Qinhui Zhang·Shihe Yu·Xiaona Pei ·Qianchun Wang·Aijun Lu·Ying Cao ·Mulualem Tigabu·Jian Feng·Xiyang Zhao

Abstract In order to quantify within- and between-population variation in seed and seedling traits of Juglans mandshurica and reveal the relationship among genetic and environmental variations and select elite families, samples of 50 J.mandshurica families from five natural populations in Liaoning Province, northeast China, were taken to measure seed and seedling traits.The results show that all seed traits varied significantly among families within the population,but only kernel weight and kernel rate showed significant variations among populations.Average values of single seed weight, length, width, lateral diameter, and average size,and kernel weight and rate were 10.1 g, 43.0 mm, 29.2 mm,28.1 mm, 33.4 mm, and 2.2 g and 22.5%, respectively.Significant variations were observed in seedling height and root collar diameter among families and interaction between families and blocks, but the block effects on height and root collar diameter were insignificant.Average values of height and root collar diameter were 94.0 cm and 8.7 mm, respectively.Family heritability of traits ranged from 0.6 gm (kernel weight) to 0.9 mm (seedling height).Correlation analysis showed a strong relationship among seed traits but a weak correlation between seed and seedling traits.Cluster analysis grouped the five natural populations of J.mandshurica into three significant clusters with different characteristics.The general combining ability analysis showed that most traits in one family (WD11) were higher, suggesting parental traits were excellent for selecting elite parent clones.Using the comprehensive evaluation method, five families with better seed traits and five families with better seedling traits were selected as elite materials with 10% selection rate.The genetic gains of these elite materials for seed weight, kernel weight, average seed size, kernel rate, seedling height, and root collar diameter were 13.1%, 10.3%, 4.1%, 2.4%, 29.7%,and 21.1%, respectively.

Keywords General combining ability·Heritability ·Comprehensive evaluation·Genetic gain

Introduction

Juglans mandshuricaMaxim.is a deciduous species of the Juglandaceae family and one of the most important hardwood species in northeast China.It is widely distributed in the Changbai Mountains, and the Greater and Lesser Khingan regions in northeast China (Yuan et al.2013).The wood ofJ.mandshuricahas high commercial value because ofits aesthetic appearance, straight texture, high density, mechanical strength, rich elasticity, and corrosion resistance, and is widely used in the production of high-quality materials such as high-grade furniture, sports equipment, and musical instruments (Zhang et al.2017; Zhu et al.2018).The fruit has high nutritional value and rich in oils, proteins, mineral elements and vitamins.The kernels can be eaten directly or as a specialty food additive and have significant economic value (Yu 2012).The shape of the kernel is unique and so it also has good economic benefits in dried fruit and antique walnut (Song et al.2017).J.mandshuricaalso has high medicinal values due to the presence of juglone, an allelopathic compound, in bark, leaves, and the fruit pericarp,is used to treat cancer, reduce blood lipids, acts as an antioxidation, and used in anti-inflammation treatments (Aithal et al.2009; Silva-Belmares et al.2014; Vardhini 2014; Wu and Sun 2019).In recent years, the demand forJ.mandshuricaproducts has been increasing in China but natural sources of the species are scarce due to anthropogenic disturbances.To protect the remaining germplasm resources and develop improved varieties for production, studies on variation in seed and seedling traits in natural populations are necessary.

Phenotypic traits are the embodiment of genetic and environmental diversity as well as interactions between them (Zhao and Si 2016).A thorough evaluation of phenotypic traits has significant importance for preserving and developing eliteJ.mandshuricagermplasm resources.Phenotypic traits have previously been studied using select populations and a limited number of families (Chu et al.2010;Zhang et al.2011, 2017), but differences persist in character performances in different regions due to differences in environments.Therefore, seed and seedling traits were characterized for five natural populations and 50J.mandshuricafamilies from five regions of Liaoning Province.The objectives were to: (1) quantify variations within- and between-populations in seed and seedling traits; (2) analyze interactions between heredity and environment; (3) explore the relationship between seed and seedling traits; and, (4)select elite families according to these traits.This study will provide information for the selection of elite cultivars and the protection of the genotypes in natural forests ofJ.mandshurica.

