Root-associated endophytic fungal community composition and structure of three medicinal licorices in Xinjiang

The purpose of this study was to explore the diversity and composition of endophytic fungal community in the root of three medicinal licorices, and learn more about its biological characteristics by analyzing its interaction with soil and root factors. A total of 2,118,633 effective sequences and 1,063 effective operational taxonomic units (OTUs) with 97% identity were obtained by high-throughput sequencing among 27 samples. In this study, a total of 8 phyla and 140 genera were annotated, among them, the phylum Ascomycota and Basidiomycota, and the genera Fusarium, Paraphoma and Helminthosporium were signicantly dominant in the 27 samples. Wilcoxon rank sum test showed that the Shannon index was signicantly different distribution between Glycyrrhiza uralensis and Glycyrrhiza inata, especially 0-20cm at the root depth, the Chao1 index in Glycyrrhiza inate was signicantly affected by root depth, and there were signicant differences in beta diversity between Glycyrrhiza uralensis and Glycyrrhiza inata. Distance-based redundancy analysis (db-RDA) showed that soil physicochemical properties (available potassium and ammonium nitrogen), and the root factor (liquiritin and water content) were the main contributing factors to the variations in the overall structure of endophytic fungal community in this study. This study provides useful information for formulating strategies to improve the quantity and quality of medicinal licorices. of three medicinal licorices at three root depths through high-throughput sequencing and explored their relationship with host plants’ bioactive compounds and soil physicochemical properties. The results will enhance researchers' understanding about the environmental and host factors that inuence endophytic fungi and the friendly relationship between endophytic fungi and medicinal plants, thus providing reference information for licorice growing for commercial medicinal purposes.


Sample collection
The roots and rhizosphere soils samples (soil depths were 0-20cm, 20-40cm and 40-60cm, respectively) of three medicinal Glycyrrhiza (Glycyrrhiza uralensis, Glycyrrhiza in ata and Glycyrrhiza glabra) were collected from August to September in 2019 from specimens growing at 3 distinct sites in 3 eco-regions in Xinjiang province, China; the geographical location of sampling points and soil physical and chemical properties are shown in Table 1. In addition, to ensure that the experiment was representative, we randomly selected three Glycyrrhiza plants in good growth condition from each geographical location according to the ve-point sampling method, and all samples were cut with sterile scissors. The roots of each plant were divided into three sections: upper (0-20cm), middle (20-40cm), and lower (40-60cm), and the roots of each section are equally divided into two parts: one part was placed into a ziplocked bag for the determination of the bioactive compounds in the root, while the other part was placed into a sterile bag and quickly transported on a piece of ice to the laboratory in preparation for the microbe determination. All the samples were labeled by combination with letters and numbers, with the rst letter representing the species (W, G and D: Glycyrrhiza uralensis, Glycyrrhiza glabra and Glycyrrhiza in ata, respectively), the second letter representing the root depth (1, 2 and 3: 0-20cm, 20-40cm, 40-60cm), and the third number representing the replicate number. For example, W.1.3 represents the third repetition of Glycyrrhiza uralensis at 0-20cm.

Surface sterilization
At the same time, to eliminate the interference of other microorganisms, the surface of roots was sterilized in the laboratory by rst rinsing soil from the roots under running water followed by washing with sterile distilled water. The roots were then soaked in 75% alcohol for 30 s for surface disinfection, and then washed ve times with sterile distilled water before soaking in 5% sodium hypochlorite for 5 min. Finally the roots were washed ve times with sterile distilled water and air-dried under sterile conditions [36]. To con rm that the surface sterilization process was successful, the last rinse solution was inoculated onto a potato dextrose agar (PDA) plate and cultured at 28 °C for 72 h. No fungi growth con rmed that the surface sterilization was successful [37]. All root samples were labeled and immediately placed on ice and then stored at liquid nitrogen prior to total DNA extraction.

