Stem hydraulics mediates leaf water status, carbon gain, nutrient use efficiencies and plant growth rates across dipterocarp species


*Correspondence author. E-mail:


  • 1Stem vascular system strongly influences structure and functioning of leaves, life-history, and distribution of plants. Xylem structure and hydraulic conductivity of branches, leaf functional traits, and growth rates in 17 dipterocarp species in a mature plantation stand were examined to explore the functional relationships between these traits.
  • 2Maximum hydraulic conductivity on the bases of both sapwood and leaf area (kL) were positively correlated with midday leaf water potential in the rainy season, stomatal conductance, area-based maximum photosynthetic rate, photosynthetic N (PNUE) and P use efficiencies (PPUE), and mean height and diameter growth rates. Moreover, kL was positively correlated with mesophyll thickness and mass-based maximum photosynthetic rate. These results revealed the mechanistic linkage between stem hydraulics and leaf photosynthesis through nutrient use efficiency and mesophyll development of leaves.
  • 3A detrended correspondence analysis (DCA) using 37 traits showed that the traits related to stem hydraulics and leaf carbon gain were loaded on the first axis whereas traits related to light harvesting were loaded on the second axis, indicating that light harvesting is a distinct ecological axis for tropical canopy plants. The DCA also revealed a trade-off between photosynthetic water use efficiency and hydraulic conductivity along with PNUE and PPUE.
  • 4The congeneric species were scattered fairly close together on the DCA diagram, indicating that the linkages between stem hydraulics, leaf functional traits, and plant growth rates are phylogenetically conserved.
  • 5These results suggest that stem hydraulics mediates leaf water status, carbon gain, nutrient use efficiencies, and growth rates across the dipterocarp species. The wide variation in functional traits and growth rates among these dipterocarp species along with the trade-offs mentioned above provide a possible explanation for their co-existence in tropical forest communities.


The xylem vascular system strongly influences life-history strategy (West et al. 1999) and distribution of plants (Brodribb & Hill 1999; Engelbrecht et al. 2007) owing to its effects on water and nutrient transportation from soil to leaves, resistance to xylem embolism induced by drought and freezing temperatures, and mechanical strength. The difference in water potential between soil and leaves is largely determined by the conducting vascular system through its regulation of water transportation from roots through stems into leaves where the water is transpired via the stomata (Meinzer 2003). Leaf is a major bottle-neck of plant hydraulic conductivity and contributes about 30% of whole plant hydraulic resistance (Sack et al. 2003). If soil water supply is ample, efficient stem hydraulic conductivity allows quick water transport into leaves to compensate leaf transpiration, resulting in high leaf water potential during the day. Stem hydraulics has therefore been found to be related to minimum daily leaf water potential in the rainy season for chaparral shrubs (Ackerly 2004) and tropical rainforest trees (Santiago et al. 2004). The strong linkage between stem hydraulic conductivity and stomatal conductance of leaves has also been demonstrated by experimental manipulations during which hydraulic conductivity was reduced through embolism injection (Hubbard et al. 2001), root chilling (Brodribb & Hill 2000) and root pruning (Meinzer & Grantz 1991). High stomatal conductance allows rapid CO2 diffusion into carboxylation sites, and consequently strong photosynthetic capacity. Species with strong photosynthetic capacity are generally fast-growing (Poorter & Bongers 2006) and consequently have low wood density (Castro-Díez et al. 1998; King et al. 2006).

Plant functional traits reflect plant adaptations to specific environments and utilization of resources, and therefore provide valuable information for the analyses of community assembly (McGill et al. 2006; Shipley et al. 2006; Ackerly & Cornwell 2007) and ecosystem functioning and services (Díaz et al. 2007). Recent studies have shown global convergence in leaf functional traits among species from diverse biomes. For example, maximum photosynthetic rate, stomatal conductance, and leaf area per unit mass are positively correlated, whereas they are negatively correlated with leaf life span across a range of sites and angiosperm taxa (Reich et al. 1997). Several recent studies have shown that stem hydraulics is correlated with leaf photosynthetic traits and photosynthetic water and nutrient use efficiencies (Brodribb & Feild 2000; Santiago et al. 2004; Ishida et al. 2008). Leaf hydraulic conductivity is also found to be correlated with the development of leaf mesophyll and stomata area index (Sack et al. 2003; Sack & Frole 2006), both of which influence photosynthesis as the mesophyll structure affects light harvesting and transmission and CO2 diffusion within leaves and the stomata geometry affects gas exchange between air and leaves. However, whether the hydraulic capacity of leaf-supporting stems influences leaf anatomy is poorly known.

