Análise multidimensional dos componentes edáfico e arbóreo em sistemas integrados de produção

Abstract

The increasing global demand for food drives deforestation of natural areas, converting them into agricultural and livestock production systems, thereby increasing greenhouse gas (GHG) emissions and accelerating ongoing climate change. Therefore, production models that reconcile food production with environmental conservation need to be implemented. This study evaluated integrated crop-livestock-forestry (ICLF) and livestock-forestry (ILF) systems, as well as forest restoration, as land use strategies to improve improve soil chemical and physical parameters and mitigate GHG emissions. Additionally, a methodology for monitoring forest metrics through remote sensing was assessed. The first chapter evaluated the physical and chemical attributes of soil in ICLF, ILF, forest restoration, degraded pasture, and forest remnants. We observed that systems incorporating a tree component improved soil quality by increasing organic matter (OM), cation exchange capacity (CEC), carbon content (C), and reducing soil bulk density. In the second chapter, we analyzed forest structure, as well as soil and aboveground biomass (AGB) carbon stocks, in areas under ICLF, ILF, forest restoration, and forest remnants. Integrated systems exhibited larger trees with lower stratification, whereas restoration and forest remnants showed higher tree density, greater stratification, and higher basal area. AGB C stocks were similar across areas, except for ICLF-N and ICLF-G. Forest remnants and ILF systems had the highest soil carbon stocks. Total C stocks were comparable among ILF, restoration, and forest remnants, with approximately 60% stored in the soil and 40% in AGB. The third chapter involved the validation of remotely piloted aircraft (RPA) and digital aerial photogrammetry (DAP) for estimating forest metrics. Field data on diameter at breast height (DBH), tree height, basal area, tree density, and AGB were compared with three-dimensional models generated from orthomosaics and point clouds, using multiple linear regression models. After model selection, we spatialized the variables of interest across the entire study area. The digital terrain model accurately represented the area's topography, contributing to model quality. The predictive model for DBH showed the lowest R² along with the lowest RMSE, while the height model performed slightly better. Basal area and tree density yielded similar results. The predictive model for AGB demonstrated satisfactory R² and RMSE values for estimating AGB and forest metrics. Therefore, integrated production systems contribute to soil quality and C stocks. Systems such as ICLF and ILF, along with forest restoration, are capable of sequestering amounts of C from AGB comparable to those in forest remnants, making them effective strategies for GHG’s emission mitigation. Lastly, we found that DAP-related methods can be applied in integrated production systems, facilitating better access to carbon stock estimates and monitoring for rural producers.