Xiaona Pei·Luping Jiang·

Ammar khalil Mohamed Ahmed4·Hongying Yu1,2·Rizheng Chong1,2·Xiangling You3·Xiyang Zhao3

Abstract Growth characteristics have complex inheritance patterns and genotype (G) by environment (E) interaction make predicting tree response to environmental changes difficult.In this study,the growth of seven poplar clones at three different sites was taken as the research focus,and heights and basal diameters were investigated in the second growing season.An ANOVA showed that all main effects,site,clone number and their interactions were highly significant in the overall F-tests.The coefficients of variation and repeatability of different traits ranged from 15.5 to 43.9%and from 0.549 to 0.912,respectively.AMMI (Additive Main Effects and Multiplicative Interaction) analysis results showed that genotype,environment and G × E interaction were significantly highly correlated.The stability analysis indicated that different clones showed different growth traits on different sites,which suggests that elite clones should be selected separately for different sites.Based on the growth traits,under a 10% selection rate,three clones were selected for different sites and the genetic gains of growth traits ranged from 4.7 to 11.2%.The three selected clones could be used to establish plantations in the future in different sites.

Keywords Poplar·Hybrid clone·Genotype·Environmental interaction·Stability·Repeatability

Introduction

Estimating growth and productivity of different sites is a central element of tree breeding and subsequent forest management (Berrill and O’Hara 2014).These estimates provide information for the potential production of biomass under different management prescriptions (Duan et al.2018).Productivity is dependent on site conditions such as the environment and soil (Duan et al.2018).Poplar (Populusspp.) plantations in favorable environments can produce large amounts of biomass in a relatively short time,maintaining their yield capacity over many years (Makeschin 1999;Hu et al.2019).However,productivity varies greatly in the first rotation on different sites (Facciotto et al.2005).This wide variability in productivity suggests that the importance of genetic characteristics of the planting material,such as site characteristics that affect growth under local climate and soils,has been underestimated (Bergante et al.2010).Therefore,

There are several methodologies to estimate and evaluate the genotype by environmental interaction,of which the most common are based on simple and multiple regressions.Despite widespread use,regression-based methods have limitations frequently reported in the literature (De Oliveira et al.2010).Crossa (1990) suggested that the application of multivariate methods can be useful to better exploit the information contained in the data.The AMMI (additive main effects and multiplicative interaction) model for estimating stability and adaptability based on the genotype,environment and their interactions has been more efficient than other stratification methods (De Oliveira et al.2010).In this study,the growth traits of seven poplar clones in three different sites was evaluated;heights and basal diameters were recorded,and several genetic and variation parameters analyzed.The objectives were to:(1) Evaluate the genotype by environmental interaction;(2) Evaluate growth performance,genetic variation and repeatability of the test clones on each site;(3) Evaluate and estimate the stability and adaptabilityof the tested clones among sites;and,(4) Suggest selection strategies for elite hybrid clones for each site.

Material and methods

Site descriptions and materials

Seven poplar clones (10,18,23,28,50,51 and 30) in this study were obtained by cross experiments (Table 1).The study was carried out on three sites in northeast China,the Fujin Forestry Center in Fujin,Heilongjiang Province(FJ);the Cuihai Forestry Center in Qiqihaer,Heilongjiang Province (CH);and the Lishu Forestry Center in Siping,Jilin Province (LS).The main climatic factors and soil characteristics of the sites are shown in Table 2.Experimental plantations were established in 2017 on each site using a randomized complete block design with five blocks,six trees per block at a 3.0 × 3.0 m spacing.

