Little change in the fir tree-line position on the southeastern Tibetan Plateau after 200 years of warming


  • Eryuan Liang,

    1. Key Laboratory of Tibetan Environment Changes and Land Surface Processes, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, PO Box 2871, Beijing 100085, China
    Search for more papers by this author
  • Yafeng Wang,

    1. Key Laboratory of Tibetan Environment Changes and Land Surface Processes, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, PO Box 2871, Beijing 100085, China
    2. Graduate University of Chinese Academy of Sciences, Beijing, 100049, China
    Search for more papers by this author
  • Dieter Eckstein,

    1. Department of Wood Science, University of Hamburg, 21031 Hamburg, Germany
    Search for more papers by this author
  • Tianxiang Luo

    1. Key Laboratory of Tibetan Environment Changes and Land Surface Processes, Institute of Tibetan Plateau Research, Chinese Academy of Sciences, PO Box 2871, Beijing 100085, China
    Search for more papers by this author

Author for correspondence:
Eryuan Liang
Tel: +86 10 62849380


  • As one of the world’s highest natural tree lines, the Smith fir (Abies georgei var. smithii) tree line on the southeastern Tibetan Plateau is expected to vary as a function of climate warming. However, the spatial patterns and dynamics of the Smith fir tree line are not yet well understood.
  • Three rectangular plots (30 m × 150 m) were established in the natural alpine tree-line ecotone on two north-facing (Plot N1, 4390 m asl; Plot N2, 4380 m asl) and one east-facing (Plot E1, 4360 m asl) slope. Dendroecological methods were used to monitor the tree-line patterns and dynamics over a 50-yr interval.
  • The three study plots showed a similar pattern of regeneration dynamics, characterized by increased recruitment after the 1950s and an abrupt increase in the 1970s. Smith fir recruitment was significantly positively correlated with both summer and winter temperatures. However, Smith fir tree lines do not show a significant upward movement, despite warming on the Tibetan Plateau.
  • The warming in the past 200 yr is already having a significant impact on the population density of the trees, but not on the position of the Smith fir tree line.


The high-altitude limit of forests, commonly referred to as the tree line, timberline or forest line, represents one of the most obvious vegetation boundaries (see a schematic representation in Fig. 1 by Körner & Paulsen, 2004). The tree-line ecotone refers to the transition from the timberline to the treeless alpine vegetation (Körner, 2003). Given the importance of temperature in controlling the elevation of alpine tree lines and in constraining tree regeneration and growth, tree lines are likely to respond quickly to climate change, showing changes in structure and position (Körner, 2003; Holtmeier & Broll, 2007; Harsch et al., 2009).

Despite the complex ecology of their dynamics, polar and high-elevation tree lines are generally expected to advance in response to global warming (Jobbagy & Jackson, 2000; Kullman, 2001; Malanson, 2001; Grace et al., 2002; Liu et al., 2002; Dai et al., 2005; Holtmeier & Broll, 2007; Payette, 2007; Harsch et al., 2009). The easily distinguishable transition between forest and tundra ecosystems allows for the measurement of shifts in tree-line location (Körner, 2003). The use of dendroecological methods can provide a rich source of data on tree-line dynamics reaching back for several centuries (Payette & Filion, 1985; Lloyd & Fastie, 2003; Camarero & Gutiérrez, 2004). For example, a warm period in the early Holocene has been associated with the expansion of the tree line beyond its current position in northern Eurasia (MacDonald et al., 2000). A growing body of evidence, accumulated over the past few decades, has also revealed that the tree line is moving upslope in the Swedish Scandes, North America and Eurasia (Kullman, 2001; Lloyd & Fastie, 2003; Esper & Schweingruber, 2004; Payette, 2007). In a global database, comprising 166 sites for which tree-line dynamics have been recorded since ad 1900 (Harsch et al., 2009), no tree lines were retreating, but tree-line advances were recorded at 52% of sites. In addition, diffuse tree lines were responding to both winter and summer warming, and abrupt or Krumholtz tree lines were responding to winter warming. Upward shifts in tree lines are often interpreted as evidence of global warming.

