A novel Bayesian approach based on wing geometric morphometry to discriminate <i>Culicoides</i> species (Diptera: Ceratopogonidae)

A Bayesian Procrustes analysis (BPA) was used to discriminate livestock-associated species: Culicoides innoxius Sen and Das Gupta, Culicoides peregrinus Kieffer, and Culicoides oxystoma Kieffer. BPA results were compared with classical geometric morphometric analysis (CGMA). Markov Chain Monte Carlo (MCMC) parameters, Kullback-Leibler (KL) divergence, Hellinger distance, and total variation distance were considered. BPA validation was further done using CGMA. BPA depicted significant differences at 95% credible intervals (CrIs) in their posterior distribution of Procrustes variance (σ) between the species with minimum overlap between closely related ones, C. innoxius and C. peregrinus, and no overlap between distantly related C. oxystoma and C. peregrinus; C. innoxius. MCMC posterior convergence plots supported the accuracy of the BPA. In the trace plots, the MCMC explored the parameter space effectively. For the estimation of divergence between the distribution of species, KL divergence, Hellinger distance, and total variance distance were calculated, which exhibited the highest dissimilarity between C. oxystoma and C. innoxius, followed by C. oxystoma and C. peregrinus and the lowest was between C. peregrinus and C. innoxius. The effectiveness of the BPA over CGMA was assessed by incorporating Culicoides regalis individuals within the analysis. In BPA, an erratic convergence plot indicated the presence of C. regalis within the C. innoxius dataset, whereas CGMA could not separate C. regalis. This is probably the first time the Bayesian approach has been used in Culicoides taxonomy. So far, the results have yielded reliable, sensitive, and accurate species identification.