studies have shown that plants with different genotypes may differ in their morphological characteristics, ecological habits, and physiological traits. There may also be differences in the characteristics of the rhizomicroorganisms they recruit. For example, (a) species with different ecological habits have more discrepancy’s in their colonizing environments, e.g., rice vs. maize, and their rhizomicrobes also differ significantly (Li et al., 2014; Nevita et al., 2018). (b) Plants with different phenotypes have disparate environmental effects, e.g., variances in the depth, number and size of plant roots makes distinct effects on soil amelioration, ultimately leads to differences in enriched rhizobacteria. (c) Plants with different physiological characteristics may differ in their metabolites, and the secretion and release of specific metabolites may inhibit some genera that are not adapted to the type and concentration of the products, while attract those who use the products as a nutritional feedstocks and structural materials. It is worth mentioning that legumes actively recruit rhizobia for forming a mutually beneficial symbiosis (Glyan Ko, 2015).

After comprehensive analysis, we found that plants with similar dominant genera to C. migao were mostly derived from forestland (C. migao’s samples also from forests), and these dominant genera reflected the bacterial structure characteristics in specific ecological environment, which differed from the rhizobacterial compositions of crops or herbs in farmland (Zhang et al., 2022). This may be a commonality, as arboreal (forestland) recruited bacteria which have longer colonization periods and more stable ecotope than herbs (farmland). Besides, because of the abundant diversities of bacteria and the more complex mechanisms of interaction between arbors and microorganisms, there are still a large number of undefined bacterial taxa in forests (Du et al., 2017; Zhang et al., 2022), more than half taxa are undefined in C. migao rhizobacteria, and the defined are basically rare. Researches on the composition and function of sylvatic microorganisms should be strengthened in the future.

Analysis of soil chemical and nutritional properties indicated that there were significant differences in pH, TK, TP, and OM between sites, while differences in AN, AK, and AP differed only slightly, and TN did not differ at all between sites. Both the pH and nutrient content of soils from LB and LY were higher, while ZF soils had the lowest pH and nutrient content. Furthermore, the contents of four medicinally important bioactive compounds (1,8-cineole, sabinene, limonene, and α-terpineol) from C. migao fruit were different significantly between sites, with fruit from LY, WM, and TE containing the highest content of bioactive compounds, the minimum was ZF. This may be related to G × E (genotype × environment) interactions, Studies have confirmed that different genotypes have genetic traits for variable chemical compositions. For example, the active constituents of Taxus (Hao et al., 2016), Cornus officinalis (Sun et al., 2022), Astragalus mongholicus (Li et al., 2021), and Baphicacanthus cusia (Zeng et al., 2018) differ from C. migao. Besides, widespread species in order to adapt to different environments, leading to the creation of stably inheritable variation that results in multiple genotypes of the same species. The chemical components and contents of these genotypes may differ, e.g., the polyphenols and antioxidants contents of Sorghum were higher in two genotypes and simultaneously did not differ significantly between some genotypes (Aruna et al., 2020). Based on the small genetic differentiation among populations of C. migao (Li et al., 2018), we speculate that environmental heterogeneity are more likely to be responsible for the differences in active ingredients and dominance of homogeneous genera in C. migao. This is substantiated by the reports of others (Li et al., 2016), indicating that differences in environmental conditions, soil properties, and microorganisms can result in substantial differences in the quality of medicinal plant materials.

The structures and diversities of microbial community are meaningful indicators for evaluating ecosystem stability (Chapin et al., 2011; Romaniuk et al., 2011; Wigginton and Amador, 2020). Several environmental factors affect soil microbial community structure and diversity, including primarily climatic, phytic, and soil-specific factors (Marschner et al., 2001; Yang et al., 2019; Wang et al., 2021). Of these, soil properties have been found to be the primary driver of microbial community structure (Philippot et al., 2013), particularly soil pH (Scola et al., 2018). Our results are in agreement with these previous studies, indicating that pH is a key driver of rhizobacterial community structure (Jiang et al., 2017). Differences in response patterns of bacteria-specific to pH gradients were observed, for example, Nitrospira, Gaiellales (unidentified), Xanthobacteraceae (unidentified), and H16 were positively correlated with pH, while Acidothermus, Variibacter, Bradyrhizobium, Acidibacter, Acidobacterianace (unidentified), Alphaproteobacteria and Candidatus-Solibacter were negatively correlated with pH. This dual response is consistent with other studies (Jones et al., 2009; Dimitriu and Grayston, 2010; Rousk et al., 2010). The acidic soil environment of C. migao may provide suitable conditions for acidophiles to colonize and enrich, such as, Acidothermus, Alphaproteobacteria, Acidibacter, and Acidobacterianace (unidentified). Additionally, Nitrospira and Xanthobacteraceae (unidentified) were found to have a close positive correlation with TN and AN, nitrogen levels are also controlling factors for C. migao rhizobacteria.

