Based on previous literature research, the sampling locations and methods were determined. The sampling sites were chosen in Guangxi, Guangdong, and Fujian provinces, which are the major producing areas of Z. nitidum. Soil samples were collected from an area with a radius of 5 mm and a depth of 30 cm around the roots of Z. nitidum, shown in Figure 1A . Impurities were removed from the samples before they were thoroughly mixed and placed in ice boxes. Finally, the soil samples were transported back to the laboratory and stored in a refrigerator at 4 °C for future use.

Specifically, sample A1 was obtained from Nanning, Guangxi; sample A2 from Yulin, Guangxi; sample A3 from Guangzhou, Guangdong; sample B1 from Hezhou, Guangxi; sample B2 from Fuzhou, Fujian; and sample B3 from Ningde, Fujian. These soil samples were collected from the root zones of wild plant seedlings. The details of the samples (such as longitude, latitude, height and ecological environment information) are presented in Table 1 and Figure 1B .

According to the agricultural industry standard of China (NY/T 1377 - 2007), the pH value of the soil samples was measured. The organic matter content was determined by referring to the Chinese agricultural industry standard (NY/T 1121.6 - 2006). The soil texture was analyzed using the pipette method specified in the China forestry industry standard (LY/T 1225 - 1999).

DNA was extracted from the soil samples using a soil DNA extraction kit (Mag - Bind Soil DNA kit from OMEGA company). The extraction steps were performed following the kit’s instructions. The concentration of genomic DNA extracted from each sample was determined by Qubit 2.0 and loaded on 1% agarose gel to detect DNA integrity. Sample DNA meeting the requirements of concentration (0.05 ng/ul~120 ng/ul) and integrity was amplified by PCR.

Bacterial 16S rDNA Amplification and Sequencing: the V3 - V4 variable region of bacterial 16S rDNA was chosen as the amplification target region. Illumina MiSeq paired - end sequencing (2 × 300 bp) was employed to perform the sequencing. PCR Primers: the primers utilized in the PCR incorporated the V3 - V4 universal primers specific to the sequencing platform. The 341F primer sequence was 5’-CCTACGGGNGGCWGCAG-3’, and the 805R primer sequence was 5’-GACTACHVGGGTATCTAATCC-3’. First - Round PCR Amplification: the reaction system for the first - cycle PCR amplification was as follows: 15 μl of 2×Taq master Mix, with Bar - PCR primer F (10 μM) and Primer R (10 μM) added to a total volume of 30 μl. The PCR amplification program consisted of the following steps: initial denaturation at 94 °C for 3 min, five cycles of: 94 °C for 30 s, 45 °C for 20 s, and 65 °C for 30 s, twenty cycles of: 94 °C for 20 s, 55 °C for 20 s, and 72 °C for 30 s, final extension at 72 °C for 7 min. Second - Round PCR Amplification: in the second round of PCR amplification, primers compatible with Illumina bridge PCR were introduced. The amplification reaction system contained 15 μl of 2×Taq master Mix, 1μl each of primer F (10 μM) and Primer R (10 μM), and 2 μl of the PCR products from the previous round, with the total volume adjusted to 30 μl. The PCR amplification program was as follows: initial denaturation at 95 °C for 3 min, thirty cycles of: 94 °C for 20 s and 72 °C for 30 s, final extension at 72 °C for 5 min. PCR products were detected by agarose electrophoresis. The amplification products that meet the requirements of more than 400bp are recovered by magnetic bead method. The recovered DNA was accurately quantified by using Qubit 3.0 DNA detection kit, and was mixed in equal amount of 1: 1 for sequencing. When mixed equally, the amount of DNA in each sample was 10ng, and the final sequencing concentration was 20 pmol. Finally, the samples were sent to Sangon Biotech (Shanghai) Co., Ltd. for sequencing.

