OTU composition of the bacterial community
The high throughput sequencing finally obtained 892,262 high-quality sequences (Clean tags) for subsequent analysis with a high-quality rate of 90.80%. Sequences were clustered at 97% similar levels yielding 1184 OUTs, remaining 1150 after leveling. In addition, to obtain the taxonomic information corresponding to each OTU, the RDP Classifier algorithm was used to annotate OUT at phylum to genus level (Table 1).
Table 1
The endophytic bacterial OTU composition of five species of Azolla
Sample
|
Phylum
|
Class
|
Order
|
Family
|
Genus
|
Afi
|
19
|
37
|
73
|
121
|
156
|
Ame
|
14
|
25
|
49
|
83
|
103
|
Aca
|
12
|
21
|
41
|
63
|
79
|
Api
|
19
|
38
|
66
|
125
|
161
|
Aim
|
26
|
52
|
95
|
156
|
194
|
To intuitively show the difference of OTU in different Azolla species, a Venn diagram showing the distribution of common or unique OTU in five species of Azolla was detailed in Fig. 1. At the OTU level, OUT common to the five species of Azolla represented 6.17% (71) of the total amount of OUT, and the remaining OUT were cross-shared or independently owned by the samples. Among them, Aim contained the most specific OTU, for 46.39% (283) of its total OTU, indicating the most abundant specific bacterial groups owned by Aim. Furthermore, Afi and Api jointly owned 328 OTU, for 56.07% and 55.59% of their total OTU, respectively, indicating more bacterial taxa shared by both. Aim and Api shared 276 OTU, accounting for 45.25% and 46.78% of their total OTU, respectively. Ame and Aca jointly owned 111 OTU, accounting for 49.78% and 66.07% of their total OTU, respectively, suggesting similarities in these two communities.
α diversity analysis of bacterial communities
To further assess the abundance and diversity of the endophytic bacterial community in Azolla, the α diversity index of samples were calculated and results were detailed in Table 2. The PD whole tree was the diversity index that takes into account the species abundance and evolutionary distance, and the results indicated that the abundance of the bacteria in Afi, Api, Aim was significantly higher than it was in Ame and Aca. Chao1 index was used to estimate the abundance of species in the community, the larger index indicating a higher community richness. Chao1 index indicated the abundance of bacterial communities ordered as Afi > Api > Aim > Ame > Aca. Shannon index was used to estimate bacterial diversity, the larger values indicating higher community diversity. The Shannon index of the bacterial community in Aim was the highest, while the Shannon index of Aca was the lowest. Simpson index was used to estimate bacterial diversity in samples, considering both abundance and uniformity, and greater values indicate higher community diversity. The Simpson index indicated that the bacterial community diversity was ranked as Aim > Api > Afi > Ame > Aca, this result was consistent with the Shannon index, but it was slightly different from the PD whole tree and the Chao1 index.
Table 2
Richness and diversity index of endophytic bacteria in five species of Azolla
Samples
|
PD whole tree
|
Chao1 index
|
Shannon index
|
Simpson index
|
Afi
|
38.41 ± 2.44a
|
505.14 ± 26.07a
|
4.58 ± 0.13b
|
0.91 ± 0.01b
|
Ame
|
20.10 ± 5.94b
|
181.82 ± 67.65b
|
2.27 ± 0.14c
|
0.68 ± 0.02c
|
Aca
|
8.93 ± 2.56b
|
134.05 ± 64.38c
|
1.69 ± 0.08d
|
0.64 ± 0.01c
|
Api
|
37.41 ± 5.22a
|
431.30 ± 61.61ab
|
4.92 ± 0.10b
|
0.91 ± 0.01b
|
Aim
|
32.38 ± 3.02a
|
367.25 ± 42.24b
|
6.45 ± 0.06a
|
0.98 ± 0.00a
|
Data are means of triplicates (Mean ± SEM, n = 3). Different letters indicate the significant difference between samples (P < 0.05). |
β Diversity Analysis Of Bacterial Communities
The bray Curtis-based principal coordinate analysis (PCoA) method was used to investigate the similarities or differences in the community structure in different species of Azolla. Bray-Curtis distance considers only the presence of species and their abundance, no evolutionary relationship between species, and distance 0 indicates the complete consistency of the species in two communities. The PCoA of the bacterial community exhibited four distinct clusters of samples (Fig. 2). PCoA 1 and PCoA 2 explained 59.16% and 21.95% of the variation, respectively. The bacterial communities of Ame and Aca were very close to each other, indicating the similarity in their endophytic bacterial communities. The placement of Api and Aim communities was closer, indicating that their species structure was more similar. The distance of Afi from the other four Azolla species was the greatest indicating it owned most species-specific bacterial communities.
Points of different colors or shapes represent different grouping situations, and the scale of the horizontal and vertical axes is a relative distance and meaningless. PC1, PC2 represent the suspected influencing factors of the microbial composition of the two groups, which need to be combined with the sample characteristic information.
