Wing geometric morphometrics for identification of mosquito species (Diptera: Culicidae) of neglected epidemiological importance

Abstract

Culicidae comprises more than 3500 species, some of which are responsible for the spread of various human diseases, causing millions of deaths worldwide. Correct identification of these species is essential for the development of surveillance and control strategies. The most common method of mosquito identification is based on specific traits of the external morphology of species. However, identification of mosquitoes by morphological characters can be inaccurate or even unfeasible if the specimen is damaged or there is a lack of distinguishing features, as in the case of cryptic species complexes. Wing geometric morphometrics is a reliable, affordable tool for the identification of mosquito species, including sibling species. More importantly, it can be used in addition to both traditional morphologic identification methods as well as genetic approaches. Here, wing geometric morphometrics was used to identify sixteen mosquito species from eight genera: Aedes, Coquillettidia, Culex, Limatus, Mansonia, Psorophora, Runchomyia, and Wyeomyia. The 390 specimens used here were collected in Sao Paulo, Brazil using CDC traps, aspiration, and Shannon traps. Allometry was assessed by multivariate regression of the Procrustes coordinates on centroid size followed by canonical variate analysis and a pairwise cross-validated reclassification test. A Neighbor-Joining tree based on Mahalanobis distances was constructed with 1,000 bootstrap replicates using MorphoJ 1.02 and Past 2.17c. The canonical variate analysis of genera resulted in distinct clusters for Culex, Limatus, and Psorophora and partial overlapping between Aedes, Coquilettidia, and Mansonia, and between Runchomyia and Wyeomyia. Pairwise cross-validated reclassification tests indicated that genera were identified with an accuracy of at least 99% and subgenera with a mean accuracy of 96% and that in 160 of the 240 possible comparisons species were identified with an accuracy of 100%. Our results show that the eight genera in the study were correctly distinguished by wing shape, as were subgenera and most species, demonstrating that wing geometric morphometrics can be used for the identification of the mosquito species studied here.