Land Use and Its Impact on Phosphorus Adsorption Capacity in a Tropical Dryland Agricultural Frontier: Insights and Predictive Modeling

• Main Author: de Souza, José Wilker G.
• Publication Date: 2025
• Other Authors: da Luz, João H. S., Silva, Dayane M. R., Silva, Ricardo B., Paulino, Sivaldo S., da S. Marques, Isabelly C. [UNESP], de O. Galdino, Wesley, de Sousa, Jadielson I., Do N. S. Barbosa, Luana, Melo, Alan F., de Souza, Weslly G., da S. Lira, Viviane, dos Santos, Valdevan R.
• Format: Article
• Language: eng
• Source: Repositório Institucional da UNESP
• Download full: http://dx.doi.org/10.1080/00103624.2025.2452995; https://hdl.handle.net/11449/307095
• topic: Adsorption isotherm; phosphate sorption; random forest; SEALBA

Land use significantly impacts the physical-chemical properties of soil, particularly in tropical regions where the maximum phosphorus adsorption capacity (MPAC) plays a crucial role in agricultural sustainability. This study evaluates the influence of land uses (cropland, grassland, and forest) on MPAC in dryland soils of Sergipe, Alagoas, and Bahia (SEALBA) agricultural frontier in Northeast Brazil. Soil samples from nine sites were analyzed for their physical-chemical characteristics (potential of hydrogen (pH), available phosphorus (P), remaining P, soil organic matter (SOM), calcium and magnesium content, exchangeable aluminum, potential acidity and texture) in the 0–20 cm soil depth. MPAC was determined using two isotherm models, and three predictive models were applied. The results revealed that cropland soils exhibit the highest MPAC, contrary to expectations of lower values due to legacy P from P inputs. However, MPAC varied slightly across sites, indicating specific dependencies on soil physical-chemical properties and suggesting that land use may have a more significant influence over the long term. Random forest and decision tree models identified SOM and available P as the most significant predictors of MPAC, yet with these routine soil fertility indicators alone, the maximum explanatory power (R2) was 75% with a precision (RMSE) of 77 mg kg−1 using the random forest model. Consequently, the conversion of forest to cropland can increase MPAC, and predicting MPAC can be a valuable tool for guiding land use and assisting in the formulation of more sustainable agricultural and environmental policies, particularly in dryland agricultural frontiers.