Kyungmi Lee·In Sik Kim· Seok Woo Lee

Abstract The effects of genetic and environmental factors on Pinus koraiensis growth were studied based on a 35 year-old progeny trial composed of open-pollinated off-spring of twenty-one plus trees.Height,DBH and volume of the offspring was analyzed using restricted maximum likelihood/best linear unbiased prediction in mixed model analysis.Significant site and family effects on the three traits were observed.The distinct growth of offspring by site with disparate climates corroborated the importance of planting species in suitable conditions.Growth differences by family was significant,emphasizing the importance of identifying families with either superior or inferior performance.The parental ranking was assigned in the sites,inferring the breeding value of each plus tree.The estimates of individual heritabilityof height,DBH and volume growth were 0.169—0.645,0.108—0.331and0.129—0.343respectively,with higherofthe height than DBH on each site.Coefficient of variance of genetic effect was higher with DBH in some cases,indicating the scope for selection is larger for this trait despite the lower heritability compared to height.For the variation between families in terms of the performance stability across sites,consideration of the genotype by environment interaction is required in selecting materials to be used in reforestation with Korean pine.A few families with either superior or inferior performance retained their parental ranking for at least a decade.Other families with increased growth on a particular site were identified,indicating their high breeding value and low stability.Differences in the genetic performance of the families by site requires delineation of the breeding region of the species.

Keywords Pinus koraiensis·Genetic parameters ·Genotype by environment interaction·Tree improvement ·Progeny trial

Introduction

Progeny testing is a principal step to determine the breeding value (BV) of selected individual trees in tree improvement programs (White 1987).Parental ranking based on the progeny performance is critical in genetic progress,particularly in forest tree improvement in which traits of interest have low variability (White et al.1988;Vidal et al.2017).Precise and accurate prediction of breeding values and quantitative genetic parameters,including heritability,is a requisite for effective tree improvement (White 1987;De Lima et al.2019).

Pinus koraiensisSiebold &Zucc.(Korean pine) is a native conifer mainly distributed in northeastern Asia from the Korean peninsula to China and the Russian Far East,inhabiting northern mountainous areas in Korea (Xu and Yan 2001;Chun et al.2015;Belyanin and Belyanina 2019).The species is important economically with its high-quality timber,edible seeds and medicinal properties (Kang and Lindgren 1998;Li and Löfgen 2000;Xu and Yan 2001;Liang et al.2018;Li et al.2020).Although the natural range of Korean pine is limited in South Korea,the area planted with the species from 1953 to 1990 was the largest compared to other native species (Kim and Zsuffa 1994).As a consequence,plantations of the species are currently the third largest in the country (Chun et al.2015).In Korea,wood produced from Korean pine is mainly used in the lumber industry with increased consumption over the last five years unlike most other species,although plantation areas have decreased (Han et al.2018;Korea Forest Service 2019a,2019b).In addition,the edible seed of the species is a major forest product in northeast Asia,with maximum production over 9,600 tons per year over the last five years in Korea(Korea Forest Service 2019c).

A tree improvement program for Korean pine was initiated in 1959 to improve its genetic quality and economic timber value (Kim et al.2019).To carry out genetic testing of selected plus trees,progeny trials with half-sib and fullsib offspring were established.There has been 244 openpollinated families included in ten open-pollinated progeny trials with two to three replication sites for each since 1975.In addition,there are two progeny trials consisting of fullsib progenies from forty-six combinations from the artificial cross.Each of twelve trials there are important and valuable for Korean pine tree improvement with their distinct constituent families or combinations.A second generation seed orchard with superior full-sib progenies has been established.An improvement program of the species is also ongoing in northeastern China,where the species is naturally distributed according to its economic value (Liang et al.2019).However,research on the genetic variation of Korean pine is scarce in China as well as in northeastern Asia in spite of a long history of breeding of the species (Kaviriri et al.2020).

To thoroughly understand the genetic parameters from progeny trials,restricted maximum likelihood (REML) and best linear unbiased prediction (BLUP) have been prevalent in mixed model analysis in plant and animal breeding(Williams et al.2002) and have the advantage of increasing the precision of estimating genetic parameters compared to ANOVA when dealing with unbalanced data in random locations (Hu 2015).

The stability of the performance of the plus clones or families by environment is important information to support the decision of deploying advanced breeding materials in the improvement program (Li et al.2017).Although genotype and environment interaction (GEI) is a major research topic in tree improvement,the stability of the performance of selected families of the Korean pine across various environments has not been analyzed.

