Multivariate soil–microbial covariation and constrained recognition of fungal-dominated community-state extremes using Evidence- Constrained Residual Fusion
This study demonstrates that multivariate soil physicochemical conditions covary with a fungal-dominated microbial community state in northern China and evaluates the Evidence-Constrained Residual Fusion (ECRF) method, which achieves high balanced accuracy in identifying community extremes using combined soil and spatial predictors, though its incremental performance over a transparent anchor model is modest and dependent on data partitioning.