Variation of mobile carbon reserves in trees at the alpine treeline ecotone is under environmental control


Author for correspondence:
Alex Fajardo
Tel: +56 67 247821


  • In low temperature-adapted plants, including treeline trees, light-saturated photosynthesis is considerably less sensitive to temperature than growth. As a consequence, all plants tested so far show increased nonstructural carbohydrate (NSC) tissue concentrations when exposed to low temperatures. Reduced carbon supply is thus an unlikely cause for low temperature range limits of plants. For altitudinal treeline trees there is, however, a possibility that high NSC genotypes have been selected.
  • Here, we explored this possibility using afforestations with single-provenance conifers along elevational gradients in the Southern Chilean Andes and the Swiss Alps.
  • Tree growth was measured at each of four approximately equidistant elevations at and below the treeline. Additionally, at the same elevations, needle, branch and stem sapwood tissues were collected to determine NSC concentrations.
  • Overall, growth decreased and NSC concentrations increased with elevation. Along with previous empirical and experimental studies, the findings of this study provide no indication of NSC reduction at the treeline; NSC increased in most species (each represented by one common population) towards their upper climatic limit. The disparity between carbon acquisition and structural carbon investment at low temperature (accumulation of NSC) thus does occur even among genotypes not adapted to treeline environments.


During the last few decades there has been increasing evidence that low temperature during the growing season is responsible for the formation of high-elevation treelines (Tranquillini, 1979; Körner, 1998; Jobbágy & Jackson, 2000; Grace et al., 2002). In low temperature-adapted plants, including altitudinal treeline trees, light-saturated photosynthesis reaches c. 50% of full capacity at +5°C, whereas no tree has ever been shown to exhibit significant growth below 5°C (James et al., 1994; Körner, 2006; Solfjeld & Johnsen, 2006; Alvarez-Uria & Körner, 2007). As a consequence, all plants tested so far show an increment in nonstructural carbohydrate (NSC = soluble sugars + starch) tissue concentrations when exposed to low temperatures. Thus, a carbon supply limitation – as proposed by, for example, Stevens & Fox (1991) and Wardle (1993)– is an unlikely cause for low temperature range limits of plants. Rather, the growth-limitation hypothesis (GLH), which claims that cell and tissue formation are the processes that are first limited by the altitudinal decrease in temperature (Körner, 1998; Grace et al., 2002), seems to work given that a growing number of empirical studies have found no decrease, and most often an increase in NSC concentrations with elevation (e.g. Piper et al., 2006; Shi et al., 2008; Fajardo et al., 2011; Hoch & Körner, 2012). It remains, however, unresolved whether this phenotypic pattern reflects a short-term physiological response driven by stronger limitations on growth relative to carbon gain or, rather, an adaptive response of treeline populations to better cope with the harsh conditions at treeline (Sveinbjörnsson, 2000; Smith et al., 2003). Consequently, NSC accumulation with elevation could be an evolutionary response to resist harsh conditions (Monson et al., 2006), a consequence of genetically limited growth (Premoli et al., 2007), with trees under low-temperature conditions allocating more carbon into storage, or the phenotypic result of a direct reduction in growth in cold environments (Körner, 1998). It is the purpose of this study to determine if the increase in carbon reserves with elevation is an immediate physiological response of trees to lower temperatures or not.

Field surveys assessing the GLH have the inherent disadvantage of being unable to measure and account for probable intraspecific, genetic adaptation (i.e. natural selection) differences along elevation. So far, several studies have demonstrated that variation in many morphological and physiological traits along the altitudinal gradient is under genetic control (Oleksyn et al., 1998; Premoli, 2003; Li et al., 2004; Piper et al., 2006; Premoli et al., 2007; Vitasse et al., 2009). For example, Premoli et al. (2007), working in a common garden, found that saplings of Nothofagus pumilio (a treeline species of the southern Andes) originating from a higher elevation grew slower than saplings from a lower elevation. Thus, the possibility remains that the higher levels of NSC found in most of the treeline studies to date are life history traits of treeline populations (ecotypes) selected by fitness gains: a clinal (genetic) adaptation. It would be of great interest to solve this conundrum, in order to determine whether the GLH is acceptable as a functional explanation for treeline formation, and to assess predictions of how treelines will react to increasing temperature and CO2. The fact that many studies thus far have supported the idea that treeline trees are directly growth-limited by low temperature (environmental effect) suggests that trees would not be responsive to increasing atmospheric CO2 (Körner, 1998), but to increasing global temperature. However, if the increasing mobile carbon charge at treeline is attributable to genetic control, the NSC assessment would fail as a proxy to resolve the dilemma between carbon and sink limitations. This could imply that treeline trees are not necessarily CO2-saturated and that they may be directly responsive to increasing atmospheric CO2 supply (e.g. Handa et al., 2005) and not to increasing temperature.

