The invasive species Hovenia dulcis is considered the main invasive species in the Atlantic Forest, capable of altering environmental conditions at a local scale and provoking profound changes in the composition of the plant community. Combining drone and satellite images can make forest monitoring more efficient, enabling a more targeted and effective response to contain the spread of invasive species. This research aimed to use high-resolution CBERS-4A satellite combined with drone images to detect invasive trees in forested areas of the Atlantic Forest. An object-oriented, supervised automatic classification was performed using the Dzetsaka Classification Tool and the Gaussian Mixture Model method. Additionally, georeferenced orthomosaics obtained by drones, totaling 150 ha, were used to confirm the identification of the invasive species. The entire forest area was surveyed to determine the tree community, where 72 random sample plots, each with a fixed area of 100 m2, were established. The calculated indices, such as the Shannon index (H') = 3.65 and uniformity (J') = 78%, demonstrate that the plant community has a high diversity of species. However, the invasive H. dulcis had the highest number of sampled individuals (146), being the species with the highest relative density (9.14) within the community and the second highest in relative frequency (5.10%), coverage importance value (8.85%), and importance value index (7.60%). The methodology employed to identify the invasive species through satellite, and drone images allowed for rapid and precise data collection and quantification of the invasive species, covering an area of 86.44 ha of the forest fragment, which corroborates the field data.
Keywords: Applied remote sensing; Automatic image classification; Biodiversity of forest fragments; Deciduous seasonal forest; Invasion ecology; Phytosociology.
© 2024. The Author(s), under exclusive licence to Springer Nature Switzerland AG.