Leaf Traits and Associated Ecosystem Characteristics Across Subtropical and Timberline Forests in the Gongga Mountains, Eastern Tibetan Plateau

Oecologia (2005) 142: 261-273

DO1 10.1007/~00442-004- 1729-6 Tianxiang Luo + Ji Luo Yude Pan Leaf traits and associated ecosystem characteristics across subtropical and timberline forests in the Gongga Mountains, Eastern Tibetan Plateau

Received: 1 Aug~st2003 / Accepted: 10 September 2004 / Published online: 10 November 2004 O Springer-Verlag 2004

Abstract Knowledge of how leaf characteristics might be threshold-like logistic relationships with annual mean used to deduce information on ecosystem functioning temperature and soil available-N variables. Our results and how this scaling task could be done is limited. In are further supported by additional literature data in the this study, we present field data for leaf lifespan, specific Americas and eastern China. leaf area (SLA) and mass and area-based leaf nitrogen concentrations (N,,,,, N,,,,) of dominant tree species Keywords Altitudinal patterns . Leaves - and the associated stand foliage N-pool, leaf area index Tree canopies . Communities . Scaling-up (LAI), root biomass, aboveground biomass, net primary productivity (NPP) and soil available-N content in six undisturbed forest plots along subtropical to timberline gradlents on the eastern slope of the Gongga Moun- Introduction tains. We developed a methodology to calculate the whole-canopy mean leaf traits to include all tree species Leaf lifespan, specific leaf area (SLA, a measure of leaf (groups) in each of the six plots through a series of surface area per unit mass), and mass- and area-based weighted averages scaled up from leaf-level measure- leaf nitrogen concentrations (Nmass,N,,,,) are funda- ments. These defined whole-canopy mean leaf traits were mental plant traits (Reich et al. 199 1; Korner 199 1). Leaf equivalent to the traits of a leaf in regard to their in- traits interact to determine plant behavior and produc- terrelationships and altitudinal trends, but were more tion and provide a useful conceptual link between pro- useful for large-scale pattern analysis of ecosystem cesses at short-term leaf scales and long-term whole structure and function. The whole-canopy mean leaf plant and stand-level scales (Chabot and Hicks 1982; lifespan and leaf N,,,, mainly showed significant re- Field 1983; Reich et al. 1992; Schulze et al. 1994; Meir lationships with stand foliage N-pool, NPP, LA1 and et al. 2002). However, it is unknown to what extent the root biomass. In general, as elevation increased, the altitudinal/latitudinal trends in the traits of a leaf reflect whole-canopy mean leaf lifespan and leaf N,,,, and geographical variations in ecosystem characteristics of stand LA1 and foliage N-pool increased to their max- structure and function. The available measurements of imum, whereas the whole-canopy mean SLA and leaf the leaf traits associated with altitude and latitude are N,,,, and stand NPP and root biomass decreased from almost limited to fully sunlit, fully expanded current their maximum. The whole-canopy mean leaf lifespan leaves (e.g., Korner et al. 1986; Korner 1989; Niinemets and stand foliage N-pool both converged towards 2001) in which little information on whole plant and stand-level characteristics is available. These studies in- T. Luo (B) dicate that SLA and related structural features appear to Institute of Tibetan Plateau Research, be controlled by temperature because they show similar Chinese Academy of Sciences, Shuangqing Road 18, altitudinal changes under diEerent light and moisture Waidian District, P.O. Box 2871, Beijing, 100085, China E-mail: [email protected] gradients. However, leaf lifespan varies substantially Fax: + 86-10-64889271 among species (Ewers and Schmid 1981; Chabot and Hicks 1982; Reich et al. 1996), and associated leaf traits J. Luo Chengdu Institute of Mountain Hazards and Environment, within species vary with leaf age as well (Field and Chinese Academy of Sciences, Chengdu, 61 0041, China Mooney 1983; Reich et al. 1991). It is common that Y. Pan different plant species with different leaf traits coexist in Northern Global Change Research Program, a plant community (e.g., Reich et al. 1991, 1999). We USDA Forest Service, PA 19073, USA need a new integrated approach to associate the traits of a leaf with ecosystem characteristics of structure and sets from 22 sites along the Tibetan Alpine Vegetation function. Transects (TAVT) (1999-2000) also indicate that the Whole-canopy mean leaf lifespan and associated leaf general distribution patterns of stand aboveground traits that include different age-group leaves from all tree biomass, root biomass, LA1 and NPP have threshold- species in a plot might be useful for indicating adapta- like logistic relationships with climatic factors of tem- tions of natural forests to climatic gradients. Abundant perature and precipitation (Luo et al. 2002b, 2004; evidence illustrates that needle longevity of evergreen T. LUOet al., submitted manuscript). We hypothesize conifers, such as pine, spruce and fir, increases with that whole-canopy mean leaf traits across biomes also elevation and latitude (Ewers and Schmid 1981; Reich show the threshold-like logistic pattern in response to et al. 1995). In general, leaf photosynthetic capacity and the climatic and soil gradients. If proved, this would be maintenance cost decrease (Reich et al. 1997, 1999), helpful to understanding the mechanisms underlying the whereas leaf construction cost increases with increasing NPP, LA1 and biomass variations and for further tests leaf lifespan (Culmon and IFu1ooney 1986). Under given of the generality of Weber's Law. environmental constraints, such as low temperatures, In this report, we explore how leaf characteristics length of growing season or light and nutrient avail- might be used to deduct ecosystem functioning in- ability, the trade-off between leaf carbon costs and formation and how this scaling task could be done based benefits for maximizing carbon gain (Chabot and Hicks on our field data of leaf traits and associated stand 1982; Chapin et al. 1987; Pearcy et al. 1987; Kikuzawa variables in undisturbed forests along the eastern slope 199 1; Sobrado 1991; Cordell et al. 2001) might result in of the Gongga Mountains within the TAVT. Our tasks the optimal whole-canopy mean leaf lifespan and asso- are to: (1) calculate the whole-canopy mean leaf lifespan, ciated leaf traits. In closed-canopy forest stands, foliage SLA, leaf N,,,, and N,,,, to include different age-group mass increases with increasing mean leaf lifespan (Ta- leaves from all tree species (groups) for each forest site, daki 1977; Reich et al. 1992, 1995). The foliage mass (2) compare relationships among leaf traits at leaf-level reaches an essentially steady state early in succession, and stand-level scales, (3) compare altitudinal trends in with a maximum leaf area index (LAI) in the nearly leaf traits at leaf-level and stand-level scales, (4) analyze mature or mature stages (Mooney 1972; Crier and relationships of the whole-canopy mean leaf traits to Running 1977; Tadaki 1977). Inside a plant canopy, the stand variables, climatic factors and soil available-N total leaf area is controlled in a way that leaf N content content to understand the mechanisms underlying the is optimal for the given light, temperature and soil nu- general geographical distribution patterns in stand LA1 trient regimes (Field 1983; Korner et al. 1986; Pearcy and NPP (Luo et al. 2004), aboveground biomass (Luo et al. 1987; Hirose et al. 1997). The seasonal increase in et al. 2002b) and root biomass (T. Luo et al., submitted total foliage N is almost proportional to the increase in manuscript) along the TAVT. total LAI. Hence the ratio of total foliage N to LA1 changes little during the season (Kull and Jarvis 1995; Kull et al. 1998). Based on published data from different Study sites and methods ecological studies of forests across North America, the synthesis by Yin (1993) reveals that stand-level average Transect study sites in the Gongga Mountains leaf Nmassis strongly correlated with climatic factors. More recently, Smith et al. (2002) found a highly pre- This study was conducted in the Hailuogou Natural dictive relationship between whole-canopy rnean leaf Reserve at the eastern slope of the Gongga Mountains. Nmass and aboveground forest productivity in diverse Six field plots of undisturbed forests (0.1-0.5 ha) were forested stands of varying age and species composition. selected along the altitudinal transect, which includes a Weber's Law given in Duvigneaud (1987) indicates subtropical evergreen broadleaf forest (EBLF) at 1,900 that a well-balanced natural plant community, regard- m, a subtropical evergreen-deciduous broadleaf forest less of species composition, should have a similar dry (EDBLF) at 2,200 m, an alpine needle-shaped leaf and matter production, such as net primary productivity broadleaf mixed forest (ANBMF) at 2,850 m, alpine (NPP) and biomass accumulation, under the same en- needle-shaped leaf forests (ANLF) at 3,000 m and vironmental conditions. Lieth (1975) describes the re- 3,050 m, and a timberline needle-shaped leaf forest lationship between climatic factors (annual rnean (TNLF) at 3,700 m (Table 1). The altitudinal transect temperature, annual precipitation, and annual evapo- covered latitudes from N2g032' to N29'37' and long- transpiration) and NPP by a logistic function at a global itudes from El01"58' to E102O03'. During July and scale. We applied Weber's Law and a threshold-like lo- August 1999, we measured leaf traits for lifespan, SLA, gistic function to establish a climate-based statistical leaf N,,,, and N,,,, across dominant tree species along model of NPP of natural vegetation on the Tibetan the transect, and associated stand characteristics for Plateau, in which the product of annual mean tem- aboveground biomass, root biomass, LA1 and NPP, as perature and annual precipitation could explain 70% of well as soil available-N content. The data for stand the NPP variation for the 180 vegetation site data de- biomass, LAI, and NPP are reported in other works rived from 1970-1980s inventory plots over the plateau (Luo et a1. 2002b, 2004; T. Luo et al., submitted (Luo et al. 20024. More recently, our independent data manuscript). Table 1 Stand variables of diameter at breast height (DBH), shoot broadleaf mixed forest, ANLF alpine needle-shaped leaf forests, height, stem basal area, and foliage biomass and leaf area index TlVLF timberline needle-shaped leaf forest, AF Abies fabri, A +B (LA4 among dominant tree species (groups) in the six forest plots Acer sp. + Betula sp., A + P Acer sp. + Populus sp., CO Cyclobnla- along the eastern slope of the Gongga Mountains. EBLF Sub- nopsis oxyodon, LC Lithocarpus cleistocarpz~s,LP Linclera pul- tropical evergreen broadleaf forest, EDBLF subtropical evergreen- clzerrina, PC Phoebe chinensis, PB Picen brachytj~la,P + B Populus deciduous broadleaf forest, AiVBM;F alpine needle-shaped leaf and sp. + Betula sp., Rh Rhododendron sp., I% Tsuga chinensis

