Skip to main content
SLU publication database (SLUpub)

Doctoral thesis2015Open access

Applications of visible and near infrared spectroscopy for sorting and identification of tree seeds

Farhadi, Mostafa


Seeds are the most commonly used regeneration material for reforestation purpose; hence interest in “precision sowing” among nurseries is high due to the high cost of containerized seedling production. In addition, the increased interest in growing hybrid larch in commercial forestry has raised concerns about the purity of hybrid larch seed lots, as there are large proportions of pure parental seeds mixed with hybrid larch seed lots. The aims of the studies presented in this thesis were to evaluate the application of visible (Vis) and/or near infrared (NIR) spectroscopy combined with multivariate modelling for sorting filled-viable, empty and petrified seeds of Larix sibirica Ledeb., verification of hybrid larch, Larix × eurolepis Henry seeds, identification of seeds of Betula pendula Roth and Betula pubescens Ehrh., and authentication of the origin of Picea abies (L.) Karst seed lots. For these purposes, reflectance spectra were recorded on single seeds using XDS Rapid Content Analyzer (FOSS NIRSystems, Inc.) from 400 – 2500 nm with a resolution of 0.5 nm, and multivariate classification models were developed. The results showed that filled-viable, empty and petrified seeds of L. sibirica can be sorted with 98%, 82% and 87% accuracies, respectively. When the seed lot was sorted into viable and non-viable (empty and petrified combined) classes, the predicted class membership reached 100% for both classes. The technique could separate the hybrid larch seeds from pure parental seeds with 100% accuracy. Seeds of B. pubescens and B. pendula were differentiated with 100% and 99% classification accuracy, respectively. Also, the overall classification accuracy among three B. pendula families was 93% and that of B. pubescens was 98%. NIR spectroscopy discriminated Swedish, Finnish, Norwegian, Polish and Lithuanian seed lots of P. abies with 92% - 100% accuracy. Absorption bands that were accounted for distinguishing the various seed lots examined in this thesis were attributed to differences in seed color, moisture content and chemical composition of the seeds, presumably polysaccharides, proteins and fatty acids, which are the common seed storage reserves. The findings demonstrate the feasibility of Vis + NIR spectroscopy as a robust technique for sorting seed lots according to their viability and for certification of seed lots by species and origin. Thus, concerted efforts should be made to scale-up the technique to on-line sorting system for large-scale tree seed handling operations.


seed quality; near infrared spectroscopy; multivariate analysis; signal pre-processing; Siberian larch; hybrid larch; silver birch; downy birch; Norway spruce

Published in

Acta Universitatis Agriculturae Sueciae
2015, number: 2015:103ISBN: 978-91-576-8404-2, eISBN: 978-91-576-8405-9
Publisher: Department of Southern Swedish Forest Research Centre, Swedish University of Agricultural Sciences

    UKÄ Subject classification

    Forest Science

    Permanent link to this page (URI)