Testing the growth limitation hypothesis for subarctic Scots pine

Journal of Ecology 2010, 98, 1186–1195

doi: 10.1111/j.1365-2745.2010.01684.x

Testing the growth limitation hypothesis for subarctic Scots pine Sanna Susiluoto1*, Emmi Hilasvuori2 and Frank Berninger1,3 1

Department of Forest Ecology, University of Helsinki, PO BOX 27, FI-00014 Helsinki, Finland; 2Dating Laboratory, University of Helsinki, PO BOX 64, FI-00014 Helsinki, Finland; and 3De´partement des sciences biologiques, Universite´ du Que´bec a` Montreál, CP 8888 Succ Centre Ville, Montreál QC H3C 3P8, Canada

Summary 1. We tested if tree line trees are sink or resource limited by comparing the reactions in Scots pine trees (Pinus sylvestris L.) to changes in resource availability (through nutrient addition and defoliation), and in sink strength (through debudding) in north-eastern Finland. 2. Height extension growth increased by over 200% and shoot extension growth by 27%, as a result of nutrient addition, whereas changes in radial increment were smaller and not statistically significant. 3. Although in defoliated trees height and shoot extension were not significantly affected, defoliated trees decreased their radial growth by 50%. 4. Branch extension growth of the remaining buds was increased due to debudding treatment by 47%, but there were no changes in radial increment. 5. Non-structural carbohydrate concentrations in twigs, however, remained relatively constant in all treatments. 6. There were no statistically significant changes in carbon isotope ratio due to any of the treatments, indicating that the manipulation treatments did not modify tree water relations. 7. Synthesis. Our results do not support the growth limitation theory for tree line trees and suggest that resource limitations might be more important. Key-words: carbohydrate, carbon isotope, debudding, defoliation, growth, sink–source relationships, tree line Introduction There is an ongoing debate on the factors and processes limiting the growth of trees in the vicinity of the tree line. A strong candidate for a general explanation of the location of the tree line is the sink limitation hypothesis, also known as growth limitation hypothesis (e.g. Ko¨rner 1999). According to this hypothesis, the actual growth processes (cell division and expansion) of tree line trees are restricted by low temperatures (e.g. Ko¨rner & Hoch 2006). This implies that tree line trees are able to acquire photoassimilates more efficiently than they can be used for growth. Consequently, carbohydrates accumulate in tissues and therefore trees have more available carbon than sinks are able to consume. The winter time respiratory costs in tree line trees usually do not exceed c. 10–15% of the carbon acquired during the growing season (Wieser 1997; Ko¨rner 1999), and the requirement for carbohydrate storage can be assumed to remain fairly stable in the long run. The main assumption of the growth limitation hypothesis is that the demand of carbon for the formation of new tissues is smaller *Correspondence author. E-mail: sannamaija.susiluoto@helsinki.fi

than the tree’s ability to acquire carbon. It is well documented that cell growth and elongation slow down – and finally cease – at low temperatures (Ha¨sler 1982; Pietarinen et al. 1982). Evidence for the growth limitation hypothesis has focused on the carbohydrate concentrations of trees. Recent studies have shown that the carbohydrate reserves of tree line trees are not severely depleted at any time of the year (Hoch, Popp & Ko¨rner 2002; Li, Hoch & Ko¨rner 2002; Hoch & Ko¨rner 2003; Handa, Ko¨rner & Ha¨ttenschwiler 2005; Shi, Ko¨rner & Hoch 2006). Furthermore, carbohydrate concentrations are usually higher in trees growing at the tree line than in trees growing at lower altitudes (Hoch & Ko¨rner 2003; Shi, Ko¨rner & Hoch 2006). However, high carbohydrate concentrations have also been interpreted as a physiological adaptation to the respiration costs during the long winters and ⁄ or as an adaptation to the high probability of physical damage to tree tissues due to snow accumulation and high winds (Tranquillini 1979; Sveinbjo¨rnsson 2000). The competing hypothesis, called the resource limitation hypothesis, claims that the uptake of carbon or nutrients is limiting the growth of tree line trees, meaning that trees are not able to acquire as much carbon or nutrients as they would be

2010 The Authors. Journal compilation 2010 British Ecological Society

Growth limitation hypothesis test for Scots pine 1187 able to use for growth. Photosynthesis might be limited by low rates of photosynthetic processes, or it may also be possible that the length of the photosynthetically active season is too short for assimilating enough photosynthates for growth and tissue renewal. According to Sveinbjo¨rnsson (2000), the high labile carbohydrate concentrations in tree line trees could be an adaptation to high rates of foliage loss and other damages in these trees. It has been reported that the net photosynthesis rate is not limited much by cold temperatures over a relatively broad range of air temperature. For example, Ha¨sler (1982) reported that temperatures between 10 and 24 C did not significantly change the rate of photosynthesis of alpine Pinus montana trees. Low temperature in conjunction with high light levels may, however, reduce photosynthetic production for several days (e.g. Ma¨kela¨ et al. 2004). James, Grace & Hoad (1994), who studied the growth and photosynthesis of tree line P. sylvestris population in Scotland demonstrated that the rate of photosynthesis in pine saplings (around 1 m tall) belonging to a tree line population was much lower during the early summer than the photosynthesis of trees in a valley population at similar temperatures, even though the photosynthetic rate of tree line population caught up with the valley population later in summer. While photosynthetic capacities of alpine herbaceous plants tend to increase from low to high altitudes (e.g. Ko¨rner 1999), changes in photosynthetic capacity of trees with latitude or altitude are not clear (e.g. Luoma 1997; Cordell et al. 1998; Benowicz, Guy & El-Kassay 2000). Nitrogen is usually not considered as a limiting factor for tree growth near the tree line, though there are a few studies which discuss this in Grace, Berninger & Nagy (2002). Sveinbjo¨rnsson, Nordell & Kauhanen (1992) found that nitrogen enrichment increased the leaf area and weight of mountain birch (Betula pubescens ssp. tortuosa) in Swedish Lapland. Nevertheless, some studies have established that nitrogen concentration in the needles of tree line trees is lower compared to the trees at lower altitudes; probably due to slower litter decomposition at lower temperatures. For example, Schulze, Chapin & Gebauer (1994) found out that in Alaska low nitrogen levels might be limiting the growth of spruce. Loomis et al. (2006) concluded that nitrogen cycling at the tree line in Alaska is limited by both low substrate quality and cold temperatures, which in turn slow down the rate of nitrogen turnover. Carbon sink-source relationships in trees have been studied extensively using manipulation experiments (e.g. Honkanen, Haukioja & Suomela 1994; Myers, Thomas & DeLucia 1999; Li, Hoch & Ko¨rner 2002). One way to reduce the strength of sinks in trees is to remove buds (debudding; e.g. Honkanen, Haukioja & Suomela 1994; Li, Hoch & Ko¨rner 2002). The sources can be manipulated either by increasing the CO2 concentration of air (e.g. Handa, Ko¨rner & Ha¨ttenschwiler 2005; Handa, Ko¨rner & Ha¨ttenschweiler 2006) or by removing the photosynthetically active tissues (defoliation; e.g. Honkanen, Haukioja & Suomela 1994; Lyytikaïnen-Saarenmaa 1999; Li, Hoch & Ko¨rner 2002; Handa, Ko¨rner & Ha¨ttenschwiler 2005; Handa, Ko¨rner & Ha¨ttenschweiler 2006).