This study aimed at:(1) investigating growth characteristics ofP.koraiensisin Korea;(2) estimating genetic parameters by dissection of the phenotypic variance;(3)evaluating the relative performance of the plus trees with the corresponding genetic and environmental factors by the REML/BLUP approach;and,(4) analyzing the stability of the performance of the families over the environments based on a 35 year-old Korean pine progeny trial.

Materials and methods

Half-sib progenies of twenty-one plus trees with seedlings produced from natural stands as reference materials were planted in 1988.The names of the plus trees included are:gg45,gg46,gg51,gg56,gg58,gg61,gg62,gg63,kw2,kw8,kw11,kw16,kw17,kw21,kw24,kw27,kw29,kw30,kw82,kw83,kw84.The experimental design was tentree family row plots 1.8 m × 1.8 m in a four randomizedcomplete block with three replications,located in Chuncheon (CC),Chungju (CJ) and Gunpo (GP) (Fig.1).Distance between individuals at 35 years is generally more than 3.6 m due to natural death or thinning.Average climatic conditions of the sites are shown in Table 1.TheCCsite is located in the mid-region of the Korean peninsula on mountainous terrain with both montane and continental climates (Korea Meteorological Administration 2020).TheCJsite is a general temperate region with a continental climate.TheGPsite is classified as having a continental climate.

Table 1 Site description of Pinus koraiensis progeny trials established in 1988

Fig.1 a Natural distribution of Pinus koraiensis and b locations of P.koraiensis progeny trials started in 1988

Height (H) and diameter at breast height (DBH) of 870 trees,consisting of 831 progenies and 39 reference individuals,were measured and volume (Vol) calculated based on the measurements (Lee et al.2017):

Statistical analysis

Growth comparisons by site,family and between progeny and reference materials were assessed based on parametric estimations.The variance components of family and block on each site were estimated using the mixed model to detect growth characteristics of the progenies (Eq.2).

whereYis the vector of observation,bthe vectors of fixed effects of block,fandfbare those of random effects of family,family-block andethe vector of residual terms.

Genetic gain (ΔG) by 10% selection for each trait was estimated with aselection intensity (i) of 1.76 and mean valueof observation

BV of each plus tree was estimated as twice its general combining ability (GCA) by BLUP.Spearman’s rank correlation analysis between the ranks of BV in pairwise sites was performed for each trait to examine the existence of relatedness between the rankings of performance of families on different sites.

Coefficient of variance of genetic,environment and phenotypic variances (CVG,CVEandCVPrespectively),the dimensionless criterion for comparing genetic variabilities(Houle 1992),were provided (Eq.8 − 10):

In addition to identifying genetic and environmental effects,the genotype by environment interaction (GEI)examines family performance across the environments(Yan et al.2007;Bose et al.2014).The stability of the performance of the families over sites was evaluated by the additive main effects and multiplicative interaction model(AMMI).The model is utilized in breeding studies of many species as a constructive methodology in detecting GEI even if it was proposed for fixed effects (Bocianowki et al.2019).It uses both analysis of variance (ANOVA)and principal component analysis (PCA) to identify GEI.Mean value of each trait by family is used as the dependent variable in the model.

All analyzes were performed using R program (R Development Core Team 2019).R packagelme4andagricolaewere used in the mixed model and for the AMMI analyses,respectively (De Mendiburu 2009;Bates et al.2014).

Results

Growth characteristics by site

Site effects on growth traits were significant in the test model (P<0.001).Mean height and DBH at age 35 were 12.0 m and 21.0 cm,respectively (Table 2).Height of the offspring was the lowest on theGPsite and DBH the lowest on theCJsite.Volume growth on the three sites was significantly different,showing the best growth on theCCsite.The stand on theCJsite showed large height growth but slow diameter growth.

Growth of offspring of the plus trees was compared to the reference originated from natural stands to identify growth increment by the plus tree selection (Fig.2;Table S1).Height differences between the groups wasinsignificant.However,DBH and volume of the offspring were higher than those from natural stands,except for DBH on theCJsite.Significant differences in DBH and volume between the groups were observed on theCCsite where the best growth appeared.

Fig.2 Mean height (H),DBH and volume (Vol) of half-sib progenies of plus trees (HS),reference materials from natural stands (REF) by 35 year -old Pinus koraiensis on each site(CC,CJ,GP).Whole on the x-axis means the three sites combined (*:P <0.05;**:P <0.01;ns:non-significant)

Table 2 Mean values of height,DBH and volume by site in a 35 year-old progeny trial of Pinus koraiensis

Estimation of genetic variation,heritability and genetic gain

Growth differences in the traits by family were significant except for DBH andvolume ontheCJsite.Among the sites,of volume growthwas the highest intheCCsite(0.343),as inthe previous assessment atage23 (Han et al.2007) when mean values of height,DBH and volume were 9.0m,14.7 cmand0.09m3(Table3).

