Baseline high-resolution maps of soil nutrients in Morocco to support sustainable agriculture.

Abstract

Phosphorus (P) and potassium (K) are essential fertilizer elements whose recommendations require soil analysis tailored to crop-specific needs. However, in developing countries like Morocco, access to open databases on nutrient bioavailability remains limited, hindering data-driven agricultural strategies. This paper presents the first national reference maps of available P and exchangeable K at 250 m resolution over Morocco's croplands using digital soil mapping with machine learning algorithms and environmental covariates. Unlike previous efforts employing traditional interpolation methods, these maps were developed using Random Forest by integrating 5,276 soil samples for P and 6,978 for K with 76 environmental covariates representing climate, topography, vegetation, and parent material. Model validation using independent test datasets demonstrated strong performance, with R² values of 0.78 for P and 0.80 for K. Uncertainty assessment through bootstrapping confirmed prediction stability across diverse agricultural landscapes. These baseline maps enhance fertilizer recommendations, promote precision farming, and support agricultural sustainability. The maps are freely available through open-access repository, enabling evidence-based decision-making for researchers, practitioners, and policymakers to improve nutrient management efficiency.