Responses of Pinus sylvestris branches to simulated herbivory are modified by tree sink/source dynamics and by external resources

Abstract

1. It is claimed that the quality of foliage following defoliation depends on carbon/nutrient balance of the tree. To study the importance of sink/source regulation for the quality of foliage, as well as for its quantity, Pinus sylvestris trees were defoliated and fertilized both in southern Finland and at the tree line in northern Finland.

2. The pattern of defoliation in a shoot was more decisive for quantitative changes in new foliage than its extent: removal of similar amounts of foliage from different branch parts led to different outcomes.

3. Defoliation of 50% spread evenly within a shoot, or applied to the basal part of a shoot only, did not alter production of new foliage, whereas defoliation applied to the apical part of a shoot decreased the mass and length of needles in the new shoot. Defoliation of apically located 1-year-old needles of the branch leader shoot, but not of 2-year-old ones, significantly reduced the mass and length of needles in new shoots.

4. These results are consistent with the explanation that damage alters the ability of shoots and branches to form strong meristematic sinks and that sink strength determines the ability of these meristems to draw resources from the common pool of the tree.

5. Defoliation of the main photosynthate source lowered concentrations of the fructose and glucose, indicating shortage of carbon. However, whole-tree defoliation did not affect the concentrations of individual foliar sugars.

6. Traits describing pine shoot growth correlated negatively with foliar phenolic concentrations but not with concentrations of other secondary compounds. Concentrations of foliage phenolics consistently increased after defoliation, while terpenoids, putatively the main class of defensive compounds in Scots Pine, did not respond to defoliation. Defoliation of a branch or a whole tree had only slight effects on the concentrations of fibre, mono- and sesquiterpenes, resin acids or nitrogen.

7. Likewise, fertilization significantly increased the concentration of some sesquiterpenes only in pine foliage. Whole-branch defoliation and fertilization together had no effect on the concentration of fibre or nitrogen in pine foliage.

8. Altogether, the amount of foliar biomass removed, nutrients or carbon did not explain in any consistent way the qualitative changes in the pine foliage. Instead, results were consistent with simple physiological dependence of foliar traits on sink strength. Production of terpenoids reflected increased sink strength, but the production of phenolics was negatively correlated with sink strength.

9. The difference between shoot growth characteristics and foliage concentrations of phenolics and, on the other hand, terpenoids, may have a biosynthetic, instead of an ecological or evolutionary explanation. Protein synthesis, and thereby possibilities for growth, competes with phenolic synthesis (via phenylalanine) but not with terpenoid synthesis.

10. These results indicate that in Scots Pine the predicted trade-off between growth and production of pooled carbon-rich secondary compounds was found in phenolics but presumably for reasons external to the carbon/nutrient balance and growth/differentiation balance hypotheses. Instead, terpenoids did not behave as predicted by these theories.

Introduction

The nature and significance of herbivore-induced, potentially defensive responses in plants have been controversial for more than a decade (e.g. Haukioja 1980, 1990; Fowler & Lawton 1985; Tallamy & Raupp 1991; Karban 1993). The earliest models explaining herbivore-induced responses assumed at least implicitly that plant quality for herbivores had been fine tuned by herbivory and emphasized that plant chemical defence varied with exposure to herbivores (e.g. McKey 1979; Rhoades 1979, 1985; Haukioja 1980). Subsequent hypotheses have particularly emphasized the role of the plant's resources in modifying defensive responses: e.g. nutrient vs carbon availability for the plant, amounts of carbon vs nutrients lost in defoliation, as well as the inherent growth rate of plants (Bryant, Chapin & Klein 1983; Coley, Bryant & Chapin 1985; Herms & Mattson 1992). According to the growth/differentiation balance hypothesis (Herms & Mattson 1992), plant growth rates may be linked to their chemical defence owing to physiological constraints between cell growth and differentiation. According to the carbon/nutrient balance hypothesis (Bryant et al. 1983), on the other hand, both the type and amount of chemical defence will vary with the environmental availability of nutrients and on the basis of the carbon/nutrient balance in the plant tissue. The cornerstone for both hypotheses is the strong negative correlation between concentrations of plant nitrogen and phenolics, the trade-off having a physiological basis in phenylalanine, a product of the shikimate acid pathway, which can be used either for protein or for phenolic synthesis (Margna 1977; Margna, Vainjärv & Laanast 1989).

Plants, as modular organisms, do not necessarily respond to environmental challenges such as herbivory as whole individuals. Instead, the behaviour of the individual plant results from responses of individual, competing modules. Furthermore, different modules within the same individual plant may carry foliage of different ages. The age of foliage often determines its value for herbivores. Accordingly, the value of foliage for herbivores may vary even within an individual plant as much as, or more than, the value of foliage of different plant species. Such within-plant variability, and the consequences of responses of individual modules, have received much less attention in plant-herbivore studies than the responses of whole plants (but see e.g. Whitham & Slobodchikoff 1981; Suomela, Kaitaniemi & Nilson 1995). All this indicates that the elements of whole-plant physiology also have to be explicitly incorporated into ecological and evolutionary explanations of herbivore-induced plant responses.

