Biological invasions – a process in which exotic species establish themselves and spread in natural ecosystems, competing with native ones and altering the environmental balance – are considered the second main cause of loss of water in the world, making the development of efficient methods for their detection and control urgent.
To face this challenge, a study carried out by researchers from the Lagoa do Sino campus of the Federal University of São Carlos (UFSCar) and published in the Journal for Nature Conservation presented a solution for diagnosing these invasions in terrestrial ecosystems, combining drone remote sensing techniques with deep learning algorithms, involving artificial intelligence (AI).
Developed in an area of 92 hectares of humid field in the transition between Cerrado and Atlantic Forest, in the municipality of Buri (SP), the study applied AI algorithms to detect the presence of the invasive Pinus elliottiia species of pine from the southern region of the United States, used for the production of wood and cellulose. Outside its original cultivation areas, it spreads rapidly and alters the soil and fire regime, compromising the diversity of species native to Brazil.
The research is developed by Giovanna de Andrade Ferreira, a student at UFSCar, with support from FAPESP and guidance from Paulo Guilherme Molin, also from UFSCar.
“The technology we developed has the potential to revolutionize the way we combat biological invasions”, highlighted Ferreira to the UFSCar Social Communication Coordination.
Practical application
The success of the methodology aroused the interest of the company Bracell, which, in partnership with the Forestry Foundation of the State of São Paulo, requested its application at the Itapeva Ecological Station (SP), an area affected by the pine invasion.
In collaboration with Bioflore, partner of the Geotechnology Research and Extension Center (CePE-Geo) in the international XPrize Rainforest competition, the team diagnosed a 25% coverage of Pinus elliottii in the conservation unit.
“The project demonstrates the importance of our methodology for correctly sizing the problem and its potential impact in terms of public policies,” said Molin.
The method offers a quick, cheap and scalable alternative, contributing to global goals such as the United Nations (UN) Restoration Decade and other issues that will be highlighted at the 30th UN Conference on Climate Change (), which will be held in November, in Belém (PA).
In addition to identifying Pinus elliottiithe team works on developing algorithms to detect other invasive species, such as brachiaria, leucena and acacia, as well as species of high conservation value, such as araucaria, brazilwood, chestnut and açaí.
The article can be read at .