(1) Processing and statistical analysis of soil physical and chemical data: Excel 2013 was used for data processing and statistical analysis, which was listed in the format of “mean ± standard deviation”. (2) Processing and statistical analysis of Miseq sequencing sequence data: After the sequencing data was stripped of primer joints, the sequences were spliced by PEAR (version: 0.9.6), and each sample data was identified and cut by PRINSEQ software (version: 0.20.4) according to the tag sequence. Then, the low-quality data in each sample data was filtered through quality control to obtain qualified sequences for each sample (see quality control data chart, which was placed in Supplementary Material ). After chimeras and nonspecific amplified sequences were removed from the quality-controlled data using USEARCH (version: 5.2.236) and UCHIME (version: 4.2.40) software, the effective sequences of each sample were obtained. Using a similarity threshold of 97%, the valid sequences of each sample were clustered into operational classification units (OTU), and the analysis, screening, and classification of OTUs were completed using USEARCH software (version: 5.2.236). RDP classifier software was used to classify species, with the confidence threshold set to 0.8. The OTU sequences were compared with the 16S bacterial database in RDP (https://sourceforge.net/projects/rdp-classifier/), and sequences with similarity > 90% and coverage > 90% were set to be used for subsequent classification, while sequences that did not meet the conditions were classified as unclassified. R software (version: 3.2) was used to complete the analysis and mapping of α and β diversity of samples, the function of differential flora, and the influence of soil environmental factors on flora structure. GenBank’s functional gene database was used for functional analysis of different flora.

The results of the soil pH determination for six samples are presented in Table 2 . Among these samples, all except A1 (pH = 7.7) were acidic (pH < 7). The pH values of the six samples followed the order: A1 > A3 > B3 > B2 > A2 = B1. In terms of organic matter content, sample A1 (54.4 ± 0.02 g/kg)exhibited the highest value among all the samples, while sample B3 (9.55 ± 0.25 g/kg) had the lowest. Regarding the clay particle level, sample A1 (593.11 ± 0.03 g/kg) had the highest content, whereas sample A3 (278.05 ± 22.96 g/kg) had the lowest. Conversely, sample A3 (523.05 ± 42.73 g/kg) had the highest content of the grain level, and sample A1 (192.84 ± 3.79 g/kg) had the lowest. For the powder (sand) grain level, sample B2 (221.58 ± 0.38 g/kg) had the highest content, and sample A3 (198.9 ± 16.29 g/kg) had the lowest. In terms of soil texture, all samples except A3, which was classified as sandy clay loam, were identified as clay.

Through high - throughput sequencing analysis of the V3 - V4 variable region of bacterial 16S rDNA from six rhizosphere soil samples, a total of 511,904 original readings were obtained. Subsequently, after merging the double - terminal reads and performing quality control filtration, 499,430 clean tags were acquired. The ARCH software (version 5.2.236) was employed to eliminate some non -amplified sequences in all samples and correct sequencing errors. Chimeras were identified using the Uchime software (version 5.2.236). Then, the sequences with chimeras removed were compared with the representative sequences in the database using BLASTN to detect and remove out - of - target sequences. Finally, 484,452 effective sequences were successfully obtained. The processing results of the sequencing data for each sample were presented in Table 3 .

Alpha diversity analysis reflected the richness and diversity of fungal communities in rhizosphere soil samples of Z. nitidum from different habitats. In the context of α-diversity indices, the Chao 1 index and the ACE index were indicative of the richness of the sample community, while the Shannon index and the Simpson index reflected community diversity. A higher Shannon index implied greater biodiversity in the samples, whereas a higher Simpson index suggested lower biodiversity. The coverage index indicated the extent to which the sequencing results represented the actual situation of the samples (Cai et al., 2018; Fu et al., 2019). As shown in Table 4 , Chao 1 index and ACE index of six soil samples were ranked as follows: A3 > B3 > A2 > B2 > A1 > B1. Both indices demonstrated the same trend in soil fungal abundance: sample A3 had the highest abundance, followed by B3, and B1 had the lowest. Besides, the Shannon index and the Simpson index of the six soil samples showed the same order in terms of sample diversity: A3 > A1 > A2 > B1 > B3 > B1. To sum up, it showed that the bacteria in the six soil samples all had high species richness and high similarity, but there were some differences in their community structure and relative abundance of fungal groups. The bacteria in sample A3 had the highest species diversity, followed by sample A1, and sample B1 had the lowest species diversity.

