Soil Properties Related to Phosphorus Sorption and Prediction of Maximum Phosphorus Adsorption Capacity of Soils from Southern Brazil

Purpose: This study aimed to assess the relationship of soil chemical and mineralogical properties with P sorption and develop a maximum phosphorus adsorption capacity (MPAC) prediction model for Brazilian subtropical soils. Methods: The data were analyzed using descriptive statistics and Pearson’s correlation analysis. The following prediction models were tested: multiple linear regression, multiple linear regression with interaction, elastic net regression, and partial least squares regression. Model quality was assessed using mean squared error (MSE), root mean squared error (RMSE), and coefficient of determination. Results: MPAC was positively correlated with clay content and cation-exchange capacity (CEC), which were negatively correlated with remaining P (P_rem). Soil samples collected in Rio Grande do Sul State showed stronger correlations between MPAC and P_rem in the highly and poorly crystalline aluminum fractions than in the respective iron fractions. The low-crystallinity aluminum fraction had a stronger correlation with CEC than the low-crystallinity iron fraction. The multiple linear regression model with interaction, which includes soil properties easily analyzed in routine laboratories, was the most efficient in predicting MPAC, with MSE and RMSE values of 169.20 and 13.01 mg kg^− 1, respectively. Conclusions: Clay content and CEC were the soil chemical attributes most strongly associated with MPAC and P_rem. The proposed model may be a valuable tool for estimating MPAC in Brazilian subtropical soils.