Objective To develop a near-infrared spectroscopy analysis model integrating the main quality traits of Arachis hypogaea L. (peanut), provide an efficient and convenient identification method for screening of peanut quality trait mutants, shorten the breeding process, and improve the breeding efficiency.

Method A total of 115 peanut germplasm materials were collected from major peanut-producing areas in China, 100 of which were utilised as a calibration set and the remaining 15 as a validation set. The Swedish Broadcom DA7200 near-infrared analyzer was used to collect spectral information. The fat content was determined by using the Soxhlet extraction method, the protein content by Kjeldahl determination, the total sugar and sucrose contents by acid hydrolysis-Rein-Einon’s method, and the content of each fatty acid by gas chromatography. The full wavelength spectral range was selected, and models were constructed by using partial least squares regression. The single and composite preprocessing methods were compared to select the best model that performed optimally under this spectral preprocessing by comparing the determination coefficients (R²) and root mean square errors of calibration (RMSEC) of different models. The 15 peanut germplasm materials in validation set were used for external validation of the optimal model for each trait. The mutants in the offspring of aerospace mutagenesis materials were screened by the best model to investigate the application value of the model.

Result A near-infrared spectral analysis model for 12 peanut quality traits was constructed. The R² of the traits was higher than 0.85, with the exception of fat and arachidic acid contents. External validation also demonstrated that the R² of the constructed model was greater than 0.85, with the exception of fat and arachidic acid contents. Furthermore, the oleic acid near-infrared analysis model screened 12 peanut oleic acid mutants from 805 aerospace mutagenic materials, with the oleic acid contents significantly higher than that of the wild type (P<0.001).

Conclusion The constructed model is an effective predictor of quality traits in peanuts and is suitable for the efficient detection of peanut kernel quality in mutants, germplasm resources and hybrid offspring populations.