Soil horizons regulate bacterial community structure and functions in Dabie Mountain of the East China

Soil bacterial communities regulate nutrient cycling and plant growth in forests. Although these bacterial communities vary with soil nutrients and plant traits, the variation and degree with soil horizons in different forest types remain unclear. Here, bacterial communities of 44 soil samples from organic horizon (O horizon) and mineral horizon (M horizon) of three forest types (Cunninghamia, broad-leaved and Pinus forests) in subtropical forests of Dabie Mountain, China were analyzed based on amplicon sequencing. We assessed the effects of soil horizons and forest types on bacterial communities. The results showed that the bacterial richness and diversity were significantly higher in the O horizon than in the M horizon. Furthermore, the bacterial community composition and functions were also remarkably different between the two soil horizons. Furthermore, forest types could affect bacterial community composition but not for diversity and functions. Moreover, soil organic matter, including the total organic carbon, available phosphorus, total organic nitrogen, available potassium, ammonium nitrogen, and pH were main drivers for bacterial community composition. The results propose robust evidence that soil horizons strongly driven bacterial community composition and diversity, and suggest that microhabitat of soil bacterial communities is important to maintain the stability of forest ecosystem.


Effects of soil horizons on bacterial communities
Bacterial community composition varied according to soil horizons.Similar bacterial community composition was clustered because of the same soil horizon based on Hierarchical clustering analysis.1027 and 997 OTUs were shared in all samples of subplots in three forest types of the two sites, respectively (Fig. 1a,b).The number of shared OTUs of the O and M horizon was 12,459 (31.56%) and 9330 (23.64%) in the two sites, respectively (Fig. 1d).Furthermore, the OTU abundance in the O horizon was higher than that in the M horizon.The soil bacterial richness and Shannon index were higher in the O horizon than in the M horizon in Cunninghamia and broad-leaved forests (Figs. 2).The dominant bacterial communities varied with soil horizons in the three forest types (Fig. 1c), and the two soil horizons shared 560 OTUs (Fig. 1d).For example, the abundance of Chloroflexi and Firmicutes was higher in the M horizon, whereas Proteobacteria and Bacteroidetes were higher in the O horizon (Fig. 1e).Notably, the bacterial community composition in the O horizon based on genus level was distinguished from the M horizon based on the NMDS analysis (Fig. 2a,b).The results of the PERMANOVA and ANOSIM analyses confirmed that bacterial community composition in the two horizons had significant differences in Cunninghamia and broad-leaved forests in TNR (PERMANOVA and ANOSIM, P < 0.05) and in Pinus forest in YNR (Fig. 2c,d).The different biomarkers of the O and M horizon were revealed based on LEfSe analysis (Fig. 2e,f, Figs. 3).Proteobacteria, Alphaproteobacteria, and Rhizobiales were main biomarkers in TO while Deltaproteobacteria, Micrococcaceae, and Arthrobacter were main biomarkers in TM.Planctomycetaceae, Rhodospirillales, and Planctomycetia were main biomarkers in YO while Acidobacteria were main biomarkers in YM.Soil bacterial community composition was markedly different between the two horizons in two sites.The bacterial composition in the O horizon was separated from that in the M horizon along the PCoA axis 1, which explained up to 48.0% and 53.0% of dissimilarity separately in YNR and TNR (Fig. 3a,b).PCoA showed that the O horizon differed from the M horizon, and the results were further proof of the remarkable effect of soil horizons on bacterial diversity.

Effects of forest types on soil communities
Bacterial beta diversity based on OTUs was varied across the three forest types with no significant difference (Fig. 2a-d).Bacterial alpha diversity was also weakly affected by forest types although soil bacterial richness and Shannon indices were the lowest and highest in Pinus and Cunninghamia forests in TNR respectively (P < 0.05) (Figs. 2).However, the soil bacterial community structure in the O horizon in Cunninghamia forest was significantly different from Pinus and broad-leaved forests in TNR (PERMANOVA and ANOSIM, P < 0.05).LEfSe analysis revealed different biomarkers in the three forest types (Figs.2e,f).Key bacterial groups varied with different forest types in the two sites, although groups such as Kofleria, Candidatus Solibacter, Phenylobacterium, Burkholderia, Gaiella, Rhizomicrobium, and Gemmatimonas were common genera in the three forest types.Furthermore, at the phylum level, the abundance of Parcubacteria and Fibrobacteres in the O horizon showed significant differences between Cunninghamia and broad-leaved forests in TNR.The abundance of Actinobacteria in the M horizon was the highest in Cunninghamia forest and the lowest in broad-leaved forest.The abundance of Bacteroidetes in the M horizon was the highest in Cunninghamia forest and the lowest in Pinus forest.At the genus level, Acidipila, Acidobacterium, and Anaeromyxobacter in the O horizon were the highest in Cunninghamia forest and the lowest in broad-leaved forest in YNR, and they showed no significant difference in the M horizon in YNR and in both horizons in TNR.These results suggested that forest types strongly effect soil bacterial community composition but they had no strong influence on bacterial beta-and alpha-diversity.

