Introduction

Most plant secondary metabolites are thought to enhance the prospects for survival of the producer or its offspring through complex interactions with the environment (Waterman, 1992). The effect of environment on secondary chemistry has been considered to be of prime importance, as suggested by many ecological hypotheses such as the carbon nutrient balance hypothesis (see Bryant et al., 1983), the growth differentiation hypothesis (Herms and Mattson, 1992), and the protein competition model (Jones and Hartley, 1999). Although the role of environmental and genetic control of plant secondary chemistry is of major importance, few studies have concentrated on this question in forest trees. Hanover (1966) studied the heritability of five terpenes present in pine, and Orians et al. (1996) studied heritability of two phenolics in Salix (see also Hamilton et al., 2001). The reason for the paucity of studies on the heritability of tree secondary compounds might be the need for time-consuming field experiments and the complicated chemical analyses.

When birch species (Betula spp.), and even individuals within a species, have been compared, a high inter- and intraspecific variation in the types and concentrations of secondary compounds in leaves has been found (Keinänen etal., 1999a, b; Laitinen et al., 2000, 2002; Salminen et al., 2002; Valkama et al., 2003). Similar results have also been reported in studies of bark chemistry, where different birch species have been compared (Taipale et al., 1994; Julkunen-Tiitto et al., 1996). Variations in the amounts of secondary compounds within birch species are known to be induced by both biotic and abiotic environmental factors such as defoliation, available nutrients, UV-light, temperature, etc. (Keinänen et al., 1999a; Tegelberg et al., 2001, 2002; Kuokkanen etal., 2001, 2003). However, in previous studies, the high phytochemical variation in naturally regenerated individual European white birch trees (Laitinen et al., 2000), and bark secondary chemistry among birch seedlings, were suggested to be mainly controlled by genotype (Tahvanainen et al., 1991). Accordingly, our previous study indicated large clonal differences among 1-yr-old greenhouse-grown plantlets, suggesting that intraspecific variation in bark phenolic chemistry could be explained, to a large extent, by genetic differences (Laitinen et al., 2004). However, the high variation in terpenoid chemistry among clones and plantlets of the same clone suggests that for some genotypes, the environment might also modify the secondary chemistry and, thus,affect the resistance of birch against browsing of mammals (Laitinen et al., 2004).

During plant growth and ontogenetic development, secondary chemistry can change considerably. Generally, juvenile stages express stronger defenses than the mature stage (Bryant and Julkunen-Tiitto, 1995). At the birch population level, this would mean even more variation in secondary chemistry, which allows populations to adapt to different kinds of stresses. However, the causes of variation in chemistry within a natural population of Betula pendula are still scanty.

In the present study, we used samples from mature birch trees from our long-term study population (see Laitinen et al., 2000, 2002) and micropropagated clonal plantlets from the same trees. Clonal plantlets were grown in four experimental sites, to study the effect of genotype, environment, and ontogeny on the chemistry of B. pendula. The objectives were to: (1) examine variation in shoot secondary chemistry within a naturally regenerated population among individual trees and within a tree, (2) determine the degree to which genotype and environment affect birch shoot secondary chemistry, (3) estimate the heritability of secondary chemistry by studying clonal repeatability, (4) test how individual clones differ in their response to the environment, and (5) compare parental trees to clonal plantlets of the same trees in order to quantify the effect of ontogeny.

Methods and Materials

Plant Material and Growing Conditions.

A naturally regenerated birch (B.pendula) stand in Punkaharju (southeastern Finland; 61°48′N, 29°18′E) was selected for long-term studies on genetic variation within a natural population. From this stand, a random sample of 30 B. pendula trees were selected (for details, see Laitinen et al., 2000). The aim was to micropropagate 300 plantlets from each tree. In some cases, the micropropagation was not successful, which limited the number of clones in this study to 14. The identification numbers of the available trees and clones were 3, 4, 5, 6, 9, 15, 16, 17, 18, 19, 20, 24, 25, and 30. These are the same parent trees and clones that were also used in our previous studies (see Laitinen et al., 2000, 2002, 2004).

