*4.4. QTL Mapping and Statistical Analysis*

Using the high-density genetic map, QTL analysis for GPC was performed by R/qtl software with the method of composite interval mapping (CIM) [33], taking the two years for the RIL population as two environments. A threshold of LOD > 2.0 was used for detecting a putative QTL. Using the low-density genetic map, QTLs controlling GPC were identified by Windows QTL Cartographer 2.5 [34]. Identification of QTLs was performed using CIM. The LOD threshold for claiming a QTL was also fixed as a LOD score of 2.0. QTL analysis in the three RH-derived populations was also conducted by CIM using Windows QTL Cartographer 2.5 [34].

Basic descriptive statistics, including mean, standard deviation, coefficient of variation, range, skewness and kurtosis, and correlation analysis for GPC in the RIL population were performed with Microsoft Excel 2016 for Windows. The *t*-test of the two parents was performed with the SAS program [35].
