将植物数量性状的主基因-多基因遗传体系检测的分离分析方法拓展到回交自交系群体。建立了包括无遗传控制、只有多基因控制、1～3对主基因控制和1～3对主基因＋多基因控制遗传模型下的混合分布函数，并由ML和ECM算法估计分布参数，以最大熵（最小AIC）准则和适合性测验选择最优最适遗传模型，进而利用所估计的分布参数估算遗传参数。在方法推演基础上通过1个模拟数据例题说明其应用。
A great number of genetic studies of quantitative traits, especially of QTL marker analysis, indicated that there existed both major genes and minor genes in a quantitative trait genetic system, not necessary all being minor genes even with equal effects. Gai et al. (2003) indicated for a QTL system, major gene plus minor gene model was the general model, while pure major gene model, or pure minor gene model was only the specific case of the general model. Based on it, they established the procedures of segregation analysis of quantitative trait to detect the genetic system. Among the genetic materials used, the permanent population, such as RILs is preferred in replication tests for precisely detecting QTLs. In the present paper, the segregation analysis was extended to backcross inbred line population (BIL). The procedures were as follows: At first, mixture distribution functions were established for the following genetic models: non-genetic, polygene, one major gene, two major genes, three major genes, one major gene plus polygene, two major genes plus polygene and three major genes plus polygene (Table 1). Then, the maximum likelihood method and ECM algorithm were used to estimate the parameters of component distributions. Furthermore, the best optimal genetic model was chosen among the possible genetic models through maximum entropy or Akaike’s information criterion and a set of tests for goodness of fit. Finally, the genetic parameters were computed from the maximum likelihood estimates of component distributions of the chosen genetic model (Table 4 and 5). An example from a simulated complete random block experiment with BIL population and their two parents is given for explanation of its usefulness (Table 7 and Fig.1).
全 文 ：