Dipterocarpaceae is the most important tree family both ecologically and commercially in Asian tropical forests. The species of this family occur in tropical wet and seasonal rainforests, heath and peat swamp forests, and some occur as far north as the Tropic of Cancer in southern China (Ashton 1964; Appanah & Turnbull 1998; Cao 2000). They vary greatly in stature, with species from the genera Shorea and Hopea usually being tall emergent and upper canopy components whereas those from the genera Vatica and Dipterocarpus being short-statured components of Asian tropical forests. Dipterocarps also vary largely in light requirement with the majority being shade-tolerant but a large number being light-demanding (Appanah & Turnbull 1998). However, information on the functional traits particularly hydraulic properties of this prominent family is scarce.

In this study, we characterized stem xylem structural properties, hydraulic conductivity, leaf functional traits, and plant growth rates in 17 dipterocarp tree species grown in a mature plantation stand, which were introduced from southern China and adjacent tropical countries. Since these plants were grown in a common garden with the same environment, the differences in plant traits and growth across species can be attributed to inherited adaptive responses of the plants (Monson 1996). We attempted to answer two questions: (i) how do leaf functional traits, stem hydraulics, and stem diameter and height growth rates vary among these dipterocarp species? (ii) Is stem hydraulics correlated with leaf structural and functional traits and growth rates? This study could provide important information on the functional association between stems and leaves, the ecology of dipterocarps, the mechanism of their co-existence and functioning in tropical forest ecosystems, and their use for reforestation.


study site and species

This study was conducted in the dipterocarp plantation stand at Xishuangbanna Tropical Botanical Garden (21°41′ N, 101°25′ E, altitude 570 m) in southern Yunnan Province, China. The garden is surrounded by the Luosuo River, a tributary of the upper Mekong River. Mean annual temperature is 21·7 °C and mean annual precipitation is 1560 mm with 80% occurring from May to October. A distinct dry period lasts from November to April. The soil of the stand is sandy alluvium, containing 0·875 mg g−1 N, 0·329 mg g−1 P and 9·693 mg g−1 K at 0–20 cm depth.

Since 1980s, seeds of about 40 dipterocarp species from southeastern Asia including southern China have been collected and germinated to seedlings in the botanical garden. The 1-year-old seedlings of these species were planted in a common stand of about 7 ha, with a density of about 1100 trees per ha. Seventeen dipterocarp species (Table 1) from this stand were selected for the study based on the following criteria: (i) whether they originated from China and adjacent countries (11 species from southern China, one species from northern Vietnam and five species from Thailand; Fig. 1); (ii) whether they grew near paths to allow access to the canopy; and (iii) whether they were sun-exposed and not overtopped by neighbouring trees. These species belong to 6 genera, that is, Anisoptera (1 species), Dipterocarpus (5 species), Hopea (4 species), Parashorea (1 species), Shorea (3 species) and Vatica (3 species). At the generic level, Hopea, Shorea and Parashorea are phylogenetically closely related (see Fig. S1 in Supporting Information). Among the study species, 16 species are evergreen species whereas Dipterocarpus tuberculatus is a deciduous species with a leafless period of 1·5 months from February to March. All study species are deep-rooted emergent or canopy layer components in primary forests. In 2005 the ages of the sampled trees were 12–24 years, the heights were 9–29 m, and the diameters at breast height (1·3 m height d.b.h.) were 10·5–47·8 cm.

Table 1.  The age, height and diameter at breast height (d.b.h.) of the sampled trees, and the maximum height of the 17 dipterocarp species
SpeciesSpecies codeAge (years)Height of sampled trees (m)D.b.h. of sampled trees (cm)Maximum height (m)
  1. Nomenclature and maximum height of the species follow the Flora of China (Tong & Tao 1990) and Flora of Vietnam (Ho 1999). Data are means ± SE, N = 10.

