Results
A total of 535430 bacterial 16S rRNA V4 amplicon sequence were obtained from 9 samples covering 3 biotopes: soil, phyllosphere and faeces in the grazed temperate steppe in Inner Mongolia and further clustered into 2136 OTUs. Simultaneously, 840851 ITS gene sequence and 614820 16S gene sequence were also generated and clustered into 1340 fungal and 621 archaeal OTUs respectively.
The α-diversity revealed distinct differences in richness, diversity and evenness of communities among biotopes of soil, phyllosphere and faeces (Table S1). The Shannon and Simpson index indicated that the soil fungal and bacterial diversity were induced in phyllosphere and faeces contrasted to the soil while fungi and bacteria showed diversity at the lowest level in faeces. The ace and chao index demonstrated the highest fungal richness in the soil than in the other two biotopes. The fungal and bacterial richness also diminished in the faeces.
All the microbiotas showed significant different distributions among the biotopes of soil, phyllosphere and faeces. The microbial communities were visualized via polar coordinate analysis (PCoA) (Fig. 1a, 1b and 1c). The PCoA graphs indicated that distinct clusters were shaped among the biotopes of soil, phyllosphere and faeces for fungal, bacterial and archaeal phyla (Fig. 1a, 1b and 1c). These clusters were confirmed by the sample within-between analysis which consisted of permutational multivariate analysis of variance (ADONIS) and similarity analysis (ANOSIM) (Table 1). These examinations also revealed that bacterial community varied among the three biotopes more significantly than fungi and archaea. The Hierarchical clustering tree at order level, based on bray- curtis distance, further supported these microbial phylogenetic differentiations among the three biotopes (Figure S1).
The network like venn analysis revealed that soil was the potential microbial source reservoir of phyllosphere while faeces was a much less relavant biotope to them. Figure. 2a indicated that phyllosphere shared large amount of OTUs with soil. However, faeces shared much fewer OTUs with the other two biotopes, especially rarely with the soil. The OTUs shared by all the three biotopes also contributed a fraction of low percentage (Fig. 2a). To be precise, the venn analysis of fungi, bacteria and archaea (Fig. 2b, 2c and 2d) showed that the main proportion of shared OTUs are contributed by fungi and bacteria. The 916 and 419 OTUs were shared by soil and phyllosphere for fungi and bacteria respectively while only 26 OTUs for archaea. However, with respect to the OTUs between soil and faeces, only 20 occurred in fungi, and both the OTUs numbers in bacteria and archaea are 0. The OTUs shared among all the three biotopes were also mainly in fungi. The three biotopes communal OTUs numbers are 104, 33 and 6 for fungi, bacteria and archaea successively.
The microbial composition also showed remarkable differentiation among the biotopes of soil, phyllosphere and archaea. For the fungi, samples from the soil biotope were more diverse than those of the phyllophere and faeces biotopes. Ascomycota (62.94%) was the dominant phylum followed with Zygomycota (14.97%) and Basidiomycota (10.01%) in the soil. Ascomycota was also the most predominant phylum in phyllosphere while its relative abundance increased to 89.69% and it was 2.63% and 5.99% for Zygomycota and Basidiomycota respectively. Moreover, Ascomycota (97.36%) almost completely dominated in the faeces fungal microbiome (Fig. 3a). For the bacterial community, the fecal microbiota was not as diverse as the soil and phyllosphere microbiota. The dominant bacterial phyla in the soil and phyllosphere were both Actionobacteria, Fimicutes, Proteobacteria, Bacteroidetes, Acidobacteria, Chloroflexi and Cyanobacteria. Actinobacteria (soil: 44.02%, phyllosphere:34.69%) contributed most in both biotopes. The relative abundance of Bacteroidetes (soil: 0.92%, phyllosphere:10.99%) and Acidobacteria (soil: 20.29%, phyllosphere:1.83%) increased or decreased most acutely from the soil to phyllosphere biotopes. Nevertheless, Firmicutes (61.71%) and Bacteroidetes (24.63%) dominated in the fecal microbiota (Fig. 3b). In the archaeal community, the dominant phyla in the soil and phyllosphere biotopes were Thaumarchaeota while Euryarchaeota dominated in the faeces biotope (Fig. 3c).
