TITLE:
Near Infrared Spectroscopy (NIRS) Model-Based Prediction for Protein Content in Cowpea
AUTHORS:
Kavera Biradar, Waltram Ravelombola, Aurora Manley, Caroline Ruhl
KEYWORDS:
Cowpea, Germplasm, Protein, Near-Infrared Spectroscopy (NIRS), Partial Least Squares (PLS)
JOURNAL NAME:
American Journal of Plant Sciences,
Vol.15 No.3,
March
15,
2024
ABSTRACT: Cowpea
(Vigna unguiculata L. Walp) is a
multi-purpose legume with high quality protein for human consumption and
livestock. The objective of this work was to develop near-infrared spectroscopy (NIRS) prediction models to estimate protein content in cowpea. A
total of 116 cowpea breeding lines with a wide range of protein contents (19.28
% to 32.04%) were selected to build the model using whole seed and ground seed
samples. Partial least-squares discriminant analysis (PLS-DA) regression technique
with different pre-treatments (derivatives, standard normal variate, and
multiplicative scatter correction) were carried out to develop the protein prediction
model. Results showed: 1) spectral plots of both the whole seed and ground seed
showed higher spectral scatter at higher wavelengths (>1450 nm), 2) data
pre-processing affects prediction accuracy for bot whole seed and ground seed
samples, 3) prediction using ground seed samples (0.64 R2 0.85) is better than the whole seed (0.33 R2 0.78), and 4) the data
pre-processing second derivative with standard normal variate has the best
prediction (R2_whole seed = 0.78, R2_ground
seed = 0.85). The results will be of interest in cowpea
breeding programs aimed at improving total seed protein content.