Parental trees were about 20-yr-old at the time of micropropagation. Small twigs were taken from the upper third of the trees, twice in June and once in August 1997, to get material for micropropagation. In the beginning of April 1998, micropropagated plantlets were transferred from laboratory cultures into 0.021 l plastic pots (EK 144) that contained perlite, unfertilized peat, and birch forest soil (2:2:1). Plantlets were grown in a greenhouse, where plantlets from each clone were divided onto three tables in which the pots of all clones were placed randomly.

In May 1998, plantlets were transferred into 49 × 30 cm plastic containers (EK 28). Each container contained 28 pots of 0.28 l each, and each pot was filled with prefertilized commercial peat (VAPO, Finland). On the July 9, 1998, plantlets were transferred to an open-air exposure field in Punkaharju. Plantlets were watered as needed and fertilized once (April 22) with 0.1%, and six times (May 27, June 17 and 24, and July 1, 8, and 15) with 0.2% Kekkilä 9-Superex (N 19.4%, P 5.3%, K 20%) fertilizer. At the end of October 1998, winter-hardened plantlets were transferred to an unheated greenhouse where the roots were protected from freezing.

Plantlets for this study were planted in June 1999 into one of three growing sites in Punkaharju (Kuikanniitty 61°47′N, 29°21′E, 79 m above sea level, Putikko 61°43′N, 29°26′E 98 m above sea level, and Vaahersalo; 61°47′N, 29°17′E, 95 m above sea level) and one site in Parikkala (Parikkala 61°36′N, 29°36′E, 93 m above sea level), in southeastern Finland. The sites at Kuikanniitty and Vaahersalo were abandoned cultivated fields, whereas Parikkala and Putikko were Myrtillus forest-type areas. The soil type was defined as fine sandy till at all growing sites.

Plantations were established using a randomized complete block design. At each of the growing sites, there were six to nine blocks. In each block, four replicates (plantlets) of each clone were planted. A random selection of six blocks per site and one plantlet per clone per block (= six samples per clone per growing site) was made. A current year’s growth from the first branch, from the top of clonal plantlets, and five branches of current year’s growth from the upper third part of parental trees were collected on June 26, 2001. Branches were air-dried and stored at −20°C for chemical analysis.

Extraction of Compounds.

Both phenolics and triterpenoids were analyzed from the same extract. For extraction, leaves were removed from study branches, and the current year’s growth (≈ 10–15 cm) was cut into 2-mm pieces and weighed. Methanol (20 ml; 100%) was added, the sample was allowed to stand for 20 min onice (≈ +4°C), and homogenized for 2 min using an Ultra-Turrax clipping homogenizer. The extraction was repeated ×2 with 15 ml of methanol and once more with 15 ml of diethyl ether for complete extraction of triterpenoids. The residue was washed with methanol. Extracts and washings were filtered and combined, and the solvents removed by vacuum evaporation. Each sample was redissolved in 10 ml of 100% methanol, and aliquots of 2×1, 1×2, and 1×3 ml were evaporated to dryness under nitrogen. Different aliquots were taken to ensure suitable concentrations for both HPLC-DAD and HPLC-MS runs. Dried samples were stored at −20°C for analysis by HPLC-DAD and HPLC-MS.

HPLC-DAD and HPLC-MS Analysis.

