Study Examines Aflatoxin Contamination in Texas Corn Crops

Arlington, TX (April 15, 2025) - A team including mathematics researchers from The University of Texas at Arlington and scientists from the U.S. Department of Agriculture (USDA) has published a new study about finding effective methods to predict outbreaks of aflatoxins in corn crops in Texas.

Aflatoxins are a family of mycotoxins that are found on agricultural crops such as corn (maize) and some types of nuts. Mycotoxins are toxic compounds that are naturally produced by certain types of fungi. Aflatoxins are carcinogenic and pose serious health risks to humans and animals.

Angela Avila, a postdoctoral fellow who earned her Ph.D. in mathematics at UTA, is second author of the study, which is titled “Prediction of aflatoxin contamination outbreaks in Texas corn using mechanistic and machine learning models” . It was published in the March 4 edition of the journal Frontiers in Microbiology. The interdisciplinary team included experts in plant biology, genetics, molecular biology, geoinformatics, geochemistry, remote sensing, and predictive modeling.

“Our research focuses on predicting aflatoxin outbreaks of maize in Texas, using remote sensing satellite, soil property and meteorological data,” Avila said. “One of the key challenges is that contamination can be present with no visible signs of fungal infection. This makes early risk prediction especially important for allowing targeted prevention and mitigation strategies.”

Jianzhong Su, professor and chair of the UTA Department of Mathematics, is a co-author of the study and was Avila’s doctoral mentor. Avila is working with Su and Lina Castano-Duque, lead author of the study and plant pathologist at the USDA Agricultural Research Service (ARS) Southern Regional Research Center in New Orleans.

“I have known Dr. Avila since early 2023, when she was a Ph.D. student on Dr. Su’s research team at UTA working on crop phenological mathematical models,” Castano-Duque said. “Her deep knowledge of mathematical modeling in the area of agricultural applications was exactly what I was looking for to grow our current research program on prediction of mycotoxin contamination in corn at the USDA-ARS.”

The team explored development of the aflatoxin risk index (ARI) and multiple machine learning methods for prediction of aflatoxin outbreaks in Texas. ARI is a predictive model output that indicates the cumulative risk of aflatoxin contamination in crops during their development. The neural network model, developed using an Aspergillus flavus fungal strain, performed best, achieving 73 percent accuracy for predicting outbreaks. A. flavus is an opportunistic pathogen of crops and the key fungus for aflatoxin production.

“My main contribution was calculating historical planting dates for each county in Texas using time-series satellite imagery,” Avila said. “Since maize is most susceptible to aflatoxin contamination at specific growth stages, having precise planting dates is critical. My contributions for planting date estimations significantly improved our risk assessment, enhancing the accuracy of our machine learning models by 20–30 percent.”

“As part of her research contributions to our mycotoxin research, Dr. Avila integrated a new input. She used the normalized difference vegetation index (NDVI), acquired from satellite imagery, to predict planting times,” Castano-Duque said. “She will continue growing her phenological model to apply it to the rest of the U.S.”

Avila noted that the study has important implications for farmers, processors, and consumers, as mycotoxin contamination leads to billions of dollars in economic losses each year.

“Our research will allow farmers to make informed decisions to implement effective mitigation strategies, helping protect crops, food security, sustainability, and economic stability,” she said.

“This cutting-edge research will revolutionize the management of mycotoxin contamination in corn, addressing its associated challenges,” Castano-Duque said. “Farmers will benefit from expert guidance on the risk levels of mycotoxin contamination that will aid in future crop selection and the ability to adapt input variables, such as fungicide and biocontrol application, as needed.”