On the southeastern Tibetan Plateau, alpine tree lines are among the highest worldwide (Li, 1985; Miehe et al., 2007; Opgenoorth et al., 2010), but their dynamics have not been studied in detail (Luo et al., 2005; Schickhoff, 2005; Li et al., 2008; Shi et al., 2008). Recent studies have suggested that warming has caused unprecedented glacial retreat on the Tibetan Plateau (Yao et al., 2004). There is also some photographic evidence of a rapid tree-line advance in accordance with recent glacial retreat in the Hengduan Mountains on the southeastern Tibetan Plateau (Baker & Moseley, 2007). In spite of the great potential of this ecosystem for monitoring the effects of climate warming, the patterns and dynamics of tree lines on the Tibetan Plateau have received little attention. Considering the negligible human influence, the Smith fir tree line in the Sygera Mountains provides a model for the detection of fluctuations in natural alpine tree lines on the Tibetan Plateau.

Therefore, the objective of this study was to examine the spatio-temporal dynamics of natural Smith fir tree lines in response to climate change over the past 400 yr in the Sygera Mountains on the southeastern Tibetan Plateau. We hypothesized that the upslope movement of Smith fir diffuse tree lines in the Sygera Mountains would be expected to be proportional to both winter and summer warming (as reported by Harsch et al., 2009), and that trees would be expected to grow more densely in response to warming. In order to detect a distinct tree-line upslope movement signal, the upslope regeneration (trees with a height ≥ 2 m) beyond the height of the tree canopy (c. 10 m in the three research plots) served as a minimum movement criterion.

Materials and Methods

Study area and climate

The study area is located in the Sygera Mountains (29°10′–30°15′N, 93°12′–95°35′E) on the northwestern side of the Great Canyon of the Yarlung Zangbo River in southeastern Tibet (Fig. 1).

Figure 1.

Map showing the location of the study sites in the Sygera Mountains (E1, N1 and N2) and of the temperature proxy sites on the Tibetan Plateau. Numbers 1–4 represent the Dunde, Guliya, Puruogangri and Dasuopu glaciers, respectively.

Southeastern Tibet is characterized by a semi-humid climate. The south Asian monsoon reaches the Sygera Mountains through the Yarlung Zangbo River valley, resulting in ample summer rainfall. Records from the meteorological station in Nyingchi (Linzhi) (29°34′N, 94°28′E, 3000 m asl) (range of 18–25 km from the three plots) show an average annual precipitation (1960–2009) of 672.7 mm, 71.8% of which is brought by the summer monsoon, prevailing from June to September. July (mean temperature of 15.8°C) and January (0.5°C) are the warmest and coldest months, respectively. As recorded by an automatic weather station at the tree line (4400 m asl), the mean annual temperatures were 0.56, 0.03 and 0.83°C in 2007, 2008 and 2009, respectively, and the annual precipitation values were 882, 960 and 754 mm, respectively. Snow of 50–100 cm in depth occurs from November to May. The mean daily temperature ranged from −15 to 10°C from November 2006 to May 2010, and soil temperature varied from −5.1 to 11.7°C (June 2007–May 2010) at 10 cm depth.

The temperatures in southeastern Tibet have shown a clear warming trend in both annual and seasonal means over recent decades (Liang et al., 2009). As shown for the meteorological station in Nyingchi, the regional mean and minimum temperatures in all seasons have increased significantly since 1960 (Liang et al., 2009); the increase in maximum temperature is significant only in spring (March–May) and winter (December–February). In addition, annual and summer precipitation values have also increased in recent decades.

