Adaptation of Norway spruce populations in Europe: a case study from northern Poland§
© The Author(s). 2017
Received: 28 April 2016
Accepted: 13 February 2017
Published: 7 March 2017
The productive potential of European species of forest tree assumes particular importance in the context of populations adapting to accelerating climatic change. Genotype-environment interaction (G × E) was studied to determine Picea abies (L.) H.Karst. (Norway spruce) inter-population variation, characterising their adaptability to the growing conditions in north-eastern Poland. The data were analysed from 22 populations evaluated in four experimental sites based on 5-year height. To identify best-adapted as well as specifically adapted populations, GGE biplots were performed.
Analysis of multi-environment trial (MET) data revealed significant differences between four experimental sites, as well as interactions between populations and sites. However, it proved possible to identify specifically adapted populations achieving high values for the trait at specific sites only, although some performed relatively well across several sites.
The productive potential of the Norway spruce populations in north-eastern Poland is associated with specific adaptation of given populations to growth conditions at the experimental sites. However, in the set of populations studied can also be found some capable of average but stable growth in all experimental sites.
KeywordsPicea abies GGE biplot Genotype × environment interaction Stability Multi-environment trials Growth rate Adaptation
The observed increase in the rate of climate change and accompanying weather anomalies in Europe necessitate ongoing adaptation processes in forest trees, as well as generating changes in the productive potential of stands (Hanewinkel et al. 2012; IPCC 2015; Lindner et al. 2014). The adaptability of trees to new conditions is dependent on genetic diversity, which is largely controlled by population size, the nature and direction of observed climate change, and the direction and intensity of selection (Alberto et al. 2013). Among Europe’s native tree species, Picea abies (L.) H.Karst. (Norway spruce) is characterised by broad adaptability, including the potential to adjust to the predicted water deficit in Central Europe (Lévesque et al. 2013; Zang et al. 2014).
Many studies have considered the potential of Norway spruce populations to adapt to climate change (Ununger et al. 1988; Krajmerová et al. 2009; Kapeller et al. 2012; Nilsson et al. 2012; Ulbrichová et al. 2015). However, the level of knowledge of the productive potential of Polish populations of Norway spruce are still inadequate for forecasting capacity to adapt to foreseen climate change (Matras 2002, 2009).
The results of phylogenetic studies based on molecular markers suggest separate evolutionary origin of Polish populations of Norway spruce with southern populations descended from Alpine refugia while north-eastern populations are from Scandinavian refugia (Lewandowski and Burczyk 2002; Nowakowska 2009; Dering and Lewandowski 2009).
Evaluation of the genotype × environment interaction (G × E) in a multi-environment trials (MET) makes it possible to characterise the genetic conditioning behind the productive potential of forest trees in relation to site conditions (Kang and Gauch 1996). The use of the GGE biplots (graphical method) in MET-based analysis allows the most-representative sites to be identified. Their discriminating ability can also be defined, along with the genotypes best adapted to grow in groups of experimental sites (MET). In addition, it is possible to characterise the stability of genotypes with respect to traits studied, i.e. in terms of rankings or relative performances of genotypes (Yan and Kang 2002). Among the broad range of possible analyses, the methods presented to assess the G × E are being used more and more in METs involving forest tree species (Murillo 2001; Kim et al. 2008; Ding et al. 2008; Correia et al. 2010; Sixto et al. 2015; Ukalski and Klisz 2016; Zhao et al. 2016).
We attempted to assess the influence of genetic and site factors on the productive potential, as assayed by height and survival 5 years after field planting, of populations of Norway spruce from north-eastern Poland. The assessment of the G × E was made based on GGE biplots, with the data used concerning measurements of the heights of trees in the MET.
Materials and methods
Mean values and coefficients of variation (CV%) of tree heights at the individual-tree level and survival in 22 populations of spruce at four experimental sites: CBL Czarna Bialostocka, CDW Czerwony Dwor, GOL Goldap, and SCB Szczebra Czarna sites
Mean height [cm] (CV [%])
Mean survival [%]
The seedlings were produced in Kołaki Forest Nursery (N 53° 18′, E 22° 04′). Sowing was carried out manually in March 2010, one seed per container cell. For producing the seedlings, HIKO V120/40 containers were used (cell volume, 120 cm3/7.32 in.3). Seedlings were kept in polyhouses until June 2010 then they were transferred to open-air conditions. A late frost in April 2011 led to the death of a number of apical shoots so some seedlings were culled from the planting stock at the nursery stage. During all nursery stages, the containers were grouped on pallets with populations distributed randomly. Up to the planting stage, seedlings were not removed from the container in order to avoid mistakes between populations. After 1 year growing in the nursery, planting material was transferred directly to the experimental sites. To acclimatise the seedlings to local growing conditions, planting material was stored close to planting places.