These treatments may also induce other changes in tree physiology, like changes in water relations. For example, changes in water relations to defoliation and nitrogen availability have been proposed by Betson et al. (2007) and Berninger et al. (2000). Stable isotope ratios of carbon, measured as d13C values, can be used as an indicator of tree water use response in manipulation experiments. The photosynthetic carbon isotope discrimination is sensitive to the ratio of CO2 concentration inside leaves to that outside the leaves (Ci ⁄ Ca) (Farquhar, O¢Leary & Berry 1982). d13C varies as the value of Ci ⁄ Ca changes as a function of stomatal conductance and photosynthetic activity. Consequently, carbon isotope discrimination has also been commonly used as a measure of plant water use efficiency (WUE) (Farquhar, Ehleringer & Hubick 1989; Saurer, Siegwolf & Schweingruber 2004). We used manipulation experiments to study the growth response of Scots pine (Pinus sylvestris L.) trees belonging to a tree line population in eastern Finnish Lapland. We increased the availability of nitrogen by fertilization, restricted the availability of carbon by removing a part of the needles (defoliation) and improved the availability of carbon for growth by removing a part of the buds (debudding). We approached the growth restrictions of tree line trees with two hypotheses: the growth limitation hypothesis and the resource limitation hypothesis. We hypothesize that trees are sink limited, since they form the tree line at our site. A sink limitation would mean, according to our interpretation, that: (i) tree growth of neither shoots nor stems will increase as a response to debudding; (ii) growth of stems and shoots will, according to the sink limitation hypothesis, not change as a response to fertilization or defoliation; (iii) carbohydrate concentrations will increase as a response to debudding, but will remain unchanged as a response to the other treatments; and (iv) d13C will indicate if changes in the tree’s water relations occur due to the treatments.

Materials and methods The study took place in the Va¨rrio¨ nature park, which is located in the north-east part of Finland, near the Russian border (6748¢ N, 2940¢ E). Lowlands in the nature park are covered by taiga vegetation and the main tree species in the area are Scots pine (P. sylvestris L.) and Picea abies (L.) Karst. The upper slopes of the fells are dominated by Betula pubescens ssp. czerepanovii (Orlova) with scattered P. sylvestris trees forming the tree line and highest fell tops being treeless. The altitudinal tree line is c. 470 m. The mean annual precipitation of the area is about 600 mm and average mean temperature is )1 C (at Va¨rrio¨ research station, altitude 380 m). This study was made on the Va¨rrio¨ fell at an altitude of about 470 m a.s.l. The vegetation consisted of scattered small Scots pine trees (tree height usually less than 5 m) and fell field vegetation (mostly dwarf shrubs, mosses and arboricolous lichen). We established a manipulation experiment, which was started in spring 2003. In spring 2003 we selected 25 healthy trees, which were about 50 years old. Each tree was randomly assigned to one of four groups: debudding (six trees), defoliation (six trees), nutrient addition (six trees) and control (seven trees). The trees were situated on the north slope of the fell in an area of c. 1 ha. The maximum difference in altitude between

2010 The Authors. Journal compilation 2010 British Ecological Society, Journal of Ecology, 98, 1186–1195

1188 S. Susiluoto, E. Hilasvuori & F. Berninger the sample trees was 10 m. We did not measure tree density, but we estimated it to be about 100 trees ha)1 and the basal area less than 3 m2 ha)1. Average tree diameter at breast height (d.b.h.) was 7.2 cm (SD = 1.9 cm). Average tree height in 2003 was 270 cm (SD = 59 cm). From the trees belonging to the defoliation group, we removed all needles except the ones produced during the previous growing season (meaning that at the beginning of the growing season the trees had one needle cohort and at the end of the growing season each tree had two needle cohorts). From the trees belonging to the debudding group, we removed three buds out of four except from the apically dominant shoot of each branch, which were all left intact. These treatments were applied during spring 2003 and repeated in spring 2004. The apical buds of all branches were left untreated since they are known to create hormonal signals that affect growth (Forest et al. 2004). The trees belonging to the nutrient addition group received a slow-release N-fertilizer (Metsa¨n kestotyppi, Kemira GrowHow, Finland; 35% N, 1% Mg, 0.15% B) that was spread in a c. 2 m-wide zone around the trees on 25 May 2003 (175 g N tree)1; corresponding to 139 kg N ha)2), 22 August 2003 (52.5 g N tree)1 corresponding to 42 kg N ha)2), on May 27th, 2004 (175 g N tree)1, corresponding to 139 kg N ha)2) and on 9 June 2005 (35 g N tree)1; corresponding to 28 kg N ha)2). As the manufacturing of Metsa¨n kestotyppi fertilizer ended in 2004, we also used 28.5 g N tree)1 (23 kg N ha)2) of the slow-release N and P fertilizer (Metsa¨n NP1, Kemira GrowHow, Finland, 25% N, 2.2% P, 5.5% Ca, 1% Mg, 0.15% B and 0.15% Zn) on each tree from the nutrient addition treatment on 9 June 2005. Individual trees in our tree line site were scattered over a large area and root overlap between the trees is not likely to have occurred. To analyse the possible changes in carbohydrate concentration in wood tissue, we collected branch xylem samples from the trees every second week during the growing season of 2004. We chose first year xylem as an indicator of whole tree carbohydrate concentration, which is more clearly a reserve than foliage concentrations, in addition to being easy to sample with minimal damage to the tree. For each sample, we chose a healthy side branch from the northern side of the tree. Each sample consisted of the growth of the previous and ongoing growing seasons (if the growth during the ongoing season had already started). The bark and phloem were then peeled off the samples during the same day. Afterwards the samples were put into the microwave oven on full power for 1 min in order to kill all living cells and to end changes in carbohydrate composition. After this procedure, the samples were dried at 60 C for 3 h and then frozen. As the sample sizes were small, we combined two samples with each other (belonging to the same tree). These samples were collected on 17 and 31 May (spring), on 28 June and 12 July (summer) and on 20 August and 11 October (autumn), respectively. Branch growth started mid-June and most of it was completed by the beginning of July. Carbohydrate analyses were done at the Institute of Botany of the University of Basel, Switzerland, by using an enzymatic digest technique with subsequent spectrophotometric glucose test as described in Hoch, Popp & Ko¨rner (2002), Ko¨rner, Pelaez-Riedl & van Bel (1995) and Wong (1990). Approximately 10 mg of grinned branch xylem was boiled in 2 mL of distilled water for 30 min. The samples were then centrifuged and aliquots of 500 lL of the extracts were treated with isomerase and invertase in order to convert fructose and sucrose into glucose. After the enzymatic conversion to gluconate-6-phosphate (hexocinase reaction, hexocinase from Sigma Diagnostics, St. Louis, MO, USA), the total amount of glucose was determined photometrically using a 95-well microplate photometer (HR 7000; Hamilton, Reno, NE, USA). The remaining insoluble material, including starch, was then incubated at 40 C for 15 h with

