P. notoginseng yields in different sites
P. notoginseng yields were significantly diverse in different sites in the range of 0.10–1.35 kg m− 2 (Fig. 1). P. notoginseng yields were 0.10, 0.25, 0.68, 0.76 and 1.35 kg m− 2 in Pingba A (PBA), Pingba B (PBB), Pingba C (PBC), Yanshan (YS) and Qiubei (QB), respectively. The yield was markedly higher in QB than those of PBA, PBB, PBC and YS (P < 0.001).
Diversity of P. notoginseng rhizosphere microbiome
A total of 1,104,569,780 paired-end clean reads were obtained, and approximately 62.9–86.6 million clean reads were obtained per sample (Additional file 1: Table S1). A total of 6,651,849 contigs were generated, with the longest contig at 611,672 bp and N50 at 1083 bp (Additional file 1: Table S2). After removing redundant sequences (identity > 95% and coverage > 90%), 6,228,225 unigenes with an average length of 546.43 bp were generated. The alpha diversity (Chao1 and Shannon index) of the rhizosphere microbiome showed the difference in the sample sites (Table 1). Shannon index at phylum, class, order and family levels were higher in the soils of YS and QB than that in the soils of PBA, PBB and PBC.
Table 1
The alpha diversity (chao1 and Shannon index) of the microbial communities revealed by metagenome data
Sample sites | Phylum | Class | Order | Family | Genus |
Chao1 |
PBA | 174.01 ± 3.45ab | 291.92 ± 3.76a | 495.19 ± 2.66a | 901.77 ± 1.59a | 2654.37 ± 9.43a |
PBB | 159.75 ± 3.72b | 283.89 ± 9.67a | 498.07 ± 12.24a | 883.88 ± 15.11a | 2640.47 ± 49.43a |
PBC | 166.37 ± 4.47ab | 278.42 ± 8.48a | 485.99 ± 10.88a | 887.60 ± 28.37a | 2662.11 ± 54.94a |
YS | 172.20 ± 6.72ab | 290.10 ± 3.85a | 506.66 ± 6.37a | 922.45 ± 16.83a | 2680.45 ± 18.87a |
QB | 176.32 ± 3.30a | 293.71 ± 5.05a | 510.46 ± 17.58a | 938.83 ± 25.37a | 2660.01 ± 6.08a |
Shannon index |
PBA | 1.99 ± 0.02b | 3.40 ± 0.03b | 4.85 ± 0.03c | 5.91 ± 0.02b | 7.27 ± 0.03a |
PBB | 1.99 ± 0.02b | 3.34 ± 0.07b | 4.76 ± 0.07c | 5.86 ± 0.05b | 7.30 ± 0.04a |
PBC | 2.01 ± 0.12b | 3.34 ± 0.15b | 4.77 ± 0.10c | 5.85 ± 0.09b | 7.24 ± 0.06a |
YS | 2.63 ± 0.03a | 3.97 ± 0.04a | 5.38 ± 0.03a | 6.25 ± 0.03a | 7.35 ± 0.03a |
QB | 2.39 ± 0.10a | 3.70 ± 0.08a | 5.14 ± 0.09b | 6.05 ± 0.12ab | 7.18 ± 0.18a |
PBA, Pingba village A; PBB, Pingba village B; PBC, Pingba village C; YS, Yanshan village; QB, Qiubei village. The mean values of three replicates per site are show, followed by the standard error of the mean. Different letters represent significant difference among five sample sites at the level of 0.05. |
Composition of P. notoginseng rhizosphere microbiome
PCoA was performed based on metagenomic sequencing using the Bray–Curtis metric to visualise the difference in microbial communities among soil samples, thereby revealing significant difference in the microbial community in the rhizosphere soils using adonis test analysis (R2 = 0.42, P = 0.001) (Fig. 2a). The first principal component axis (18.7% contributions) demonstrated that the microbial communities in the soils of YS and QB differed from those of the soils of PBA, PBB and PBC; the second principal component (13.61% contributions) suggested that the microbial communities in the soils of YS significantly differed from those of other sites.
