Abstract: In order to predict the fiber quality of cotton through agronomic traits, Xinjiang Line 9 (G. hirsutum L.), Xinhai 16 (G. barbadense L.) and their backross populations (BC₄F_2:4) were used as research materials in this study, agronomic traits and fiber quality traits of 165 cotton accessions were investigated and identified using correlation analysis, principal component analysis, regression analysis and path analysis for exploring the relationship between agronomic traits and fiber qualities. The results showed that the phenotypes of the population were abundant, and the coefficients of variation on 12 traits were ranged from 0.11% to 48.14%, of which boll number per plant was the largest. Correlation analysis showed that fiber length was significantly positively correlated with plant height, effective fruit branch number, boll number per plant and effective boll number per plant, while micronaire was significantly negatively correlated with these agronomic traits, and was extremely significantly positively correlated with lint percentage. Fiber strength was significantly positively correlated with height of the first branch, and fiber uniformity was significantly positively correlated with plant height and effective fruit branch number. By principal component analysis, four principal components were extracted and the cumulative contribution rate was reached 75252%, including cotton boll number correlation factor (PC1), fiber quality selection factor (PC2), low quality control factor (PC3) and int percentage and micronaire comprehensive factor (PC4) for selection. The stepwise regression equation showed that number of the first branch and effective fruit branch number had no effects on fiber quality. The results of path analysis indicated that fruit branch number had a negatively direct effect on fiber length, and boll number per plant had a positively direct effect on fiber length, both boll number per plant and height of the first branch had positively direct effects on cotton fiber strength, effective boll number per plant had negatively effect on micronaire, the lint percentage had direct impact on it, and fiber uniformity was only directly affected by plant height. Among them, the indirect effects of plant height, boll number per plant and effective boll number per plant on fiber quality were dominant. Therefore, the indirect selection of fiber quality could be realized through the selection of agronomic traits, which provided a theoretical reference for the efficient selection of high-yield and high-quality new cotton varieties in Xinjiang and surrounding areas.

Key words: Gossypium barbadense L., Gossypium hirsutum L., agronomic trait, fiber quality, correlation

摘要： 为了通过农艺性状来预测棉花纤维品质，以新疆陆地棉系9和海岛棉新海16号及其回交自交后代BC₄F_2:4群体为研究材料，对165份材料的农艺性状和纤维品质性状进行鉴定，利用相关分析、主成分分析、回归分析和通径分析等方法，探讨群体主要农艺性状与纤维品质的关系。结果表明,海陆回交群体的表型非常丰富，12个性状的变异系数为0.11%～48.14%，其中，铃数的变异系数最大。相关分析表明，株高、有效果枝数、铃数和有效铃数与纤维长度和衣分呈显著正相关，与马克隆值呈显著负相关；始节高与纤维强度呈显著正相关；株高和有效果枝数与纤维整齐度呈显著正相关。主成分分析提取了4个主成分，累计贡献率达75.252%。其中，PC1为成铃数相关因子，PC2为纤维品质选择因子，PC3为低品质控制因子（株高和纤维品质），PC4为衣分与马克隆值选择的综合因子。逐步回归方程表明，始节数和有效果枝数对纤维品质的影响较小。通径分析表明,果枝数对纤维长度产生负向直接效应；铃数对纤维长度产生正向直接效应；铃数和始节高对棉花纤维强度产生正向直接效应；有效铃数对马克隆值产生负向直接效应，而衣分对其产生正向直接效应；纤维整齐度只受株高的正向直接效应；株高、铃数和有效铃数对纤维品质的间接影响占主导地位。因此，可通过对农艺性状的选择实现对纤维品质的间接选择，为新疆及周边地区高产优质棉花新品种的选育提供理论参考。

关键词: 海岛棉, 陆地棉, 农艺性状, 纤维品质, 相关性