Forest type Dominant Mean DBH Mean height Basal area Leaf mass Trees LA1 species (minimum-maximum) (minimum-maximum) (m2 ha-9 (Mg ha-') (ha ha-') (groups) icm)

EBLF (1,900 m) CO LC PC LP A+P EDBLF (2,200 m) GO TC PC LC LP A+P ANBMF (2,850 m) AF PB TG Rh A+B ANLF (3,050 m) AF Rh P+B ANLF (3,000 m) AF Rh P+B TNLF (3,700 rn) AF Rh

Associated stand variables in a forest plot crops of live vegetation based on the data for live-biomass, recent stem growth rate and leaf lifespan. The Table 1 summarized general stand variables of diameter annual woody production (stem, branch and root) of at breast height (DBH), shoot height, stem basal area, trees was the product of the woody biomass multiplied and foliage mass (biomass and LAI) according to by the average annual growth rate (%) of stem volume dominant tree species (groups). Tree height and DBH of during the past 2 or 5 years. The annual production of each forest plot were measured for all trees > 3 cm leaves equaled the result of green leaf mass divided by DBH. The species-specific stem basal area was the species-specific leaf lifespan. The annual woody calculated from measurements of tree DBH. The production of undergrowth shrubs was calculated as the species-specific foliage biomass was calculated from ratio of their biomass divided by their average ages measurements of DBH and tree height using species- ranging between 10 and 25 years. The annual production specific allometric regressions that were developed in our of undergrowth shrub leaves, herb and/or moss equaled earlier study (Luo et al. 2002b). The species-specific LA1 the result of green biomass divided by the leaf lifespan as was the product of the species-specific foliage biomass follows: 2 years for evergreen shrubs and mosses, 1 year multiplied by the species-specific SLA from Table 2. for deciduous shrubs and herbs. More information Stand foliage N-pool was the sum of products of the about field methods of live-biomass both aboveground species-mean leaf Nmass (Table 2) multiplied by the and belowground and the NPP estimates is found in Luo species-specific foliage biomass (Table 1). Stand LAI et al. (2002b, 2004; T. Luo et al., submitted manuscript). was the sum of the tree LAI (Table 1) plus the undergrowth LA1 (Luo et al. 2004). The total root biomass of trees and undergrowth was measured by digging up all Leaf samples at leaf and shoot levels the roots in 0.5x0.5-m2 quadrats to the depth of the deepest visible root (50-100 cm), which included live We selected three to five standard trees with average tree medium and coarse roots and live and dead fine roots. height and quadratic mean DBH calculated from the Aboveground live-biomass of trees in the six forest plots average per-tree basal area according to dominant tree was estimated by species-specific allometric regressions species (groups). For each tree, three sample shoots with on measured tree height and DBH, and undergrowth twigs and leaves were cut from the upper, middle and biomass was measured by harvesting quadrats (2x2 m2). lower canopy positions by climbing access. In total, we NPP was estimated as the sum of increases in standing collected 90 sample shoots from 30 standard trees for 20 Table 2 Species-mean leaf traits of lifespan, specific leaf area (SLA),mass-based leaf nitrogen concentrations (N,,,)and area-based leaf nitrogen concentrations (Na,,,) among dominant tree species (groups) in the six forest plots along the east slope of the Gongga Mountains. For otber abbreviations, see Table 1

