Analysis of 16S rRNA sequencing results
Six samples were taken from each group, but only 5, 5 and 3 samples from BL, BH and SH groups respectively accorded with the requirements of library construction, because the rest samples of PCR products were no purpose bands or low concentrations. The sequencing results were treated by a series of purification and filtration processes. Finally, a total of 246,700 sequences were obtained from the 13 samples, and the effective sequences of 13,740 to 28,143 were collected from each sample. The filtered sequencing data of each sample were counted in Table S1 (Additional file 1). The proportion of effective tags among Raw_Tags was in the 60% to 80% range at each sample. After quality control filtration, the reads sequences within the corresponding length range of each sample were calculated. Distribution of the effective sequence length is shown in Fig. S1 (Additional file 1). 400 bp to 440 bp were the most effective sequence length distribution region.
The sequences number of each sample OTU was distributed in the 97% sequence similarity threshold. Meanwhile, the classification information for each species corresponding to each OTU was also obtained by comparing the OTU representative sequences with a microbial reference database. The current results showed that the detected bacteria can be classified into 13 phyla, 21 classes, 41 orders, 77 families, and 120 genera. The composition of each sample community was calculated at phylum, class, order, family, genus and species levels, respectively. Table S2 (Additional file 1) was showed the number of each species at different levels, and Table S3 (Additional file 1) was recorded the total number of OTUs covered by each sample in their subordinate levels. The number of OTUs increased with the depth of sequencing, which was confirmed by the dilution curves of the OTUs measured in this study. The amount of sequencing data is reasonable because the final curve became stable (Fig. 1).
Comparison of core intestinal microbiota in sturgeons from three experimental groups
The core intestinal microbiota of the sturgeons in three experimental groups was shown by Venn diagrams (Fig. 2). The core microflora of sturgeons meant the common bacterial populations in three experimental groups. All sturgeons in three experimental groups were shared 243 OTUs, while sturgeons in BL and BH groups were shared 510 OTUs, and 278 OTUs for sturgeon in BH and SH groups (Fig. 2). The main bacterial phyla in the intestines of three groups’ sturgeons are shown in Fig. 3. Proteobacteria, Firmicutes and Cyanobacteria were three dominant phyla in BL and BH groups, while SH group was Proteobacteria, Firmicutes and Actinobacteria. The core intestinal microflora in fish from three experimental groups were the same, but their relative abundances were different. The top five core intestinal microflora at phyla in BL group showed the order from high to low was Proteobacteria, Firmicutes, Cyanobacteria, Bacteroidetes and Actinobacteria (P<0.05), whereas the core intestinal microflora in BH groups showed the difference with the order of Cyanobacteria>Proteobacteria>Firmicutes>Actinobacteria>Bacteroidetes. And the order from high to low of core intestinal microflora in SH group was Proteobacteria, Firmicutes, Actinobacteria, Cyanobacteria, and Bacteroidetes.
A heatmap (Fig. 4) is a graphical representation that uses a system of colored gradients to represent the size of values in a data matrix, and the cluster data are also expressed in heatmap according to species or the abundance similarity of samples [20]. In order to reflect the similarities and differences between multiple sample communities, high-abundance and low-abundance species are clustered by color gradient and similarity [21]. Based on the species composition and relative abundance of each sample, a heatmap analysis was performed to extract the species at each taxonomic level [21]. Mapping was achieved using R language tools, and a heatmap cluster analysis was performed at the levels of the phylum, class, order, family, genus, and species, respectively [21]. The result was found in Fig. 4 that the vertical clustering between BH and SH groups showed the long branch length, which indicated the richness of intestinal microbiota between two groups was clearly different. Similarly, BL and BH groups have a certain degree of similarity in richness, because the short branch length was found in BL and BH groups at Fig. 4. In addition, the relative abundance of intestinal flora at genus level among BL, BH and SH groups were also showed in Fig. 4. The major intestinal microflora at genus level in BL group were Acinetobacter, Pseudomonas, Bacillus, and so on, as same as the BH group, while the SH group included Bacillus, Staphylococcus and Acinetobacter intestinal microflora at genus level.
Alpha diversity analysis of microbial communities in sturgeons from three experimental groups
Alpha diversity indices mainly include Chao1, Observed_species, Shannon, PD_whole_tree, and Good’s coverage, which can response to the richness and diversity of a single sample species [21]. The Chao1 index measures the richness of species (i.e., the number of species), whereas the Shannon index measure the diversity of species [21]. Observed_species shows the number of OTU was observed with the increase of sequencing depth. PD_whole_tree refers to the number of species observed, reflecting the abundance of the colony. Good’s coverage reflects the completeness of the sequencing. In this study, 99% good’s coverage indicated that the most bacterial species present in the sample had been detected.
