The overviews of microbial community and function
Here, metagenome sequencing was performed on 91 samples that represented six (donkeys) and nine (cows) digestive tract regions with swab and fecal samples of monogastric (donkeys) and polygastric (cows) animals. Using high-throughput sequencing, we obtained approximately 1.66 TB of metagenomic sequencing data (4,605,907,824 raw reads). After quality control, a total of 1.5 Tb of clean, high-quality data remained, with an effective data quality control rate of 98.5%.
We assembled 4,004,115 (cow) and 2,938,653 (donkey) contigs, and obtained 9,060,744 genes with an average of 553 bp (range 102 bp ~ 10,912 bp) were obtained by clustering at 95% identity and 90% coverage (Additional file 1: Fig. S1a). The gene length mainly distributed between 100 and 1,000 (Additional file 1: Fig. S1b). To evaluate the total number of genes that could be identified from these samples, rarefaction analysis was performed with random sampling 50 times and the core-pan gene rarefaction curve was close to plateaus (Fig. 1b, Additional file 1: Fig. S1c). We used heatmap to calculate the correlation between the samples with the gene table (Fig. 1b). The results showed the positive correlation between digestive tract segments within the same animal and negative otherwise. What's more, different digestive tract segments and individuals of cow showed high consistency especially within four stomachs, large intestine (cecum, colon, rectum) and anus.
In the analysis of microbial composition, we focused on exploring the bacteria that accounted for 76% of all microbial community (Fig. 1c).The number of microorganism at other level were 135 (phylum), 90 (class), 189 (order), 436 (family), 2,133 (genus) and 12,213 (species). The most abundant five phyla were Firmicutes, Bacteroidetes, Proteobacteriam, Actinobacteria and Spirochaetes and the more microbial composition of all samples at different levels were listed in the additional file 1: Fig. S2. At the species level, the number of microbes was more abundant in the fecal samples and digestive tract segments of donkey besides stomach (Fig. 1d). Further, we performed a functional analysis of microbiota using several databases to investigate the functional differences (Fig. 1d). We counted the abundance of metabolic pathways of the KEGG and more microbial functions were related to metabolism (Fig. 1e). As for the other database, ‘Replication, recombination and repair’ was the most abundant function in the eggNOG, ‘ABC1, ABC2, ABC3 Superfamilies’ accounting for 41% in the TCDB, ‘antibiotic efflux’ and ‘Glycoside Hydrolases’ were particularly enriched in the CARD and CAZy, respectively (Additional file 1: Fig. S3, S4).
Microbiota was more susceptible affected by sampling method and fecal samples embodied more abundant microbes
To investigate the sampling method differences in the intestinal microbiota of monogastric and polygastric animals, we performed correlation analysis on donkey (Fig. 2) and cow (Fig. 3) respectively.
We firstly calculated the effects of sampling method on donkey. At the phylum level, Partial Least Squares Discriminant Analysis (PL-SDA) showed the differences with 9.75% and 10.07% variations explained by PC1 and PC2 (Fig. 2a, Additional file 1: Fig. S5). We then conducted ANOSIM similarity analysis, and the results showed that the sampling methods of the donkey were significantly different (P < 0.05, R = 0.1454) (Fig. 2b, Additional file 1: Fig. S6). In the analysis of microbial composition, we focused on exploring the top 10 bacteria at the phylum and genus levels. We analyzed the composition of microorganisms at each taxonomic level. At the phylum level, Firmicutes (75.47%, 64.76%) and Bacteroidetes (17.23%, 24.69%) were still the dominant bacteria with swab and fecal samples (Fig. 2c). Other levels showed the similar differential results (Additional file 1: Fig. S7), and at species level, there were 389 and 109 special species recognized by single method (Fig. 2d, Additional file 1: Fig. S8). We performed differential analysis using STAMP, differential bacteria at different taxonomic level were obtained, and the differential bacteria only appeared at the genus level that were Pseudoscardovia, Raoultibacter, Enteroscipio and Senegalimassilia, and other levels’ results were all showed in the Additional file 1: Fig. S9 and Table S1. All the differential functions by stamp analysis were showed in the Additional file 2: Table S2. The number of unique species was more abundant in the fecal samples, which was reflected in the every digestive tract segments (Fig. 2e). We also used random forest algorithm that could effectively and accurately classify microbial community and identify Firmicutes as a key phylum that could distinguish between groups (Fig. 2f), and Bacilli, Eggerthellales, Eggerthellaceae, Libanicoccus and hyointestinalis were the most important microbes at the other levels from class to species in order (Additional file 1: Fig. S10). To further explore the different microorganisms between the methods, we performed the LEfSe analysis on the donkey (Fig. 2g). The phylum, more abundant bacteria in the swab samples, was Proteobacteria (LDA = 4.25), and Tannerellaceae under Bacteroidetes was also a significantly abundant bacteria at the family level. The genus of more abundant bacteria was Tannerella which was also under the class of Bacteroidetes. What’s more, other 18 significantly abundant bacteria were all at the species level, in which Campylobacter jejuni was the most significantly higher than another sampling method (LDA = 4.18). In the fecal samples, Morganellaceae under Enterobacterales and Moraxellaceae under Pseudomonadales were more abundant bacteria at the family level. Xenorhabdus under Morganellaceae was also more abundant. These three bacteria were all belonged to Proteobacteria phylum. The other 45 significant microorganisms were all at the species level, and the Lactobacillus hayakitensis was the most significantly abundant species (LDA = 4.29).