Within-plant control among meristems is exerted via differences in their sink strengths (Clifford 1992) and operates by means of a complex interplay involving at least hormonal cues (Hillman 1984; Cline 1991). Effective control at the whole-plant level assumes that meristems in the most favourable positions (in relation to abiotic factors but also in relation to other meristems) for genet fitness dominate over meristems located in inferior positions (Haukioja 1991; Sachs, Novoplansky & Cohen 1993). Interactions among modules form the basis for the sink/source hypothesis explaining herbivore-induced responses in plants (Haukioja 1990; Honkanen & Haukioja 1994; Honkanen, Haukioja & Suomela 1994; Haukioja & Honkanen 1996). According to this hypothesis, one primary way in which the plant is affected by damage is via damage-induced changes in the abilities of meristems to compete for resources (Haukioja et al. 1990; Honkanen & Haukioja 1994; Ruohomäki et al. 1997). Damage by herbivores may modify sink strengths either by directly disturbing them or via disturbances to source leaves during sink formation (Honkanen et al. 1994). Recognizing that the lost biomass may have temporarily and spatially variable physiological roles, sink vs source, this scenario allows for several alternative outcomes even after damage which removes the same amount of biomass and nutrients. For instance, if the leaves on a branch, at a given time, feed developing local meristems, damage to these leaves will presumably lead to weak new meristems. The insufficient sink strength of weak meristems seems to be an important reason why a branch which has been defoliated at a time when the meristems for the next-year leaves are still developing, in the following growth season will compete poorly with other branches and will not receive its normal share from the total resource pool of the tree. Among-module alterations alone may lead to herbivore-induced responses (Haukioja & Honkanen 1996; Honkanen & Haukioja 1998). This emphasizes the need to understand whole-plant sink/source relationships at the time when the damage occurs and the importance of the timing of the damage in relation to seasonal shifts in resource allocation (Marquis 1992; Haukioja & Honkanen 1996).

There were two aims in this study. First, to test the predictions of the sink/source hypothesis, sink/source relationships were altered by debudding and defoliation experiments were conducted on whole trees and on individual branches and shoots of Scots Pine Pinus sylvestris. Second, to test the role of sink/source regulation on the quality of Scots Pine foliage for herbivores after fertilization and defoliations, the concentrations of the main putative defensive compounds, phenolics, resin acids and terpenoids, which in Scots Pine are crucial defences, were determined (Niemelä, Mannila. & Mäntsälä 1982; Haukioja et al. 1983; Danell, Gref & Yazadani 1990; Björkman, Larsson & Gref 1991; Sjöberg & Linden 1991; Saikkonen, Neuvonen & Kainulainen 1995). In addition, the concentrations of structural carbohydrates (fibre and lignin) which contribute to toughness of foliage as well as some sugars and nitrogen, i.e. important metabolites for insect growth, were determined.

Materials and methods

Scots Pine is an evergreen conifer with a determinate growth habit and a single annual flush of growth. The growth of Scots Pine normally involves the addition of one internode and one whorl of shoots per year to the main axis and to each branch axis. A branch is an extension arising directly from a lateral bud on the tree main axis, and a shoot as a branch part formed in a certain year; in the rest of this paper, C indicates the current-year shoot and C-1 a shoot which had elongated in the previous year. Canopy growth in Scots Pine takes place in June–July by elongation of shoots (‘long shoots’, from buds formed the previous season). Normally the terminal bud produces the largest shoot in the whorl, i.e. the terminal shoot dominates its laterals.

EXPERIMENTAL TREATMENTS

Branch debudding, defoliation and fertilization

To find out how whole branches respond to defoliation, a single 3-year-old branch in 144 trees (2–3 m tall, 10–20-years old) at Kullaa, SW Finland (62° N, 22° E), was randomly assigned to six treatments. The experimental area was a cultivated Scots Pine stand representing a single successional stage and local seed origin.

The treatments were conducted in May 1988 and consisted of (1) unmanipulated controls, (2) 50% defoliation (every other needle fascicle totally removed) over the whole branch, (3) 100% defoliation (all needles totally removed) over the whole branch, (4) debudding (removal of the terminal bud from the branch leader shoot), (5) treatments 2 and 4 combined, and (6) 3 and 4 combined (Fig. 1). Half the trees in each of the six treatment categories were fertilized with commercial NP fertilizers, in line with forestry recommendations (120 g tree^–1, equivalent to 27·6 g N and 0·84 g P).

Figure 1. . Coding of the treatments in branch debudding, defoliation and fertilization experiment.

Needle-class and within-shoot defoliations

These two experiments were carried out at Seili, at the Archipelago Research Institute of the University of Turku (60° N, 21° E). The experimental area was a predominantly coniferous forest, with Scots Pines at several succesional stages. All the experimental trees were 1–2 m tall, 10–20-years old and had been established by natural regeneration from local seed.