Soil physicochemical
Soil samples from the rhizosphere were air-dried and sieved through a 2-mm mesh for soil physicochemical properties analysis. The following soil physicochemical characteristics were analyzed according to the methods described by the Bao et al [38]: The content of organic matter (SOM) was determined by external heating with potassium dichromate. Soil pH (1:2.5 = soil: distilled water) was determined using a pH meter. Soil Water content (SWC) was determined by weighing. The total nitrogen (STN) content was determined using the perchloric acid-sulfuric acid digestion method. The total phosphorus (STP) content was determined by acid digestion (molybdenum-antimony colorimetry). The total potassium (STK) content was determined by acid digestion (atomic absorption spectrometry). The total salt (TS) content was determined by atomic absorption spectrometry. Nitrate nitrogen (SNN) and ammonium nitrogen (SAN) contents were analyzed using 0.01 M calcium chloride extraction. The available phosphorus (SAP) content was determined by sodium bicarbonate extraction (molybdenum-antimony colorimetry). The available potassium (SAK) content was determined by ammonium acetate extract method (atomic absorption spectrometry).

Determination of active components
The root samples were dried at 60 °C for 72 h to constant weight (it has been con rmed that glycyrrhizic acid (GIA) and liquiritin (LI) do not decompose at this temperature [39]. The dried root samples were ground to a powder with a pestle and mortar and passed through a 60 mesh sieve. An aliquot (0.2 g) of powdered root sample was extracted with 71% chromatographic methanol in an ultrasonic bath (250 W, 40 kHz) at room temperature. The extract was then centrifuged at 12,000 rpm for 10 minutes and the supernatant was ltered (0.22-µm pore size) (Agilent, USA). The GIA and LI contents in the dried root samples (0.2 g) of the medicinal licorices were determined by high-performance liquid chromatography (HPLC, Agilent-1260 In nity, USA) using an Agilent ZORBAX SB-C18 column (150 mm × 4.6 mm, 5 µm) with mobile phase (chromatographic methanol: ultra-pure water: 36% glacial acetic acid = 71:28:1) and mobile phase (acetonitrile:0.5% glacial acetic acid = 1:4) respectively, and a gradient elution ow rate of 1.0 mL•min − 1 . GIA and LI were detected at 254 nm and 276 nm, respectively. The injection volume was 5 µL and the column temperature was 30 °C. The GIA and LI reference materials (CAS#1405-86-3 and CAS#551-15-5, respectively) were purchased from Solarbio and used for calibration purposes. The total avonoid content (GTF) in medicinal licorices was determined by ultraviolet spectrophotometry at 334 nm with the liquiritin standard (CAS#551-15-5) as the control.

DNA extraction and library construction
After immersion in liquid nitrogen, genomic DNA was extracted from the samples using the DNA Quick Plant System kit (Tiangen, China) according to the manufacturer's instructions. The purity and concentration of DNA were detected evaluated using a NanoDrop2000 (Thermo Fisher Scienti c, USA). According to the concentration, each DNA sample was diluted a nal concentration to 1 ng/µL with sterile distilled water for use as a DNA template.

