Application of Latent Semantic Analysis for Open-Ended Responses in a Large, Epidemiologic Study

摘要

Environmental geospatial data and adult and larval mosquito collection data for up to 106 sites throughout the Republic of Korea (ROK) were used to develop ecological niche models (ENMs) of the potential geographic distribution for eight anopheline species known to occur there. The areas predicted suitable for the Hyrcanus Group species were the most extensive for Anopheles sinensis Wiedemann, An. kleini Rueda, An. belenrae Rueda, and An. pullus Yamada, intermediate for An. sineroides Yamada, and the most restricted for An. lesteri Baisas and Hu and the non-Hyrcanus Group species An. koreicus Yamada and Watanabe and An. lindesayi Yamada. The relative vectorial importance of these species is unknown, and all, except An. koreicus and An. lindesayi, are predicted to occur widely in the northwest of the ROK where malaria transmission has been sporadic since its resurgence in 1993. Our ENMs suggest that it is unlikely that An. koreicus and An. lindesayi are vectors but we do not document consistent geographic differentiation that might incriminate any of the other species as vectors. Because all species are predicted to occur in North Korea, we also cannot reject the hypothesis that malaria infected mosquitoes from North Korea may have been the cause of the resurgence of malaria in the ROK. Ecological differentiation of the eight species is inferred from collection locations and 34 environmental layers based on remote sensing and global climatic averages. Interspecific differences were noted, and characterizing mosquito habitats by ground-based and remote sensing methods is proposed.