The purpose of the needle-class experiment was to determine whether the defoliation of young and old needles, which play different roles in resource storage (Ericsson 1979), had a similar effect on the growth of new needles. Thirty pines were randomly assigned to three treatments, conducted in May 1988: (1) unmanipulated controls; (2) removal of the C-2 needle year-class; (c) removal of the C-1 needle year-class of the 5-year-old branch leader shoot (Fig. 2).

Figure 2. . Coding of the treatments in needle-class defoliation experiment.

In the within-shoot defoliation experiment, in May 1988, the effects of equally large but differently distributed needle losses within a C-1 shoot on the growth of new needles were studied. Forty pines were randomly assigned to four treatments: (1) unmanipulated controls; (2) total removal of needles from the apical half of the C-1 shoot; (3) total removal of needles from the basal half of the C-1 shoot; (4) removal of half the needles evenly all over the C-1 shoot of the 5-year-old branch leader shoot (Fig. 3).

Figure 3. . Coding of treatments in within-shoot defoliation experiment.

Sink/source defoliation and whole-tree defoliation and fertilization experiment

To study effects of defoliation under extreme environmental conditions, experiments were also conducted at the Kevo Subarctic Research Institute of the University of Turku, in northernmost Finland (69° N, 27° E), close to the tree-line. The experimental area was a mixed Scots Pine and Mountain Birch (Betula pubescens ssp. czerepanovii) forest, with Scots Pines representing several successional stages. All experimental trees were 1–2 m tall, 20–40-years old, were established by natural regeneration and again represented local seed origin.

The purpose of the sink/source defoliation experiment was to find out whether defoliation of sink and/or source foliage had a similar effect on the growth and chemical composition of the foliage. Thirty-two pines were randomly assigned to four treatments, carried out in early July 1993: (1) unmanipulated controls; (2) cutting of the distal part of every other sink needle pair in the C shoot; (3) removal of the C-1 and C-2 (source) needle year-class of the branch leader shoot; (4) treatments 2 and 3 combined (Fig. 4). Only the leader shoot of one 5-year-old branch per tree was treated.

Figure 4. . Coding of the treatments in sink/ source defoliation experiment.

The purpose of the whole-tree defoliation and fertilization experiment was to study the interaction between defoliation and the resource status of trees in determining the growth and chemical composition of Scots Pine. In early June 1993, 32 pine trees (about 1·5 m high) were randomized among four treatments: (1) unmanipulated controls; (2) defoliation of all needle classes in the whole tree; (3) fertilization; (4) treatements 2 and 3 combined. Commercial NP fertilizers were used and to achieve realistic nutrient levels 140 g of fertilizer per tree was applied. This is equivalent to 35 g N and 0·14 g P.

GROWTH MEASUREMENTS

In all the experiments, except in the whole-tree and fertilization experiment, needle samples from the treated and corresponding control branches were collected at the end of the first growth season after treatments. In the branch debudding, defoliation and fertilization experiments, and in the needle-class and within-shoot defoliation experiments, six needle fascicles were sampled from both the leader and the lateral shoots of the uppermost whorl in the branches. In the sink/source defoliation six needle fascicles from the leader shoot of the branch were collected. In the sink/source defoliation experiment, needle samples were also taken from the C-1 needle-class (treatments 1 and 2). In the whole-tree defoliation and fertilization experiment, needle samples from both the C and C-1 year-classes from the leader shoot of the branch were collected at the end of the second growth season after treatments. The needles were dried (+ 80 °C, 24 h) and weighed to the nearest 0·01 g. Needle length was measured to the nearest 0·1 cm.

In the branch debudding, defoliation and fertilization experiment, branches of trees were excised and brought to the laboratory for measurement at the end of the second growth season after treatments. The branches were cut into segments corresponding to yearly growth; the segments contained C and C-1 year-class shoots. All needles were dried and weighed to the nearest 0·01 g; shoot length was measured to the nearest 0·1 cm.

In the whole-tree defoliation and fertilization experiment, the trunk base diameter (mm) and the length (cm) and diameter (mm) of the C and C-1 year-classes were measured.

CHEMICAL ANALYSES

Fibre and nitrogen analyses

Foliage samples from trees in the four treatments in the branch debudding, defoliation and fertilization experiment were analysed for nitrogen and fibre: control, 100% defoliation, fertilization and 100% defoliation + fertilization. The samples were dried (+ 80 °C, 48 h) and milled to pass a 0·5 mm sieve. Nitrogen concentrations were analysed by a CHN-analyser (Leco) and fibre as described by Van Soest, Robertson & Lewis (1991).

Sugars

In the sink/source defoliation and whole-tree defoliation and fertilization experiments, foliage was analysed for sugars (fructose, glucose and sucrose, mg g^–1) at the end of the first and the second growth season after treatments, respectively.

