Abstract: The genetic variation of 18 traits of leaf phenotype and photosynthetic physiology in F1 progeny（including 183 individuals）of interspecific cross between Castanea mollissima and C. henryi was studied by using the statistical methods of discrete character description，Pearson correlation analysis and cluster analysis. The results indicated：the variation coefficients of 18 traits for leaf phenotype and photosynthetic physiology in F1 generation ranged from 6.51% to 32.92%. The variation coefficients of 11 traits，such as petiole length and stomatal conductance，were all above 20%，which indicated that there were extensive genetic variations in F1 population. The values of leaf phenotypic traits and photosynthetic physiological characters of 183 individuals in F1 generations were mostly distributed between that of their parents，and showed a good trend of continuous normal or skewed distribution. The mid-parent heterosis rate of 18 traits ranged from–33.97% to 19.64%. Seven traits such as leaf length showed positive mid-parent heterosis，while the other 11 traits showed negative mid-parent heterosis. Further，two and six traits showed positive and negative over-parent heterosis respectively，while the other 10 traits did not show over-parent heterosis. For 153 pairs of correlations produced by 18 traits，there were 85 and 8 pairs of correlations reached extremely significant level（P < 0.01）and significant level（P < 0.05），respectively. The 183 offsprings and their parents were divided into three groups at genetic distance 2.33. At genetic distance 1.2，the second and third groups were divided into two subgroups respectively. The clustering results reflected the characteristics of each group，and the hybrid offspring could be selected more accurately according to the results of classification and target traits. The results provide important theoretical basis for early selection and breeding of chestnut hybrids.

Key words: Castanea, interspecific hybridization, heterosis, genetic variation, phenotypic trait, photosynthesis