Anisoptera costata Korth.Ac1318·1 ± 0·425·7 ± 1·140
Dipterocarpus alatus Roxb. ex G. DonDa2429·4 ± 0·747·8 ± 5·245
D. intricatus DyerDi2418·1 ± 1·026·8 ± 2·230
D. retusus Bl.Dr1819·0 ± 1·122·0 ± 1·845
D. tuberculatus Roxb.Dtb2423·3 ± 1·430·2 ± 2·925
D. turbinatus Gaertn. f.Dtr1617·7 ± 0·723·1 ± 2·035
Hopea chinensis Hand.-Mazz.Hc129·2 ± 0·610·5 ± 0·420
H. hainanensis Merr. et ChunHha1717·3 ± 0·520·9 ± 0·720
H. hongayensis Tard.-Blot.Hho1513·0 ± 0·313·4 ± 1·030
H. mollissima C.Y. WuHm2411·7 ± 0·217·6 ± 3·430
Parashorea chinensis Wang HsiePc2016·8 ± 0·517·6 ± 0·960
Shorea assamica DyerSa148·7 ± 0·522·4 ± 1·250
S. robusta Gaertn. f.Sr1514·3 ± 0·420·6 ± 0·640
S. spp.Ss2416·1 ± 1·224·4 ± 2·830
Vatica guangxiensis X.L. MoVg1310·1 ± 1·011·2 ± 1·040
V. mangachapoi Bl.Vm169·3 ± 0·711·2 ± 1·020
V. xishuangbannaensis G.D. Tao et J.H. ZhangVx2016·5 ± 1·019·7 ± 1·440
Figure 1.

The location of the study site (solid black circle) and the sites (open stars) where the 17 dipterocarp species were collected. XTBG: Xishuangbanna Tropical Botanical Garden (21°41′ N, 101°25′ E); I: Bangkok (13°45′, 100°30′); II: Yingjiang (24°52′, 97°55′), Yunnan; III: Mengla (21°25′, 101°32′), Yunnan; IV: Hekou (22°32′, 103°57′), Yunnan; V: North Vietnam (22°00′, 105°25′); VI: Napo (23°22′, 105°52′), Guangxi: VII: Shangsi (22°10′, 108°00′), Guangxi; VIII: Jianfengling (18°40′, 108°49′), Hainan island. See the species codes in Table 1.

stem hydraulic conductivity and anatomical properties

In the rainy season of 2005, four to six sun-exposed terminal branches from three to four trees per species were excised from the upper canopy in the morning. They were re-cut in water and transported to the laboratory to determine maximum hydraulic conductivity (kh) following the method described by Santiago et al. (2004). De-ionized water under a pressure of 0·2 MPa was pumped through a 30-cm-long shoot segment for 15–20 min to remove embolisms, then kh was measured as the water flow rate through the shoot segment under a low gravitational pressure (5 kPa) generated by a hydraulic head of 50 cm (Sperry et al. 1988). Maximum stem sapwood specific conductivity (kS) was calculated as the ratio of kh to the cross-sectional area of the sapwood. Sapwood thickness of a stem segment was determined using a dye solution after the measurement of kh. Maximum leaf specific hydraulic conductivity (kL) was calculated as the ratio of kh to the total leaf area distal to the stem. Leaf area was measured using a portable leaf area meter (LI-3000A, Li-Cor, Lincoln, NE).

Because shoot morphology varies largely among the study species, the 30-cm-long shoot segment was used for the hydraulic conductivity measurement for all species to facilitate the comparison. Although these stem segments may contain some open vessels (cf. Brodribb & Holbrook 2003; maximum vessel length of some tropical rainforest tree species ranged from 15 to 35 cm), our measurements on the hydraulic conductivity should provide a good basis for comparison between the species in this study. Certainly, care should be taken when comparing values of kS and kL presented here with those measured using different methods in other studies.

After the hydraulic measurement, the fresh volume of a small sapwood sample was measured by displacement of water and then the sapwood was dried at 80 °C for 48 h to calculate sapwood density. Some sapwood samples were stored in 1 : 1 (v/v) ethanol : glycerol for 2 months and then transverse sections of about 5 µm thickness were made with a microtome. Vessel anatomy was examined with the aid of a microscope. The number and diameters of vessels in a field were measured under 10 × and 40 × objectives calibrated with an ocular micrometer. Vessels of the dipterocarp species examined were elliptical (see Fig. 2) and thus average vessel diameter in a field was calculated according to the method described by Becker et al. (2003).

Figure 2.