Materials and methods

Plant materials

The seed samples ofindividual trees (n=50) were collected in 2013 from five populations in Liaoning Province (Tables 1 and 2).Ten trees were collected for each population and ten seeds/tree collected for characterizing phenotypic traits.All sample trees were healthy, mature indigenous specimens.The following year, seeds were planted in a randomized block design with five blocks of six trees.Different families were randomly planted in row plots at 3.0 m × 4.0 m.

Table 1 The main geographical and climatic conditions of the sampled populations in J.mandshurica

Table 2 Names of different families

Table 3 ANOVA analysis of different seed traits of J.mandshurica

Table 4 ANOVA analysis of different seedling traits of J.mandshurica

Seed trait measurements

These traits were measured in November 2013.Seed weight(SWt,W St), kernel weight (KW,W k), seed length (SL,L s), seed width (SW,W s), lateral diameter (LD,D l), average seed size(ASS,S as), and kernel rate (KR,R k) were determined.Seed weight and kernel weight were measured using an electronic balance.L s,W s, andD lwere measured using a vernier caliper.The average seed size (ASS,S as) was calculated as:

was calculated as

Seedling trait measurements

Two-year-old seedlings were measured in October 2015.Seedling height (SH,H s) and root collar diameter (RCD,D rc) were determined for each seedling.H swas measured using a sliding ruler, whileD rcwere measured with a Vernier caliper.

Statistical analysis

All data were analyzed using data processing system (DPS)version 18.10 and SPSS Statistical Package (SPSS version 26.0, IBM Corp., Armonk, NY, USA).The following linear model was used to analyze seed traits (Ji et al.2013):

whereYijkis the observed value of seedkin familyjgrowing in populationi;μis the overall mean;τiis the fixed effect of populationi,δj(i)is the random effect of familyjwithin populationi, andεk(ij)is the random error.

The following linear model was used for joint analysis of families and blocks of seedling traits (Pan et al.2018):

whereXijklis the observed value of an individual treelin familyjwithin populationigrowing in blockk,μis the family mean,Bkis the fixed effect of blockk,Piis the fixed effect of populationi,Fj(i)is the random effect of familyjwithin populationi,BPikis the fixed interactive effect of blockkand populationi,BFj(i)kis the random interactive effect of blockkand familyjwithin populationi, andeijklis the random error.

The phenotypic differentiation coefficient (V ST) was calculated using the following formula (Wei et al.2020):

wheresis the number of populations,tis the number of plants in populations,is the variance component among populations, andis the variance component within populations.

The coefficient of variation (CV) was calculated using the following formula (Munilla and Guitián 2014):

Broad-sense heritability (h 2) among families for each trait was calculated according to Xu ( 2006) as:

whereFis the F-test value in the ANOVA.

The phenotypic correlation between traitsxandywas calculated according to Han et al.( 2017):

whereCovP(x,y) is the phenotypic covariance between traitsxandy,andare the phenotypic variance component for traitxandy,respectively.

General combining ability (GCA) of each trait among different families was calculated according to Wang et al.( 2016):

wheregis the parent’s general combining ability,xis the mean value of the offspring of a specific parental combination in a specific trait, andμis the total mean value of all combinations of this trait.

Comprehensive evaluations of different families were calculated as described by Zhao et al.( 2016):

wherea i=X ij/X jmax; theQivalue is the comprehensive evaluation value of each family;X ijis the mean value of one trait.X jmaxis the maximum value of that trait;nis the number of traits;

Genetic gain was estimated using the formula of Silva et al.( 2008):

where ΔGis the genetic gains,h2andare the heritability and mean value of a given trait, respectively.ΔSis the selection differential.

