Abstract: 【Objective】GYT biplot analysis was employed to analyze agronomic traits, yields and cultivar-trait interactions of test proso millet varieties in different trial sites and comprehensively asses and classify these test varieties in term of their multiple traits, so as to provide a theoretical basis for reasonable arrangements and zonations of proso millet varieties in China.【Method】GGE biplot, GT biplot and GYT biplot along with the analysis of correlation were employed to comprehensively assess 20 varieties tested in the State Multi-region Trial for Proso Millet in 15 trial sites in 2019-2020 in terms of their growth period, plant heights, node numbers, main spike lengths, grain weights per spike, 1000-kernel weights and yields.【Result】It was showed by the analysis of correlation that in both non-waxy and waxy proso millet varieties, the yields were significantly and positively correlated with the growth period and negatively correlated with the main spike lengths, and the plant heights was significantly and positively correlated with the growth period and main spike lengths. In non-waxy proso millet varieties, the yields were also significantly negatively correlated with the node numbers, and the node numbers and 1000-kernel weights were significantly and positively correlated. In waxy proso millet varieties, the yields were significantly and positively correlated with the 1000-kernel weights, and the growth period were significantly and positively correlated with the plant heights and 1000-kernel weights. The analysis of genotype-trait interactive effects by GT biplot showed that the main components of GT biplot PC1 and PC2 explained 61.81% and 69.96% of genotype-trait interactive effects in no-waxy and waxy proso millets, respectively. The correlations between agronomic traits of the tested varieties displayed by GT biplot were basically consistent with Pearson correlation coefficients. The correlations between the yield-trait combinations of the tested varieties were analyzed by GYT biplot, and all the traits of these combinations were significantly and positively correlated. In terms of calculated ideal indexes, Yi 11-02-92-4, Gu 19-63, 0515-2-2, Yi 11-03-3-2-2 and Zhenglongmi 1 were identified as varieties with a good yield-traits combination performance.  In terms of comprehensive performances, Chimi 3, Xinong 2018-N02, Xinong 2018-N10, Y1660, Gu 15-173 and Xinong 18-W06 poorly performed. It followed that Yi 11-02-92-4 and 0515-2-2 had wider adaptabilities and better yields than the other varieties in different planting areas, showing an absolute regional yield advantage. 【Conclusion】In assessing multiple traits of millet varieties, GYT biplot analysis was more reliable than GGE and GT biplot analysis, so that it was an effective method to scientifically assess merits of proso millet varieties. Among the varieties tested in the state multi-region trial for proso millet, the no-waxy variety with the best comprehensive performance was Yi 11-02-92-4 and suitable to be planted in spring planting areas of proso millet in Northeast China and spring and summer-planting areas of proso millet on the Loess Plateau. The waxy variety of 0515-2-2 with a better comprehensive performance was suitable to be planted in spring-planting areas of proso millet in northern China and spring- and summer-planting areas of proso millet on the Loess Plateau. 

Key words: GYT biplot, ideal index, proso millet, multi-region trial, yield-trait combination