Description of the sequencing data
The samples we obtained were from the feces of a great variety of sheep, which were comprised 40 subjects, namely 10 Dorset sheep (DrS), 15 Small Tail Han sheep (STHS), 5 Tibetan sheep (TS) and 10 Dorper sheep (DrS). We received 1, 694, 264 raw bases on a PE250 instrument, as mentioned in the experiment method. Behind quality–filtering (also as described in the methods), 1, 359, 405 total sequences with an average of 433 bp in length were obtained for the following analysis. In order to normalize the total count for further alpha and beta diversity analysis, the high-quality sequences in all samples were optimized to the minimum number.
Gut microbiota is associated with sheep breeds
After simplifying the original data, 1, 359, 405 high-quality available sequences were obtained. According to 97% breeds similarity, 7039, 6887, 4112, and 8257 OTUs were separately acquired from samples at groups DrS, DsS, TS, and STHS (Table S2), respectively. A total of 26295 OTUs were detected from all samples, of which 2448 obtained in all groups mentioned as core OTUs (Figure 1(A)). The core OTUs contained nearly 9.31 % of the entire OTUs. Furthermore, 225, 203 OTUs were uniquely detected in DrS and DsS groups, and 168, 654 unique OTUs were found in group TS and STHS, respectively. The number of observed OTUs in the TS samples was fewer than that of the other threes.
To confirm the quality of our sequencing data, we examined alpha and beta diversities of bacterial fraction of the sheep microbiota. Several alpha diversity indices diverged significantly between the four sheep breeds (Fig.1 (B) and (C)). The Shannon-Wiener index may straightly expose the heterogeneity of a community according to the amount of breeds present and their related abundance . The Shannon-Wiener estimator of groups DsS, STHS, TS and DrS were 9.92, 9.33, 8.09 and 9.06, respectively. Within the groups, Chao1 and Shannon index visually reflected that the abundance and diversity of intestinal microbial population in the TS group were lower than those in groups DsS, STHS and DrS, and the difference between these four groups was significant (P < 0.05; Figure 1 (B) and (C)). In terms of the Chao1 estimator, 3160, 2598, 1585 and 2207 average OTUs were noted in samples at groups DsS, STHS, TS and DrS, respectively, and the ACE estimator indicated 3263, 2646, 1627 and 2267 (Table S3), which was consistent with the results shown in Chao1, suggested that lower OTU richness was got in TS samples than the other threes (Fig.1 (B) and (C)). Collectively, these data pointed towards a more diverse bacterial population in TS compared to the other threes, and indicated differences in intestinal microbial composition in association with sheep breeds.
Comparison of bacterial microbiome diversity among different sheep breeds
The Bray-Curtis distance matrices were measured according to the OTUs abundance of each sample. Based on the distance matrices, the unweighted Unifrac similarity analysis indicated that the similarities among different sheep breeds were significant. The principal coordinates analysis (PCoA) was performed according to the phylogenetic-tree-based Unifrac metric. As shown in Fig. 2 (A), all samples were assembled into three clusters, scattered points in the principal component denoted different breeds and their relationship between each other. There were significant differences between breeds in relation to microbiome composition (PERMANOVA, p < 0.01). TSs were mainly aggregated in cluster B, whereas DsSs were mostly converged in cluster A. Moreover, DrSs and STHSs were more scattered and found between cluster A and cluster B. Both principal components accounted for 28.1% (PC1) and 8.5% (PC2) of the explained variance. Interestingly, two lambs in STHSs were clustered separately as shown in Fig. 2 (A), suggesting that there was a general difference in gut microbiome between adult sheep and lambs.
Nonmetric multidimensional scaling (NMDS) was used to further clarify the difference among all the breeds in bacterial population composition, which was performed using the Bray-Curtis similarity for all the samples at OUT level. Be regarded as a dominant ordination method that could exhibit the non-linear relationship between samples, NMDS has been widely applied in the study of gut microbiome. As shown in Fig. 2 (B), there was distinguishing clustering of TSs samples, meanwhile samples from DsSs were very close to DrSs. However, the samples from STHS were more dispersed (Fig. 2 (B)).
Additionally, hierarchical clustering analysis of all samples was used to exhibit the similarity between samples, which was performed with Unweighted pair-group method with arithmetic means (UPGMA) and used the Bray-Curtis similarity. Two primary groups were perceived in this analysis. One cluster contains the whole samples of TS and the other cluster contains the all samples from DrS (Fig. 3). It was similar with the results above, that TS samples were distinctive compared with other breeds. And the gut bacterial composition in sheep is largely influenced by the type of samples.
Gut microbial diversities and community composition among different sheep breeds
In order to clarify the diversity of gut bacterial composition in different sheep breeds, we estimated the gut microbiota in different taxonomical levels. The overall bacterial composition of different groups at the phylum level is identified in Fig. 4 (A)， which shows that Firmicutes was the most predominant phylum in the 40 samples, followed by Bacteroidetes. The higher abundance of phylum Spirochaetes, Proteobacteria and Verrucomicrobia were found in TSs than those in the other three breeds, but the Deferribacteres was lacked in TSs (Fig. 4 (A)).
When analyzed on the family level from all samples, as shown in Fig. 4 (B), no significant differences were detected between these four groups. Ruminococcaceae and WCHB1-25 were the most abundant families in DsS, DrS and STHS groups, whereas Alcaligenaceae, Desulfovibrionaceae and Barnesiellaceae were almost absent. As for TS groups, the most abundant families were Spirochaetaceae, S24-7, Prevotellaceae, Barnesiellaceae and Succinivibrionaceae, while BS11 and WCHB1-25 were almost absent in the TS samples (Fig. 4 (B)).
In contrast to the family level, there are significant differences between TS group and the other three groups on the genus level. The main genera in TS group included Treponema, Succinivibrio, 5-7N15 and Prevotella (Table 1), while Bifidobacterium, Sharpea and YRC22 were absent (Fig. 4 (C)). Moreover, in the groups of DsS, STHS and DrS, Treponema remained the predominant population, and Coprococcus and Roseburia were relatively less abundant. However, it's worth noting that a large number of microbes in TS samples were relatively abundant, when compared to other three groups.