|Home | About | Journals | Submit | Contact Us | Français|
The mobile carbon supply to different compartments of a tree is affected by climate, but its impact on cell-wall chemistry and mechanics remains unknown. To understand better the variability in root growth and biomechanics in mountain forests subjected to substrate mass movement, we investigated root chemical and mechanical properties of mature Abies georgei var. smithii (Smith fir) growing at different elevations on the Tibet–Qinghai Plateau.
Thin and fine roots (0·1–4·0 mm in diameter) were sampled at three different elevations (3480, 3900 and 4330 m, the last corresponding to the treeline). Tensile resistance of roots of different diameter classes was measured along with holocellulose and non-structural carbon (NSC) content.
The mean force necessary to break roots in tension decreased significantly with increasing altitude and was attributed to a decrease in holocellulose content. Holocellulose was significantly lower in roots at the treeline (29·5 ± 1·3 %) compared with those at 3480 m (39·1 ± 1·0 %). Roots also differed significantly in NSC, with 35·6 ± 4·1 mg g−1 dry mass of mean total soluble sugars in roots at 3480 m and 18·8 ± 2·1 mg g−1 dry mass in roots at the treeline.
Root mechanical resistance, holocellulose and NSC content all decreased with increasing altitude. Holocellulose is made up principally of cellulose, the biosynthesis of which depends largely on NSC supply. Plants synthesize cellulose when conditions are optimal and NSC is not limiting. Thus, cellulose synthesis in the thin and fine roots measured in our study is probably not a priority in mature trees growing at very high altitudes, where climatic factors will be limiting for growth. Root NSC stocks at the treeline may be depleted through over-demand for carbon supply due to increased fine root production or winter root growth.
In mountainous regions, vegetation may be subjected to abiotic stress through wind loading, snow gliding and various substrate mass movements (Sidle et al., 1985; Sidle and Ochiai, 2006). Soil degradation also occurs as a result of human activity and, in particular, road-building is a major cause of shallow landslides in hilly areas (Norris et al., 2008; Stokes et al., 2010). Vegetation is usually seen as having a positive effect on slope stability, as root systems reinforce soil, thus acting to prevent shallow landslides (Wu, 1976, 2007; Burroughs and Thomas, 1977; Coppin and Richards, 1990; Gray and Sotir, 1996; Roering et al., 2003; Stokes et al., 2009). Although studies have been carried out on how tree root systems react to mechanical stresses in mountain forests, in particular with regard to growth adaptation on steep slopes (Chiatante et al., 2003; see Stokes et al., 2009), little information exists concerning the influence of altitude on different types of stress adaptation (Soethe et al., 2006). As root systems not only provide trees with water and nutrients, but also anchor it to the soil, the effects of elevation should be considered when trying to understand root growth in mountain forests, and the possible consequences for slope stability.
At higher elevations, air and soil temperature and the length of the growing season become increasingly limiting for tree growth. At the treeline, which represents the high altitudinal limit of forest growth (Körner and Paulsen, 2004), non-structural carbon (NSC) in leaves and branches usually increases (Hoch et al., 2002; Hoch and Körner, 2003, 2009; Shi et al., 2008) whereas root NSC decreases, but variability is high between species (Shi et al., 2006; Li et al., 2008). The concentration of NSC represents a measure of carbon supply available for growth. A reduction of NSC concentration indicates either that carbon demand exceeds supply (source limitation) or that both source and sink activity are low. An increase of NSC pools is likely to indicate that photosynthesis exceeds that needed for growth (sink limitation) (Körner, 2003). NSC exists in the form of starch and low-molecular-weight sugars, namely sucrose, glucose and fructose (Hoch et al., 2002). These sugars provide the substrate required for the formation of certain cell-wall components such as cellulose, the foundational polymer in plant cell walls (Amor et al., 1995; Haigler et al., 2001). If increases in tree NSC concentrations occur at high altitudes, it can thus be assumed that cellulose synthesis will also rise due to the increased supply of substrate.
