Fruit flies (Diptera: Tephritidae) include some of the world’s most serious agricultural pests. Besides causing billions of dollars in direct losses to a wide variety of fruit, vegetable and flower crops (e.g., citrus, apple, mango, sunflower), they limit the development of agriculture in many countries because of the strict trade quarantines imposed to prevent their spread. Of the more than 4,400 species known worldwide (Norrbom 2004), nearly 200 are considered pests. Because there are so many species, many of which are extremely similar, the identification of pest fruit flies is very difficult, even for professional identifiers, such as the port identifiers of USDA APHIS-PPQ for whom this system was originally conceived. We therefore developed this system, which stores the most up to date taxonomic information on pest fruit flies in DELTA format (Dallwitz 1980; Dallwitz, Paine, and Zurcher 1993, 1999), and uses it to make descriptions and interactive keys.
The Pest Fruit Fly Identification System was originally developed as part of a larger project to make taxonomic information on Tephritidae readily available (Thompson 1999). Other components of that project included a comprehensive data base of fruit fly names (Norrbom et al. 1999; Norrbom 2004) and incorporated into the Biosystematic Database of World Diptera; and an extensive data base of Fruit Fly (Diptera: Tephritidae) Literature (updated from Carroll, Norrbom, Thompson & Evenhuis 1999). Updated versions of both are now available on the USDA/Smithsonian Diptera Web Site (also see the Fruit Fly Taxonomy Pages for additional information). After making an identification with the interactive keys, the user may query these databases for complete nomenclatural and distributional data, as well as pertinent references.
The Pest Fruit Fly Identification System includes interactive keys for adults and for larvae, using Intkey (Dallwitz, Paine, and Zurcher 1995) and Lucid Versions 2 and 3. These keys include data and images to identify the most economically important species of fruit flies from throughout the world. Separate keys for adults and larvae were developed because the larvae of many species have yet to be thoroughly described. The adult system includes about 190 species, including a few Anastrepha species that are non-pests but commonly encountered in U.S. ports, and the larval system includes about 80 species.
For recording the data, our objective was to use a data format that the systematics community endorses and widely uses. We also wanted a data format that could encode all kinds of character data and was not proprietary, so data sets could be shared. The DELTA data format (Dallwitz 1980; Dallwitz, Paine & Zurcher 1993) was the only available one that matched our criteria.
The DELTA data sets for the original versions were published as Carroll, White, Freidberg & Norrbom (1999) and Carroll (1999), respectively. Both were rechecked and slightly modified by MJD and ALN, mainly to correct character dependencies, although a few scoring errors were corrected, several character states were reworded, added, or deleted, and two text characters (synonymy, common name) were added to both data sets.
A distribution character, five classification characters (giving subfamily and tribe), and a morphological character (width of cell r2+3 subapical hyaline area) were added to the adult key. The character images, which were originally used with the Online interactive key program (Pankhurst and Aitchison 1975) were modified by MJD for use with Intkey and Lucid. New or updated adult images were also added for many species by ALN.
The characters, character dependencies, and data matrix were converted to Lucid2 Interchange Format (LIF) by means of the Lucid Translator, and the character notes and other DELTA data by means of ad hoc programs written by MJD. The resulting LIF file was imported into Lucid2 and Lucid3.
In the DELTA dataset for larvae, there is a character with more than 90 states, specifying the families of the host plants. Because Lucid2 allows at most 15 states, this character was divided into 8 additional characters, and the original character was excluded from the Lucid2 key.
As Lucid has no equivalent of DELTA’s free-text characters, these characters were also excluded from the Lucid keys.
The ‘expert route’ in the Lucid2 key to adults was generated with the aid of Intkey. In selecting the ‘Best’ characters for an identification, Intkey takes into account the ‘reliabilities’ of the characters as well as their separating power. Therefore, the characters in the expert route are quite different from the ‘Best’ characters selected by Lucid, which has no provision for character reliabilities.
To facilitate comparison of the key programs, the DELTA character numbers were included in the text of the Lucid characters.
Natural-language descriptions were generated from the DELTA data by the program Confor (Dallwitz, Paine, and Zurcher 1993). They are available directly as Web pages, and are also accessible from within the Intkey and Lucid keys.
The Intkey and Lucid2 keys run under MS Windows 95/NT or later. Intkey will also run on Macintosh computers with Windows emulation software. The Lucid3 keys require the Java Runtime Environment Version 1.4.2 or later, which is available for computers running: Windows 98SE, ME, 2000 Professional, Server 2003, and XP; Solaris; Linux; and Mac OS X.
The keys are designed to be totally self-contained, and have built-in ‘Help’.
We have assumed that our users are professional identifiers, such as those working for USDA APHIS-PPQ. Hence, they are already familiar with traditional identification aids, such as keys, and are familiar with their organisms.
An interactive key differs from traditional identification tools in allowing for random access to character data. The user is free to choose any of the available characters in any order, whereas a traditional key allows only for the use of specific characters in a rigid sequence. The program can select the best characters for use, based on their ability to separate the remaining taxa under consideration.
The verification process is also much easier. Complete descriptions are available, just as in traditional taxonomic publications. Intkey can also give the differences between the specimen and a taxon, or within any set of taxa, and can generate diagnostic descriptions of taxa.
The programs can accommodate errors (e.g., a poor or aberrant specimen, a mistake by the user, or an error in the database): a taxon is eliminated only when the number of discrepancies between the specimen and the taxon exceeds a specified value.
Numeric characters and characters with more than two states are allowed. This speeds up the identification process by reducing the total number of decisions that must be made, because more than the traditional two possibilities can be efficiently evaluated at one time. Characters are accompanied by illustrations and notes that can be accessed at any time.
To keep costs within budget, existing images were re-used wherever possible and only the minimal number of new ones were created. However, many images were improved, for instance, many black and white habitus figures were colored. So, don’t be surprised if these images look familiar!
The user should be aware, of course, that interactive keys are not a panacea. Unusual specimens, those outside the domain of the data or with distorted features will still have to be sent to the systematist. Non-pest species of fruit flies not included in the system may also be misidentified. See Dallwitz, Paine and Zurcher (2000) for additional explanation of the features and advantages of interactive keys in general, and Dallwitz (2000a,b) for comparisons of various interactive-key programs.
An earlier version of the Intkey interactive key for adults has been published on CD-ROM (Carroll et al. 2004).