Bacteria is the predominant domain (97.32% ± 0.90%, mean relative abundance ± SD, n = 15), with small proportions of Eukaryota, Archaea and viruses detected based on the annotated unigenes. Proteobacteria, Actinobacteria, Acidobacteria, Chloroflexi, Bacteroidetes, Gemmatimonadetes and Verrucomicrobia were present at high relative abundances (> 1.0%) in the rhizosphere microbiome of sample sites at the phylum level (Fig. 2b). The relative abundance of Proteobacteria was 67.29%, 67.34%, 66.42%, 48.05% and 54.57%, respectively, in rhizosphere soils of PBA, PBB, PBC, YS and QB. The relative abundance of Bacteroidetes was 3.73%, 5.68%, 2.95%, 2.30% and 1.95%, respectively, in rhizosphere soils of PBA, PBB, PBC, YS and QB. The relative abundance of Proteobacteria and Bacteroidetes were significantly higher in rhizosphere soils of PBA, PBB and PBC than YS and QB (P < 0.01). The relative abundance of Actinobactera was significantly higher in rhizosphere soils of YS (17.62%) and QB (16.91%) than PBA (8.43%), PBB (7.93%) and PBC (9.36%) (P < 0.001).
Correlations between taxonomic taxa and P. notoginseng yields
A total of 43 families and 45 genera (relative abundance > 0.1%) were obtained from rhizosphere soils of P. notoginseng plants (Fig. 3a,b). At the family level, the relative abundance of Caulobacteraceae was 2.08%, 2.45%, 1.94%, 1.42% and 1.27%, respectively, in rhizosphere soils of PBA, PBB, PBC, YS and QB. The relative abundance of Methylobacteriaceae was 0.21%, 0.18%, 0.19%, 0.18% and 0.13%, respectively, in rhizosphere soils of PBA, PBB, PBC, YS and QB. The relative abundance of Caulobacteraceae and Methylobacteriaceae were significantly lower in rhizosphere soils of QB (higher yields) than PBA, PBB, PBC and YS (P < 0.05). The relative abundance of Micrococcaceae was significantly higher in rhizosphere soils of QB (0.76%) than PBA (0.49%), PBB (0.40%), PBC (0.21%) and YS (0.50%) (P < 0.05). Pearsonʼs correlation analysis showed that the relative abundance of Comamonadaceae (R = − 0.89), Opitutaceae (R = − 0.88), Rhodobacteraceae (R = − 0.96) and Sphingobacteriaceae (R = − 0.90) were negatively related to P. notoginseng yield (P < 0.05). Relative abundance of Ktedonobacteraceae (R = 0.95), Streptosporangiaceae (R = 0.91) and Thermomonosporaceae (R = 0.90) were positively correlated to P. notoginseng yield (P < 0.05).
At the genus level, the relative abundance of Leifsonia was significantly higher in rhizosphere soils of QB (0.40%) than PBA (0.21%), PBB (0.15%), PBC (0.17%) and YS (0.20%) (P < 0.01). The relative abundance of Pseudomonas was significantly higher in rhizosphere soils of PBA (lower yields, 1.55%) than PBB (0.93%), PBC (0.60%), YS (0.90%) and QB (1.48%) (P < 0.05). Pearsonʼs correlation analysis showed that the relative abundance of Actinomadura (R = 0.89), Arthrobacter (R = 0.41), Enhydrobacter (R = 0.73), Leifsonia (R = 0.79), Mycolicibacterium (R = 0.83), Rhizophagus (R = 0.74) were positively correlated with P. notoginseng yield. Whereas the relative abundance of Bosea (R = − 0.85), Cupriavidus (R = − 0.91), Mucilaginibacter (R = − 0.89), Novosphingobium (R = − 0.79), phenylobacterium (R = − 0.89), Opitutus (R = − 0.85), Phenylobacterium (R = − 0.89), Polaromonas (R = − 0.82), Reyranella (R = − 0.70), Rhizobium (R = − 0.74), Sphingobium (R = − 0.87), Sphingomonas (R = − 0.83) and Variovorax (R = − 0.85) were negatively correlated with P. notoginseng yield .
