Predicting microbial responses to changes in soil physical and chemical properties under different land management

Microbial abundance and community structure can be altered directly and indirectly by soil physical and chemical characteristics which, in turn, can be influenced by land use management. This study utilized the cubist model to predict soil microbial communities based on soil properties at different depths and under different agricultural management in Dawson County, Montana, USA. A total of 538 soil samples were collected from three management treatments (control, no-tillage (NT), and no-tillage with livestock grazing in winter (NTLS)) from three depths (0-5, 5-15, and 15-30 cm). Soil physical and chemical properties and total phospholipid fatty acid (PLFA) analysis were used to predict soil biological properties. Root mean square error (RMSE), mean absolute error (MAE), relative error (RE), mean bias error (MBE), and R squared (R-2) were used to assess the performance of predictions. Results showed that the strongest correlation was between the total PLFA and soil microorganisms. Different soil chemical and physical properties were useful to predict soil microbial communities; ammonium-N, phosphorus, potassium, electrical conductivity, pH, organic matter, bulk density, sand, and clay significantly correlated with most soil microorganisms. Results indicated that the cubist algorithm produced promising results to predict most soil microorganism responses to various treatments and depths. However, this model did not perform well when attempting to predict the ratio of bacteria to fungi. The most important variable to predict all soil microorganisms was the total PLFA, with >90 % effectiveness. These results imply that applying pedotransfer functions (PTFs) to predict soil microbial communities in areas with limited soil data and monetary resources shows promise.