Quantitative analysis of the low molecular weight phenolics (LMWP) in stem samples was done by HPLC using gradient elution and diode array detection (DAD) as previously described by Laitinen et al. (2000). The wavelengths used for detection were 220, 270, 280, 320, and 360 nm. Identification of quercetin derivatives, 3,4′-dihydroxypropiophenone-3-β-d-glucopyranoside (DHPPG), caffeoyl quinic acids, cinnamic acid derivatives, and flavonoid aglycones, was based on comparison of HPLC retention times and spectral characteristics described in Keinänen and Julkunen-Tiitto (1998) and Julkunen-Tiitto et al. (1996). Secondary metabolites were quantified against commercial standards: salicin (Roth, Karlsruhe, Germany) for salidroside, DHPPG, rhododendrin, and platyphylloside; (+)-catechin hydrate (Aldrich Chemical Company, USA) for catechin derivatives; chlorogenic acid (Aldrich, Steinheim, Germany) for all phenolic acids, quercetin-3-O-glucoside (Extrasynthèse, Genay, France) for all quercetin derivatives; gallic acid (Aldrich) for gallotannins, and apigenin (Roth) for flavonoid aglycones. Before HPLC-DAD analysis, dried 1 ml phenolic samples were redissolved in 0.4 or 0.6 ml water/methanol (1:1), depending on sample concentration. Quantitative analysis of triterpenoid components was done by HPLC-MS. Isocratic elution was used with EtOH (94%): aqueous 1.5% tetrahydrofuran + 0.25% acetic acid (75:25, pH 5.4). HPLC/atmospheric pressure electrospray ionization (HPLC/API-ES) conditions were as follows: the column was a Hypersil RP C-18, 2-mm ID × 10 cm, the ES fragmentor voltage was 100 V, and the flow rate was 0.2 ml/min. Before HPLC-MS analysis, dried 1 ml terpenoid samples were redissolved in 5 ml clonal plantlets, and dried 3-ml samples were redissolved in 3 ml parental trees, 94% ethanol. Triterpenoid components were quantified by using purified papyriferic acid, which was kindly supplied by Prof. Paul Reichardt, University of Alaska, USA. The concentrations of pendulic acid were considered to be relative because papyriferic acid was used as a standard.

In our figures, phenolic acids are divided into two groups: neochlorogenic acid and chlorogenic acid, and six chlorogenic acid derivatives were grouped together as caffeoyl quinic acids, and three p-hydroxycinnamic acid-like compounds were grouped as total cinnamic acid derivatives. Catechin derivatives were combined as the total of four components [(+)-catechin and three catechin derivatives]. Total quercetin derivatives contained quercetin-3-galactoside, quercetin-3-glucuronide, quercetin-3-glucoside, quercetin-3-arabinoside, and four unknown quercetin derivatives. The flavonoid aglycone group contained several components, some of which could not be resolved in all samples by our method. Apigenin and one luteolin derivative were counted together, as was also done for chrysoeriol and one other luteolin derivative. In addition, a third luteolin derivative and two apigenin derivatives were included in the flavonoid aglycone group. Total low molecular weight phenolics (LMWP) include all of the analyzed phenolics, except condensed tannins. For triterpenoids, the papyriferic acid and four papyriferic acid derivatives were grouped as total papyriferic acid derivatives, and two pendulic acid derivatives were grouped to represent the total of pendulic acid derivatives. Total triterpenoids include all analyzed triterpenoids.

Analysis of Condensed Tannins.

Condensed tannins were analyzed from stem extract by an acid butanol assay as previously described in Porter et al. (1986). Quantification of tannins was based on purified B. nana leaf tannin.

Statistics.

For evaluating chemical similarity in phenolic composition between parental trees within a population, a classification of parental trees to putative chemotypes was made by UPGMA clustering as previously described in Laitinen et al. (2000). Mean values of five samples per tree of all analyzed phenolic compounds were used for UPGMA-clustering analysis.

