The results of one-way ANOVA showed that the contents of the effective components (glycyrrhizic acid (GIA), liquiritin (LI) and total flavonoid (GTF)) in the roots of medicinal licorices (Glycyrrhiza uralensis, Glycyrrhiza inflata, and Glycyrrhiza glabra) were not affected by the variety, but were significantly affected by the growing year (1–3) (P < 0.05), with a trend of stable increase in the contents observed with each growing year (Table 1). Specifically, the contents of GIA, LI and GTF in the year 3 were significantly higher than those in year 1, the contents of GIA and GTF in year 2 were significantly higher than those in year 1, and the contents of GIA and LI in year 3 were significantly higher than those in year 2.
In addition, Spearman correlation analysis showed that the content of effective components was significantly correlated with soil physicochemical properties and nutritional components of leaves (Table 2). In terms of soil physicochemical factors, GIA and LI had a very significant positive correlation with soil ammonium nitrogen (SAN) (R2 > 0; P < 0.01), but had a very significant negative correlation with total salt (TS) (R2 < 0; P < 0.01); GTF had a very significant positive correlation with SAN (R2 > 0; P < 0.01), but had a very significant negative correlation with nitrate nitrogen (SNN) (R2 < 0; P < 0.01), and had a significant negative correlation with TS (R2 < 0.01; P < 0.05). In terms of leaf nutrition, GIA, LI and GTF were very negatively correlated with water content (PWC), total potassium (PTK) and crude fiber (CF). In addition, GTF also showed a very significant very negative correlation with total phosphorus (PTP). There was a significant relationship between leaf nutrients and a small number of variables. Specifically, PWC was positively correlated with SNN, but negatively correlated with SAN; organic carbon (POC) was positively correlated with soil total potassium (STK); total nitrogen (PTN) was positively correlated with POC; total phosphorus (PTP) was very positively correlated with SNN, TS and PTN; PTK was very positively correlated with SNN, TS, PWC, PTN and PTP, but negatively correlated with SAN; CF was very positively correlated with PWC and PTK.
1.2 Composition of bacterial community in the root of medicinal liquorices
In 27 samples, 73,316 effective sequences with an average length of 372 nucleotides (nt) were obtained after quality control. The sequencing results for each sample are listed in Supplementary table S1. The effective sequences were clustered into operational taxonomic units (OTUs) with 97% identity, and a total of 2,432 OTUs were obtained. The rarefaction curves different colors represent different samples, showed that the number of OTU in each sample increased gradually with quantity of sequence, thus confirming that the amount of sequencing data is adequate (Fig. 1). According to the OTUs, 28 phyla, 45 classes, 106 orders, 216 families, 508 genera and 313 species were annotated. Figure 2a shows the 10 bacterial phyla with the greatest abundance in the bacterial community in the roots of medicinal licorices. Proteobacteria (76.28%) dominated the observed sequences at the phylum level, and Actinobacteria (10.62%), Bacteroidetes (6.92%) and Firmicutes (3.42%) were found to be the predominant phyla in each group, this was followed by unidentified_bacteria (0.80%), Chlamydiae (0.42%), Tenericutes (0.30%), Acidobacteria (0.21%), Verrucomicrobia (0.20%), Planctomycetes (0.13%) and others (0.70%).
In terms of genus (Fig. 2b), unified-Rhizobiaceae (17.62%) was more abundant than other genera, and Pseudomonas (7.49%), Novosphingobium (4.52%), Pantoea (2.23%), Bacteroides (1.43%), and Halomonas (1.14%) were found to be the main genera in all groups, followed by Nocardiopsis (0.88%), Sinomicrobium (0.55%), unidentified-bacteria (0.77%), Methylophilus (0.90%) and others (62.46%). Details of the composition information of top 10 dominant bacteria at other classification levels are listed in Supplementary table S2. Specifically speaking, Alphaproteobacteria (46.20%), Gammaproteobacteria (28.54%), unidentified-Actinobacteria (9.86%) dominate at the class taxonomic level; the dominant species at the order taxonomic level are Rhizobiales (26.63%), Sphingomonadales (10.90%), Gammaproteobacteria (8.58%); the dominant species at the family taxonomic level are Rhizobiaceae (19.16%), Pseudomonadaceae (7.49%), Sphingomonadaceae (10.90%); the dominant species at the species taxonomic level are Pantoea-brenneri (1.97%), Neorhizobium-huautlense (2.73%), and Pseudomonas-psychrotolerans (1.08%).
