Variation and Genetic Parameters of Leaf Morphological Traits of Eight Families from Populus simonii × P . nigra

: Leaf morphology in Populus L. varies extensively among sections, species and clones under strong genetic control. P. nigra L. (section Aigeiros ), with large and triangular leaves, is a commercial forest tree of economic importance for fast growth and high yield in Europe. P . simonii Carr. (section Tacamahaca ) with small land rhomboid ovate leaves performs cold and dry resistance / tolerance in the semi-arid region of Northern China. Leaf morphological traits could be used as early indicators to improve the e ﬃ ciency of selection. In order to investigate the genetic variation pattern of leaf morphology traits, estimate breeding values (combining ability), as well as evaluate crossing combinations of parents, 1872 intersectional progenies from eight families ( P. simonii × P. nigra ) and their parents were planted with cuttings for the clonal replicate ﬁeld trial in Northern China. Four leaf size traits (area, perimeter, length, width) and roundness were measured with leaf samples from the 1-year-old clonal plantation. Signiﬁcant di ﬀ erences regarding leaf traits were found between and among three female clones of P. simonii from Inner Mongolia, China and six male clones of P. nigra from Casale Monferrato, Italy. The genetic variation coe ﬃ cient, heritability and genetic variance component of most traits in male parents were greater than these of female parents. Heritability estimates of male and female parents were above 0.56 and 0.17, respectively. Plentiful leaf variations with normal and continuous distributions exited in the hybrid progenies among and within families with the genetic variation coe ﬃ cient and heritability above 28.49 and 0.24, respectively. Heritability estimates showed that leaf area was the most heritable trait, followed by leaf width. The breeding value ranking of parents allowed us to select the parental clones for new crosses and extend the mating design. Two male parental clones (N430 and N429) had greater breeding values (general combining ability, GCA) of leaf size traits than other clones. The special combining ability (SCA) of the crossing combination between P. simonii cl. ZL-3 and P. nigra cl. N430 was greater than that of others. Eight putatively superior genotypes, most combined with the female parental clone ZL-3, can be selected for future testing under near-commercial conditions. Signiﬁcant genetic and phenotypic correlations were found between ﬁve leaf morphology traits with the coe ﬃ cients above 0.9, except for leaf roundness. The results showed that leaf morphology traits were under strong genetic control and the parental clones with high GCA and SCA e ﬀ ects could be utilized in heterosis breeding, which will provide a starting point for devising a new selection strategy of parents and progenies.

having high densities and lengths at the abaxial leaf surface; cell expansion is the main reason for the increase in leaf area. On the other hand, the leaves of P. deltoides (section Aigeiros) were thin and small, and had stomata with low density and small stomatal length at the abaxial leaf surface; cell division is the main reason for the increase in leaf area [35]. Moreover, evolution of differential cellular polarity plays a significant role in leaf morphological variation observed in the subgenera of Populus [36]. The formation of leaf morphological characteristics is regulated by many functional genes' expression, transcription regulators and microRNAs [37][38][39]. The mechanism of poplar heterosis was discussed widely in terms of growth variation, photosynthetic capacity, water use efficiency, hormone content, genetic mapping and allelic variation [40][41][42][43][44]. However, there are few reports on the genetic variation of leaf traits, genetic relationship between hybrid progenies and parents and heterosis of these leaf morphology traits of poplar [16,45].
Accurate estimation of genetic parameters of leaf traits is of great significance to poplar breeding and selection. With the aim to realize the estimated genetic variance components without bias, the key is to eliminate the influence of environmental factors on genetic variance. In breeding practice, a large amount of non-equilibrium data is generated due to the limitation of measuring the number of offspring or preservation conditions; it is difficult to obtain accurate genetic parameters by classical genetic methods. The genetic model and statistical analysis method based on the mixed linear model can easily solve this problem even with a large population size, a complex population structure and unbalanced observed data [46,47]. With the development of computational methods and technology, a mixed linear model approach has become the main method of genetic evaluation and has been widely used in crop [48,49], tree [50,51] and animal [52,53] breeding. Breeding value, also known as additive effect value, is an important parameter in forest selection and breeding. Improving the prediction accuracy of tree breeding values can improve the breeding efficiency and readily to estimate genetic gain, but its value cannot be directly measured. Statistical methods can be used to dissect genetic effect and environmental effect through phenotypic values and the genetic relationships between individuals [54]. The Best Linear Unbiased Prediction (BLUP) method [55,56] can predict the value of random effects, effectively reduce the error variance of prediction breeding value, improve the accuracy of selection and make it possible to predict the breeding value of different generations of individuals. An accurate BLUP prediction value and forward selection based on an individual breeding value will improve the selection efficiency and genetic gain in the process of genetic improvement [57,58]. With the rapid development of genomics in recent years, the combination of genomics and quantitative genetics has a significant impact on the genetic gain of traits. It can accurately predict the phenotypes of candidate individuals in the early stages, while the prediction of multiple traits also shows great advantages. The combination of genomics and quantitative genetics, based on the accurate prediction of BLUP breeding value, laid a solid foundation for the estimation of forest genetic parameters, early selection and accurate evaluation of parents. The BLUP breeding value has been widely used in animal breeding and selection research. In recent years, breeding value estimation and genome breeding value estimation have also been applied to Eucalyptus nitens [59][60][61], Pinus elliottii [62], Paulownia [63] and Picea [64,65] with reference to animal breeding experience, but there are few reports on BLUP applications in poplar [58].
In this study, 1872 intersectional progenies of eight families (P. simonii × P. nigra) and their parents were used and measured leaf morphological traits (area, perimeter, length, width, roundness) with leaf samples from the 1-year-old clonal plantation. The aim of the study was to analyze: (1) genetic variation on leaf traits among and within eight families and their parental clones, (2) prediction on BLUP of leaf traits based on mixed linear model and (3) GCA and SCA on leaf traits of eight crossing combinations with the different parental clones. Such analyses were expected to provide a theoretical basis for analyzing leaf variation of the F 1 progenies between three P. simonii clones and six P. nigra clones, helping to explore the selection in early hybrid progeny and the combination of parental clones and breeding strategies.