The impact of the environment and genotype on height growth
According to a linear model with genotype and block fixed effects, analysis of variance was performed for each site separately. Based on Tukey HSD test, homogeneous groups were determined for the sites where the genotype effect was significant (p < 0.05).
GGE and GREG biplots
To identify stable populations in terms of height growth at the experimental sites, the GGE biplot was used (Gabriel 1971; Yan et al. 2001). The significance of the individual G × E interaction effects was examined using the Bonferonni test (Dunn 1961), with a 0.01 probability threshold experiment-wise. In turn, GREG biplot (Cornelius et al. 1996; Cornelius and Seyedsadr 1997) was used to determine the similarity between sites, along with the AEC method (Yan and Hunt 2001; Yan 2002). The choice of the GREG biplot was dictated by the fact that first and second principal components for interaction effects (PC1 and PC2, respectively) explained 9.95% more of the variation than did the GGE biplot, as well as the fact that the interaction in this model comprises the effects of E and the G × E, a combination that better characterises the interrelationships between the experimental sites (Ukalski and Klisz 2016).
The similarities between all of the pairs of experimental sites were determined in two ways. In the first method, the value of α ij angles between vectors OAi and OBj were determined (where OAi and OBj are vector lengths between the centre of the coordinate system and position of sites A and B, respectively). In the second method, Pearson correlation coefficients were calculated for each pair of sites on the basis of the principal component values utilising G × E interaction effects.
Statistical analyses and presented biplots were performed using SAS/STAT 13.1 software (SAS Institute Inc 2013) with procedures GLM, MIXED (Littell et al. 1996), PRINCOMP (Khattree and Naik 2000), and GPLOT.
Results and discussion
Analysis of variance
Between-population differences were confirmed for height growth only for the CDW site, while for survival only for the CDW and GOL sites. Details regarding the analyses of variance are given in Additional file 1: Table S1.
Results of analysis of variance for tree heights in 22 populations of spruce at four trial sites, along with the percentage of the variation explained (Q(G) is a quadratic form associated with genotype fixed effect; F statistics are calculated according to Hocking’s approach)
Source of variation
Expected mean square
% (G + E + GE)
Var(ERROR) + 3.69 Var(GE) + Q(G)
0.989*MS(GE) + 0.011*MS(ERROR)
Var(ERROR) + 3.73Var(GE) + 22Var(Block (Site)) + 82.13Var(E)
MS(Block (Site)) + MS(GE) − MS(ERROR)
Var(ERROR) + 22Var(Block (Site))
Genotype × site (GE)
Var(ERROR) + 3.73Var(GE)
Experimental error (ERROR)
The principal components analysis (PCA) showed that the division of effects (G + E + GE) with GGE biplot was as follows: the PC1 explained 51.08% of the variation (G + E + GE), while PC2 explained 25.85%. This made 76.93% in total.
Similarity of environments
The location of sites relative to the position of the AE axis shows that the average height of spruce trees at CDW was almost identical to the height overall mean. Where the comparisons among the other sites were concerned, two—SCB and GOL—were rotated through 90° in the direction of the location of site CBL. A new AE point for these three sites was designated. It emerged that, at the CBL site, the heights of spruce trees were greater than the overall mean, as was also the case for the GOL site, albeit with the differences in comparison with the mean being slight in this case. In contrast, at the SCB site, heights of spruce trees were much below the overall mean.
The vector length of the site is determined by a line drawn through the biplot origin and the site marker and is a measure of discriminating ability of the site. Thus, the sites CBL and CDW, which had the longest vectors, were the most discriminating for genotypes. The line that passes through the plot origin and is perpendicular to the AE axis is a measure of representativeness of the sites (Yan and Hunt 2001). Therefore, site CDW was the most representative (as it had a near-zero length of projection from marker of a site onto the AE axis) and CBL was the least.