Ethanol (4 ml) containing 1 mg ml^–1 of an internal reference compound (phenyl β-D-glucoside), was added to 300 mg of cut needles in a Teflon-lined tube; samples were extracted in a ultrasound bath (30 min) and left overnight in ethanol (Marcy & Carroll 1982). The samples were shaken and centrifuged (5000r.p.m.). Clear extract (500:l) was evaporated and 400:l of silytation reagent (21% trimethylsilylimidatzol in pyridine) was added to the sample. The samples were analysed with GC/MS, using split injection with a HP-5 capillary column. Helium was used as the carrier gas. The temperature programme was 110 °C for 0·5 min followed by 20 °C min^–1 to 300 °C for 20 min.

Mono- and sesquiterpenes and resin acids

In the sink/source defoliation and whole-tree defoliation and fertilization experiments the foliage (C and C-1 needle year-classes and C needle year-class at the end of the experiments, respectively) was analysed for mono- and sesquiterpenes and resin acids at the end of the first and the second growth season after the treatment, respectively.

Needle samples were frozen in liquid nitrogen in the field. In the laboratory, a bundle of needles was sliced into 1 mm pieces. For the analyses, 300 mg of sliced needles was extracted in Teflon-lined screw-cap culture tubes (16 mm × 100 mm) with 3 ml hexane. The samples were mixed, allowed to bond, agitated in an ultrasound bath (30 min) and left in the tubes overnight. The samples were then mixed thoroughly and centrifuged (5000 r.p.m.). The samples were analysed with gas chromatography (HP 5890) and mass spectrometry (HP 5988) using an NB-315 capillary column (25 m × 0·2 mm × 0·2:m, HNU-Nordion). Helium was used as the carrier gas. The temperature programme consisted of 60°C (0·5 min, splitless injection), followed by 5 °C min^–1 to 230 °C and finally 10 min at 230°C. Identification of mono- and sesquiterpenes was based on mass-spectral and retention data (Pohjola 1993). Monoterpenes were quantified based on the response factor of α-pinene. Sesquiterpenes were quantified using an internal reference compound (1-chloro-dodecane). The response factor was β-caryophyllene.

For the resin acid analysis, 300 mg of sliced needles and 1 mg of heptadecanoic acid were extracted twice for 2 h in an ultrasonic bath with 2 ml ethanol. The ethanol phases were combined and centrifuged. Clear extract (2 ml) was used in the GC–MS analysis. Before the analysis the samples were evaporated to dryness and methylated with freshly prepared diazomethane in diethylether-methanol (9:1). The samples were analysed using an HP 5 capillary column (25 m × 0·25 mm × 0·3:m, Hewlett-Packard). Helium was used as the carrier gas and the temperature programme consisted of 11°C (0 min) followed by 10 °C min^–1 to 300 °C (5 min) . The injector and transfer-line temperatures were 260 °C and 300°C, respectively. Resin acids were identified by their mass-spectral data and retention data according to Gref & Tenow (1987) and Morales, Birkholz & Hrudey (1992). Quantification of resin acids was conducted by the response factor of abietic acid.

Total phenolics

In the sink/source defoliation experiment, total phenolics were analysed from damaged and undamaged current-year needles as well as from C-1 needles. In the whole-tree defoliation and fertilization experiment, total phenolics were analysed from C needle year-class and from C-1 needle year-class. These needles were produced one and 2 years after the defoliation, respectively.

Oven-dried samples were ground to a fine powder. About 50mg of powdered needle material was extracted with 80% methanol and the residue washed three times with methanol. Total phenolics were analysed using the Folin–Ciocalteu procedure (see e.g. Abrahamson et al. 1991).

STATISTICAL ANALYSES

The effects of the treatments on shoot and needle traits were analysed by MANOVA and by two- or three-way factorial ANOVA. In the MANOVA tests we used Wilk's lambda. All nitrogen, NDF, ADF, lignin, terpenoid, resin acid and phenolics data were log-transformed before the ANOVA analyses.

Results

CONSEQUENCES OF BRANCH DEFOLIATION, DEBUDDING AND FERTILIZATION

Debudding significantly decreased the total needle biomass and the length of lateral shoots in the first growth season after the treatment (C-1; Fig. 5, Table 1) but did not affect the mass or length of individual needles in the lateral shoots (Fig. 5, Table 2). In the second growth season after the treatments, the total needle biomass of the newest shoots was still significantly lower in the debudding treatment than in the control treatment but the pooled length of shoots was not affected (Fig. 5, Table 1).

Table 1.  .ANOVA table of treatment effects on total needle biomass and total length of C and C-1 shoots in branch debudding (Deb), defoliation (Def) and fertilization (Fer) experiment at the end of the second growth season after treatments. Adjusted values of the dependent variables were used in the tests to account for different growth rates of trees. The pre-treatment annual growth of the main axis of the branch was used as a covariate

Figure 5. . Mean values and SE of (a) needle mass and (b) needle length at the end of the first growth season after treatments, and mean values and SE of (c) needle biomass and (d) shoot length in control and debudding groups at the end of the second growth season after treatments in branch debudding and fertilization experiment.