Table 1 Parent mating combination of hybrid clones on poplar

Table 2 Main climatic factors and soil characteristics of each site

Measurements of traits and statistical analyses

Stem height (HT) and basal diameter (D0) of all healthy trees were measured after leaf fall in 2018.Statistical analysis was performed using SPSS and DPS software.The significance of fixed effects was tested by analysis of variance (ANOVA)F-tests.The linear model used for joint analysis of the three sites (Dhillon et al.2013) was as follows:

where yijklis the performance of the individuallof familykgrowing in blockjof sitei,μis the overall mean,S iis the effect of sitei(i=1,…,3),B j(i)is the effect of blockjwithin sitei(j=1,…,5),C kis the effect of clonek(k=1,…,7),CS ikis the interactive effect of clonekand sitei,BC j(i)kis the interactive effect of clonekand blockj(within sitei),andε ijklis the random error.

Variations among clones were analyzed by ANOVA within sites according to Hansen and Roulund (1997) using the following equation:

wherey ijis the performance of an individual of cloneiwithin blockj,μis the overall mean,α iis the clone effect(i=1,…,7),β jis the block effect (i=1,…,5),αβ i(j)is the random effect of familyiby blockjinteraction,andε ijis the random error.

The phenotypic coefficient of variation PCV (%) and genotypic coefficient of variance (GCV) were estimated by the following formula (Xie and Ying 1996):

Repeatability (Rc) among clones in the same site was estimated according to Razafimahatratra et al.(2016):

Repeatability (Rs) among different site were estimated according to Senger et al.(2016):

The additive main effects and multiplicative interactions(AMMI) of height (HT) and basal diameter (D0) were calculated following Zhao et al.(2016) with the following linear model:

whereμis the grand mean,αiis the clone mean deviation,βjis the site effect withj=1,…,3,λkis the singular value for axis k with k=1,…,n,ψikis the clone eigenvector value for axis k,δjkis the site eigenvector for axis k,εijris the error,εijr=Yjirminus the meanμij,andεijris distributed normally with mean zero and standard deviation r.These parameters were calculated by DPS software and the output supplied the parameter principal component analysis (PCA),which can explain the stability of different clones.

Results

Average values of different clones on different sites

Average HT and D0of the clones on the three sites were 3.27 m and 36.99 mm,respectively (Table 3).Average heights and basal diameters were highest on site CH,with values of 4.00 m and 48.49 mm,respectively,higher than the average values on site FJ by 2.14 m and 32.02 mm,and slightly higher than the average values on site LS by 0.04 m and 2.73 mm,respectively.Clone 28 on site CH showed the highest average HT value of 4.38 m,higher than the average value on CH,FJ and LS sites by 0.38,2.52 and 0.42 m,respectively.Clone 28 on site CH also had the highest average D0values of 53.06 mm,higher than the averagevalues on CH,FJ and LS sites by 4.57,36.60 and 7.30 mm,respectively.

Table 3 Statistics of HT and D0 of tested hybrid clones in each site

ANOVA analyses of HT and D0 at different sites

The analysis of variance showed that site,clone number and their interactions were highly significant in overallF-tests(Table 4).The repeatability of HT and D0exceeded 0.50 and varied from 0.65 (D0) to 0.94 (HT).

Genetic and variation parameters of different traits on the same site

Significant differences were observed among clones on the same sites (Table 5).GCVs and PCVs of HT and D0were higher than 15% and ranged from 15.5 to 42.1% and 18.9 to 43.9%,respectively.PCV and GCV of heights on site FJ showed the highest values,43.9% and 42.1%,respectively.Moreover,PCV and GCV of D0on site LS had the lowest values,15.3 and 16.0%,respectively.The repeatability of Ht and D0exceeded 0.50 and varied from 0.55 (D0in site CH)to 0.91 (HT in site FJ).

Table 4 Variance analyses of growth traits of hybrid poplar clones in sites experiment

AMMI analysis of different clones on different sites

AMMI analyses of height and basal diameter for the seven clones on the three sites are presented in Table 6.All three components,i.e.,genotype (G),environment (E) and G × E interaction were significant.Genotype and environment accounted for 3.7% and 94.1% in height and 0.9% and 96.5% in basal diameter of the total variations,respectively.Meanwhile,interaction between genotype and environment of height and basal diameter accounted for 2.1% and 2.6%of the total variation,respectively.The genotype and environment interaction was further partitioned in the principal component analysis (PCA1 and PCA2).The PCA1 of height and basal diameter accounted for 77.9% and 90.1%of the total variation in the interaction,whereas PCA2 was not significant.

