Discrimination of termite species using Near-Infrared Spectroscopy (NIRS)

Abstract

The classification of most taxa is based exclusively on morphological characters which have limited capacity to discriminate species with phenotypic plasticity and to detect the existence of cryptic species. These limitations can be reduced by the use of recent techniques that quantify molecular attributes in a sample. Near-Infrared Spectroscopy (NIRS) is a non-destructive method that has been used as an efficient tool to discriminate species from several plant and animal groups in controlled laboratory conditions. We tested the efficacy of this method as an alternative to morphology to discriminate termite castes, species, genera, and families. Seven species were selected: Cylindrotermes flangiatus; Cylindrotermes parvignathus; Cornitermes pugnax; Cornitermes ovatus; Heterotermes tenuis; Heterotermes crinitus and Coptotermes testaceus. Discriminant models were created with the separation of 70% of the samples for training (creation of the model) and 30% for testing. We found a high level of accuracy for the discrimination of families, genera, species and castes, ranging from 76% to 100% of accuracy (mean of 90%). This high level of accuracy, using the soldier and worker castes, indicates that Near-Infrared Spectroscopy serves as a reliable alternative to identify termite species. © 2019 Elsevier Masson SAS