In fact, there is an interaction, soil physicochemical factors not only affect microorganisms, but also microorganisms can react to soil properties. For example, Acidibacter as an acidophilic ferric reducing gammaproteobacterium can causing the pH to increase when it catalyzed the reductive dissolution of the ferric iron mineral schwertmannite (Falagán and Johnson, 2014); Nitrospira plays pivotal role in nitrification as an aerobic chemolithoautotrophic nitrite-oxidizing bacterium, metabolic versatility enables Nitrospira to colonize a broad range of habitats and to sustain shifts in environmental conditions such as changing oxygen concentrations and promoting nitrogen-cycling (Daims et al., 2015; Daims and Wagner, 2018), this is verified by the higher Nitrospira abundance and AN contents in LB samples. It’s showed that the enrichment of acidophiles and Nitrospira may be primarily responsible for the acidity and nitrogen-rich in C. migao rhizosphere soil. The further description is that C-, N- and P-cycles in soil are subject to the allocation of microbial community structures and functions, the growth and activity of endemic rhizomicroorganisms can promote the transformation and storage of effective nutrients in rhizosphere soil, which stimulate the uptake of nutrients by plant roots.

In addition, we found connection between soil pH, soil nutritional properties, and rhizobacterial diversity. On the one hand, pH and soil nutrients are positively correlated, especially with AK, AP and OM, pH can affect the effectiveness of soil nutrients, and it may be a key factor in regulating soil nutrients and promoting plant nutrient uptake. On the other hand, there are also positive correlation among pH, nutrients and rhizobacterial diversity. For example, soils with very low pHs tended to have lower nutritional content and, consequentially, lower rhizobacterial diversity, as was the case for ZF. Contrarily, soils with higher pH tended to have higher nutritional content and rhizobacterial diversity, as was the case for LY. This supported that soils with higher nutrient content tend to have higher microbial diversity (Maestre et al., 2015; Gebhardt et al., 2017), but other samples (WM, CH, LB, etc.) do not follow this universal rule. So other factors affecting rhizomicrobial diversity should be taken into account.

We comprehensively analyzed the geoclimatic information across all sample sites, the results showed that although a geoclimatic analysis of the nine sample sites indicated a significant elevation gradient, no significant correlations were found between elevation and soil properties, rhizobacterial diversity. This is consistent with reports indicating that bacterial community distribution may not follow the same elevational distribution patterns exhibited by plants or animals (Shen et al., 2013), but contrary to other reports indicating that bacterial communities do differ by elevation, despite perhaps indirectly (Fierer et al., 2011). Alternatively, temperature may be an important factor in the differentiation of the rhizobacterial diversity, for example, the low rhizobacterial diversity of ZF may be related to its relatively lower mean annual temperature (17.98°C), and the higher bacterial diversity of WM may be related to its relatively higher mean annual temperature (20.29°C). Soil properties explained 75.54% (RDA1 57.57%, RDA2 17.97%) of the variation in rhizobacterial community structure. This explanation rate was close to apple trees of around Bohai Gulf, with a cumulative explanation rate of 78.88% (RDA1 50.43%, RDA2 28.45%) for RDA (rhizobacteria × soil properties; Jiang et al., 2017), and both major contributing factors included pH and AN; Betula albosinensis was 57.43% (RDA1 42.65%, RDA2 14.72%), and its dominating factors showed SOC > C:N > TN > pH (Du et al., 2017). This discrepancy may be related to the different soil factors selected. The explanation rate of 75.54% indicates that soil chemistry is the dominant influence of rhizobacteria in C. migao, which may be connected with the existence of long-term, direct × indirect interactions with rhizomicroorganisms in a combination of multiple ways. The related percentage of variation unexplained by factors studied here (24.46%) suggests that there are other environmental or biological factors influencing C. migao rhizobacterial community structure. Future studies should examine the effect of other climatic, geographical, soil physiochemical properties (e.g., soil enzymatic activity, aeration, and water permeability), and associated plant communities on C. migao rhizobacterial community structure and diversity.