Forest type Dominant Species-mean Species-mean SLA Species-mean leaf Nm,,, Species-mean leaf N,,,, species lifespan (years)(m + SD) (cm2 g-"(range) (mg g-l) (range) (g m-7 (range)

EBLF (1,900 m) CO 2.0 rt 0.0 LC 2.0 LF' 0.0 PC 2.0 LF' 0.0 LP 2.0 LF' 0.0 A. +P. 0.5 (6 months) EDBLF (2,200 m) CO 2.0 i 0.0 TC 2.7 rt 0.6" PC 2.0 i 0.0 LC 2.0 =I=0.0 LP 2.0 LF' O.ob A. + P. 0.5 (6 months) ANBMF (2,850 m) AF 6.7 rt 0.6 PB 6.0 LF' 0.0 TC 2.7f 0.6 Rh 1.0 + 0.0 A. 4- B. 0.4 (5 months) ANLF (3,050 m) AF 7.lf 1.7 Rh 2.0 * 0.0 P. + B. 0.3 (4 months) ANLF (3,000 m) AF 6.9 + 1.3 Rh 2.0 + o.oc P. + B. 0.3 (4 months) TNLF (3,700 m) AF 7.5 LF'0.6 Rh 2.0 f 0.0

"Using the measurements at 2,850 m the measurements at 1,900 m "Using the measurements at 3,050 m

woody plant species along the altitudinal transect. We for Abies and Tsuga trees and double rectangular area separated leaf age classes and measured the dry mass of for Picea trees, where side length and width (mm) of a different age-group leaves for each sampled shoot. In single needle were measured using Vernier callipers. For evergreen conifers, the apex of a long shoot produces a each of the 146 leaf samples, leaf areas of 30-50 fresh yearly growth increment that bears a single age class of leaves were measured, and the leaves were then dried to needles (Ewers and Schmid 1981). We counted back leaf a constant weight at 70°C. age classes from shoot tips, which were validated by counting tree-rings at the base of the shoots. In broadleaved evergreen trees, leaf age classes were determined Species-mean leaf traits among dominant tree species by counting back the internodes from the shoot tips (groups) in a forest plot (Wang et al. 2000). The leaf age of current leaves of deciduous and evergreen trees was determined as the Based on leaf sample measurements of the leaf and duration between leaf onset and sampled date. The leaf shoot levels given above, we calculated species-mean leaf lifespan of evergreen trees was determined by the max- lifespan, SLA, leaf N,,,, and N,,, concentrations ac- imum leaf age in each sampled shoot. The leaf lifespan cording to dominant tree species (groups) in the six of deciduous trees were determined as the duration be- forest plots (Table 2). The species-mean leaf lifespan was tween leaf onset and leaf fall according to local pheno- the arithmetic average of maximum leaf ages from each logical observations. sampled shoot in different canopy positions of a tree. We collected 146 samples of age-group leaves from Other species-mean leaf traits were the shoot-level the 90 sample shoots according to different tree species weighted-average leaf traits from different canopy posi- and forest sites. We measured leaf N,,,, and N,,,, tions of the tree based on weighted dry mass or area of concentrations and SLA. The leaf N,,,, was analyzed leaves in different age classes. with a micro-Kjeldahl assay. The leaf N,,,, was calculated from the leaf N,,,, and associated SLA (the ratio of fresh leaf area to dry mass). The one-side leaf area for Whole-canopy mean leaf traits in a forest plot broadleaved trees was measured using a (21-203 portable laser area meter (CID). The curve surface leaf area ex- The whole-canopy mean leaf traits (??') to include all tree posed to sunlight for coniferous trees was determined species (groups) in a forest plot were defined as the according to the needle shapes: single rectangular area weighted averages of leaf traits using the species-mean leaf traits (Si) (Table 2) and the species-specific stem 0.6"C per 100 m of altitude. Annual precipitation at basal area or foliage mass (Ki) (Table I): W=C(Six Ki)/ below 2,500 m in elevation was estimated from the me- x(Ki). The whole-canopy mean leaf lifespan was based teorological observatory at 1,600 m in increments of 120 on the species-mean leaf lifespan and the species-specific mm per 100 m (1,600-2,500 m). Annual precipitation stem basal area. The whole-canopy mean SLA and leaf above 2,500 m was estimated from the meteorological N,,,, were based on the species-mean SLA and leaf observatory at 3,000 m in decrements (2,500-3,000 m) or N,,,, and the species-specific foliage biomass. The increments (above 3,000 m) of 74 mm per 100 m (Zhong whole-canopy mean leaf N,,,, was based on the species- et at. 1997). mean leaf N,,,, and the species-specific LAI.

Modeling relationships between whole-canopy mean leaf Soil available-N content traits and climatic and soil variables

We collected soil samples and measured soil bulk den- We hypothesized that the relationships between the sities by layer (Ao, Al and/or B horizons) from the soil whole-canopy mean leaf traits and climatic and soil pits (0.5x0.5-mZ quadrats) to the depth of the deepest variables follow Weber's Law with simple logistic visible root. Four soil pits were dug under an average equations: tree in each forest plot where the layer-specific soil samples were from different soil pits. The soil samples were chemically analyzed for available-N content by layer. The FeS04+ Zn + NaOH distilled water extrac- tion analysis determined the soil available-N content. where, y is dependent variables of the whole-canopy The soil bulk density was measured with the cutting ring mean leaf traits, and x independent climatic and soil method in most plots, or by weighting soil mass in some variables, including annual mean temperature and soil forest plots with rocky soils. The storage of available-N available-N content. Temperature was considered the in soils was calculated from the soil-mass-weighted limiting factor for the vegetation distribution in the averages for available-N content and soil bulk density Gongga Mountains because actual evapotranspiration is and the maximum plant root depth average (Table 3). generally low and accounts for only 27% of tlie annual precipitation at 3,000m (Zhong et al. 1994). Only annual mean temperature was used in the modeling because the Estimates of climatic factors along the transect estimated mean temperatures for January, July and the year and annual precipitation along the altitudinal The climatic data of the six forest sites along the alti- transect were highly correlated (r2= 0.99, P < 0.001). k is tudinal transect were estimated from 10 years of me- the maximum leaf trait measurements estimated from teorological observations measured at 1,600 and 3,000 m Table 2: eight for leaf lifespan (year), 30 for leaf Nmass by the Alpine Ecosystem Observation and Experiment concentration (mg/g DM), four for leaf N,,,, con- Station, Chinese Academy of Sciences (Table 3). Annual centration (g/m2 leaf area) and 250 for SLA (cm2/g mean temperature was calculated using a lapse rate of DM). exp is the base of the natural logarithm, and the