The functional traits of the P. notoginseng rhizosphere microbiome
In total, 1,330,812 genes were hit in the KEGG databases and were assigned to 4,436 KEGG orthology (KO) functional categories (Additional file 2). The KOs were mainly involved in 6 KEGG level 1 pathways and 43 KEGG level 2 pathways (Fig. 4a,b). The relative abundance of cellular processes, environmental information processing, genetic information processing, human diseases, metabolism and organismal systems pathways were 0.89%−1.24%, 2.56%−3.54%, 2.83%−3.72%, 1.60%−2.08%, 39.62%−51.83% and 0.94%−1.09%, respectively, in rhizosphere soils of five sample sites at the first KEGG level. The relative abundance of cellular processes, environmental information processing, genetic information processing and human diseases were significantly higher in rhizosphere soils of PBA (lower yields) than PBB, PBC, YS and QB (P < 0.05). The relative abundance of biosynthesis of other secondary metabolites, drug resistance, environmental adaptation, glycan biosynthesis and metabolism, lipid metabolism, metabolism of cofactors and vitamins, membrane transport and signal transduction were 0.46%−0.61%, 0.28%−0.41%, 0.05%−0.07%, 0.40%−0.55%, 1.24%−1.63%, 1.53%−2.10%, 1.38%−1.98% and 1.18%−1.56%, respectively, in rhizosphere soils of five sample sites at the second KEGG level. The relative abundance of biosynthesis of other secondary metabolites, drug resistance, environmental adaptation, glycan biosynthesis and metabolism, metabolism of cofactors and vitamins, membrane transport and signal transduction were significantly higher in rhizosphere soils of PBA (lower yields) than PBB, PBC,YS and QB (P < 0.05). The relative abundance of lipid metabolism was significantly lower in rhizosphere soils of QB (higher yields) than PBA, PBB, PBC and YS (P < 0.05).
PCoA was performed based on KO functional categories using the Bray–Curtis metric, and an adonis test (R2 = 0.62, P = 0.001) showed significant difference in sample sites (Fig. 4c). KO functional categories with a relative abundance exceeding 0.15% are described in Fig. 4d to clarify which KO functional categories were dominant among rhizosphere microbiome of P. notoginseng and which metabolism pathways were the main components in sample sites. Ninety-three KO functional categories were obtained with differences among the sample sites. The K01999 (mean 0.61%), K00626 (mean 0.50%), K00799 (mean 0.43%), K00059 (mean 0.42%) and K03701 (mean 0.40%) were the top five categories that were enriched in rhizosphere soils of PBA, PBB, PBC, YS and QB. Notably, the relative abundance of K01999 was the highest among all the KO functional categories, and K00626 was the second highest. K01999, a branched-chain amino acid transport system substrate-binding protein, is a member of membrane transport pathway. K00626 (atoB, acetyl-CoA C-acetyltransferase) is involved in carbon metabolism, pyruvate metabolism, carbon fixation pathways in prokaryotes, two-component system, fatty acid metabolism and biosynthesis of antibiotics pathways. K00799 (gst, glutathione S-transferase) is involved in glutathione metabolism, metabolism of xenobiotics by cytochrome P450 and drug metabolism–cytochrome P450 pathways. K00059 (fabG, 3-oxoacyl-[acyl-carrier protein] reductase) is involved in metabolic, fatty acid metabolism, fatty acid biosynthesis, biosynthesis of unsaturated fatty acids and biotin metabolism pathways. K03701 (uvrA, excinuclease ABC subunit A) is mainly involved in the nucleotide excision repair pathway.
Correlations between functional traits and P. notoginseng yields
A total of 389 functional classifications were obtained, and the relative abundance showed differences among the sample sites at the third KEGG level (Additional file 3 and Fig. 5). The relative abundance of functional traits, likely steroid hormone biosynthesis (ko00140), lysine biosynthesis (ko00300), ABC transporters (ko02010), two-component system (ko02020) and plant-pathogen interaction (ko04626) were significantly lower in rhizosphere soils of QB (higher yields) than PBA, PBB, PBC and YS (P < 0.05). Eighty-five functional traits were significantly correlated (P < 0.05) with P. notoginseng yields using Pearsonʼs correlation analysis, among 12 and 73 functional traits were positively and negatively correlated with yields, respectively. A total of 304 functional traits were correlated with P. notoginseng yields, among which 140 and 164 functional traits were positively and negatively correlated with yields, respectively.
The KOs involved in known plant–microbe and microbe–microbe interactions, such as bacterial secretion system (ko03070), flagellar assembly (ko02040), bacterial chemotaxis (ko02030) and two-component system (ko02020), were negatively correlated with P. notoginseng yields (Additional file 3 and Fig. 5). Pearsonʼs correlation analysis showed that the abundances of plant–pathogen interaction (ko04626), ABC transporters (ko02010), metabolism of xenobiotics by cytochrome P450 (ko00980), drug metabolism–cytochrome P450 (ko00982), dioxin degradation (ko00621), chloroalkane and chloroalkene degradation (ko00625) and degradation of aromatic compounds (ko01220) were negatively correlated with P. notoginseng yields. Meanwhile the abundances of MAPK signaling pathway (ko04010) and steroid biosynthesis (ko00100) were positively correlated with P. notoginseng yields.