Microscopy images of transverse sections of stem xylem for (a) Dipterocarpus retusus, (b) Hopea hongayensis, (c) H. mollissima and (d) Shorea assamica, as examples of xylem anatomy for dipterocarps. Scale bar = 100 µm.

leaf anatomy

Thickness of leaf, upper and lower epidermis, and palisade and spongy mesophylls were measured on transverse sections of leaves using the light microscope. The ratio of palisade to spongy mesophyll thickness was calculated. Stomatal density and guard cell length on abaxial surfaces of leaves were measured from the epidermal impressions made with colourless nail polish. Stomatal pore area index was calculated as SPI = stomatal density × guard cell length2 (Sack et al. 2003). Six leaves from 4–6 plants per species were used and at least three fields of each leaf were observed. We observed that all study species had heterobaric leaves, that is, with bundle sheath extensions (Kenzo et al. 2007).

gas exchange and water status

In the rainy season, we accessed the upper canopies of the trees using a crane mounted on a truck. Between 08.30 and 11.00 h area-based maximum photosynthetic capacity (Aa) and stomatal conductance (gs) were measured from 3 to 4 sunlit canopy leaves from each branch of 4–6 plants per species using a portable photosynthetic gas exchange system (LI-6400, Li-Cor). Prior to the measurements, leaves were fully induced by sunlight or by an artificial light with photosynthetic photon flux density of 1500 µmol m−2 s−1 for 10 min provided by a LED light source. CO2 concentration inside the leaf chamber was maintained at 380 µmol mol−1 through the CO2 controlling system of the gas analyzer with attachment of a tiny CO2 cylinder. During the measurements inside the leaf chamber the relative air humidity was 47–54%, leaf-to-air vapor pressure deficit was 1·5–2·0 kPa, and leaf temperature was 27–30 °C. The intrinsic photosynthetic water use efficiency (WUE) was calculated as Aa/gs. The average from the repeated measurements of each plant was used to represent the gas exchange value for one plant.

At midday on clear days in the rainy season, three to seven fully-expanded sunlit leaves from four to five sampled trees per species were collected from the upper canopies reached by the crane and then midday leaf water potentials (Ψmd) were measured using a pressure chamber (SKPM 1400, Skye Instruments, Powys, UK).

lma and foliar chlorophyll and nutrient concentrations

After the gas exchange measurements the measured leaves were harvested and each leaf was cut into two parts along the midrib. One half of the leaf was used to measure chlorophyll concentration using 80% acetone according to Johnston et al. (1984). The projection area of the other half was measured with the leaf area meter (LI-3000A, Li-Cor), and then the leaves were dried at 80 °C for 48 h to determine the leaf mass per unit area (LMA). Leaf density was calculated as LMA/leaf thickness.

The remaining harvested leaves were used to determine foliar concentrations of N, P and K. Total N concentration was determined using an auto Kjeldahl unit (K370, BÜCHI Labortechnik AG, Flawil, Switzerland) after the leaf samples were digested with concentrated H2SO4. Total foliar P and K concentrations were analysed using an inductively coupled plasma atomic-emission spectrometer (IRIS Advantage-ER, Thermo Jarrell Ash Corporation, MA) after the samples were digested with concentrated HNO3–HClO4. Photosynthetic N (PNUE) and P use efficiencies (PPUE) were calculated as the ratio of mass-based maximum photosynthetic capacity (Am) to foliar N and P concentrations, respectively.

growth rates

Height from ground to tree tops and d.b.h. of ten individuals per species were measured with tapes. Average height (HGR) and diameter growth rates (DGR) were calculated by dividing height and d.b.h. by tree age.

statistical analysis

Relationships between stem hydraulics, leaf-level traits, and plant growth rates were analysed with Pearson's correlation. The associations between 37 plant traits listed in Table 2 and between the dipterocarp species studied were tested by a detrended correspondence analysis (DCA) using DECORANA function in R (R v. 2·6·2; The R Foundation for Statistical Computing, Vienna, Austria). Some traits (e.g. ratio of palisade to spongy mesophyll, layers of palisade) deviated significantly from the normal distribution (Shapiro-Wilk W test), even after they were log-transformed. Therefore, for our data DCA was more suitable than principle component analysis (Hill 1979).