Table 3 Estimation of genetic parameters in a 35 yearold progeny trial of Pinus koraiensis by site

Breeding value estimates by family

Means and breeding values for each trait were compared to check the relationships between the estimates (Fig.3).Pearson correlation coefficients between means and BVsfor each trait varied from 0.94—0.99.The least means to have positive BVs were on a similar level or slightly lower compared to the mean estimates for each site.Although their correlation was high,it was possible to find the reverse in terms of mean versus BV.For example,when comparing the volume ofgg63 andkw24 on theCCsite,the latter had a higher mean value but lower BV.

Fig.3 Comparison of mean values and breeding values (BVs) of height (H),DBH and volume (Vol) in a 35 year-old progeny trial of Pinus koraiensis

Among the twenty-one plus trees,gg61 andkw21 showed higher BVs of volume growth across the sites(Fig.4;Table S2).Thegg62 andgg63 families had poor volume growth across the sites.Thegg51 plus tree was highly ranked on theCCsite but estimates on the other sites decreased.Thekw21 family,which showed large volume growth in a previous analysis a decade ago,still appeared high in the ranking of the families.Breeding values of DBH and therefore the volume growth ofgg62 andgg63 were low,consistent with the previous study.The significant correlation was identified in the ranks of BVs between theCJandGPsites for DBH and volume by Spearman’s rank correlation test.In the remaining pairs analyzed,the rank correlation was insignificant in showing changes in the ranks of BVs of the traits by GEI.

Fig.4 Changes in breeding values (BVs) of height (H),DBH and volume (Vol) of the plus trees on the three sites (CC,CJ and GP) in a 35 year-old Pinus koraiensis progeny trial

Growth characteristics of each family by site

The stability of the performance of each family over the sites was assessed by the AMMI stability value (rASV)and visualized by AMMI biplot analysis (Fig.5;Table S2).According to the stability analyses to evaluate GEI,the first principal component accounted for 62.9%,71.5%and 85.5% of the total interaction between genotype and environment in height,DBH and volume,respectively.Different patterns of performance were detected by family in the relationship between BV and stability.Among the families with high BVs,kw21 andgg61 showed stable growth of overall traits on the sites,promoting their common effectiveness in the advanced generation.Thekw11 andkw24 were ranked as the most stable families in terms of volume growth,although their height growth was unstable by site.The performance of the offspring ofgg46,for which growth was not superior,was stable in the traits.Thegg51 andgg58 were the superior families on theCCsites with the largest volume growth,in accordance with DBH.However,both were low ranking on theGPsite,showing relative susceptibility to environmental effects.The fluctuation of the performance of these two families by site was conf irmed in the rASV and the biplot.

Fig.5 Stability analysis by AMMI biplot between height (H),DBH,volume (Vol) and principal component 1 (PC1) of 21 half-sib families over the three sites in a 35 year-old Pinus koraiensis progeny trial(numbers in parenthesis is the percentage accounting for Genotypeby-Environmental Interaction)

Discussion

Growth characteristics by site

Growth differences by site was significant (Table 2).Traits on theCCsite were 12.3 m for height and 26.0 cm for DBH,showing the best growth among the sites.The best growth in volume on theCCsite was in agreement with the results from the same stands at age 23 (Han et al.2007).Since theCCsite is in mountainous areas in northeastern South Korea with cool-summer climates (Kong et al.2016) (Fig.1;Table 1),the better volume growth is due to the favorable environmental conditions (Wang et al.2013,2019).

The difference inP.koraiensisgrowth by site in South Korea under different climate conditions raises the consideration of the delineation of the breeding region of the species.Breeding materials used in this study were selected fromKangwonandGyeonggiprovinces,higher latitudes than theCJsite (Fig.1).The best volume growth on theCCsite ref lects the better performance of the local population(Gray et al.2016).

In the context of the relationship between growth and climate,it was regarded worthwhile to consider this over the range of its biogeographical distribution.Comparing the growth of selected breeding materials at age 30 in Korean pine with several studies in northeast China (Liang et al.2018;Wang et al.2018;Kaviriri et al.2020),growth in different locations was unexpectedly comparable (Table 4).In particular,growth characteristics in Linjang,where on the east side is the Baekdusan Mountain,was surprisingly different,showing better height growth with lower diameters than those of Korean pine in South Korea.Without considering environmental effects on tree growth,the difference is contrary to the general understanding of the relationship between diameter growth and planting spacing.Wider spacing promotes diameter growth after a particular period (Hébert et al.2016).Considering the natural distribution ofP.koraiensis(Belyanin and Belyanina 2019),the Korean peninsula is its southern marginal habitat (Fig.1).The enhanced growth of the peripheral population in South Korea might be affected by elevated temperatures,especially in theCCsite where growth exceeded the other sites (Way and Oren 2010).