a dialysed crude fungal amylase (Clarase, from Aspergillius oryzae, Enzyme Solutions Pty Ltd., Croydon South, Victoria, Australia) to break down the starch into glucose. The method previously described was used for determining the glucose concentration. As standards, a standard plant powder (Orchad leaves, Leco, St. Joseph, MI, USA), pure starch and glucose, fructose and sucrose solutions were used. The different types of sugars analysed in the study are a sum of glucose, fructose and sucrose. The starch was calculated as NSC (non-structural carbohydrates) minus low molecular weight sugars. For the carbon stable isotope analyses and for the estimation of the annual stem radial increment, we sampled the trees using an increment borer. Two samples from each tree, from the north and east side below the lowest living branches, were collected on 12 October 2004, and the growth of the last 5 years was used in the calculations. First, the tree ring widths were measured under a microscope, after which the selected samples were prepared for the carbon isotope analysis. Tree rings were cut into slivers with a surgical blade and a-cellulose was extracted from the wood following the method described by Loader et al. (1997). Cellulose samples of 70–100 lg were weighed into tin capsules. The samples were then combusted and CO2 separated in an elemental analyser (NC 2500). Gas was introduced into a mass spectrometer (Delta Advantage, Finnigan, Bremen, Germany) via an interface (ConFlo II or III). Carbon isotope results are expressed using the conventional d (delta) notation, where isotope ratios are expressed relative to the VPDB (Vienna PeeDee Belemnite) standard: d13 C ¼ 1000 ðRsample =Rstandard 1Þ where Rsample and Rstandard are the 13C ⁄ 12C ratios of the sample and a VPDB standard, respectively. Reproducibility for laboratory reference cellulose, measured in parallel with the studied samples, is 0.10% for d13C. Two replicate samples were analysed and the final d13C value was calculated as an average. Carbon isotopes were not measured from the debudded group as we did not have a clear hypothesis for the changes due to the treatment, did not consider them to be important for the interpretation of the results, and also because the analyses are fairly expensive. We collected foliage samples from each tree on 24 May 2003, 27 August 2003, 31 May 2004, 11 October 2004 and 2 October 2005 to analyse the possible changes in needle nitrogen concentration before and after growing season. Side branches from the northern side of the tree were chosen as samples. Each sample consisted of the growth of previous and ongoing years (on fall samples). The samples were dried at 60 C for 12 h. The analyses for needle nitrogen concentration were done with Leco elemental analyser (model CNS-1000; LECO Corporation, St. Joseph, MI, USA). Branch extension was measured from the branch samples that were collected in April 2005. From each tree, 3–8 samples were collected and the growth of the last 4 years was measured from the internodes. On 29 September 2005 we measured the height increment of the trees by measuring the length of the internodes of the terminal shoot of the main stem grown from 2001 to 2004, respectively. A time table for all the treatments and measurements is presented in Table 1.

STATISTICAL ANALYSES

We acknowledge that we worked in a spatially and temporally very heterogeneous environment and our initial tree population differed in size and growth history. We, therefore, chose to analyse changes in the growth rates of each tree rather than absolute values of production and physiology to reduce the level of heteroscedasticity in the data. We chose the pre-treatment growth levels as the reference.


Growth limitation hypothesis test for Scots pine 1189 Table 1. Timetable for the treatments and the measurements. Black squares denote actual measurements while grey squares denote annual values derived from the measurement of tree rings or annual extension growths, which were done after the experiment. NSC = non-structural carbohydrates, N = nitrogen Pre-experiment

Intensive survey + treatments

2001

Spring 2003

2002

Summer 2003

Autumn 2003

Post-experiment

Winter 2004

Spring 2004

Summer 2004

Autumn 2004

Winter 2005

Spring 2005

Autumn 2005

2005

Defoliation Debudding Fertilization NSC Sampling N Sampling Isotopes Branch Increment Height Increment

Table 2. Nitrogen concentrations in pine needles during the experiment. Values presented are average percentages of nitrogen from dry weight. Values in parentheses are SE of means. According to analysis of variance, the nitrogen concentrations differed between treatments only during falls 2004 and 2005. Letters in Dunnett’s test represent following statistically significant differences (at P < 0.05): A = Control – Fertilization and NA = No statistically significant differences. N = number of sampled trees Control n Sampling dates 24 May 2003 27 August 2003 31 May 2004 11 October 2004 2 October 2005

7 1.34 1.35 1.19 1.16 1.27

Debudding 6

(0.027) (0.041) (0.050) (0.031) (0.042)

1.32 1.39 1.19 1.18 1.33

Defoliation 6

(0.046) (0067) (0.053) (0.068) (0.066)

As there are large variations in all of the parameters between trees living in tree line environments, we subtracted the average values of the previous years of carbon isotope ratio, tree ring width (TRW) and height growth increment from the results of 2003, 2004 and 2005 (depending on the measured parameter). Data from 2000 to 2002 were considered as pre-treatment reference growth. The variable used in the statistical tests was the difference of the observed growth to the average pre-treatment growth. For the isotope ratio and TRW, two samples were collected from each tree (east and north sides). We used the average of both values for the calculations. For the statistical analyses of the height increment, we did not use the results from 2003 as branch extension is widely known to be pre-determined. We used a slightly different approach to analyse branch extension since we had 3–8 branch samples per tree and an initial analysis of residuals revealed that the branch extension depended on the growth of previous years in a nonlinear manner. In addition, we had several branch measurements per tree available. We used, therefore, a nonlinear mixed effect model (with treatments as fixed, and trees as random effects). The data was adjusted for heteroscedasticity by weighting the residual using a variable power function. The value of the parameter b was kept constant for all treatments. The model presented was better than other models we tested (including models estimating treatment effects on b and linear models) based on the Akaike Information criterion. As carbohydrate samples were collected only during one summer, no modifications were performed for the results; they are presented as averages of each treatment.

1.38 1.40 1.16 1.08 1.25

Nutrient addition

Dunnett’s test

6 (0.046) (0.037) (0.030) (0.029) (0.041)

1.31 1.44 1.25 1.34 1.62

(0.050) (0.036) (0.046) (0.035) (0.030)

NA NA NA A A

All statistical analyses were performed using R 2.5.1 statistical software package (R Development Core Team 2007) and the STAT, NLME (Pinheiro et al. 2007) and multcomp libraries (Hothorn et al. 2008). The usual analysis was the analysis of variance and Dunnett’s test was used to test the differences between the treatments and the control.