Due to nonnormality and heterogeneity of variances, the concentrations of the following compounds were transformed for statistical tests: a square-root (x+ 1) transformation was used for parental tree concentrations of quercetin-3-glucuronide + quercetin-3-glucoside; a natural logarithm (x + 1) transformation of concentration was used for salidroside, chlorogenic acid, and quercetin-3-galactoside; a natural logarithm (x + 0.5) transformation of concentration was used for rhododendrin, quercetin-3-arabinoside, and quercetin derivative 3; and a natural logarithm (x + 0.1) transformation of concentration was used for platyphylloside, gallotannins, and all terpenoid compounds and compound groups. Data from micropropagated plantlets were treated by a square-root transformation for concentrations of quercetin-3-glucuronide + quercetin-3-glucoside. A natural logarithm of concentration + 1 transformation was used for papyriferic acid and total pendulic acids; a natural logarithm of concentration +0.5 was used for DHPPG, catechin, chlorogenic acid, quercetin-3-galactoside, quercetin-3-arabinoside, and flavonoid aglycones, and a natural logarithm of concentration + 0.1 was used for gallotannins and platyphylloside.

Differences in the amount of chemical compounds or compound groups among parental trees, growing sites, and clones were tested by analysis of covariance (ANCOVA). For analysis of chemical data between clones and growing sites, the following model was used: site (S), clone (C), S × C, S × block (B), weight of the current year’s growth was used as covariate. Growing site and clone were treated as fixed factors, and block was used as a random factor. Weight was used as a covariate to remove an additional factor that may give rise to variation, i.e., the wood-to-bark ratio, which varies with different sizes of branches. A sequential Bonferroni correction was used for evaluating the results of all ANCOVA tests for parental trees and for each growing site (Rice, 1989).

A variance component analysis was used to evaluate the effect of genotype on total chemical variation. For clonal plantlets, the observed phenotypic variance (VP) can be partitioned into genotypic variance (VG) and environmental variance (VE), i.e., VP = VG + VE. In the case of clonal material, plantlets within a clone have an equal genetic constitution, thus, the within clone, variance component is due to variation in the environment or an error in sampling and measuring (Falconer, 1989). The between-clone component of variance is an estimate of VG. The total genotypic variance component contains both the additive and nonadditive components, which cannot be separated by using clonal material. Thus, the extent to which individual phenotype is determined by genotype was expressed by the degree of genetic determination, i.e., heritability in a broad sense = VG / VP (Falconer, 1989). When studying clonal plantlets, there are environmental effects included in VG, because some part of the environmental differences between parental plants may be transmitted to all of their clonal descendants. When the degree of genetic determination is estimated by using clones, the ratio VG / VP is an overestimate and, strictly speaking, it should be referred to as clonal repeatability as suggested by Falconer (1989, p. 128). Thus, the term clonal repeatability was also used in this study. The estimation of variance components was computed for all compounds and compound groups (varcomp procedure, REML), and the results are presented as variation between parental trees with 95% confidence limits (Searle, 1997) and as clonal repeatability ± SE (Falconer, 1989; Dickerson, 1969) for clonal plantlets. All figures are based on nontransformed data. All statistical analyses were made using SPSS 10.1.0, SPSS Inc., 1989–2000.

Results

Chemical Variation among Parental Trees and within Individual Trees.

There was significant variation among parental trees for all analyzed compounds and compound groups (univariate analysis of variance with sequential Bonferroni correction, P < 0.001) (Table 1 and Figures 1–5). The differ-ences among parental trees accounted for most of the variation in chemical concentrations for phenolics, whereas for triterpenoid groups, pendulic acid derivatives, and total terpenoids, the variation within an individual parental tree was greater than that among parental trees (Table 1). For individual phenolic compounds, in many cases, the variation among trees was large. For example, inthe case of (+)-catechin, the mean concentration of tree 25 was only 1.1 mg/g(DW), whereas several other trees contained more than 5 mg/g (DW) of (+)-catechin (Figure 3). With respect to chlorogenic acid, the differences among trees were more than 10-fold and nearly fivefold for salidroside (Figure 3).