We then did the same operations for cow like donkey. PL-SDA showed the differences with 8.21% (PC1) and 6.17% (PC2) (Fig. 3a, Additional file 1: Fig. S5), and ANOSIM also showed the same results (P < 0.05, R = 0.1454) (Fig. 3b, Additional file 1: Fig. S6). The composition of microorganisms at each taxonomic level performed the both consistency and difference (Fig. 3c, Additional file 1: Fig. S7). At the species level, there are 630 and 96 species recognized by only one method respectively (Fig. 3d) and other levels’ results were shown in the Additional file 1: Fig. S8. STAMP analysis showed Actinobacteria, Firmicutes, Fusobacteria, Chloroflexi, Lentisphaerae were the significantly differential bacteria at the phylum level, and other levels’ results were showed in the Additional file 1: Fig. S9 and Table S3. All the differential functions by stamp analysis were showed in the Additional file 2: Table S4. The number of unique species was more abundant in the fecal samples, which was reflected in the every digestive tract segments and more significantly than the donkey (Fig. 3e). Random forest algorithm identified Actinobacteria as a key phylum that can distinguish between groups (Fig. 3f), and other key microbes at the other levels from class to species in order were shown in the Additional file 1: Fig. S10. Compared with donkey, the differential bacteria calculated by LEfSe analysis of cow not only had more counts but also were located at more different taxonomic level (Additional file 1: Fig. S11). In swab samples, there were three phyla Bacteroidetes, Proteobacteria and Spirochaetes were more abundant. There were 8,10,15,16, and 70 bacteria at the class, order, family, genus and species level, respectively. The rank of most significantly abundant bacteria were Bacteroidetes, Bacteroidia, Bacteroidales, Prevotellaceae, Prevotella, Prevotella ruminicola in order from phylum to species. However, in the fecal samples, the differential bacteria’ taxonomic level were focused on the species level, and there were 100 significantly abundant bacteria at the species level. Fusobacteria, Fusobacteriia, Fusobacteriales, Fusobacteriaceae and Fusobacterium were all significantly higher from phylum to genus level. Alkalilimnicola under Ectothiorhodospiraceae and Ruminobacter under Succinivibrionaceae were another two differential bacteria and they were both belonged to Proteobacteria. The Sharpea azabuensis, Sarcina sp. DSM 11001, Kandleria vitulina and Bifidobacterium pseudolongum were the four significantly abundant species in the fecal samples with the LDA score greater than 4.
Bacteria community in the stomach was significantly different from other segments and colon held the most abundant bacteria in the other intestine
In the analysis of the microbes on different digestive tract segments, we analyzed the bacterial composition in the donkey. At the different level, bacterial composition and abundance presented commonality and peculiarity, and stomach was obviously different from other intestinal segments (Fig. 4a-b, Additional file 1: Fig. S12). There were 14 core phyla among all donkey samples (Fig. 4c), and most phyla were distributed in the all digestive tract segments (40/126) and intestinal segments (66/126) (Fig. 4d). At the species level, intestinal segments and all digestive tract segments contained the most abundant microbes consistent with phyla (Fig. 4e). Ranked by importance value, the top 30 microbial characteristics cumulatively accounted for 50.7% of the importance of the total characteristic value, and Armatimonadetes was the most important one(Fig. 4F).
In the cow, bacterial composition presented that stomach was different from other intestinal segments and small intestine was also significantly different from hindgut (Fig. 6a-b, Additional file 1: Fig. S13). The most microbes were distributed in the all digestive tract segments at the phylum (138/176) and species level (Fig. 6c, 6d). The number of peculiar species in the stomach of cow was significantly higher than that in other intestinal segments and cow have more core species than donkeys. Ranked by importance value, the top 30 microbial characteristics cumulatively accounted for 44.9% of the importance of the total characteristic value, and Elusimicrobia was the most important one but not significantly different with other characteristics which was different from donkey (Fig. 6e).