Table 3 Whole-canopy mean leaf traitsa in relation to stand foliage N-pool, LAI, net primary productivity (NPP), root biomass, aboveground biomass, soil available-N content and storage, and climatic factors of annual mean temperature and annual precipitation. DM Dry mass; for other abbreviations, see Table 1 Leaf traits and associated ecosystem variables EBLF EDBLF ANBMF ANLF ANLF TNLF (1,900 m) (2,200 m) (2,850 m) (3,000 m) (3,050 m) (3,700 m)

Whole-canopy mean leaf lifespan (year) Whole-canopy mean leaf N,,,, (mg g;' DM) Whole-canopy mean leaf N,,,, (g F- leaf area) Whole-canopy mean SLA (cdgg- DM) Foliage N-pool (Kg ha-' land) Stand LA1 (ha ha-' land) NPP (Mg DM ha-' Root biomass (Mg DM ha-' land) Aboveground biomass (Mg DM ha-' land) Soil available-N content (mg per l0Og soil) Soil available-N storage (kg ha-' land) Annual mean temperature ("C) Annual precipitation (cm)

"The whole-canopy mean leaf traits (W) to include all tree species species-specific stem basal area or foliage mass (Ki) (Table 1): (groups) in a forest plot were defined as the weighted averages of W=~(SixKi)j~Ki leaf traits using the species-mean leaf traits (Si) (Table 2) and the b~tationobservations at 3,000 rn other parameters of a, b and c are equation coefficients. across tree species groups. Accounting for the age effects The site-specific data in Table 3 were used to determine is the basis for scaling up the traits of a leaf to whole the equation coefficients by the least squares regression plant and stand-level scales. At a stand-level scale, the method. When there was a non-robust logistic re- analysis of simple linear relationships among the scaled- lationship (P> 0.05), we further tested the data with < up leaf traits indicated similar trends (from Table 3): the other models such as linear, power, log and expon ential whole-canopy mean leaf N,,,, and SLA both deceased functions. (r2= 0.88-0.93, P < 0.0 l), whereas the whole-canopy mean leaf N,,,, increased (r2= 0.75, P < 0.05) with increasing whole-canopy mean leaf lifespan. The whole- Results canopy mean leaf N,,,, and SLA were significantly correlated (r2= 0.71, P < 0.05). Comparisons of relationships among leaf traits at leaf and whole-canopy scales Comparisons of altitudinal trends in leaf traits At a leaf scale, the leaf N,,,, and SLA both ex- at leaf and whole-canopy scales ponentially decreased with increasing leaf age (r2= 0.50- 0.53, P < 0.001) (Fig. la, c). However, the leaf N,,,, Figure 2 presents altitudinal trends of the within-age leaf showed a distinct pattern in relation to leaf age (Fig. lb), trait measurements in 1-year-old leaves across evergreen which logarithmically increased with age (r2= 0.23, tree species groups. As elevation increased, the leaf P < 0.001). The leaf N,,,, logarithmically increased with N,,,, was generally stable (r2=0.03, not statistically increasing SLA (r2= 0.55, P < 0.00 1) (Fig 1d) because significant at P<0.10) (Fig. 2c), but the leaf N,,,, in- both were strongly correlated with leaf age. The results creased (r2= 0.16, P < 0.10) (Fig. 2a), and the SLA de- indicated that as a leaf aged, the leaf N,,,, and SLA creased (r2= 0.16, P < 0.10) (Fig. 2b). At a stand-level sharply decreased, while the leaf N,,,, slightly increased scale, the percentage of stem basal area of evergreen

Fig. la-d Relationships among the traits of a leaf across tree 0 Spruce- fir trees species groups along th-G eastern A Evergreen broadleaved trees slope of the Gongga A A Deciduous broadleaved trees Mountains. Because leaf mass- Undergrowth Rhododendron based nitrogen concentrations O (N,,) (a), area-based leaf nitrogen concentrations (N,,,,) (b) and specific leaf area (SLA) (c) were closely related to leaf age, both SLA and leaf N,,,, were well correlated (d). yr Year

Leaf age (yr.) Leaf age (yr.)

Leaf age (yr.) SL~(crn2 g-'1 trees significantly increased with increasing altitude six forest plots of the Gongga Mountains (Table 4). (r2= 0.73, P < 0.05) (Fig. 2d) where the whole-canopy Stand foliage N-pool increased with increasing whole- mean leaf lifespan also significantly increased (r2= 0.9 1, canopy mean leaf lifespan (P< 0.01) and leaf N,,,, P<0.01) (from Table 1). As elevation increased, the (P < 0.05), but decreased with increasing SLA (P < 0.0 I) whole-canopy mean SLA significantly deceased and leaf N,,,, (P < 0.05). Stand LA1 had a significant (r2= 0.80, P<0.02) and the leaf N,,, slightly increased positive relationship with the whole-canopy mean leaf (r2= 0.63, P < 0.10) with similar patterns in the traits of a lifespan (P < 0.05) and a negative relationship with the leaf. However, the whole-canopy yean leaf N,,,, sig- leaf Nm,,, (P<0.05). However, the LA1 showed weak nificantly decreased with altitude (r= 0.74, P < 0.05) in relationships with the whole-canopy mean leaf N,,,, and contrast to the leaf-level variation. Although the traits of SLA (not statistically significant at P < 0.05). Stand root a leaf varied greatly with plant species groups, the biomass and NPP, in contrast, both had a negative re- whole-canopy mean leaf traits indicated more significant lationship with the whole-canopy mean leaf lifespan altitudinal patterns consistent with the change in relative (P < 0.01) and a positive relationship with the leaf N,,,, percentages of evergreen versus deciduous canopy. (P < 0.0 I) and SLA (P < 0.05). AIL four whole-canopy mean leaf traits had no linear relationship with stand aboveground biomass (data not shown). The results in- Relationships between whole-canopy mean leaf traits dicated that the whole-canopy mean leaf traits had sig- and stand characteristics nificant implications for the structure and function of forest ecosystems. We did a correlation analysis between whole-canopy mean leaf traits and stand characteristic variables for the

