Assessment of carbon use efficiency in a livestock-forest integration system in eastern Maranhão: an approach with ARP remote sensing

The quantification of carbon dioxide flux (CO₂ Flux) in agricultural systems is essential for understanding the carbon balance and its relationship with sustainable management practices. Remote sensing, through spectral indices, allows for vegetation monitoring and the analysis of CO₂ fixation and release. In Livestock-Forest Integration (LFI) systems, this approach helps define more efficient and sustainable agricultural management this study aims to evaluate the efficiency of carbon use in an LFI system in eastern Maranhão. Remote sensing techniques were used through multispectral images obtained via UAV in July 2024 over the coordinates 3.69939° S and 42.92656° W. The objective was to identify the most representative indices for estimating CO₂ Flux variability. Monitoring was conducted using a Matrice 350 RTK Remotely Piloted Aircraft (UAV) equipped with a Micasense RedEdge-P Blue multispectral camera, which captures data in six spectral bands with a pixel size of 3.45 x 3.45 μm. The images were processed using Metashape software to generate orthomosaics, and spatial analysis was performed in QGIS using zonal statistics to extract the mean spectral band values within 4-meter buffers. The NDVI, NDRE, GNDVI, EVI, and SAVI indices were calculated in Excel, and their relationships with CO₂ Flux were analyzed using Pearson correlation and descriptive statistics. The normality of the data was verified using the Shapiro-Wilk test. NDVI, NDRE, and GNDVI showed the highest correlations with CO₂ Flux (r = 0.95), indicating that areas with greater vegetative vigor have a higher capacity for CO₂ fixation. The observed mean values were: NDVI (0.65 ± 0.08), NDRE (0.30 ± 0.05), GNDVI (0.65 ± 0.07), EVI (0.75 ± 0.15), and CO₂ Flux (0.22 ± 0.04). The Shapiro-Wilk test indicated normality for NDVI, NDRE, GNDVI, and EVI (p > 0.05), while DVI, SAVI, and sPRI showed significant deviations (p < 0.05). CO₂ Flux (p = 0.0790) was close to the normality threshold. GNDVI, NDRE, and NDVI were the most representative indices for estimating CO₂ Flux variation in LFI systems. The adopted methodology proved efficient for monitoring the carbon balance. The results reinforce the importance of using remote sensing to enhance agricultural practices aimed at carbon sequestration, contributing to sustainable management strategies aligned with Sustainable Development Goals (SDG) 13 (Climate Action) and 15 (Terrestrial Life).