Tree species and tree line

Smith fir, with a pencil-shaped crown, mainly grows on shady or semi-shady slopes. Pollination occurs in May and seeds become mature in October of the current year. The seed cones are sessile, black at maturity and of an ovoid-cylindrical shape (4.8–7.8 cm in length and 2.6–3.9 cm in diameter) based on a random selection of 30 cones in October 2010. According to our weekly wood formation study at the tree line at E1 in 2007, cambial cell division started in mid-May and ended in mid-August, whereas cell wall thickening continued to the middle or end of September. The oldest individual is c. 400 yr old (Liang et al., 2009, 2010). For trees with an average age of 180 yr, the mean annual ring width is c. 1.40 mm (Liang et al., 2010). A successful recruitment is partly related to forest gaps resulting from the breakdown of rotten stems (Luo et al., 2002).

Smith fir grows from the river valley (c. 3600 m asl) to the upper tree line, which varies from 4250 m (south-facing slopes) to 4400 m (north-facing slopes) asl depending on the topography and site exposure. The tree height diminishes from c. 30 m at 3600–3850 m asl to c. 10 m at the timberline (the maximum tree heights at N1 and E1 are 10 and 12 m, respectively). The stand density decreases with increasing altitude. There are no stumps or dead trees above the current tree line. It is a typical diffuse tree line, without any flagging of trees or Krumholtz formation.

Along the altitudinal gradient, mixed forests of Smith fir, Quercus aquifolioides and Picea likiangensis var. linzhiensis are found at c. 3600 m asl, whereas, from 3700 to 4000 m asl, mixed forests of Smith fir and Lonicera spp. are prevalent. Smith fir and lichen (Actinothuidium spp.) communities dominate from 4000 to 4320 m asl, whereas Smith fir and Rhododendron spp. communities occur from 4320 to 4400 m asl, and alpine shrubs (Rhododendron spp.) and meadows are found above 4400 m asl (Ren et al., 2007); above the tree line, rhododendron shrubs, 2–3 m high, cover c. 80% of the area.

At the tree-line ecotone, the soil is podzolic (an organic mat over a gray leached layer) with an average pH value of 4.5. With a denser vegetation cover and damp air, the organic matter on the surface increases and the leaching action intensifies, resulting in an eluviation of most of the iron and aluminum and an accumulation of silicon (thickness of c. 30 cm) (Fang, 1997). The soil is generally covered by organic matter (c. 5 cm) and moss (from 1.0 to 8.2 cm in thickness).

The Smith fir tree-line sites selected in the Sygera Mountains have not been disturbed by animals or humans (e.g. grazing or logging), probably because of their remoteness and the low human population density. In addition, the forest understory at the three sites is dominated by rhododendron shrubs. No grasses are available for yak grazing. Moreover, we did not find any evidence of disturbance by insects or fire (Liang et al., 2009). The three sites are located away from avalanche paths and major rocky outcrops, and the plant community has the potential to shift without being constrained by major terrain features, such as talus or cliffs. Thus, the Smith fir tree lines in the Sygera Mountains are highly representative of natural tree lines on the southeastern Tibetan Plateau.

Research plots and sampling

Three sites, one on an east-facing slope (Plot E1: slope angle 13°, 29°39.468′N, 94°42.596′E, 4360 m asl) and two on north-facing slopes (Plot N1: slope angle 10°, 29°37.918′N, 94°42.136′E, 4420 m asl; Plot N2: slope angle 15°, 29°38.47′N, 94°42.462′E, 4380 m asl), were selected. At each site, a rectangular plot (30 m × 150 m) was placed on a topographically uniform area of the alpine tree-line ecotone to include the timberline and the upper species’ limit of Smith fir. The longer side (y-axis) of each plot was parallel to the altitudinal gradient of subalpine forest to alpine shrub land. For the three plots, the point (x, y) = (0, 0) was located in the bottom left corner. The altitudes of the lower and upper parts of the plots were determined by GPS. The location of each Smith fir individual within the plots was mapped on the xy-plane; the coordinates for a given tree locate the centre of the main stem to the nearest 0.1 m. The following measurements were made on all Smith fir trees; diameter at breast height (DBH) (1.3 m), tree height and diameter of tree’s canopy along the x- and y-axes. Tree height was determined by a measuring stick if the tree was ≤ 2 m, and with a clinometer if the tree was > 2 m. Field work was conducted in autumn 2010.

