Artificial neural networks (ANN) methodology, molecular analyses and comparative morphology of the male postabdomen were used successfully in parallel for species identification and resolution of some taxonomic problems concerning West Palaearctic species of the genus Tachina Meigen, 1803. Supervised feed-forward ANN with back-propagation of errors was applied on morphometric and qualitative characters to solve known taxonomic discrepancies. Background molecular analyses based on mitochondrial markers CO I, Cyt b, 12S and 16S rDNA and study of male postabdominal structures were published separately. All three approaches resolved taxonomic doubts with identical results in the following five cases: case 1, the four presently recognized subgenera of the genus Tachina were confirmed and the description of a new subgenus was recommended; case 2, the validity of a new boreo-alpine species (sp.n.) was confirmed; case 3, the previously supposed presence of T. casta (Rondani, 1859) in central Europe was not supported; case 4, West Palaearctic T. nupta (Rondani, 1859) was contrasted with East Palaearctic specimens from Japan, which seem to represent a valid species not conspecific with central European specimens; T. nupta needs detailed further study; case 5, T. nigrohirta (Stein, 1924) resurrected recently from synonymy with T. ursina Meigen, 1824 was confirmed as a valid species. This parallel application of three alternative methods has enabled the principle of 'polyphasic taxonomy' to be tested and verified using these separate results. For the first time, the value of using the ANN approach in taxonomy was justified by two non-mathematical methods (molecular and morphological).

Masaryk University Faculty of Science Kotlářská 2 Brno Czech Republic

Citace poskytuje Crossref.org