Genetic Variation in Growth and Cone Traits of Pinus Koraiensis Half-Sib Families in Northeast China

: Genetic parameters were evaluated for growth and cone characteristics (tree height, diameter at breast height, volume, cone number, thousand seeds weight and single cone seeds weight) on 86 half-sib families of Pinus koraiensis aged 31 years. Analyses of variance revealed significant differences ( p < 0.001) in all growth and cone traits among families while no significant differences were detected among blocks and the interaction between blocks and families. The average family values for growth traits were 17.22 m, 8.67 cm and 0.43 m 3 for tree height, diameter at breast height and volume, respectively. The average cone number, thousand seeds weight and single cone seeds weight were 17.57, 748.91 g and 77.25 g, respectively. Genotypic additive variance and phenotypic variances ranged from 0.00009 to 3.820 and from 0.0005 to 23.066, while genotypic and phenotypic coefficients of variation ranged from 2.693% to 37.196% and 4.963% to 60.595%, respectively. Heritability at the individual and family level ranged from 0.152 to 0.215 and 0.611 to 0.862, respectively. Growth traits were significantly positively correlated with each other, but cone traits showed a weak correlation with growth traits. Based on 10% selection rate, nine families each were selected as elite materials in terms of high performance in volume and cone numbers, with 22.16% and 43.82% genetic gain in volume and cone number, respectively. These results provide beneficial information to select excellent families and establish orchards of P. koraiensis from improved seeds. heritability was observed for growth and cone traits at the individual and average/family level. This study identified volume and cone numbers per tree as indices in the selection and evaluation of families. Thus, 18 elite families were selected based on 10% selection rate, with 22.16% and 43.82% genetic gain in volume and cone number, respectively. These results provide beneficial information to select excellent families and establish orchards of P. koraiensis from improved seeds. The families that were selected could be used for reforestation, and female parents would be useful for establishing improved seed orchard.


Introduction
Pinus koraiensis Siebold & Zucc is a dominant tree species in mixed-climate broadleaf-conifer forests at the Xiaoxing'anling mountain, Zhangguangcai ridge, Laoye ridge, and Changbai Mountains in northeastern China [1,2]. The distribution area extends to the northern regions of North Korea, the central region of Japan and the southern part of Far East Russia [3,4]. Because of natural P. koraiensis forests' edible nuts and excellent timber quality, these forests have been overexploited [5], particularly during the years 1975-1988, which was characterized by large-scale timber harvesting and wood production by the clear-cut method practiced in China [6]. The devastated natural areas have been offset by artificial forests [7].
Due to poor quality production in certain artificial forest stands [8], slow growth and late sexual maturity of this species [9], there have been efforts to improve P. koraiensis for decades in China. Thus, a series of studies were conducted at different stages of improvement program of this species. As an example, the selection of plus trees and different provenances in natural forests [10,11], the implementation of seed orchards [12] and selection of superior families and parents [13][14][15]. Because of high demand for timber in past several years, the improved genotypes of P. koraiensis have been selected based on growth traits and wood properties [16]. But according to the economic value of seeds and the protection project of natural forest, the cone yield turned into the main objective in P. koraiensis breeding in recent years.
Quantifying genetic variation between generations is a fundamental element in the tree improvement process [17]. For P. koraiensis, though several seed orchards were established and a large number of progeny test plantation have been established, rigorous statistical analyses of genetic variation of these seedlings were rare [18,19]. Breeding objectives determine breeding methods, and growth traits and wood properties are the most important traits for breeder's consideration when elite families or clones are to be selected [20]. For evaluation of genetic differences, most research focus on several growth traits or together with wood properties by comprehensive evaluation methods [21][22]. Though the seed yield has more importance than wood properties due to higher economic values, the cone trait had rarely been investigated by breeders in recent years. In this study, growth and cone traits of 86 half-sib P. koraiensis families at the age of 31 years were investigated, and the main objectives were (1) to investigate the genetic variation of each traits among different families; (2) explore the relationship between growth traits and cone traits; (3) evaluate and select elite families according to their growth and cone traits, which can provide a theoretical basis for improved variety breeding and upgraded seed orchard.

Site Description
The experiment was conducted in Naozhi Forestry Seed Orchard (41°05'N, 126°06'E), located on the western hillside of Changbai Mountain, Linjiang City, Northeast China. The climate type is temperate monsoon, and the average frost-free season, average elevation, mean annual precipitation and mean annual temperature were 128 days, 510 m, 744 mm and 8.2°C respectively. The soil type in this region is dark brown soil of silty texture with varying proportion of sand (15.13%), Silt (63.31%) and Clay (21.56%) [23]. Pure P. koraiensis monoculture plantations and broadleaf mixed forest are typical at this region at the foot of the Changbai Mountain [24].