Cellulose is made up of polymer chains linked together by highly resistant hydrogen bonds grouped in a hemicellulose matrix that is optimized to resist failure in tension (Sjostrom, 1993; Delmer and Amor, 1995). Holocellulose content, i.e. both cellulose and hemicellulose, increases with decreasing root diameter in fine (0·1–2·0 mm) and thin roots (2·1–10 mm) of woody plants (for roots of an equivalent age), thus rendering these roots highly resistant in tension (Ifju and Kennedy, 1962; Hathaway and Penny, 1975; Commandeur and Pyles, 1991; Genet et al., 2005; Hales et al., 2009). Root tensile strength is an important parameter contributing to the stability of forested slopes (see Norris et al., 2008). Trees reinforce the soil matrix through their root system, by either increasing soil shear strength (Wu, 1976; Coppin and Richards, 1990), providing structural support, or lowering pore-water pressures in the soil (Coppin and Richards, 1990; Gray and Sotir, 1996). The presence of plant roots results in an increase in apparent cohesion via root fibre reinforcement. Thin and fine roots act in tension during failure on slopes and, if they cross the potential slip surface, provide a major contribution to slope stability (Wu, 1976, 2007; Reubens et al., 2007; Stokes et al., 2009). Variations in root tensile strength have been studied in detail (Burroughs and Thomas, 1977; Operstein and Frydman, 2000; Abernethy and Rutherfurd, 2001; Bischetti et al., 2005, 2009; Genet et al., 2005, 2008) and depend on species and site factors, for example immediate environment, tree age, season, root diameter and orientation along the slope (Gray and Sotir, 1996; Lindström and Rune, 1999; Genet et al., 2005, 2008). The resistance of a root to failure in tension is controlled by cellulose content (Genet et al., 2005; Hales et al., 2009). However, the influence of environmental conditions on cellulose content is not well documented, although Hales et al. (2009) showed that cellulose variability along a slope is probably related to topographical differences in soil water potential. If changes in root cellulose content occur with altitude, we can suppose that modifications will be reflected in tensile strength. Understanding the environmental conditions that control the distribution of cellulose for roots in the same diameter class may therefore allow us to better comprehend the mechanisms that control root reinforcement (Hales et al., 2009). Such information will be useful for engineers studying soil reinforcement and who use these data in models for calculating the stability of a mountain slope (Greenwood, 2006; Kokutse et al., 2006).
This paper describes a field experiment designed to analyse the links between root tensile resistance (which contributes to soil shear resistance), the chemistry of tree roots and how each are affected by environmental changes incurred with increasing altitude on a mountain slope. Our objectives were to (1) characterize the influence of elevation on root tensile resistance and cellulose content and to (2) understand the interactions between these two parameters and root NSC content, taking into consideration environmental influences.
The study sites were located on Sergyemla Mountain, 50 km from the town of Linzhi, in south-east Tibet, China. Mean annual air temperature obtained from 30-year records at a meteorological weather station located at 2900 m is 8·5 °C with mean minimal temperature in January of –0·2 °C and mean maximal temperature in July of 15·5 °C. Average annual precipitation is around 654·1 mm with 75 % of the annual average amount from June to September. At 4320 m, a meterological weather station installed in 2005 (Southeast Tibet Station for Alpine Environment Observation and Research of the Chinese Academy of Sciences) gave mean air temperatures for the year, January and July as 0·6, –6·9 and 8·7 °C, respectively, and mean annual precipitation was 847·8 mm over a 4-year period.
Two plots were selected at altitudes of 3480 and 4330 m (treeline) along the east slope of Sergyemla Mountain. The slope angle was approx. 35 ° at each plot. Tree species and stand characteristics of the plots at 3480 m and at the treeline were collected (Table 1). A third site at 3900 m, situated between the two elevations and with comparable characteristics, was added later, so that we could compare mechanical and chemical properties from roots at an intermediate altitude. Ten trees of Abies georgei var. smithii (Abies georgei) in each site were selected. Abies georgei is dominant in the region, having an elevation range from 3400 to 4400 m along the western and eastern slopes of Sergyemla Mountain. Abies georgei was studied because this species was the only tree species present at the three different altitudes, thus providing an opportunity to investigate altitudinal effects on root characteristics. Trees selected were aged 100–200 years. Mean diameter at breast height (dbh) was 201·6 ± 7·12 mm (s.e.) for trees growing at 3480 m, 191 ± 11·16 mm at 3900 m and 113 ± 8·13 mm for trees growing at the treeline. Thin and fine live roots (within a diameter range 0·1–4·0 mm) were manually excavated to a depth of about 0·6–0·7 m below the soil surface and a distance of 0·5 m from the stem base. Care was taken to avoid any damage to roots during the excavation process. Samples were collected randomly from the root system around each tree in three directions along the slope (upslope, downslope and perpendicular to the slope direction) in order to have representative samples of roots of all order and diameter ranges. Once the roots had been removed from the trees, they were washed, dried rapidly in the open air and stored for less than 1 month in a dry location until laboratory tests were performed.
