Identification Ofmegaselia(diptera: Phoridae) Species Using Wing Vein Landmarking

A semi-automated identification system using wing venation is described for the large, taxonomically challenging genus Megaselia Rondani (Diptera: Phoridae). Wing photographs make two-dimensional images that can be landmarked and analysed to produce transformed (standardised) x-y coordinates, and are well suited for a semi-automated approach. We landmarked wing photographs of individuals of 108 specimens of Megaselia from the ZADBI (Zurqui All Diptera Biodiversity Inventory) project, as well as 262 specimens of Megaselia from the Los Angeles based BioSCAN (Biodiversity Science: City and Nature) project and ran them against libraries of one specimen of each species of the fauna from each site, respectively. Accuracy was about 52-67% for the first identification, with 86-95% of correct identifications in the top 5 candidates list. Expanding the library to include all identified specimens raised the likelihood that the first identification was correct to 93-96% of specimens, with 97-98% correct identifications in the top 5 list. Incorporating major body colour (dark brown vs yellowish) in the identification system increased accuracy to 100% for the ZADBI specimens; other, easily seen characters were used for problematic LA specimens. We suggest that a wing photograph be included in all Megaselia species descriptions to allow this semi-automation of identification based on morphology, and further that this system could provide a bridge to DNA-based species descriptions from old holotypes in collections.