Table 1 Results for Analyzed Compound Groups and Individual Compounds: F-Values and Significance Levels with Sequentical Bonferroni Correction for Tree (Parental Trees) by Analysis of Covariance, and Results of Variance Component Analysis for Variation Between Parental Trees and Between Clones
Full size table

Four distinct chemotypes were found among the 30 study trees by qualitative UPGMA clustering (Figure 1, groups 1–4). In the chemotype grouping (Figure 1), (+)-catechin and chlorogenic acid (compounds 7 and 8) were the main compounds in groups 1 and 2. One chlorogenic acid derivative (compound 6) was lacking from group 2; also, other chlorogenic acid derivatives (compounds 27 and 28) were lacking or only present in trace amounts. In group 3, the (+)-catechin and quercetin-3-galactoside were the main compounds. Tree 13 was separated from all other trees. It had no clear main compounds; that is, salidroside, (+)-catechin, chlorogenic acid, rhododendrin, catechin derivative 1, and quercetin-3-galactoside (compounds 1, 7, 8, 10, 11, and 14 in Figure 1) were all found in moderate concentrations. Minor compounds, such as p-OH-cinnamic acid derivatives, catechin derivatives, and platyphylloside, were present in all chemotype groups.

Fig. 1
figure 1figure 1

UPGMA dendrogram for the chemical similarity in parental trees. Trees with an example chemical profile are marked with bold numbers. The chemical structures of main compounds used for defining different chemotypes are shown: for groups 1 and 2, (+)-catechin and chlorogenic acid; for group 3, (+)-catechin and quercetin-3-galactoside. Small arrows indicate the other compounds, which aided to define different chemotypes. Individual compounds are listed in order of retention times to illustrate the HPLC profile. Analyzed compounds that were used in UPGMA clustering are: 1 = salidroside; 2 = DHPPG; 3 = neochlorogenic acid; 4, 5, and 9 = p-OH-cinnamic acid derivatives; 6, 17, 21, 25, 27, and 28 = chlorogenic acid derivatives; 7 = (+)-catechin; 8 = chlorogenic acid; 10 = rhododendrin; 11, 12, and 13 = catechin derivatives; 14 = quercetin-3-galactoside; 15 = quercetin-3-glucoside + quercetin-3-glucuronide; 16 = quercetin-3-arabinoside; 18, 20, 23, and 24 = quercetin derivatives; 19 = pentagalloylglucose; 22 = platyphylloside; 26 and 29 = gallotannin derivatives; 30 = apigenin + luteolin derivative; 31 = chrysoeriol + luteolin derivative; 32 and 33 = apigenin derivatives.

Full size image

Clonal Variation in Chemistry: Effects of Genotype and Environment.

There were significant differences among clones for all analyzed compounds and compound groups (Figures 2–5). For compound groups such as cinnamic acid derivatives, total quercetin derivatives, flavonoid aglycones, and gallotannins (Figure 2), and for individual compounds such as quercetin derivatives, DHPPG, and platyphylloside (Figures 3 and 4), there were significant differences only among clones; that is, there was no detected environmental effect in these cases. Especially high variation among clones for individual compounds was found for quercetin-3-glucoside + quercetin-3-glucuronide (these compounds could not be separated by our analytical method); the concentrations varied from 0.1 mg/g (DW) in clone 25 to 1.1 mg/g (DW) in clone 19. For DHPPG, the concentrations varied from 0 mg/g (DW) in clone 3 to 2.1 mg/g (DW) in clone 5 (Figures 3 and 4).

Fig. 2
figure 2figure 2

Concentrations per parental tree or clone for analyzed compound groups. For parental trees, an average ± SE of five stem samples was used, and for individual clones, an average ± SE of six stem samples was counted for each growing site. Results are expressed as milligrams per gram of dry weight (DW) of plant material. Trees and clones are grouped by chemotype (see UPGMA clustering in Figure 1).

Full size image
Fig. 3
figure 3figure 3

Concentrations per parental tree or clone for main phenolic compounds: (+)-catechin, chlorogenic acid, DHPPG, rhododendrin, salidroside, and platyphylloside. For parental trees, an average ± SE of five stem samples was used, and for individual clones, an average ± SE of six stem samples was counted for each growing site. Results are expressed as millgrams per gram of dry weight (DW) of plant material. Trees and clones are grouped by chemotype (see UPGMA clustering in Figure 1).