Results

Variations in seed and seedling traits

All seed traits varied significantly among families within the populations but only kernel weight and rate showed significant variations among populations (Table 3).Similarly,seedling height and root collar diameter showed significant variation among populations, families, interaction between block and population and interaction between block and families within populations (Table 4), but the block effects for these seedling traits were insignificant.Descriptive statistics for seed and seedling traits are presented in Table 4.The mean seed weight and kernel weight of all families were 10.1 g and 2.2 g, respectively, and the maximum values were 1.8- and 1.7-fold the minimum values, respectively.The mean seed length, seed width, lateral diameter, and average seed size were 43.0 cm, 29.2 cm, 28.1 cm, and 33.4 cm,respectively, and the maximum values were 1.5-, 1.5-, 1.4-,and 1.3-fold the minimum values, respectively (Table 4).The mean kernel rate was 22.5%, and the maximum value was 1.8- fold the minimum value.The mean seedling height and root collar diameter were 0.9 m and 8.7 mm, respectively, and the maximum values were 2.7- and 2.1-fold the minimum values, respectively.TheCVs of all traits ranged from 6.4% (average seed size) to 34.2% (root collar diameter), among which theCVs of seed length, seed width,lateral diameter, and average seed size were lower and did not surpass 10%, but theCVs of seedling height and root collar diameter were higher than 30% (Fig.1).The broad sense heritability,h 2, for kernel weight (0.56) and kernel rate(0.69) was moderate, and for other traits was high, ranging from 0.87 to 0.95 (Table 5).The mean values of each trait for different populations are shown in Fig.2.The DBG population showed the largest seed weight (10.7 g ± 0.4 g), seed width (29.8 mm ± 0.4 mm), seedling height (1.1 m ± 0.2 m),and root collar diameter (10.2 mm ± 2.0 mm).The DGJ population showed the largest kernel weight(2.37 g ± 0.15 g), seed length (44.50 mm ± 2.98 mm),lateral diameter (28.66 mm ± 1.76 mm), and average seed size (34.22 mm ± 1.83 mm).The WD population had the highest kernel rate (23.32% ± 3.15%).The KD population showed the lowest seed weight(9.5 g ± 0.7 g), kernel weight (2.1 g ± 0.1 g), seed length(41.39 mm ± 3.24 mm), seed width (28.43 mm ± 0.98 mm),lateral diameter (27.73 mm ± 0.60 mm), average seed size (32.52 mm ± 1.62 mm), and seedling height(0.74 m ± 0.14 m).The DBG population showed the lowest lateral diameter (27.73 mm ± 1.03 mm), and kernel rate(20.5% ± 1.6%), while the JC population had the lowest root collar diameter (8.10 mm ± 0.82 mm).

Fig.1 Coefficients of variation(CV, %) of seed and seedling traits among populations.For traits, SWt, KW, SL, SW, LD,ASS, KR, SH and RCD stand for seed weight, kernel weight,seed length, seed width, lateral seed diameter, average seed size, kernel rate, seedling height, and root collar diameter,respectively

Fig.2 Mean values of different traits among different populations.Bars followed by the same letter are not significantly different.For traits, SWt, KW, SL, SW, LD, ASS, KR, SH and RCD stand for seed weight, kernel weight, seed length, seed width, lateral seed diameter,average seed size, kernel rate, seedling height, and root collar diameter, respectively

Table 5 Descriptive statistics for seed and seedling traits for J.mandshurica families

Phenotypic differentiation among natural populations

The proportion of different variance components to the total variation was analyzed, and phenotypic differentiation coefficients calculated.The variance component among populations accounted for 32.53% of the total variation, the variance component within populations accounted for 33.98%of the total variation, and the random error accounted for 22.62% (Table 6).The variance component for seed length and average seed size among populations was greater than 40% while that of lateral diameter was lower than 20%.The variance component for seed width and lateral diameter within populations was higher than 40%, whereas it was lower than 20% for kernel weight and kernel rate.The phenotypic differentiation coefficient ranged from 21.71%(lateral diameter) to 68.75% (kernel weight), and the mean phenotypic differentiation coefficient of all traits was 50.31%among populations and 49.69% within populations (Fig.3).

Fig.3 The differentiation coefficient of morphological traits among/within populations in J.mandshurica.For traits, SWt, KW, SL, SW,LD, ASS, and KR stand for seed weight, kernel weight, seed length,seed width, lateral seed diameter, average seed size, and kernel rate,respectively

Table 6 Variance components for seed traits among/within populations in J.mandshurica

Correlation analysis

The correlation coefficients between different traits are shown in Fig.4.Seed weight had a significantly positive correlation with kernel weight, seed length, seed width, lateral diameter, and average seed size.Kernel weight was significantly positively correlated with seed length, seed width,average seed size, and kernel rate.Average seed size was significantly positively correlated with seed length, seed width and lateral diameter.The correlation between the seed width and lateral diameter was significant and positive, but seed width was significantly negatively correlated with kernel rate(0.437).There was a weak correlation between average seed size and kernel rate (0.336).Significant positive correlation was also observed between seedling height and root collar diameter (0.784).However, there was no significant correlation between seed and seedling straits.