Fig. 2 Altitudinal trends in leaf- 3.6 - (a) 100 (a), level N,,,, SLA (b) and - N,,,, (c) of 1-year-old leaves 3.4 across evergreen tree species 3.2 - groups were compared with the - change in percentages of stem N? 3.0 basal area of evergreen versus E 2.8 - Cn V deciduous trees (d) along the 2.6 -0 eastern slope of the Gongga Mountains. For abbreviations, zg 2-4 -0 see Fig. I

Altitude (m) Altitude (m)

0 Spruce-fir trees Evergreen broadleaved trees Evergreen trees a Undergrowth Rhododendron Deciduous trees

2000 2500 3000 3500 1900 2200 2850 3000 3050 3700 Altitude (m) Altitude (m) Whole-canopy mean leaf traits in relation positively correlated with SLA, and all three traits de- to climatic and soil factors cline with increasing leaf lifespan (see Reich et al. 1997, 1999). Generality in the relationships among leaf traits The logistic function (y= k/[1 + exp(a + hex)]) fitted the across diverse communities and ecosystems has sig- relationship between whole-canopy mean leaf lifespan nificant implications for global-scale modeling of vege- and annual mean temperature significantly (r2 = 0.94, tation-atmosphere C02 exchange (Schulze et al. 1994; P < 0.01) (Fig 3a), while another logistic function (y= k/ Reich et al. 1999). However, limited knowledge exists [1 + exp(a + b-x+ c.x2)]] fitted the relationship between about their relationships with ecosystem characteristics the leaf lifespan and soil available-N content well of structure and function (Reich et al. 1992). (r2= 0.82, P < 0.02) (Fig. 3b). Because whole-canopy Taking into account age effects and species variations mean SLA, Nmassand N,,,, were closely correlated with is the basis for scaling up the traits of a leaf to a stand- the leaf lifespan (as indicated above), they also showed level scale. In this study, we developed a methodology to similar logistic relationshps with annual mean tem- calculate the whole-canopy mean leaf traits to include all perature (r2= 0.64-0.78) and soil available-N content tree species (groups) in a forest plot through a series of (r2 = 0.76-0.92) (data not shown). Because the whole- weighted averages scaled up from leaf-level measure- canopy mean leaf lifespan was closely related to the al- ments. Such defined whole-canopy mean leaf traits were titudinal distribution of evergreen forest trees (from Fig. well correlated and consistent with the general re- 2d, Table 3), it would be an integrated indicator for lationships among the traits of a leaf in this study (Fig. ecological adaptations of natural forests to the climate 1) and many previous studies of diverse taxonomic gradients in the Gongga Mountains. groups and biomes (e.g., Field and Mooney 1986; Reich et al. 1999). Much evidence from altitudinal transect studies on within-age leaf traits indicates consistent Discussion trends that as elevation increases, SLA decreases, whereas leaf N,,,, increases (Woodward 1986; Korner Approach to scaling up the traits of a leaf et al. 1986; Korner 1989) or remains unchanged in to the stand-level scale evergreen conifers (Hultine and Marshall 2000). Fur- thermore, leaf N,,,, increases with altitude in herbac- Leaf mass-based photosynthetic capacity and leaf N eous and deciduous woody plants (Woodward 1986; concentration are usually positively correlated. Both are K6rner 1989; Weih and ICarlsson 1999) but is

Table 4 Correlation coefficients for linear relationships between whole-canopy mean leaf traits and stand variables among the six forest sites in the Gongga Mountains. For abbreviations, see Tables 1, 2 and 3

Independent variables Foliage N-pool LA1 Root biomass NPP

Whole-canopy mean leaf lifespan 0.9542"" Whole-canopy mean leaf N,,,, 0.8819" Whole-canopy mean leaf N,,,, -0.8313" Whole-canopy mean SLA -0.9298"" - *P< 0.05, ""P < 0.01

Fig. 3 Whole-canopy mean leaf lifespan showed logistic relationships with annual mean temperature (a) and soil available-N content (b) across the six forest sites in the Gongga Mountains

Annual mean temperature ('c) Soil available-N content (mg per IOOg soil) remarkably stable in evergreen woody plants (Korner Generality of logistic relationships between 1989) or decreases in evergreen conifers (Hultine and whole-canopy mean leaf traits and climatic factors Marshall 2000). Our field data for the within-age leaf traits and their whole-canopy weighted averages along The estimated annual mean temperature and annual the east slope of Gongga Mountains generally confirm precipitation in the Gongga Mountains are highly cor- previous conclusions. Moreover, the scaled-up whole- related, and the ternperature in the subtropical forests is canopy mean leaf traits indicated more significant alti- low. We need to test for a generality of the logistic re- tudinal patterns consistent with the change in relative lationships between the whole-canopy mean leaf traits percentages of evergreen versus deciduous canopy. and mean temperature found in this study. More importantly, the whole-canopy mean leaf life- We collected more data from Reich et al. (1999) and span and leaf N,,,, generally showed significant re- Wang et al. (2000) and then estimated the whole-canopy lationships to stand foliage N-pool, LAI, root biomass mean leaf traits for an additional six forest communities and NPP (Table 4). We believe that the whole-canopy in the Americas and eastern China (Appendix 1). Be- mean leaf traits are equivalent to the traits of a leaf but cause of the lack of data for stem basal area and foliage are more useful for large-scale pattern analysis of eco- biomass in the literature, the whole-canopy mean leaf system functioning. traits in the additional forest stands were estimated as the arithmetic average of the leaf trait measurements available for the dominant tree species. Then we mixed Relationships between foliage N-pool and abiotic the additional plot data from other studies with our plot factors of climate and soil data and did the same logistic regression analysis for the pooled data sets from 12 plots (Fig. 5). The data sets Stand foliage N-pool has important implications for cover a wide range of forests along tropical/subtropical ecosystem structure and function (Chapin et al. 1990). to temperate/alpine gradients where annual mean tem- Because estimates of foliage N-pool were based on the perature and annual precipitation have a low relation- data for leaf NmaSsand stand foliage biomass, we applied ship (r2 = 0.16, not statistical significance at P < 0.10). a similar logistic function to the one above to analyze All three whole-canopy mean leaf traits for leaf lifespan, their relationships with annual mean temperature and SLA and leaf N,,,, fit the logistic patterns associated soil available-N storage. Here, we set the maximum fo- with annual mean temperature (r2 = 0.49-0.76, P < 0.02) liage N-pool (k) to equal 300 kg/ha land that was esti- (Fig. 5a, c, d). The whole-canopy mean leaf lifespan also mated from the data in Table 3. Annual mean fits the logistic relationship for annual preci itation temperature and soil available-N storage explain 92 and across tropical and temperate forests (r4 =0.92, 78% of the variation in foliage N-pool, respectively P < 0.001) where annual mean temperature is > 8°C (Fig. 4). The soil available-N storage also exponentially (Fig. 5b). However, annual precipitation shows no re- increased with increasing altitude (r2= 0.71, P < 0.05) lationship with the canopy-mean SLA and leaf Nmass (from Table 3). Along the altitudinal transect, foliage N- (data not shown). pool was linearly correlated with stand NPP (r2=0.86, Leaf lifespan has highly important implications for P< 0.01), root biomass (r2=0.78, P< 0.02) and LA1 the altitudinal distribution of evergreen forest trees (r2= 0.67, PC 0.05) (from Table 3). The results indicate (from Fig. 2d). The pooled data from this study and that the climatic gradient characterizes not only the ve- additional literature indicate that the whole-canopy getation distribution but also the soil N conditions of the mean leaf lifespan shows higher correlations with forest ecosystems. temperature and precipitation than the other three leaf