To compare wood cellulose content in the stem with that in the roots, increment cores of outer sapwood were taken at dbh perpendicular to the slope direction using a Suunto wood corer. Cores were taken from trees growing at altitudes of 4330 and 3480 m only.
To determine soil mechanical properties, cohesion and friction angle, eight soil samples were removed from the upper and lower altitudes by manually pushing cylindrical shear boxes of a known volume (62 mm diameter × 20 mm height) into non-rooted soil at depths of 50 and 300 mm (Genet et al., 2008). Strain-controlled direct shear tests were carried out using a standard Chinese shear testing procedure (Anon., 1996). The undisturbed non-saturated soil samples were removed from the shear boxes and placed in a shear testing device (Nanjing Soil Shear Machine SDJ-1, China). Normal loads of 100, 200, 300 and 400 kPa were applied as weights on consecutive samples. A lateral displacement was applied at 0·8 mm min−1 until failure occurred and the peak shear force was noted.
Total and available nitrogen (N), total and available phosphorous (P) and total carbon (C) were analysed to characterize chemical soil properties of the sites at 3480 and 4330 m. Soil samples were collected from the topsoil after removing the humus layer. Each measurement was repeated three times. Total N % was analysed using HClO4 + H2SO4 digestion and NaOH distilling neutralization analysis and the available N % using FeSO4 + Zn + 20 % NaOH distilled water extraction analysis (ISS-CAS, 1978). Total P % was evaluated by vanado-molybdate colorimetry and available P % by 20 % (NH4)2CO3 distilled water extraction analysis. Total C % was measured with a TOC analyser (TOC-VCPH, Shimadzu, Kyoto, Japan) (ISS-CAS, 1978).
Tensile tests were carried out successfully on 568 root samples from the three different altitudes, using a Universal Testing machine (ADAMEL Lhomargy, Roissy en Brie, France) (see method in Genet et al., 2005, 2008). A load cell with a maximal capacity of 1·0 kN was used to measure the maximum force required to break roots in tension (Fmax). Crosshead speed was kept constant at 2·0 mm min−1 and both force and speed were measured constantly. Tests were considered successful only when specimens failed approximately in the middle of the root so that root rupture was assumed to be related to the applied tensile force only and not due to stress concentration near the jaws. Tensile strength is defined as the maximal force required to break the root divided by the root cross-sectional area (CSA) at the point of breakage. The central diameter of each root was measured before the test with an electronic slide gauge with 0·02 mm accuracy.
To verify that differences observed in cellulose and NSC content were not due to changes in root tissue density (RTD), the mean RTD between the three sites was calculated. Mean RTD was evaluated gravimetrically by dividing the weight of 30 dry root samples from each site by their volume (determined from root length and diameter).
The method used to measure total holocellulose content was based on that developed by Leavitt and Danzer (1993) and consisted of removing as many non-cellulosic compounds as possible from the root material (Genet et al., 2005, 2006). Hemicelluloses, which are polysaccharides linked to the cellulose present in the cell walls, are not separated from the crystalline cellulose. The quantity obtained at the end of the experiment therefore represents both cellulose and hemicelluloses. After removing the bark, each root/stem sample was ground into a fine powder with a vibration mill (Retsch MM 300, Haan, Germany) and poured into a Teflon sachet (no. 11842, pore size 5·0 µm). The first step consisted of eliminating lipids compounds (waxes, oils and resins) from the ground tissue by means of a mixture of toluene 99 %/ethanol 96 % (2 : 1; v/v) followed by 100 % ethanol. The samples were then immersed in distilled water heated to 100 °C for 6 h to remove hydrosoluble molecules. The lignin was finally removed with 700 mL of distilled water, 7·0 g sodium chlorite (NaClO2) and 1·0 mL acetic acid (C2H4O2) heated to 60 °C for 12 h and concentrated by 100 % three times. The percentage of holocellulose was evaluated by calculating the relative difference in the initial and final weight of each sample. Holocellulose content was calculated on a dry matter basis (% d.m.).