Table 2.  .ANOVA table of treatment effects on mass and length of needles in terminal and lateral shoots in branch debudding (Deb), defoliation (Def) and fertilization (Fer) experiment at the end of the first growth season after treatments

Total defoliation of a branch significantly decreased needle length and mass of the current-year foliage in terminal and lateral shoots after the first growth season, while 50% defoliation did not (Fig. 6, Table 2). Total defoliation of the branch had significantly decreased the total needle biomass but not the total length of the C-1 shoots; 50% defoliation had no effect (Fig. 6, Table 1). Defoliation significantly affected the total needle biomass of the C shoots. Fifty per cent defoliation of a branch, unlike total defoliation, significantly increased the length of C shoots (Fig. 6, Table 1), indicating local overcompensation after partial defoliation.

Figure 6. . Mean values and SE of (a) needle mass and (b) needle length at the end of the first growth season after treatments, and mean values and SE of (c) needle biomass and (d) shoot length in control and defoliated groups at the end of the second growth season after the treatments in branch debudding, defoliation and fertilization experiment; values with different letters are significantly different (P<0.05) (Tukey-test).

Fertilization significantly increased the mass and length of current-year needles in terminal and lateral shoots measured at the end of the first growth season after fertilization (Fig. 7, Table 2), demonstrating that the relatively young trees at Kullaa responded to fertilization. Fertilization did not significantly affect the total needle biomass or length of shoots in the C-1 or C year-classes (Fig. 7, Table 1).

Figure 7. . Mean values and SE of (a) needle mass and (b) needle length at the end of the first growth season after treatments, and mean values and SE of (c) needle biomass and (d) shoot length in control and fertilized groups at the end of the second growth season after the treatments in branch debudding, defoliation and fertilization experiment.

CHEMICAL CHANGES AFTER DEFOLIATION AND FERTILIZATION

The results of MANOVA showed that defoliation and fertilization did not modify patterns of resource allocation among NDF, ADF or lignin in the C-1 or C shoots (C-1 foliage: 100% Defoliation F = 0·4037 NS, Fertilization F = 0·9464 NS, 100% Defoliation × Fertilization F = 0·1565 NS, df = 3,18; C foliage: 100% Defoliation F = 1·2851 NS, Fertilization F=2·8541 NS, 100% Defoliation × Fertilization F=0·9840 NS, df = 3,18) (Fig. 8).

Figure 8. . Mean values and SE of ADF, NDF and lignin in C-1 (a and c) and C (b and d) year-class shoots: at the end of the second growth season after the treatments in branch debudding, defoliation and fertilzation experiment.

Total branch defoliation did not significantly affect the concentration of nitrogen in the C-1 or C year-class Scots Pine foliage (Table 3); this was in accordance with previous findings (Honkanen & Haukioja 1994). Although the nitrogen content of pine needles was thus fairly invariable, the pool of foliage nitrogen naturally decreased because of the lower total mass of foliage.

Table 3.  . Mean values (SE in parentheses) and effects of defoliation and fertilization on the nitrogen concentration (mg g^–1) of the foliage of the C-1 and C year-class shoots in branch defoliation (Def), debudding (Deb) and fertilization (Fer) experiment after the second post-treatment season

Fertilization significantly increased the nitrogen concentration of foliage in the C-1 year-class shoots but not in the C year-class shoots, when measured after the second post-treatment growth season (Table 3). In C year-class shoots, the interaction between defoliation and fertilization was significant because foliage nitrogen concentrations increased in trees which were both fertilized and defoliated (Table 3).

CONSEQUENCES OF NEEDLE-CLASS AND WITHIN-SHOOT DEFOLIATIONS

In the needle-class defoliation experiment, the removal of C-1 year-class needles significantly decreased the mass and length of needles in the C year-class in the terminal and lateral shoots, while the removal of C-2 year-class needles had no effect on current needle growth (needle mass: terminal shoot F = 4·90*, lateral shoots F = 7·46**; needle length: terminal shoot F = 6·76**, lateral shoots F = 2·53 NS; df = 2,27 in all cases) (Fig. 9).

Figure 9. . Mean values and SE of (a) needle mass and (b) needle length in terminal and lateral shoots at the end of the first growth season after treatment in the needle class defoliation experiment. Means with different letters are significantly different (P<0.05) (Tukey-test).

In the within-shoot defoliation experiment, removal of the same amount of needles from different positions in C-1 year-class shoots led to different outcomes. Defoliation of the apical part of a shoot decreased the mass and length of needles in terminal, but not in lateral, current-year shoots, whereas removal of basal needles or evenly distributed defoliation had no effect on the mass and length of needles in current-year terminal or lateral shoots (needle mass: terminal shoot F=3·57*, lateral shoots F=2·52 NS; needle length: terminal shoot F=2·94*, lateral shoots F=2·53 NS; df = 3,34 for terminal shoot and df = 3,36 for lateral shoots (Fig. 10).