Experimental Materials
In the spring of 2014, F 1 progeny seedlings of 8 families were obtained from the intersectional hybridizations between P. simonii and P. nigra and raised in the greenhouse of CAF (Chinese Academy of Forestry), Beijing, China. Three female clones of P. simonii ('1-XY', 'XY-5', 'ZL-3') were selected from the gene bank of P. simonii in Inner Mongolia, Northern China. Six male clones of P. nigra ('N188', 'N020', 'N139', 'N151', 'N429', 'N430') were selected and introduced from the gene pool of P. nigra in Casale Monferrato, Italy. After this, the seedlings of 1872 genotypes from 8 families (Table 1) were planted at the Tongzhou Base of RIF (Research Institute of Forestry) in Beijing, China. In the spring of 2016, cuttings of F 1 progenies and their parents were planted for the clonal replicate field trial at Fengnan Base of CAF in Tangshan City, Hebei Province. Five plants for each genotype of F 1 progenies and each clone of parents were grown in a randomized complete block design, one plot per block with spacing of 30 × 50 cm and conventional management with field practices.

Experimental Site
The experimental area is located at the Fengnan Base of CAF in Tangshan City, Hebei Province, Northern China and lies between 39 • 29 N latitude and 118 • 16 E longitude. Due to the semi-humid continental monsoon climate, its annual average temperature is 12.5 • C with a minimum of −9 • C in winter and a maximum of 32 • C in summer. The annual precipitation and the annual frost-free period are 596.4 mm and 190 days, respectively. The soil is sandy-loam type with a pH of 7.5.

Leaf Traits Measurements
In August 2017, three plants per genotype/clone performing with uniform growth and without disease and insect damage were selected as the average for subsequent measures for study use. One of the leaves from the sixth to the ninth from top to down of each plant were collected, then put into plastic bags with moistened filter paper and brought back to the laboratory. Each leaf sample was scanned with an HP ScanJetG4010 scanner and saved as JPG (Joint Photographic Experts Group) files. Finally, Digimizer (MedCalc Software bvba) software was used to measure leaf morphologic traits (length, width, area and perimeter) and calculate leaf roundness [66].