The alpha diversity results of the gut microbiota in sturgeons from three experimental groups were showed in Table 2. A total of 826 OTUs were obtained at the 97% similarity level. There was a significant difference in the OTUs, Chao1, Good’s coverage, Observed species and Shannon indices between the BH and SH groups (P<0.05), but no significant difference was found in the PD whole tree of the two groups (P>0.05). The result showed that the alpha diversity indices were considered to be significantly different between Siberian sturgeon and Beluga sturgeon when they fed with the same diet (P<0.05).
Meanwhile, although the most alpha diversity indice were no significant difference between BL and BH groups (P>0.05), but the Shannon index was obviously higher in BL group than that in BH group (P<0.05), which pointed that the species diversity of the intestinal microbiota was enhanced when Beluga sturgeon fed with the low fishmeal diet.
Beta diversity analysis of microbial communities in sturgeons from three experimental groups
Non-metric multidimensional scaling (NMDS) (Fig. 5) is a data analysis method that simplifies research objects (samples or variables) in multidimensional space to low-dimensional space for positioning, analysis and classification, while retaining the original relationship between objects. It is applicable to the case that the exact similarity or heterogeneity data between the research objects cannot be obtained, but only the hierarchical relationship data between them can be obtained. Its basic feature is to treat the data of similarity or dissimilarity between objects as a monotone function of the distance between points, and replace the original data with a new data column of the same order for the metric multidimensional scale analysis on the basis of maintaining the original data order relationship. In other words, when the data is not suitable for the direct multidimensional scaling analysis of variable type, the variable transformation is carried out and then the multidimensional scaling analysis of variable type is adopted. For the original data, it is called non-metric multidimensional scaling analysis. Its characteristics are reflected in multi-dimensional space in the form of points according to the species information contained in the sample, and the degree of difference between different samples is reflected by the distance between points, and finally the spatial locus map of the sample is obtained.
The closer the distance between sample points, the higher the similarity. Generally speaking, the samples within the same circle meant the difference between samples was not obvious, while the sample points within circles with no intersection indicated that there was significant difference between the samples. As shown in Fig. 5, the circles were intersect between BL and BH groups, but no intersection was found in BH and SH groups, which indicated the beta diversity between BH and SH groups was significant difference, whereas BH and BH groups was not.
Analysis of the differences in intestinal microbiota among sturgeons in three experimental groups at the phylum and genus levels
LEfSe (Linear discriminant analysis Effect Size) (Fig. 6) is an algorithm for high-dimensional biomarker discovery and explanation that identifies bacteria of each level of phylum, class, order, family, or genus characterizing the differences among BH, BL and SH three groups (Fig. 6). The cladogram (Fig. 6a) showed differences in 102 taxa among fish in BL, BH and SH groups. BL fish (enrichment in the Gammaproteobacteria and Proteobacteria) have no similar to BH fish, with a lot of changes occurring in the BH group such as enrichment in Actinobacteria, Corynebacteria and Norynebacteria (Fig. 6b). And the SH group was mainly enriched in Bacillales and Bacillus genus, which was clearly difference with BH group (Fig. 6b).
In addition, Fig. 7 shows the relative abundance differences in phylum, family and genus levels among the three experimental groups. The relative abundance of Proteobacteria and Bacteroidetes in BL group sturgeons were significantly higher than that in BH group (P<0.05), while the relative abundances of the Cyanobacteria phyla in BL group were significantly lower than in BH group (P<0.05). The relative abundances of Firmicutes and Actinobacteria have no significant differences in both groups (P> 0.05) (Fig. 7a).
Compared with the sturgeons of BH group, the relative abundance of Proteobacteria and Firmicutes were significantly increased in sturgeons from SH group, whereas the relative abundance of Cyanobacteria were clearly decreased (P<0.05). The Bacteroidetes and Actinobacteria relative abundance were no significant differences in sturgeons of BH and SH groups (P> 0.05) (Fig. 7b).
At the genus level, the relative abundances of Pseudomonas and Citrobacter in BL group were significantly higher compared with BH group (P<0.05) (Fig. 7c), while the relative abundance of Bacillus, Luteibacter, Staphylococcus, and Oceanobacillus was lower in BH group than in SH group (P<0.05) (Fig. 7d). Three groups did not show any significant differences in the relative abundances of Alpinimonas and Acinetobacter (P>0.05).