3.1 Species Composition Analysis
Through the statistics of the ASV/OTU table after pumping, the specific composition table of microbial communities in each sample at each classification level can be obtained. Then we use the R script to plot the data in the table into a histogram (Fig. 1) to visually show the number of taxa at each classification level for different samples. In order to show the composition of all taxa at the same time, we draw a microbial classification hierarchy tree using ggtree in the R language.
The advantage bacterium group are Proteobacteria (70%), Firmicutes (13%), Actinobacteria (9%) and Bacteroidetes (7%) in all microbiome samples coming from corn and alfalfa in Hengshui and Xingtai (Fig. 2). The four dominant bacteria have more branches, which indicates that the genotypes of the dominant bacteria in the samples are diversified in evolutionary relations. Lactobacillales were found mainly in the samples of corn (Fig. 2). Enterobacteriaceae (24%) belonging to the Proteobacteria phylum are the most predominant bacteria on both corn and alfalfa samples. More than 1% of the reads from 4 genera belonging to Proteobacteria and Bacteroidetes including Pseudomonas (8%), Acinetobacter (4%), Chryseobacterium (3%) and Hymenobacter (1%) (Fig. 3).
3.2 Microbial diversity analysis of maize and alfalfa samples
3.2.1 Alpha diversity analysis
Alpha diversity represents the diversity of species within habitats. Chao1 (Chao. 1984) and Observed species indices measure community richness. Shannon and Simpson (Simpson. 1949) indices measure community diversity. Faith's PD (Faith., 1992) index represents diversity based on evolution. Pielou’s evenness (Pielou. 1966) index represents the evenness. Good's coverage (Good. 1958) index represents coverage. And the specific results were plotted into a boxplot using the R script to visually show the difference of alpha diversity between different groups. It can be seen that the microbial community richness, diversity, evenness and evolutionary diversity of microbiome samples of alfalfa (group E and G) is higher than that of maize (group D and F) on average. However, the coverage of species in a community of microbiome samples of alfalfa is lower than that of maize (Fig. 4). The microbiome samples of alfalfa and maize in Xingtai have extreme significant differences in community richness and evolutionary diversity. And the microbiome samples of alfalfa and maize in Hengshui have highly significant differences in community richness, diversity and evenness. Thus, the diversity of epiphytic microorganism community is significantly affected by plant species. All alpha diversity indices of alfalfa in different areas have no significant difference. And alpha diversity indices of maize except the evolutionary diversity have no significant difference in Xingtai and Hengshui. We can conclude that the region has no significant effect on the diversity of epiphytic microbial community. The epiphytic microbial diversity of Shengrui 565 is lower than other breeds in the two places.
3.2.2 Beta diversity analysis
The microbial communities in alfalfa and maize samples were compared using NMDS based on the weighted UniFrac distance (Lozupone and Knight. 2005). Each point in the diagram represents a sample, and different colored dots indicate different samples (Fig. 5). Samples are clustered according to their similarity, and the closer the distance between two points is, the more similar the tow samples are. Alfalfa group samples were aggregated in the NMDS analysis diagram, while the samples of the maize group were dispersed. Samples of corn (group D and F) are similar and the samples of alfalfa (group E and G) are similar. The results showed that epiphytic bacteria were more affected by species than by region.
3.3 Species difference analysis and biomarker
The number of ASV/OUT in group D, E, F and G are 6683, 8305, 6920, 8080 respectively (Fig. 6). And there are 545 ASV/OUTs in common.
We use the abundance data of the top 50 genera in average abundance to make a heat map. In the genus-level species composition heat map for species clustering, red patches indicate that the genera are more abundant in this sample than other samples, and blue patches indicate that the genera are less abundant in this sample than other samples. Lactic acid bacteria have an important effect on the silage fermentation, such as Leuconostoc and Lactobacillus in the top 50 genera in average abundance. Leuconostoc mainly exist in group D2, F1, F2 and F3. Lactobacillus mainly exist in group D1, D3, D4, F1, F2, F3, E1, E2 and E3 (Fig. 7). Group F1 and F2 have more Clostridium sensu stricto 1 that are harmful to fermentation (Fig. 7).
Through the algorithm analysis of Random Forests (Breiman. 2001), we obtained the distribution of important species in each group (Fig. 8). The abscissa is the importance of species to the classifier model, the ordinate is the taxon name at the level of genus from top to bottom, and the importance of species in influencing grouping decreases successively. These highly important species can be considered markers of differences in these groups, and they are Pedobacter, Nocardioides, Chryseobacterium, Burkholderia − Caballeronia − Paraburkholderia, Paracoccus, Pseudomonas, Acinetobacter, Allorhizobium − Neorhizobium − Pararhizobium − Rhizobium, Larkinella, Mucilaginibacter, Sphingomonas, Brevundimonas, Siphonobacter, Methylobacterium, Spirosoma, Hymenobacter, Bacillus, Actinomycetospora, Taibaiella, Sphingobacterium. Bacillus mainly exist in SR4030 and Saidi 5 in Xingtai.