Statistical Analysis
All summary statistics were analyzed using the R program. The variance components were estimated by restricted maximum likelihood, using the average information REML algorithm implemented in the ASREML program. The analysis of variance components were tested according to the following general mixed linear model: where Y ij is the phenotypic value of the jth individual in clone i, µ is the overall mean, C i is the effect of clone i and e ij is the residual effect.
-For F 1 progenies: Y ijklm = µ + T k + F ij + C l(Tk) + e ijklm where Y ijklm is the observation of the mth ramet of the lth clonal plant within kth genotype from the ijth full-sib family, µ is the mean population and T k is the random effect of the kth additive genetic N(0, σ 2 A ), F ij is the random specific combining ability (SCA) effect of the ijth full-sib family~N(0, σ 2 SCA ), C l(Tk) is the fixed effect of the lth clonal plant within the kth genotype~N(0, σ 2 C(G) ), and e ijklm is the residual random error~N(0, σ 2 e ).
Individual-tree narrow-sense heritability (h 2 s ) was estimated for each leaf trait at each clone within each genotype using the individual animal models as follows: where σ 2 a was the additive genetic variance and σ 2 p was the phenotypic variance. The additive genetic coefficient of variation (CV A ) was calculated as follows: where σ a was the square root of the additive genetic variance for a trait and X was the trial mean for the trait. The correlation estimates were obtained using the above model in the multivariate formulation [59]. The correlations related to genetic and environmental effects between traits 1 and 2 were calculated with the following formulas: the additive genetic correlation: ρ a = (cov a(1,2) )/(σ a1 )(σ a2 ); the phenotypic correlation: where cov (1,2) is the covariance between traits 1 and 2 and σ 1 , σ 2 are the standard deviations of traits 1 and 2, respectively.
The best linear unbiased predictors (BLUPs) were computed by solving the mixed model equations using ASReml 3.5.3 software. Considering that the leaf shapes of these plants exhibited a large difference, Fisher's Least Significant Difference (LSD) test was used to carry out multiple comparisons of 5 leaf morphology traits among the parental clones and the 8 families. The individual BLUP for the male and female parents is called GCA, and for the full-sib families is called SCA. Both parameters were calculated to estimate the combination ability of crossing combinations between different parental clones so that favorable genes or characters can transmit to their progenies.

Genetic Parameters and Frequency Distribution of Leaf Traits in Parents and F 1 Progenies
Five investigated leaf morphology traits (area, perimeter, length, width, roundness) differed significantly between three female parental clones of P. simonii (section Tacamahaca) and six male parental clones of P. nigra (section Aigeiros) ( Table 2). Four leaf size traits of the male parents, P. nigra, had greater means than those of the female parents, P. simonii. The genetic effects were stronger in male parental clones, and the clonal variance components for leaf traits in the male parents were much higher than these of the female parents. The male parental clones varied from 56.4% of leaf roundness to 99% of leaf area, while the female parental clones varied from 2.4% of leaf roundness to 23.8% of leaf width. Leaf areas of the female parental clones had, significantly, the highest CV A (92.69%), followed by those of the male parental clones (81.13%), while leaf perimeter, length and width obtained similar values around 40% and that of leaf roundness was around 11%. Heritability estimates on four leaf size traits of the male parental clones were all above 0.96 while that of the female parental clones ranged from 0.173 (leaf area) to 0.238 (leaf width) as leaf size traits were weakly controlled by additive effects. F 1 progenies of eight families resulting from interspecific crossing of P. simonii × P. nigra showed the intermediate means of four leaf size traits between those shown by their two parents (Table 3). In F 1 progenies, the CV A of four leaf size traits were above 25%-leaf perimeter had the largest (76%). The estimated heritability of all five leaf traits was above 0.2, and leaf area had the highest heritability (0.74). The genetic variance components ranged from 79.49% for leaf roundness to 83.08% for leaf perimeter, which indicated an obvious additive genetic effect and strong heredity. Data from five leaf morphology traits measured in the F 1 progenies of eight families resulting from P. simonii × P. nigra exhibited a nearly normal distribution ( Figure 1, the data not shown here) with a kurtosis from 0.40 to 3.38 and skewness from −1.31 to 0.48, while that of leaf roundness showed the highest kurtosis (3.38) and the lowest skewness (−1.31). Among eight families, the F 1 progenies from two families, ZL-3×N188 and 1-XY×N139, showed the most concentrated probability distribution of five leaf traits. The F 1 progenies of each family exhibited the distribution, which biased their parent with higher leaf traits than another parent. The distributions of leaf size traits were biased toward the fathers while the genotypes with longer and wider leaves than their parents were isolated from the offspring population. On the contrary, the distribution of leaf roundness was biased towards the mother and the genotypes with larger leaf roundness than their parents were isolated from the offspring population.