Plant growth-promoting rhizobacteria (PGPR) colonize plant roots and stimulate plant growth, stress tolerance, and defense (Kloepper and Schroth, 1978). The widely reported PGPR in C. migao include Bradyrhizobium and Bacillus (Podile and Kishore, 2006; Lugtenberg and Kamilova, 2009; Haney et al., 2015; Xie et al., 2019), which have disease-preventing and growth-promoting effect on C. migao. They promote plant growth by enhancing nutrient acquisition, niche competition, the production of antimicrobial compounds, or inducing systemic resistance to prevent pathogen infection (Zamioudis et al., 2013; Pieterse et al., 2014, 2016; Chowdhury et al., 2015). Recent studies have demonstrated that not only do rhizobacteria and their metabolites affect the growth and stress resistance of plants, but they can also impact plant accumulation and synthesis of medicinal bioactive compounds (Köberl et al., 2013; Huang et al., 2018; Zhang et al., 2020). There are three primary mechanisms by which rhizomicroorganisms affect the accumulation of bioactive secondary metabolites in medicinal plants. First, microorganisms activate secondary metabolite signaling pathways in plants, resulting in the upregulation of secondary metabolite biosynthesis (Trivedi et al., 2020). Second, microorganisms themselves produce bioactive secondary metabolites which stimulate the upregulation of secondary metabolite biosynthesis in plants (Schmidt et al., 2014). Third, microorganisms improve the stress resistance of medicinal plants, which is often mediated by the biosynthesis of secondary metabolites (Taghinasab and Jabaji, 2020). Relevant representative studies, Piriformospora indica and Azotobacter chroococcum can significantly enhance the artemisinin content of Arteuaisia annua (Arora et al., 2016). Stenotrophomonas maltophilia and Pseudomonas fluorescens can significantly enhance the hypericin and hyperforin content of Hypericum perforatum (Manero et al., 2012). Pseudomonas can regulate gene expression in Taxus cells by releasing the phytotoxin-coronatine, which may be responsible for the resultant increase in taxane content (Onrubia et al., 2013). Burkholderia sp. increases the synthesis of indigo in Baphicacanthus cusia (Zeng et al., 2018). Bacillus pumilus enhances glycyrrhizic acid content by increasing the expression of key enzymes (Xie et al., 2019).

Here, we found that several bacteria, including Acidothermus, Acidibacter, Bryobacter, Variibacter, Candidatus_Solibacter, and Acidimicrobiales (unidentified) were positively correlated with the content of bioactive compounds in C. migao. Interestingly, Acidobacteria (unidentified) was positively correlated with all compounds except limonene, and Nitrospira, Alphaproteobacteria, and Xanthobacteraceae (unidentified) were negatively correlated with all four bioactive compounds. Another important finding is that the functional classification of C. migao rhizobacteria tended to be homogenous across regions, with metabolism-related functions being the most common. Therefore, we speculated that C. migao may selectively recruit specific microorganisms in rhizosphere that could metabolize certain substances to effect C. migao. Based on the data obtained in this study, the enrichment of bacterial genera Acidothermus, Acidibacter, Bryobacter, Variibacter, Candidatus_Solibacter, and Acidimicrobiales potentially promote the biosynthesis and accumulation of 1,8-cineole, cypressene, limonene, and α-terpineol in C. migao, while Nitrospira and Alphaproteobacteria may be repressive for the bioactive ingredients. However, the mechanism by which these rhizobacteria alter the accumulation of bioactive compounds remains to be identified.