Correlation coefficients (upper triangle in the table) and α ij angles (lower triangle) for pairs of sites as of 2010
Stability of populations
Most of the observed variation among populations is explained by the G × E rather than across-site main effects of populations. The growth potential of the northern populations shows no clear geographical trend. Population 52692 from Białowieża Forest shows specific adaptation to growth at experimental sites CBL and CDW. Population 1 (national standard from southern Poland) shows some lack of adaptation to the growing conditions in northern Poland. Site CDW seems to be the most-representative site, while CBL and SCB are characterised by the greatest discriminating ability.
Average environment coordination
Best linear unbiased predictor
Czarna Białostocka experimental site
Czerwony Dwór experimental site
Effect of environment
Effect of genotype
- G × E:
Genotype and genotype-by-environment interaction biplot
Gołdap experimental site
Genotype regression biplot
First principal component
Second principal component
Principal components analysis
Szczebra experimental site
The study was funded by the State Forests National Forest Holding as part of research project BLP-375. The authors are grateful to Dr Rowland Burdon for his insightful comments helping to make this a stronger manuscript.
MK is responsible for the data collection, preparation of manuscript, text corrections, coordinating work; SzJ is responsible for the initial preparation, data collection, text corrections; KU is responsible for the proposal and performance of statistical analysis and contribution to interpretation of results; JU is responsible for the description and performance of statistical analysis and correction of manuscript; PP is responsible for the data collection, text corrections. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
- Alberto, F. J., Aitken, S. N., Alía, R., González-Martínez, S. C., Hänninen, H., Kremer, A., et al. (2013). Potential for evolutionary responses to climate change—evidence from tree populations. Global Change Biology, 19(6), 1645–1661. doi:10.1111/gcb.12181.View ArticlePubMedPubMed CentralGoogle Scholar
- Balzarini, M. (2002). Applications of mixed models in plant breeding. In M. S. Kang (Ed.), Quantitative genetics, genomics, and plant breeding (pp. 353–365). New York: CABI Publishing.Google Scholar
- Bentzer, B. G., Foster, G. S., Hellberg, A. R., & Podzorski, A. C. (1988). Genotype × environment interaction in Norway spruce involving three levels of genetic control: seed source, clone mixture, and clone. Canadian Journal of Forest Research, 18(9), 1172–1181. doi:10.1139/x88-180.View ArticleGoogle Scholar
- Cornelius, P. L., & Seyedsadr, M. S. (1997). Estimation of general linear-bilinear models for two-way tables. Journal of Statistical Computation and Simulation, 58(4), 287–322. doi:10.1080/00949659708811837.View ArticleGoogle Scholar
- Cornelius, P. L., Crossa, J., & Seyedsadr, M. (1996). Statistical tests and estimators of multiplicative models for cultivar trials. In M. S. Kang & H. G. Gauch Jr. (Eds.), Genotype- by-Environment Interaction (pp. 199–234). Boca Raton: CRC Press.Google Scholar
- Correia, I., Alia, R., Yan, W., David, T., Aguiar, A., & Almeida, M. H. (2010). Genotype × Environment interactions in Pinus pinaster at age 10 in a multi-environment trial in Portugal: a maximum likelihood approach. Annals of Forest Science, 67(6), 612p1–612p9. doi:10.1051/forest/2010025.View ArticleGoogle Scholar
- Dering, M., & Lewandowski, A. (2009). Finding the meeting zone: where have the northern and southern ranges of Norway spruce overlapped? Forest Ecology and Management, 259, 229–235. doi:10.1016/j.foreco.2009.10.018.View ArticleGoogle Scholar
- Ding, M., Tier, B., Yan, W., Wu, H. X., Powell, M. B., & McRae, T. A. (2008). Application of GGE biplot analysis to evaluate Genotype (G), Environment (E), and G × E interaction on Pinus radiata: a case study. New Zealand Journal of Forestry Science, 38(1), 132–142.Google Scholar
