A Multi-Tool Statistical Approach for Predicting Soil Electrical Conductivity (ECe): A Case Study of the Manouba Province, North East Tunisia

Soil electrical conductivity (ECe) is a key indicator of soil salinity and sustainability, particularly in semi-arid and arid regions. Estimation of ECe is vital for managing soil salinity and ensuring crops productivity. A multi-tool approach that integrates statistical software, various soil parameters, including Texture, SOC, TN, CEC and EC 1:5, measured, was applied to predict ECe and develop a robust predictive model. The performance of each pedotransfer function was systematically evaluated, demonstrating a significant accuracy: PTFs (1) yielded an R2 value of 0.85, PTFs (2) shows a R2 of 0.71 for Stepwise regression model, another PTFs (3) has developed with an R2 of 0.84, Lasso/Ridge re-gression as PTFs (4) achieved an R2 value of 0.89, lastly the PTFs (5) model provided an R2 of 0.83. Our findings revealed a regional variation in soil salinity, with certain areas showing elevated ECe levels that could potentially affect environment sustainability. Consequently, C-Predictive Regression Tree (C-RT), highlighted the importance of integrating non-linear approaches to improve predictive accuracy, presenting an R2 of 0.77. This research underscores the importance of PTFs in predicting soil salinity from soil properties, especially EC1:5 provides valuable insights for sustainable soil management strategies in Northeastern Tunisia.