Results Needle nitrogen concentrations in the trees belonging to the nutrient addition group increased (Table 2). Nitrogen concentration in the needles differed statistically from the control group starting from fall 2004 (at P < 0.01). The nitrogen concentrations of the other groups did not differ statistically from the control group. The differences in nitrogen concentration between the non-nutrient added treatments and the control were small (less than 0.075%) and the confidence intervals around estimated means varied between±0.17% and 0.09%. There were statistically significant differences in the annual height increment (at P < 0.05; Fig. 1; Table 3). The nutrient addition treatment differed significantly during both years (2004 and 2005) from the control treatment (at P < 0.05). Other treatments were not significantly different from the control. On average, the annual height increment of the terminal shoot was 11.8 cm in 2001, 11.5 cm in 2002 and 10.9 cm in


1190 S. Susiluoto, E. Hilasvuori & F. Berninger 1

16 Change in branch extension (cm)

Normalized height increment (cm)

*

*

14 12 * 10 8 6 4 2

0.5

* 0

*

–0.5

–1

0 –2

–1.5 2002

–4 2002

2003

2004

Fig. 1. The height growth increment in trees belonging to different treatments compared to the height growth in 2001 (values for 2001 were subtracted from each value). Lines in the figure represent the standard error of means. White = control, light grey = debudding, dark grey = defoliation and black = nutrient addition. N for control = 7, debudding = 6, defoliation = 6 and nutrient addition = 6. Stars refer to significant treatment differences compared to control using a Dunnett’s test (as described in Materials and methods).

Table 3. Results of Analysis of Variance (anova) and Dunnett’s test (at P < 0.05) for carbon isotopes, tree ring growth and length increment. Letters is Dunett’s test mean that the results differ statistically from each other at P < 0.05. A = Defoliation-Control, B = Nutrient addition-Control and NA = No statistically significant differences Carbon isotopes

Tree ring width

Height growth

Year

2004

2004

2005

anova

F = 3.979, P < 0.047 NA

F = 5.531, P < 0.009 A

F = 6.237, P < 0.003 B

2003 F = 1.280, P < 0.314 NA

2003 F = 4.134, P < 0.025 A

2004 F = 4.715, P < 0.011 B

Year anova Dunnett’s test

2004

2005

Year

Dunnett’s test

2003 Year

2003. The average growth rate of the control treatment was between 9.0 and 11.6 cm year)1 from 2001 to 2005. During the experiment period, trees in the nutrient addition treatment increased their growth by over 200%, growing on average 21.3 cm in 2005. It is worth noting that growth of debudded trees increased by c. 35%, with an average growth of 19.3 cm in 2005, but this difference was not statistically significant. From the confidence intervals of our data we can conclude that during 2005, differences of 9.3 cm of height growth would have been significant. The side branch extension growth in both nutrient addition and debudded groups was higher than in the control group

Fig. 2. Change in branch extension during 2002–04 compared to the branch extension in 2001 (values for 2001 were subtracted from each value). Lines in the figure represent the standard error of means. White = control, light grey = debudding, dark grey = defoliation and black = nutrient addition. N for control = 7, debudding = 6, defoliation = 6 and nutrient addition = 6. Stars refer to significant treatment differences compared to control using a nonlinear mixed effect model (Table 4) (as described in Materials and methods).

and both groups differed statistically from the control group in 2004. However, the only statistically significant difference in 2003 was between control and debudded group (at P < 0.05; Fig. 2; Table 4). The average annual side branch extension in all treatments was between 1.8 and 2.6 cm. Growth increases in 2004 (as calculated from changes of the parameter a) compared to the control were 47% for the debudding, 17% for the defoliation and 27% for the fertilization treatment. There were 188 branch measurements and the coefficient of determination (also called pseudo-R2) of the model was 0.54. There were also statistically significant differences in the annual TRW (at P < 0.05; Fig. 3; Table 3). On average, the tree diameter growth was 3.35 mm in 2000, 3.06 mm in 2001, 3.11 mm in 2002, 2.86 mm in 2003 and 2.87 mm in 2004 across all the treatments. During 2004, the average tree diameter increment was 3.92 mm for the control group, 3.44 mm for the debudded group, 1.87 mm for the defoliated group and 3.26 mm for the nutrient addition group. Using changes in growth, rather than absolute growth as shown in the materials and methods, the annual TRW differed significantly at Table 4. Parameter values for the model of branch extension (branch extension = a*((branch extension2002 + branch extension2001) ⁄ 2))b. Note that P-values denote the probability that a treatment is different from the control. The parameters a and b were significantly different from 0 (P < 0.001) for both years

Treatment

Value a Value a for branch for branch P-value P-value extension2003 extension2004 2003 2004

Control Debudding Defoliation Nutrient addition Parameter b

0.90 1.17 0.80 0.91 0.72

1.21 1.78 1.42 1.54 0.61

1 0.02 0.37 0.90

1 0.003 0.20 0.04


Growth limitation hypothesis test for Scots pine 1191

Change in annual tree ring width (mm)

1 0.8 0.6 0.4 0.2 0 –0.2 –0.4 *

–0.6

*

–0.8 –1 –1.2

2001

2002

2003

2004

Year

Fig. 3. Change in annual tree ring width. From each measurement we subtracted the average growth of a tree during year 2000. Lines in the figure represent the standard error of means. White = control, light grey = debudding, dark grey = defoliation and black = nutrient addition. N for control and debudding = 4, defoliation = 6 and nutrient addition = 5. Stars refer to significant treatment differences compared to control using a Dunnett’s test (as described in the methods). Debudding and defoliation treatments were applied in spring 2003 and repeated during spring 2004. Nutrient addition treatment was applied in spring 2003, fall 2003, spring 2004, fall 2004 and spring 2005.

