Estimation of the Exchangeable Sodium Percentage from the Sodium Adsorption Ratio for salt-affected soils in the High Valley (Bolivia)

In order to obtain a more cost-time efficient way to predict the Exchangeable Sodium Percentage (ESP) function to Sodium Adsorption Ratio (SAR_e) of salt-affected soils in the High Valley of Cochabamba (Bolivia), two regression models were generated: ESP= 0.972 SAR_e + 1.576 (R²=0.85, P <  0.0005) and ESP= 6.522 SAR_e^0.5 – 5.723 (R²=0.78, P < 0.0005), based on 84 soil samples. The efficiency of the models was evaluated through an independent test set with 18 samples. The predicted ESP values showed a significant relationship with the measured ESP values: (R²=0.69, P < 0.0001) and (R²=0.68, P < 0.0001), respectively, and according to T-test of paired samples were not significantly different (P > 0.05).  Both models are relatively similar in terms of performance and could be recommended to predict ESP from SAR_e in the High Valley. To improve the prediction, additional samples for modelling and data stratification in terms of sodicity might be necessary.