Mapping soil organic carbon as soil health indicator of cocoa landscapes in middle Côte d’Ivoire

The spatial distribution of soil organic carbon (SOC) in cocoa-growing landscapes is a vital indicator of soil health and essential for environmental management. However, documented work on the mapping of SOC is often unavailable for such landscape. At the same time, spatial information is required for specific soil fertility management and the implementation of knowledge-based policies that aim at increasing cocoa production. This study aimed to establish baseline SOC maps in cocoa-growing landscape for topsoil (0 - 20 cm) and subsoil (20 - 50 cm) and investigate underlying factors for its spatial distribution using a digital soil mapping approach. For that purpose, two variants of the Random Forest algorithm were used, considering the normal random cross-validation (random CV) and the Leave-Location-Out cross-validation (LLOCV), which accounts for spatial autocorrelation. Our findings revealed that SOC content was better predicted by considering LLOCV compared to random CV. The spatial variability of SOC content was primarily related to elevation at all depths, with areas with more natural vegetation located at high altitude having higher SOC content than lower regions. However, both methods indicate that SOC content in the studied cocoa-growing landscapes is less than 2%, which is below the standard required for supporting a sustainable cocoa production. These results put forward the need to increase the awareness among all cocoa stakeholders and policymakers towards actions that promote the building up of soil organic matter as a cost-effective way to boost productivity.