RESUMO
The large numbers of samples processed in breeding and biodiversity programmes require the development of efficient methods for the nondestructive evaluation of basic seed properties. Near-infrared spectroscopy is the state-of-the-art solution for this analytical demand, but it also has some limitations. Here, we present a novel, rapid, accurate procedure based on time domain-nuclear magnetic resonance (TD-NMR), designed to simultaneously quantify a number of basic seed traits without any seed destruction. Using a low-field, benchtop (1) H-NMR instrument, the procedure gives a high-accuracy measurement of oil content (R(2) = 0.98), carbohydrate content (R(2) = 0.99), water content (R(2) = 0.98) and both fresh and dry weight of seeds/grains (R(2) = 0.99). The method requires a minimum of ~20 mg biomass per sample and thus enables to screen individual, intact seeds. When combined with an automated sample delivery system, a throughput of ~1400 samples per day is achievable. The procedure has been trialled as a proof of concept on cereal grains (collection of ~3000 accessions of Avena spp. curated at the IPK genebank). A mathematical multitrait selection approach has been designed to simplify the selection of outlying (most contrasting) accessions. To provide deeper insights into storage oil topology, some oat accessions were further analysed by three-dimensional seed modelling and lipid imaging. We conclude that the novel TD-NMR-based screening tool opens perspectives for breeding and plant biology in general.