Dry matter and sugar concentrations of bulk plant parts were determined after lyophilization. Samples were ground with liquid nitrogen using a ball grinder (Mixer MillMM200, Retsch, www.retsch.com). Soluble sugars were extracted three times from 30-mg samples with 1 mL 80 % ethanol for 30 min at 80 °C, and then centrifuged. Soluble sugars were contained in the supernatant and starch in the sediment. The supernatant was filtered in the presence of polyvinyl polypyrrolidone and activated carbon to eliminate pigments and polyphenols. After evaporation of solutes with a Speedvac (RC 1022 and RCT 90, Jouan SA, Saint Herblain, France), fructose, glucose and sucrose were quantified by high-performance ionic chromatography (HPIC, standard Dionex, http://www.dionex.com/en-us/index.html) with pulsed amperometric detection (HPAE-PAD). The starch contained in the sediment was solubilized with 0·02 m sodium hydroxide at 90 °C for 2 h and then hydrolysed with α-amyloglucosidase at pH 4·2 for 1·5 h. Glucose was quantified as described by Boehringer (1984) with hexokinase and glucose 6-phosphate dehydrogenase, followed by spectrophotometry of NADPH at 340 nm (UV/VIS V-530 spectrophotometer, Jasco Corp., Tokyo, Japan). Starch content was expressed in mg per g dry matter of equivalent glucose.
The normality of all data was tested using Kolmogorov–Smirnov tests and data were log-transformed when they were not normally distributed. Power regressions were carried out initially to evaluate the relationship between Fmax and root diameter. Data were log-transformed, before analysis, to reflect the power relationship in linear regressions. Multiple regressions were tested between Fmax, holocellulose content and root diameter. Analysis of variance (ANOVA) and analysis of covariance (ANCOVA) with root diameter as the covariate were used to detect differences in Fmax, holocellulose content, NSC and RTD between the three field-sites at 3480, 3900 and 4330 m and between different positions around the tree (upslope, downslope and perpendicular to the slope direction). F and probability (P) are presented as results of ANOVA and ANCOVA tests. F was calculated by dividing the factor mean squares (found by dividing the sum of squares by the degrees of freedom) by the error of the mean squares. Significant differences between data analysed with ANOVA and ANCOVA were determined using post-hoc Fisher's least significant difference tests. A t-test was used to detect significant differences in soil mechanical and chemical properties between the treeline and the lowest altitude. Data were analysed with Minitab software 13 and Statistica (StatSoft, Bedford, UK).
No significant differences were found in mean soil cohesion (C) of pure soil and mean friction angle (ϕ) at altitudes of 3480 and 4330 m (Table 2). No significant differences were found in the chemical properties of soil from 3480 and 4330 m (Table 2).
The force necessary to cause failure in roots increased significantly with diameter at all three altitudes (Fig. 1). Power regressions between Fmax and diameter were significant for roots growing at the three different altitudes (Fig. 1). A significant difference between roots from the treeline and the lowest elevation was found (F2,474 = 99·22, P < 0·001, ANCOVA) taking into account root diameter (F1,474 = 1126·33, P < 0·001, ANCOVA), with significantly more force required to cause breakage in roots from 3480 m (Fig. 1). No significant difference was found in Fmax between roots growing at 3480 and 3900 m (F1,295 = 0·39, P = 0·53, ANCOVA) taking into account root diameter (F1,295 = 650·09, P < 0·001, ANCOVA). Mean root tensile strength was 28·47 ± 1·09 MPa for trees growing at 3480 m, 22·26 ± 1·52 MPa for trees growing at 3900 m and 13·46 ± 0·58 MPa for trees growing at 4330 m. Root tensile strength varied significantly along the altitudinal gradient (F2,386 = 73·80, P < 0·001, ANCOVA). Power regressions between root tensile strength and diameter were significant at all altitudes, with tensile strength increasing with decreasing root diameter (Table 3).
No differences in root mechanical properties were found at different positions around the tree (F2,308 = 1·59, P = 0·206, ANCOVA; Fig. 2).
