Evidence of a trans-kingdom plant disease complex between a fungus and plant-parasitic nematodes.

Disease prediction tools improve management efforts for many plant diseases. Prediction and downstream prevention demand information about disease etiology, which can be complicated for some diseases, like those caused by soilborne microorganisms. Fortunately, the availability of machine learning methods has enabled researchers to elucidate complex relationships between hosts and pathogens without invoking difficult-to-satisfy assumptions. The etiology of a destructive plant disease, Verticillium wilt of mint, caused by the fungus Verticillium dahliae was reevaluated with several supervised machine learning methods. Specifically, the objective of this research was to identify drivers of wilt in commercial mint fields, describe the relationships between these drivers, and predict wilt. Soil samples were collected from commercial mint fields. Wilt foci, V. dahliae, and plant-parasitic nematodes that can exacerbate wilt were quantified. Multiple linear regression, a generalized additive model, random forest, and an artificial neural network were fit to the data, validated with 10-fold cross-validation, and measures of explanatory and predictive performance were compared. All models selected nematodes within the genus Pratylenchus as the most important predictor of wilt. The fungus after which this disease is named, V. dahliae, was the fourth most important predictor of wilt, after crop age and cultivar. All models explained around 50% of the total variation (R2 ≤ 0.46), and exhibited comparable predictive error (RMSE ≤ 1.21). Collectively, these models revealed that the quantitative relationships between two pathogens, mint cultivars and age are required to explain wilt. The ascendance of Pratylenchus spp. in predicting symptoms of a disease assumed to primarily be caused by V. dahliae exposes the underestimated contribution of these nematodes to wilt. This research provides a foundation on which predictive forecasting tools can be developed for mint growers and reminds us of the lessons that can be learned by revisiting assumptions about disease etiology.