Brazilian university researchers develop AI to combat Asian soybean rust

Researchers in the Brazilian state of Paraná are developing an Artificial Intelligence (AI) application for agriculture, specifically targeting the monitoring and control of Asian soybean rust. 

Caused by the fungus Phakopsora pachyrhizi, this disease remains Brazil's primary soybean threat and has caused significant crop losses nationwide.

The research is led by Professor Marcelo Giovanetti Canteri from the State University of Londrina's (UEL) Agronomy Department, who is also the Director of the Center for Artificial Intelligence in Agriculture (Ciagro). His team includes researchers from various Paraná institutions focused on agricultural technology development and organizations monitoring more than 200 collectors distributed throughout soybean plantations.

These collectors are "traps" mounted above plants that capture wind-borne fungal spores. While researchers previously needed to inspect traps in person and analyze samples under microscopes, AI now enables remote spore detection from the laboratory through the automatic identification of collected spores.

"Despite all control measures, including annual fallow periods, biological control research, and development of more resistant soybean varieties, the disease returns yearly and requires fungicide applications," Canteri said. On average, products must be applied annually across 40 million hectares.

Edivan Possamai, State Coordinator of IDR-Paraná's Sustainable Grains project, emphasized that approximately 160 collectors are installed annually to aid farmer decision-making. "This work can reduce fungicide applications by 35%, lowering costs for producers and environmental impact," he stated.

The project aims to reduce production costs through more effective monitoring, as other controllable factors also influence disease development. Three aspects are considered: the plant, the climate, and the fungus itself. Ciagro is automating observation of all three factors, accelerating data collection and enabling faster decision-making.

The 2023-2024 harvest monitoring tests analyzed two aspects: climate and pathogen (fungus). Plant monitoring will be added for the next season (2024-2025). Canteri noted that the fungus has increasingly developed fungicide resistance, highlighting the urgent need for new control technologies. While new fungicides are being developed, AI can indicate optimal application timing and whether treatment is warranted based on regional characteristics.

Climate data represents a crucial research component. Previously taking up to two weeks to process, this information is now available within two days. This improves the effectiveness of virtual alerts issued by Embrapa's Anti-Rust Consortium, which provides nationwide rust dispersion information. Planting timing is also critical: late planting means the fungus arrives when plants are less developed, increasing loss risk and necessitating more fungicide applications and costs.

"These legislative measures are crucial for Asian Soybean Rust Management, as they form the foundation for reducing disease inoculum pressure. Additionally, a more restricted planting window can lead to reduced fungicide applications for rust control, resulting in less pressure on active ingredients used for field disease management," explained Marcílio Araújo, Head of Agricultural and Forest Pest Surveillance and Prevention at Adapar (Paraná Agricultural Defense Agency).

(Editing by Leonardo Gottems, reporter for AgroPages)