Development and validation of IIKC: an interactive identification key for Culicoides (Diptera: Ceratopogonidae) females from the Western Palaearctic region

Abstract

Background and methods: The appearance of bluetongue virus (BTV) in 2006 within northern Europe exposed a lack of expertise and resources available across this region to enable the accurate morphological identification of species of Culicoides Latreille biting midges, some of which are the major vectors of this pathogen. This work aims to organise extant Culicoides taxonomic knowledge into a database and to produce an interactive identification key for females of Culicoides in the Western Palaearctic (IIKC: Interactive identification key for Culicoides). We then validated IIKC using a trial carried out by six entomologists based in this region with variable degrees of experience in identifying Culicoides.

Results: The current version of the key includes 98 Culicoides species with 10 morphological variants, 61 descriptors and 837 pictures and schemes. Validation was carried out by six entomologists as a blind trial with two users allocated to three classes of expertise (beginner, intermediate and advanced). Slides were identified using a median of seven steps and seven minutes and user confidence in the identification varied from 60% for failed identifications to a maximum of 80% for successful ones. By user class, the beginner group successfully identified 44.6% of slides, the intermediate 56.8% and the advanced 74.3%.

Conclusions: Structured as a multi-entry key, IIKC is a powerful database for the morphological identification of female Culicoides from the Western Palaearctic region. First developed for use as an interactive identification key, it was revealed to be a powerful back-up tool for training new taxonomists and to maintain expertise level. The development of tools for arthropod involvement in pathogen transmission will allow clearer insights into the ecology and dynamics of Culicoides and in turn assist in understanding arbovirus epidemiology.

Figure 1

Initial screen of IIKC upon opening program. On the left side, the descriptor list and their states (red rectangle); on the right side the remaining and discarded taxa (green rectangle) and, in the middle definitions and pictures (violet rectangle) of either descriptors, states or taxa according to the selection.

Figure 2

IIKC database structure. The four hierarchical levels are represented by grey circles and are numbered 1 to 4 (black). The first level gathered the 23 descriptors available at the start of identification. Descriptor logical dependency between two descriptors was shown by a black line meaning a particular state of the descriptor into level n have to be selected to “unlock” the one into the level n + 1. The unlocked descriptors were incremented in the list of the descriptors available to user.

Figure 3

Examples of end user pathways,intermediateandadvanceduser, for the identification ofC. newsteadi, and the pathway following the xper sort. Histogram showed the percentage of discarded taxa at each step for each pathway. Letters a to f or g corresponded to the identification step 1 to 6 or 7 on the scheme and on the histogram; the black, green and red letters/arrows corresponded to respectively intermediate, advanced user and the xper sort.

Figure 4

Numbers of observations for each descriptor used (n = 1,397). Black bars represent successful observations and grey ones those which failed. The light area gathered the seven descriptors, which represents 50% of total observations and the pigmented area 95% of the whole. Bars were ordered from the above to the top by total decreasing. The stars were added for the descriptors leading to error superior to 10%.

Figure 5

Descriptor groups used by the different users (in percentage) with n as the total number of observations made by each user.