The mangroves of Diawling National Park (PND) in Mauritania form an ecosystem of high ecological value that is exposed to severe climatic and anthropogenic pressures. This study analyses the spatial variability of four soil physicochemical properties (electrical conductivity (EC), pH, soil organic carbon (SOC) and total nitrogen (N)) and compares two spatial interpolation methods, inverse distance weighting (IDW) and ordinary kriging (OK). Fifty samples were taken at a depth of 20 cm beneath stands of Avicennia germinans and Rhizophora racemosa and analysed in the laboratory. The descriptive statistics show pronounced variability for EC (CV = 38.4%) and SOC (CV = 35.9%), whereas pH (CV = 10.2%) and N (CV = 16.4%) are distributed more evenly. The mean organic carbon content (0.680%) falls below the values reported for humid tropical mangroves, a difference that reflects the combined saline and water stress affecting Sahelian mangroves. Cross-validation based on six statistical criteria (ME, RMSE, R, MSE, RMSSE, ASE) confirms the clear advantage of IDW, whose correlation coefficients approach unity (R > 0.999) for all four properties, while OK performs more unevenly, from very weak for pH (R = 0.091) to moderate for EC and SOC (R = 0.647 and 0.445) and good for N (R = 0.989). Semivariogram modelling shows a nugget-to-sill ratio of 0.7569 for EC, denoting weak spatial dependence, and a fitted nugget exceeding the fitted sill for pH, SOC and N (ratios of 5.3158, 2.9924 and 16.3692, respectively), a pattern functionally equivalent to a pure nugget effect; together these results reveal that unstructured short-range variability dominates the spatial signal of all four properties and provide a mechanistic basis for the systematic underperformance of OK. This contrast stems from the strong short-range heterogeneity produced by the redox processes typical of the hydromorphic environment, which weakens geostatistical approaches that depend on structured spatial autocorrelation. The IDW maps display an organised spatial pattern, with a salinity gradient that follows hydraulic connectivity and a mosaic of pH values governed by local redox cycles. These results offer a spatial reference for managing and conserving the PND mangroves and add to the understanding of biogeochemical processes in mangrove ecosystems at the Sahelian margin.