Fig.4 Correlation coefficients among different traits in J.mandshurica families.*correlation is significant at 5% level;**correlation is significant at 1% level; Numbers are the correlation coefficients among different traits

The correlation coefficients between seed traits and geoclimatic factors are shown in Fig.5.Kernel weight was significantly negatively correlated with elevation (- 0.889);annual rainfall had a significantly negative correlation with seed length (- 0.906) and average seed size (- 0.910) while there was no significant correlation between other seed traits and ecological factors.

Fig.5 Correlations between seed traits and geo-climatic factors in J.mandshurica families.LOG: Longitude; LAT:Latitude; ELE: Elevation; ATE:Annual temperature; ARA:Annual rainfall; FFD: Frost-free days.*correlation is significant at 5% level; Numbers are the correlation coefficients between seed traits and geo-climatic factors

Cluster analysis of different populations

The Between-groups Linkage of Euclidean distance was used to cluster all the traits of five natural populations ofJ.mandshurica(Fig.6).According to the cluster analysis, the natural population ofJ.mandshuricacould be divided into three major clusters when the genetic distance was 7.The first cluster included the population DBG, chracterized by families with large seed weight, seed width, seedling height,and root collar diameter.The second cluster included the population DGJ and this population was characterized by families with large kernel weight, seed length, lateral diameter, average seed size, and kernel rate.The third cluster included JC, WD, and KD populations with characteristics oflow seed weight, seed length, seed width, average seed size, seedling height, and root collar diameter.

Fig.6 Cluster analysis of the seed and seedling traits for J.mandshurica in different natural populations JC (Jianchang), WD (Wendao),KD (Kuandian), DGJ (Dagujia) and DBG (Dabiangou)

General combining ability

The general combining ability (GCA) values of different traits among families are shown in Table S1.The GCAs for seed weight ranged from - 2.333 (WD10) to 4.007 (WD11),for kernel weight, from - 0.457 (WD7) to 0.774 (WD11),for seed length, from - 8.078 (KD16) to 8.491 (DGJ4), for seed width, from - 6.104 (WD10) to 5.315 (WD11), for lateral diameter, from - 3.383 (WD10) to 5.360 (WD11), for average seed size from - 3.407 (WD16) to 4.639 (WD11),for kernel rate from - 5.223 (DBG10) to 8.877 (WD10),for seedling height, from - 0.373 (WD8) to 0.585 (KD10),and for root collar diameter, from - 2.505 (WD7) to 4.432(DBG5).

Selection of elite family

Based on the results of correlation analysis, seed weight,kernel weight, average seed size and kernel rate were selected as the comprehensive evaluation indices of seed traits.The comprehensive evaluation value,Qi,of seed and seedling traits are shown in Table S2.Family WD11 had the largestQivalue (1.918), followed by families DGJ8(1.841), DGJ4 (1.824), WD9 (1.795), and JC17 (1.790),while the family WD7 had the lowestQivalue (1.633).Comprehensive evaluation for seedling traits showed that family DBG5 had the highestQivalue (1.391), followed by families KD10 (1.370), DBG8 (1.311), DBG7 (1.312) and DBG1 (1.297); the family KD7 had the lowestQivalue(0.949).

Table 7 Genetic gains for seed weight (SWt), kernel weight (KW),average seed size (ASS), kernel rate (KR), seedling height (SH) and root collar diameter (RCD)

Genetic gains

Based on theQivalues in Table S1 and a selection rate of 10%,families WD11, DGJ8, DGJ4, WD9, and JC11 were selected as elite families for seed traits, and the genetic gains in seed weight, kernel weight, average seed size and kernel rate of these elite families were 12.82%, 10.25%, 4.08%, and 2.42%,respectively (Table 7).Families DBG5, KD10, DBG8, DBG7 and DBG1 were selected as elite families according to seedling traits, and the genetic gains in seedling height and root collar diameter of these elite families were 29.72% and 21.05%,respectively (Table 7).