Effects of soil properties on bacterial communities
The relationships between soil physical and chemical properties and bacterial composition of different samples with RDA analysis (Fig. 3d,e).The content of TOC, AP, TON, AK, and AHN in the O horizon was significantly higher than that in the M horizon.Furthermore, pH was the lowest in the Pinus forest than the Cunninghamia and broad-leaved forest although soil from the three forest types were acid soil (Table 1).The soil organic matter had no significant difference among the three forest types in O horizon, except that AP and TON were the lowest and highest in the Pinus forest in M horizon, respectively.There was positive relationship among TOC, AP, TON, AK, and AHN while they had negative relationship with pH.Meanwhile, TOC, AP, TON, AK, AHN, and pH were significant influences on bacterial community composition.For examples, TOC, AP, TON, AK, AHN, and pH were significantly related to the abundance of Phenylobacterium and Rhizomicrobium at the two www.nature.com/scientificreports/

Functional profiles of bacterial communities
Metabolic functions of bacterial communities were predicted in KEGG pathways using PICRUSt software.The dominant metabolic functions of soil bacterial communities in the study sites were amino acid metabolism, circulatory system, replication and repair, and membrane transport.Hierarchical clustering analysis showed that bacterial community functions varied with soil horizons.The NMDS analysis also showed that bacterial community functions were driven by soil horizons (Fig. 4a,b).Metabolic functions, such as the metabolism of cofactors and vitamins, and membrane transport, were different between the O and M horizons, and they were significantly different in the O and M horizons of Pinus and broad-leaved forests in TNR.The bacterial community functions in the three forest types showed high similarity according to PCoA, especially in the M horizon (Fig. 4c,d).The abundance of biosynthesis of secondary metabolites, glycan biosynthesis and metabolism, carbohydrate metabolism, enzyme families, and transcription were higher in O horizon while the abundance of metabolism of terpenoids and polyketides, cell motility, and metabolism of other amino acids were higher in M horizon (Fig. 4e,f).Bacterial community functions in broad-leaved forest differed from Pinus forest in TNR, whereas they were clustered in YNR according to PCoA.

