Submitted to: Plant Disease
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/20/2014
Publication Date: 1/1/2015
Citation: Copes, W.E. 2015. Weather-Based forecasting of Rhizoctonia web blight development on container-grown azalea. Plant Disease. 99:100-105.
Interpretive Summary: Azalea web blight is an annual problem on some evergreen azalea cultivars grown in containerized nursery production in the southern and eastern United States. Eight weather-based models were defined to generate daily risk values and were tested for their ability to predict web blight development using weather and web blight incidence measurements from 12 locations and years. Risk classifications of 18 or more hours of temperatures between 20 and 30°C with a maximum daily temperature below 35°C and/or 16 or more hours of leaf wetness predicted the correct outcome with an 80% accuracy. The models establish a relationship for accumulated weather based risks to predict need for fungicide applications and can be integrated with calendar scheduling and scouting techniques to adjust for yearly differences in weather that affect web blight development. Practical recommendations were developed for producers of container-grown azalea, and biologically important information was developed for extension specialists and research scientists.
Technical Abstract: Fungicides are the only approach currently used to control Rhizoctonia web blight on container-grown azalea. The most reliable criterion for timing fungicides has been a fixed calendar date with adjustment for year-to-year differences in disease progression made by monitoring early-season increase of blighted leaves within the shrub’s interior canopy. The purpose of this study was to model infection periods as a function of weather variables as a proxy to signal fungicide applications within a discrete period preceding significant increases in leaf blight incidence (LBI). The analysis used continuous weather measurements from 12 site-year data sets, from 2006 to 2011, representing diverse weather patterns from wetter than average, average, and hotter and dryer than average; and weekly or biweekly LBI assessments to significant increases in LBI. Eight weather-based models were defined to generate daily numerical risk values based on temperature, leaf wetness, and rainfall summations that were accumulated daily over time. Using a randomly selected development set of seven site-years, receiver operating characteristic (ROC) curves were used to derive a critical risk index that preemptively signaled a significant LBI increase far enough in advance to empirically justify the need for fungicide applications through the web blight season from June 16 to September 4. Four models were selected for further evaluation based on sensitivity, specificity, Youden’s Index, and positive and negative likelihood ratios to evaluate the five site-year validation data sets. Similar results were obtained from both the developmental and validation data sets. The most consistent results were obtained with two models which used only temperature and leaf wetness variables. Both models had identical values of sensitivity (0.833), specificity (0.750), Youden’s Index (0.583), and positive and negative likelihood ratios (3.333 and 0.222, respectively) from validation data sets. Although all of the final four weather-based models successfully denoted infection periods identified from significant increases in LBI, they showed a similar tendency to miss some infection periods as well as signal some infection periods that were not followed by disease increases. The models establish a relationship for accumulated weather based risks to predict need for fungicide applications, and their practical application to manage web blight in nurseries is discussed.