Table 5 Variance analysis of growth traits of hybrid poplar clones in each site

Table 6 AMMI ANOVA of 7 genotypes under 3 different environmental conditions

Biplot of genotype-environment interactions

A genotype showing a low interaction with environment is stable over a range of environments.The PCA1 scores of genotypes close to zero showed that the clones were reliable in different environments.In contrast,the genotypes with large absolute PCA1 values were less reliable.Clone 28 and 30 were far from the horizontal ordinate according to Fig.1 of height,which indicated that both clones were the most unstable of all materials in the height trait.For the basal diameter trait,clones 10 and 23 were more unstable than the other clones.Clone 23 and clone 28 were more stable than the other clones in height andbasal diameter traits.As for the average HT and D0,clone 10 and clone 28 showed higher D0or HT than the other clones.The PCA1 (principal component analysis) scores of sites close to zero also showed clone growth to be reliable in the same environment.Although the average values of LS and CH sites were higher than the FJ site in HT and D0traits,the LS and CH sites also showed higher absolute PCA1 values than the FJ,indicating that the clones were unstable in the two sites.

Fig.1 HT (D0) biplot of G × E interaction of different poplar clones.Abscissa indicates the HT (D0) value,and the ordinate shows the PCA1 value of different clones

Genotypes and locations combined with PCA1 scores of the same sign produced positive specific interactions,whereas combinations of positive signs had negative specific interactions.From the AMMI model (Figs.1 and 2),it was determined that LS and FJ sites were positive for clones 10,23,30,and 50 in height traits.Similarly,the LS site had positive interactions with clone 10,18,30,and 50 in basal diameter traits.In contrast,the CH site was positive for clones 18,28,and 51 in height traits,and FJ and CHsites had positive interactions with clones 23,28 and 51 in trait D0.

Fig.2 HT (D0) biplot of G × E interaction of the growth of different clones in different sites.Abscissa indicates the HT (D0)values,and the ordinate shows the PCA1 value of different sites.The average HT and D0 in site FJ are lower than in the other two sites,but in site FJ they were much more stable than in the other two sites

Elite clone selection

In the seedling stage,heights were more important than diameter when elite materials are evaluated.Based on height,clone 10 showed the highest average height among the different sites but clones 23,28 and 10 showed higher HT values on site FJ,CH and LS,respectively.The genetic gain in height by clones 10,23and 28 on sites FJ,CH and LS were 10.4%,7.8% and 11.2%,respectively.Clone 23 showed more stability than clones 10 and 28 but low average values in height.

Discussion

Understanding the variations within and between clones is important in breeding programs (Safavi et al.2010).The ANOVA is an important statistical analysis for estimating and understanding the extent of variability in a breeding population (Zhao et al.2015).In this study,the variations of height and basal diameter showed significant differences among all variance sources,indicating that elite clones and suitable site selection were effective.

Different population parameters,such as phenotypic coefficient of variations and genotypic coefficient of variations,are commonly used to estimate the extent of variability in breeding programs (Liu et al.2015).In this study,the PCV and GCV of height and basal diameter on different sites ranged from 15.3 to 43.9%,exhibiting a wide range of genotype performance across different environments.GCV values were lower than PCV values but were approximate.The results are in agreement with Zhao et al.(2013),suggesting that variations are controlled by genetic factors.Height traits had higher PCV and GCV values than basal diameter traits on different sites;the results are in agreement with Zhao et al.(2014a,2016),indicating that stem height should be considered as a selection trait in the early stages of growth.