No significant differences in RTD were found between roots growing at the three different altitudes (Table 3) nor in different positions around the tree.
No significant differences in stem holocellulose content were observed between the upper and lower altitudes. Mean holocellulose content in stemwood was 46·8 ± 1·0 % at 3480 m and 44·9 ± 1·3 % at the treeline. Holocellulose content was significantly higher in stemwood than in roots in trees growing at 3480 m (F1,53 = 34·75, P < 0·001, ANOVA) and also at 4330 m (F1,40 = 74·11, P < 0·001, ANOVA).
Holocellulose content was significantly lower in roots at 4330 m than in roots at 3480 and 3900 m (F2,161 = 19·04, P < 0·001, ANCOVA, Table 3) when root diameter was taken into account (F1,161 = 13·04, P < 0·001). A significant relationship existed between log Fmax, log holocellulose content and log root diameter (d) for roots growing at 3480 and 4330 m (Fig. 3), but not at 3900 m. No significant differences in holocellulose content were found in roots from different positions around the trees (F2,60 = 3·78, P > 0·05, ANCOVA).
Across elevations, roots differed significantly in total soluble sugar concentrations per unit of dry matter (Table 4). Glucose, fructose and sucrose content were significantly different between roots from trees growing at 3480 m and those growing at 4330 m (Table 4). Mean total soluble sugars was 35·6 ± 4·1 mg g−1 d.m. in roots at 3480 m and 18·8 ± 2·1 mg g−1 d.m. in roots at 4330 m. No significant differences were found in starch content between roots from trees growing at upper and lower elevations. Values of NSC in roots growing at 3900 m could not be analysed correctly because some decay had occurred in these samples and it was not possible to return immediately to the field site to obtain further samples. Samples need to be taken at the same time period to be comparative.
The force needed to cause root failure in tension, Fmax, decreased significantly with increasing altitude for roots of the same diameter. Although more force was required to cause failure in larger roots, when translated into tensile strength values, thinner roots were significantly stronger in tension. Although not measured in this study, if root number and size were equivalent between the three elevations, the decrease in strength would imply a reduction in root reinforcement at higher altitudes. It is possible that to compensate for the decrease in tensile strength, other modifications may occur within the root system, so that tree anchorage is not compromised. Soethe et al. (2006) investigated root architecture of tree species at three different altitudes between 1900 and 3000 m in tropical montane forests in Ecuador and observed that root system morphology differed between altitudes. The mean ratio of root CSA to tree height, as well as the extent of root asymmetry, decreased significantly when altitude increased. Soethe et al. (2006) also observed that native tree species growing in tropical montane forest possessed a variety of roots traits that improved tree stability, but they did not measure mechanical properties. To determine if a decrease in root resistance to failure in tension with elevation is compensated for by morphological adaptations, further studies of whole root system architecture should be performed. The position of roots within the root system should also be taken into account. In agreement with the results of Khuder et al. (2006), we found no differences in the mechanics of roots from different positions around the root system. However, Abdi et al. (2010) found that downhill roots of Parrotia persica were significantly stronger in tension compared with uphill roots. There appears to be no general rule for tree root growth and structure on sloping ground; variability is governed largely by species and soil type (see Stokes et al., 2009).
It is possible that modifications in root mechanical resistance with altitude were due to changes in soil chemical and physical properties. Although no difference was found in soil chemical properties and in cohesion and friction angle between 3480 and 4330 m, it was difficult to reach any conclusions due to the small number of samples. Nevertheless, Ba (2008) showed that different levels of soil compaction had no significant influence on the tensile strength of roots of Acacia senegal and Prosopis juliflora. However, Goodman and Ennos (1999) showed that Zea mays possessed stiffer roots when growing in soil with a low bulk density compared with soil with a higher bulk density, but that no differences occurred in sunflower (Helianthus annuus) roots in the same homogeneous, agricultural soils. Therefore, to verify whether soil physical properties influence root mechanical properties, it would be necessary to carry out further detailed experiments on several species in different types of soils.
