Abstract
Mosquito-borne diseases remain a significant public health concern, underscoring the need for accurate species-level identification of vector species, including Aedes mosquitoes. Identification based solely on morphology is often limited by interspecific overlap, environmentally induced phenotypic plasticity, and physical damage to field-collected specimens. This study evaluated nine Aedes species (Ae. aegypti, Ae. albopictus, Ae. chrysolineatus, Ae. lineatopennis, Ae. macfarlanei, Ae. poicilius, Ae. vexans, Ae. vigilax, and Ae. vittatus) and a related taxon (Aedeomyia catasticta) in Thailand, using DNA barcoding, wing geometric morphometric (WGM) analysis, and the Random Forests (RF) machine learning algorithm. DNA barcoding of the cytochrome c oxidase subunit 1 (cox1) gene showed strong concordance with morphological classifications, confirming its reliability for species-level identification. Across all 10 species, sequence similarity with GenBank and the Barcode of Life Data System ranged from 96% to 100%, highlighting reliable identification when robust references are available. WGM analysis revealed significant wing shape differences among species (P < 0.05), with 91.05% classification accuracy. The Mahalanobis distance and RF algorithms, applied to newly field-collected specimens assigned as unknown species, demonstrated strong discriminatory power, both achieving 100% accuracy for seven species based on wing shape. Slightly lower accuracy was observed for three species, with Mahalanobis distance achieving 90% (one misclassified individual) and the RF algorithm 80% (two misclassified individuals). These findings present a practical guideline for identifying Aedes mosquitoes and a related taxon in Thailand by integrating approaches. Accurate species identification is essential for selecting targeted vector control strategies and enhancing the effectiveness of Aedes-borne disease surveillance and management.