Genetic Variation of Leaf Traits in Parents and F 1 Progenies
The results of ANOVA on five leaf morphology traits of the parents showed that the variations were considered significant or extremely significant among three female parental clones of P. simonii and significant among six male parental clones of P. nigra when the P value of the ANOVA F-test was below 0.05 or 0.01 (Table 2). In the same way, the variations of five leaf morphology traits of their offspring were extremely significant among eight families and among the F 1 genotypes within families (p ≤ 0.01) ( Table 3).
The multiple comparison Fisher's Least Significant Difference (LSD) test assessed the differences among the three female parental clones, six male parental clones and eight families after parametric ANOVA (Table 4). Five leaf morphology traits were significantly different among three clones of P. simonii and among six clones of P. nigra. The differences of leaf traits among three female parental clones of P. simonii were found that 1-XY had the largest mean leaf area, width and roundness and ZL-3 had the largest mean leaf perimeter and length. Among six male parental clones of P. nigra, N429 and N020 had the highest means regarding four leaf size traits in correspondence with the lowest mean of roundness (4.33 and 4.54), while N151 and N188 had the smallest mean of four leaf size traits. The difference of the F 1 progenies within eight families revealed that the progenies of two families, ZL-3×N429 and ZL-3×N430, had the largest means of leaf size traits in correspondence with the lowest means of roundness (5.94 and 6.12). The leaves of three families, 1-XY×N151, XY-5×N188 and 1-XY×N188, were significantly smaller than these of others and their leaf roundness showed the intermediate mean of the parents. The progenies of other three families, 1-XY×N020, 1-XY×N139 and ZL-3×N188, exhibited the highest means of roundness as well as the intermediate means of four leaf size traits.

Breeding Values and Combining Ability Analysis on Leaf Traits of Parents
The estimated results of breeding value (i.e., GCA) regarding leaf traits (Table 5) showed that the families with ZL-3 as their female parent had the largest GCA (0.45), and the families with 1-XY as their female parent had the smallest GCA (−0.36) regarding leaf area. GCAs regarding leaf roundness and width were the largest in the families of ZL-3 as their female parent (0.38 and 0.33, respectively). For the families with 1-XY as their female parent, the GCA regarding leaf roundness of 1-XY was the lowest (−0.53), followed by leaf area and width (−0.36 and −0.32). Among the families of six male parental clones of P. nigra, GCA regarding four leaf size traits (leaf area, perimeter, length and width) of N430 was the largest (1.36, 0.58, 1.43, 0.73), followed by these of N429 (0.52, 0.50, 0.87, 0.28). The families of the remaining four male clones had GCAs less than 0, while GCA regarding leaf roundness of the families with N020, N139 and N430 as their male parents was the largest (0.86, 0.87). Table 5. General combining ability (GCA) and special combining ability (SCA) on leaf morphology traits of 8 families resulting from P. simonii×P. nigra. Based on the mixed linear model, the breeding values of eight families (i.e., SCA) were obtained ( Table 6). The results showed that the highest SCA observed with respect to leaf area of crossing combination ZL-3×N430 (1.06) among the eight families. Among the families of ZL-3 with high GCAs, ZL-3×N430 (1.06, 0.33, 0.63) and ZL-3×N429 (0.63, 0.28, 0.41) had higher SCAs regarding leaf area, perimeter and width. For the trait of leaf length, the highest SCA appeared in the crossing combination ZL-3×N430 (0.81), but the GCAs of the families with ZL-3 as female parent were negative values. The same went for the leaf roundness. The highest SCA appeared in two crossing combinations, 1-XY×N020 (0.24) and 1-XY×N139 (0.24), which fully reflected the heterosis and indicated the greater degree of family differentiation obtained in this experiment. Two crossing combinations, ZL-3×N430 and ZL-3×N429, had the top two breeding values above 0 with regard to leaf size traits, which indicated that in these two families' leaf size traits were strongly controlled by heredity. 1-XY×N020 −0.03 1-XY×N020 −0.08 1-XY×N020 −0.13 1-XY×N020 0.01 XY-5×N188 0.09

Traits
The values in bold indicated that GCA or SCA regarding leaf traits were the maximum among the families.

