Bayesian Inference of Soil Traits from Green Manure Fields in a Tropical Sandy Soil

Abstract

Green manure represents a crucial soil management practice for soil traits and potentially sequestering organic carbon (OC) within the soil profile. Understanding the biomass dynamics of Fabaceae and Poaceae plants has become essential for refining existing models of soil fertility and organic carbon. In this work, we have developed six models of the fertility and carbon stock of the soil collected from the plots where ten plant species were cultivated as green manuring crop. Two of them are named aboveground and belowground biomass models which use specific biomass production parameters α and adapted to both Fabaceae and Poaceae stands, and three other models are named Ca²⁺, K⁺ and P models adapted to green manure stands. The last one is named soil organic carbon stock model, and it is adapted to both Fabaceae and Poaceae stands. A Bayesian inference was carried out to determine parameters values according to the 6-years field experiment database. The highest significant values for SOC stock, aboveground biomass model, and belowground biomass modes were found for the Fabaceae model with 9.99 t ha⁻¹, 5.37 t ha⁻¹, and 0.61 g cm⁻³, respectively. All proposed models into this study (density, soil, and biomass models) were explained by the geometric reliability index (GRI) and efficiency factor (EF) with a more dispersive fitting. This study underscores the importance of considering adapted models from the Fabaceae and Poaceae families, particularly those with high growth rate index (GRI) and efficiency factor (EF). We found that shoot dry biomass exhibited a polynomial decrease, whereas root density showed an exponential decrease over time for both Poaceae and Fabaceae plants. Furthermore, our study revealed that long-term cultivation of cover crops with green manure significantly augmented the contents of exchangeable cations (Ca²⁺ and K⁺) as well as soil organic carbon (SOC) stock.