The power law relationship observed between root strength and diameter was similar to that found in previous studies (Burroughs and Thomas, 1977; Gray and Sotir, 1996; Nilaweera and Nutalaya, 1999; Operstein and Frydman, 2000; Genet et al., 2005, 2008). Genet et al. (2005) showed that the major determinant governing root tensile strength was a modification in cellulose content, which is greater in thinner roots. In the present study, we showed that the negative relationship between Fmax and holocellulose content was highly significant (taking into account root diameter, Fig. 3). Cellulose comprises polymer chains consisting of glucose units grouped together in a hemicellulose matrix (Delmer and Amor, 1995). The structure of cellulose has been found to be optimal for resisting failure in tension (Sjostrom, 1993). However, variability was high in our data, which may be related to root age. It is extremely difficult to determine root age: although young roots are thin, older roots can also have a small diameter, but may be more lignified (Stokes et al., 2009). Other parameters also influence variations in root tensile strength, such as lignin content and microfibril angle (Genet et al., 2005). However, no detailed studies have yet been carried out to examine these phenomena.
The quantity of holocellulose differed significantly for roots from trees growing at 3480 and 3900 m compared with those at 4330 m, with roots at the treeline possessing significantly less cellulose per dry mass. Cellulose biosynthesis depends largely on free sugars (sucrose or glucose and fructose), which probably make up the substrate required for the cellulose biosynthesis pathway (Amor et al., 1995; Haigler et al., 2001). Haigler et al. (2001) suggest that cellulose biosynthesis in an established plant is optional at any given moment. When assimilated carbon is limited under stress, evolutionarily successful plants must partition carbon preferentially towards survival or starch reserves and not cellulose synthesis. Plants therefore synthesize cellulose when conditions are optimal, especially the large amount of cellulose required for cell walls. Cellulose synthesis in the thin and fine roots measured in our study is therefore probably not a priority in mature trees growing at the treeline.
Root cellulose content can also be influenced by environmental conditions; for example, Hales et al. (2009) observed that in the southern Appalachian mountains, cellulose content and tensile force of roots from Quercus rubra and Liriodendron tulipifera varied as a function of root diameter and slope topography. The increase in cellulose content in roots on slope noses compared with roots from hollows was explained by an increase in vegetation density and leaf area index associated with increased soil moisture, suggesting a relationship between slope hydrology and the hydraulic architecture of root tissue. We did not measure soil water content over time in our study, although mean annual precipitation was 194 mm higher at the treeline and vegetation density was lower. Slope hydrology can influence the density of wood in tree stems (Barij et al., 2007), but root tissue density in our study was not different between elevations. Thus, it seems likely that another climatic factor, such as temperature changes associated with increasing elevation, affects root chemistry. It is unlikely that there is a direct effect of air temperature on root cellulose, but soil temperature might affect root cell-wall chemistry, as changes in root morphology have been observed when soil temperature was varied (Lyr and Hoffman, 1967). The link between temperature and cellulose synthesis in plants has yet to be elucidated, although Roberts et al. (1992) showed that in cultured cotton (Gossypium hirsutum) ovules, the optimal temperature for synthesis was 28 °C. The use of roots would be a good model for studying cellulose synthesis at low temperatures, as roots can continue growing at soil temperatures approaching 0 °C (Lyr and Hoffman, 1967).
Gindl et al. (2001) observed a higher lignin content in stemwood of Norway spruce (Picea abies) as elevation increased, and we can therefore assume that cellulose content decreased. The authors supposed that the increase helped maintain the mechanical integrity of the xylem. We observed no differences in stem cellulose content with altitude. Beadle et al. (1996) showed that a significant decrease in kraft pulp yield occurred with increasing elevation for Eucalyptus globules and Eucalyptus nitens, but offered no explanation. Richardson (2004) observed a sharp and significant decrease of lignin and cellulose with increasing elevation in foliar concentrations of Picea rubens and Abies balsamea growing at alpine treelines in the north-eastern USA. Carbon limitation, due to the shorter season and possibly also reduced rates of photosynthesis, may further contribute to the reduced concentrations of expensive compounds (in terms of carbon) at high elevations (Richardson, 2004), although our results suggest that whilst this phenomenon was found in roots, it did not occur in mature stems of A. georgei.
To test if carbon limitation in roots could explain the decrease of cellulose content (and thus root tensile strength) observed at the treeline, NSC of A. georgei roots was analysed. NSC content represents locally available carbon for metabolic activities or carbon source strength (Guo et al., 2004). These mobile compounds represent carbon reserves used when the plant's internal demand for carbon is higher than the supply by current photosynthesis. An increase of NSC in roots would signal a surplus in assimilate accumulation, whereas a decrease of these reserves would indicate either that carbon demand exceeds supply or that both source and sink activity are low (Körner, 2003).
