PDF Humphreys_et_al...pdf | Request a copy Full text restricted Available under University of Tasmania Standard License.

Abstract

Identification of plant hybrids produced from closely
related species can be difficult using morphological characteristics
alone, particularly when identifying young
seedlings. In this study, we compared the performance of
three calibration models developed to discriminate
between seedlings of Eucalyptus globulus, E. nitens and
their first-generation hybrid using either foliar oil chemistry
or near-infrared reflectance spectral data from
fresh, whole leaves. Both oil and near-infrared
reflectance spectroscopy (NIRS) models were developed
using partial least-squares discriminant analysis and
showed high classification accuracy, all correctly classifying
more than 91% of samples in cross-validation.
Additionally, we developed a larger, “global” and independently
validated NIRS model specifically to discriminate
between E. globulus and F1 hybrid seedlings of different
ages. This model correctly classified 98.1% of
samples in cross-validation and 95.1% of samples from
an independent test set. These results show that NIRS
analysis of fresh, whole leaves can be used as a rapid
and accurate alternative to chemical analysis for the
purpose of hybrid identification.

Item Type:	Article
Authors/Creators:	Humphreys, JR and O'Reilly-Wapstra, J and Harbard, JL and Davies, NW and Griffin, AR and Jordan, GJ and Potts, BM
Keywords:	eucalypt, spectra, 1,8-cineole, NIRS, fresh leaves, discriminant analysis, partial least-squares regression
Journal or Publication Title:	Silvae Genetica
ISSN:	0037-5349
Item Statistics:	View statistics for this item

Actions (login required)

Item Control Page