Figure 10. . Mean values and SE of (a) needle mass and (b) needle length in terminal and lateral shoots at the end of the first growth season after the treatment within-shoot defoliation experiment. Means with different letters are significantly different (P<0.05) (Tukey-test).

CONSEQUENCES OF REMOVING SINK VS SOURCE NEEDLES ON FOLIAGE CHEMISTRY

Defoliation, whether of source needles alone or of both sink and source needles, significantly altered the carbohydrate composition in new foliage (C year-class shoots) (MANOVA, Sink defoliation F=1·3052 NS, Source defoliation F=5·3594**, Sink × Source defoliation F=5·1539**, df = 6,21). The removal of the main photosynthate sources of sink needles was reflected in reduced concentrations of the main photosynthetic products, fructose and glucose, but not sucrose, in C year-class needles (two-way ANOVAs, fructose: Sink defoliation F=0·10 NS, Source defoliation F=10·77**, Sink × Source defoliation F = 0·23 NS; glucose: Sink defoliation F=0·50 NS, Source defoliation F=16·57**, Sink × Source defoliation F = 1·47 NS; sucrose: Sink defoliation F=1·60 NS, Source defoliation F=0·20 NS, Sink × Source defoliation F = 3·27 NS; df = 1,1,1,26 for all compounds) (Table 4).

Table 4.  . Mean values (SE in parentheses) of carbohydrates, mono- and sesquiterpenes, resin acids and phenolics (mg g^–1) in C year-class foliage in sink/source defoliation experiment

Two-way ANOVA, testing the effects of defoliation on damaged sink needles, revealed that the damaging of sink needles significantly increased the total concentration of phenolics in the damaged needles themselves (Sink defoliation F = 31·47 ***, Source defoliation F = 0·42 NS, Sink × Source defoliation F = 2·19 NS; df = 1,1,1,27) (Table 4). However, two-way ANOVA, testing the effects of defoliation on undamaged sink needles, revealed only very local effects and no changes in the concentrations of phenolics in undamaged sink needles either after damage to other sink needles or after defoliation of source needles (Sink defoliation F = 0·02 NS, Source defoliation F = 0·02 NS, Sink × Source defoliation F = 3·98 NS; df = 1,1,1,27) (Table 4). The composition and concentration of different sesquiterpenes and resin acids in the C year-class foliage did not respond to removal of the main photosynthate source (C-1 and C-2 needles at the time of the treatments) or to damage to the sink needles themselves (sesquiterpenes: Sink defoliation F=0·9923 NS, Source defoliation F=0·5251 NS, Sink × Source defoliation F=0·6457 NS, df = 7,21; resin acids: Sink defoliation F=0·7188 NS, Source defoliation F=0·1851 NS, Sink × Source defoliation F=1·3857 NS, df = 3,25) (Table 4). Again, concentrations of these secondary compounds did not directly depend on carbon gain of needles in C shoots, although defoliation of source needles significantly decreased the concentration of a monoterpene, e-β-ocimene, in the C year-class foliage (MANOVA conducted for monoterpenes, Sink defoliation F=1·6400 NS, Source defoliation F=3·7844*, Sink × Source defoliation F=0·9088 NS, df = 5,23; two-way ANOVA conducted for e-β-ocimene, Sink defoliation F = 0·42 NS, Source defoliation F = 19·34**, Sink × Source defoliation F = 0·03 NS, df = 1,1,1,26) (Table 4).

Weakening of the sinks did not affect the source foliage, because defoliation of sink needles did not significantly change allocations to different carbohydrate compounds in the C-1 year-class foliage (MANOVA: df = 3,11, F = 0·4739 NS) (Table 5). Similarly, sink defoliation did not significantly affect the composition of different mono- and sesquiterpenes in the needle class of C-1 year-class shoots [MANOVA: monoterpenes F = 1·5755 NS; sesquiterpenes F = 1·7904 NS (df = 6,8, df = 5,9, respectively)] (Table 5). The same was true for phenolics: allocation to phenolics in C-1 shoots did not change after sink defoliation (F = 2·12 NS, df = 1,10) (Table 8), but the composition of different resin acids altered significantly (MANOVA: df = 7,7, F = 4·5266*) (Table 5). Two-way ANOVAs showed that the concentration of 4-epiimbricatolic resin acid significantly increased in control foliage after sink defoliation (F = 6·49*, df = 1,13).

