Spatial variability of chemical parameters based on predictive models in cocoa-producing soils in Antioquia, Colombia

The evaluation of soil chemical parameters is a fundamental element in the sustainable planning of cocoa cultivation. Lack of knowledge about soil properties leads to the inappropriate use of amendments and fertilizers, which increases production costs and generates negative environmental impacts. The objective of this study was to describe the spatial variability of cation exchange capacity (CEC), total nitrogen (TN), and organic matter (OM) in soils intended for cocoa production in Antioquia, Colombia, based on predictive models. A total of 439 soil samples were collected from eight cocoa-producing subregions and analyzed using wet chemistry. Data were fitted using generalized additive models (GAM) with Poisson distribution, considering longitude, latitude, altitude, pH, texture, and sand and clay content as covariates. Spatial distribution maps were generated for each parameter using the predicted values. Results showed that most soils had CEC values below 5 cmol.kg⁻¹, OM contents around 2.5%, and TN levels near 0.30%. The subregions with the highest cocoa production recorded the lowest values for the evaluated variables. It is concluded that cocoa-growing soils in Antioquia exhibit chemical deficiencies, especially in areas with greater production, highlighting the need to implement fertilization and amendment programs to improve soil quality and ensure the productive sustainability of cocoa systems.