Grain losses pose a threat to agricultural sustainability, particularly in developing countries. Mechanized harvesting is a key process in which losses occur; thus, its investigation is essential. Therefore, a systematic and meta-analytical review was conducted to encompass studies on quantitative soybean losses during harvesting in South America from the last two decades. The initial search yielded 1,094 scientific articles; however, only 25 studies met the inclusion criteria. This study provides an overview of technical aspects monitored on farms, guidelines to ensure efficient harvesting and literature gaps for further innovations. Because studies on soybean losses were predominantly based on continuous observational data and lacked methodological quality according to a quality scoring, only four studies were included in the meta-analysis. The meta-analysis results showed a significant relation between losses and the combine age, indicating that combines with over 10 years of use result in higher losses than newer ones (p ≤ 0.05). Conversely, operating at speeds exceeding 5 km h^-1 did not lead to significant increases in soybean losses (p > 0.05). Many decisions can be taken at the farm-level to reduce losses, such as proper training of workers and adequation of combine harvesters according to each crop condition. The insights described here are timely for paving the way towards innovation in harvesting systems and minimizing grain losses by understanding the critical points within the context improving yields during soybean harvesting.

Remote Sensing enables the acquisition of planting information through the spatial resolution of multispectral images obtained from Remotely Piloted Aircraft (RPA) in situ. This study aimed to evaluate the application of agricultural gypsum in citrus orchards using precision agriculture tools and image-based analysis. Multispectral images were acquired at three flight altitudes (80, 100, and 120 m) to assess their influence on canopy area delineation and the Normalized Difference Vegetation Index (NDVI). The experimental design consisted of five gypsum application rates (0.0, 0.6, 1.2, 1.8, and 2.4 t ha⁻ ¹ ) applied to a citrus orchard located in Capitão Poço, Pará, Brazil. The variables analyzed were NDVI – derived from near-infrared and red reflectance bands – and canopy area, extracted from vector polygons of tree crowns. The results showed that the 0.6 t ha⁻ ¹ dose yielded the greatest canopy area. Additionally, flights at 80 m altitude provided higher spatial resolution, enabling more precise delineation of citrus canopies. This enhanced detail supports improved vegetation monitoring, emphasizing the advantage of higher spatial resolution in precision agriculture.