Title: Prediction of maize seed attributes using a rapid single kernel near infrared instrument Authors
|Palacios-Rojas, Natalia - INTL MAIZE & WHEAT IMPROV|
Submitted to: Journal of Cereal Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: August 10, 2009
Publication Date: November 1, 2009
Repository URL: http://handle.nal.usda.gov/10113/37523
Citation: Tallada, J.G., Palacios-Rojas, N., Armstrong, P.R. 2009. Prediction of maize seed attributes using a rapid single kernel near infrared instrument. Journal of Cereal Science. 50(3):381-387. Interpretive Summary: Non-destructive measurements of seed attributes would significantly enhance breeder selection of seeds with specific traits and potentially improve hybrid development. A single-kernel near infrared reflectance (NIR) instrument was tested for rapidly measuring maize seed attributes. At a throughput of five kernels/s, the instrument enables plant breeders to quickly select individual seeds that possess specific desired traits. Accuracy of the instrument was tested on 87 maize samples representing a wide variability in the essential amino acids, tryptophan and lysine, crude protein, oil and sugar content. Results showed crude protein and kernel mass were measured well. Tryptophan, lysine and oil measurements were less accurate but have good potential for sorting individual seeds into high, medium and low values. Sugar content was not measured accurately. The instrument has good potential to augment breeder development of nutritionally enhanced maize hybrids.
Technical Abstract: Non-destructive measurements of seed attributes would significantly enhance breeder selection of seeds with specific traits and potentially improve hybrid development. A single kernel near infrared reflectance (NIR) instrument was developed for rapidly predicting maize grain attributes, which would enable plant breeders to quickly select promising individual seeds. With the overall goal to develop spectrometric calibrations, absorbance spectra from 904 to 1685 nm were collected from 87 maize samples, 30 kernels each, (2610 kernels total) representing a wide variability in the essential amino acids, tryptophan and lysine, and crude protein, oil, and soluble sugar contents. Average sample spectra were matched to bulk reference values. Partial least squares calibration models with cross-validation were developed for both relative (% dry matter) and absolute (mg per kernel) constituent contents. Similarly, models using bagging PLSR were developed. The best model obtained was for relative contents of crude protein with an R2p of 0.751 and a SEP of 0.472 mg/kernel. Kernel mass was also highly predictable (R2p = 0.761, SEC = 0.029 g). Tryptophan, lysine, and oil were less predictable but had good potential for segregating individual seeds using NIR. Soluble sugar contents had poor model statistics. Bagging PLSR yielded models with similar levels of prediction.