RESUMO
Seed samples from 117 genetically diverse pea breeding lines were used to determine the robustness of Fourier transform mid-infrared spectroscopy (FT-MIR) for the rapid nutritional profiling of seeds. The FT-MIR results were compared to wet chemistry methods for assessing the concentrations of total protein, starch, fiber, phytic acid, and carotenoids in pea seed samples. Of the five partial least square regression models (PLSR) developed, protein, fiber and phytic acid concentrations predicted by the models exhibited correlation coefficients greater than 0.83 when compared with data obtained using the wet chemistry methods for both the calibration and validation sets. The starch PLSR model had a correlation greater than 0.75, and carotenoids had correlation of 0.71 for the validation sets. The methods implemented in this research show the novelty and usefulness of FT-MIR as a simple, fast, and cost-effective technique to determine multiple seed constituents simultaneously.
