Development and validation of a “Real-Time” Apple IPM Website for New York

Abstract: Apple growers in the eastern US have faced challenges in managing the complex ofinsects and diseases of apples using conventional pesticides during the last decade because ofincreasing pesticide regulatory restrictions, public concerns about food safety and environmentalquality, and the development of resistance to older materials by key insect and disease pests.Growers are attempting to turn to newer reduced-risk pesticides, but these are more expensiveand require more precise use patterns because of their different modes of action. In addition,many current IPM protocols were designed for older conventional materials. An interdisciplinarygroup of researchers at Cornell University has developed a web-based, “Real-Time” Apple IPMDecision Support System that can deliver relevant, current information on weather data and pestpopulations to facilitate grower pest management decisions throughout the growing season. Thissystem tracks seasonal development of key insect pests and diseases using Degree Day (DD) andInfection Risk models that indicate pest status, pest management advice and sampling options,and are linked to an interactive system that helps growers choose appropriate materials whenpesticide use is recommended. Insect pest developmental stages are calculated from DDaccumulations at New York State IPM and National Weather Service airport weather stationsthroughout the state. The insect pests addressed by this website are: apple maggot, oriental fruitmoth, codling moth, plum curculio, obliquebanded leafroller and spotted tentiform leafminer.Disease predictions are available for apple scab, fire blight, and a summer disease complex (sootyblotch and flyspeck). We compared web predictions with population trends observed in the fieldfor as many pest species as possible, although not all populations of all species were large enoughor distinct enough to make a practical assessment of the website’s accuracy in all cases. Predictionswere generally accurate, although some pest occurrences were predicted too early or toolate. The main sources of error in the website predictions were: 1 – Traps were sometimes set outtoo late, so that we missed the first flight, and therefore the biofix was wrong. 2 –- The trap checkinterval was sometimes too long to precisely identify moth catch trends; our 7-day schedule couldhave been shortened at times, to better track important events, such as dates near the anticipatedfirst or peak catches. 3 – Some target insect populations were too low to make good predictionsof their developmental events; this was generally a result of the cool, wet summer weather in2009, and so was out of our control to remedy. 4 – Model predictions based on historical datawere not precise enough to be accurate every time; for instance, we did not have extensiverecords on codling moth peak flight periods. 5 – The weather stations were often not numerousenough or close enough to individual sites to be representative of true DD conditions in theorchards. This would be difficult to rectify without investing in a large number of additionalgrower-owned ground weather stations, or else obtaining our DD information from nationalweather databases.