Estimating Soil Bacterial Abundance and Diversity in the Southeast Qinghai-Tibet Plateau

Yuanyuan Yang,Qianqian Chen,Wu Yu,Zhou Shi
DOI: https://doi.org/10.1016/j.geoderma.2022.115807
IF: 6.1
2022-01-01
Geoderma
Abstract:Soil bacteria play important functional roles in ecosystems but are challenging to investigate because of time-consuming and costly laboratory analyses. Digital soil mapping (DSM) technology is an emerging and efficient tool for regionalization of soil bacteria. It has been introduced as a viable complementary approach to the traditional methods for expression of the spatial distribution of soil properties due to being fast and cost-effective. This study was conducted to develop a strategy for mapping the relative abundance of the dominant phyla and community diversities of bacteria to have a better understanding of their biogeography in the highly heterogeneous area of Southeast Tibet, China. Here, we developed state-factor models using predictor variables that were already mapped and publicly available soil and environmental proxies for edaphic, climatic, biotic and topographic factors. We evaluated seven statistical and machine learning algorithms; namely, partial least squares regression (PLSR), random forest (RF), Cubist, support vector machines (SVM), Gaussian process regression (GPR), XG-boost (XGB) and convolutional neural networks (CNNs). Ten-fold cross-validation with all observations revealed that the CNNs outperformed other algorithms and could explain between 48% and 72% of the variation in bacterial abundance and diversity. Estimates of the relative abundance of Actinobacteria and Proteobacteria produced the largest R-2 values (& GE;0.70), while estimates of Acidobacteria, Gemmatimonadetes, Chloroflexi, and Planctomycetes produced values of 0.6 < R-2 < 0.7, and estimates of Verrucomicrobia, Bacteroidetes, Nitrospirae, OTUs, and Shannon diversity produced values of 0.5 < R-2 < 0.6. The estimates for Firmicutes were poorest, with R-2 values < 0.5. The estimated phyla abundances and diversities clearly exhibited regional patterns and local characteristics. Soil total nitrogen (TN), carbon to nitrogen ratio (C/N), pH, clay content, and temperature were prominent controls that regulated bacterial community distribution.
What problem does this paper attempt to address?