Composition Analysis Of The Bacterial Community
The bacterial communities showed high diversity at the phylum level, totally up to 19 bacterial phyla distributed among five species of Azolla. The relative abundance of Proteobacteria in Afi (92.96%), Ame (95.60%), and Aca (99.58%) was significantly higher than in Api (86.10%) and Aim (52.23%). While the relative abundance of Actinobacteria and Firmicutes in Afi (0.389%, 0.12%), Ame (0.22%, 1.13%), Aca (0.02%, 0.1%) was significantly lower than in Aim (24.33%, 7.02%) and Api (0.67%, 1.59%), respectively (Fig. 3A).
Bacterial communities are more diverse at the class level (Fig. 3B). Moreover, the bacterial groups and relative abundance of different species of Azolla are different. The bacterial community of Afi mainly include 62.63% α-Proteobacteria, 24.32% β-Proteobacteria, 7.06% γ-Proteobacteria, 2.10% δ- Proteobacteria, and 1.12% Sphingobacteriia. The bacterial community of Ame is mainly composed of 78.35% β-Proteobacteria, 18.58% α-Proteobacteria, and 1.11% Bacilli. And it is mainly composed of 82.83% β-Proteobacteria and 16.66% α-Proteobacteria of Aca. There are 71.18% β-Proteobacteria, 9.14% α-Proteobacteria, 7.46% γ-Proteobacteria, 2.67% δ-Proteobacteria, 2.25% Sphingobacteriia and 2.15% Clostridia of Api. While it is mainly composed of 27.64% β-Proteobacteria, 23.84% Actinobacteria, 20.17% α-Proteobacteria, 7.97% γ-Proteobacteria, 5.82% Bacilli, 2.98% Chloroflexia, 2.78% δ-Proteobacteria and 1.81% Clostridia of Aim (Fig. 3B).
The endophytic microbial community composition of the five species of Azolla was different at the family level (Fig. 4A). In detail, the bacteria inhabiting Aca and Ame were relatively similar, and the bacterial diversity of these two Azolla was significantly lower than that of the other three. The family of Oxalobacteraceae belonged to the order of Burkholderiales was the predominant bacterial group in Ame (46.00%) and Aca (44.30%), however, it was less than 1% in the other three species. Comamonadaceae was the dominant group in Api, with a relative abundance of 57.00%. While it accounted for 18.56% and 21.70% of the total bacteria of Afi and Aim, respectively. The relative abundances of Methylophilaceae in Ame and Aca were 30.00% and 38.00%, respectively, but their relative abundances in Api was 11.30%, and it was both less than 5.00% in Afi and Aim. The relative abundance of Rhizobiaceae was 17.17% in Afi, 14.27%, and 16.03% in Ame and Aca, respectively, while it was only 7.44% and 1.42% in Aim and Api, respectively. The most abundant bacterial group was Rhodospirillaceae in Afi (39.61% ), but it was extremely low in the other four species of Azolla.
Further analysis revealed that endophytic bacterial communities showed high diversity and significant differences at the genus level among different species of Azolla (Fig. 4B). The bacterial communities of Ame and Aca were mainly composed of Herbaspirillum (45.91%, 44.08%) and Methylophilus (29.97, 37.96%), however, they were less than 5.00% in Afi and Aim. The relative abundance of Ideonella in Afi, Api, and Aim were 8.87%, 14.00%, and 3.43%, respectively, while it was less than 1.00% in both Ame and Aca. The genus of Ferrovibrio was the most abundant bacterial group in Afi (18.53%) and it was less than 1.00% in the other four species of Azolla. Rhizobium was the second abundant in Afi (16.68%), and it was 6.49% in Aim, but it was less than 1.00% in others.The relative abundance of Nitrospirillum and Dongia in Afi was 10.92% and 9.72%, respectively, but it was less than 1.00% in others. The relative abundance of Cellvibrio in Afi, Aim, and Api was 5.40%, 3.79%, and 3.31%, respectively, but less than 1.00% in Ame and Aca. Bacillus, Asticcacaulis, Variovorax were specific taxa in Aim, however, their relative abundances were relatively low. These results showed that different species of Azolla have an enrichment preference for the endophytic bacterial community.
Drivers of variation in endophytic bacterial communities composition in Azolla
To further explore the variation of the bacterial communities in different species of Azolla, the LEfSe analysis method was used to find bacteria with significant differences in abundance between samples (Fig. 5). In detail, at different taxa levels, specific bacteria inhabiting in Aim than in other species of Azolla. The genus of Methylophilus was specific in Aca. The Herbaspirillum, Devosia were specific in Ame. The phyla of Actinobacteria, Firmicutes, Chloroflexi, Cyanobacteria were more abundant in Aim compared to other species of Azolla. The family of Comamonadaceae was the richest group in Api. Moreover, the phyla of Bacteroidetes, Spirochaetae, and Planctomycetes exhibited a difference in relative abundance in Api compared to other Azolla. The class of Alphaproteobacteria was the most abundant in Afi, especially the genera of Ferrovibrio, Rhizobium, and Nitrospirillum (Fig. 5).
Circles radiating from inside to outside represent classification levels from phylum to genus. Each small circle at different taxonomic levels represents a classification at that level, with the small circle diameter size being proportional to the relative abundance size. Coloring principle: species with no significant differences were colored to yellow, and different species biomarker follows the group. Red nodes indicate microbial taxa that play an important role in the red group, green nodes indicate microbial taxa that play an important role in the green group, and other circle colors mean the same. The species names indicated by the English letters in Fig were shown in the legend to the right.