Electrical conductivity modelling in calcareous soils of the Maltese Islands : developing a soil:water suspension to saturated paste conversion model

Abstract

Soil salinity is a critical factor influencing soil health and agricultural productivity, typically measured through the electrical conductivity (EC) of a saturated soil paste extract (ECe). However, this standard method is time-consuming, and alternative soil: water suspension methods (EC(soil : water)) offer quicker results, though conversion models are often soil-specific. This study aimed to develop models to convert EC (1:1, 1:2, and 1:5) to ECe for Malta's calcareous soils, while also assessing the impact soil texture, carbonate, and organic matter content have on model accuracy. Moreover, the effect these variables have on soil salinity in the field was also investigated. A total of 134 soil samples with varying textures, carbonate, and organic matter levels were analysed. The results indicate a considerable variability in salinity levels among the sampled soils with a mean average of 3772 µS/cm. Out of the different soils analysed, 104 can be classified as non-saline, with an electrical conductivity (e) less than 4,000 µS/cm. This study produced generalized models (ECe = 10(a(logEC(soil))+b)) with strong correlation coefficients (r² = 0.91-0.93, p < .001). Models specific to fine- and medium-textured soils, soils with 35-50 % carbonate content, and those with 2.5-4.2 % organic matter demonstrated even higher accuracy (mean r² = 0.96). Validation using 22 independent samples confirmed the reliability of these models (RMSE = 0.93-0.97 dS m⁻¹; NSE = 0.95-0.97), except for coarse-textured soils. These findings suggest that general models can be reliably applied to most of Malta’s soils, with exceptions for high sand content soils. This methodology was also applied to analyse metal concentrations, and the resulting general model can estimate the concentrations of key metal ions Na⁺, Ca²⁺, Mg²⁺, and K⁺ in the soil saturation extract (ECe) using measurements derived from soil:water suspensions at 1:1, 1:2, and 1:5 ratios. No significant correlation between ECe and the levels of sand, silt, and clay was observed, nor were significant differences in EC found across contrasting textural classes (p = .465). Weak positive correlations were detected between EC and both carbonate (rs = 0.216, p < .05) and OM content (rs = 0.269, p < .05). These results suggest that salinity in these soils is not primarily governed by physical properties like texture. Organic matter levels, when plotted against sand and clay content, showed no significant correlation, indicating that other factors such as soil management practices and environmental factors seem to have a more pronounced impact on soil salinity.