Trait Measurement
Different traits were measured on living plants of each family on September 2015 at the age of 31 years. The traits measured included: Tree height (Ht), Diameter at breast height (DBH), Cone numbers per plant (CN), thousand seeds weight (TSW) and single cone seeds weight (SCSW). Tree height was directly measured by the Vertex Laser instrument (Haglof, Sweden), and diameter was measured with a circumference ruler. Cone number per tree was directly counted, while the single cone seedS weight and one thousand seeds weight were measured with a high precision balance. The volume (V) of trees was estimated using the formula proposed by Huber for conifers with a form factor of 0.41 [25].

Statistical Analyses
The IBM SPSS statistics software version 20 [26] was used to perform the statistical analyses. The significance of fixed effects (Eq. (1)) was tested by analysis of variance (ANOVA) F test [27][28]. The linear model used to calculate the performance of individual trees in different families is shown below [29]: where ijk X is the performance of k individual trees from i family growing in block j; µ is the overall mean; i α is the effect of family i; j β is the effect of block j,    The phenotypic coefficient of variation PCV (%) and genotypic coefficient of variance (GCV) were estimated by the formula below [31].  (8) where was mean value of the trait.
The trait relationships were estimated by Pearson's correlation coefficient, ) (xy r A , which was derived from the covariance of related traits ) , ( y x P COV divided by the product of multiplying their respective variances The best linear unbiased predictors (BLUP) of families were calculated according to Eq. (10) [33], which was used to obtain the general combining ability (GCA) of parent trees in order to deduce the family's breeding values according to Eq. (11) [29].
where and were the family phenotypic value and the average of offspring population from the same family, respectively.
where was the selection difference between the selected families performance and 2 HS h was the heritability at the family level as defined above.

Average Values for Growth and Cone Traits
Significant differences in growth and cone traits were detected among P. koraiensis families (P < 0.001). The other source of variance (block and the interaction of block and family) had shown no significant differences for all traits analyzed. The average values of different traits of all families are shown in Tab. 2. The overall average values for growth traits were 17.43 cm, 8.73 m and 0.044 m 3 in Ht, DBH and V, respectively. The overall means for cone traits were 17.71, 750.32 g and 77.28 g for cone number per plant, thousand seeds weight and single cone seeds weight, respectively. The highest average values differed from the lowest average values by 17.78 cm, 3.49 m and 0.66 m 3 for DBH, Ht and V, respectively. Three families (F84, F65 and F56) had higher average in DBH, two families (F56 and FA079) had higher average in Ht and family F56 had higher value in V. Eight families (F66, F55, F35, F128, FA093, F106, F25 and F81) showed particularly low average values in DBH, Ht and V. The highest cone number per plant was observed in two families (F38 and F107), and two families (F38 and F48) also showed higher thousand seeds weight than others. The differences between the highest and the lowest average value in cone number, thousand seed weight and single cone seeds weight were 33.67, 130.90 g and 23.30 g, respectively.

Genetic Parameters
The genetic parameters for all traits in all families are given in Tab. 3. For growth traits, the higher and lower phenotypic and genotypic variances were observed in DBH and V, which ranged from 4.601% to 42.511%. The higher and lower genotypic and phenotypic coefficient of variation in cone traits was observed for TSW and CN, which ranged from 2.693% to 60.595%. Cone number showed higher genotypic and phenotypic variances. The highest and lowest heritability at the individual and average/family level were observed for CN and Ht, respectively. Heritability ranged from 0.15 to 0.21 and 0.61 to 0.84 at the individual and average/family levels, respectively.

Correlation Analysis Between Traits
The correlation coefficients between different traits were determined. All cone traits (CN, TSW and SCSW) showed weak positive or negative correlation with growth traits (Ht, DBH and V) and among each other (r = -0.026-0.101). TSW and SCSW were negatively correlated to Ht and DBH, respectively whereas CN was negatively correlated with DBH and V. Correlation coefficient was also significantly positive between DBH and V (r = 0.840), while a weak significant positive correlation was noted between DBH and Ht (r = 0.226).

Breeding Values and Genetic Gains of Elite Families
With 10% selection rate, 9 elite families (Tab. 4) were selected for stem volume (F51, F102, F68, F65, F127, FA085, F130, F126 and F72) and cone number (F38, F107, F11, F108, F55, F59, F2, F130, F57) each. The range of the breeding values for the selected families was from 0.285 m 3 to 0.360 m 3 and from 15.10 to 17.43 in volume and cone numbers per plant, respectively. Family F130 stood out as the best family in terms of high-performance value in both volume and cone numbers per plant (Tab. 4). The genetic gains from selection of families are shown in Tab. 5, which ranged from 5.20% to 22.16% and 10.12% to 43.82% in growth and cone traits, respectively. The V and CN had the highest genetic gain among growth and cone traits. Among all traits, Ht had the lowest genetic gain, with 5.20% genetic gain.

