Validation of Sclerotinia sclerotiorum Apothecial Models to Predict Sclerotinia Stem Rot in Soybean Fields of North Dakota

Sclerotinia sclerotiorum, the causal agent of Sclerotinia stem rot (SSR), is a major fungal pathogen of soybeans, causing an estimated 31 million bushels (0.84 million metric tons) of yield loss in 2019 alone in the northern United States and Ontario, Canada. Fungicide applications are the primary management strategy, and their effectiveness depends on correct timing during the flowering period when apothecia are present on the soil surface. Accurate prediction of apothecial development can guide precise fungicide applications and reduce ineffective applications. In 2018, the University of Wisconsin-Madison developed three weather-based logistic regression models for irrigated conditions and three for non-irrigated conditions to predict the probability of S. sclerotiorum apothecial development, each incorporating different environmental variables. This current study aimed to validate the non-irrigated models and determine an appropriate action threshold for North Dakota. Models were evaluated using disease monitoring data collected from 52 non-irrigated commercial soybean fields across North Dakota over three years. To improve predictive accuracy, probabilities from non-irrigated models were averaged to create an ensemble model. The performance of individual and ensemble models was assessed. Results showed that the ensemble model achieved the highest accuracy (68%) among all the models, in predicting end-of-season disease incidence (DI, %) across all fields using a 10% DI threshold and a 35% risk probability action threshold. The model also demonstrated balanced sensitivity (70%) and specificity (67%). These findings can improve disease risk predictions, reduce unnecessary fungicide applications, and guide soybean farmers in North Dakota on optimal fungicide application timing.