- Dunn, O. J. (1961). Multiple comparisons among means. Journal of the American Statistical Association, 56(293), 52–64. doi:10.1080/01621459.1961.10482090.View ArticleGoogle Scholar
- Gabriel, R. K. (1971). The biplot graphic display of matrices with application to principal component analysis. Biometrika, 58(3), 453–467. doi:10.1093/biomet/58.3.453.View ArticleGoogle Scholar
- Hanewinkel, M., Cullmann, D. a., Schelhaas, M.-J., Nabuurs, G.-J., & Zimmermann, N. E. (2012). Climate change may cause severe loss in the economic value of European forest land. Nature Climate Change, 3(3), 203–207. doi:10.1038/nclimate1687.View ArticleGoogle Scholar
- Hocking, R. R., & Speed, F. M. (1975). A full rank analysis of some linear model problems. Journal of the American Statistical Association, 70(351), 706–712. doi:10.2307/2285959.View ArticleGoogle Scholar
- IPCC. (2015). In Core Writing Team, R. K. Pachauri, & L. A. Meyer (Eds.), Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Geneva: IPCC.Google Scholar
- Kang, M. S., & Gauch, H. G. (1996). Genotype-by-Environment Interaction. London: CRC Press. doi:10.1201/9781420049374.fmatt.Google Scholar
- Kapeller, S., Lexer, M. J., Geburek, T., Hiebl, J., & Schueler, S. (2012). Intraspecific variation in climate response of Norway spruce in the eastern Alpine range: selecting appropriate provenances for future climate. Forest Ecology and Management, 271, 46–57. doi:10.1016/j.foreco.2012.01.039.View ArticleGoogle Scholar
- Karlsson, B., Wellendorf, H., Roulund, H., & Werner, M. (2001). Genotype × trial interaction and stability across sites in 11 combined provenance and clone experiments with Picea abies in Denmark and Sweden. Canadian Journal of Forest Research, 31(10), 1826–1836. doi:10.1139/cjfr-31-10-1826.View ArticleGoogle Scholar
- Khattree, R., & Naik, D. N. (2000). Multivariate Data Reduction and Discrimination with SAS Software. Cary: SAS Institute Inc.Google Scholar
- Kim, I. S., Kwon, H. Y., Ryu, K. O., & Wan, Y. C. (2008). Provenance by site interaction of Pinus densiflora in Korea. Silvae Genetica, 57(3), 131–139.Google Scholar
- Krajmerová, D., Longauer, R., Pacalaj, M., & Gömöry, D. (2009). Influence of provenance transfer on the growth and survival of Picea abies provenances. Dendrobiology, 61(SUPPL. 1), 17–23.Google Scholar
- Lévesque, M., Saurer, M., Siegwolf, R., Eilmann, B., Brang, P., Bugmann, H., & Rigling, A. (2013). Drought response of five conifer species under contrasting water availability suggests high vulnerability of Norway spruce and European larch. Global Change Biology, 19(10), 3184–3199. doi:10.1111/gcb.12268.View ArticlePubMedGoogle Scholar
- Lewandowski, A., & Burczyk, J. (2002). Allozyme variation of Picea abies in Poland. Scandinavian Journal of Forest Research, 17(6), 487–494. doi:10.1080/02827580260417134.View ArticleGoogle Scholar
- Lindner, M., Fitzgerald, J. B., Zimmermann, N. E., Reyer, C., Delzon, S., Van Der Maaten, E., et al. (2014). Climate change and European forests: what do we know, what are the uncertainties, and what are the implications for forest management? Journal of Environmental Management, 146, 69–83. doi:10.1016/j.jenvman.2014.07.030.View ArticlePubMedGoogle Scholar
- Littell, R. C., Milliken, G. A., Stroup, W. W., & Wolfinger, R. D. (1996). SAS System for Mixed Models. Cary: SAS Institute Inc.Google Scholar
- Matras, J. (2002). Growth and development of polish provenances of Norway spruce (Picea abies Karst.) in the IUFRO 1972 experiment. Forest Research Papers, 947(4), 73–97. https://www.ibles.pl/web/lesne-prace-badawcze/-/prace-instytutu-badawczego-lesnictwa-a-2002-4-73-97.Google Scholar
- Matras, J. (2009). Growth and development of Polish provenances of Norway spruce (Picea abies Karst.) in the IUFRO 1972 experiment. Dendrobiology, 61, 145–158.Google Scholar
- Murillo, O. (2001). Genotype by environment interaction and genetic gain on unbalanced Pinus oocarpa provenances trials. Agronomia Costarricense, 25(1), 21–31.Google Scholar