Discussion

Genetic variation is the driving force of evolution, enabling organisms to overcome environmental challenges (Cardoso et al.2015).Thus, assessment of genetic variation within and among populations is an essential process for the effective conservation of forest genetic resources (Papi et al.2012).In natural environments, different populations often produce significant variations due to the interaction of genetic and environmental diversity (Ming and Gu 2006).In this study,there were significant differences in all seed and seedling traits within-population, while there existed significant differences in kernel weight and kernel rate between populations.The heritability of the seed traits, such asW st,L s, andW swere as high as 0.85, 0.87 and 0.87, indicating that the variation of these traits can be stably inherited, which indicates that detecting genetic gain by observing phenotypic variations makes sense.In addition, the variance component of seed traits among and within populations were 32.53%and 33.98%, respectively.Furthermore, the mean phenotypical differentiation coefficients were 50.31% and 49.69%,respectively.This indicates that the genetic differentiation among populations was similar to within populations, and the primary phenotypic sources variation ofJ.mandshuricain natural populations included both among and within populations.The phenotypic coefficients observed in our study were lower than those observed inAmygdalus mira(Koehne) Yu et Lu (Wei et al.2020),Carya dabieshanensisM.C Liu & Z.J.Li (Zhang et al.2020),Liquidambar formosanaHance (He et al.2018), andSophora japonicaL.(Sun et al.2011), but higher than that of some coniferous species (Loha et al.2009; Li et al.2013; Chen et al.2015;Deng et al.2017).This discrepancy reflects the complexity of genotypes by environment interaction and the universality of phenotypic variation, resulting from a different environmental selection (Arag?o et al.2015).The mean phenotypic differentiation coefficients of kernel weight and kernel rate were more than 60%, which indicates that the phenotypic variation of these two traits mainly came from among populations (Zhang et al.2008).The diversity variation among populations reflects the population’s adaptation in different environments; larger phenotypic differentiation coefficient means wider adaptation range.

In this study, family variation accounted for more than 30% of the total phenotypic variation for seed traits(Table 5), suggesting the presence of adequate genetic variability in the present material, as observed in other studies(Zhang et al.2014a).Generally, variations in seed size are the results of environmental influences during seed development combined with genetic variability.This study also confirms that seed size was more under genetic influence than environmental influence.This is further confirmed from lower CVs for seed traits than for seedling traits, which were relatively higher compared with studies of Chen et al.( 2015)and Zhang et al.( 2014b).This in turn indicates that seed traits are more stable than seedling traits that have higherCVs, although a high degree of variability in seedling traits has relatively great selection potential.

Heritability reflects the relative role of genetics and environment in the expression of various traits, and is also useful for ranking the importance of each trait in crossbreeding programs (Jaisankar et al.2014).In this study, theh2for of seed weight, seed length, seed width, lateral diameter, average seed size, seedling height and root collar diameter were all higher than 0.8, suggesting the variation in these traits was greatly controlled by genetic factors, and it is more accurate to use these traits as selection indicators.The heritability values found in our study were higher than that observed inPinus wallichianaA.B.Jacks (Singh and Thapliyal 2012),indicating that the selection of elite families is feasible, and the selection intensity can be increased appropriately.Strong and intermediate genetic control is favorable for selection in breeding programs as it allows the use of small numbers of families to achieve high genetic gains (Munthali et al.2012),providing a reliable guarantee for the genetic improvement ofJ.mandshurica.As a whole, the pattern of genetic variations in seed and seedling traits observed in this study were consistent with previous studies made on several tropical tree species (Zhang et al.2014b, 2015, 2020; Chen et al.2015; Xu et al.2016; Wei et al.2020).

Strong positive correlations were found between seed weight, kernel weight, and other seed size traits, which was similar to the results by Li et al.( 2017).The strong linkage between seed traits could provide a basis for multiple-trait selection (Li et al.2018).Besides, this study found that a strong negative correlation between kernel rate and seed width and average seed size, which was different from other studies (Xiong et al.2017; Zhang et al.2014a; Wang et al.2012).This could be the reason for some seeds with higher weight and size but lower kernel content due to seed coat thickness.Seeds with increasing thickness during maturation are more influenced by environmental factors during late growth and are prone to measurement errors, which may have contributed to this negative correlation.There was also a significant positive correlation between seedling height and root collar diameter, which was similar to numerous other species (Wu et al.2018; Bai et al.2019).In this study,seed phenotypic traits had weak correlation with seedling traits.Similar results were found withCordia africanaLam.(Loha et al.2006).This suggests that any single trait could not be employed for selecting elite materials for tree improvement programs.