There are at least two types of study that may assist in distinguishing the cause of the increasing levels of NSC of treeline trees: environmentally controlled experiments, either common garden or reciprocal transplants, and natural experiments. In both cases, the idea is to control for the genetic origin of plants and make temperature vary. In the first type of study, using same-provenance seedlings placed in growth chambers (Hoch & Körner, 2009) or hydroponic installations (Solfjeld & Johnsen, 2006), it was shown that lower mean (constant or variable) temperatures (comparable to the milieu of treeline) led to an increase in NSC, demonstrating a phenotypic component in the seedlings’ response. Alternatively, a reciprocal transplant experiment could also work; in this case, treeline seedlings are brought to lower elevation and after several growing seasons NSC concentrations are assessed. Experiments with seedlings have the inconvenience of being an abstraction of what is actually occurring at treeline. Adult trees are the ones ultimately defining the natural treeline position, being the net outcome of many years of growth and reproduction; genetically controlled responses may occur in seedlings but not in adult trees and vice versa (Premoli & Brewer, 2007). Additionally, the GLH has been proposed to explain treeline formation for tall trees as these are more closely coupled to atmospheric conditions as a consequence of their arboreal architecture, whereas seedlings or saplings growing near the ground decouple from the free atmosphere and therefore generally experience warmer conditions than arborescence, taller trees (Körner, 1998; Körner, 2007). To respond to our objective, the existence of natural experiments with adult, tall trees can be reduced to two cases: permafrost tree populations and afforestations. Hoch (2008) found that trees of Picea abies growing on low elevation permafrost accumulated significantly higher NSC concentrations than their immediate, and probably same-origin, neighbours growing on warmer soil. In relation to alpine treelines, NSC accumulation over permafrost has been interpreted as a consequence of growth limitation and not a direct response of acclimation to low temperature (e.g. cryoprotection), as starch rather than soluble sugars (osmotically active compounds) was largely responsible for the increase in NSC concentrations in dwarfed trees over permafrost. Although permafrost soils experience low-temperature conditions comparable to treeline soils, the alpine treeline milieu is not the same, that is, trees do not experience low air temperatures (Körner & Hoch, 2006). Hence, afforestations established along elevational gradients up to treeline elevation and originating from a single provenance (i.e. the same population) are optimal systems with which to test whether the documented changes in the trees’ carbon budget (NSC) at natural treelines are under phenotypic control.

It was the objective of this study to test the following hypothesis: the variation in NSC concentrations of treeline trees is attributable to an immediate (i.e. merely phenotypic) response of the trees’ carbon balance to low temperature. We investigated mature trees of three conifer species, each deriving from a single provenance, in afforestations along altitudinal gradients from lower sites to the elevation of the natural treeline. Our expectations can be summarized as follows (Fig. 1): if a decrease of temperature with elevation is having a direct effect on the carbon status of treeline trees, then there will be a variation in NSC concentrations when trees from a single population are planted along the elevational gradient up to the treeline. Alternatively, if the observed NSC increase in natural treeline species is related to genotypic (evolutionary) differentiation along the elevational gradient, but absolutely unrelated to the decrease in temperature with elevation, then the carbon status of afforested (single provenance) trees should not change with elevation (Fig. 1). Moreover, these trees may show similar carbon status between lower and higher stands or show even a decreasing trend with elevation, if the unadapted trees run into carbon limitation.

Figure 1.

Expected trends of nonstructural carbohydrate (NSC) concentration variation with elevation (temperature effect) for trees belonging to one population (plantation of a single provenance) according to possible causes: absolute environment or phenotypic control (left) and absolute genetic control (right). If a decrease in temperature with elevation is altering the NSC concentration of trees naturally occurring in treeline ecotones, then there will be a similar variation when a tree species is planted along the altitudinal gradient up to treeline (i.e. NSC concentration varies with elevation; left side). Alternatively, if there were intraspecific, genotypic (evolutionary) differences with elevation in natural treelines, then the NSC concentration of afforested trees should not be altered along the altitudinal gradient between lower and higher stands (right). In more detail, if there is any variation of NSC with elevation (left), such variation can be explained in terms of carbon limitation (carbon-limitation hypothesis (CLH): a reduction in NSC with elevation) or growth (sink) limitation (growth-limitation hypothesis (GLH): an increase in NSC with elevation).