Table 2.  Summary of traits used in this study
Group, traitCodeUnitMeanCoefficient of variation (%)
Wood and xylem
 Vessel densityVdeno. mm−270·048
 Vessel diameterVdiµm66·128
 Sapwood densityWDg cm−30·54316
 Maximum sapwood specific hydraulic conductivitykSkg m−1 s−1 MPa−18·8739
 Maximum leaf specific hydraulic conductivitykL×10−4 kg m−1 s−1 MPa−18·5745
 Ratio of leaf area to sapwood areaL/Sm2 cm−21·2128
Leaf anatomy and morphology
 Leaf thicknessLTµm19027
 Thickness of upper epidermisTCµm23·739
 Thickness of palisade mesophyllTPµm61·329
 Thickness of spongy mesophyllTSµm88·535
 Mesophyll thickness (= TP + TS)MTµm15027
 Thickness of lower epidermisTLµm16·540
 Ratio of palisade to spongy mesophyllP/S 0·79855
 Layer of palisade mesophyllLPno.1·64748
 Stomatal densitySDno. mm−247035
 Guard cell lengthGCLµm18·713
 Stomatal pore area index (= SD × GCL2)SPI 0·16644
 Leaf mass per unit areaLMAg m−290·427
 Leaf densityLDkg m−348419
Leaf nutrients
 Mass-based N concentrationNmmg g−119·712
 Mass-based P concentrationPmmg g−11·2216
 Mass-based K concentrationKmmg g−15·5925
 Area-based N concentrationNag m−21·7524
 Area -based P concentrationPag m−20·10927
 Area -based K concentrationKag m−20·49331
 Ratio of N to P concentrationN/P 16·310
 Mass-based chlorophyll concentrationChmmg g−15·5027
 Area-based chlorophyll concentrationChaµg cm−247·825
 Ratio of chlorophyll a/ba/b 2·3414
Leaf water potential
 Midday leaf water potential in the rainy seasonΨmdMPa–0·61137
Gas exchange and leaf function
 Stomatal conductancegsmol m−2 s−10·23350
 Area-based maximum photosynthetic rateAaµmol m−2 s−111·739
 Mass-based maximum photosynthetic rateAmnmol g−1 s−113233
 Photosynthetic N use efficiencyPNUEµmol mol−1 s−193·528
 Photosynthetic P use efficiencyPPUEmmol mol−1 s−13·2628
 Photosynthetic water use efficiencyWUEµmol mol−152·517
 Mean annual height growth rateHGRm year−10·87926
 Mean annual diameter growth rateDGRcm year−11·2132


The dipterocarp species varied largely in wood and leaf traits (Table 2; see Fig. 2 for xylem anatomy). For example, sapwood density varied 1·9-fold among species (0·342–0·655 g cm−3), kL 4·5-fold (3·1–13·9 × 10−4 kg m−1 s−1 MPa−1), LMA 2·9-fold (52–152 g m−2), mass-based N concentration (Nm) 1·7-fold (16·2–26·8 mg g−1), gs 6·5-fold (0·085–0·558 mol m−2 s−1), Aa 3·9-fold (5·1–20·3 µmol m−2 s−1), PNUE 2·5-fold (54–135 µmol mol−1 s−1), HGR 2·9-fold (0·49–1·39 m y−1), and DGR 2·9-fold (0·70–1·99 cm y−1) (see Table S1).

Stem hydraulic conductivity was highly correlated with vessel diameter (Fig. 3) and also correlated with a suite of leaf traits among species (see Table S2). Specifically, both kS and kL were positively correlated with Ψmd in the rainy season (Fig. 4b), gs (Fig. 4 c), Aa (Fig. 4d), PNUE (Fig. 4e), and PPUE (Fig. 4f). Moreover, both kS and kL were positively correlated with HGR (Fig. 4g) and DGR (Fig. 4h), with stronger correlation with DGR than with HGR. Stem kS was negatively correlated with WUE, and kL was positively correlated with thickness of mesophyll (Fig. 4a) and lamina, and Am (see Table S2). Stem hydraulic conductivity was not significantly correlated with LMA and stomatal density or SPI (see Table S2). Wood density was not significantly correlated with hydraulic conductivity, vessel diameter and density, and growth rates.

Figure 3.