A precise explanation of the relationship between growth and environment,including climate,is limited in this study.Nevertheless,growth differences by site reveal the importance of planting the species on suitable sites.It is emphasized by the susceptibility to environmental change and reduction of the natural range predicted for this species(Han and Park 1988;Ahn et al.2015).Difference in volume growth by plus tree selection was significant on theCCsite where environmental conditions are considered to be more favorable.

Estimation of genetic variation,heritability and genetic gain

Significant differences in growth traits among families of the species has been reported in previous studies (Liang et al.2018;Wang et al.2018;Kaviriri et al.2020).Whileand DBH ranged from 0.169—0.645,0.108—0.331 by site,the estimates of individual heritability in Linjiang were 0.21 for height and 0.17 for DBH,respectively(Table 4).Theof the traits 0.384—0.678 and 0.288—0.502 in this study were lower than previous studies.

Table 4 Description of the materials and results from the genetic studies

Heritability of all traits were lowest on theCJsite which was considered to have unfavorable conditions for Korean pine.Decrease of the estimate in the site exhibits larger environmental effects hindering the expression of genetic factors.Lower heritability was often reported in field trials than in greenhouse experiments by increased environmental variance and decreased additive genetic variance (Conner et al.2003).

Breeding value estimates by family

In this study,the breeding value of each mother tree was estimated based on best linear unbiased prediction (BLUP) by restricted maximum likelihood (REML).Rankings of BVs in growth traits by family were compared across sites (Fig.4;Table S2).Changes in the BVs ranking between sites calls for the delineation of the breeding region ofP.koraiensisin South Korea.Superior families on the high ranks in each site are recommended for utilization as planting materials for the purpose of the timber volume enhancement.

The results are comparable with the study of the same progenies at age 23 using an ANOVA.The offspring ofgg51,kw16 andkw21 which had superior volume growth in the previous study,still appeared in the relatively high ranks of the 21 families (Fig.4).Low BVs of volume growth ofgg62 andgg63 were also consistent with the results of the previous study.As the duration of a progeny trial is related to the cost of the program,the age-age correlation is one of the major factors affecting the economic benefits of tree improvement programs (Rweyongeza 2016).Both the superior and inferior families in this trial maintained stable ranks at least for a decade.It may be suggested with some caution that the selection at age 23 would correspond with the later selection derived from the analysis more than a decade after,at least for those families.However integrated analysis with periodical growth data of the families is needed to estimate the precise age-age correlation.The progeny trials with the different families ofP.koraiensismust also be investigated to verify and define the optimum age for effective selection for the improvement of the species.In terms of the integrated analyses,future studies on the different sets of the progeny trial are required to test the 244 plus trees of Korean pine using the growth data acquired from the periodical survey covering the age classes.

Growth characteristics by family across the sites

The stability of the families examining the existence of GEI was analyzed by the AMMI model (Fig.5).The rASV provides additional information to assist with the selection of breeding and planting materials under the selection strategy with the separated breeding region.Among the superior families in each site,kw21 had stable volume growth,whereasgg62 andgg63 families had inferior growth with low BVs on all the three sites.The former is evaluated as highly appropriate as the resources among the trees with various combined patterns in terms of performance and stability.At the same time,trees ofgg51 andgg58 showed special adaptability on theCCsite with the lowest rASV.Under the delineation of the breeding region suggested above,it is possible to maximize improvement by planting site-specific superior families within the region.

Conclusions

As one of the decisive procedures in the tree improvement program of the Korean pine in South Korea,the growth of a 35 year-old progeny trial of twenty-one half-sib families was explored.Our study indicates the existence of significant genetic and environmental effects on the growth which is informative in plantation establishment and for advancing the breeding generation.The genotype by environmental interaction examined in this study suggests consideration of the breeding region for the species to make growth enhancement feasible in the improvement program.The reforested sites of Korean pine had either succeeded or failed,depending on their environment even within a province in South Korea.

The augmented genetic gain is achievable using the genetically improved materials derived from the plus trees with superior growth and stability.To promote the successful tree improvement program,it is necessary to understand the underlying genetic effects on the growth of the species.Future research would appraise the influence of genetic factors on growth with the expanded progeny trial data collated from more families in the time scale.

With regards to the environmental effect,several incidental results from the previous studies were referred to for growth comparisons of the Korean pine covering its natural range.Future studies with specific experimental designs are necessary to clarify suitable environmental conditions for optimum growth of Korean pine.Attention to the speciessite relationship is fundamental preceding plantation establishment for timber production or conservation.