The age structure of the tree populations was identified by a number of methods. In order to obtain a core through the pith of the tree (or as close as possible to the pith), Smith fir trees with DBH > 5 cm were cored at breast height using an increment borer. The age of saplings and seedlings (height < 2.5 m and DBH ≤ 5 cm) was nondestructively determined in the field by counting the terminal bud scars (internodes or branch whorls) along the main stem (according to Camarero & Gutiérrez, 2004; Batllori & Gutiérrez, 2008). Just outside of the three plots, we collected five seedlings with their roots and counted the number of rings in the root collar zone in the laboratory. Seedling ages obtained from the root collars were compared with the ages obtained from counting the internodes in the field; counting the internodes may underestimate the true seedling age by up to 4 yr.

In order to obtain a more accurate estimate of the age of Smith fir trees in the two height classes of 1.3–2 m and ≥ 2 m, we determined the age of 20 individuals (located both inside and next to the plots) by counting the internodes. We found that Smith fir seedlings took 29 ± 5 yr (standard deviation) (at E1) and 31 ± 7 yr (at N1 and N2) to reach 1.3 m in height, and 33 ± 5 yr and 34 ± 5 yr to attain a height of 2 m, respectively. As performed in other studies (Camarero & Gutiérrez, 2004), we assumed that the time required for a seedling to reach breast height, or 2 m, was statistically the same at the three sites.

Linear regression was used to characterize the relationship between tree age, tree height and DBH; dead trees or trees with a rotten centre were excluded from this analysis. The age of trees with a rotten centre was instead estimated on the basis of the tree age: DBH relationship. We found one dead tree at N1 and eight at E1, all located at low elevation; from these trees, increment cores of the outer part of the stem or disc were collected and the year of tree death was identified by cross-dating. Then, their age was estimated using the relationship between tree age and DBH.

Identification of the tree line and current timberline

The location of the tree line was identified by the presence of upright trees with a minimum height of 2 m at the highest altitude (Kullman, 2001; Holtmeier, 2003; Camarero & Gutiérrez, 2004). A change in the altitudinal position of the alpine tree-line ecotone includes changes in stand density, tree growth form, tree height, DBH and tree location. The simplest descriptor of an upward shift of an alpine tree-line ecotone is the change in elevation at which the highest (altitude) tree is found. The location of the tree line was determined over a 50-yr interval. On this timescale, the 5-yr error associated with age estimation for 2-m tall trees is negligible. Hence, we proceeded to document the variability in the location of the upper tree line over the past 400 yr in each study plot. The germination date of large trees was estimated using the tree ages derived from increment cores (taken at 1.3 m) plus 29 or 31 yr for trees at E1 and N1/N2, respectively.

The timberline (location of the highest closed forest) refers to the maximum elevation at which the coverage with trees > 5 m tall amounts to at least 30% (Holtmeier, 2003) (along an area of 30 m (x) × 10 m (y)). The canopy coverage was calculated as π(d1/2 + d2/2)/2, where d1 and d2 are the diameters of the canopy measured along the x- and y-axes, respectively.

Recruitment and tree-line dynamics

The recruitment dynamics of Smith fir (> 3 yr old) were reconstructed using the age structure of the community for each plot. The reconstruction of stand densities enables a comprehensive view of the change in stand dynamics through time. A large pool of 1–3-yr-old seedlings was not included because most cannot survive. Temporal recruitment patterns were compared between the sites using the two-sample Kolmogorov–Smirnov test. Abrupt recruitment increases were defined when the tree abundance increased by at least 50% when compared with the preceding age class.