1.3 Effects of years and varieties on alpha diversity and beta diversity in the bacterial community of roots.
The alpha diversity index of each group is shown in Supplementary table S3. One-way ANOVA showed that there was no significant difference between the alpha diversity indexes (Shannon, Simpson, Chao1, ACE) among the groups, reflecting the diversity and richness of bacterial communities. The results showed that the planting year and varieties did not affect the diversity and richness of endophytic bacterial communities in the root of medicinal licorices (Table 3). However, beta diversity analysis showed significant differences in the endophytic bacterial community structures among the different groups. Specifically, we used the weighted-UniFrac and unweighted-UniFrac distances to construct a beta diversity distance heat map to measure the coefficient of the difference between the two groups (Fig. 3). The distances among the nine groups based on weighted-UniFrac ranged from 0.2 to 0.674, which indicated the existence of differences in endophytic bacterial community of roots of medicinal licorices between different varieties and different planting years. More specifically, the samples collected from E.D showed a high degree of difference compared with other sampling groups: 0.481 in E.W, 0.570 in E.G, 0.631 in R.D, and 0.526 in S.D.
Table 3
Endophytic bacteria community richness estimators and diversity index in licorice root with different growth year and variety.
| | Shannon | Simpson | Chao1 | ACE |
year | E | 5.8688 ± 1.3289a | 0.9347 ± 0.0611a | 626.172 ± 152.8335a | 623.8419 ± 148.8065a |
R | 6.2889 ± 0.7270a | 0.9572 ± 0.0349a | 689.9542 ± 100.8890a | 690.1123 ± 100.0530a |
S | 5.4513 ± 1.2254a | 0.9132 ± 0.0687a | 633.2774 ± 159.4251a | 637.8597 ± 145.1567a |
variety | W | 6.0936 ± 1.0528a | 0.9424 ± 0.0642a | 634.6612 ± 165.1966a | 632.2988 ± 154.0842a |
D | 5.2879 ± 1.1309a | 0.9146 ± 0.0643a | 611.0410 ± 106.3678a | 620.3057 ± 92.8245a |
G | 6.2276 ± 1.1055a | 0.9481 ± 0.0421a | 703.7014 ± 133.7225a | 699.2094 ± 139.7966a |
Description: the mean is followed by the standard deviation. Different letters indicated statistically significant difference (P < 0.05) (E, R and S: years 1, 2, and 3, respectively; W, G and D: Glycyrrhiza uralensis Fisch., Glycyrrhiza glabra L.and Glycyrrhiza inflata Batal., respectively). |
We used linear discriminant analysis (LDA) effect size (LEfSe) to identify discriminative taxa among different varieties of medicinal licorices and different planting years. As shown in Fig. 4, the results of LEfSe analysis of all years and varieties based on the rank sum test revealed a total of 16 biomarkers with significant differences that were contained in the E.D (5 taxa), E.W (7 taxa), R.G (2 taxa) and R.W (2 taxa) groups. No discriminative taxa were observed in year 3. These biomarkers included Variovorax, Nocardiopsis, Methylophaga, Pelagibacterium, Halomonas and Sinomicrobium at the genus taxonomic level.
1.4 The relationship between the dominant phylum and genus of endophytic bacteria and the effective components, soil physicochemical properties and leaf nutrition.
Spearman correlation analysis showed that there was a significant relationship between dominant bacteria phylum and effective ingredients, soil physicochemical properties and leaf nutrition (Table 4). Specifically, Proteobacteria showed a very significant negative correlation with PTN; Actinobacteria showed a significant negative correlation with SAN, GIA and LI (R2 < 0; P < 0.05); Firmicutes showed a significant positive correlation with PTN and CF; and Acidobacteria showed a significant positive correlation with POC (R2 > 0; P < 0.05).
Table 4
Spearman correlation analyses testing the relationship between the relative abundance of dominant bacterial phyla in the root of licorice and effective ingredients, soil physicochemical properties and leaf nutrition.
| Proteobacteria | Actinobacteria | Bacteroidetes | Firmicutes | Unidentified-Bacteria | Chlamydiae | Tenericutes | Acidobacteria | Verrucomicrobia | Planctomycetes |
SOM | 0.098 | -0.219 | -0.096 | -0.190 | -0.253 | -0.167 | -0.305 | -0.246 | 0.092 | -0.332 |
STN | 0.242 | -0.358 | -0.004 | -0.178 | -0.236 | -0.207 | -0.131 | -0.252 | -0.020 | -0.155 |
STP | -0.024 | -0.236 | 0.209 | 0.099 | -0.129 | 0.125 | 0.165 | -0.319 | -0.070 | -0.033 |
STK | -0.220 | 0.087 | 0.216 | 0.091 | -0.127 | 0.060 | -0.225 | 0.111 | 0.256 | -0.116 |
SNN | 0.075 | 0.154 | -0.137 | -0.005 | -0.128 | -0.241 | 0.292 | 0.128 | -0.355 | -0.026 |
SAN | 0.073 | -0.455* | 0.250 | 0.138 | 0.223 | 0.107 | 0.067 | -0.068 | 0.208 | -0.222 |
TS | 0.034 | 0.274 | -0.228 | 0.045 | -0.195 | -0.276 | 0.311 | 0.214 | -0.289 | 0.211 |
PWC | -0.088 | 0.111 | 0.020 | 0.173 | -0.270 | -0.094 | 0.047 | 0.038 | -0.204 | -0.046 |