The coverage rate of each sample exceeded 0.97, suggesting that the sequencing results accurately represented the actual bacterial composition in the samples.

The optimal comparison results for the OTU sequences were then selected. The classification results of six bacterial community samples at different taxonomic levels were presented in Table 5 . At the phylum classification level, the composition of the bacterial communities in the six soil samples was shown in Figure 2A . The data from Table 5 and Figure 2A indicated that the bacteria in the samples primarily belonged to ten bacterial phyla, namely Proteobacteria, Acidobacteria, Actinobacteria, Verrucomicrobia, Bacteroidetes, Planctomycetes, Firmicutes, Chloroflexi, Gemmatimonadetes, and Unclassified. Notably, there were significant differences in the composition and species richness of the bacterial communities across all samples. Proteobacteria and Acidobacteria emerged as the dominant bacteria in the bacterial communities (> 10% relative abundance). Among them, Proteobacteria showed the most prominent dominance and the highest species richness. For instance, in sample B3, the abundance of Proteobacteria was the highest, reaching 56.77%, followed by sample B2 with 53.39%, and sample B1 with the lowest abundance at 35.03%.

In a heat map ( Figure 2B ), each color block represented the relative richness of a species. A redder color block indicated a higher relative abundance, while a bluer color signified a lower relative abundance. Simultaneously, heat map analysis could also present the results of sample clustering (Wang et al., 2018; Qin et al., 2019). Based on the heat map clustering trees at each taxonomic level, when the similarity of community structure and relative species abundance of samples was used as the criterion for clustering analysis, the clustering results of the six samples were consistent across all taxonomic levels. The six samples could be roughly divided into two groups. Samples A2, B1, and B2 were classified into one group, while samples A1, A3, and B3 were placed in another. Within the group of A2, B1, and B2, samples A2 and B1 exhibited a high degree of similarity in community structure and relative species richness. Similarly, within the group of A1, A3, and B3, samples A1 and A3 showed a high level of similarity in community structure and relative species richness.

However, there were certain disparities in the relative species richness. Excluding the Unclassified group, the top 5 genera in terms of relative abundance for each sample were as follows: For sample A1, the top 5 genera and their relative abundances were: Unclassified at 30.91%, GP6 at 10.74%, povalicharacter at 3.52%, Sphingomonas at 3.41%, and Lacibacterium at 2.02%. For sample A2, they were: Unclassified at 22.6%, Gp1 at 9.52%, Gp2 at 8.93%, Burkholderia at 7.36%, and Subdivision3_genera_incertae_sedis at 3.48%. For sample A3, the values were: Unclassified at 28.84%, Sphingomonas at 14.73%, Gp6 at 6.66%, Lysobacter at 3.27%, and Spartobacteria_genera_incertae_sedis at 1.56%. For sample B1, the top 5 genera had relative abundances of: Unclassified at 25.99%, Gp1 at 10.28%, Spartobacteria_genera_incertae_sedis at 8.2%, Gp2 at 6.64%, and Povallibacter at 4.56%. For sample B2, the relative abundances were: Unclassified at 20.95%, Erwinia at 14.72%, Pantoea at 12.2%, Subdivision3_genera_incertae_sedis at 3.4%, and Gp1 at 3.18%. For sample B3, the top 5 genera and their abundances were: Sphingomonas at 19.14%, Unclassified at 18.32%, Rudaea at 4.04%, Bradyrhizobium at 3.07%, and Gemmatimonas at 2.93%.

