Investigating loci compositions by conventional methods is limited in fully addressing complex gene information. We applied self-organizing map (SOM) to characterize Amplified Fragment Length Polymorphism (AFLP) of aquatic insects in six streams in Japan in responding to environmental variables. Locus band presence patterns were clustered by the trained SOM. Presence and absence data of loci were altered and cluster change through recognition was Subsequently expressed to indicate sensitivity to environmental variables. The outlier loci were determined based on the 90th percentile. Subsequently environmental responsiveness was obtained for each outlier in different species. Outlier loci were overall sensitive to pollutants and feeding material. Poly-loci like responsiveness was detected in adapting to environmental constraints. SOM training combined with recognition could be an efficient means of characterizing loci information without knowledge on population genetics a prior.