DNA barcoding reveals taxonomic gaps and cryptic diversity in mosquito vectors from arbovirus-active coastal...

Mangrove-coastal ecosystems in the Colombian Caribbean combine high mosquito diversity, migratory birds, vertebrate reservoirs, and documented circulation of medically important arboviruses. Reliable vector identification in these systems is limited by morphological complexity and incomplete public barcode reference libraries.

Adult mosquitoes collected in coastal rural localities of Córdoba department, Colombia, were morphologically identified and analyzed using COI DNA barcoding. A curated dataset of 181 quality-controlled COI sequences standardized to 696 alignment positions was generated after removal of exact duplicate records and sequences with internal stop codons. K2P distances and Neighbor-Joining trees were estimated in MEGA12. To address the limitations of distance-based inference, maximum-likelihood analysis was performed in IQ-TREE with ModelFinder, SH-aLRT, and ultrafast bootstrap support; Bayesian inference was performed in MrBayes; and maximum parsimony was used as a complementary topology check. Public BOLD searches were used as an independent reference-library check.

The dataset represented 15 mosquito taxa across medically and ecologically relevant genera. Mean intraspecific K2P distances ranged from 0.0000 to 0.0143, with the highest values in An. aquasalis and Ps. confinnis. Mean between-taxon K2P distances ranged from 0.0638 to 0.2542. IQ-TREE selected GTR+F+G4 as the best-fit model and recovered the principal barcode clusters observed in the NJ/K2P tree; Bayesian inference produced a broadly congruent topology with acceptable convergence diagnostics. Public BOLD searches supported several morphology-based identifications but returned genus-level, ambiguous, or discordant placements for difficult groups, especially Culex, Psorophora, Haemagogus, and Culex grupo Salinarius.

This expanded COI barcode dataset provides a molecular baseline for mosquito surveillance in arbovirus-active coastal ecosystems of Colombia. The results support the operational value of COI while showing that single-marker identifications must be interpreted cautiously in groups affected by incomplete reference libraries, species complexes, and unresolved mitochondrial structure.