Materials and Methods

Study site and research design

The first study location was within the Cerro Castillo National Reserve (45°59′ S and 71°52′ W, 900–1300 m above sea level (asl); Fig. 2), Coyhaique Province, Patagonia (Chile), where Pinus sylvestris L. was planted in 1970 in an effort to cover slopes with vegetation against the risk of landslides after a catastrophic human-induced forest fire. Here, where the forest was not reached by fire, Nothofagus pumilio is the native tree species at treeline, and the natural treeline occurs at c. 1300 m asl (Fajardo et al., 2011). The afforestation of P. sylvestris was considered feasible and convenient because of its potentially higher growth rates compared with native species. Seedlings of P. sylvestris were produced in a nursery in the region (Tejas Verdes, Coyhaique, Corporación Nacional Forestal, pers. comm.) and seeds came from one provenance (German Forest Service at Wolfgang, Hesse, Germany); seedlings were planted at random, that is, there was not a priori selection for some genotypes to be planted at different elevations. Trees were originally planted in an altitudinal gradient from c. 1000 to 1300 m asl, that is, up to the natural treeline elevation in this region (Fig. 2). Mean annual precipitation for 1992–2007 was 1100 mm at the Villa Cerro Castillo climate station (46°07′ S and 72°09′ W, 340 m asl) located c. 45 km southwest of the study site; most of the precipitation falls as snow from April to October (Informe Meteorológico de Chile, Dirección General de Aguas, 2008). The duration of the growing season at a nearby treeline was found to be 144 d for the period measured between March 2008 and April 2009, with mean temperatures across the growing season of 6.9°C (Hoch & Körner, 2012). The soil is derived from aeolian volcanic ash deposits (Veit & Garleff, 1995).

Figure 2.

Photographs showing (a) afforestation of Pinus sylvestris along with a natural forest of Nothofagus pumilio in the southern Andes (Reserva Cerro Castillo, Chile, 45°59′ S and 71°52′ W, 1000–1300 m above sea level (asl)). Light-green forest corresponds to P. sylvestris; darker green corresponds to N. pumilio, the natural treeline species of the southern Andes; (b) afforestation plot composed of Pinus cembra and Larix decidua in the Swiss Alps (Julier Pass, 46°28′ N and 9°43′ E, 1970–2150 m asl).

The second study location is in the eastern Swiss Alps at the Julier Pass (46°28′ N and 9°43′ E, 1970–2150 m asl), Graubünden (Switzerland) and uses afforestations with Pinus cembra L. and Larix decidua Mill., two native treeline species of the Alps. Small afforestation plots were established in the mid-1970s, and are located alongside the road leading to the Julier Pass in order to assess the afforestation potential on slopes that were previously clear-cut (Fig. 2). Although the exact provenances of the afforested trees here were not recorded, all plots were simultaneously afforested with trees from local provenances (Tiefencastel Forestry District, Graubünden, pers. comm.). Trees were planted in four separated rectangular plots with sides of length 50–100 m (protected with game fences) along an altitudinal gradient from lower elevations to the potential treeline (at 1970, 2065, 2090 and 2150 m asl). In addition to L. decidua and P. cembra, plots also included Pinus mugo Turra, but this species died off in the lower plots and was present at higher abundance at the highest plot only in the sampling year (2008). Thus, P. mugo was not included in the present study. The climate here is temperate with a mean annual precipitation of c. 700 mm (long-term records from a nearby meteorological station in Samedan, c. 12 km east of the Julier Pass; data provided by MeteoSchweiz). Temperatures were logged at the uppermost plot (2160 m) during the year 2008 and revealed a growing season length of 191 d with a growing season mean temperature of 7.0°C.

