Indicadores da saúde do solo em sistemas integrados lavoura-pecuária-floresta: atividades das enzimas β-glicosidase e arilsulfatase e modelagem preditiva por Infravermelho Próximo (NIRS)

Abstract

Soil organic carbon is one of the main components associated with the quality and functionality of terrestrial ecosystems, directly influencing microbial activity, nutrient cycling, and the stability of organic matter. In this context, soil enzymes such as β-glycosidase (BG) and arylsulfatase (ARYL) stand out as bioindicators sensitive to management-induced changes, as they are related, respectively, to the carbon and sulfur cycles. The objective of this study was to evaluate biochemical, spectroscopic, and soil carbon indicators in long-term integrated crop-livestock-forestry (SILPF) systems conducted at Embrapa Pecuária Sudeste in São Carlos, São Paulo. A total of 114 soil samples were analyzed, collected from the 0–10 cm layer in ILPF systems with different variations, including: the complete system, with all components, that is, integrated crop-livestock-forestry (ILPF); livestock-forestry (IPF) system; crop-livestock (ILP) system; intensive pasture (INT); extensive pasture (EXT); and native forest (FLO), used as a reference area. BG and ARYL activities were determined chemically by measuring the release of p-nitrophenol; total carbon was determined by CHN elemental analysis; the humification index of soil organic matter (HLIFS) was assessed by laser-induced fluorescence; and the average specific enzyme activity (ASEA) was calculated from enzyme activities normalized by soil carbon. In addition, near-infrared (NIRS) spectra were obtained to assess the technique’s potential for estimating biochemical attributes and soil carbon. The highest BG activity was observed in the ILP system, with a mean of 155.91 μg pNP g⁻¹ soil h⁻¹, followed by the INT system, with 105.64 μg pNP g⁻¹ soil h⁻¹, while the EXT system had the lowest value, at 63.17 μg pNP g⁻¹ soil h⁻¹. For ARYL, the FLO treatment showed the highest mean value, at 745.5 μg pNP g⁻¹ soil h⁻¹, followed by the INT treatment, at 475.8 μg pNP g⁻¹ soil h⁻¹, while the EXT treatment showed low activity, at 191.3 μg pNP g⁻¹ soil h⁻¹. In general, systems with a tree component exhibited lower enzymatic activities compared to ILP, INT, and FLO. ASEA was highest in FLO (13.22), followed by INT (11.71) and ILP (8.83), indicating higher enzymatic activity relative to available carbon in these systems. FLO had the lowest HLIFS (1350.1), while the EXT pasture had the highest HLIFS (3935.4), highlighting differences in the degree of organic matter transformation among the evaluated systems. The models obtained by NIRS showed an R² of 0.39 and an RMSECV of 22.42 for BG, and an R² of 0.72 and an RMSECV of 83.8 for ARYL, indicating better predictive performance for arylsulfatase. Overall, the integration of enzymes, total carbon, HLIFS, ASEA, and NIRS allowed for a more comprehensive assessment of soil health, indicating that conservation-based management systems with higher plant biomass favor soil biochemical quality.