Soil texture and organic matter prediction using Mehlich-3 extractable nutrients

Soil organic matter (SOM) and texture are key properties influencing soil nutrient and water dynamics but are time-consuming procedures for analytical laboratories. Our objective was to evaluate SOM and soil texture predictions using Mehlich-3 nutrients and pH in Arkansas soils. Particle size was determined by the hydrometer method (2- and 8-h readings) and SOM by loss on ignition. Two datasets were used to calibrate clay and sand (n = 409) and SOM (n = 1019) prediction models using simple and multiple regression. Estimated cation exchange capacity was highly correlated with clay, resulting in significant prediction models alone or combined with phosphorus (P); pH and copper (Cu); or pH, sodium (Na), and Cu (R² = 0.84, 0.88, 0.89, and 0.90; p < 0.0001, respectively). Soil nutrients were weakly correlated with sand, resulting in a prediction model with moderate accuracy when using Mehlich-3 P, calcium (Ca), Na, iron (Fe), and manganese (Mn) (R² = 0.49; p < 0.0001). Clay and sand prediction models presented comparable accuracy when validated on a new dataset (n = 103). Predicted sand and clay showed good accuracy in grouping soils into medium (65%) and fine (96%) textural categories but had limited ability to define the coarse-textural group (9%). SOM had moderate goodness-of-fit statistics for calibration and validation datasets using pH, P, Ca, Na, Mn, and zinc (R² = 0.65 and 0.70, respectively; p < 0.0001). Mehlich-3 nutrients can be used to estimate soil texture and assist with crop management decisions, but further research is needed to improve SOM prediction.