Modelagem não linear da dinâmica do carbono em solo tratado com lodo de curtume

Abstract

In 2020, the Brazilian commercial cattle herd was the largest in the world, representing 14.3% of the world herd, with 217 million heads. Although this activity produces significant profits, contributing to the country’s economic and social development, it has been the target of concerns, mainly due to the large production of residues/effluents observed during the processing of bovine hide. Tanning is the processing of the animal’s raw skin, which consists of transforming it into leather. During the tanning process, it is estimated that each processed hide produces about 15 kg of waste, of which 7.5 kg settles at the bottom of the primary treatment pond in the form of sludge. Tannery sludge is a potential agricultural fertilizer, as it is rich in several nutrients for vegetables, however it is restricted, mainly due to the presence of heavy metals in high concentrations. When an organic waste is added to the soil, microbial activity is stimulated and microorganisms use carbon as an energy source for their metabolic processes, and thus, part of the carbon is incorporated into the microbial cells and part is released in the form of dioxide of carbon (CO2). Mineralization is very important, as the decomposition process of these organic residues also releases nutrients that are essential for plant growth. The carbon release dynamics can be described by nonlinear models. Thus, the present work aims to model the mineralization of organic carbon in the soil at doses of 6, 12, 24 and 36 milligrams per hectare of tannery sludge using the nonlinear models Stanford & Smith, Cabrera and Juma. Soil samples with a very clayey texture were used: Typical Eutroferric Red Nitosol (Nvef), the mineralized carbon was measured in 21 observations over time until the 105th day of incubation. Nonlinear models were adjusted considering the structure of first-order autoregressive errors (AR1) and heteroscedasticity of variances when necessary. Parameter estimation was performed using the Generalized Least Squares method. The fits were compared using the following selection criteria: Corrected akaike information criterion (AICc) and the adjusted determination coefficient (R2a j). The nonlinear Cabrera and Juma models showed good fits. Although the Stanford & Smith is the most used in the description of carbon mineralization from organic residues, in this work it did not present good fits. The Cabrera model showed the best fits for doses 6, 12 and 24 and the Juma for dose 36 based on the selection criteria used.