Phenotype differences are the result ofinteractions between genes and the environment, and environmental factors play crucial roles in shaping plant phenotypes (Nicotra et al.2010).The strong negative correlation between kernel weight and elevation in this study was similar to the findings of Diao et al.( 2014) forSapindus mukorossiGaertn.It may be due to decreases in atmospheric pressure and oxygen as elevation increases, affecting seed development ofJ.mandshurica(Hou et al.2017).Annual rainfall was strongly negatively correlated with seed length and average seed size, contrary to the study of plant communities by Wang ( 2015).Seed size was significantly negatively correlated with rainfall in deserts and grasslands, while size was significantly positively correlated with rainfall in forest species.This indicates that factors influencing seed size are different in different vegetation types, and the reproductive strategies of plants are also different.The rest of the environmental factors were not significantly correlated with seed traits, indicating that these had less influence on seed traits ofJ.mandshurica, which was also an external manifestation of environmental adaptation in various parts of the seed.In summary, the variation in selected seed traits in natural populations ofJ.mandshuricais correlated with environmental and geographic factors, reflecting a clear trend in geographic variation.This is further evidenced from the cluster analysis in which the natural populations ofJ.mandshuricawere divided into the major clusters.The first cluster included the DBG population of Liaoning province with usually the highest seed weights, seed widths, seedling heights and root collar diameters, which could be used to screen elite germplasm resources for breeding improvement ofJ.mandshurica.The second cluster contained population GDJ, with the highest seedling growth; growth in this population was rapid, which could provide materials for the study of fast-growing genotypes for timber production.

Using the general combining ability analysis, family WD11 has the highest value for most traits.The results showed that the additive effect of genes controlled the inheritance of growth traits in this material.General combining ability is an essential parameter in tree improvement programs that reflects the ability of parents to transmit valuable traits to their offspring, and is often used to screen for elite parents or cross combinations to provide theoretical guidance for seed orchard construction (Zhou et al.2004; Biabani et al.2012).However, GCA values for different traits varied widely among the families, so it was necessary to use a comprehensive evaluation to choose elite families.

In tree breeding, comprehensive evaluation is a good analytical method for selecting elite families or clones (Liang et al.2018).There are several methods of comprehensive evaluation, such as the optimal liner breeding value prediction method (White and Hodge 1988).Different methods have their appropriate territoriality, so it is necessary to choose an applicable method for data analysis (Chen et al.2004).In this study,Qivalues of differentJ.mandshuricafamilies were calculated for multiple-trait comprehensive evaluation, and elite families were selected based on the results of correlation analysis.When families were evaluated according to their seed traits, families WD11, DGJ18,DGJ4, WD9 and JC17 were considered elite (genetic gains in seed weight, kernel weight, average seed size and kernel rate were 12.82%, 10.25%, 4.08% and 2.42%, respectively).Similar results have been reported forCarya illinoensisK.Koch (Li et al.2011).The selected families have better seed characters and economic benefits and are suitable for seed production.When evaluated based on seedling traits, families DBG5, KD10, DBG8, DBG7, and DBG1 were considered elite (genetic gains in seedling height and root collar diameter were 29.72% and 21.25%), higher than the results by Li and Liu ( 2014).These elite families had good growth performances which could provide superior material for timber production.Breeding objectives determine breeding methods, andJ.mandshuricais an excellent species for both nut and timber production, and its premium materials should be selected from multiple perspectives.These materials have high potential for improvement, and their genetic quality for major economic traits and their flowering and fruiting selection should be studied in the next step to obtain elite materials with higher yield and quality.Studies have shown that the juvenile period ofJ.mandshuricawas representative and that a one-year-old seedling can predict the growth of s six-year-old seedling (Yuan 2013).Therefore, this study has practical implications for early selection using two-year-old seedlings ofJ.mandshurica.

Conclusion

Juglans mandshuricais a species with considerable economic value and has been extensively studied in medicine because ofits juglone component, while research on conventional breeding has not been studied as much.Besides,J mandshuricais a heterodichogamous plant, which has always been propagated by seedlings and has been wild or semi-wild for a long time, resulting in lower improved varieties and benefits.Coupled with over-logging in natural forests, natural resources ofJ.mandshuricaare near depletion.Therefore, within- and between-population variations in seed and seedling traits were analyzed in this study.The results confirmed the existence of good variation in all seed and seedling traits among families within a population.Several families with elite seed traits or seedling traits were selected, and will contribute to the genetic improvement ofJ.mandshuricaand the collection, preservation, evaluation,and utilization of germplasm resources.

Author contributionsQHZ and XYZ conceived and designed the research.JF conducted the experiments.SHY, QHW and AJL collected data, XNP and YC analyzed the data.QHZ wrote the manuscript.MT provided expert knowledge and revision of the manuscript.All authors read and approved the manuscript.