To further understand the association of the microbial structure among the soil, phyllosphere and faeces, the fungal, bacterial and archaeal composition of three biotopes were visualized by ternary analysis (Fig. 1d, 1e and 1f). A plenty of fungal classes were shared between the soil and phyllosphere with low abundance. Few fungal classes coexisted between faeces and the other two biotopes. Several fungal classes with a certain amount of relative abundance were communal among three biotopes. Most of these classes belonged to phylum Ascomycota and some classes with low abundance belonged to Basidiomycota. As for bacteria, classes shared between the soil and phyllosphere were also common while much fewer classes occurred between faeces and phyllosphere. Furthermore, there was no bacterial class between the soil and faeces. In the archaeal community, all the classes shared by the soil and phyllosphere belonged to Thaumarchaeota. A few classes belong to Euryarchaeota were shared by faeces and phyllosphere but it was at a low level of abundance in phyllosphere. The archaeal classes did not coexist between soil and faeces either (Fig. 1d, 1e and 1f). The radar analysis and coxcomb analysis of the microbial composition also supported these statements. It was found that Ascomycota and Basidiomycota were communal among three fungal microbiotas but the relative abundance of Basidiomycota was lower than Ascomycota (Fig. 3d and S2). The random forest model further classfied the fungal distribution among the three biotopes. The communal genuses with the top relative abundance such as unclassified Dothideales, Neostagonospora, unclassified Sporopmiaceae, unclassified Xylariales and Pyrenophora all belonged to the fungal lineages within Ascomycota (Fig. 3f).Actinobacteria, Proteobacteria and Acidobacteria were the shared bacterial phyla between the soil and phyllosphere while Bacteroidetes were shared between phyllosphere and faeces. As for archaea, Thaumarchaeota abounded in both soil and phyllosphere biotopes. Euryarchaeota abounded in faeces but it occurred rarely in phyllosphere (Fig. S2).
To identify the influence of faeces input to the soil and phyllosphere microbiota, the microbial community of soil and phyllosphere biotopes from grazed and ungrazed grassland were presented via network analysis. Multiple network topological metrics were employed to reveal the remarkable difference between the microbial assemblages of grazed and ungrazed grassland. Although the co-occurrence networks constructed by significant correlation indicated a larger proportion of bacterial and fungal OTUs were included in the networks of grazed grassland, there were similar edges in these assemblages with grazing (Fig. 4a and 4b). The average weighed degree, which indicated the normalized number of connections to a node, diminished under the grazing conditions. The decreasing cluster coefficient and increasing average path length also exhibited the attenuated connectivity of fungal and bacterial networks in the grazed grassland. Likewise, the margin attenuation of fungal and bacterial networks density under grazing conditions reflected that enormous potential connection between nodes of fungal network lost efficacy in the grazed grassland. Nevertheless, the modularity index indicated that the bacterial and fungal assemblages were more modular in the grazed grassland than ungrazed grassland while it was reverse in the archaeal assemblages (Table S2).
To clarify the influence factor of the microbiota among biotopes of the soil, phyllosphere and faeces, the environmental characteristics were determined (Table S3). The Mantel Test revealed that total nitrogen (TN), pH and temperature influenced all the fungal, bacterial and archaeal microbiota highly significantly and total organic carbon (TOC) also influenced fungi very remarkably (Table S4). Although all the environmental factors such as TOC, TN, total phosphorus (TP), nitrate nitrogen (NO3--N), ammonia nitrogen (NH4+-N), carbon nitrogen ratio (C/N), pH, temperature and moisture could influence the fungal and bacterial communities, there were only TOC, TN, pH and temperature exerting impact on archaeal community. The canonical correspondence analysis (CCA) indicated that there was no predominant influence factor in these environmental variables (Fig. 5). Carbon, nitrogen and phosphorus impacts were strongly and positively correlated while they were negatively correlated to pH impact. Moisture and ammonia nitrogen were in another strongly positive correlated impact pair while they were significantly and negatively correlated to nitrate nitrogen. Notably, soil bacteria were most strongly influenced by nitrate nitrogen and the phyllosphere bacterial community were most correlated to temperature.
The structure equation model (SEM) was also constructed to assess the direct and indirect environmental impacts on the microbiota of the grazed grassland ecosystem which consisted of biotopes of soil, phyllosphere and faeces (Fig. 6). The multigroup modelling approaches were applied to assess the hypothesized relationships among significant environmental physicochemical factors and microbial communities. Temperature directly and strongly influenced fungal and bacterial compositions. Nevertheless, moisture and carbon were weakly correlated to fungal compositions. Nitrogen directly influenced fungal and bacterial compositions, meanwhile, it was impacted by carbon, temperature, moisture, fungal and bacterial compositions. The only factor directly linked to fungal, bacterial and archaeal compositions was pH. The fungal impact on bacterial community was weak while there was strong correlation between bacterial and archaeal community.