P < 0.05 between the defoliated group and the control group in both years. Differences in the changes in growth of the debudded group and the nutrient addition treated trees, compared to the control, were modest (0.42 mm for the fertilized and )0.014 mm for the debudding treatment). There was also a significant difference in d13C change values (P < 0.05) between treatments during 2004, but the results did not differ from each other during 2003 (Fig. 4; Table 3). However, no statistically significant differences were observed

Change in carbon isotope ratio (‰)

1.2 0.9 0.6 0.3 0 –0.3 –0.6

2001

2002

2003

2004

Year

Fig. 4. Average carbon isotope composition between 2001 and 2004 in nutrient addition, defoliation and control treatments. From each value we reduced the isotope composition value during year 2000 to get the basic level the same for each tree. Lines in the figure represent the standard error of means. White = control, dark grey = defoliation and black = nutrient addition. Number of samples in each treatment was four for control, five for fertilization and six for defoliation.

between the control group and any of the treated groups in d13Cchange value. A significant difference would have been c. 0.81& and the observed differences between the treatments and the control were less than 0.45&. The average carbon isotopes value for all trees was )26.31& on 2000 and )26.43& on 2004. The sugar concentrations in the branch xylem were the highest during spring (the average for all trees was 77.5 mg g)1) and then dropped to about half their value after growth initiation (average value 38.4 mg g)1; Fig. 5a). The sugar concentrations increased again during fall to the average value of 69.2 mg g)1. The concentrations of starch were the highest in the middle of the growing season (57.4 mg g)1) and lowest in fall (31.3 mg g)1; Fig. 5b). The proportion of starch from NSC was 35% in spring, 60% in summer and 33% in fall. According to Dunnett’s test, the only statistically significant difference was found in the NSC levels in spring, as trees in the defoliated group had lower NSC concentration than trees in the control group (at P < 0.05 using a Dunnett’s test; Fig. 5a–c). Analysis of the confidence intervals of the treatment effects reveals that differences of 2.42% during spring, 2.61% during summer and 1.65% during fall would have been statistically significant but all calculated differences between the treatments and the control group were less than 0.8% (fall) and 1% (summer). During the spring, all the treatments (except for defoliation) differed less than 0.6% from the control.

Discussion The experiment reveals that, perhaps not surprisingly, the growth of the trees at the tree line changes if the resources available for the growth of new plant organs are manipulated. The results, however, did not concord well with the hypothesis that trees are sink limited, but they corroborate the alternative resource limitation theory. In the light of our results, we argue that resource limitation might, at least at this particular tree line site, be predominant since the trees were able to increase the growth of the remaining meristems, as a response to the removal of meristems, and were increasing their overall growth in response to changes in resource availability. As a response to the treatments, we observed changes in allocation since some treatments affected height and diameter growth differently. Non-structural carbohydrate concentrations in twigs, however, remained pretty constant and only one minor change in total carbohydrate concentrations was statistically significant. Defoliation and debudding might change the hormonal balance of the tree. However, in our experiment the treatments were less radical than in other experiments, where similar theories were tested (Li, Hoch & Ko¨rner 2002; Palacio et al. 2008). We also attempted to minimize the effects of the treatments on the hormonal balance of the trees by leaving apical buds and the newest year of foliage intact. Another potential effect is that debudding reduces the growth of new foliage. This could result in a reduction of total tree foliage biomass and finally in a reduction of tree growth. Our data, however, shows that


1192 S. Susiluoto, E. Hilasvuori & F. Berninger

80.0

80.0 Starch (mg g–1)

(b) 100.0

Sugars (mg g–1)

(a) 100.0

60.0

40.0

60.0

40.0

20.0

20.0

0.0

0.0 Spring

Summer Season

Fall

Spring

Summer Season

Fall

(c) 150.0

NSC (mg g–1)

120.0

*

90.0

60.0

30.0

0.0 Spring

Summer Season

Fall

debudded trees increased the growth of the remaining buds and that the tree diameter growth was not reduced. Altogether, we think that the unwanted artefacts of our treatments are reasonably small and do not affect the validity of our results. Defoliation did not change the branch extension (Fig. 2; Table 3) or the apical growth of the terminal shoot (Fig. 1; Table 2). However, the diameter growth of the main stem decreased (Fig. 2; Table 2). Also, Handa, Ko¨rner & Ha¨ttenschweiler (2006) described that Larix decidua decreased the diameter growth during the year as a response to artificial defoliation and the growth of both defoliated L. decidua and Pinus cembra was clearly decreased during the following year. The differences in the response of the diameter and height growth to the treatments indicate changes in allocation, which seem to maintain or restore the functional and structural balances of the trees (e.g. Ma¨kela¨ 1999). Ericsson et al. (1985) treated Scots pine trees by shoot pruning (as a simulation of a Tomicus pinperda L. attack). As a result, the impacts of defoliation on foliage biomass were only moderate since new shoots had a higher foliage-to-shoot biomass ratio. In another paper of these authors (Ericsson, Larsson & Tenow 1980), shoot extension growth was reduced by foliage removal. The actual reduction in photosynthetic production will, however, be lower than the percentage of foliage removed, since clipping reduces self shading and the physiologically more active 1-year-old needles remain on the tree. Our Scots pines had a tight apical control and we did not observe any sprouting of secondary buds in the debudding or any other treatment. Also, height growth of Scots pine in Scandinavia is strongly predetermined (e.g. Junttila 1986) and we

Fig. 5. Sugars (a), starch (b) and total NSC (non-structural carbohydrate) concentration (c) in branch xylem. Lines above the bars describe the standard error of means. According to Dunnett’s test the only statistically significant result was found in the results of NSC on spring between defoliation and control. White = control, light grey = defoliation, dark grey = nutrient addition and black = debudding. N for control = 7, debudding = 6, defoliation = 6 and nutrient addition = 6.

did not observe any formation of internodal whorls. The branch extension increased in the debudded trees by 45%, as previously reported by Susiluoto et al. (2007) in a similar experiment (using a slightly different statistical analysis). On the other hand, TRW did not change due to the debudding treatment. It seems that after a large destruction of the apices, the trees are compensating for the damage by allocating a larger proportion of the resources for the extension of new branches without using secondary dormant buds. We assumed that trees would increase their foliar nitrogen concentrations after debudding, since debudding would decrease the growth of new foliage without changing fine root biomass. However, we did not observe any increase in foliar nitrogen content, probably because the increased production of the remaining buds was sufficient to provide a sink for most of the ‘surplus’ nitrogen. Nutrient addition clearly increased the height and branch extension of the trees, whereas radial growth remained unaffected. Also, Schulze, Chapin & Gebauer (1994) studied Picea glauca and P. mariana at a tree line in Alaska and found that the twig length correlated strongly with the nitrogen concentration in the needles. Sveinbjo¨rnsson, Nordell & Kauhanen (1992) demonstrated a stronger correlation with growth and leaf N concentration in Betula pubescens ssp. czerepanovii growing at a tree line than in a valley in the Swedish Lapland. They also found that the height increment of trees growing at tree line responded more strongly to the nitrogen fertilization than the trees growing in valley conditions, even though the a priori nitrogen concentration in the leaves was substantially higher in the needles of tree line trees than in valley trees. How-