At the phylum level, Firmicutes and Bacteroidetes were the dominant bacteria (> 80%) in the intestinal segment. But there were a few differences in stomach between donkeys and cows, that Firmicutes were accounted for the vast majority in donkeys and Bacteroidetes had more proportion in cows. Besides, Proteobacteria and Actinobacteria were the another dominant bacterial phyla. At the genus level, Lactobacillus (~ 80%) and Prevotella (40%~60%) were the dominant genera in the donkey and cow’ s stomaches. Compared to the stomach, most of the annotated bacteria were homogeneity in the hindgut. At the species level, equigenerosi, hayakitensi, crispatus and equicursoris were the dominant species in the donkey stomach (> 60%) and they all belonged to Lactobacillus(Additional file 1: Fig. S12, S13).We also used the venn diagram to calculate the species composition shared by rumen, reticulum, omasum, abomasum and donkey stomach. Most species in the donkey stomach (1,992/2,330) also existed in the cow and there were nearly 80% (5950/7503) microbes in all cow stomaches (Fig. 6F). The differential bacteria calculated by LEfSe analysis of rumen, reticulum, omasum and abomasum were located at different taxonomic level (Additional file 1: Fig. S14). In the rumen samples, there was a genus (Dongia) and 15 species (Streptomyces sp. BK335, Bacillus alkalitelluris, etc) significantly abundant than other stomaches. As for the reticulum, 2 genera (Psychrosphaera, Ochrobactrum) and 32 species (Clostridiales bacterium, Bacteroides_sp_43_108, etc) were more abuntant. And there were 4 genera (Trinickia, Sphaerochaeta, Fibrisoma and Mannheimia) and 10 species (Micromonospora globispora, Trinickia soli, etc) abundant in the omasum. Differ from the other stomaches, there were more significantly abundant bacteria in the abomasum which were distributed in the different levels including phylum Proteobacteria, class Alphaproteobacteria, 5 order (Rickettsiales, Enterobacterales, etc), 6 families (Anaplasmataceae, Enterobacteriaceae, etc), 10 genera (Anaplasma, Klebsiella, etc) and 42 species (Anaplasma phagocytophilum, Staphylococcus aureus, etc)
The microbial community structure of donkey was obviously different from cow for both dominant and endemic microbes
To confirm the correlation between the animals, we analyzed the bacterial composition of all samples. At the different level, bacterial composition and abundance presented commonality and peculiarity between donkey and cow (Fig. 6a, 6b, Additional file 1: Fig. S15). PLS-DA showed the differences with 10.6% and 6.28% variations explained by principal component 1 (PC1) and PC2, respectively (Fig. 6c). In addition, NMDS, OPLS-DA and PCA analysis drawn the similar result with PLS-DA (Additional file 1: Fig. S16). We also used a venn diagram to calculate the number of species shared by different animals (Fig. 6d). There are 6,860 common species, accounting for 68% and 78% of the donkey and cow respectively. However, there was only six common species of the cow and donkey between the most abundant 30 microbes. The number of unique species was more abundant in the donkey, which was reflected in the every digestive tract segments besides stomach (Fig. 3e).
The phylum of more abundant bacteria in the cow was Spirochaetes, and Acidobacteria in the donkey was a significantly abundant bacteria at the phylum level (LDA > 2, P < 0.05; Additional file 1: Fig. S17). Alphaproteobacteria, Spirochaetia, Gammaproteobacteria were the abundant classes in the cow while in the donkey were Flavobacteriia, Betaproteobacteria, Cytophagia, Sphingobacteriia, Chitinophagia. Besides Betaproteobacteria was under the class of Proteobacteria, others were belonged to Bacteroidetes. At the order level, there were five and eight significant bacteria in the cow and donkey respectively. Former included Rickettsiales, Spirochaetales, Pseudomonadales, Aeromonadales, Brachyspirales and the latter included Flavobacteriales, Marinilabiliales, Burkholderiales, Pasteurellales, Cytophagales, Rhizobiales, Sphingobacteriales, Chitinophagales. The number of significantly bacteria in the cow and donkey were nine and fourteen at the family level. The top three genera of more abundant bacteria in the cow were Prevotella under Bacteroidetes, Anaplasma under Proteobacteria, Treponema under Spirochaetes and another eleven genera. Here, Prevotella was the most significantly higher than that in the donkey of all the levels (LDA = 4.70). Meanwhile, the significantly abundant bacteria at the genus of the donkey included Bacteroides under Bacteroidaceae, Tannerella under Tannerellaceae, Xenorhabdus under Proteobacteria and ten more. Finally, the species of more abundant bacteria in the cow and donkey were 142 and 176 respectively. The species of more abundant bacteria in the cow included Prevotella ruminicola, Sharpea azabuensis and Kandleria vitulina, while Lactobacillus equigeneros, Lactobacillus hayakitensis and Lactobacillus crispatus were the top three more abundant species in the donkey compared to cow, which may be related to the digestive function.
We also used random forest algorithm that can effectively and accurately classify microbial community samples and identify Lactobacillus equi as a key species that can distinguish between groups (Fig. 6f, Additional file 1: Fig. S18). All the function annotation results between animals of the other databases were listed in the Additional file 1: Fig. S19. All the differential microbes and functions by stamp analysis were showed in the Additional file 2: Table S5 and S6.