Bioinformatics analysis and statistical analysis
Single-end reads was assigned to samples using Cutadapt [41] software based on their unique barcode and truncated by cutting off the barcode and primer sequence. To avoid the in uence of non-microbiota sequences (such as, chloroplast and mitochondrial sequences), the raw sequences were further ltered by Cutadapt software to remove non-microbiota taxa before subsequent analysis. Then raw tags were subjected to a strict quality controlled process using Cutadapt software to obtain high-quality clean reads. Clean reads were obtained by comparison with the reference database (Unite database) [42] using UCHIME algorithm to detect and remove chimeric sequences.
UPARSE software [43] (UPARSE v7.0.1001) was used to cluster the effective tags of all samples into the same operational taxonomic units (OTUs) with ≥ 97% identity, and taking the sequence with the highest frequency as the representative sequence of each OTU. The taxonomic information for each representative sequence was annotated using the Unite database, and multiple sequence alignment was performed using MUSCLE (Version 3.8.31) software to study the phylogenetic relationship of the representative sequences of OTUs among the 27 root samples. OTU abundance information was normalized using a standard sequence number corresponding to the sample with the least sequences (54,262 reads for sample D.2.1). Subsequent analysis of alpha diversity and beta diversity were performed based on this output normalized data. The raw sequence reads have been deposited in the NCBI Sequence Read Archive (SRA) with BioProject accession number PRJNA664554.
Using the FunGuild database based on the species information obtained from amplicon analysis, the ecological functions of existing species in the environment can be inquired.
Alpha Diversity analysis was used to study the complexity of species diversity in a sample through six indices (observed-species, Shannon, Simpson, Chao1, ACE, and Good-coverage) [44]. All indices in the samples were calculated with QIIME (Version 1.7.0) and displayed with R software (Version 2.15.3).
Beta diversity analysis was used to evaluate differences in sample species complexity, which based on weighted Unifrac was calculated by QIIME software.
The Un-weighted Pair-group Method with Arithmetic Mean (UPGMA) clustering analysis was conducted by QIIME software (Version 1.7.0). In addition, R software (Version 2.15.3) was also used to rarefaction curve generation, Wilcoxon rank sum test, Metastat statistical test, Spearman correlation analysis of heat maps and Distance-based Redundancy Analysis (db-RDA). Pearson correlation analysis was run among the bioactive compounds and the soil physicochemical properties. Two-way ANOVA was performed with SPSS 19.0 (IBM Inc., Armonk, USA), and displayed with GraphPad Prism 5.

Sequencing Results
In the root of the three medicinal Glycyrrhiza species (Glycyrrhiza uralensis, Glycyrrhiza glabra, and Glycyrrhiza in ata), a total of 2,118,633 effective sequences were obtained after ltering out low-quality and short sequence reads. The sequencing results for each sample are listed in Supplementary table S1. The effective sequences were clustered into OTUs with 97% identity, and a total of 1,063 OTUs were obtained, among them, 91.53% of the effective sequences were assigned to the Kingdom level, 59.27% to the phylum level, 54.37% to the class level, 53.72% to the order level, 46.19% to the family level, 38.01% to the genus level, and 23.52% to the species level by the Illumina HiSeq (Fig. 1a). The rarefaction curves showed that the number of OTU in each sample increased gradually with quantity of sequence, thus con rming that the amount of sequencing data was adequate (Fig. 1b).

Composition of fungal community in the root of medicinal liquorices
According to the OTUs sequence and Unite database, 8 phyla, 23 classes, 53 orders, 102 families, 140 genera and 141 species were annotated. The endophytic fungal phyla with the greatest abundance from nine groups were enumerated in Fig. 2a D3, G1, G2, G3, W1, W2 and W3, respectively. In addition, Basidiomycota occupied a large part of the relative abundance in D2 (21.348%), D3 (28.440%), G2 (10.631%), G3 (12.523%), W2 (6.749%) and W3 (5.110%), respectively. Meanwhile, our results showed that the relative abundance of Ascomycota gradually decreased with the downward movement of root depths (Fig. 2b). For the difference analysis at the Phylum classi cation level, a MetaStat statistical test based on species abundance was constructed, and the results showed that the relative abundance of Ascomycota in Glycyrrhiza in ata signi cant difference distribution at different root depth (Fig. 2b). Speci cally speaking, the relative abundance of Ascomycota at D1 sample (91.821%) was signi cantly higher than D3 sample (39.956%) (Fig. 2b).
In terms of genus, we listed the top 10 dominant fungal genera in each group in Fig. 2c Based on the ITS amplicons analysis, we obtained the classi cation and abundance information of endophytic fungal community in root of medicinal licorices; we also pay attention to what role these species play in the ecosystem. The top 25 main ecological function of fungal species based on FunGuild analysis was shown in Fig. 3. Plant_ Pathogen-Soil_ Saprotroph-Wood_ Saprotroph (32.072%) was found to be the most dominant in G2 sample; Fungal_ Parasite-Plant_ Pathogen (26.567%) was found to be the most dominant in D1 sample; the abundance of Unde ned_ Saprotroph was high in the G1, G2 and G3 samples, accounting for 21.271%, 28.330% and 35.555%, respectively. Ectomycorrhizal (21.187%) and Endophyte (13.208%) were found to be the most dominant in D2 sample. However, a much higher proportion of unassigned ecological function existed in all groups.

