Three Cubital Cell Wing Two Cubital Cell

Solitary bees (Hymenoptera: Apoidea) are endangered by extinction. Furthermore they are highly suited for nature conservancy indication purposes. Nevertheless their appearence in development and ecological planning tasks is quite seldom due to their diicult classii-cation. Our approach to classiication of solitary bees is based on two ideas: (1) to use only the veining of the wings for discrimination, (2) to automate measurement, evaluation, and classiication using image analysis and a hierarchical classiication approach. First results show the functionality and eeciency of our approach even for the classiica-tion of species which are very diicult to separate by classical taxonomy. 1 Motivation There are several reasons for an intensive research interest in solitary bees (Hy-menoptera: Apoidea): (1) The hazard, especially by modern agriculture techniques , for the about 550 Central European species of solitary bees is obvious. (2) Solitary bees are economically and ecologically important for the pollination of many owers and plants. (3) From all pollinating insects the species of solitary bees show the greatest dependencies on territorial conditions ((owers, timbers etc.). Thus solitary bees are very sensitive to agricultural interventions (structural loss, pesticide etc.) and highly suited as bioindicators within ecological planning and monitoring tasks (Drescher 1982, Schwenninger 1992). However there is a lack of suucient classiication results for solitary bees in traditional taxonomy as well as in morphometrical methods which hampers the ecological analysis and the use of these species as bioindicators. 2 Related Work Classical taxonomy reveals several drawbacks: insuucient literature about the determination of zoological species, unsolved relationships between several gen-era and species, and for all an amazing deeciency about the ecology and diversity of species, i.e. worldwide only about 15% of all species are known (Vogt 1994). Computer-assisted approaches to morphometrical analysis (i.e. classiication based on the measurement of characteristic morphological features) can only be found for honey bees and show a lack of eeciency within extensive applications due to an assembly of quite diierent features used for classiication: (1) features of diierent body parts, i.e.