Growth limitation hypothesis test for Scots pine 1193 ever, in Sveinbjo¨rnsson’s study in south central Alaska (Sveinbjo¨rnsson 2000), no difference was found in growth of Picea glauca growing at or below the tree line due to nitrogen fertilization. Altogether, the data provides evidence against our hypothesis that trees are sink limited. Growth was sensitive to changes in the supply of carbon or nutrients. Sink removal resulted in an increase of the growth of the remaining meristems. Our manipulations resulted in large changes in the above-ground allocation of the trees since they responded to some treatments only by changing their height growth and to other treatments by changing their diameter growth. The results do not give a clear answer to whether nitrogen or carbon limits the growth. This fits well into the ideas of Millard, Sommerkorn & Grelet (2007) who reviewed nitrogen and carbon limitations: carbon and nitrogen acquisition are linked and trees could, to some extent, accelerate the nitrogen cycle by allocating more carbon to mycorrhizal fungi. In our experiment, tree growth changed notably while the NSC concentration remained constant (Fig. 5a–c). This implies that trees react to changing conditions primarily by adjusting their growth and keeping their carbohydrate reserves fairly constant. Observed changes in the growth of shoots and stems fit into this picture. This conflicts with the works of Ericsson, Larsson & Tenow (1980) who observed in more southerly Scots pine that carbohydrate concentrations were easily changed as a response to defoliation and shoot pruning. On the other hand Li, Hoch & Ko¨rner (2002) showed that, not surprisingly, complete defoliation decreased the carbon reserves in Pinus cembra at an alpine tree line in the central Alps. It is worth noting that there is little evidence that high carbohydrate concentrations in tree line trees would limit photosynthetic rates through sucrose feedback (as described by the works of Stitt (e.g. Stitt et al. 1988)). According to Bansal & Germino (2008), tree seedlings at the tree line in the Rocky Mountains had higher NSC concentrations than seedlings at lower elevations, but as there were no indications of negative feedback of high carbohydrate concentrations on gas exchange, they concluded that the accumulation of the NSC did not suppress photosynthesis. Susiluoto et al. (2007) reached similar conclusions based on sapflow data from the experiment used in this study. It is well-established that northern and high elevation tree line populations of pines have higher carbohydrate concentrations than more southern or low elevation trees. Hoch (Hoch, Popp & Ko¨rner 2002; Hoch & Ko¨rner 2003) studied carbohydrate dynamics of pines in the Alps, the Mexican Sierra and the Scandinavian Alps and found that tree line trees had higher or at least equal carbohydrate concentrations than lowland trees. Also, Kaipiainen & Sofronova (2004) compared the carbohydrate dynamics of Scots pine populations growing in the Russian area of Karelia (at geographical latitude of 62 N) to a population growing in the more northern Kola Pennisula (latitude of 69 N) and found that the carbohydrate concentrations in branches and needles of the population growing in Kola Peninsula were higher than in the southern population. The establishment of whole tree carbohydrate reserves is diffi-

cult and depends much on how well the concentrations and masses of different tree parts are estimated (e.g. Berninger & Salas 2002). In this study we used carbohydrate concentrations in branches as indicators of the whole tree carbohydrate status as was done by Li, Hoch & Ko¨rner (2002). We acknowledge that pines have important mobile carbon reservoirs in the form of lipids (e.g. Hoch & Ko¨rner 2003) but we believe that branch xylem carbohydrate concentrations are probably valid indicators of whole tree carbohydrate status. In many areas, tree line trees occasionally experience biomass losses during winter and ⁄ or death of tissues due to cold nights during growth season. Even though these are not necessarily very frequent, trees in such risky environments must have developed strategies to cope with these losses of foliage, which might otherwise be fatal. A sensible solution for surviving at the tree line would be to have a basic mechanism to cope with unexpected – and often large – loss of biomass from snow, frost or wind damages. Sveinbjo¨rnsson (2000) proposed that the high carbohydrate reserves of tree line trees are part of a strategy to survive from frequent biomass losses caused by an extreme climate. Palacio et al. (2008) showed by measuring Betula pubescens that even though the NSC levels decreased in short time period due to herbivory, the NSC levels will recover to the normal level after a longer time period (7 years in the study in question) even though the herbivory continues. It is worth noting that Palacio’s simulated grazing treatments resemble a combination of our debudding and pruning treatments rather than a pure debudding or pure defoliation treatment. The work of Vanderklein & Reich (1999) shows that both growth and carbohydrate reserves of temperate pines does not necessarily change as a reaction to partial defoliation. Also, Ho´dar et al. (2008) observed that Scots pine in subalpine conditions in Spain can compensate to a large extent for foliage and meristem losses, but that no overcompensation of growth occurs even under favourable light and nutrient conditions. All in all, we think that carbohydrate reserves are well buffered and trees seem to maintain their carbohydrate reserves at fairly constant levels. These constant levels are maintained even under stress: sink removal (debudding) did not increase carbohydrate concentrations but the effect of source removal (defoliation) induced only a short term decrease of the concentrations during spring, which indicates that the amount of usable carbon for growth was slightly diminished during that period. As a response to debudding, the growth of the remaining apical meristems (in the branches) was increased. We conclude that trees can avoid growth limitation by adjusting their patterns of biomass allocation. Several papers have reported that trees at the tree line are xeromorphic and drought adapted (e.g. Li, Hoch & Ko¨rner 2002). We, therefore, hypothesized that the defoliation treatment will change the tree’s water relations (as speculated before by Berninger et al. 2000). This would lead to decreased d13C abundances in defoliated trees as the reduced leaf surface enables higher transpiration rates and higher stomatal conductances. This would increase the diffusion of CO2 into the leaf (Meinzer & Grantz 1990; Reich et al. 1993). The observed differences in carbon isotope ratios were, however, not statisti-


1194 S. Susiluoto, E. Hilasvuori & F. Berninger cally significant, although the direction of the change was consistent with our expectations. In the literature there are different ideas of how defoliation could affect the d13C of trees. Some studies observed unaffected d13C signatures in tree rings of host species during insect infestation (Ellsworth et al. 1994; Haavik et al. 2008; Kress et al. 2009) and others enriched d13C values due to compensatory increase in photosynthetic rate (Simard et al. 2008). We predicted that nutrient addition would increase d13C (Scha¨fer et al. 2002), since the increased needle nitrogen concentration enhances the carboxylation capacity by increasing the amount of Rubisco enzyme in Calvin cycle. In consequence, water use efficiency of the tree increases (Ripullone et al. 2004). The higher nitrogen concentrations in the foliage of fertilized trees indicate that photosynthetic efficiency was probably increased. In the long-term N-fertilization study of Betson et al. (2007) in the north of Sweden pine needle, d13C values were found to be 0.45% higher compared to control. This is in agreement with the direction and the order of magnitude of the change in our results. Even though the individual treatments did not differ from the control, the differences between treatments were statistically significant in 2004 according to analysis of variance. Nevertheless, changes in carbon isotopic abundances in our trees were small (less than 0.3% from the control) and the effect of our treatments on the water use efficiency of trees seems to be modest. The observed changes would result in changes in substomatal carbon dioxide concentration of less than 5 p.p.m. (calculated using the equations of Farquhar, O¢Leary & Berry (1982)). This indicates that the hydraulic adjustment of our trees to changes in the foliage-fine root ratio or source-sink relationships is, at best, minor. Our overall results suggest that the tree growth of Scots pine with the current climatic conditions is not limited by its capacity to grow but by the available resources in our study area. They could adjust their architecture in a plastic way to a reduction in sink strength (like defoliation) by increasing branch extension. Nitrogen is a strongly limiting factor for our tree line trees, since reactions of growth to nitrogen addition were substantial. However, throughout this study we have to keep in mind that growth increases are not necessarily a good proxy of the fitness and persistence of trees as a lifeform in the arctic alpine transition zone. Also, it is worth noting that the late summers (July and August) of 2003 and 2004 were warmer than usual at the site. This should, however, not have affected the extension growth that takes place in early summer (late June) when temperatures were close to their long-term averages. This research shows that resource limitation dominates the variations of productivity at this tree line.