Effects of root depth and plant species on alpha diversity in the fungal community of roots
The alpha diversity index of each group was shown in Table 2. Some indexes (Shannon and Chao1) respectively re ected the diversity and richness of microbial communities in samples, the greater the index, the higher the species diversity, the richer the distribution. The Shannon index of the W1 (4.910) sample was the highest. In contrast, that of the D1 (3.393) sample was the lowest. Moreover, we found that D1 had the lowest Chao1 (238.678) and ACE (253.105), while the D3 sample had the highest Chao1 (356.317) and ACE (355.694), respectively. Meanwhile, the results based on Wilcoxon rank sum test showed that the Shannon index was signi cantly different distribution between Glycyrrhiza uralensis and Glycyrrhiza in ata, especially 0-20cm at the root depth (Fig. 4a). Speci cally, the Shannon index in W1 sample was signi cantly higher than D1 sample (p < 0.05). Furthermore, the Chao1 index in Glycyrrhiza in ata increased gradually with the downward movement of root depths, and based on Wilcoxon rank sum test showed that the Chao1 index in Glycyrrhiza in ate was signi cantly affected by root depth (Fig. 4b). Speci cally, the Chao1 index in D3 sample was signi cantly higher than D1 sample (p < 0.01); D2 sample was signi cantly higher than D1 sample (p < 0.05).

Effects of root depth and plant species on beta diversity in the fungal community of roots
Beta diversity analysis was used to evaluate differences in microbial community composition among the samples. The Unweighted Pair-Group Method with Arithmetic (UPGMA) cluster analysis was performed to study similarity in the composition of endophytic fungal community among different samples, and the clustering results were integrated with species relative abundance at phyla taxon level in each group. As shown in Fig. 5a, the results of UPGMA cluster tree based on Weighted Unifrac distances showed that samples from G3, G2, G1 and W1 were clustered together, and samples from W3, D2, W2 and D1 were clustered together (Fig. 5a). Meanwhile, for the difference analysis between the beta diversity, a Wilcoxon rank sum test based on Weighted Unifrac distances was constructed (Fig. 5b), and the results showed that there were signi cant differences in beta diversity between Glycyrrhiza uralensis and Glycyrrhiza in ata, which was consistent with UPGMA cluster tree. Speci cally, there were signi cant differences in beta diversity between D1 and D2 samples (P < 0.05), D3 and W3 samples (P < 0.05), and D1 and W1 samples (P < 0.01) (Fig. 5b), which indicated there were signi cant differences in endophytic fungal community composition in roots of medicinal licorices between different species and different root depth.