Field sampling and initial sample processing

As suggested in Hoch & Körner (2012), sampling took place in the second half of the growing season for each site: during early March 2008 for the Patagonia location and during mid-July (wood) and mid-September 2008 (needles) for the Swiss Alps location. In the case of P. sylvestris at the Patagonian site, 12 trees were sampled at four equidistant elevations from treeline elevation down to tall forest (200 m below). At the Swiss site, eight dominant trees per species were sampled at each of the four afforestation plots. Trees of dominant social status were selected for tissue collection at each sampling elevation or afforestation plot. A total of 48 (two slopes) and 64 trees were selected at the Cerro Castillo (Patagonia) and Julier Pass (Swiss Alps) sites, respectively. For each individual tree surveyed, we also measured tree height using a yardstick or, for trees taller than 3 m, a clinometer (Suunto, Vartaa, Finland), and stem diameter (at 135 and 50 cm height in Patagonia and the Swiss Alps, respectively). Tree heights ranged from 2.0 to 24.9 m at Cerro Castillo and from 1.5 to 19.4 m at the Julier Pass, respectively (Table 1).

Table 1.  Mean (SE) growth rate (basal area increment (BAI) for the last 10 yr and height) at different elevations in three species (Pinus sylvestris, Pinus cembra and Larix decidua) planted at two treeline ecotone locations in the southern Andes (Reserva Cerro Castillo, Chile, 45°59′ S and 71°52′ W, 1000–1300 m above sea level (asl)) and the Swiss Alps (Julier Pass, Switzerland, 46°28′ N and 9°43′ E, 1970–2150 m asl)
   Pinus sylvestris    Pinus cembra Larix decidua
  1. Elevations are species-specifics.

  2. Different letters in the same column indicate significantly different BAI or height means among elevations (post hoc Tukey’s honestly significant differences (HSD) test). Numbers in italics represent the percentage change between the lowest elevation and the upper elevation values for a particular growth measure and species.

BAI (cm2)m aslElevation m aslElevation  
10001141.3a (30.1)1970195.0a (10.1)249.8a (28.1)
10952101.61a (13.2)20672120.6a (13.8)112.7b (18.7)
12153109.8ab (17.5)20903119.5a (11.7)92.7b (18.1)
1300435.2b (7.5)2150496.4a (11.0)56.8b (7.8)
   −75.1    1.5 −77.3
Height (m)100017.5a (0.3)197014.1a (0.3)7.1a (0.6)
109525.0b (0.2)206724.5a (0.1)4.9b (0.3)
121533.9c (0.1)209034.7a (0.2)5.4b (0.4)
130041.6d (0.1)215043.7a (0.3)3.3c (0.3)
   −80.1    −9.7 −53.5

The collection of tissue samples was conducted between 10:00 and 17:00 h. Two stem cores were taken with 5.15-mm-diameter increment bores (Haglöf, Långsele, Sweden) at the base of the tree, one c. 10 cm in length for chemical analysis (stem sapwood) and a second to-the-pith long core at c. 30 cm high for growth determination. Additionally, from each sampled tree we collected sun-exposed branch sapwood (with bark and phloem removed in the field using a knife) from 2- to 5-yr-old branch segments, and also the previous season’s needles from that branch. All tissue samples were bagged and labelled. Back in the laboratory, woody samples for NSC analyses were immediately heated in a microwave oven at 600 W for 90 s to denature enzymes. All samples (except those for tree growth determination) were then oven-dried at 70°C for 48 h. All samples used for NSC measurements were then ground to a fine powder and stored at cold temperature until analysis.

Tree growth determination

For each site, cores were prepared following standard dendrochronological techniques (Stokes & Smiley, 1996): cores were dried, mounted and glued firmly in grooved wooden sticks and sanded with successively finer grades of sandpaper until optimal surface resolution allowed annual rings to be distinguished under magnification. The inside-bark bole radius and the annual radial increment for the last 10 yr were measured to the nearest 0.01 mm using a microscope mounted on a dendrochronometer with a Velmex sliding stage (Bloomfield, NY, USA) and an Accurite measuring system (St. Louis, MO, USA). Cross-dating accuracy was checked visually as rings were easily identified. The last 10-yr basal area increment (BAI), which represents a relative measure of gain in growth, was then computed by considering diameter at coring height and bark thickness.

Chemical analysis

NSC concentrations (the sum of the three quantitatively most important low-molecular-weight sugars, that is, glucose, fructose and sucrose, plus starch) were analysed following Wong (1990) as described in detail in Hoch et al. (2002). Approximately 10 mg of plant powder was extracted with 2 ml of distilled water at 100°C for 30 min. An aliquot of the extract was used for the determination of low-molecular-weight carbohydrates after enzymatic conversion of fructose and sucrose to glucose. The concentration of free glucose was determined photometrically after enzymatic conversion of glucose to gluconat-6-phosphate on a 96-well multiplate reader. Following the degradation of starch to glucose by a crude fungal amylase (‘Clarase’ from Aspegillus oryzae; Enzyme Solutions Pty Ltd, Crydon South, Victoria, Australia) at 40°C overnight, the NSC concentration was determined in a separate analysis. The concentration of starch was calculated as NSC minus the sum of free sugars. All concentrations are given as per cent dry matter.

