Near-infrared spectroscopic models for analysis of winter pea (Pisum sativum L.) quality constituents

Background: Winter pea (Pisum sativum L.) grows well in a wide geographic region, both as a forage and cover crop. Understanding the quality constituents of this crop is important for both end uses; however, analysis of quality constituents by conventional wet chemistry methods is laborious, slow and costly. Near infrared reflectance spectroscopy (NIRS) is a precise, accurate, rapid and cheap alternative to using wet chemistry for estimating quality constituents. We developed and validated NIRS calibration models for constituent analysis of this crop.

Results: Of the 11 constituent models developed, nine constituents including moisture, dry-matter, total-nitrogen, crude protein, acid detergent fiber, neutral detergent fiber, AD-lignin, cellulose and non-fibrous carbohydrate had low standard errors and a high coefficient of determination (R² = 0.88-0.98; 1 - VR, which is the coefficient of determination during cross-validation = 0.77-0.92) for both calibration and cross-validation, indicating their potential for quantitative predictability. The calibration models for ash (R² = 0.65; 1 - VR = 0.46) and hemicellulose (R² = 0.75; 1 - VR = 0.50) also appeared to be adequate for qualitative screening. Predictions of an independent validation set yielded reliable agreement between the NIRS predicted values and the reference values with low standard error of prediction (SEP), low bias, high coefficient of determination (r² = 0.82-0.95), high ratios of performance to deviation (RPD = SD/SEP; 2.30-3.85) and high ratios of performance to interquartile distance (RPIQ = IQ/SEP; 2.57-7.59) for all 11 constituents.

Conclusion: Precise, accurate and rapid analysis of winter pea for major forage and cover crop quality constituents can be performed at a low cost using the NIRS calibration models developed. © 2018 Society of Chemical Industry.