The relationship between endophytic fungal communities and the bioactive compounds and soil physicochemical properties
The results of two-way ANOVA showed that the content of the bioactive compounds (glycyrrhizic acid (GIA), liquiritin (LI) and total avonoid (GTF)) were not signi cantly affected by the interaction effect between root depth (0-20cm, 20-40cm, 40-60cm) and plant species (Glycyrrhiza uralensis, Glycyrrhiza in ata, and Glycyrrhiza glabra) (P > 0.05) (Table 3). However, the content of LI was signi cantly affected by the main effect plant species (P < 0.05) ( Table 3 and Fig. 6). As shown in the Fig. 6, the contents of LI in root of Glycyrrhiza uralensis were signi cantly higher than those in Glycyrrhiza in ate (P < 0.05), and the contents of LI in root of Glycyrrhiza uralensis were signi cantly higher than those in Glycyrrhiza glabra (P < 0.05) (Fig. 6a). In addition, Pearson correlation analysis showed that the content of bioactive compounds was signi cantly correlated with soil physicochemical properties (Supplementary table S3). GIA content in root had a very signi cant positive correlation with available potassium (SAK) and soil water content (SWC) (r > 0; P < 0.05), but LI content in root had a very signi cant negative correlation with SAK and total salt (TS) content (r < 0; P < 0.05).
Furthermore, spearman correlation analysis showed that the content of LI was signi cantly positive correlated with alpha diversity index (r > 0, P < 0.05) (Fig. 7). As shown in Fig. 7, the content of LI had a very signi cant positive correlation with Shannon index, Simpson index and Chao1 index (P < 0.05), which indicated that the content of LI was accountable for the diversity of endophytic fungal community in this study.
As shown in Fig. 8, there was a signi cant relationship between the dominant fungi genus and bioactive compounds and soil physicochemical properties.
Speci cally, Fusarium showed a signi cant positive correlation with LI content (P < 0.05); Paraphoma showed a signi cant positive correlation with SAN (P < 0.05), but a signi cant negative correlation with SAK, TS and SWC (P < 0.05); Helminthosporium showed a signi cant positive correlation with PH (P < 0.05); Sarocladium showed a signi cant negative correlation with SOM, STN and SNN (P < 0.05); Conocybe showed a signi cant positive correlation with SWC, but a signi cant negative correlation with SAN (P < 0.05).
Distance-based redundancy analysis (db-RDA) based on the Bray-Curtis distance showed that the bioactive compounds and soil physicochemical had signi cant effects on the differences of endophytic fungal community (Fig. 9). The differential distribution of endophytic fungal community was mainly restricted in the rst and second ordination axes, and the rst ordination axis, the second ordination axis were explained 16.23%, 13.89% of the total variability, respectively (Fig. 9). Speci cally, among the soil environment factors, SAK content was identi ed as the factor that most signi cantly affects the differences of endophytic fungal community (r 2 = 0.329, P < 0.01), followed by SAN (P < 0.05). Among the root factors, the RWC was explained the difference of endophytic fungal communities in roots to the greatest extent (r 2 = 0.247, P < 0.05), followed by LI content (P < 0.05) (Fig. 9, Table 5). According to results of the db-RDA analysis, the SAN, SAK, RWC, and LI content were the major factors contributing to the variations in the overall structure of endophytic fungal community in this study.

