Application of computer-based pattern recognition procedures in the study of biological samples. Comparison of the cuticular hydrocarbon profiles of different colonies of the black imported fire ant

Abstract

Computer-based pattern recognition procedures have been applied to gas chromatographic data obtained from the cuticular hydrocarbons of different colonies of the black imported fire ant. These included both supervised and unsupervised learning methods, based on the ARTHUR chemometrics system package. The results, obtained from ten samples, indicate significant differences in the hydrocarbon profiles of different colonies of the black imported fire ant. The relative effectiveness of the different procedures in making correct classifications was approximately 90%.