Fig. 4 Stand foliage N-pool 300 showed logistic relationships with annual mean temperature ;3- (a) and soil available-N storage 2 250 (b) across the six forest sites in y7 the Gongga Mountains $ 200 sIS)

at IS) -.-m 0

Annual mean temperature ('(2) Soil available-N storage (Kg ha'-' land) traits. It seems that temperature and/or precipitation contribute to the effects of elevation and latitude on characterize mainly the size of the whole-canopy mean conifer needle longevity (Ewers and Schmid 1981; leaf lifespan that sets the upper limit of stand canopy Chabot and Hicks 1982; Reich et al. 1995). A later leaf ages, while the canopy-mean SLA and leaf N,,,, report by Reich et al. (1996) provides evidence from are according to seasonal changes in canopy leaf-age garden experiments that longer needle longevity of structure (e.g., leaf mass ratios among different age- spruce and pine populations at high elevations and high group leaves). The lower relationship between leaf N,,,, latitudes is largely an environmentally regulated phe- and leaf age (Fig. 1b) indicates that more complicated notypic acclimation. N allocations to leaf areas exist. Inside a plant canopy, evidence shows that the vertical distribution of leaf N,,,, is mainly controlled by photosynthetic photon Strategy for nitrogen conservation and maximum flux density (Field 1983; Ellsworth and Reich 1993; carbon gain in natural forests Meir et al. 2002). In a plant community, canopy leaf- age structure is generally considered a control on al- It is still unclear why some forest types, such as tem- locating leaf N for maximization of carbon gain (Field perate/alpine evergreen conifers, maintain high LA1 1983; Kull et al. 1998). Given a climate regime, the values of more than six or seven and what their biolo- optimal leaf age structure for resource use should exist. gical function is. In the Gongga Mountains, alpine Earlier analyses among genera and species suggest that spruce-fir forests had the highest LA1 ranging from 7 to both phenotypic plasticity and genotypic variation 10 (Table 3) based on the biomass allometric regressions

Fig. 5 Based on the pooled data from this study (Table 3) and the literature (Appendix l), whole-canopy mean leaf lifespan (a), SLA (c) and leaf N,,,, (d) fit the logistic patterns associated with annual mean temperature. The whole-canopy mean leaf lifespan also fits the logistic relationship for annual precipitation across tropical and temperate forests where annual mean temperature is > 8°C. For abbreviations, see Fig. 1

Annual mean temperature (OC) Annual precipitation (cm)

- Fitting curve 0 Tropical rainforest CI Subtropical evergreen broadleaved forest A Temperate mixed forest O Alpine spruce-fir forest

