Screening of maize haploid kernels based on near infrared spectroscopy quantitative analysis

Abstract

Doubled haploid breeding (DH breeding) is one of the major maize breeding techniques. The screening of haploid seeds is an essential link of DH breeding. Oil marking method is more stable than other haploid marking methods, which makes the oil content of haploid and diploid seeds significantly different. Currently, it is the most mature way to measure the oil content of maize seeds by nuclear magnetic resonance (NMR) spectrometer. However, the slow speed, high cost, and large volume have restricted the wide applications of NMR spectrometer. Similar to NMR, near infrared spectroscopy (NIRS) can also be used for the non-destructive screening of seeds. Moreover, compared with NMR spectrometer, NIRS spectrometer has a fast speed, low cost, and small volume. This study attempts to explore a maize haploid seed screening solution based on NIRS quantitative analysis to meet the demand of large-scale DH breeding. In this study, a screening solution for Maize Haploid seeds based on the quantitative analysis of NIRS is proposed, and tests of the solution by two representative varieties show a consistent average accuracy above 90%. A model monitoring and calibration solution based on small sample set is proposed to guarantee the stability and efficiency of the screening solution. The reason for the poor separability of the non-embryo side diffuse reflectance spectra is analyzed, and the feasibility of the non-embryo side spectra is verified by analyzing the spectral information of hyperspectral data. In summary, this research provides a more economical and efficient solution for maize haploid screening.