Statistical analysis

The concentrations of soluble sugars were normally distributed so they were analysed untransformed; however, concentrations of starch and NSC, as well as BAI distributions, deviated from normality and were therefore log-transformed before statistical analyses (Kolmogorov–Smirnov test for goodness-of-fit). The influences of elevation and species were analysed with a two-way ANOVA, with elevation and species as fixed factors. Differences among elevations were compared and tested for significance at the P < 0.05 level using a Tukey–Kramer honest significant difference (HSD) test. All statistical tests were performed with jmp 8.0 (SAS Institute, Cary, NC, USA).


Tree growth

Tree growth rate (BAI in cm2 for the last 10 yr, and tree height in m) generally decreased with elevation for all three species (Table 1, Fig. 3). While this decrease was pronounced and highly significant for P. sylvestris (Southern Andes) and L. decidua (Swiss Alps), the difference between the growth rates of the lowest elevation and treeline trees was negligible for P. cembra (Swiss Alps; Table 1, Fig. 3). In particular, BAI and tree height at the treeline were significantly lower than at the lowest elevation for both P. sylvestris and L. decidua (Table 1), with a c. 75% decrease from the lowest to the highest elevation in both species. For P. sylvestris and L. decidua, the decreasing growth pattern becomes even more pronounced when the significant decline in tree height is also considered (Table 1), because this implies markedly reduced aboveground stem volumes along the gradients. For P. cembra, neither tree height nor BAI changed significantly with elevation, and trees located at intermediate elevations tended to have the highest tree height and BAI values, while both parameters tended to be smaller at the highest site (Table 1, Fig. 3).

Figure 3.

Tree growth rate (basal area increment (BAI) in cm2 for the last 10 yr) in relation to elevation for three afforested species (and all species combined) in the southern Andes (Pinus sylvestris: Reserva Cerro Castillo, Chile, 45°59′ S and 71°52′ W, 1000–1300 m above sea level (asl)) and the Swiss Alps (Pinus cembra and Larix decidua: Julier Pass, Switzerland, 46°28′ N and 9°43′ E, 1970–2150 m asl) treeline ecotones. Elevations are species-specifics (see Table 1). Slope values are shown in each panel with statistical significance when different from zero (*, < 0.05; **, < 0.01; ***, < 0.001). Different letters in the same panel indicate significantly different BAI means among elevations (post hoc Tukey’s honestly significant differences (HSD) test).

NSC concentrations

In contrast to tree growth, concentrations of soluble sugars, starch, and NSC in needles, branch and stem sapwood in general did not decrease with elevation (Table 2). In fact, across all three species, a significant increase in NSC for all tissue types was found: needles (= 0.468; P = 0.023), branch (= 0.518; < 0.001), and stem wood (= 0.111; = 0.014) (Tables 2, 3, Fig. 4). The mean increases in NSC from the lowest elevation to treeline for P. sylvestris were 23 and 99% for needles (= 0.647; = 0.019) and branch wood (= 0.269; = 0.002), respectively, with no significant change in stem wood (= 0.019; = 0.762). The increases between the lowest and the highest plots at the Swiss site in P. cembra were 35, 78 and 55% for needles (= 0.101; = 0.003), branch wood (= 0.211; = 0.001) and stem wood (= 0.163; = 0.010), respectively; for L. decidua, only NSC in stem wood changed significantly and positively with elevation (= 0.2; = 0.019), with an increase of 53% from lower to upper elevations (Table 3). For all three species, there were in general higher concentrations of soluble sugars than of starch (Fig. 4) (e.g. see branches of L. decidua at elevations 3 and 4). Starch concentrations were, however, those that increased the most with elevation (Table 3).

Table 2.  Effects of elevation (E), species (S) and their interaction (E × S) on the concentrations of soluble sugars, starch and nonstructural carbohydrates (NSC = sugars + starch) analysed using a two-way ANOVA
F ratio P-value F ratio P-value F ratio P-value
  1. Three species were considered at four equidistant elevations in two treeline ecotone locations in the southern Andes (Pinus sylvestris: Reserva Cerro Castillo, Chile, 45°59′ S and 71°52′ W, 1000–1300 m above sea level (asl)) and the Swiss Alps (Pinus cembra and Larix decidua: Julier Pass, Switzerland, 46°28′ N and 9°43′ E, 1970–2150 m asl).

