Uniformly rooted young trees with 1?year?old stems and 2?year?old roots were established using the completely randomized block design of three blocks at Changping and Langfang in the springs of 2005 and 2008, respectively. For the field trials of Changping and Langfang, 10 to 20 and 25 individuals per clone were planted in each block at spacings of 3 m×4 m and 3 m×5 m. The diameter at breast height(DBH) and height were measured for each tree after growing seasons of the 5th and 6th year at two sites and the stem volume was calculated with the formula from the tree volume table. Growth traits were analyzed by multivariate analysis of variance. Two average sample trees were selected from each block per clone in the third growing season of plantation at Langfang. According to the top?down order of natural branching, the canopy was divided into 3 layers, i.e. upper, middle and lower canopy. A representative branch was selected from the first?order branches on the south side at each canopy position of each average tree. One mature leaf was sampled from the sixth to tenth branches at three different canopies of each sample tree of Langfang. Leaf anatomy (thickness of palisade and spongy parenchyma layers in total leaf) and its stomatal traits (density and length) were measured in the laboratory after collecting the leaves in the field trial. Correlation analysis and principal components analysis (PCA) were performed with 11 traits including the growth traits (DBH, stem volume) and leaf traits at the upper, middle and lower canopy position, separately.

The comparison of genetic parameters among growth traits at two sites showed that variation coefficient of traits at Changping (15%-47%) were more than those at Langfang (13%-30%) and the broad sense heritability and the repeatability at Changping were less than those at Langfang. The broad sense heritability (0.03) and the repeatability (0.24) of tree height at Changping were the minimum, while the broad sense heritability and the repeatability of other growth traits in two sites were greater than 0.32 and 0.82. The results of multivariate analysis of variance show that, the difference of growth traits among different clones was significant. There was significant difference of growth traits between two sites. In addition to the volume at 6?year?old, the interaction effect of other traits between clones and sites was significant or extremely significant. The growth traits of the same clone were significantly different between two sites, among which the DBH and stem volume of five clones (Br, Ti, La, Por and 108) were significantly different between two sites. The correlation analysis results showed that the DBH and stem volume were negatively correlated with abaxial stomatal density at three canopy positions, and were not significantly correlated with other leaf traits. The results of PCA between growth traits and leaf traits showed that the main planes of PCA (PC1, PC2, PC3 and PC4) of 11 traits established for upper, middle and lower canopy positions independently explained 85.5%, 87.2% and 90.7% of the variability, respectively. With PC1 × PC2 plane as factors, the first principal component was the index of growth character. The second principal component was positively correlated with leaf anatomical traits, and negatively correlated with leaf stomatal density and length of the lower epidermis. However, the contribution rate of leaf anatomical traits was the largest; therefore, the second principal component was the relevant index for leaf anatomical traits. Based on the results of PCA, the 13 clones can be divided into three categories: the first class clones (Por, 111, 107, 108, Me, 36, 50 and Be) have greater growth and smaller leaf thickness, leaf epidermis stomatal density, and sponge tissue thickness than the other classes. The classification between the second and third class varied from different canopies. According to PCA of growth traits and leaf traits at the upper and middle canopy, the classification can be divided into the second class clones (Ta and La) and the third class clones (Br, Ti and 109). According to PCA of growth traits and leaf traits at the lower canopy, the classification can be divided into the second class clones (Ti, 109 and La) and the third class clones (Br and Ta). The second and third class clones have smaller growth and greater adaxial stomatal density and spongy parenchyma thickness than the first class clones.

The 5? and 6?year?old DBH and stem volume of Section Aigeiros clones were mainly affected by genetic effects and general environmental effects. There were significant differences of growth traits between the sites and the interaction between the sites and the clones, indicating that the genotypes had strong interaction effect with the environment on the growth traits. Growth traits of five clones (Br, Ti, La, Por and 108) were more affected by environment than these of other clones. There was a significant correlation between DBH and tree volume with abaxial stomatal density. The results of these studies showed that the establishment of a joint selection system for multi?trait perennials was of great significance to the research on selective breeding of poplars.