Temperature is often a critical factor in controlling tree species’ recruitment and the tree-line position (Holtmeier & Broll, 2007; Harsch et al., 2009; Smith et al., 2009). The meteorological records in Nyingchi (taken since 1960) do not date back far enough to evaluate the association between tree-line dynamics and climate variables. To date, tree-ring-based temperature reconstructions (over the past 400 yr) at high (Bräuning & Mantwill, 2004; Liang et al., 2009) and low (Yang et al., 2010) elevations in Nyingchi cannot capture century-scale temperature variability. However, temperature proxies from tree-ring chronologies on the northeastern Tibetan Plateau (Liu et al., 2005; Gou et al., 2008; Zhu et al., 2008; Liu et al., 2009) and from ice core 18O series (an indicator for summer temperature) obtained from the glaciers of Dunde (northern Tibetan Plateau), Guliya (west), Puruogangri (middle) and Dasuopu (south) (Yao et al., 1996; Thompson et al., 2006) provide a general picture of temperature variability over the past thousand years (Fig. 2). These temperature proxies show a general agreement, characterized by a continuous warming trend since around the 1820s. The tree-line position and recruitment dynamics over the past 400 yr were evaluated with respect to these temperature proxies, documenting temperature variability on a coarse scale.

Figure 2.

Reconstructed mean temperatures from the previous September to the current April for the Wulan area (Zhu et al., 2008) (a), standardized December–April temperature-sensitive tree-ring width index in the Qilian Mountains of the northeastern Tibetan Plateau (Liu et al., 2005) (b) and integrated ice core 18O series (an indicator of summer temperature) from the Dunde, Guliya, Puruogangri and Dasuopu glaciers on the Tibetan Plateau (Thompson et al., 2006) (c). The thick gray curves represent the 100-yr low-pass-filtered values and the horizontal lines represent the long-term means. Z-score is calculated by dividing the score deviation (data of each year – mean value of the series) by the standard deviation of the series.

The average September–April temperature (SAT) recorded at Nyingchi shows a higher correlation with the Climate Research Unit (CRU)-gridded SAT (the climatic variable in Fig. 2a) in Wulan (r = 0.64, P < 0.001) (Zhu et al., 2008) than with the SAT (the climatic variable in Fig. 2b) recorded from 1961 to 2002 at Zhangye (r = 0.42, P < 0.05) (Liu et al., 2005) on the northeastern Tibetan Plateau. Therefore, the tree-ring-based winter half-year temperature reconstruction by Zhu et al. (2008) (Fig. 2a) is used to represent the long-term temperature situation in Nyingchi. The integrated ice core 18O series (Fig. 2c) is considered to be a large-scale indicator of summer temperature for the Tibetan Plateau.


Age structure of the Smith fir forests

Despite topographical differences, the temporal pattern of recruitment was not significantly different (at the 0.05 level) among the three study sites, as indicated by the two-sample Kolmogorov–Smirnov tests, but showed a continuous reproduction and a high correlation of the decadal recruitment rates between E1 and N1 from 1751 onwards (r = 0.92, P < 0.001, n = 26), and between E1 and N2 (r = 0.97, P < 0.001, n = 20) and between N1 and N2 (r = 0.99, P < 0.001, n = 20) from 1801 onwards (Fig. 3). The oldest Smith fir trees (in 2010) were 429, 353 and 210 yr at E1, N1 and N2, respectively (Fig. 4). There were more trees > 100 yr of age at E1 (59 trees) than at N1 (40 trees) and N2 (18 trees). Before the 1950s, the establishment of Smith fir trees occurred at a low and variable rate.

Figure 3.

Temporal (10-yr interval) dynamics of Smith fir regeneration within the three research plots (E1, white bars; N1, gray bars; N2, black bars) in the Sygera Mountains and decadal variation in winter half-year temperature (solid line; see Fig. 2a).

Figure 4.

Spatio-temporal variability in tree density and tree-line position (maximum elevation with live individuals with stems at least 2 m high) at the three Smith fir alpine ecotones (a) E1, (b) N1, (c) N2 in the Sygera Mountains, southeastern Tibetan Plateau. Each closed symbol represents an individual that was established during the period indicated at the top; open symbols represent trees established during periods previous to that indicated at the top. Different symbols correspond to different tree establishment periods (e.g. triangles = 1661–1710). The current timberline (forest cover ≥ 30%) is indicated by a hatched rectangle (30 m × 10 m).