E × S62.490.0281.650.1403.570.003
E × S62.390.0341.390.2272.090.060
E × S60.730.7263.580.0031.870.094
Table 3.  Mean (SE) tissue concentrations of nonstructural carbohydrates (NSC = soluble sugars + starch) at different elevations in three species (Pinus sylvestris, Pinus cembra and Larix decidua) planted at two treeline ecotone locations in the southern Andes (Reserva Cerro Castillo, Chile, 45°59′ S and 71°52′ W, 1000–1300 m above sea level (asl)) and the Swiss Alps (Julier Pass, Switzerland, 46°28′ N and 9°43′ E, 1970–2150 m asl)
  1. Elevations are species-specifics (see Table 1).

  2. Different letters in the same column indicate significantly different NSC means among elevations (post hoc Tukey’s honestly significant differences (HSD) test). Numbers in italics represent the percentage change between the lowest elevation and the upper elevation values for a particular tissue and species.

Pinus sylvestris
Needles16.8a (0.4)2.0a (0.1)8.8a (0.5)
26.4a (0.4)2.2ab (0.2)8.6ab (0.5)
36.7a (0.2)2.4ab (0.4)9.2b (0.5)
47.6a (0.4)3.1b (0.4)10.7b (0.5)
  11.8 59.6 22.5
Branch11.2a (0.1)0.9ab (0.3)2.2a (0.4)
21.8ab (0.2)0.9a (0.2)2.6ab (0.3)
32.3b (0.2)1.7b (0.3)4.0bc (0.4)
42.3b (0.2)2.0b (0.4)4.3c (0.5)
  87.6 114.9 99.3
Stemwood10.9a (0.1)0.8a (0.1)1.7a (0.1)
20.9a (0.1)0.6a (0.1)1.5a (0.1)
30.9a (0.1)0.7a (0.1)1.7a (0.2)
41.1a (0.1)0.6a (0.1)1.7a (0.2)
  22.7 –23.5 0
Pinus cembra
Needles15.9a (0.2)3.3a (0.2)9.2a (0.2)
26.2ab (0.3)4.6a (0.6)10.8a (0.5)
37.1b (0.3)5.5ab (0.7)12.6ab (0.8)
46.8ab (0.3)5.7b (0.8)12.5b (0.8)
  16.1 68.8 35.3
Branch11.1a (0.1)0.7ab (0.2)1.8a (0.2)
21.3a (0.1)0.6a (0.1)1.9ab (0.1)
31.6ab (0.1)1.3bc (0.1)2.9bc (0.2)
41.9b (0.2)1.4c (0.2)3.3c (0.4)
  68.4 92.5 77.9
Stemwood10.5a (0.1)0.5a (0.1)0.9a (0.1)
20.6b (0.1)0.6b (0.1)1.3ab (0.1)
30.6b (0.1)0.7bc (0.1)1.4b (0.1)
40.7c (0.1)0.8c (0.1)1.5b (0.1)
  46.3 63.9 54.8
Larix decidua
Needles110.1a (0.5)3.0a (0.2)13.1a (0.6)
29.5a (0.6)2.5a (0.3)12.0a (0.7)
39.2a (0.4)3.6a (0.6)12.8a (1.0)
48.4a (0.6)2.8a (0.3)11.3a (0.8)
  –16.7 –6.2 –14.3
Branch11.4a (0.1)0.4a (0.1)1.8a (0.2)
21.7a (0.2)0.6b (0.1)2.3a (0.3)
31.3a (0.1)0.5a (0.1)1.8a (0.1)
41.6a (0.1)0.6b (0.1)2.3a (0.2)
  18.2 66.7 28.2
Stemwood11.0a (0.2)0.3 a (0.1)1.2a (0.3)
21.0a (0.2)0.3a (0.1)1.3a (0.2)
30.9a (0.1)0.6b (0.1)1.4a (0.1)
41.2a (0.2)0.7b (0.1)1.8b (0.2)
  21.9 168.4 52.6
Figure 4.

