Developing rapid and efficient methods for assessing quality traits in foxtail millet germplasm is crucial for identifying superior genetic resources. In this study, we analyzed 657 foxtail millet germplasms from diverse ecological regions, both domestic and international. The double-wavelength method was used to determine amylose, amylopectin, and total starch contents in seeds. Of these, 550 germplasms were selected to develop near-infrared spectroscopy (NIRS) models for predicting amylose, amylopectin, and total starch contents using Unscrambler X 10.4 chemometric software. Standard normal variate combined with scatter correction and first derivative transformation were applied for spectral preprocessing, and partial least squares regression was used to construct the predictive models. The results showed that amylose content across the 657 germplasms ranged from 2.99% to 22.40% (mean: 16.25%), amylopectin content from 52.77% to 76.09% (mean: 59.56%), total starch content from 62.53% to 83.31% (mean: 75.81%), and the amylose-to-amylopectin ratio from 0.04 to 0.40 (mean: 0.28). Among all tested germplasms, foreign accessions exhibited the highest coefficients of variation (CVs) for amylose (30.08%) and total starch (5.07%). Compared to domestic germplasms, foreign germplasms had lower average amylopectin (59.20%) and total starch (75.19%) contents, with ranges of 54.65%–65.76% and 64.65%–82.38%, respectively. Significant differences in starch content were observed among foxtail millet germplasms from five domestic ecological regions. The Inner Mongolia Plateau region exhibited the highest CVs for amylose content (29.40%), total starch content (4.07%), and the amylose-to-amylopectin ratio (30.77%) among all domestic regions. The highest CV for amylopectin content (6.00%) was observed in the Northeast Spring Millet region, whereas the Southern region exhibited the lowest CVs for amylose (8.21%), amylopectin (4.40%), total starch (2.97%), and the amylose-to-amylopectin ratio (10.71%). Germplasms with the highest amylose-to-amylopectin ratios and amylopectin contents were primarily from the North China Summer Millet region, Loess Plateau region, and Northeast Spring Millet region. Notably, Ermaojian from the Loess Plateau region had the highest amylose-to-amylopectin ratio (0.40) and amylose content (22.40%), while Banmanghonggu from the North China Summer Millet region had the highest amylopectin content (76.09%). The NIRS models developed for amylose, amylopectin, and total starch contents achieved calibration correlation coefficients of 0.910, 0.848, and 0.717, respectively; cross-validation determination coefficients of 0.902, 0.830, and 0.675; and external validation determination coefficients of 0.903, 0.826, and 0.702. The standard errors of calibration were 1.156, 1.234, and 1.367, while the root mean square errors of cross-validation were 1.208, 1.288, and 1.471, and the root mean square errors of prediction were 1.130, 1.260, and 1.649, respectively. The ratio of performance to deviation for external validation was 3.415, 2.539, and 1.765, with optimal factor numbers of 9, 10, and 10, respectively. This study highlights the substantial variation in starch content among foxtail millet germplasms from different ecological regions, both domestically and internationally. The NIRS models developed here are effective for predicting amylose and amylopectin contents in foxtail millet. However, further refinement is needed to improve the accuracy of total starch content predictions.
WeChat
WeChat