Table 5.  . Mean values (SE in parentheses) of carbohydrates, mono- and sesquiterpenes, resin acids and phenolics (mg g^–1) in C-1 year-class foliage in sink/source defoliation experiment

Table 8.  . Correlations between growth characteristics, main secondary metabolites and carbohydrates in whole-tree defoliation and fertilization experiment. Number of cases given below the correlation coefficient

CONSEQUENCES OF WHOLE-TREE DEFOLIATION AND FERTILIZATION UNDER EXTREME CONDITIONS

After the second growth season, total defoliation had significantly decreased the length and diameter of C year-class shoots, the diameter of C-1 year-class shoots and the mass of needles in C year-class shoots (Table 6). However, it did not affect the allocation of resources to different carbohydrates (MANOVA, Defoliation F=1·1134 NS, Fertilization F=2·8226 NS, Defoliation×Fertilization F=1·3495 NS, df=9,18) (Table 7). Unlike the vigorously growing young Kullaa trees (branch debudding, defoliation and fertilization experiment), the growth traits of the fertilized older and naturally slowly growing trees at Kevo did not change (Table 6).

Table 6.  . Mean values (SE in parentheses) and effects of defoliation and fertilization on growth characteristics of trees in whole-tree defoliation and fertilization experiment

Table 7.  . Mean values (SE in parentheses) of carbohydrates, mono- and sesquiterpenes, resin acids and phenolics (mg g^–1) in whole-tree defoliation and fertilization experiment

MANOVA revealed no significant changes in the composition of different mono- and sesquiterpenes and resin acids after total defoliation (monoterpenes: Defoliation F=1·7253 NS, Fertilization F=1·1797 NS, Defoliation×Fertilization F=0·3909 NS; sesquiterpenes: Defoliation F=0·7188 NS, Fertilization F=3·5767*, Defoliation×Fertilization F=0·6924 NS; resin acids: Defoliation F=0·4866 NS, Fertilization F=0·5031 NS, Defoliation × Fertilization F=0·9889 NS; df = 9,19, df = 8,20, df = 3,25, respectively) (Table 7). However, although no growth effects of fertilization were observed (Table 6), the foliage concentrations of three individual sesquiterpene compounds (β-caryophyllene, α-humulene and δ-cadinene) increased significantly after fertilization (two-way ANOVAs: β-caryophyllene, Defoliation F=0·82 NS, Fertilization F=11·87**, Defoliation×Fertilization F=2·05 NS; α-humulene, Defoliation F=0·00 NS, Fertilization F=12·38**, Defoliation×Fertilization F=2·40 NS; δ-cadinene, Defoliation F=0·03 NS, Fertilization F=6·10*, Defoliation×Fertilization F=0·24 NS; df=1,1,1,26 for all compounds) (Table 7).

After whole-tree defoliation, but not after fertilization, the concentration of total phenolics in C and C-1 year-class needles increased significantly (phenolics in C-1 year-class: Defoliation F = 7·40*, Fertilization F = 0·12 NS, Defoliation × Fertilization F = 0·84 NS; phenolics in C year-class: Defoliation F = 7·24*, Fertilization F = 0·48 NS, Defoliation × Fertilization F = 0·76 NS) (Table 7).

CORRELATIONS BETWEEN GROWTH AND SECONDARY METABOLITES

The total concentration of resin acids in trees in the whole-tree defoliation and fertilization experiment correlated positively with the diameter of tree trunk (Table 8). Because tree-trunk diameter and the relative longitudinal growth rate of the trees at Kevo showed a significant positive correlation (r = 0·670**), trees with a high growth rate allocated relatively more carbon to resin acids than more slowly growing trees. This is consistent with findings by Björkman et al. (1991). The correlations between growth characteristics and mono- and sesquiterpenes were not significant (Table 8).

Contrary to resin acids, the relationship between tree growth and concentrations of phenolic was as predicted, e.g. by the growth/differentiation balance hypothesis. The total concentrations of phenolics in C-1 foliage showed a significantly negative correlation with the length of current shoots (Table 8). Total phenolics in current-year foliage showed significantly negative correlations with the length and diameter of C and C-1 shoots as well as with the mass of needles in current (C) year shoots (Table 8). All these findings indicate that the concentrations of phenolics were inversely related to tree growth characteristics, while those of terpenoids were not. These results are consistent with findings by Muzika & Pregitzer (1992).

Discussion

This study demonstrated considerable variation in pine shoot growth and foliage chemistry after different types of defoliation and debudding. Clearly, growth responses and changes in Scots Pine foliage chemistry after local and tree-wide damage depended more on regulative mechanisms in within-tree physiology than on the amounts of lost resources as such. Furthermore, responses of phenolic and terpenoid concentrations to our treatments were very different.

SHOOT AND NEEDLE GROWTH AFTER SIMULATED HERBIVORY: REGULATING AND MODIFYING FACTORS

Early season defoliation applied to individual branches and shoots of Scots Pine reduced the length of needles and diameter of shoots in these branches but not in other ones, confirming previous findings (Långström et al. 1990; Honkanen & Haukioja 1994; Honkanen et al. 1994). Longitudinal growth of shoots decreased in the second growth season after the treatments, indicating that shoot growth was predetermined, while the growth of needles largely depended on current photosynthates (e.g. Thompson & Puttonen 1992; Lippu 1994).

