Abstract :
[en] In a broad sense, soil salinity relates to high levels of soluble salts in the soil solution phase and soil sodicity refers to an excess of sodium in the exchangeable complex, while alkalinity indicates the dominance of alkaline salts and high pH. Salt-affected soils are mainly caused by natural conditions and/or anthropogenic activities and negatively affect plant growth and soil-water properties. The High Valley of Cochabamba - Bolivia is characterized by low soil and crop productivity, and land degradation primarily due to salinization processes, which in turn, are driven by semiarid conditions, population increase, deforestation, and inadequate agricultural practices. Some studies have been conducted primarily focused on mapping and characterizing salt-affected soils in this region, but there are still gaps in soil information, prediction tools, and amelioration techniques for their proper management. Therefore, this study aimed to contribute to the sustainable management and rehabilitation of salt-affected soils in the High Valley through baseline soil information, salinity/sodicity prediction models, and insights into amendment-based remediation techniques.
Regarding the characterization and classification of soil samples and profiles, the saline-sodic and saline classes dominate among the salt-affected soil samples, and most salt-affected soil profiles’ horizons showed high levels of salinity and sodicity. The alternative classification approach can overcome the confusion caused by the – USSL – saline-sodic soil class by considering the nature of soluble ions; in this context, some differences between the two methods, for salinity and sodicity distributions were observed. The spatial interpolation was unsatisfactory due to insufficient spatial correlation. Incorporating additional soil profiles and samples might improve the representativeness of the soil information, spatial prediction, and classification system.
Concerning the performance evaluation of machine learning models to predict soil salinity/sodicity variables, random forests (RF) and support vector machines (SVM) regressions outperformed the partial least squares algorithm in estimating soil ESP and ECe, as well as for predicting salt-affected soil classes. Multivariate regressions predicting soil ESP as a function of EC, SAR, and pH showed relatively good performance, somewhat similar to simple regression predicting ESP from SAR. The models to predict soil ESP and EC from remote sensing-based and geomorphometric features showed relatively low performance. Overall, these models might contribute to the monitoring and management of salt-affected soils in the High Valley; however, validations with additional samples and predictor variables are essential to improve their accuracy.
According to the first soil-column experiment assessing the effectiveness of individual mineral and organic amendments with leaching in remediating saline-sodic soils, gypsum was more effective than sulphur, while cattle/chicken manure was better than biochar and peat in lowering soil ESP, and any organic or mineral amendment was as efficient as water alone in decreasing soil ECe. The superiority of gypsum was mainly due to its Ca2+ content which displaces exchangeable Na+, while that of manure was probably due to its contribution of organic matter and divalent cations, which also improve soil-water properties. The second soil-column experiment evaluating the combined effect of manures and gypsum showed that either cattle or chicken manure together with gypsum at any dose was more effective than gypsum alone in reducing the soil ESP to below 5%; furthermore, except for water alone, all treatments were effective in lowering the soil ECe to below 1.6 dS m−1, and any combination was effective in decreasing soil pH to below 8.7. Thus, the effectiveness of manure combined with gypsum was mainly due to their synergistic effect on adsorbed Na+ displacement and soil structure improvement. The addition of manure might enhance and hasten the effect of gypsum with leaching in ameliorating saline-sodic/sodic soils. Further validation of the most effective amendment-based remediation techniques through field experiments is recommended, and alternative approaches such as biosaline agriculture and phytoremediation should also be explored.
In sum, the proper management and rehabilitation of salt-affected soils in the High Valley of Cochabamba relies on adequate characterization, correct classification, accurate estimation, and effective amelioration of these soils; consequently, this study contributes to these goals by providing: (1) comprehensive baseline soil information, (2) tailored prediction and classification tools, and (3) insights into amendment-based remediation techniques, all of which are subject to further refinement.
