Geographical Patterns And Spatial Distribution Prediction Of Soil Microorganisms In Alpine Ecosystem Of Southeast Tibet

Understanding microbial biogeographic pattern and mechanism is important,as soil microbe play a critical role in biogeochemical cycle.Due to the limitation of technological means,the study of microbial biogeography is relatively backward.Recently,the rapid development of molecular biology technology has enabled the quantification of microbial community in natural environment,which greatly prompted the research of microbial biogeography.Numerous studies have proved that soil microbial community has certain geographic distribubtion pattern,breaking through the traditional view of “every thing is everywhere”.Southeast Tibet has the largest primitive forest area in China,and is one of most biologically diverse regions in the world.Even though bearing important ecological functions,the region is uniquely vulnerable to global climate change and anthropogenic disturbance.Thus,studying soil microbe response to different climates and land uses in this region will help to understand the changes in ecological functions under global changes in the Tibetan Plateau.Here,we employed the high-throughput sequencing and gene chip technology to explore the the biogeographic pattern and meahnism of soil bacterial communities under different land uses,elevational gradient,and timberline ecosystem in the Sygera Mountains,Southeast Tibet.Based on microbial community distribution pattern and mechanism discussed above,we took the lead in using Digital Soil Mapping(DSM)theory,Deep Learning(DL)algorithm,and big data parallel computing technology to predict and map the soil bacterial community distribution at 90 m resolution.We found that:(1)Diversity of soil bacterial community in cropland system was significantly higher than that in forest and grassland ecosystems.On average,1728 species were detected at OTU classification level in cropland ecosystem,while only 1151 and 1241 species were detected in forest and grassland,respectively.The Shannon Index of soil bacterial community in cropland(mean,6.64)was also larger than that in forest(mean,5.91)and grassland(mean,5.85).The representative sequence of OTU was annotated and divided into 45 phyla and 323 genera,Acidobacteria and Proteobacteria were the dominant taxa of the three ecosystems,with an average cumulative relative abundance of 50 %.Bacterial taxa that preferentially inhabit neutral or weak alkaline soil environments,such as Actinobacteria,Gp4 and Gp6 of Acidobacteria,were significantly more abundant in cropland(11.2 %,7.1 % and 4.3 %)than that in forest and grassland(5.0 %,4.2 % and 2.1%).We observed a unimodal distribution of bacterial species diversity along the elevation gradient in forest,the largest value observed at the mid-elevation(3800 m).No uniform pattern in the diversity trend of global mountain systems is apparent as various patterns were observed.However,it is clear that in a fixed ecological environment,an apparent community composition pattern along the elevation exists,demonstrating the importance of bacterial climate adaptation in a bacterial niche.Although climatic variables,especially temperature,strongly affects bacterial community structure,this natural effect is surpassed by the effect of land use.The soil pH was the primary edaphic property that regulated bacterial community composition across the different land uses.Additionally,pH was the main factor distinguishing bacterial communities in managed soils(i.e.,cropland)from the communities in the natural environments(i.e.,forests and grassland).(2)In timberline ecosystem,the diversity of soil bacterial community in shrubland was significantly larger than that in coniferous forest.At the taxonomic level,the Shannon Index of soil bacterial community in shrubland(mean,6.38)was larger than that in coniferous forest(mean,5.75).At the functional level,an average of 22,510 functional genes were detected in shrubland ecosystem,while only 19,600 functional genes were detected in coniferous forest.There were apparent difference in bacterial community composition and function between two forest types.Average abundance of the eutrophic bacteria including Betaproteobacteria and Gemmatimonadetes in the shrubland(6.0 % and 6.6 %)were larger than that in coniferous forest(3.9 % and 3.5 %).In contrast,average abundance of the oligotrophic bacteria including Acidobacteria,Planctomycetes and Chloroflexi in the coniferous forest(26.4 %,16.6 % and 8.6 %)were larger than that in shrubland(22.7 %,7.4 % and 3.6 %).The intensity of genes encoding enzymes for degradation of unstable organic carbon including monosaccharide,disaccharide and starch were higher in shrubland than in coniferous forest,while the intensity of genes encoding enzymes for degradation of recalcitrant organic carbon including lipids and aromatics were larger in coniferous forest than in shrubland.Therefore,shrubland was more efficient in utilizing recalcitrant organic carbon,while coniferous forest had higher efficiency in utilization of stable organic carbon.Soil temperature and C/N were the dominant factors controlling bacterial community composition and function in timberline ecosystem.Using empirical models such as multi-model inference,we discussed the importance of soil carbon balance driven by microbes in timberline ecosystem.Results showed that bacterial communitiy taxa including Betaproteobacteria and Planctomycetes and functional genes including glucose oxidase coding gene and manganese peroxidase coding gene mnp still exert unique role in explaining soil carbon variation in timberline ecosystem after accounting for various environmental variables such as soil attributes,vegetation,climate,and topographic conditions.Coupling these key microbial variables into the Earth System Model will likely improve the predictive power of the carbon feedback process of terrestrial ecosystem.(3)Based on the DSM theory,we predicted and mapped the distribution of dominant phyla abundance and diversity of soil bacterial community at 90-m spatial resolution in Southeast Tibet by DL modeling with multiple soil-envriornment factors.The R-square of independently validated models ranged between 0.32-0.60.The model performance was generally better than that of the existing ones conventionally based only on soil attributes.Through comparing the mapping results with the biogeographic pattern of soil bacterial community at different land uses,elevational gradient,and timberline ecosystem discussed above,we found mapping results well revealed the spatial and vertical variation of bacterial community.Predictive models could well excavate the relationship between microorganisms and soil-environmental factors,that is,the dominant driving factors such as,soil pH,C/N,and temperature,were the most important predictors in the models.This study provides a new technique for rapid investigation of soil microbial diversity in the area especially at large scale.