Agricultural practices significantly impact soil microbes, crucial for soil health and sustainable agriculture. This study employed machine learning to analyze prokaryotic and fungal assembly under various agricultural practices. Fertility source emerged as the most influential factor, particularly for fungi, with its effect decreasing with soil depth. Machine learning also revealed how fertility source shapes microbial co-occurrence patterns, leading to fungi-dominated modules sensitive to fertility down to 30 cm. Tillage affected soil microbiomes at shallower depths (0–20 cm), enhancing dispersal and stochastic processes but potentially harming microbial interactions. Cover crop effects were less pronounced and lacked depth-dependent patterns. The study highlights the multifaceted impact of agricultural practices on microbial communities and machine learning's power in overcoming traditional method limitations to provide enhanced insights into microbial assembly and distribution in agricultural soils.

Yujie Mo, Raven Bier, Xiaolin Li, Melinda Daniels, Andrew Smith, Lei Yu, Jinjun Kan