Concentration of nonstructural carbohydrates (NSC = soluble sugars (grey) + starch (black)), in needles, branch wood and stem sapwood in relation to elevation for three afforested species (and all species combined) in the southern Andes (Pinus sylvestris: Reserva Cerro Castillo, Chile, 45°59′ S and 71°52′ W, 1000–1300 m above sea level (asl)) and the Swiss Alps (Pinus cembra and Larix decidua: Julier Pass, Switzerland, 46°28′ N and 9°43′ E, 1970–2150 m asl) treeline ecotones. Values are mean ± SE for eight to 12 trees. Elevations are species-specifics (see Table 1). Slope values are shown in each panel with statistical significance when different from zero (*, < 0.05; **, < 0.01; ***, < 0.001). Different letters in the same panel indicate significantly different NSC means among elevations (post hoc Tukey’s honestly significant differences (HSD) test).


A phenotypic response to low temperature

Overall, growth decreased and NSC concentration increased with elevation, reaching a peak at treeline. Although some tissues in some species did not show an NSC concentration increase with elevation (e.g. stem wood in P. sylvestris and needles and branches in L. decidua), thus not showing a phenotypic response to lower temperature, we are parsimonious and state that, as afforested, adult trees from a single provenance for each species were used, the finding of at least one species that showed an increase in its NSC concentration with elevation is sufficient evidence to demonstrate that this trend is attributable to a physiological response of trees to lower environmental temperature. Hence, our results support the notion of a direct reduction in tree growth due to low temperature as the cause of the observed NSC accumulation along natural treeline ecotones. We recognize that genetic differentiation may exist and be responsible for morphological or even functional variation (plastic response) along altitudinal gradients (Oleksyn et al., 1998; Premoli, 2003; Piper et al., 2006; Premoli & Brewer, 2007; Premoli et al., 2007; Vitasse et al., 2009); however, genetic variation clearly cannot be exclusively responsible of the increasing accumulation of NSC found here. In this respect, we do not discard the possibility that some local selection is currently going on, particularly at treeline elevations where less apparent survival can be observed, that is, that we sampled a stronger genotype. We believe, however, that our results are consistent and provide additional and definite evidence to support the hypothesis of higher NSC with elevation being a physiological response of trees to lower temperature (see Solfjeld & Johnsen, 2006; Hoch, 2008; Hoch & Körner, 2009). Thus, all the studies so far, including ours, leave little leeway for the explanation of altitudinal NSC accumulation as an evolutionary response.

While temperature-induced NSC accumulation could also be caused by a cold-acclimation response to frost tolerance (Sakai, 1983), we discard this possibility as an explanation for the results obtained here. It is widely known that soluble sugars are important in stress tolerance and natural selection has favoured NSC accumulation in species inhabiting extreme environments such as alpine ecosystems (Monson et al., 2006). However, in agreement with all previous studies supporting the GLH, we found that osmotically inactive starch, and not soluble sugars, was the component responsible for the major variation in NSC with elevation. The trend found of increasing NSC towards treeline thus, cannot be directly related to freezing avoidance or tolerance, although starch can be hydrolysed to provide free sugars for increased freezing resistance in adult trees (Alberdi et al., 1989). However, as frost may occur year-round at temperate treelines and the trees were sampled towards the end of the growing season (when trees should be prepared for frosts), the starch accumulation found in the current study seems to be more a consequence of the increase in carbon reserves following the stronger limitation of growth processes than the consequence of carbon gain. Increased carbon reserve concentrations have been proposed to indicate the increased carbon source–sink imbalance not only at treeline, but also under other environmental stresses that directly limit growth, for example, water limitations associated with ontogeny (Piper & Fajardo, 2011), or drought (Sala et al., 2010; Muller et al., 2011; Piper, 2011).

