Machine learning models reveal how biochar amendment affects soil microbial communities

Abstract

AbstractThe biochar amendment plays a vital role in maintaining soil health largely due to its effects on soil microbial communities. However, individual cases and the variability in biochar properties are not sufficient to draw universal conclusions. The present study aimed to reveal how the biochar application affects soil microbial communities. Metadata of 525 ITS and 1288 16S rRNA sequencing samples from previous studies were reanalyzed and machine learning models were applied to explore the dynamics of soil microbial communities under biochar amendment. The results showed that biochar considerably changed the soil bacterial and fungal community composition and enhanced the relative abundances of Acidobacteriota, Firmicutes, Basidiomycota, and Mortierellomycota. Biochar enhanced the robustness of the soil microbial community but decreased the interactions between fungi and bacteria. The random forest model combined with tenfold cross-validation were used to predict biomarkers of biochar response, indicating that potentially beneficial microbes, such as Gemmatimonadetes, Microtrichales, Candidatus_Kaiserbacteria, and Pyrinomonadales, were enriched in the soil with biochar amendment, which promoted plant growth and soil nutrient cycling. In addition, the biochar amendment enhanced the ability of bacteria to biosynthesize and led to an increase in fungal nutrient patterns, resulting in an increase in the abundance and diversity of saprophytic fungi that enhance soil nutrient cycling. The machine learning model more accurately revealed how biochar affected soil microbial community than previous independent studies. Our study provides a basis for guiding the reasonable use of biochar in agricultural soil and minimizing its negative effects on soil microecosystem. Graphical Abstract