Comparison of nonlinear models for the description of carbon mineralization in degraded pasture soil and in soils with plant cover

Abstract

Tree planting is an important way to restore degraded areas, however, the quality of the plant residue added to the soil influences the organic matter decomposition rate and, consequently, carbon availability. Carbon mineralization curves over time make it possible to understand the decomposition of organic residues and improve soil management. Nonlinear regression models have been used to describe the dynamics of carbon mineralization over time, as they summarize the information contained in the data in just a few parameters with practical interpretations. Thus, this study aimed at evaluating the nonlinear models Cabrera, Juma and Stanford & Smith to describe the soil carbon mineralization in the following plantations: Secondary forest, Acacia auriculiformis, Mimosa caesalpiniifolia and Pasture, obtained from the first to the twentieth week. All the computational part involved in the adjustments and analyses was performed using the R statistical software. The most suitable regression model was selected for the description of soil carbon mineralization for each vegetation cover based on the following criteria: adjusted coefficient of determination (R2adj), residual standard deviation (RSD) and Akaike information criterion (AIC). For Acacia, the Cabrera model was indicated as the best to describe this treatment. For Forest and Pasture, the Juma model had the best fit, and the Stanford & Smith model best described the Mimosa treatment.