Discussion
In this study, we investigated the composition and diversity of endophytic fungal communities in different root depth (0-20cm, 20-40cm and 40-60cm) of three medicinal licorices (Glycyrrhiza uralensis, Glycyrrhiza glabra, and Glycyrrhiza in ata) using high-throughput sequencing technology, which provides a large amount of data with more accuracy than that obtained in previous studies using traditional technology [45][46][47]. We obtained the composition of endophytic fungal communities at different taxonomic levels (phylum, class, older, family, genus and species) by high-throughput sequencing (Fig. 2a, Fig. 2c and Supplementary table S2). The results showed that there was a speci c microbiome in 27 samples of tree medicinal licorices, and the relative abundance of endophytic fungi was correlated with the host plant species and root depth. For example, Ascomycota was the dominant phylum in all samples, followed by Basidiomycota, which result consistent with previous studies [48,49]. The phylum Ascomycota, as the largest phylum of fungi, has diverse populations and plays an important role in genetics [50], ecology [51] and phylogeny [52]. Such as, the Ascomycota produce large numbers of spores through both asexual and sexual reproduction. Asci can act as small water cannon, spraying spores into the air. Dispersal process of ascospores, spores is important for dissemination of many fungal plant diseases and for the dispersal of many saprophytic fungi [53].
Moreover, our results showed that the relative abundance of Ascomycota gradually decreased with the downward movement of root depths (Fig. 2b), which was consistent with the results of Ko, Daegeun et al [54]. On this basis, we found that the relative abundance of Ascomycota in Glycyrrhiza in ata had a signi cant difference at different root depth, but Glycyrrhiza uralensis and Glycyrrhiza glabra were not signi cant difference, indicating that some endophytes may preferentially proliferate in a certain ecological region and play different ecological roles from other endophytes. Overall, in addition to soil depth, the relative abundance of endophytes was also related to the genotype of the host plant species. This was consistent with the results of host genotype and soil conditions on ectomycorrhizal community of poplar clones by Karliński, Leszek et al. [55].
Alpha Diversity and Beta Diversity analysis of endophytic fungal community showed signi cant differences in root depths (0-20cm, 20-40cm and 40-60cm) between Glycyrrhiza uralensis and Glycyrrhiza in ata (Fig. 4, Fig. 5), which indicated that both genotype and ecological region of host plants contributed to the differences of endophytic fungal community. Meanwhile, numerous studies [56] have shown that the adaptation of endophytic fungal community largely depends on the adaptation of host plants to the ecological environment, which indicated that host plants largely determine the colonization and distribution of endophytic fungal community. The relationship between fungus and host plant were also often considered as a exible interaction, with orientations determined by subtle differences in the expression of fungal genes in response to the host, or conversely, by the host's recognition and response to the fungus.
Thus, slight genetic differences in the two genomes control the symbiosis [57].
Furthermore, our results showed that the root depth had a signi cant effect on the richness and composition of endophytic fungal community (Fig. 4b, Fig. 5a and Fig. 5b), which indicated that different ecological types of endophytic fungi may represent certain ecological regions (different root depth), these should become an important consideration factor that endophytic fungi inoculation. We speculated that this is related to root respiration and soil C content. On the one hand, root respiration, accounts for 60% of total soil respiration, can regulates the metabolism of roots and related microorganisms, and is an important part of terrestrial carbon budget [58]; on the other hand, the content of C in unstable soil varies greatly between different soil depths [59]. Moreover, Noah Fierer et al. [60] demonstrated that the vertical distribution of the speci c microbial species was largely related to the decrease in carbon availability with soil depth.
However, one weakness in this study was that the samples of three glycyrrhiza species were collected from areas which differed in geographical environment.
Since it is rare to nd three glycyrrhiza species in the same habitat, to a certain extent, the soil physicochemical properties can represent the environmental factors in which three glycyrrhiza species were growing. Therefore, in this study, in addition to root factor, also included the effects of the soil factors.
