Submitted to: Mycopathologia
Publication Type: Abstract Only
Publication Acceptance Date: October 25, 2002
Publication Date: May 1, 2004
Citation: Dowd, P.F. 2004. Validation of a mycotoxin predicting computer program for U.S. Midwest grown maize in commercial fields [abstract]. Proceedings of the Aflatoxin and Fungal Genomics Workshop/Mycopathologia. 157(5):463. Technical Abstract: A computer program designed to predict relative amounts of fungal inoculum at pollination and mycotoxin levels prior to harvest was developed from intensive data collected from 1994-1997 in Central Illinois in the U.S. Temperature, rainfall, and insect presence are the main factors involved. The program could predict fumonisin values close to actual field values when data collected from fields sampled in 1994-1999 was used, which included fumonisin values ranging from ca. 0.1 to 8 ppm. The computer program has now been well validated for fumonisin levels in 3 commercial fields in 2000 and 7 fields in 2001 (most fields were planted with two hybrids) using the generalized values (a version using customized values is under development). In 2000, rainfall was relatively common and insect damage was limited. A predicted level of ca. 1.5 ppm compared well to actual field means of 0.8, 1.3, and 1.6 ppm for prevously studied hybrids. In 2001, the climate was drier during the summer, and insect damage varied according to planting date, with later planted corn being more heavily damaged by European corn borers. For "early" plantings, predicted fumonisin levels (ca. 0.5 ppm) were close to actual means in fields (0.2, 0.3, 0.4, 0.4, 0.6 ppm) except for one field, which ran well below the predicted value. We have a tentative explanation for the lower values of this one field, which was restudied in 2002. Later planted corn encountered more insect damage, but predicted values of near 2.0 ppm compared well with actual values of 2.2, 2.2, and 4.2 for non-Bt maize. No aflatoxin was predicted for 2000, and none was found. In 2001, predicted aflatoxin values of circa 10 ppm were higher than mean values found in some fields, although some individual samples had 2-10 ppm. Initially, equations used for aflatoxin prediction in the program were based on published values from 1983 and 1987, but we knew this data was probably not detailed enough (no insect damage levels available) to predict values very well (although predictions of negative occurrences have all been correct for years from 1992 through 2000). The new data has allowed for adjustment in the program which will be checked against data collected in 2002, another year that the program has predicted aflatoxin will be present. The program is currently being converted from DOS Basic to a format that will run in Windows, with the intent of producing a version that is commercial quality. The predictive computer program will supply data to farmers that will allow them to make decisions on whether to control insects from silking to milk stage, or harvest early in order to help prevent accumulation of undesirable levels of mycotoxins in Midwest grown maize in the U.S.