Computer decision aids can help integrate and apply diverse sources of information and expertise to problems of integrated pest management (IPM) in agriculture and forestry. AppleMgr combines a rule-based expert system with databases and spreadsheets in a prototype decision aid intended to be expanded and modified for use by extension workers in the Northwest U.S. The program requires an IBMcompatible microcomputer with hard disk. AppleMgr concentrates on the two most important insect pests on apple in the Northwest--codling moth, Cydia, pomonella (L.), and San Jose scale, Quadraspidiotus perniciosus Comstock, and on phytophagous mites, whose control largely depends on predators. The primary goal of AppleMgr is to demonstrate an improved process of decision making in apple IPM. AppleMgr is composed of modules for diagnosis of pest injury, identification of pest and natural enemy specimens, and management. The first two modules arrive at conclusions through backward-chaining inference from user observations. The management module uses backward chaining supplemented with external calculation programs to find the net benefit of a pesticide application. A method is included to predict yield and fruit size from crop samples. Cullage from codling moth and San Jose scale, mite effect on fruit size, probability of biological mite control and pesticide efficacy are predicted from researchers' data and estimates. Selected relative beneficial and adverse side effects of pesticides are presented in spreadsheets. An analysis of packing house records for apple crops from eight orchards at three yields using 1987 and 1988 prices and packing charges showed that net crop value varied by up to $8000 per acre. The variability in crop value and the importance of adverse side effects of pesticides suggest that the commonly-used "action thresholds" for treatment are seriously inadequate. AppleMgr may point the way toward more dynamic and realistic methods of IPM decision making. / Graduation date: 1990

Direct dissemination of expert knowledge to agricultural producers through computer programs will increase product quality as well as profit margin. The construction of an expert system to help farmers manage apple orchards is reported. The system provides advice regarding specific pest management, treatment of winter injuries, drought control and general pesticide selection. The knowledge structures employed in the construction of the system are explained, and some sample interactions are provided. A model of the apple scab disease cycle is incorporated into POMME to give the system a more fundamental reasoning capability than available from the use of infection tables. Extension experts who have run trial cases on the system have approved its release for use by commercial apple growers. / Master of Science