Beta diversity was the comparison of diversity between different ecosystems, and it was the rate of change of species composition along the environmental gradient or between communities, which was used to express the response of biological species to environmental heterogeneity. Distance thermogram could intuitively show the distance relationship between samples by color, that is, the similarity between samples. The color block represented the distance value. The redder the color, the closer the distance between samples, the higher the similarity, and the bluer the distance. The samples were clustered in the heat map, and the distance relationship between samples could also be seen through the cluster tree. Using the heat map package of R, a heat map was drawn according to the Unifrac distance matrix between samples. As shown in Figure 3A , samples A1 and A3 were relatively close, indicating a relatively high similarity, so they could be classified into one category. Similarly, samples A1 and B2 could also be grouped together. The distance between other samples was relatively large, indicating that the similarity was relatively low, and each of these samples could be regarded as a different category. Among them, the high similarity between samples A1 and A3 might be attributed to their similar latitudes, while the high similarity between samples A2 and B1 might be attributed to their close geographical locations. Therefore, we speculated that geographical location (such as longitude or latitude) may have had an influence on the OTU abundance of samples.

Heat map could reflect the abundance information of flora function by color change, and could intuitively express the functional abundance value by defined color depth. At the same time, six samples and functional information were clustered and rearranged, and the results after clustering were shown in Figure 3B . Based on the COG functional classification abundances, the functional distribution of microbial communities in six samples was analyzed across 25 functions, such as General function prediction only, Amino acid transport and metabolism, Energy production and conversion, Transcription, Cell wall/membrane/envelope biogenesis, and so on. The sequences of the functions of the microbial flora in the six samples were compared to predict and analyze the differences in their abundances. Figure 4 showed the functional abundances of the microbial communities in the six samples. We could clearly and intuitively see that the highest functional abundance of the six samples was General function prediction only, and the lowest functional abundance was RNA processing and modification, Chromatin structure and dynamics, Extracellular structures, Nuclear structure, Cytoskeleton. In addition, on the Heat map, we could clearly see that the flora functions of sample A2 and sample B1 were highly similar. The highest abundance of flora functions of sample A2 and sample B1 was General function prediction only, followed by Function unknown, Transcription, Signal transduction mechanisms, Amino acid transport and metabolism, Carbohydrate transport and metabolism. Similarly, the flora functions of sample A1 and sample A3 were very similar. The highest abundance of flora functions in sample A2 and sample B1 was General function prediction only, followed by Function unknown, Amino acid transport and metabolism, Transcription. In summary, the samples with close geographical distance often had high similarity in the cluster analysis of six samples, and their microbial community functions were also closely related. This observation show the potential correlation between geographical location and the functional characteristics of microbial communities in these samples.

According to DCA analysis, the axis length was less than 3.0. Therefore, Redundancy Analysis (RDA) was employed to analyze the correlation between soil bacterial flora and environmental factors. In this study, soil pH, organic matter content (ORM), altitude (HT), soil clay (CP), soil sand (SP), and soil silt (SSG) were considered as environmental factors. RDA analysis was conducted in combination with the sample clustering results, and the findings are presented in Figure 4 . As depicted in Figure 4 , the central line of sample A1 formed an acute angle with the vectors representing soil silt content and organic matter content, while it formed an obtuse angle with those representing pH value, altitude, and soil sand content. These results indicate that the community structure at the sampling site of sample A1 was positively correlated with soil silt (SSG) and organic matter (ORM) content, and negatively correlated with pH, elevation (HT), and soil sand (SP). The patterns observed for samples A2, B1, and B2 were consistent with that of sample A1. In contrast, the patterns for samples A3 and B3 were opposite to those of the aforementioned samples.

By measuring the distance from the projection point of the A1 sample on the arrow line or its extension line of various environmental factors to the arrow, we can assess the influence of these environmental factors on the bacterial community structure of the A1 sample. The order of influence is as follows: pH > soil sand (SP) > altitude (HT) ≈ soil silt (SSG) ≈ organic matter content (ORM).