Full size image
Fig. 4
figure 4figure 4

Concentrations per parental tree or clone for main phenolic compounds: main quercetin derivatives. For parental trees, an average ± SE of five stem samples was used, and for individual clones, an average ± SE of six stem samples was counted for each growing site. Results are expressed as milligrams per gram of dry weight (DW) of plantmaterial. Trees and clones are grouped by chemotype (see UPGMA clustering in Figure 1).

Full size image

When the degree of genetic determination for the birch shoot secondary chemistry was calculated by the variance component analysis for compound groups, the highest values for clonal repeatability were found for caffeoyl quinic acids and cinnamic acid derivatives (Table 1). The genetic control of birch shoot secondary chemistry differed greatly among individual compounds or compound groups, as shown by high variation in repeatability values. For individual compounds, the clonal repeatability was highest (over 0.5) for chlorogenic acid, quercetin-3-glucoside, quercetin-3-glucoside + quercetin-3-glucuronide, and DHPPG (Table 1).

Significant environmental effects on some of the analyzed chemical groups were detected for the clonal plantlets, namely, for catechin derivatives, caffeoyl quinic acids, condensed tannins, and LMWP (Figure 2), and for total pendulic acid derivatives and total triterpenoids (Figure 5). For individual compounds, significant differences among growing sites were found for (+)-catechin, chlorogenic acid, rhododendrin, and salidroside (Figures 3 and 6a). For example, the mean concentration for (+)-catechin varied from 1.2 mg/g (DW) in Kuikanniitty to 2.7 mg/g (DW) in Vaahersalo (Figure 3). The minimum and maximum mean concentrations for chlorogenic acid were 0.83 mg/g (DW) in Kuikanniitty and 1.26 mg/g (DW) in Parikkala, respectively. Similar results were obtained for rhododendrin.

Fig. 5
figure 5figure 5

Concentrations per parental tree or clone for papyriferic acid and triterpenoid groups. For parental trees, an average ± SE of five stem samples was used, and for individual clones, an average ± SE of six stem samples was counted for each growing site. Results are expressed as milligrams per gram of dry weight (DW) of plant material. Trees and clones are grouped by chemotype (see UPGMA clustering in Figure 1).

Full size image

There was significant site × clone interaction for all individual terpenoid compounds and for total terpenoids (Figure 5), as well as for two of the phenolic compounds, rhododendrin, and salidroside (Figure 3). This indicates that the response of individual clones to differences in growing environment differed among clones; that is, the environmental sensitivity was dependent on individual plant genotype (Figure 6b).

Fig. 6
figure 6figure 6

Environmental sensitivities of individual clones. Environmental values are expressed as an average of all analyzed samples within a particular growing site, and genotypic values are expressed as an average of samples from one clone in a particular growing site. The value of each genotype is plotted against the environmental mean of all genotypes to determine differences in the environmental sensitivities of different genotypes. a Environmental values are 0.83 mg g−1 (DW) for Kuikanniitty (KU) and Putikko (PU), 1.02 mg g−1 (DW) for Vaahersalo (VA), and 1.26 mg g−1 (DW) for Parikkala (PA). b environmental values are 7.34 mg g−1 (DW) for PA, 7.44 mg g−1 (DW) for VA, 8.12 mg g−1 (DW) for PU, and 8.52 mg g−1 (DW) for KU. Linear regressions are presented as lines for all clones. The equation and R2 values are presented for those clones differing mostly from each other.

Full size image

Comparison between Parental Trees and Clonal Plantlets.