Here we observed that root NSC decreased significantly at the treeline. Similarly, in roots of Abies fabri from the Gongga Mountains on the eastern edge of the Tibet–Qinghai Plateau, Li et al. (2008) found a significant decrease in NSC at the treeline (3670 m) compared with roots from 2750 m. A decrease of sugar and starch in roots with increasing altitude was also measured by Mooney and Billings (1965) in Calyptridium umbellatum roots growing at 1829, 2134 and 2743 m. A significant linear decrease in total NSC from Artemisia tridentata roots occurred from the bottom to the top of a slope in south central Wyoming, USA, at an elevation of 2400 m (Sturges and Trlica, 1978). Our results are therefore in accordance with these different studies, as free sugars were less abundant in roots of A. georgei at the treeline compared with roots growing at 3480 m. The decrease in cellulose content of the same roots could therefore be related to carbon limitation in root compartments or a lower metabolic activity.
The decrease in NSC in roots that we observed at the treeline might be explained by an over demand of carbon at the root system level. The changes in NSC content with increasing elevation between above- and below-ground compartments could be compared with the opposite trend observed in above- and below-ground biomass (Kitayama and Aiba, 2002; Leuschner et al., 2007; Hertel and Wesche, 2008). Hertel and Wesche (2008) observed a contrasting tendency in above- and below-ground structure of Polylepsis stands along an elevational transect between 3650 and 4050 m in the eastern cordillera of Bolivia. The authors identified a decrease in tree height, stem diameter and stand basal area with increasing elevation. Carbon allocation to above-ground tree compartments was reduced, whereas the biomass and surface area of the fine roots increased markedly towards the treeline, thus requiring more resources for growth. The results are similar to the findings of Leuschner et al. (2007) for a tropical montane forest in southern Ecuador, where tree height decreased linearly with elevation by 5·2 m km−1, whereas fine root biomass increased by a factor >4 from 1050 to 3060 m, and the same trend was also observed for coarse roots. A similar observation was made by Kitayama and Aiba (2002), who noted a linear increase in fine root biomass along a transect between 650 and 3080 m in a rain forest stand at Mt Kinabalu, Malaysia. We did not measure root biomass, but as NSC decreased with increasing elevation, it is possible that resources were used for the production of fine roots.
The increase in carbon investment to the root system at high altitudes could be caused partially by a reduced nutrient supply, probably due to lower temperatures rather than water shortage (Leuschner et al., 2007; Hertel and Wesche, 2008). Another explanation for the shift in carbon investment towards the root system with increasing elevation could be found in purely physical reasons (Stevens and Fox, 1991). Under cold growth conditions the long-distance transport of water and nutrients in trees requires a disproportionally large carbon investment in the root system compared with warmer environments (Stevens and Fox, 1991). The increase in carbon investment in structural growth in below-ground parts of trees could also explain why NSC decreases with altitude, as the demand for carbon exceeds supply.
A second hypothesis for the observed decrease in root NSC is that root growth of cold-adapted tree species continues during the winter months (Crider, 1928) if soil temperatures do not drop below freezing (Lyr and Hoffman, 1967). Therefore, at the treeline, active roots may utilize stocks of NSC, which are not replenished as quickly during the summer months or to the same extent as at lower altitudes, where shoot growth starts earlier. Regular sampling of root NSC throughout the year would elucidate differences in NSC dynamics and allow us to understand better the flow of NSC between different compartments in the tree at different elevations. Using functional and structural growth models incorporating tree architecture with such data would allow us to determine the evolution of source supply and demand in time and space (Fourcaud et al., 2008).
Funding was from a LIAMA-CASIA seed project and a Bourse Dufrenoy (French Academy of Agriculture). Thanks are due to students and researchers from ITP-CAS who helped with fieldwork. AMAP (Botany and Computational Plant Architecture) is a joint research unit which associates CIRAD (UMR51), CNRS (UMR5120), INRA (UMR931), IRD (2M123) and Montpellier 2 University (UM27) (http://amap.cirad.fr/).