Discussion
Forest ecosystems provide a broad range of habitats for bacteria, especially abundant in soil and litter.Bacteria play an important role in the transformation of dead plant biomass in litter and soil.Consistent with the first hypothesis, bacterial composition, diversity and functions varied with soil horizons, and the diversity in the O horizon were significantly higher than that in the M horizon.Our results confirmed that soil horizons strongly affect bacterial communities structure 20,23 .The distribution of bacterial communities is consistent with fungal communities in O and M horizon 24 .Soil that is a strongly spatial and temporal variability in biological, physical, and chemical properties, offers variety and complexity habitats for microbial organisms 20 .Soil horizons are an essential soil characteristic and the resultant of vegetation, climate, parent material, organisms, and time 25 .Soil properties in different horizons are main drivers for the changes of microbial community structure 26 .Further research showed that soil bacterial community composition was significantly influenced by soil organic matter content.Moreover, the content of TOC, AP, TON, AK, and AHN were significant higher in the O horizon than in the M horizon.This results highly suggest that TOC, AP, TON, AK, AHN, and pH in soil horizons contribute to construction of bacterial communities.We speculate that bacterial communities usually have high growth rates under nutrient rich microhabitat, such as nitrogen is critical for Proteobacteria 27 .Variations in bacterial communities with soil horizons are sensitive to the change of availability of soil organic matter 28 .The diversity and richness of bacterial communities are decreased with the content decreases of organic matter in the two horizons that strongly provide evidence.Bacterial diversity is significantly associated with soil properties, especially carbon and nitrogen 6 .Bacterial carbon utilization can promote nitrogen availability, and then drive bacterial diversity and community construction 29 .Nitrogen tightly associates with soil organic matter and pH that indirectly affect bacterial communities 30 .This results suggest that the interaction of soil organic matter regulate the change of soil bacterial communities.Notably, soil pH is widely accepted as a main driver of variation in soil bacterial communities by influencing the bioavailability of carbon and nitrogen characteristic.Our study showed that soil pH of Pinus forest was the lowest and the bacterial richness and diversity were also the lowest among the three forests.The results indicate that the changes of soil microbial community composition are strongly correlated with pH 31 , and bacterial diversity is decreased with soil acidity 16 .Moreover, pH was negatively correlated with TOC, TON, AHN, AK, and AP, and strongly affected bacterial community composition according our results.For example, soil pH was significantly correlated with Phenylobacterium and Rhizomicrobium, and highly correlated with the diversity of dominant bacteria, including Acidobacteria, Alphaproteobacteria, Bacteroidetes, and Actinobacteria.These results further indicate that soil pH is a main predictor for bacterial diversity and community composition 17,29 .Studies of the indirect or direct effect by soil organic matter and pH on soil bacterial community composition can help understand the relationships between pH and microbial community.Consistent with the third hypothesis, forest types affected bacterial community composition, and key bacterial groups varied with the three forest types although forest types weakly effect on bacterial diversity and bacterial community functions.In forest, tree species by litter quality, litter decomposition, shading, interception of precipitation, and windbreak directly affect the soil organic horizon and indirectly affect the mineral horizon, and then provide metabolic resources and microhabitat for microorganism 31,32 .Forest type dominated by specific tree species indirectly affect soil bacterial community composition and diversity by the effect on soil chemical properties and plants functional traits 9,30,33 .However, plant diversity and richness cannot drive the bacterial alpha diversity [34][35][36] .Our studies also dominate that forest types mainly effect on bacterial community composition rather than diversity.
The PICRUSt has recently become available for the determination of the metabolic and functional profiles in a broad range of host-associated microbial communities 37 .Soil bacterial communities play an important role in maintenance of ecosystem and sustainability.Our results suggest that the bacterial functional composition of O horizon is significantly different from M horizon.The content of soil organic matter is significant higher in the O horizon than in the M horizon (Table 1).These results indicate that higher soil organic matter could improve the soil bacterial functional community.Furthermore, the different of microhabitat with soil horizon significant effects on bacterial functions, and synergistic interactions among bacterial species and the composition of the bacterial community are important in determining the level of ecosystem functioning 38 .Notably, the functions, such as biosynthesis of secondary metabolites, glycan biosynthesis and metabolism, carbohydrate metabolism, enzyme families, and transcription are enriched in O horizon while the functions, such as metabolism of terpenoids and polyketides, cell motility and other amino acids are enriched in M horizon.Our studies showed that Proteobacteria and Acidobacteria were the dominant soil phyla and higher in O horizon than in M horizon.The two phyla contribute to nitrogen, carbon, and sulfur cycling [39][40][41][42] .Rhizobiales was main biomarker in organic matter according our results.Rhizobiales commonly exert beneficial functions for their hosts by providing various nutrients, phytohormons as well as precursors for essential plant metabolites, such as nitrogen fixing, methanotrophic, microsymbiotic bacteria 43 .Furthermore, the O horizon consist of undecomposed, partially, or highly decomposed litter composed and has high soil organic matter content 25 .The M horizon is characterized by an accumulation of humified organic matter mixed with the mineral fraction 25 .The bacterial activity contributes to decomposition and utilization of litter in O horizon while they might improve environmental adaptation of plant or their own 44 .The similarity functions are showed in the three forest types further indicate that soil organic matter in different horizon is main driver for bacterial communities rather than the type of litter.A limitation of this study is that only the effects of forest types on bacterial community composition were examined.Future research should consider plant species traits, richness and diversity to provide deeper insights into the mechanisms underlying the effects of forest types on the soil bacterial community structure and functions.

Conclusions
This study reports the effects of soil horizons and forest types on bacterial communities in Dabie Mountain, China.The results suggest that dominant bacterial taxa playing different roles in the microenvironment varied with soil horizons and forest types.The diversity, composition and functions of bacterial communities were strongly correlated with soil horizon.Notably, forest types could affect soil bacterial community composition while had weak effect on bacterial diversity.Furthermore, our results highlight that soil organic matter and pH  YO YM e f

Figure 1 .Figure 2 .
Figure 1.Venn diagrams of shared and unique bacterial OTUs in the O and M horizon and three forest types (a) at Tiantangzhai and (b) and Yaoluoping Nature Reserve (d) of Dabie Mountain.(c) Visualization of taxonomic and phylogenetic based on bacterial genus level by GraPhIAn.(e) The heatmap of bacterial community composition based on phylum level.

Figure 4 .
Figure 4.The relative functional abundance of soil bacterial communities based on heatmap of KEGG using hierarchical clattering and unweighted pair group method with arithmetic mean in the two horizons and three forest types (a) at Tiantangzhai and (b) Yaoluoping Nature Reserve.Nonmetric multidimensional scaling ordination (NMDS) of soil bacterial functional communities (c) at Tiantangzhai and (d) Yaoluoping Nature Reserve.Significantly altered bacterial communities between O and M horizon as measured by the response ratio method at the 95% confidence interval (Welch's t-test) at (e) Tiantangzhai and (f) Yaoluoping Nature Reserve.

Table 1 .
Physical and chemical properties of soil at O and A horizon of forest types.TCO, TBO, TPO, TCM, TBM, and TPM represent O and M horizon of three forest types respectively at Tiantangzhai nature reserve.YCO, YBO, YPO, YCM, YBM, and YPM represent O and M horizon of three forest types respectively at Yaoluoping nature reserve.