Benjamin T. Aldrich, Elizabeth B. Maghirang, Floyd E. Dowell, Srinivas Kambhampati
Journal of Insect Science 7 (18), 1-7, (1 April 2007) https://doi.org/10.1673/031.007.1801
KEYWORDS: species identification, neural network, Zootermopsis angusticollis, Zootermopsis laticeps, Zootermopsis nevadensis, Z. n. nuttingi, Z. n. nevadensis
Dampwood termites of the genus Zootermopsis (Isoptera: Termopsidae) are an abundant group of basal termites found in temperate forests of western North America. Three species are currently recognized in the genus and one of these species is subdivided into two subspecies. Although morphological and genetic characters are useful in differentiating among the three species and the two subspecies, respectively, only hydrocarbon analysis can enable differentiation both among the three species and the two subspecies. Due to the limitations of hydrocarbon analysis, such as the need for fresh specimens, alternative methods that could rapidly and accurately identify Zootermopsis would be useful. Using a partial least squares analysis of near-infrared spectra, each of the Zootermopsis species and subspecies were identified with greater than 95% and 80% accuracy, respectively. Neural network analysis of the near-infrared spectra successfully enabled the identification of the species and subspecies with greater than 99% accuracy. The inexpensive, reproducible, and rapid nature of near-infrared spectroscopy makes it a viable alternative to morphological, hydrocarbon, or genetic analysis for identifying Zootermopsis.