Dynamic geospatial modeling of mycotoxin contamination of corn in Illinois: unveiling critical factors and predictive insights with machine learning

Authors: Castano-Duque, Lina; Winzeler, Hans - Edwin; Blackstock, Joshua; CHENG, LIU - Wageningen Ur; VERGOPOLAN, NOEMI - Princeton University; FOCKER, MARLOUS - Wageningen Ur; BARNETT, KRISTIN - Illinois Department Of Agriculture; Owens, Phillip; VAN DER FELS-KLERX, INE - Wageningen Ur; Vaughan, Martha; Rajasekaran, Kanniah - Rajah

Submitted to: Frontiers in Microbiology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/26/2023
Publication Date: 11/1/2023
Citation: Castano-Duque, L.M., Winzeler, H.E., Blackstock, J.M., Cheng, L., Vergopolan, N., Focker, M., Barnett, K., Owens, P.R., Van Der Fels-Klerx, I., Vaughan, M.M., Rajasekaran, K. 2023. Dynamic geospatial modeling of mycotoxin contamination of corn in Illinois: unveiling critical factors and predictive insights with machine learning. Frontiers in Microbiology. 14. Article 1283127. https://doi.org/10.3389/fmicb.2023.1283127.
DOI: https://doi.org/10.3389/fmicb.2023.1283127

Interpretive Summary: Mycotoxin contamination of corn is a pervasive problem that negatively impacts human and animal health and causes economic losses to the agricultural industry worldwide. This research provides a modeling study of mycotoxin contamination in Illinois (IL), a corn producing state in the USA. Historical aflatoxin (AFL) and fumonisin (FUM) mycotoxin contamination data, daily weather data, satellite data, dynamic geospatial soil properties and land usage parameters were modeled to identify factors significantly contributing to outbreaks of mycotoxin contamination. Two modeling methods were used: gradient boosting machine (GBM) and a neural network (NN). GBM identified temperature and precipitation prior to sowing as significant influential factors contributing to the outcome of both AFL and FUM contamination. Furthermore, the dynamic geospatial models showed that soil physical and chemical characteristics were correlated with AFL and FUM contamination. NN models showed high class specific performance for one-year predictive validation for AFL (73%) and FUM (85%), highlighting its accuracy for annual mycotoxin prediction. Our models revealed soil and vegetative index information derived from geospatial data along with year-specific weekly average precipitation and temperature were important factors that were correlated with mycotoxin contamination. These findings serve as reliable guidelines moving forward for future modeling efforts and to identify novel data inputs for prediction of AFL and FUM outbreaks and potential farm-level management practices.

Technical Abstract: Mycotoxin contamination of corn is a pervasive problem that negatively impacts human and animal health and causes economic losses to the agricultural industry worldwide. Historical aflatoxin (AFL) and fumonisin (FUM) mycotoxin contamination data of corn, daily weather data, satellite data, dynamic geospatial soil properties and land usage parameters were modeled to identify factors significantly contributing to outbreaks of mycotoxin contamination of corn grown in Illinois (IL), AFL >20 ppb and FUM >5 ppm. Two methods were used: a gradient boosting machine (GBM) and a neural network (NN). Both GBM and NN models were dynamic at a state-county geospatial level because they used GPS coordinates of the counties linked with soil properties. GBM identified temperature and precipitation prior to sowing as significant influential factors contributing to high AFL and FUM contamination. AFL-GBM showed that a higher aflatoxin risk index (ARI) in January, March, July, and November led to higher AFL contamination in the southern regions of IL. Higher values of corn-specific normalized difference vegetation index (NDVI) in July led to lower AFL contamination in central and southern IL, while higher wheat specific NDVI values in February led to higher AFL. FUM-GBM showed that temperature in July and October, precipitation in February, and NDVI values in March are positively correlated with high contamination throughout IL. Furthermore, the dynamic geospatial models showed that soil characteristics were correlated with AFL and FUM contamination. Greater calcium carbonate content in soil was negatively correlated with AFL contamination, which was noticeable in southern IL. Greater soil moisture and available water holding capacity throughout southern IL were positively correlated with high FUM contamination. Higher clay percentage in the northeast areas of IL negatively correlated with FUM contamination. NN models showed high class-specific performance for one-year predictive validation for AFL (73%) and FUM (85%), highlighting its accuracy for annual mycotoxin prediction. Our models revealed soil, NDVI, year-specific weekly average precipitation and temperature were the most important factors correlated with mycotoxin contamination. These findings serve as reliable guidelines for future modeling efforts to identify novel data inputs for prediction of AFL and FUM outbreaks and potential farm-level management practices.