Among the enriched taxa in C. migao rhizosphere. Acidothermus was widely known as the Acidothermus celulolyticus endoglucanase (E1; Larry et al., 2008). It has been recorded to colonize the rhizosphere of Lettuce (Cipriano et al., 2016), and found to be associated with the oxidation of sulfur, nitrogen and iron (Wang et al., 2022). This oxidation alters the uptake of sulfur, nitrogen, and iron by plants, affecting the synthesis of plant metabolites, in this study, Acidothermus may be a potentially beneficial bacteria used to promote the accumulation of bioactive compounds (especially limonene) of C. migao. Acidibacter, Bryobacter, Variibacterr, and Candidatus_Solibacter may have a slight stimulating effect on the accumulation of bioactive ingredients in the fruit of C. migao. Acidibacter colonized in many plant species, and associated with soil nutrient and iron cycles, plant blights, and soil pollution controls (Falagán and Johnson, 2014; Liu et al., 2016; Jiao et al., 2018; Huang et al., 2022), it may has a variety of potential functions. Bryobacter has been reported as beneficial bacteria in studies such as leguminous plants and Marchantia, where it plays roles in plant-growth promotion, complex exudate degradation, nitrogen fixation, methylotrophs, and disease-suppressive (Xiao et al., 2017; Luis et al., 2018). Variibacterr has been documented in soil contamination and nutrient cycle studies (Jiao et al., 2018; Wu et al., 2020), which may be a microbial community with degradation potential, although it has decreased abundance in tobacco-peanut intercropping soils with Burkholderia (Gao et al., 2019), it may also have biocontrol roles as Burkholderia. The abundance of Candidatus_Solibacter were increased in healthy sesame rhizosphere soils with Bryobacter and Bradyrhizobium, they were beneficial to improving the soil microecology and alleviating continuous cropping obstacle (Wang et al., 2017). Alphaproteobacteria is a diverse class of organisms within the phylum Proteobacteria, its members inhabit diverse environmental niches and have many important biological roles, they participate in a variety of metabolic strategies, including nitrogen fixation, photosynthesis, ammonia oxidation, and methylotrophy (Williams et al., 2007; Andreote et al., 2009; Collier, 2016), in this study, Alphaproteobacteria may suppresses the accumulation of C. migao bioactive ingredients by participating in a particular metabolisms. Nitrospira was widely reported as an aerobic autotrophic nitrite-oxidizing bacteria that plays a key role in plant nitrification and promotes efficient conversion and storage of nitrogen (Daims et al., 2015; Daims and Wagner, 2018); additionally, Nitrospira has also been reported to be associated with verticillium and fusarium wilts of cotton and watermelon (Zhu et al., 2019), the complexity of Nitrospira functions makes it possible to both as beneficial bacteria to stimulate plant absorption of effective nutrients, as well as a pathogen to induce systemic resistance of plants, Nitrospira may inhibit the accumulation of medicinal bioactive ingredients of C. migao. Overall, the enriched strains and their metabolites may promote or inhibit the growth and quality of C. migao through a variety of pathways. We suggest that the rhizosphere of wild-type-C. migao are enriched with a large number of potentially valuable bacteria. It provided a reference for the next step of screening beneficial bacteria to promote the growth and quality of C. migao. Meanwhile, the function of specific microbial strains should be validated, and the mechanisms of specific strains on bioactive ingredients in C. migao should be elucidated in the future.

The effective magnitude of rhizomicroorganisms on bioactive compounds is likely to be species-, tissue-, and contact-dependent. For example, in herbs with roots as medicinal parts, 57 OTUs are positively (0.9 ≥ R² ≥ 0.51) or negatively (−0.85 ≤ R² ≤ −0.52) correlated with the bioactive ingredient in Astragalus mongholicus (Li et al., 2021); 13 rhizobacteria genera are significantly (P < 0.05) associated with bioactive ingredient in Scutellaria baicalensis (Jiang, 2021). In herbs with the whole plant as medicinal parts, the correlation of bioactive ingredients of different medicinal parts with rhizobacteria is shown as aboveground < underground (R²) in Rumex japonicus (Ni, 2019). The relationship between three medicinal licorices endophytic bacteria and the bioactive ingredients showed that the cumulative interpretation rate was 22.33% (RDA1 12.55%, RDA2 9.78%; Dang et al., 2020). Combined with our results, soil properties explained 75.54% of the variation in rhizobacterial community structure, whereas rhizobacteria explained 19.69% of medicinal components. (a) We speculated that the above phenomena may be related to the effect of rhizobacteria on the medicinal ingredients in roots or rhizomes of herbs could rely on dual influences of direct and indirect, while fruit trees could basically depend on indirect effects. (b) The bioactive ingredients of fruits and rhizobacteria are not the dominant influencing factors for each other despite they somewhat correlated. The remaining 80.31% of unexplained factors may be related to other unmeasured indicators, such as: temperature, light, moisture, endophytes, other rhizomicroorganisms, and rare genera (relative abundance < 1%). Although the interpretation of interaction between arbor rhizobacteria and fruit medicinal ingredients is limited, it cannot be ignored, which can provide a possibility for understanding the complex ecological characteristics of C. migao and developing C. migao medicinal resources.

In summary, this study revealed the complex correlation of rhizobacteria with bioactive ingredients and soil properties, which provides us with an opportunity to improve and enhance the quality of C. migao from a microbial perspective.