- Nilsson, U., Elfving, B., & Karlsson, K. (2012). Productivity of Norway spruce compared to Scots pine in the interior of Northern Sweden. Silva Fennica, 46(2), 197–209.View ArticleGoogle Scholar
- Nowakowska, J. A. (2009). Mitochondrial and nuclear DNA differentiation of Picea abies populations in Poland. Dendrobiology, 61(SUPPL. 1), 119–129.Google Scholar
- Piepho, H. P. (1998). Empirical best linear unbiased prediction in cultivar trials using factor-analytic variance structures. Theoretical and Applied Genetics, 97(1–2), 195–201. doi:10.1007/s001220050885.View ArticleGoogle Scholar
- Piepho, H. P., & Möhring, J. (2006). Selection in cultivar trials—is it ignorable? Crop Science, 46(1), 192–201. doi:10.2135/cropsci2005.04-0038.View ArticleGoogle Scholar
- Piepho, H. P., Möhring, J., Melchinger, A. E., & Büchse, A. (2008). BLUP for phenotypic selection in plant breeding and variety testing. Euphytica, 161, 209–222. doi:10.1007/s10681-007-9449-8.View ArticleGoogle Scholar
- SAS Institute Inc. (2013). SAS/STAT 13.1 User’s Guide. Cary: SAS Institute Inc.Google Scholar
- Saxton, A. M. (2004). Genetic Analysis of Complex Traits Using SAS (pp. 1–292). Cary: SAS Institute Inc.Google Scholar
- Searle, S. R., Casella, G., & McCulloch, C. E. (1992). Variance Components (pp. 1–537). New York: John Wiley and Sons.Google Scholar
- Sixto, H., Gil, P. M., Ciria, P., Camps, F., Cañellas, I., & Voltas, J. (2015). Interpreting genotype by environment interaction for biomass production in hybrid poplars under short rotation coppice in Mediterranean environments. GCB Bioenergy. doi:10.1111/gcbb.12313.Google Scholar
- Ukalski, K., & Klisz, M. (2016). Application of GGE biplot graphs in multi-environment trials on selection of forest trees. Folia Forestalia Polonica Series A-Forestry, 58(4), 228–239. doi:10.1515/ffp-2016-0026.Google Scholar
- Ulbrichová, I., Podrázský, V., Beran, F., Zahradník, D., Fulín, M., Procházka, J., & Kubeček, J. (2015). Picea abies provenance test in the Czech Republic after 36 years – Central European provenances. Journal of Forest Science, 61(11), 465–477. doi: 10.17221/23/2015-JFS.View ArticleGoogle Scholar
- Ununger, J., Ekberg, I., & Kang, H. (1988). Genetic control and age-related changes of juvenile growth characters in Picea abies. Scandinavian Journal of Forest Research, 3, 55–66. doi:10.1080/02827588809382495.View ArticleGoogle Scholar
- Yan, W., Cornelius, P. L., Crossa, J., & Hunt, L. A. (2001). Two types of GGE biplot for analyzing multienvironmental trial data. Crop Science, 41, 656–663. doi:10.2135/cropsci2001.413656x.View ArticleGoogle Scholar
- Yan, W. (2001). GGE biplot: a Windows application for graphical analysis of multi-environment trial data and other types of two-way data. Agronomy Journal, 93, 1111–1118. doi:10.2134/agronj2001.9351111x.View ArticleGoogle Scholar
- Yan, W. (2002). Singular-value partitioning in biplot analysis of multienvironment trial data. Agronomy Journal, 94(5), 990–996. doi:10.2134/agronj2002.0990.View ArticleGoogle Scholar
- Yan, W., & Hunt, L. A. (2001). Interpretation of genotype. Crop Science, 41(1), 19–25. doi:10.2135/cropsci2001.41119x.View ArticleGoogle Scholar
- Yan, W., & Kang, M. S. (2002). GGE Biplot Analysis: A Graphical Tool for Breeders, Geneticists, and Agronomists. London: CRC Press. doi:10.1201/9781420040371.View ArticleGoogle Scholar
- Zang, C., Hartl-Meier, C., Dittmar, C., Rothe, A., & Menzel, A. (2014). Patterns of drought tolerance in major European temperate forest trees: climatic drivers and levels of variability. Global Change Biology, 20(12), 3767–3779. doi:10.1111/gcb.12637.View ArticlePubMedGoogle Scholar
- Zhao, X., Xia, H., Wang, X., Wang, C., Liang, D., Li, K., & Liu, G. (2016). Variance and stability analyses of growth characters in half-sib Betula platyphylla families at three different sites in China. Euphytica, 208(1), 173–186. doi:10.1007/s10681-015-1617-7.View ArticleGoogle Scholar