The repeatability magnitude revealed that the genotype could be reliably recognized by its phenotypic expression which is needed for estimation,and genetic gain can be achieved through selection (Montes et al.2008).In this research,the repeatability of height and basal diameter traits varied from 0.55 (basal diameter on site CH) to 0.91 (height on site FJ),indicating that much of the variation in these clones was repeatable.The result is in general agreement with Yang et al.(2018) forEucalyptus,Liu et al.(2015)forPopulusand Martins et al.(2014) forCasuarina.High repeatability in clones among different sites indicates that an elite clone selected based on height and basal diameter will be effective and less affected by environmental factors(Maniee et al.2009).

The effects of site mirror the response of species to the combined effects of local and regional climatic and edaphic conditions (Pliura et al.2007).The same genotype might show different phenotypic characteristics at different sites because of phenotypic plasticity.When assessing the suitability of a species for a particular environment,it is important to examine the interactions between the genotype and the environment (Yu and Pulkkien 2003).When such interactions are strong,tree breeders must decide whether to select for stability of performance or to develop populations specifically adapted to each environment with the view of maximizing gain (Zhao et al.2015).In this study,the three sites represent different climate types in northeast China,with different environmental conditions,especially altitude,precipitation,elevation and annual average temperatures.The remarkable differences in height and basal diameter among the clones at the different sites indicates that the environment strongly influenced growth and development(Zhao et al.2014a,b;Liu et al.2015).At the Fujin Forestry Center site,although soil nitrogen levels were higher than in the other sites,low average temperature (3.2 °C) and annual precipitation (517 mm) led to the lowest average heights and basal diameters.The same results were found at the Cuihai Forestry Center site where average temperature is 14.5 °C and annual precipitation is 871 mm;the seedlings had the highest average heights and basal diameters,indicating that,in this experiment,climatic factors were more important than soil nutrient contents.

The AMMI model is a useful tool that combines ANOVA and principal component analysis and then exports the relationship among sites and among genotypes (Balestre et al.2009).The model analysis generates a graphic representation of the major effects of the interaction (PCA1) of both genotypes and environment concurrently (Zhao et al.2015).In an AMMI biplot,genotypes that are distributed near the origin have minimal interactions with the environment but those further away are more sensitive to interactions with the environment (Misra et al.2009).In our study,clones 23 and 28 showed lower absolute values in height and basal diameter traits which indicate that these clones were more stable than other clones.The results also suggested that the two clones exhibited the same characteristics regardless of the environment (Liu et al.2015).Although yield stability data are consequently useful for concurrently selecting highyielding and stable genotypes (Kang and Pham 1991),clones 30 and 28,and clones 10 and 23 showed higher PCA1 absolute values than the other clones in height and basal diameter traits,respectively,indicating that these clones have more unstable growth than the other clones on the three sites.The results also suggest that elite clones should be selected separately on different sites.In this study,based on height traits,different elite clones were selected on different sites,clone 23 on the Fujin Forestry Center site,clone 28 on the Cuihai Forestry Center site,and clone 10 on the Lishu Forestry Center site,indicating the importance of separate selection for different environments.The genetic gains in height of the selected clones ranged from 7.8 to 11.2%,which were similar to previous research (Fries 2012;Zhang et al.2013).The three selected clones could be used to establish future plantations at the different sites.

Conclusion

Growth characteristics are complex in inheritance and are greatly influenced by various environmental conditions.In this study,significant differences among sites,clones,site × clone interactions,and high values of genotypic coefficients of variation,phenotypic coefficients of variation and repeatability indicate that the evaluation and selection of poplar clones were useful.High values of the genotype × environment interaction of growth traits and by different clones vary in different environments,indicating that elite clones should be evaluated and selected separately for different sites.Three clones were selected based on height in different sites should be used to establish future plantations in different sites.The limitations of this study were the short growth period and few investigated traits.Future research should focus on investigating the consecutive and perennial traits where results will be more meaningful.