Discussion
Each offspring in a half-sib family has unique phenotypic characteristics because of different genotype and different microenvironment, which may lead to a certain amount of variation among trees within the same family [35]. The analysis of variance helps to characterize this variation and enables selection of genetic resources in tree improvement research [17]. In the present study, significant differences (P < 0.001) were observed in all growth and cone traits among families whereas blocks and the interaction of block and family showed no significant differences. The lack of block effect may be due to the fact that all blocks closely resembled with the environmental and soil conditions of the region. Similar results have been reported for Larix olgensisand [36] and Cornus wilsoniana [37], which indicated that the selection of elite families was feasible.
Average family performance values of different traits were used to establish successive order in tree selection according to their ranks [38]. In the present study, the analysis on growth and cone traits of 86 P. koraiensis families showed great variation in average trait values among families but weak variation within families in different blocks, which was more favorable for family selection than individual selection. Although families had different mean values for various traits, with a default rounding, the growth measures presented approximate means for all growth and cone traits. In fact, the weak variation observed within trees in families led to the uniform pace in growth at the individual level. This finding could be due to the slow growing speed of trees at the maturity age [39][40], which also explains the slight decreasing trend of the difference between the maximum and minimum value observed in DBH.
Genetic variation is the basic element for evaluation and selection of excellent materials in forest tree improvement programs [41]. Genotypic variation reflects the difference between trees within families, while phenotypic variation indicates the difference between families. The higher value of these two coupled variations can enable a selection of the best descendant [42]. The phenotypic coefficients of variance were slightly higher for all traits compared to the genotypic coefficient of variance, generally because deviations from the averages in traits was greater among families than between trees of the same family sharing a half genotype although with different microsites [43] which showed similarity trend emerged in Quercus robur [44] and Larix olgensis families [45]. The narrow-sense heritability ( ) as a measure of the relative amount of genetic control for a given trait in population [46] expresses the fraction of the phenotypic variance that is accounted for by the variance among the breeding values of trees [47]. In this study, all growth traits showed moderate to higher genotypic heritability at the individual and family level that ranged from 0.152 to 0.210 and from 0.611 to 0.862, respectively. This finding shows that the phenotypic measurement completely reflected the underlying breeding value of trees and the selected families might be less influenced by environmental effects [48].
Correlation coefficients between growth traits allow the understanding of relationships among measured traits. The correlation between different growth traits guides the selection of trees based on multiple traits in research on tree improvement [49]. In this research, traits related to cone and seed were weakly negatively correlated to growth traits which indicated that the improvement objectives for wood and seeds would be equally manageable. Thus, the selection of elite families would be evaluated for growth and cone traits independently. Similar weak correlation between cone and growth traits was observed for Jatropha curcas L [42] but significant positive correlation was reported for Argania spinosa [28] and Pinus palustris [50]. The reason may be related to tree spacing used in these studies, which impacts growth in diameter and height of trees as well as the widening of the branch and consequently on the family's number of cones. There exist higher correlation coefficients between V and Ht, which were higher than correlation between DBH and V, DBH and H which also attributed to wider spacing, leading to better growth in height than diameter [51]. The correlation results corroborate those of Liang [52], who found strong correlation (0.899) between V and DBH on P. koraiensis clones.
Multiple-trait comprehensive evaluations had been developed for breeding cultivation materials with strong integrated ability, which are always used to select excellent families or clones [53]. An appropriate comprehensive evaluation method and suitable traits need to be selected for research because taking so many traits into consideration may decrease genetic gain [54]. In this study, two most important traits of P. koraiensis were volume and cone numbers, but due to weak correlation between these two traits, the selection of elite families were conducted for these two traits independently. With 10% selected rate, nine families each were selected as elite families based on volume or cone numbers. The genetic gain in volume in our study was higher than those reported for P. koraiensis clone by Liang et al. [52], as also for other tree species, such as Coastal Douglas [55], poplar clones [46] and Larix olgensis families [45], which showed an effective selection process. The genetic gain in cone traits in our study was also higher than the research by Jiang [55] and lower than the research of Jiang [56] on P. koraiensis under the same selection rate, which may be related to differences in age, environment and stand density. In all, the level of genetic gain depends on the repeatability, amplitude of genetic variation and selection rate, and most breeding programs aim to increase genetic gain via selection [57]. The families that were selected in our research could be used for forestation and female parents could be used in improved seed orchard establishment.

Conclusions
The study revealed significant differences in growth and cone traits among half-sib families of P. koraiensis, which enabled selection of elite families. Positive significant correlations were observed among growth traits, but cone traits were weakly correlated with growth traits, suggesting independent selection of elite families for these two traits. Phenotypic and genotypic variances were slightly lower for growth traits than for characters related to reproduction. Moderate to higher heritability was observed for growth and cone traits at the individual and average/family level. This study identified volume and cone numbers per tree as indices in the selection and evaluation of families. Thus, 18 elite families were selected based on 10% selection rate, with 22.16% and 43.82% genetic gain in volume and cone number, respectively. These results provide beneficial information to select excellent families and establish orchards of P. koraiensis from improved seeds. The families that were selected could be used for reforestation, and female parents would be useful for establishing improved seed orchard.
Disclosure Statement: No potential conflict of interest was reported by the authors.