Acknowledgements We want to thank Dr. Christian Ko¨rner for his helpful comments on the paper. We also want to thank his team, especially Dr. Gu¨nter Hoch for their help in performing the carbohydrate analyses. The measurements could not have been done without the support of the staff in Va¨rrio¨ Research Station. We also thank Eloni Sonninen and Igor Shevchuk for isotope analyses, and we are grateful to Henrik Brunholm for assistance with isotope sample preparation. The research

was funded by Finnish Cultural Foundation and Maj and Tor Nessling foundation (Project no. 2108018). Kasia Richer-Juraszek and Emmanuelle Frechette were kind enough to revise our English.

References Bansal, S. & Germino, M.J. (2008) Carbon balance of conifer seedlings at timberline: relative changes in uptake, storage, and utilization. Oecologia, 158, 217–227. Benowicz, A., Guy, R.D. & El-Kassay, Y.A. (2000) Geographic pattern of genetic variation in photosynthetic capacity and growth in two hardwood species from British Columbia. Oecologia, 123, 168–174. Berninger, F. & Salas, E. (2002) Effects of pruning intensity on biomass dynamics of Erythrina lanceolata. Agroforestry Systems, 57, 17–26. Berninger, F., Sonninen, E., Aalto, T. & Lloud, J. (2000) Modelling 13C discrimination in tree rings. Global Biogeochemical Cycles, 14, 213–223. Betson, N.R., Johannisson, C., Lo¨fvenius, M.O., Grip, H., Granstro¨m, A. & Ho¨gberg, P. (2007) Variation in the d13C of foliage of Pinus sylvestris L. in relation to climate and additions of nitrogen: analysis of a 32-year chronology. Global Change Biology, 13, 2317–2328. Cordell, S., Goldstein, G., Mueller-Dombois, D., Webb, D. & Vitousek, P.M. (1998) Physiological and morphological variation in Metrosideros polymorpha, a dominant Hawaiian tree species, along an altitudinal gradient: the role of phenotypic plasticity. Oecologia, 113, 188–196. Ellsworth, D.S., Tyree, M.T., Parker, B.L. & Skinner, M. (1994) Photosynthesis and water-use efficiency of sugar maple (Acer saccharum) in relation to pear thrips defoliation. Tree Physiology, 14, 619–632. Ericsson, A., Larsson, S. & Tenow, O. (1980) Effects of early and late season defoliation on growth and carbohydrate dynamics in Scots pine. Journal of Applied Ecology, 17, 747–769. Ericsson, A., Hellqvist, C., La˚ngstro¨m, B., Larsson, S. & Tenow, O. (1985) Effects on growth of simulated and induced shoot pruning by Tomicus piniperda as related to carbohydrate and nitrogen dynamics in Scots pine. Journal of Applied Ecology, 22, 105–124. Farquhar, G.D., Ehleringer, J.R. & Hubick, K.T. (1989) Carbon isotope discrimination and photosynthesis. Annual Review of Plant Physiology and Plant Molecular Biology, 40, 503–537. Farquhar, G.D., O¢Leary, M. & Berry, J.A. (1982) On the relationship between carbon isotope discrimination and the intercellular carbon dioxide concentration in leaves. Australian Journal of Plant Physiology, 9, 121–137. Forest, L., San Martin, J., Padilla, F., Chassat, F., Giroud, F. & Demongeot, J. (2004) Morphogenetic processes: application to cambial growth dynamics. Acta Biotheoretica, 52, 415–438. Grace, J., Berninger, F. & Nagy, L. (2002) Impacts of climate change on the tree line. Annals of Botany, 90, 537–544. Haavik, L.J., Stephen, F.M., Fierke, M.K., Salisbury, V.B., Leavitt, S.W. & Billings, S.A. (2008) Dendrochronological parameters of northern red oak (Quercus rubra_L. (Fagaceae)) infested with red oak borer (Enaphalodes rufulus (Hadelman) (Coleoptera: Cerambycidae)). Forest Ecology and Management, 255, 1501–1509. Handa, I.T., Ko¨rner, C. & Ha¨ttenschweiler, S. (2006) Conifer stem growth at the altitudinal treeline in response to four years of CO2 enrichment. Global Change Biology, 12, 2417–2430. Handa, I.T., Ko¨rner, C. & Ha¨ttenschwiler, S. (2005) A test of the tree-line carbon limitation hypothesis by in situ CO2 enrichment and defoliation. Ecology, 86, 1288–1300. Ha¨sler, R. (1982) Net photosynthesis and transpiration of Pinus montana on east and north facing slopes at alpine timberline. Oecologia, 54, 14–22. Hoch, G. & Ko¨rner, C. (2003) The carbon charging of pines at the climatic treeline: a global comparison. Oecologia, 135, 10–21. Hoch, G., Popp, M. & Ko¨rner, C. (2002) Altitudinal increase of mobile carbon pools in Pinus cembra suggests sink limitation of growth at the Swiss treeline. Oikos, 98, 361–374. Ho´dar, J.A., Zamora, R., Castro, J., Go´mez, J.M. & Garcı´ a, D. (2008) Biomass allocation and growth responses of Scots pine saplings to simulated herbivory depend on plant age and light availability. Plant Ecology, 197, 229–238. Honkanen, T., Haukioja, E. & Suomela, J. (1994) Effects of simulated defoliation and debudding on needle and shoot growth in Scots pine (Pinus-sylvestris) – implications of plant source-sink relationships for plant – herbivore studies. Functional Ecology, 8, 631–639. Hothorn, T., Bretz, F., Westfall, P. & Heiberger, R.M. (2008) Multicomp: simultaneous inference in general parametric models. URL. R-project.org. R package version 1.0-0.


