The features of branching and growth studied included height, diameter at breast height (DBH), total number of branches, annual height growth, annual branch elongation in the year of elongating, annual branch number for four consecutive years, diameter of branches of different ages, and diameter of stem where branch-whorl originates. For features of total growth and overall branching, no significant differences were found between families, except for DBH. For annual features, no significant differences were found in annual stem height growth, annual branch elongation in the year of elongation and diameter of branches. In the last four years, differences in number of branches were not significant in the first two years but were significant in the last two year; differences in stem diameter where branch-whorls grow were significant for the four consecutive years. Trend of annual growth and branching features of families can be divided into three types as increasing type, stable type and fluctuating type. Most of families have an increasing type with respect of annual height growth and annual branch elongation, while most families belong to a fluctuating type with annual branch number. The results indicated that in the fifth year after planted in seedling seed orchard, differences between families were mostly insignificant. This result may have two main explanations: one is the growth rhyme in early ages of Masson pine, the other one is the complex paternal components to form the open-pollinated families. Family selection seemed to be not useful based on the result. It is suggested to select some of families in the nursery instead of to use all the families when establishing seedling seed orchards with open-pollinated families from plus-trees.
According to the genetic difference detected by means of RAPD markers, the methods of distinguish among 2 Liriodendron species and their interspecific hybrid were discussed, and the relationship between genetic distances of parents and their hybrids performance were analysed. Of the total 136 bands amplified with 25 10 mer primers, 3 presented in all 5 L.chinense sample trees, but absented in all 5 L.tulipifera sample trees. On the contrary, 8 presented in all 5 L.tulipifera sample trees, but absented in all 5 L.chinense sample trees. The 11 fragments could be detected in all the 6 hybrids sample trees. It was considered that 11 fragments could be used as markers in the identification. The range of genetic distance among 5 L.chinense sample trees, among the sample trees between L.chinense and L.tulipifera and between L.chinense and the hybrid were not overlapped. Therefore, the genetic distance between an unidentified tree and a L.chinense tree could also be used in the hybrid identification. The 16 sample trees were grouped correctly. So, the UPGMA (unweighted pair group methods of arithmetic (average))cluster analysis method was useful in the (hybrid) species identification. The mean height and basae diameter increment of families after the first growing season showed a significant guadratic cure correlation with genetic distances of parents. Therefore, the genetic distance had high potential for the choice of parents and the heterosis prediction in cross breeding of Liriodendron.