Numerous studies [61,62] showed the accumulation of bioactive compounds in medicinal licorice roots is affected by various factors. In this study, the content of LI were more affected by main effect plant species than main effect root depth (Table 3), among them, the content of LI in root of Glycyrrhiza uralensis were signi cantly higher than those in Glycyrrhiza in ata and Glycyrrhiza glabra (Fig. 6a), which is consistent with the results of Zhang et al. [63]. We speculate that this is related to the expressions of some functional genes that are closely associated with the content of bioactive compounds including glycyrrhizic acid and liquiritin in root of licorice species. Some studies [64][65][66] have shown that key functional genes, such as chalcone synthase gene, 3-Hydroxy-3-methylglutary CoA reductase (HMGR) and squalene synthase (SQS), are involved in transcriptional level regulation process in glycyrrhizic acid and liquiritin biosynthesis. Although further studies are required to characterize the expression of functional genes of bioactive compounds, this study provides a theoretical basis for the development of strategies to expand the Glycyrrhiza uralensis cultivation. On the other hand, the content of bioactive compounds is the result of the interaction between plants and their growing environment, therefore, the accumulation of bioactive compounds in root is in uenced by the ecological environment of its. In this study, GIA, GTF and LI content had a positive correlation with soil total nitrogen (STN) (r > 0), indicating that soil nutrients can promote the accumulation of bioactive compounds, but not all soil nutrients, such as soil total potassium (STK), have such a function. Although potassium can be involved in many enzyme activation systems in plants and improve plant stress resistance [67], the content of GIA, GTF and LI were negatively correlated with STK (r < 0) in this study, which is consistent with the results of Liu et al [68]. In addition, soil available potassium (SAK) had a signi cant positive correlation with GIA, but had a signi cant negative correlation with LI (Supplementary table S3), indicating that the utilization mechanism of soil nutrients by bioactive compounds is completely different. Although the mechanism by which available potassium regulate bioactive compounds is still unclear, this discovery may form the basis of further in-depth research. In general, these soil factors exhibit habitat speci c characteristics are related to the regulation of bioactive compounds in root of licorice.
In recent years, a growing number of studies [69][70][71] have demonstrated that the dynamics of the micro ora is driven to a large extent by environmental factors including soil characteristics (pH, nitrogen, phosphorus and potassium) and climate condition (rainfall and temperature). Consistent with these reports, our results showed that LI, RWC, SAN and SAK content were the major factors contributing to the variations in the overall structure of endophytic fungal community ( Fig. 9 and Table 5). In addition, we found that the content of LI in root had a very signi cant positive correlation with diversity of endophytic fungal community (Shannon and Simpson index) (P < 0.05) (Fig. 7). Liquiritin (LI), the main bioactive compounds of avonoids, is one of the material basis for clinical e cacy and an important index of the quality of medicinal licorices. Flavonoids can be speci cally induced by symbiotic fungus to respond to puri ed signaling molecules from these organisms when the fungus colonizes. Chen et al. [72] demonstrated that with inoculation of fungi Glomus mosseae, Glycyrrhiza uralensis plants signi cantly increased stem and root biomass and liquiritin content in the main root.
Meanwhile, our results showed that soil physicochemical and bioactive compounds had a signi cant effect on composition of endophytic fungal communities (such as phylum and genus) ( Fig. 8 and Table 4), which showed that there is an interaction among endophytic fungal community, root and soil factor. This suggests that we may be able to alter the fungal composition by altering soil factors [73], thereby promoting the accumulation of bioactive compounds in plants [74]. In the case of medicinal licorices, Wei Xie et al. [75] shown that P addition and arbuscular mycorrhizal (AM) inoculation could improve plant growth and facilitated glycyrrhizic acid and liquiritin accumulation in Glycyrrhiza uralensis. Meanwhile, Y. Orujei et al. [76] also shown that two species of arbuscular mycorrhizal fungi (AMF) were successful inoculation, the increase in the growth rate and the accumulation of bioactive compounds in licorice roots ( Glycyrrhiza glabra) were observed compared to control. In general, this study provided useful an information for the development of strategies to improve the production and quality of medicinal licorices, although further studies are required to characterize the functions of these endophytic fungi.

Conclusions
In this study, numerous endophytic fungal communities were detected in roots of medicinal licorices based on high-throughput sequencing. Furthermore, we identi ed signi cant differences in the relative abundance of Ascomycota among root depth. Furthermore, the alpha diversity analysis and beta diversity analysis showed that the endophytic fungal community structure and composition differed among the species and root depth in medicinal licorices. Moreover, the SAN, SAK, RWC, and LI content were the major factors contributing to the variations in the overall structure of endophytic fungal community in this study.
This study clari ed the ecological role of non-biological factor (soil and root) in the endophytic fungal community of medicinal licorices, which may provide theoretical basis for the synthesis of bioactive compounds and rational utilization of medicinal plants in production practice. Availability of data and material All data generated or analysed during this study are included in this published article and its supplementary information les.