In both developmental stages of birches, the main phenolic compound group was condensed tannins (Figure 2). The greatest difference in chemical quality between parental trees and clonal plantlets was found in triterpenoids (Figure 5). The main phenolic compounds in the shoots of parental trees and in the shoots of young plantlets were salidroside, (+)-catechin chlorogenic acid, and quercetin-3-galactoside (Figures 1–4). For phenolics in general, the chemical profile was similar in parental trees and in their micropropagated plantlets. Two triterpenoid compounds, papyriferic acid, and one pendulic acid derivative were found insamples from parental trees (Figure 5), whereas in clonal plantlets, papyriferic acid and four papyriferic acid derivatives were found. In addition, two pendulic acid derivatives were found only in young shoots of micropropagated plantlets.

Quantitatively, there were great differences in secondary chemistry between parental trees and clonal plantlets. This difference can be seen when the results from phenolic compound groups are compared. The amount of condensed tannins in parental trees was twice the amount found in clonal plantlets (Figure 2). Also, the concentrations of catechin derivatives and caffeoyl quinic acid derivatives were clearly higher in parental trees. For triterpenoid concentrations, the difference between parents and clonal plantlets was even more pronounced (Figure 5). All parental trees had less than 1 mg/g (DW) of terpenoids in their current year growth. In clonal plantlets, the amounts of terpenoids varied from 10.2 mg/g (DW) (clone 30 in Vaahersalo) to 64.6 mg/g (DW) (clone 19 in Vaahersalo) (Figure 5).

When individual phenolic compounds such as (+)-catechin, chlorogenic acid, and salidroside were analyzed, the concentrations were clearly higher in samples from parental trees than in samples from clonal plantlets. On the contrary, DHPPG was only found in low concentration in samples from one of the parental trees (tree 17), whereas 0.2–2.1 mg/g (DW) was found in all clones except clone no. 3 (Figure 3).

Discussion

Between parental trees, a high variation in birch shoot secondary chemistry was found. Variation within an individual tree and between branches at the same level of tree canopy was low for most of the studied compounds and compound groups (Table 1). This indicates a high population level variation in the secondary chemistry of the European white birch, a result that is in accordance with our previous results where leaf chemistry from these same parental trees was investigated (Laitinen et al., 2002). UPGMA clustering of shoot chemistry (Figure 1) gave different chemotype groups than grouping on the basis of leaf chemistry (see Laitinen et al., 2000). This is apparently due to qualitative differences between leaf and shoot chemistry’s.

The main phenolic compounds were the same in both mature trees and young clonal plantlets of the same trees; that is, the chemical profiles were similar. This indicates genetic control of these secondary compounds in European white birch. Generally, the concentrations of defensive compounds in deciduous trees are thought to be higher during the juvenile stage (Bryant and Julkunen-Tiitto, 1995), which was also the case in our present study—especially for triterpenoids. The amounts of total terpenoids for some clones were almost 100 times greater in clonal plantlets than in parental trees (Figure 5). This is in accordance with a previous study on Alaska paper birch (Betula resinifera), where Reichardt et al. (1984) found 25 times greater amounts of papyriferic acid in juvenile stems than in mature twigs. Contrary to terpenoids, the concentrations of some phenolic compound groups and individual compounds, e.g., catechin derivatives, caffeoyl quinic acids, condensed tannins, (+)-catechin, and salidroside, were higher in parental trees (Figures 3 and 4). The reason for this difference may be related to the different roles for those compounds. Catechins play a role as important structural elements of condensed tannins (Strack, 1997; Seigler, 1998), whereas triterpenoids are effective defenses against hares particularly in the juvenile developmental stages of white birches (e.g., Reichardt et al., 1984). Ontogenic differences make it difficult to predict the quantity of triterpenoids in clonal plantlets by only using results gained from parental trees, whereas a higher similarity between clonal plantlets and parental trees occurs for many studied phenolic compounds.

