Link to home

Variation in Disease Incidence of Phomopsis Cane and Leaf Spot of Grape in Commercial Vineyards in Ohio

July 2008 , Volume 92 , Number  7
Pages  1,053 - 1,061

M. Nita, Kansas State University, Manhattan 66506; and M. A. Ellis and L. V. Madden, The Ohio State University, Ohio Agricultural Research and Development Center, Wooster 44691



Go to article:
Accepted for publication 21 February 2008.
ABSTRACT

A statewide survey for incidence of Phomopsis cane and leaf spot of grape (caused by Phomopsis viticola) was conducted during the 2002 to 2004 growing seasons. Over the 3 years, disease was observed in all surveyed vineyards, and mean disease incidence for leaves and internodes was 42 and 50%, respectively. A hierarchical linear mixed model was used to evaluate effects of region, farm within region, vineyard within farm, sampling site (i.e., vine) within vineyard, and shoot (i.e., cane) within vine on disease incidence. Region of the state did not have a significant effect on incidence but there was significant variation at all other levels of the hierarchy (P < 0.05); the greatest variation was at the lowest scale (shoots within vines). The potential effects of weather and management practices on disease risk at the vineyard scale were determined by using nonparametric correlation and binary logistic analyses after first classifying mean incidence per vineyard as being below or above 20% (D20 = 0,1) and 40% (D40 = 0,1). Overall results indicated that variables for predicted number of moderate infection events (DM; based on ambient temperature and hours when either there was measured rainfall or relative humidity above 90%), the extent of fungicide application (C) during early- and mid-May (M1 and M2, respectively), and the use of a dormant-period application of fungicide (DOR) were the key factors in predicting disease risk (for either D20 or D40). Accuracy (percentage of high and low disease vineyards correctly predicted) and area under the receiver operating characteristic curve (an overall measure of the accuracy of a model) for a generic model combining these predictor variables were 74 and 0.84, respectively, for D40 and 87 and 0.97, respectively, for D20. Models based on management practices were as accurate as those that incorporated weather variables. Although the degree of control of this disease is inadequate in Ohio, based on the survey results for incidence, the results from the risk-model analysis showed that improved management might be obtained by applying fungicide early during the growing season.


Additional keywords:Kendall correlation, sensitivity, specificity, ROC

© 2008 The American Phytopathological Society