Data Sheet 1_Geogenic perspectives on potassium dynamics and plant uptake: insights from natural and submerged conditions across different soil types with machine learning predictions.docx

Four different soil types including red, alluvial, calcareous, and black soils along with rice cultivated on them were collected from various parts of India and analyzed for potassium dynamics in the soil plant continuum. Soil potassium (K) dynamics were studied under submerged and non-submerged conditions, and potassium content was analyzed in rice roots, shoots, and grains, along with other soil properties. Red (S1: 5.9) and alluvial (S5: 5.16) soils were moderately acidic, while black (S8: 8.01) and calcareous (S7: 8.1) soils were alkaline. Black soil (S8) had the highest cation exchange capacity (CEC: 31.25 cmol (p+)/kg) and clay content (41.2%), while alluvial soil had the most organic carbon (S5: 1.74%). Submerged conditions enhanced potassium availability, with red soil showing the highest levels of water-soluble K (WsK), exchangeable K (ExK), and non-exchangeable K (NEK), particularly Step-K and constant rate K (CR-K) forms. Rice potassium content was highest in grains, followed by shoots and roots, with red soil containing the most available potassium. A strong correlation was found between soil potassium forms and rice plant potassium uptake. Sensitivity analysis indicated that WsK and ExK from non-submerged soil to be the most favorable forms for potassium uptake, especially in the rice roots and grains. Machine learning models, particularly Random Forest, accurately predicted potassium availability and uptake, highlighting their potential in optimizing soil fertility and advancing precision agriculture for better crop yields and soil health.