Abstract:

This study aims to establish an evaluation method for identifying new maize varieties, selecting those with high and stable yields as well as wide adaptability, to accelerate the promotion and application of new varieties and support the sustainable and stable increase of maize production in Henan Province. 57 summer maize varieties were used as experimental materials. A comprehensive evaluation of varietal performance was conducted by analyzing the growth period, lodging resistance, yield, and related agronomic traits, integrating the high stability coefficient method, the Eberhart-Russell model, and the random forest algorithm. The average growth period of the tested varieties was 107.0 days, the plant height ranged from 225.5 to 280.8 cm, and the sum of lodging and lodging-breaking rates varied from 0.1% to 14.4%. The yield range was 9202.81 to 11093.58 kg/hm². Among them, ‘Chengyu 668’ had the highest yield, reaching 11093.58 kg/hm²; while ‘Dingyou 221’ had the highest high stability coefficient of 88.65%. Correlation and random forest model analysis showed that yield was significantly positively correlated with traits such as 100-grain weight and ear diameter. Seed yield had the greatest contribution to yield prediction in the random forest model, and the path coefficient of seed rate to yield was also the highest. Through cross-validation of the high stability coefficient method, four-quadrant analysis and Eberhart-Russell model, two varieties, ‘Keteng 918’ and ‘Xianyu 534’, with both high yield and stable yield were screened out, providing a scientific basis for optimizing the maize variety structure in Henan Province.

Key words: summer maize, new variety evaluation, high and stable yield, high stability coefficient, Eberhart-Russell model, random forest, path analysis, Henan Province