At least two factors may have contributed to growth reduction after damage. First, losses of photosynthates and/or nutrients in the clipped needles may have been decisive. In our study this is not a plausible explanation because the reduction in the growth of new foliage did not depend on whether whole branches, individual shoots within these branches or subsets within shoots had been defoliated. After all these unequal defoliations, the mass of new needles in terminal shoots decreased by 31–39% and needle length by 23–25%. This finding demonstrates the high priority of growing needles for investments and indicates that growing foliage was able to obtain resources from storage outside the branch. The result also excludes the second possibility, i.e. that growth retardation is a simple consequence of the distance between sink and source, as proposed by Hardwick (1986).

Accordingly, growth reduction after foliar damage was determined neither by the total amount of resources lost in defoliation nor by the distance between sink and source. Instead, the growth retardation in defoliated branches and shoots was consistent with the sink/source hypothesis: weakened transport of resources (photosynthates and/or mineral nutrients) from the common or local resource pools of the tree to the defoliated branch. A simple potential mechanism is that defoliation indirectly decreased the meristematic activity of new shoots. Physiologically this may have occurred via defoliation-reduced auxin production (Sundberg & Little 1987).

The results of this study point to the importance of the sink/source relationship also in recovery from damage. After removal of apical buds, debudded branches were unable to compensate. This contrasts with the outcome of debuddings of the leader shoot of Scots Pine stems (Honkanen et al. 1994). Obviously, in a branch, but not in the stem, the removal of the apical bud of the leader shoot reduced the photosynthetic capacity of the remaining foliage (see also Gezelius et al. 1981), thus decreasing the total productivity of the branch. These results are consistent with the stronger dominance of the leader shoot of the stem, compared to the leader shoots of branches.

The results of the debudding treatment showed that seemingly minor changes in the damage pattern may cause significant differences in the outcomes of experiments. If these different patterns translate to plant fitness (see e.g. Marquis 1992), then the ways in which herbivores may affect plants become highly complex. These results also hint that determining the true value of leaves for a plant is more complex than usually assumed. For instance, leaves which subtend developing meristems and growing leaves must have high value for a plant, because they determine the productivity of new foliage.

QUALITATIVE CHEMICAL CHANGES IN FOLIAGE AFTER TREATMENTS

Total phenolics was the only class among the secondary metabolites whose concentrations consistently changed after defoliations. In agreement with numerous previous studies (see e.g. Sunnerheim-Sjöberg & Hämäläinen 1992 for pine and Herms & Mattson 1992 for other species), shoot growth and content of foliage phenolics correlated negatively. However, we are uncertain as to the extent to which the high concentration of phenolics in slowly growing pines and pine parts can be equated with high defence. In this study, just as in birch (Tuomi et al. 1988), the post-defoliation increase in phenolics was a strictly local phenomenon. We cannot even exclude the possibility that the negative correlation between phenolic compounds and growth has a non-defensive causal component, because some phenolic compounds have been shown to act as growth regulators (Jacobs & Rubery 1988; Kuiters 1989).

The behaviour of phenolics could not in all cases be directly predicted from the carbon gain of needles, nor did the behaviour of phenolics represent the behaviour of all carbon-rich secondary metabolites in pine foliage (for similar results with conifer seedlings, see Muzika & Pregitzer 1992; Holopainen et al. 1995; Kainulainen et al. 1996). Regardless of fertilization and severe defoliations, terpenoids in Scots Pine foliage retained their percentual levels and thus were insensitive to changes in the availability of resources. Scots Pine has been assumed to be an unsuitable tree for testing the carbon/nutrient balance hypothesis (Edenius 1993). In our opinion it is not, instead it offers a really risky test for the general validity of the central claim of the carbon/nutrient balance and growth/differentiation hypotheses, i.e. of the trade-off between tree growth and its investments into carbon-based secondary compounds. This is an ecological/evolutionary prediction but testing it with phenolics may give confirming results simply because biosyntehsis of phenolics takes place via phenylalanine. Phenylalanine is produced in the branching point from shikimate pathway either to protein synthesis or to synthesis of phenylpropanoids (see Muzika & Pregitzer 1992). Terpenoids, in turn, are products of a different pathway, the mevalonic acid pathway (Waterman & Mole 1989; Gershenzon 1994), and therefore a trade-off between terpenoids and growth, but not of phenolics and growth, would be a real proof an ecological/evolutionary, instead of biochemical, causation behind the assumed trade-off between tree growth and defensive allocations (Haukioja et al. 1998; Koricheva et al. 1998). Altogether, these results suggest that, contrary to the carbon/nutrient balance and growth/differentiation balance hypotheses, growth and defence may not be in all cases fundamentally alternative options for trees.

Acknowledgements

We thank S. Larsson for comments on an earlier version of the manuscript. We are grateful to the staff of the Satakunta Enviromental Research Center and the Kevo Subarctic Research Station for help with this research project. The work was supported financially by the Maj and Tor Nessling foundation, the Lapland Forest Damage Project and the Academy of Finland.