Correlation between maximum sapwood specific hydraulic conductivity (kS) and vessel diameter across the 17 dipterocarp species. Species symbols: inline image, Anisoptera costata; inline image, Dipterocarpus alatus; inline image, D. intricatus; inline image, D. retusus; inline image, D. tuberculatus; inline image, D. turbinatus; inline image, Hopea chinensis; inline image, H. hainanensis; inline image, H. hongayensis; inline image, H. mollissima; inline image, Parashorea chinensis; inline image, Shorea assamica; inline image, S. robusta; inline image, S. spp; inline image, Vatica guangxiensis; inline image, V. mangachapoi; inline image, V. xishuangbannaensis. ***P < 0·001.

Figure 4.

Correlations between maximum leaf specific hydraulic conductivity (kL) and: (a) mesophyll thickness (MT); (b) midday leaf water potential (Ψmd) in the rainy season; (c) maximum stomatal conductance (gs); (d) area-based maximum photosynthetic rate (Aa); (e) photosynthetic N use efficiency (PNUE); (f) photosynthetic P use efficiency (PPUE); (g) height growth rate (HGR); and (h) diameter growth rate (DGR) across the 17 dipterocarp species. The symbols for the species are noted in Fig. 3. *P < 0·05; **P < 0·01, ***P < 0·001.

Interestingly, several traits were correlated with maximum height (Hmax) of the species as reported in literature, such as kS (r = 0·52, P < 0·05), WUE (r = −0·73, P < 0·001), thickness of spongy mesophyll (r = −0·55, P < 0·05), and the ratio of palisade to spongy mesophyll thickness (r = 0·66, P < 0·01). With increasing leaf thickness, area-based N, P, K, and chlorophyll concentrations increased (see Table S2). There were positive correlations between gs and Aa, Am, PNUE, and PPUE. With increasing LMA, Nm and mass-based chlorophyll concentration decreased. Both HGR and DGR were positively correlated with Nm, Aa, Am, PNUE and PPUE.

The associations between stem xylem anatomical properties, hydraulic conductivity, leaf-level traits, and plant growth rates were further analysed using DCA (Fig. 5a). The first axis of the DCA reflects the continuum in hydraulic conductivity vs. vessel density. The negative side of the axis represents the species with high hydraulic conductivity, which was characterized by large vessels, high kS, kL, Ψmd, gs, Aa, thick mesophyll, and rapid growth rates. The positive side of the axis indicates the species with small diameter vessels, dense vessel packing, and high sapwood density. In other words, the first DCA axis was correlated with traits related to stem hydraulics and leaf carbon gain. The second axis of the DCA represents species possessing leaves with high chlorophyll concentration, larger palisade to spongy mesophyll ratio and low LMA on the positive side, which are related to light harvesting and interception. Congeners from Shorea, Dipterocarpus, Hopea and Vatica were scattered fairly close together in the DCA diagram (Fig. 5b).

Figure 5.

Detrended correspondence analysis (DCA) using mean values of plant traits across 17 dipterocarp tree species. Plant trait (a) and species (b) loadings on the first and second axes. See the explanations of the trait abbreviations in Table 2 and the species codes in Table 1.


The present study showed that stem hydraulic conductivity of dipterocarps is correlated with a large suite of leaf structural and functional traits such as leaf mesophyll thickness, maximum photosynthetic gas exchange rates, and photosynthetic water (negative) and nutrient use efficiencies (positive) (Fig. 4; see Table S2). Several previous studies have also shown the correlation between stem hydraulic conductivity and leaf photosynthetic capacity (Brodribb & Feild 2000; Santiago et al. 2004; Ishida et al. 2008). This relationship is likely mediated through stomatal regulation (Meinzer & Grantz 1991; Brodribb & Feild 2000; Hubbard et al. 2001) to balance transpiration, photosynthesis, and hydraulic dysfunction. Although the leaf is a major bottle-neck in the whole plant hydraulic conductivity (Sack et al. 2003), stomatal regulation links stem and leaf hydraulic systems. Under optimal soil water supply, high stem hydraulic conductivity allows quick water transport from stems into leaves to compensate leaf transpiration water loss and consequently maintain high daily water potential (Meinzer 2003; Ackerly 2004; Santiago et al. 2004). High daily leaf water potential mitigates the stomatal limitation to gas exchange, especially during the midday when the transpiration demand is high, and thus increase daily carbon assimilation. Maintenance of high leaf water potential should also contribute to maintaining a high cell turgor, and thus benefit the plant cell expansion and tree growth (Koch et al. 2004; Woodruff et al. 2004). Therefore, an efficient water supply to the upper canopy could buffer the high xylem tension in tall trees. This mechanism is also considered to support the theory of hydraulic limitation to tree height (Ryan & Yoder 1997), and could explain the positive correlation between ks and maximum tree height found in the present study.

