Validation of tuber blight (Phytophthora infestans) prediction model

Abstract Potato tuber blight caused by Phytophthora infestans accounts for significant losses of tubers in storage. Despite research on infection and management of tuber blight, there is paucity of information on the prediction of the occurrence tuber blight or modelling of tuber infection by P. infestans under field conditions. A tuber blight prediction model was developed in New York in experiments conducted using cultivars Allegany, NY101, and Katahdin in 1998 and 1999. This model was validated using data collected from the potato cultivar Snowden in field experiments in Laingsburg, Michigan from 2000 to 2009. In both New York and Michigan experiments, disease was initiated by artificial inoculation of cultivars with a US-8 isolate of P. infestans . Mean leaf area affected ranged from 0 to 94% at New York, and 0 to 93% at Michigan. At New York and Michigan, mean tuber blight incidences ranged from 1 to 40% and 0 to 15%, respectively. In the validation of the model using data collected at Laingsburg, Michigan, the model correctly predicted tuber blight incidence in 7 out of 9 years. Comparison of observed with predicted values indicated that slopes of the regression line between observed and predicted germination and infection data were not significantly different ( P  > 0.3547). Correlation coefficient between observed and predicted values was high ( r 2  > 0.65) and the coefficient of variation of the residuals of error was about 12%. Although inoculum availability is assumed in the model, incorporation of relationships of inoculum density, propagule survival in soil, and tuber blight incidence would greatly improve the prediction of tuber blight under field conditions.