Annual mean temperature ('(2) Annual mean temperature (OC) with harvested trees ranging from 10 to 74 cm in DBH. cing nutrient leaching losses from leaves and soil as a Water balance simulations further suggest that available result of high rainfall and temperature (Monk 1966; soil water is enough to support such a high forest LA1 Chabot and Hicks, 1982; Sobrado 1991; Cordell et al. (Luo et al. 2002~).High measurements of forest LA1 200 1). have led to questions concerning the methodology used to calculate them since Marshall and Waring (1986) reported that estimates of leaf area based on tree dia- Implications from convergence towards logistic meter appear to be inaccurate, and therefore the ex- patterns in whole-canopy mean leaf traits ceedingly high leaf-area indices previously reported for and stand characteristics Douglas-fir forests are thought to be unreliable. How- ever, Ren and Peng (1997) present different results in The community-oriented growth analysis can provide a their study on comparisons of digerent LA1 rneasure- better basis to predict effects of climate change on plant ment methods in three forest types in the Dinghushan growth than the species-based analysis (Korner 199 1). In Reserve, south China. The study indicates that three response to climatic and soil gradients in the Gongga methods, including empirical allometric regressions, in- Mountains, whole-canopy mean leaf lifespan and stand clined point quadrats and light interception, give the foliage N-pool converged towards threshold-like logistic same LA1 estimates and the serious ~lnderestimatefrom relationships with annual mean temperature and soil litterfall is because of the influences of frequent ty- available-N variables (Figs. 3,4). In general, as elevation phoons and storms in the region. increased, the whole-canopy mean leaf lifespan and leaf The LA1 of alpine spruce-fir forests at high altitudes N,,,, and stand LA1 and foliage N-pool increased to seems to be controlled by low soil temperatures and the their maximum, whereas the whole-canopy mean SLA interaction between foliage production and soil N and leaf Nm,,, and stand NPP and root biomass de- availability. Approximately 75% of the nitrogen in a creased from their maximum. Such threshold-like lo- plant leaf with C3 photosynthesis is invested in photo- gistic patterns are also found in stand NPP, LA1 and synthetic components, and nitrogen acquisition by roots live-biomass both aboveground and belowground across is a major carbon expense of a plant (Chapin et al. 1987). s~lbtropicalforests to alpine vegetation on the Tibetan In the Gongga Mountains, forest ecosystems at higher Plateau (Luo et al. 2002a, b, 2004; T. Luo et al., sub- altitudes tended to have higher foliage N-pool and mitted manuscript). These results confirm that in plant higher soil N storage where stand canopy had longer leaf growth, natural selection favors a high carbon gain, longevity and higher foliage mass. We found that root close to the maximum that can be maintained in any biomass was negatively correlated with whole-canopy given environment (Mooney 1972; Grime 1977). mean leaf lifespan and foliage N-pool. Our synthesis in Understanding the mechanisms underlying these re- Fig. 5 suggests that the whole-canopy mean leaf lifespan lationships will increase our capacity to predict future appears to be an integrated indicator for such ecological ecosystem behaviors under global climate change. adaptations of natural forests to temperature and/or precipitation gradients. Many theories explain that Acknowledgements This study was funded by the Key Project of the variations in leaf lifespan are a strategy for optimizing Chinese Academy of Sciences (KZCX3-SW-339-04), the National plant carbon gain (Grime 1977; Chabot and Hicks 1982; Key Projects for Basic Research of China (2002CB111504), the National Natural Science Foundation of China (30370290), the Pearcy et al. 1987; Kikuzawa 1991) and/or plant adap- Special President Foundation of the Chinese Academy of Sciences tations to specific temperature, moisture and nutrient (2002), and the USDA Forest Service Northern Global Change regimes (Monk 1966; Waring and Franklin 1979; Coley Program (02-IC-11242343-029). We thank Dr E.-D. Schulze for his et al. 1985; Chapin et al. 1987). Low temperature and comments on this manuscript. then low growth rates favor plant longevity (Grime 1977; Coley et al. 1985) where leaf and tree lifespan are correlated (Reich et al. 1992). The growth of leaves at high altitude seems to be controlled in a way that leads Appendix 1 to comparatively high nutrient contents, which in turn support high metabolic activity (Kiirner 1989). On the Estimated whole-canopy mean leaf traits of forests in other hand, the closed canopy of trees with high leaf the Americas and eastern China based on the data in mass and long leaf longevity generally creates a low soil Reich et al. (1999) and Wang et al. (2000). Because of temperature, which impairs root activity (Korner 1998). the lack of data for stem basal area and foliage bio- Then carbon costs for nitrogen absorption by roots in mass, the whole-canopy mean leaf traits were esti- alpine plants become more expensive and extending the mated as the arithmetic average of the leaf trait leaf lifespan is more economic. For tropical or sub- measurements available for the dominant tree species tropical evergreen broadleaf forests, evergreen canopy in each forest stand. AMT Annual mean temperature, leaves with a relatively shorter mean lifespan (generally AP annual precipitation, SLA specific leaf area, N,,, approximately 1-3 years) would favor mineral con- leaf mass-based nitrogen concentrations, yr. year, DM servation and maintain optimal growth rates by redu- dry mass Locations Vegetation type AMT AP Whole-canopy Whole-canopy Whole-canopy Authors ("C) (cm) mean leaf mean SLA mean leaf N,,,, lifespan (yr.) (cm2 g-l DM) (mg g-"DM)

San Carlos, Tropical rainforest (22 tree 26.0 356 2.31 + 1.46 88 rir 25 14.0 3Z 4.7 Reich et al. (1999) Amazonas, species) Venezuela South Wisconsin Cold temperate forest of 8.0 82 1.023~1.03 98 + 58 18.1 + 11.2 Reich et al. (1999) pines and deciduous broad leaved trees (25 tree species) Coweeta, Montane humid temperate 12.5 183 1.9011.83 122k72 18.1 i5.1 Reich et al. (1999) North Carolina forest of pines and deciduous broadleaved trees (eight tree species) Hobcaw, Warm temperate forest of pines 18.3 130 1.42 * 0.88 67 3Z 26 11.913.5 Reich et al. (1999) South Carolina and deciduous broadleaved trees (six tree species) Niwot Ridge, Sub-alpine spruce-fir and pine - 1.2 90 6.17 Ilt: 2.75 33 rt 6 10.6 3Z 0.5 Reich et al. (1999) Colorado forest ecotone (three tree species) Tiantong National Subtropical evergreen broadleaf 16.2 138 1.69 10.73 - - Wang et al. (2000) Forest Park, forest(35 tree species) Zhejiang, China

