In creating this morphological database for Ichneumonoidea 2011
, 9,175 publications were consulted. Due to language limitation on the part of DYU, only papers in English, Chinese and German were deconstructed. Every morphological entry is derived from published information, and can be verified by left clicking on the information to show the source. Because of the coverage of the publications used in producing this database, this study is a good reflection as to which parts of an insect are deemed to be important for discrete measurements by the authors of Ichneumonoidea when publishing their papers.
The 360 morphological characters (see Morphological index
) recorded in Ichneumonoidea 2011
include measurements (in units 1/10 mm) of whole insects and body parts (e.g. wing, head, mesosoma, metasoma, ovipositor), ratios (in percent) of one part of the body with another (e.g. ovipositor length vs wing length, malar space in relation to eye height ), shape of parts of the body (e.g. length vs width of tergite 1) and count of subparts (e.g. number of antennal segments, number of distal hamuli on the hind wing). Only a few non numerical characters are included (e.g. characters of the antenna, occiputal carina, notauli, tarsal claws etc.) as these maybe a bit more variable as there is no unified way of describing these body parts. Due to limitation of time, colour characters are recorded only for a limited number of taxa. Hopefully a more extensive colour database will be available in the next version.
The major divisions for character processing are head, mesosoma, metasoma, wing, antenna, leg and genitalia. The head is rich in characters and is being used quite extensively by authors in their numerical descriptions. Most mesosomal characters are descriptive with very few discrete measurements, because a globe-like object is not easy to measure, and the variation with respect to propodeal carinae makes a unified description difficult. Although metasoma seems to be amenable to repeatable measurements, very little numerical descriptions outside of the first two tergites are available. Wing venation is not tackled in this version, as the more appropriate way to use wing venation for identification is by image analysis. I hope to present an automatic wing identification system in the next version. The legs are a rich source of numerical descriptions, but unfortunately they are usually passed over because they do not contribute much to species level identification. Genitalia is also not well studied and published description is limited, so very little use is made of the genitalia in this work.
The main purpose of creating this morphological database is that it may eventually form the basis for an alternative to dichotomous key approach to identification. By deconstructing a description to its most basis elements, each element can be used as a separating point. With enough separating points, a specimen can be identified to subfamily, genus or species. The more separating points used, the further down one can go. Thus it can be used by workers without pre-knowledge of the group, just any people who knows the names of the parts of an insect body. Also the advantage of using separating points method is that it is additive, that is, a database of separating points can be added to another to make a better identification tool. On the other hand, a dichotomous key is not additive, and a new key has to be made when a new taxon is added. However, deconstructing descriptions is time consuming and the results can be variable because of the absence of a common methodology. An example is the width to height ratio of the face. The most common width measurement is the narrowest width, but some use the 'antennal socket level' width. Also some use the 'lower face' width and height. But the narrowest width measurement is actual a mixture of two elements - the width/height ratio and the lower/upper face width ratio (which indicates diverging, parallel or converging compound eye). And the 'lower face' ratio is a mixture of face ratio and clypeus ratio. In this case, the better method is the use of 'antennal socket level' width for width/height ratio because the ratio is not affected by the form of the compound eye. The dilemma therefore is deciding what should be included in a database such as this - include only the perceived correct ratio and excluding the bulk of the information available, or include all ratios and make the result very variable. The current work applied the latter with the intention of fine tuning it a bit later.