The link between photosynthetic capacity and stem or leaf hydraulic conductivity seems obvious because stem and leaf hydraulic systems carry water to replace the water lost in the leaves during photosynthetic gas exchange (Brodribb & Feild 2000; Brodribb et al. 2002). However, both stem hydraulic traits and maximum photosynthetic rates are adapted to long-term circumstances and the physiological mechanisms explaining the coordination between these two systems and the involvement of nutrient supply in this process are poorly understood. Our results provide a better insight into the coordination between hydraulic and photosynthetic traits in terms of nutrient supply and nutrient use efficiency. Higher nutrient availability or addition of nutrients has been found to enhance water transport efficiency and carbon assimilation rates in Neotropical savanna trees (Bucci et al. 2006). When soil nutrients are comparable, species with higher photosynthetic nutrient use efficiency will enhance development of hydraulic and photosynthetic systems, resulting in higher growth rates and probably higher maximum tree height. However, better water supply leads to a luxurious use of water and hence a low WUE. Consistent with the present study, other studies have also reported such a trade-off between photosynthetic nutrient and water use efficiency (Santiago et al. 2004; Cai et al. 2007).

The positive correlation of stem hydraulics with leaf or mesophyll thickness (Fig. 4a) might be partly due to the functional linkage with the leaf vascular system. For example, species with thick mesophyll have dense veins (Sack & Frole 2006). As discussed above, efficient water transportation in stems and leaves allows the maintenance of high leaf water potential and leaf turgor which benefits mesophyll development. Increased nutrient use efficiency with increasing stem hydraulic conductivity may also help mesophyll cell expansion and growth. Moreover, the dipterocarp species of the present study all possess heterobaric leaves, that is, with extended bundle sheath (Kenzo et al. 2007). Therefore, their leaf thickness is approximately the width of the vascular bundles. Dense veins and wide vascular bundles of the leaves allow the rapid diffusion of water and nutrients within leaves (Sack & Frole 2006), which coincides with the positive correlation between stem hydraulic conductivity and nutrient use efficiency.

The large variation in functional traits, including photosynthesis and hydraulics, across the dipterocarp species studied (Table 2) is consistent with the established knowledge on the wide ecological adaptations of dipterocarps (Ashton 1964). The range in wood density of the present dipterocarp species was relatively narrow, whereas photosynthetic gas exchange rates and stem hydraulic conductivity varied widely among species (see Table S1). This is in agreement with the wide variation in vessel traits and a relatively narrow range of wood density in 51 Californian angiosperm woody species (Preston et al. 2006). In addition, these authors have found that vessel traits were related to Hmax of the species. Although the dipterocarp species of the present study originated from different regions, their performance in a common environment may discern their inherited differences in adaptations to the environment, and in growth rate and resource use. The Hmax in their native habitats was positively correlated with kS and negatively with WUE, implying the importance of hydraulic traits in relation to tree Hmax (Ryan & Yoder 1997; Koch et al. 2004; Woodruff et al. 2004). It was also positively correlated with the ratio of palisade to spongy mesophyll thickness and negatively with thickness of spongy mesophyll. High kS, a large ratio of palisade to spongy mesophyll and lower WUE are usually characteristics of light-demanding tree species (Swaine & Whitmore 1988). Our results therefore also support the idea that Hmax of tropical forest trees is a light capture adaptation (Thomas 1996), suggesting a mechanistic link between light capture characteristics and hydraulic traits (Campanello et al. 2008). Vatica xishuangbannaensis, a relatively short dipterocarp species, had relatively lower photosynthesis and growth rate (see Table S1), and is a typical shade-tolerant species (Zhu 2000). It is also relatively drought-tolerant as it occurs on relatively dry upper slopes (Zhu 2000). However, being the tallest species in this study, Parashorea chinensis is also shade-tolerant. Its seedlings can grow and survive in the shaded forest understorey though they grow much better in high irradiance sites such as canopy gaps and secondary forest stands (Zhu 2000; Tang 2008). Fast-growing species have a better capability of adjusting tree hydraulics to different light regimes (Campanello et al. 2008) and P. chinensis may be an example of the fast-growing species with large plasticity in hydraulic traits.