POC | -0.178 | 0.016 | 0.181 | 0.354 | -0.290 | -0.225 | 0.074 | 0.434* | 0.077 | 0.151 |
PTN | -0.499** | 0.291 | 0.187 | 0.430* | 0.236 | 0.008 | 0.339 | 0.263 | 0.066 | 0.184 |
PTP | 0.022 | 0.122 | -0.134 | 0.090 | -0.162 | -0.177 | 0.238 | 0.130 | -0.199 | 0.188 |
PTK | -0.202 | 0.218 | -0.010 | 0.267 | -0.339 | -0.172 | 0.279 | 0.117 | -0.186 | 0.214 |
CF | -0.330 | 0.206 | 0.095 | 0.408* | -0.054 | 0.170 | 0.343 | 0.121 | -0.022 | 0.212 |
GlA | 0.358 | -0.463* | 0.049 | -0.274 | 0.185 | 0.040 | -0.097 | -0.204 | -0.012 | -0.100 |
GTF | 0.168 | -0.354 | 0.031 | -0.241 | 0.248 | 0.083 | -0.038 | -0.165 | 0.221 | -0.070 |
LI | 0.331 | -0.458* | 0.079 | -0.242 | 0.189 | 0.074 | -0.111 | -0.218 | 0.037 | -0.109 |
Description: the values are the correlation coefficients. ** means P < 0.01; * means P < 0.05. |
As shown in Fig. 5, there was a significant relationship between the dominant bacteria genus and the effective components of roots, soil physicochemical factors and leaf nutrition. Specifically, unidentified-Rhizobiaceae had a significant positive correlation with SAN, GIA, GTF and LI, but a very significant negative correlation with CF; Pantoea had a significant positive correlation with PTK; Pseudomonas had a significant positive correlation with PTP; Halomonas had a significant positive correlation with SNN, TS, PWC, PTK and CF, but a very significant negative correlation with SAN, GIA, GTF and LI; Nocardiopsis had a significant positive correlation with TS, PWC, PTP, PTK and CF, but a very significant negative correlation with SAN, GIA, GTF and LI; Novosphingobium had a significant negative correlation with PWC, but a very significant positive correlation with GTF; and Sinomicrobium had significant positive correlation with SNN, PWC, PTK and CF, but a significant negative correlation with GIA, GTF and LI. It is worth noting that we found that the biomarkers with significant differences at the genus taxonomic level of the dominant bacterial community (including Nocardiopsis, Halomonas and Sinomicrobium) had a significant negative correlation with GIA, GTF, and LI (R2 < 0; P < 0.05). Details of the interrelationships between the biomarkers with significant differences in the class, orders, families and genera of dominant bacteria and effective components, soil physicochemistry and leaf nutrients are shown in Supplementary table S4. These results showed that the biomarkers with significant differences (except Methylophaga) at each classification level were significantly negatively correlated with GIA, GTF, and LI, while all showed a significant positive correlation with PTK.
1.5 Total flavonoids explained the difference in composition and distribution of endophytic bacteria community in the roots of planting licorices to the greatest extent.
Distance-based redundancy analysis (db-RDA) based on the Bray–Curtis distance showed that the effective components, soil physicochemical and leaf nutrient components had significant effects on the endophytic bacterial community in the roots of medicinal licorices (Fig. 6, Table 5). Specifically, the contents of GIA, GTF, and LI all had a significant effect on the endophytic bacterial community in the roots (P < 0.05). Among them, the content of GTF explained the difference in the composition and distribution of endophytic bacterial communities in the roots of cultivated medicinal liquorices to the greatest extent (R2 = 0.638, P < 0.01). Among the soil environment factors, TS of soil was identified as the factor that most significantly affects the endophytic bacterial community, followed by SAN and SNN. Among the leaf nutrients factors, PWC was identified as the factor that most significantly affects the endophytic bacterial composition, followed by PTK and PTP. In addition, Mantel tests revealed that the combination of medicinal components is the environmental factor with the greatest correlation with the endophytic bacterial community, indicating that the combination has the greatest impact on the microbial community (r = 0.260; P = 0.008) (Supplementary table S5).
Table 5
Results for db-RDA testing effects of soil physicochemical properties, leaf nutrients and medicinal components on the composition and distribution of bacterial community in licorice root.
| r2 | P Value |
SOM | 0.118 | 0.213 |
STN | 0.163 | 0.113 |
STP | 0.165 | 0.110 |
STK | 0.035 | 0.666 |
SNN | 0.296 | 0.010 |
SAN | 0.428 | 0.002 |
TS | 0.497 | 0.000 |
PWC | 0.498 | 0.001 |
POC | 0.014 | 0.849 |
PTN | 0.044 | 0.607 |
PTP | 0.260 | 0.028 |
PTK | 0.442 | 0.001 |
CF | 0.247 | 0.033 |
GlA | 0.555 | 0.000 |
GTF | 0.638 | 0.000 |
LI | 0.393 | 0.005 |
Description: r2 is the determinant coefficients of the distribution of the bacterial community by environmental factors. |