CHAPTER 1

INTRODUCTION

Within the continental United States, the major commercial production area for subtropical and tropical fruits is limited to Arizona, California, Texas and southern Florida. Citrus production in Florida alone occurs in more than 30 counties with more than 14,000 citrus growers cultivating 103 million citrus trees on more than 853,742 acres (Crawford, 1994a; Crawford, 1994b). In 1993-94, Florida citrus production was valued at over 1.3 billion dollars, making it one of Florida's largest agricultural industries (Fla. Ag. Stat. Serv., 1994). In 1993-94, Florida accounted for 81 percent of the citrus produced in the United States and 18 percent of citrus production worldwide (Crawford, 1994a). Several tropical fruits including avocado, mango and 'Tahiti' lime have been of major economic importance in south Florida. The estimated 1993-94 value of these three commodities is in excess of $34.2 million (J. H. Crane, Univ. of Florida, personal comm.). Although 'Tahiti' lime is taxonomically classified as a citrus species, it is considered to be a tropical fruit crop by the Florida fruit industry since it is best suited to a tropical climate and exclusively grown in the commercial tropical fruit producing regions of southern Florida, primarily in Dade County. Other specialty tropical fruit crops such as carambola, papaya, and lychee are also emerging as important alternative fruit commodities of considerable economic importance.

Florida's warm and subtropical climate with relatively high annual rainfall is conducive to a diverse complex of diseases, disorders and pests. It is well known that plant diseases and insect pest damage, as well as the lack of essential nutrients, are some of the major factors which limit fruit crop productivity. In order to manage these problems and so limit their negative impact on crop production, correct symptom identification and rapid diagnosis is essential.

The diagnosis of plant disease, disorders and pest damage is dependent mainly on observations of their respective symptoms and the presence of an insect pest or pathogenic agent in or on diseased tissues. The determination of the causal agent is usually the first step in an attempt to control the disease, disorder or pest damage. When the problem is common and an experienced knowledgeable grower is available, it may not be necessary for them to request assistance from experts. However, as is more often the case, an experienced expert is required to provide diagnosis of plant diseases, disorders and pest damage because growers may be unfamiliar with the diagnosis of its symptoms. On the other hand, many experts are rarely able to devote adequate time to the assistance of all diagnostic requests. Unfortunately, in many instances diagnosis by an expert is too untimely or impractical for growers. In some instances experts cannot make a diagnosis because multiple symptoms appear and recommend soil samples or samples of diseased or damaged plant tissue be sent to diagnostic laboratories or other competent experts for identification of a casual agent. Currently, several of these methods of diagnosis are utilized by growers including soil and plant tissue analysis. However, laboratory analysis is often both untimely and costly to a grower and requires producers to be competent in collecting representative field samples. Consequently, it has been suggested by Florida fruit crop producers that the development and application of alternative means to traditional diagnostic methods currently utilized would be welcomed. By providing a new diagnostic tool with diagnostic capabilities directly to producers, the producers self-reliance may be improved, and may compliment the current diagnostic methods and possibly reduce the dependency of growers upon outside diagnostic assistance.

An expert system is a computer program that uses expert knowledge to attain high levels of performance in a problem area (Plant, 1989). An expert system can be used to supplement the availability of human experts and to increase the availability of expert advice. These programs typically represent knowledge symbolically, examine and explain reasoning processes, and address problem areas that require years of specialized training and education for individuals to acquire and master confidently (Peart, 1989). In order for producers to be able to better deal with diminishing profit margins, elimination of costly and controversial pesticides, the decision making process of plant disease and insect pest damage must be more objective and based upon current up-to-date research information and technology. Consequently, we proposed to develop a computer-based diagnostic expert system which will provide assistance in the diagnosing of diseases, disorders and pest damage of citrus and six selected subtropical and tropical fruit crops.

The primary objective of this project was to develop a computer-based diagnostic expert system that performs at the level of knowledgeable human experts which could be easily understood and utilized by producers as a valuable tool to assist them in the commercial management of subtropical and tropical fruits. Since expert systems are based on reasoning methods that reflect expert human logic, the diagnostic system was patterned on knowledge acquired from a single expert. The reasoning methods were verified by an accordance of prominent experts not directly involved with the knowledge-base development. In instances when the knowledge of the participating experts was limited, additional information was drawn from either current scientific literature or obtained from existing IFAS extension publications.

Well developed expert systems should perform at a level equivalent to the diagnostic ability of human experts, especially when used by 'laypersons' or nonexperts (Bergsma, 1993). Therefore, several analyses were conducted based on questionnaires and actual user participation to determine if the expert systems performance was equivalent to the level of experts. First, we requested outside experts to validate the diagnosis made by the expert system. Secondly, we asked nonexperts which included commercial producers and county extension agents to use the expert system to evaluate if nonexpert users could utilize the expert systems to obtain accurate diagnoses equivalent to the performance of experts. Finally, since symptom identification is vital in the diagnosis of plant disease, we requested the participants to indicate the confidence of both their subjective as well as computer-based diagnoses. We hypothesize that if nonexperts can use the expert system to perform at the level of experts, we have attained our objective and can declare that we have successfully developed a computer-based diagnostic expert system that can increase the availability of expert diagnosis of subtropical and tropical fruits diseases, disorders and pest damage.