Phenotypic and Genetic Correlations between Leaf Morphology Traits
The results of correlation analysis on five leaf traits indicated that both genetic and phenotypic correlation coefficients between leaf size traits (leaf area, perimeter, length, width) of hybrid progenies were greater than 0.910 significantly (Figure 2). Leaf roundness exhibited strong negative correlations with four leaf size traits which ranged from −0.439 to −0.233 (p < 0.001).

Phenotypic and Genetic Correlations between Leaf Morphology Traits
The results of correlation analysis on five leaf traits indicated that both genetic and phenotypic correlation coefficients between leaf size traits (leaf area, perimeter, length, width) of hybrid progenies were greater than 0.910 significantly (Figure 2). Leaf roundness exhibited strong negative correlations with four leaf size traits which ranged from −0.439 to −0.233 (p < 0.001).

Variation of Leaf Traits of the Parental Clones and F 1 Progenies
The leaf morphology trait in Populus varies extensively and significantly among sections, species and clones under strong genetic control [24,25,36]. Overall, the performance results together with large variation in leaf parameters showed that considerably higher genetic gain is possible with larger leaf size and higher resistance. In the study, intraspecific variation of leaf traits was significant between the female parental clones of P. simonii and the male parental clones of P. nigra. P. simonii (section Tacamahaca) with distribution of arid and cold areas in Inner Mongolia revealed the clonal variation of smaller leaves, while P. nigra (section Aigeiros) naturally distributed in humid and warm regions of Europe showed the clonal variation of larger leaves. Heritability estimates of four leaf size traits of the male parental clones (P. nigra) were greater than these of the female parental clones (P. simonii), indicating that the measured traits were under low genetic control in the female parents which contained small additive genetic variance components and were largely influenced by environment. High variable environmental conditions such as climate and soil in different sites exert a large influence on the phenotypic variation and plasticity of the parental clones with an adaptive advantage. In addition, the F 1 progenies of eight families from the crossing of three female parental clones of P. simonii and six male parental clones of P. nigra showed significant variation of leaf traits among families and among genotypes within families. Leaf size traits of the F 1 progenies were moderate as compared with their female and male parental clones while some progenies were larger than their parents with regard to leaf traits, indicating that the phenomenon of super heterosis was controlled by strong heredity [4,35,67,68].
Remarkable phenotypic variation of leaf shape and size reflects the natural selection operating function, which is considered to be the result of adaptation to a particular environment [3]. Leaves play a major role in photosynthesis, as the light capturing organ and the site of photochemical reactions, and are responsible for most of the carbon fixation in a plant and therefore are critical factors influencing plant success. To absorb sufficient light energy, leaves must be as wide as possible, while leaves must be as flat and thin as possible to facilitate gas exchange [69]. Leaf shape is considered as a highly complex trait controlled by a large number of loci, each contributing only a small effect; these loci likely act via modulation of gene expression [16] and hence the substantiated obvious additive effect of leaf traits. Drost et al. [36] found in their study of backcross population (BC 1 ) between P. deltoides and P. trichocarpa that leaf width and aspect ratio represented the vast majority of information on leaf shape variation in BC 1 which was the main difference of leaf shapes between their parents. In addition, the range of variation with regard to leaf shape of hybrid progenies mostly depended on the difference between their parents [4], which probably provides the perspective of utilization in the heterosis of breeding. Moreover, the data presented by Tsarev et al. [70] found that for different genotypes, even for the same tree species, the results of hybridization were different, and the difference of hybrids with different hybridization variants was significant. This indicates that parental differences as well as variation took a significant effect on the genetic variation of leaf morphology traits in the F 1 progenies. An optimistic outlook was retained on the abundant variation among progeny populations which would stabilize the genetic properties, substantiate greater selection potential and improve the efficiency and effect of breeding.