Hirose T, Ackerly DD, Traw MB, Ramseier D, Bazzaz FA (1997) References C02 elevation, canopy photosynthesis, and optimal leaf area index. Ecology 78:2339-2350 Chabot BF, Hicks DJ (1982) The ecology of leaf life spans. Annu Hultine KR, Marshall JD (2000) Altitudinal trends in conifer leaf Rev Ecol Syst 13:229-259 morphology and stable carbon isotope composition. Oecologia Chapin FS 111, Bloom AJ, Field CB, Waring RH (1987) Plant 123:32-40 responses to multiple environmental factors. BioScience 37:49- Kikuzawa K (1991) A cost-benefit analysis of leaf habit and leaf 57 longevity of trees and their geographical pattern. Am Nat Chapin FS 111, Schulze E-D, Mooney HA (1990) The ecology and 138:1250-1263 economics of storage in plants. Annu Rev Ecol Syst 21:423-447 Korner CH (1989) The nutritional status of plants from high alti- Coley PD, Bryant JP, Chapin FS I11 (1985) Resource availability tudes: a worldwide comparison. Oecologia 8 1:379-39 1 and plant antiherbivore defense. Science 230: 895-899 Korner CH (1991) Some often overlooked plant characteristics as Cordell S, Goldstein G, Meinzer FC, Vitousek PM (2001) Reg- determinants of plant growth: a reconsideration. Funct Ecol ulation of leaf life-span and nutrient-use efficiency of Me- 5:162-173 trosidero polymorpha trees at two extremes of a long Korner Ch (1998) A re-assessment of high elevation treeline posi- chronosequence in Hawaii. Oecologia 127:198-206 tions and their explanation. Oecologia 115:445-459 Duvigneaud P (1987) La synthise icologique. Chinese Science Korner Ch, Bannister P, Mark AF (1986) Altitudinal variation in Press, Beijing, p 122 stomata1 conductance, nitrogen content and leaf anatomy in Ellsworth DS, Reich PB (1993) Canopy structure and vertical different plant life forms in New Zealand. Oecologia 69:577-588 patterns of photosynthesis and related leaf traits in a deciduous Kull 0,Jarvis PG (1995) The role of nitrogen in a simple scheme to forest. Oecologia 96: 169-1 78 scale up photosynthesis from leaf to canopy. Plant Cell Environ Ewers FW, Schmid R (1981) Longevity of needle fascicles of Pinus 18:1174-1182 longaeva (bristlecone pine) and other North American pines. Kull 0, Koppel A, Noormets A (1998) Seasonal changes in leaf Oecologia 5 1: 107-1 15 nitrogen pools in two Salix species. Tree Physiol 18:45-51 Field C (1983) Allocating leaf nitrogen for the maximization of Lieth H (1975) Modeling the primary productivity of the world. carbon gain: leaf age as a control on the allocation program. In: Lieth H, Whittacker RH (eds) Primary productivity of Oecologia 56: 34 1-347 the biosphere. Springer, Berlin Heidelberg New York, pp Field C, Mooney HA (1983) Leaf age and seasonal effects on light, 237-64 water, and nitrogen use efficiency in a California shrub. Oeco- Luo T, Li W, Zhu H (2002a) Estimated biomass and productivity logia 56:348-355 of natural vegetation on the Tibetan Plateau. Ecol Appl12:980- Field C, Mooney HA (1986) The photosynthesis-nitrogen 997 relationship in wild plants. In: Givnish TJ (ed) On the economy Luo T, Shi P, Luo J, Ouyang H (2002b) Distribution patterns of of plant form and function. Cambridge University Press, New aboveground biomass in Tibetan alpine vegetation transects. York., pp 25-55 Acta Phytoecol Sin 26:668-674 Grier CC, Running SW (1977) Leaf area of mature northwestern Luo T, Neilson RP, Tian H, Vorosmarty CJ, Zhu H, Liu S (2002~) coniferous forests: relation to site water balance. Ecology A model for seasonality and distribution of leaf area index of 58:893-899 forests and its application to China. J Veg Sci 13:817-830 Grime JP (1977) Evidence for the existence of three primary stra- Luo T, Pan Y, Ouyang H, Shi P, Luo J, Yu Z, Lu Q (2004) Leaf tegies in plants and its relevance to ecological and evolutionary area index and net primary productivity along subtropical to theory. Am Nat 11 1 : 1169-1 194 alpine gradients in the Tibetan Plateau. Global Ecol Biogeogr Gulmon SL, Mooney HA (1986) Costs of defense and their effects 13:345-358 on plant productivity. In: Givnish TJ (ed) On the economy of Marshall JD, Waring RH (1986) Comparison of methods of esti- plant form and function. Cambridge University Press, New mating leaf-area index in old-growth Douglas-fir. Ecology York, pp 681-698 67:975-979 Meir P, Kruijt B, Broadmeadow M, Barbosa E, Kull 0, Carswell Schulze E-D, Kelliher FM, Korner CH, Lloyd J, Leuning R (1994) F, Nobre A, Jarvis PG (2002) Acclimation of photosynthetic Relationships among maximum stomata1 conductance, ecosys- capacity to irradiance in tree canopies in relation to Ieaf ni- tem surface conductance, carbon assimilation rate, and plant trogen concentration and leaf mass per unit area. Plant Cell nitrogen nutrition: a global ecology scaling exercise. Annu Rev Environ 25:343-357 Ecol Syst 25529-660 Monk CD (1966) An ecological significance of evergreenness. Smith M-L, Ollinger SV, Martin ME, Aber JD, Hallett RA, Ecology 47:504--505 Goodale CL (2002) Direct estimation of aboveground forest Mooney HA (1972) The carbon balance of plants. Annu Rev Ecol productivity through hyperspectral remote sensing of canopy Syst 3:315-346 nitrogen. Ecol Appl 12: 1286-1302 Niinemets U (2001) Global-scale climatic controls of Ieaf dry mass Sobrado MA (1991) Cost-benefit relationships in deciduous and per area, density, and thickness in trees and shrubs. Ecology evergreen leaves of tropical dry forest species. Funct Ecol 82:453-469 5:608-616 Pearcy RW, Bjorkman 0, Caldwell MM, Keeley JE, Monson RK, Tadaki Y (1977) Leaf biomass. In: Shidei T, Kira T (eds) Primary Strain BR (1987) Carbon gain by plants in natural environ- productivity of Japanese forests: productivity of terrestrial ments. BioScience 37:21-29 communities. JIBP synthesis 16. University of Tokyo Press, Reich PB, Uhl C, Walters MB, Ellsworth DS (1991) Leaf life span Tokyo, pp 39-44 as a determinant of leaf structtlre and function among 23 Wang X, Zhang J, Zhang Z (2000) Leaf longevity of evergreen Amazonian tree species. Oecologia 86: 16-24 broad-leaved species of Tiantong National Forest Park, Zhe- Reich PB, Walters MB, Ellsworth DS (1992) Leaf life-span in re- jiang Province (in Chinese). Acta Phytoecol Sin 24:625-629 lation to leaf, plant, and stand characteristics among diverse Waring RII, Franklin JF (1979) Evergreen coniferous forests of the ecosystems. Ecol Monogr 62:365-392 Pacific Northwest. Science 204: 1380-1 386 Reich PB, Koike T, Gower ST, Schoette AW (1995) Causes and Weih M, Karlsson PS (1999) Growth response of altitudinal consequences of variation in conifer leaf life span. In: Smith ecotypes of mountain birch to temperature and fertilization. WK, Hinckley TM (eds) Ecophysiology of coniferous forest. Oecologia 119: 16-23 Academic Press, San Diego, Calif., pp 225-254 Woodward FI (1986) Ecophysiological studies on the shrub Vac- Reich PB, Oleksyn J, Modrzynski J, Tjoelker MG (1996) Evidence cinizlm myrtillus L. taken from a wide altitudinal range. Oeco- that longer needle retention of spruce and pine populations at logia 70:580-586 high elevations and high latitudes is largely a phenotypic re- Yin X (1993) Variation .in foliage nitrogen concentration by forest sponse. Tree Physiol 16:643-647 type and climatic gradients in North America. Can J For Res Reich PB, Walters MB, Ellsworth DS (1997) From tropics to 23: 1587-1602 tundra: global convergence in plant functioning. Proc Natl Zhong X, Wu N, Luo J, Yin K, Tang Y, Pan Z (1997) Researches Acad Sci USA 94:13730-13734 of the forest ecosystems on Gongga Mountains (in Chinese). Reich PB, Ellsworth DS, Walters MB, Vose JM, Gresham C, Volin Chengd~~Science and Technology University Press, Chengdu, JC, Bowman WD (1999) Generality of leaf trait relationships: a pp 1-171 test across six biomes. Ecology 80: 1955-1969 Ren H, Peng S (1997) Comparison of methods for estimating leaf area index in Dinghushan forests (in Chinese). Acta Ecol Sin 17(2):220-223