Trends in the age structure of the tree populations were similar for the three plots; most striking was the dominance of young trees (established after 1960); they accounted for 81.4%, 84.0% and 81.95% at E1, N1 and N2, respectively (Figs 3, 4). The age structure approximately follows a reverse J-shape. This densification trend does not occur only inside the ecotone, but also above the current tree line. An abrupt increase in tree recruitment took place in the 1970s, and the recruitment rate has continued to accelerate since then. In addition, there are c. 120–200 seedlings (1–3 yr old) in the three plots. They were not included in Figs 3 and 4, as it cannot be foreseen how many of them will survive in subsequent years.

Tree height and DBH were correlated significantly with tree age; a strong correlation was also found between tree height and DBH (Fig. 5).

Figure 5.

Relationship between Smith fir age, height and diameter at breast height (DBH) in the E1, N1 and N2 plots in the Sygera Mountains.

Tree-line dynamics

The age structure of the tree populations, measured in 50-yr intervals, depicts the tree-line dynamics over the past 400 yr. The two timberlines at the north-facing slopes (N1 and N2) are located in a rather similar elevation, higher than that at the east-facing slope (E1). There was only one tree higher than 2 m from 1611 to 1660 at E1, and at N1 from 1661 to 1710. From 1711 onwards, the tree lines in the three plots only showed a slight movement or were stable for most of the time (Fig. 4).

At E1, the tree-line position was stable from 1761 onwards, but advanced by 3.3 m between 1711 and 1760, and by 9.7 m between 1611 and 1660 (Fig. 4). Considering that a 10-m upslope movement is a minimum criterion for the identification of a significant advancement, the tree-line position at E1 has not changed significantly in the past 400 yr. We should also keep in mind that, from 1611 to 1710, there were only one to three trees growing in the plot, so that the tree-line position cannot be robustly identified.

At N1, the tree line advanced by 16.8 m from 1761 to 1810, when compared with the interval from 1611 to 1710, displaying a significant upslope movement. However, it should be considered that this change referred to only one old tree established before 1611. In addition, altogether, there were only two trees between 1611 and 1710, precluding a confident analysis for a tree-line shift. From 1711 to the present time, the tree-line position is considered not to have changed significantly given a total upslope movement of 3.6 m.

At N2, there was only one tree from 1761 to 1810. In comparison with the period 1811–1910, the tree line moved 9.2 m from 1911 onwards, not showing a significant shift.

Potential driving forces for recruitment

The decadal recruitment rates were strongly correlated with the reconstructed mean winter half-year temperature (Fig. 2a) (r = 0.68, 0.66 and 0.71 for E1, N1 and N2, respectively; P < 0.001, n = 42 for E1 and N1; n = 20 for N2) and with summer temperature (Fig. 2c) (r = 0.65, 0.64 and 0.64 for E1, N1 and N2, respectively; P < 0.001). A 10-yr lag of the decadal winter or summer temperature showed less significant correlations with the decadal recruitment rate at E1 (r = 0.52 and 0.58, respectively), N1 (r = 0.52 and 0.60) and N2 (r = 0.57 and 0.56) (P < 0.001 for all). Given an insignificant tree-line upslope movement since 1711, the driving forces underlying the tree-line dynamics were not evaluated statistically.


Regeneration dynamics

The age class distribution in our tree-line stands was used as a rough indicator of the forest state at the landscape scale (Holtmeier, 2003; Körner, 2003). It followed a reverse J-shaped curve, which is considered to be characteristic for a stable forest (Hett & Loucks, 1976). Despite a less significant tree-line advance, a progressively more recent seedling establishment at higher elevations was evident in the ecotone.