Combining Ability and Parental Evaluation of Parent Cross Combinations
Based on the mixed linear model, the GCA of the leaf morphological traits of eight crossing combinations between P. simonii and P. nigra were obtained by using the BLUP method. The breeding values (GCA) regarding leaf area, width and roundness of the female parental clone ZL-3 (P. simonii) were the largest and above 0.33. Similarly, the female clone 1-XY could pass on a large leaf perimeter to their offspring. Among the male parental clones of P. nigra, the GCA of leaf size traits of N429 and N430 were above 0.28, while those of N188 and N151 were below -1.18. The male parental clones, P. nigra N430 and N429, showed higher GCAs in cross combinations, ZL-3×N430 and ZL-3×N429. Interestingly, however, the female parental clones, such as ZL-3, showed lower GCAs but the highest SCAs of leaf length in the crossing combination with N430, which indicated a certain degree of heterosis and strong genetic control. By the breeding value ranking the families, two topcross combinations were selected, i.e., ZL-3×N430 and ZL-3×N429, whose leaf size traits had strong heritability. These two hybrid combinations can be used in poplar breeding and selection in the future.
We delightedly found widespread evidence of hybridization under natural conditions between the species of section Aigeiros and section Tacamahaca with high occurrence of hybrids characterized by rapid growth of section Aigeiros and diversified resistance of section Tacamahaca. Typically, high crossing compatibility among the interspecific hybridization between the species of both sections made it therefore become an important direction to carry out resource-saving resistance breeding for the diversity of ecological conditions in poplar cultivation areas in China. In terms of combining ability, assessment of GCA and SCA from the crossings of superior parental clones is probably more important than their own characters. The families and the clones should be focused on rather than poplar hybrid types and parental species alone [71]. There has been cases where the combining ability of the superior parental clones was high; their hybrids probably showed heterosis on the superior characters of parents [72][73][74]. Therefore, as long as the parents were carefully selected, even though the GCA of the parents was not high, the best effect of hybridization was probably obtained by combining with breeding target and using the crossing combination of the parents with high SCA [70].

Correlation and Early Selection of Leaf Morphological Traits between Parents and F 1 Progenies
Both the genetic and phenotypic correlation coefficients among four leaf size traits (leaf area, perimeter, length, width) of hybrid progenies were all significant and above 0.9 in the study. The strongest relationship among all correlations was presented between leaf area and length, at an extremely significant level. Moreover, leaf area is known to be an important functional trait, the relative importance of which is significantly correlated with leaf length and width [75], and even became a major determinant of leaf shape [28]. The proportional relationship between leaf area and the product of leaf length and width also generally remained stable during leaf evolution [75]. As the characters of early rapid growth, strong ability of photosynthetic and biosynthesis, and large production in a single growing season, poplar is extremely well suited to high-yield biomass production [35]. Strong positive correlations between leaf traits and plant growth indicated that leaf area development and leaf size were the robust indicator of biomass production in Populus [13,[76][77][78], therefore the larger their leaf size the greater potential is for larger growth. It has been widely confirmed that biomass yield is dependent on leaf number and leaf area, whatever the growth conditions of P. deltoides and P. euramericana as well as their hybrids [77,79,80]. On the other hand, the leaf increment rate and the largest leaf area appeared to be robust indicators of biomass production which can readily promote poplar growth regardless of the conditions and species [77,80,81]. Leaf area therefore can be used as an early indirect selection index for high yield of future poplar breeding and improving [12,14,82]. For further correlation studies, different and various leaf traits, e.g., leaf dry weight, leaf number, leaf thickness, specific leaf area, petiole traits, leaf anatomy and stomatal characteristics, allowed us to comprehensively and systematically analyze the correlation between leaf shape and growth which made leaf traits an alternative to precise and direct selection for biomass production.
Based on the mixed linear model, using the BLUP method to estimate the breeding value makes the selection of the poplar breeding more accurate and provides a more reliable theoretical basis for the selection and mating of hybrid parents and the design of breeding strategies. The results of leaf morphological traits in the study will be used as important phenotypic traits for constructing genetic maps in poplar and association analysis (GWAS), together with a recent boom in high through-put technologies. These can lay a foundation for the potential genetic and molecular mechanism which controlled leaf morphological variation as well as further mapping quantitative trait loci (QTL) [83].

Conclusions
In this study, intraspecific and interspecific variations of leaf morphology traits were significant between and among the parental clones of P. simonii and P. nigra. The F 1 progenies showed intermediate morphological characteristics between those of their parents, but they also displayed heterosis on leaf traits under strong genetic control. Leaf morphology traits were under strong genetic control. The heritability estimates showed that leaf area was the most heritable trait, followed by leaf width. The significant genetic and phenotypic correlations were found between five leaf morphology traits with coefficients above than 0.9, except for leaf roundness. Based on the mixed linear model, two male parental clones (N430 and N429) had the largest breeding value (GCA) for leaf size traits. The SCA of the crossing combination between P. simonii cl. ZL-3 and P. nigra cl. N430 was greater than that of other crossing combinations.