17.3.1. Identification keys
The output of taxonomic studies usually includes keys for determining the names (i.e. for identification) of organisms. Traditionally, keys involve a series of questions, concerning the presence, shape, or color of a structure, which are presented in the form of choices. For example, one might have to determine whether the specimen has wings or not — in the case that the specimen of interest has wings then all possibilities without wings are eliminated. The next question might concern whether there is one or two pairs of wings, and if there are two pairs, whether one pair of wings is modified in some way relative to the other pair. This means of proceeding by a choice of one out of two (couplets), thereby eliminating one option at each step, is termed a “dichotomous key” because at each consecutive step there is a dichotomy, or branch. One works down the key until eventually the choice is between two alternatives that lead no further: these are the terminals in the key, which may be of any rank (section 1.4) — families, genera, or species. This final choice gives a name and although it is satisfying to believe that this is the “answer”, it is necessary to check the identification against some form of description. An error in interpretation early on in a key (by either the user or the compiler) can lead to correct answers to all subsequent questions but a wrong final determination. However, an erroneous conclusion can be recognized only following comparison of the specimen with some “diagnostic” statements for the taxon name that was obtained from the key.
Sometimes a key may provide several choices at one point, and as long as each possibility is mutually exclusive (i.e. all taxa fall clearly into one of the multiple choices), this can provide a shorter route through the available choices. Other factors that can assist in helping the user through such keys is to provide clear illustrations of what is expected to be observed at each point. Of necessity, as we discuss in the introduction to the Glossary, there is a language associated with the morphological structures that are used in keys. This nomenclature can be rather off-putting, especially if different names are used for structures that appear to be the same, or very similar, between different taxonomic groups.
A good illustration can be worth a thousand words — but nonetheless there are also lurking problems with illustrated keys. It is difficult to relate a drawing of a structure to what is seen in the hand or under the microscope. Photography, which seems to be an obvious aid, actually can hinder because it is always tempting to look at the complete organism or structure (and in doing so to recognize or deny overall similarity to the study organism) and fail to see that the key requires only a particular detail. Another major difficulty with any branching key, even if well illustrated, is that the compiler enforces the route through the key — and even if the feature required to be observed is elusive, the structure must be recognized and a choice made between alternatives in order to proceed. There is little or no room for error by compiler or user. Even the best constructed keys may require information on a structure that the best intentioned user cannot see — for example, a choice in a key may require assessment of a feature of one sex, and the user has only the alternative sex, or an immature specimen.
The answer to identification undoubtedly requires a different structure to the questioning, using the power of computers to allow multiple access to the data needed for identification. Instead of a dichotomous structure, the compiler builds a matrix of all features that in any way can help in identification, and allows the user to select (with some guidance available for those that want it) which features to examine. Thus, it may not matter if a specimen lacks a head (through damage), whereas a conventional key may require assessment of the antennal features at an early stage.
Using a computer-based, so-called interactive key, it may be possible to proceed using options that do not involve “missing” anatomy, and yet still make an identification. Possibilities of linking illustrations and photographs, with choices of looking “like this, or this, or this”, rather than dichotomous choice, can allow efficient movement through less-constrained options than paper keys. Computer keys proceed by elimination of possible answers until one (or a few) possibilities remain — at which stage detailed descriptions may be called up to allow optimal comparisons. The ability to attach compendious information concerning the included taxa allows confirmation of identifications against illustrations and summarized diagnostic features.