中文摘要：

以5个高粱不育系材料为母本, 18个优质苏丹草材料为父本, 按照遗传交配设计（NCII）配制成90个杂交组合。分别在内蒙呼市和包头两地, 利用与高丹草产量相关的QTL位点标记检测亲本间的遗传差异, 并将F1的8个性状表型值对亲本材料进行标记位点的筛选, 建立标记效应和标记型值估算体系。估算特异性位点对性状表现的效应及杂种标记型值, 进而分析杂种标记型值与杂种表现的相关性, 应用逐步回归分析法建立8个性状杂种表现的预测模型, 并通过Jackknife抽样技术检测模型的精确度和稳定性。结果显示, 在分别考虑显性、加性作用下8个性状的标记型值与表型值的相关系数平均为0.65, 各性状的可决系数较大（0.51~0.88）, 而且两地结果趋势一致, 表明该预测模型稳定性强, 精确度较高。该模型对高丹草的杂种表现预测以及亲本选配都具有一定的指导意义。

英文摘要：

Molecular markers-based heterosis prediction can provide some advices for Sorghum ×Sudan grass breeding. The 90 cross combinations were made according to the North Carolina design II （NCII） with five sorghum sterile lines as maternal parents and 18 Sudan grass lines as paternal parents. The field trials were carried out on the farms in Hohhot and Baotou to evaluate environmental effects. The parental genetic differences were compared by yield related QTL markers. The phenotypic values of eight traits for all F1 hybrids were investigated. The values were used for selecting significant marker loci on parental lines for setting up the evaluation system of marker effect and marker value. The specific loci were used for evaluating the trait effects and hybrid marker value, and for analyzing the correlation between the marker value and heterosis. The prediction models of the eight traits for the hybrid were constructed with the stepwise regression analysis. The Jackknife sampling method was used to test the accuracy and stability of the model. The result indicated that, considering dominance and additive effect separately, eight traits showed the average correlation coefficient of 0.65 between vigor value and phenotypic value. The coefficient of determination was from 0.51 to 0.88 in the eight traits. The results in two places were similar. The model could be instructive for heterosis prediction and parents selection.