Multivariate Statistical Analysis of Spectral Reflectance in Coleoptera Elytra Thesis uri icon

Overview

abstract

  • Thesis (M.S., Statistical Sciences)--University of Idaho, June 2014 | Coleoptera, commonly known as beetles, comprises 40% of all insects. Spectral readings of the elytra, the hardened outer wing, may be used to identify Coleoptera. Utilizing normal mixture models, eighteen peak wavelengths were identified across taxonomic groups and genders creating a multivariate structure. Multivariate procedures including principal component and discriminant analyses were employed to differentiate taxonomic groups and genders. The first three axes of the principal component analysis provided a clustering of genus and gender for a subset of taxonomic groups. The nearest neighbor discriminant analysis with proportional priors gave a misclassification rate of 5.2%. Internal bootstrap validation of the discriminant model yielded an average error rate of 3.5%. An external cross validation of the same model, conducted on independent samples resulted in an average misclassification of 6.5%. Given the low misclassification rate, multivariate statistical approaches are recommended for analysis of spectral reflectance in Coleoptera and other similar insect groups.

publication date

  • June 1, 2014

Other