Growth limitation hypothesis test for Scots pine 1195 James, J.C., Grace, J. & Hoad, S.P. (1994) Growth and photosynthesis of Pinus sylvestris at its altitudinal limit in Scotland. Journal of Ecology, 82, 297–306. Junttila, O. (1986) Effects of temperature on shoot growth in northern provenances of Pinus sylvestris L Tree Physiology, 1, 185–192. Kaipiainen, L.K. & Sofronova, G.I. (2004) The role of the transport system in the control of the source-sink relations in Pinus sylvestris. Russian Journal of Plant Physiology, 50, 125–132. Ko¨rner, Ch. (1999) Alpine Plant Life. Springer-Verlag, Berlin. pp. 109–139. Ko¨rner, Ch. & Hoch, G. (2006) A test of treeline theory on a montane permafrost island. Arctic, Antarctic and Alpine Research, 38, 113–119. Ko¨rner, Ch., Pelaez-Riedl, S. & van Bel, A.J.E. (1995) CO2 responsiveness of plants: a possible link to phloem loading. Plant, Cell and Environment, 18, 595–600. Kress, A., Saurer, M., Bu¨ntgen, U., Treydte, K.S., Bugmann, H. & Siegwolf, R.T.W. (2009) Summer temperature dependency of larch budmoth outbreaks revealed by Alpine tree-ring isotope chronologies. Oecologia, 160, 353–365. Li, M.H., Hoch, G. & Ko¨rner, Ch. (2002) Source ⁄ sink removal affects mobile carbohydrates in Pinus cembra at the Swiss treeline. Trees – Structure and Function, 16, 331–337. Loader, N.J., Robertson, I., Barker, A.C., Switsur, V.R. & Waterhouse, J.S. (1997) A modified method for the batch processing of small wholewood samples to a-cellulose. Chemical Geology, 136, 313–317. Loomis, P.F., Ruess, R.W., Sveinbjo¨rnsson, B. & Kjelland, K. (2006) Nitrogen cycling at treeline: latitudinal and elevational patterns across a boreal landscape. Ecoscience, 13, 544–556. Luoma, S. (1997) Geographical pattern in photosynthetic light response of Pinus sylvestris in Europe. Functional Ecology, 11, 273–281. Lyytikaïnen-Saarenmaa, P. (1999) The responses of Scots pine, Pinus sylvestris, to natural and artificial defoliation stress. Ecological Applications, 9, 469– 474. Ma¨kela¨, A. (1999) Acclimation in dynamic models based on structural relationships. Functional Ecology, 13, 145–156. Ma¨kela¨, A., Hari, P., Berninger, F., Ha¨nninen, H. & Nikinmaa, E. (2004) Acclimation of photosynthetic capacity in Scots pine to the annual cycle of temperature. Tree Physiology, 24, 369–376. Meinzer, F.C. & Grantz, D.A. (1990) Stomatal and hydraulic conductance in growing sugarcane: stomatal adjustment to water transport capacity. Plant, Cell and Environment, 13, 383–388. Millard, P., Sommerkorn, M. & Grelet, G.-A. (2007) Environmental change and carbon limitation in trees: a biochemical, ecophysiological and ecosystem appraisal. New Phytologist, 175, 11–28. Myers, D.A., Thomas, R.B. & DeLucia, E.H. (1999) Photosynthetic responses of loblolly pine (Pinus taeda) needles to experimental reduction in sink demand. Tree Physiology, 19, 235–242. Palacio, S., Hester, A.J., Maestro, M. & Millard, P. (2008) Browsed Betula pubescens trees are not carbon-limited. Functional Ecology, 22, 808–815. Pietarinen, I., Kanninen, M., Hari, P. & Kelloma¨ki, S. (1982) A simulation model for daily growth of shoots, needles, and stem diameter in Scots pine trees. Forest Science, 28, 573–581.

Pinheiro, J., Bates, D., DebRoy, S. & Sarkar, D. (2007) nlme: Linear and Nonlinear Mixed Effects Models. R package version 3.1-83. R Foundation for Statistical Computing, Vienna, Austria. R Development Core Team (2007) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Reich, P.B., Walter, M.B., Krause, S.C., Vanderklein, D.W., Raffa, K.F. & Tabone, T. (1993) Growth, nutrition and gas-exchange of Pinus resinosa following artificial defoliation. Trees – Structure and Function, 7, 67–77. Ripullone, F., Lauteri, M., Grassi, G., Amato, M. & Borghetti, M. (2004) Variation in nitrogen supply changes water-use efficiency of Pseudotsuga menziesii and Populus · euroamericana; a comparison of three approaches to determine water-use efficiency. Tree Physiology, 24, 671–679. Saurer, M., Siegwolf, R. & Schweingruber, F. (2004) Carbon isotope discrimination indicates improving water-use efficiency of trees in northern Eurasia over the last 100 years. Global Change Biology, 10, 2109–2120. Scha¨fer, K.V.R., Oren, R., Lai, C.T. & Katul, G.G. (2002) Hydrologic balance in an intact temperate forest ecosystem under ambient and elevated atmospheric CO2 concentration. Global Change Biology, 8, 895–911. Schulze, E.-D., Chapin, F.S., III & Gebauer, G. (1994) Nitrogen nutrition and isotope differences among life forms at the northern treeline of Alaska. Oecologia, 100, 406–412. Shi, P.L., Ko¨rner, C. & Hoch, G. (2006) End of season carbon supply status of woody species near the treeline in western China. Basic and Applied Ecology, 7, 370–377. Simard, S., Elhani, S., Morin, H., Krause, C. & Cherubini, P. (2008) Carbon and oxygen stable isotopes from tree-rings to identify spruce budworm outbreaks in the boreal forest of Quebec. Chemical Geology, 252, 80–87. Stitt, M., Wilke, I., Feil, R. & Heldt, H.W. (1988) Coarse control of sucrosephosphate synthase in leaves: alterations of the kinetic properties in response to the rate of photosynthesis and the accumulation of sucrose. Planta, 174, 217–230. Susiluoto, S., Pera¨ma¨ki, M., Nikinmaa, E. & Berninger, F. (2007) Effects of sink removal on transpiration at the treeline: implications for growth limitation hypothesis. Environmental and Experimental Botany, 60, 334–339. Sveinbjo¨rnsson, B. (2000) North American and European treelines: external forces and internal processes controlling position. Ambio, 29, 388–395. Sveinbjo¨rnsson, B., Nordell, O. & Kauhanen, H. (1992) Nutrient relations of mountain birch growth at and below the elevational tree-line in Swedish Lapland. Functional Ecology, 6, 213–220. Tranquillini, W. (1979) Physiological Ecology of the Alpine Timberline. Springer-Verlag, Berlin. Vanderklein, D.W. & Reich, P.B. (1999) The effect of defoliation intensity and history on photosynthesis, growth and carbon reserves of two conifers with contrasting leaf lifespans and growth habits. New Phytologist, 144, 121–132. Wieser, G. (1997) Carbon dioxide gas exchange of cembran pine (Pinus cembra) at the alpine timberline during winter. Tree Physiology, 17, 473–477. Wong, S.-C. (1990) Elevated atmospheric partial pressure of CO2 and plant growth. Photosynthesis Research, 23, 171–180. Received 22 November 2009; accepted 7 May 2010 Handling Editor: Richard Bardgett