In the present study wood density of terminal branches was not significantly correlated with hydraulic conductivity, vessel diameter and vessel density (see Table S2). This supports the idea that wood density and vessel traits are two distinct ecological axes (Preston et al. 2006). Wood density of angiosperms is largely determined by wall thickness of vessels and density of fibre cells and therefore strongly related to the mechanical strength and capacity to prevent vessel implosion induced by the pressure gradient between actively conducting and embolized vessels (Hacke et al. 2001). Hydraulic conductivity of stems, on the other hand, is largely determined by vessel size and density (James et al. 2003). Hydraulically, wood density is inversely correlated to water storage capacity (Bucci et al. 2004; Scholz et al. 2007; Meinzer et al. 2008). In dry ecosystems like Brazilian savanna, high water storage capacity provides a better buffering to the water transport system and consequently would allow high water transport efficiency indicated by high kS (Bucci et al. 2004; Scholz et al. 2007). This indirect correlation between wood density and hydraulic conductivity, however, may not show up in environments with optimal soil water supply because of the relatively lower demand on the stored water in buffering seasonal and daily water deficits. Wood density of the terminal branches was not significantly correlated with mean diameter or height growth rates (see Table S2). This is in contrast with the trade-off between allocation of biomass to tissue density and growth rate suggested by an allometric model (Enquist et al. 2000), and the finding of a significant inverse relationship between mean diameter growth and wood density among woody species in primary tropical rain forests (King et al. 2006). In our case, the ages of the terminal branches were only 1–2 years whereas the growth rates were averaged over past 12–24 years, which could obscure the correlation between them. Nevertheless, both diameter and height growth rates of the dipterocarps were positively correlated with vessel diameter, while height growth rate was negatively correlated with vessel density (see Table S2). It appears that the vessel size and density which are determined by cambium growth and related to hydraulic conductivity regulate the diameter growth rate. The relationship between height growth and vessel diameter and density probably reflects the allometric relationship between cambium growth and apical meristematic growth.

A trade-off for the species with high stem hydraulic conductivity, as in those with tall stature, is the low photosynthetic water use efficiency (see Table S2) which could limit these plants to establish successfully in dry habitats. This trade-off as well as the trade-off between photosynthetic nutrient and water use efficiencies discussed above combined with the large variation in functional traits and growth rates could allow niche differentiation among the dipterocarp species and consequently their co-existence in a forest community or a landscape with heterogeneous environments (Meinzer 2003).

The DCA results (Fig. 5a) revealed that light harvesting of leaves (i.e. chlorophyll concentration) is a distinct ecological axis, and was mainly regulated by leaf structure, for example, LMA (also see Poorter & Evans 1998; Cao 2000), among the dipterocarp species. Because hydraulic conductivity may affect the development of leaf mesophylls as mentioned above, it is likely that there is an indirect relationship between light harvesting and water transport capacity of plant. The irradiance over the canopy leaves of tropical forests is usually intense and particularly so in the dry season; therefore, it is important to optimize light harvesting to balance the light energy used for photosynthesis and surplus energy that may induce photoinhibition.

On the DCA diagram, the cogeneric species were packed fairly close together, with species from Vatica scattered at the opposite end of the scatter plot in comparison with Dipterocarpus and Shorea (Fig. 5b). This spatial distribution of the dipterocarp species in the DCA diagram corresponds to their positions in the phylogenetic consensus tree constructed using chloroplast DNA sequences (Kajita et al. 1998; Gamage et al. 2003; Li et al. 2004; also see Fig. S1). These results suggest that stem hydraulic characteristics and their association with leaf functional traits and plant growth are phylogenetically conserved. Phylogenetic conservatism of hydraulic properties has also been reported by other studies (Preston et al. 2006; Hao et al. 2008; Willson et al. 2008), indicating large evolutionary constraints on the changes in these traits.


We thank two anonymous referees, our colleagues J.W. F. Slik, G.-Y. Hao, and Y.-J. Zhang for their helpful comments on our manuscript. The Biogeochemistry Laboratory of our botanical garden made the analyses of soil and foliar nutrient concentrations of the present study. This study was financially supported by the National Natural Science Foundation of China (grant No. 90302013).