Support for the growth-limitation hypothesis

To find no reduction of NSC with elevation in the three afforested species despite reduced growth is equivalent to saying that there is no support for carbon limitation as an explanation for treeline formation, but suggests that cold temperatures limit carbon sink activities at higher temperatures than photosynthesis. A higher cold sensitivity of growth processes compared with photoassimilation was previously shown for P. sylvestris trees in Scotland (James et al., 1994). Körner (1998) suggested that cell and tissue formation of treeline adult trees were the processes being limited by low temperature rather than photosynthesis, thus introducing this alternative explanation, and proxy (i.e. the use of NSC), into the treeline theory. We discard, then, the possibility that an insufficient annual carbon balance in trees is occurring at the climatic tree limit, and also the possibility of the existence of a disproportionate reduction of autotrophic relative to heterotrophic tissue with elevation (Körner, 2012). There is one key aspect of this hypothesis that still needs clarification to avoid misunderstandings: the GLH necessarily deals with adult, tall trees (> 2 m), not seedlings, given that they are the ones experiencing the two drivers of NSC accumulation, that is, coupling cooler atmospheric temperatures and experiencing self-shading of their own roots (Alvarez-Uria & Körner, 2007; Körner, 1998, 2007). As seedlings normally decouple from atmospheric temperatures by being near the ground, we state that the use of seedlings as an in situ study model to test the GLH is not adequate and should be ruled out. In this respect, several studies have properly tested the GLH and found support for it in natural treeline ecotones in a variety of environments (e.g. Piper et al., 2006; Shi et al., 2006, 2008; Fajardo et al., 2011; Hoch & Körner, 2012). Our study provides further support for the GLH from a perspective never before adopted.

Leaf habit: are all treeline species similar?

Among the three coniferous species we studied, L. decidua was the species showing the least NSC concentration variation with elevation (slopes did not differ from zero for needles and branches; Fig. 4); only stem sapwood showed slightly (but significantly) higher NSC concentrations at treeline. Although the trend of no variation of NSC with elevation may still support the GLH, given that there is no reduction of NSC with elevation, the deciduous character of L. decidua seems to hold some clues that may expand our knowledge of treeline formation. For instance, deciduous species differ from evergreens in the dynamic of their carbon storage, and it is still unknown whether their NSC dynamic responds similarly to stressors (Hoch et al., 2003). In this respect, the most important piece of evidence against the GLH so far has been obtained in a treeline in situ CO2-enrichment experiment, where Handa et al. (2005) found both support for the carbon-limitation hypothesis (CLH) in L. decidua and support for the GLH in Pinus uncinata. They showed that L. decidua increased its growth as a response to the addition of CO2, that is, it is carbon limited. However, this positive response of growth was found to disappear with time, becoming insignificant after 8 yr of increased CO2 treatment (Dawes et al., 2011). Also, when natural treelines of Larix species were considered, a more consistent increase in NSC was found in L. decidua (Switzerland; Hoch & Körner, 2012) and Larix potaninii (Himalayas; Shi et al., 2008). Thus, it remains uncertain whether the leaf functional type (i.e. deciduous vs evergreen) or any species-specific difference (e.g. L. decidua being an early successional vs P. uncinata being a later successional species) can explain the different responses in this CO2-enrichment experiment.

Conclusions and perspectives

This study provides new and conclusive evidence regarding the carbon supply status of trees at high-elevation treelines. By default, we suggest that trees at the alpine treeline have a direct low-temperature restriction of tissue formation (i.e. an increase of NSC with elevation), thus supporting the GLH, and that this increase in NSC is not necessarily attributable to an ecotypic adaptation (genetic origin) of trees but is certainly attributable to an immediate physiological response of trees to lower temperature. The latter is probably caused by the imbalance between carbon source and sink activities. Thus, our study additionally validates the use of NSC as a proxy to test the GLH. The fact that many studies thus far have supported the idea that treeline trees are directly growth-limited by low temperature (environment) suggests that trees here would be responsive not to increasing atmospheric CO2, but to increasing global temperature (Körner, 2012).

In the light of the findings of the current study, and other observations and natural experiments at treelines that indicate a direct low temperature-driven limitation of growth processes at the upper limit of tree growth, the next step towards elucidation of the physiological mechanisms underlying this growth limitation should be controlled experiments that clarify the thermal limit for meristematic growth in trees and reveal those processes during cell division, elongation and differentiation that are first affected by low temperatures. These investigations might include and combine micro-anatomical (Rossi et al., 2008), biochemical and genetic methods, as already employed in agricultural studies of growth limitation (Muller et al., 2011). Only by understanding the physiology behind tissue formation at alpine treelines will one be able to design functional models for the development of treelines in response to ongoing climatic change.


This work was supported by the Chilean Fondo Nacional de Desarrollo Científico y Tecnológico (FONDECYT) post-doctoral project 3070050 (grant awarded to A.F.). We appreciate the field assistance of Sean Sweeney during part of the fieldwork. F.I.P. is grateful for support received via FONDECYT post-doctoral project 3080057. G.H. is the recipient of funding from European Research Council (ERC) grant 233399 (project TREELIM). We finally thank the Corporación Nacional Forestal (CONAF) for facilitating access to Reserva Cerro Castillo.