For the A2 sample, the impact on its bacterial community structure is ranked as: pH > soil sand (SP) > organic matter content (ORM) > altitude (HT) ≈ soil silt (SSG). The influence on the bacterial community structures of samples A3 and B1 is consistent with that of the A1 sample. Regarding the B2 sample, the effect on its microbial community structure is ranked as: pH > soil sand (SP) > soil silt (SSG) > altitude (HT) > organic matter content (ORM). For the B3 sample, the order of influence on its microbial community structure is: soil sand (SP) > pH > altitude (HT) > soil silt (SSG) ≈ organic matter content (ORM) (Jiang and Song, 2010).

Correlation analysis is a classical method to test the interaction between microorganisms. In this study, we aim to determine the significant correlation, strong correlation, positive correlation and negative correlation in the microbial community of six samples. In the correlation matrix diagram based on OTU, the size of the ellipse in the figure represented the absolute value of the correlation coefficient (the greater the absolute value of the correlation, the smaller the ellipse), the left oblique was positively correlated, the right oblique was negatively correlated, and the color changed with the right color scale. For analysis, we selected the top 100 most abundant OTU information and conducted bilateral tests. Only the results with p value less than 0.05 are shown in the figure. The results are shown in Figure 5 . Significant negative correlations were observed between the following pairs of OTUs: Otu33165 and Otu228, Otu472 and Otu14485, Otu472 and Otu33129, Otu462 and Otu2087, Otu18302 and Otu2087, Otu32039 and Otu28824. Negative correlations were detected between: Otu33148 and Otu33165, Otu2087 and Otu556, Otu2075 and Otu469, Otu2075 and Otu33148, Otu34911 and Otu33143, Otu447 and Otu33223. Significant positive correlations were found in the following pairings: Otu14485 and Otu33157, Otu519 and Otu34911, Otu2087 and Otu23005, Otu2087 and Otu34910, Otu33157 and Otu33165, Otu29445 and Otu31135. Positive correlations were identified between: Otu29452 and Otu30363, Otu14485 and Otu2126, Otu2111 and Otu2126, Otu33223 and Otu220, Otu2059 and Otu508, Otu2113 and Otu2067.

Z.nitidum is rich in a diverse array of active ingredients, such as alkaloids, flavonoids, volatile oils, and polysaccharides. These components endow Z. nitidum with a wide spectrum of pharmacological effects. Among the alkaloids present in Z. nitidum, nitidine chloride is particularly well - known. It exhibits anti - tumor, analgesic, anti - inflammatory, and cardiovascular protective properties, which are emblematic of the remarkable medicinal value of Z. nitidum.

Cluster analysis of heat map based on 6 samples ( Figure 3A ), a grey correlation analysis was conducted between the soil bacterial community abundance and nitidine chloride content. Fifteen bacterial taxa, including unclassified bacteria, Sphingomonas, Povallibacter, Rudaea, Bradyrhizobium, and Gemmatimonas, were selected as characteristic sequences, while nitidine chloride served as the parent sequence. A resolution coefficient of 0.50 was employed. The correlation values ranged from 0 to 1, with higher values indicating a stronger correlation with nitidine chloride. The data regarding the soil flora abundance and nitidine chloride content of Z. nitidum from different habitats are presented in Table 6 . The results of the correlation coefficients are shown in Table 7 , and the correlation degrees are presented in Table 8 . As can be observed from Table 7 , Table 8 , and Figure 6 , among the 15 soil bacterial taxa and nitidine chloride content, Rudaea had the highest comprehensive evaluation, with a correlation degree of 0.762. It was followed by Bradyrhizobium (correlation degree: 0.744), Gemmatimonas (correlation degree: 0.739), Sphingomonas (correlation degree: 0.690), and Erwinia (correlation degree: 0.668). In contrast, Lacibacterium had the lowest correlation degree, at 0.481.

Therefore, it can be inferred that in Z. nitidum soils from different habitats, a higher abundance of Rudaea, Bradyrhizobium, and Gemmatimonas may be associated with a higher nitidine chloride content in Z. nitidum.