Previous studies have documented significant differences in the concentrations of foliar phenolic compounds, e.g., among clones of Salix sp. (Nichols-Orians et al., 1993; Hakulinen et al., 1995; Hakulinen, 1998) and B. pendula (Keinänen et al., 1999a). In this study, a statistically significant variation in shoot chemistry was also found among clones (Figures 2–5), which agrees with our previous study that used greenhouse-grown B. pendula clonal plantlets (Laitinen et al., 2004). Apparently, the higher environmental variation at the four growing sites in the present study increased the chemical variation within clones and may partly explain the slight differences in the pattern of chemical variation between these studies. Regardless of the differences, the results in these two studies are similar in cases where clonal repeatability (i.e., heritability in a broad sense) was found to be high. This indicates that genetic control of the accumulation of many birch shoot secondary compounds, such as chlorogenic acid, quercetin-3-glucoside + quercetin-3-glucuronide, and DHPPG, is strong (Table 1).

In clonal repeatability, the additive and nonadditive component of genetic variation cannot be separated (Falconer, 1989). However, it is used especially when the possible gain obtainable by vegetative production of trees is considered (van Buijtenen, 1992) and also because of the huge plant material needed for reliable estimates of narrow sense heritability (see King et al., 1997). Differences in heritability among different compounds may be due to both the origin and/or different bioactivities of the compounds (Orians et al., 1996). Intense selection for adaptive traits may lead to low heritability values. Thus, high heritabilities for compounds with high bioactivity may suggest a short period of selection time (Falconer, 1989; Orians et al., 1996).

The especially low heritability (i.e., clonal repeatability) in our study for (+)-catechin and catechin derivatives may be due to their role as important precursors for condensed tannin synthesis (e.g., Strack, 1997). Condensed tannins occur in ferns, gymnosperms, and angiospermous plants (e.g., Seigler, 1998), which indicates that they are of rather ancient origin. Some condensed tannins are also known to have a high degree of bioactivity (e.g., Waterman, 1988; Hagerman et al., 1998; Kraus et al., 2003); thus, they can be assumed to have been under selective pressures for a long time. Therefore, the low heritability found in this study for condensed tannins might be due to a strong, long-lasting selection that decreased heritability during generations of time. Some other phenolic compounds or compound groups had high heritabilities that could indicate either low selection pressures or their more recent origin.

Many triterpenoids are characteristic especially for rapidly evolving (e.g., Dugle, 1966) birch species (e.g., Reichardt, 1981; Vainiotalo et al., 1991; Taipale et al., 1993; Julkunen-Tiitto et al., 1996). Triterpenoids are defensive compounds against mammalian herbivory (Reichardt et al., 1984; Risenhoover et al., 1985; Rousi et al., 1991; Laitinen et al., 2004). In spite of their ecological importance (i.e., high bioactivity), the heritabilities of triterpenoids were high. Therefore, we postulate that these compounds have a rather recent evolutionary origin. To further study the possibilities for the selection of genotypes, the additive genetic variation for these traits should be studied.

The profound effect of environment on plant secondary chemistry was seen in our study as statistically significant differences between growing sites for the accumulation of 45% of the studied phenolic compound groups and individual compounds, e.g., catechin derivatives, condensed tannins, (+)-catechin, chlorogenic acid, rhododendrin, and salidroside (Figures 2 and 3). The site with the highest or lowest concentrations differed, depending on the compound studied, indicating that the immediate environment affected different compounds or compound groups differently. For rhododendrin, salidroside, and terpenoids, both the effect of growing site and the genotype by growing site interaction were statistically significant (Figures 3 and 5), suggesting that different clones may respond differently to environmental variation; that is, genotypes are specialized (Via, 1984) in their production of such compounds in different environments (e.g., Figure 6b). These results agree with previous studies that have shown thatin B. pendula leaves, there are significant genotype × environment (different fertilization, defoliation, and ozone treatments) interactions for some of the phenolic compounds (Keinänen et al., 1999a; Yamaji et al., 2003). Also, the secondary chemistry of some willow clones is more sensitive to environmental factors than other willow clones (Hakulinen et al., 1995; Veteli et al., 2002). It seems that there is a clear genetic component in the determination of secondary chemistry profiles in European white birches, and the sensitivity of chemical accumulation to environmental factors is highly dependent on the studied compound and tree genotype.