Our study plots were characterized by increasing recruitment after the 1950s. It was also shown that the stand density had increased at the upper elevation sites since 1950. Many studies have demonstrated substantial increases in tree-line population density during the 20th century in both high-latitude and high-elevation sites in the Northern Hemisphere (Payette & Filion, 1985; Szeicz & Macdonald, 1995; Camarero & Gutiérrez, 2004; Esper & Schweingruber, 2004; Baker & Moseley, 2007; Danby & Hik, 2007; Kullman, 2007; Shiyatov et al., 2007; Hallinger et al., 2010). However, these densifications of tree-line stands commenced in the first half of the 20th century. In contrast with our findings, poor recruitment after the 1950s occurred in the central Tianshan Mountains, northwest China (Wang et al., 2006), and a seedling establishment deficit was observed after the 1970s and 1980s at alpine tree lines in southern America (Villalba & Veblen, 1997). The Smith fir regeneration patterns and dynamics in the Sygera Mountains may reflect an influence of the regional climate in southeastern Tibet. In addition, we should clarify that trees that died c. > 200 yr ago might have been rotten. Thus, there might have been more trees growing in the earlier periods than identified by the reconstructed population density for the earlier periods.

The synchronous recruitment trends at these three tree-line sites suggest that Smith fir regeneration has been driven by a common external factor, such as climate. The most severe bottlenecks for tree recruitment at tree lines tend to be seed production, seedling emergence and survival (Körner, 2003). At the alpine tree-line ecotone, both winter and summer temperatures are often key constraints on tree recruitment (Lloyd & Fastie, 2003; Holtmeier & Broll, 2007; Harsch et al., 2009). A growing body of literature has also demonstrated that tree population density at tree lines can respond quickly to rising temperatures (Camarero & Gutiérrez, 2004; Esper & Schweingruber, 2004). In our study, Smith fir recruitment, measured on a decadal scale, was also strongly correlated with mean summer and winter temperatures. This seems reasonable considering the harsh winter conditions at the natural Smith fir tree lines in the Sygera Mountains. Severe soil frosts during cold winters were considered to be critical factors in the control of seedling survival by causing needle and shoot desiccation or fine root mortality (Tranquillini, 1979; Körner, 2003; Kullman, 2007). As a confident support, Kullman (2007) found a strong and positive link between winter temperatures and survival rates of Pinus sylvestris tree-line populations in the Swedish Scandes. The second half of the 20th century was the warmest period on record during the past 1000 yr on the Tibetan Plateau, probably stimulating a rapid pulse of viable seed production, dispersal, germination, seedling establishment and survival at the Smith fir tree lines of the Sygera Mountains. In addition, enhanced precipitation in association with recent warming in the study area could also have lowered the mortality rate. The abrupt change in the population age structure observed in the 1970s may also be indicative of the importance of a positive feedback occurring between trees and site-specific environmental conditions, which could influence the spatial patterning of subsequent tree establishment (Alftine & Malanson, 2004; Bekker, 2005; Batllori & Gutiérrez, 2008).

Tree-line movement

The Smith fir tree-line position did not display a proportional upslope movement in response to climatic warming on the Tibetan Plateau (Liu et al., 2005; Thompson et al., 2006; Zhu et al., 2008), in contrast with our hypothesis. Considering that the oldest Smith fir tree is c. 400 yr old, the entire sequence from ad 1611 to 1950 might be viewed as the consequence of a low recruitment into the pre-existing forest during the 200-yr-long Little Ice Age period (ad 1600–1820) after a rather long warm period (ad 1100–1600). If the forest established between ad 1100 and 1600 was of mixed age, we would expect a gradual loss of trees with little replacement over the next 200 yr. The average temperature from 1600 to 1820, reconstructed from tree rings, appeared to be 0.5°C lower than the long-term average, and thus would possibly have lowered the recruitment zone over that time. As an apparent confirmation of this, a continuous recruitment of trees began again at the three study sites at 1800 and a lack of trees from 1611 to 1760 at N2. The once higher altitude trees may have been lost after these were recruited. The oldest tree recorded in this study was actually found close to the current timberline, and the difference in the early placement of the tree line among the three sites is almost certainly a consequence of random survivorship. For an altitudinal lapse rate of 0.6°C per 100 m, the upward movement in the past 400 yr has been a total of 10 m in altitude or 0.06°C at E1, a total of 19.2 m in altitude or 0.12°C at N1, and a total of 9.2 m in altitude or 0.06°C at N2, where the low placement of the previous tree line may have masked upslope movement. Given that c. 2°C warming has taken place in the past 400 yr (Fig. 2), the safest conclusion has to be that there was no significant alteration in tree-line position over this period. On the other hand, an exact number (e.g. a total tree-line advance of 10 m in altitude at E1), derived from the arbitrary tree-line position in the study plots, should not be over-interpreted.

Temperature is generally considered to be the primary determinant for tree-line location; however, only strong and long-term climatic changes are able to produce significant changes in vegetation patterns in alpine regions (Körner, 2003; Holtmeier & Broll, 2007). Strong ecological inertia of Smith fir forests may counteract the influence of warming on the tree-line position, being consistent with other alpine tree lines at low or mid-latitudes (Cuevas, 2000; Cullen et al., 2001; Cui et al., 2005; Dalen & Hofgaard, 2005; Wang et al., 2006; Green, 2009). At some alpine tree lines, the stand density increased, but the tree line did not advance significantly in the 20th century (Luckman & Kavanagh, 1998; Klasner & Fagre, 2002; Camarero & Gutiérrez, 2004). In the Alps, an altitudinal rise in tree occurrence was closely related to changes in land use, such as abandonment of pasture (Gehrig-Fasel et al., 2007; Wieser & Tausz, 2007). There have also been reports that recent warming has caused the consolidation of pre-existing tree populations, rather than changes in the absolute position of the tree line (Payette & Filion, 1985; Danby & Hik, 2007). Thus, a lag response of alpine tree-line position to climate change has been widely reported. It may be a result of local site conditions, species’ traits and feedback effects, as suggested by some alpine tree-line studies (Lloyd & Fastie, 2003; Dalen & Hofgaard, 2005; Batllori & Gutiérrez, 2008).

Seed dispersal and a lack of disturbance are possible limiting factors for Smith fir tree-line movement. Smith fir seeds become mature in October and fall to the ground in November and December. Based on observations by our automatic weather station, the maximum wind speed is c. 3 m s−1 during this period, which is insufficient to blow the seed far away from the crown. The absence of wind-flagging trees at the tree-line ecotone also indicates that the average wind force is not strong. According to Yao et al. (2008), 95% of the seeds fall within the vertical projection of the tree crown, and only 5% can reach 2 m away from the crown, 2% of which can germinate. The lack of animals for the seed dispersal of Smith fir is probably another limiting factor for tree-line upslope movement. Feedback effects may also constrain the response of the Smith fir tree-line position to climate warming. Further studies are necessary to explore why there has been no significant expansion of the high-elevation Smith fir tree line in response to global warming.


Although the Smith fir stands at the tree line in the Sygera Mountains have become increasingly dense because of an increasing number of seedlings after the 1950s, the tree-line position has only moved slightly and insignificantly upslope in response to climate warming. The lack of understanding of the spatio-temporal dynamics of Smith fir tree lines precludes a robust statistical analysis of the underlying causes of a tree-line change. Therefore, the current results should not be over-interpreted, but remain noteworthy because of the significant warming trend in the study area. In order to obtain a more general picture of trends in recruitment and tree-line dynamics, further tree-line studies across different topographical aspects in the Sygera Mountains are needed.


This work was supported by the National Basic Research Program of China (2010CB951301), the Knowledge Innovation Program of the Chinese Academy of Sciences (KZCX2-YWQN111) and the National Natural Science Foundation of China (40871097). We thank Prof. Ian Woodward and the reviewers for useful comments and suggestions, and E. B. M. Drummond at the University of British Columbia for her assistance with English language and grammatical editing of the manuscript.