Contrasting patterns of leaf water potential and gas exchange responses to drought in seedlings of tropical rainforest species



  • 1 The leaf gas exchange and water potential (Ψw) responses to a soil drying cycle and to increasing atmospheric water vapour pressure deficit were compared in seedlings of three tropical rainforest canopy species in a growth chamber.
  • 2Eperuafalcata Aub. presented an anisohydric behaviour in relation to soil drought. The decrease in predawn Ψw at the beginning of the drying cycle was accompanied by a more pronounced decrease in midday Ψw. Stomatal closure occurred from a volumetric soil water content (Θ) of 0·1 m3 m−3, which corresponded to approximately one third of the available soil water.
  • 3 Diplotropispurpurea (Rich.) Amsh. displayed an isohydric behaviour. Midday Ψw did not decrease – and the difference between predawn and midday Ψw did not increase – over a large range of Θ (to less than 0·1 m3 m−3). Stomatal conductance (gs) was more sensitive to soil drought than in Eperua.
  • 4 A unique behaviour was found in Virola michelii: Heckel. Midday Ψw remained remarkably stable, and close to predawn Ψw, over the drying cycle as long as leaves did not dry, and gs was extremely sensitive to soil drought. These results led us to postulate the existence of a homeostatic regulation of Ψw in Virola.
  • 5 The three species differed in their stomatal sensitivity to atmospheric drought. Eperua was the least and Diplotropis the most sensitive. Virola, even though extremely sensitive to soil drought, displayed an intermediate sensitivity of gs to atmospheric drought.
  • 6 These findings suggest that the survival of Eperua, Diplotropis and Virola in the tropical rainforest of French Guiana may depend partly on local hydrological conditions.


A major finding in tropical rainforest ecophysiology over the past decade is that tree species within a given habitat (e.g. upper canopy trees) differ greatly in their leaf gas-exchange regulation, pointing to an important aspect of functional diversity. Studies based on direct CO2- and H2O-exchange measurements, or on indirect assessments by the carbon isotope discrimination (Δ) approach (Farquhar, O’Leary & Berry 1982), have demonstrated a huge diversity among species (Bonal et al. 2000b; Hogan, Smith & Samaniego 1995; Huc, Ferhi & Guehl 1994; Guehl et al. 1998; Meinzer et al. 1993). Δ screening studies carried out on more than 100 canopy tree species in rainforests of French Guiana suggested that estimated intrinsic water-use efficiency (ratio of CO2 assimilation rates to stomatal conductance of water vapour, A/gs) varied over a threefold range (Bonal et al. 2000b; Guehl et al. 1998). The ecological interpretation of these results remains incomplete.

Tropical tree species may also differ in their gas-exchange sensitivity to either atmospheric or soil drought (Franks, Cowan & Farquhar 1997; Hogan et al. 1995; Huc et al. 1994; Meinzer et al. 1993; Roberts et al. 1990). The mechanisms underlying stomatal responses to drought remain unclear, even though both hydraulic and metabolic root-to-shoot signals act as triggers for stomatal closure (see reviews in Davies & Zhang 1991; Jones 1998; Schulze 1993; Whitehead 1998).

In a previous field study with three tree species growing in monospecific plantations (Bonal et al. 2000a), leaf gas exchange and water potential were compared in the wet and the dry season. Eperua falcata Aub. (Caesalpiniaceae), a species with leaf δ13C of −28·6‰, had similar gs and minimum leaf water potential (Ψwm) in both seasons. Diplotropis purpurea (Rich.) Amsh. (Caesalpiniaceae), a species with very negative δ13C (−30·9‰), partly closed its stomata in the dry season, while Ψwm increased slightly from wet to dry season. A unique behaviour was observed in Virola surinamensis Warb. (Myristicaceae), a species with intermediate leaf δ13C (−29·9‰). Extremely high Ψw (about −0·30 MPa) were observed during the day in both seasons, Ψwm being very close to predawn water potentials (Ψwp). Stomatal conductance was high in the wet season, whereas stomata were almost completely closed in the dry season.

It is important to clarify whether these three distinct behaviours can be interpreted as response mechanisms to soil and atmospheric drought. As compared to the field study by Bonal et al. (2000a), in the controlled environment study reported here we were able (1) to separate the effects of soil and atmospheric drought; and (2) to follow the dynamic response of leaf gas exchange and Ψw to increasing soil drought.

Materials and methods

Plant material

In this study, V. surinamensis was replaced by Virola michelii, which has similar δ13C and Ψwm responses to environment (Bonal et al. 2000b). The three species studied are abundant in the tropical rainforest of French Guiana. Eperua is a shade hemitolerant species, whereas Virola and Diplotropis are heliophilic species (Favrichon 1994). Seeds of the three species were collected in April 1997 in the natural forest of Paracou, French Guiana (5°20′ N, 52°50′ W). They were sown in 22 l cylindrical plastic bags filled with a sandy loam forest soil (A horizon). Plants were installed in a open-sided greenhouse covered with shade nets reducing the incident light by 65%, and automatically watered to keep soil at field capacity.

Plants were grown under these conditions for 20 months. At the beginning of February 1999, seven plants per species were transferred to a growth chamber where they acclimated to the following environmental conditions for 1 month: photoperiod, 12 h; ambient CO2 concentration, 360 µmol mol−1; photosynthetic photon flux density, 500 µmol m−2 s−1; air temperature, 27/21 °C (day/night); water vapour pressure deficit (VPD), 1·1/0·4 kPa. Plants were randomly distributed in the chamber. Soil water content (Θ, m3 water m−3 soil) was measured weekly with a Theta probe ML1-108 (Delta-T devices Ltd, Cambridge, UK). Water was added to maintain Θ at field capacity (≈0·32 m3 m−3).

The experiment started on 9 March 1999 (d = 0). Environmental conditions in the growth chamber were kept as in the acclimation phase. Three plants per species were assigned randomly to the well-watered treatment, and were watered three times a week in order to keep soil at field capacity. Four plants per species were assigned to the drought treatment, and received no water over the whole experiment. The drying experiment ended at d = 77. Stem height and diameter at ground level were measured throughout the experiment for each plant.

Water vapour pressure deficit was varied stepwise 1 day a week to test the influence of leaf-to-air vapour pressure difference (ν) on the leaf gas exchange of the well-watered plants. To obtain a VPD range similar to that observed in the field (0·3–2·6 kPa) (Bonal et al. 2000a), different air temperatures and humidity were imposed (Table 1). We assessed not single VPD effects, but the combination of increasing temperature and increasing VPD as it occurs naturally. Controlled changes in VPD started with the coolest air temperature and maximum humidity (minimum VPD). Air temperature was increased and humidity decreased to increase VPD by ≈0·5 kPa at each step.

Table 1.  Environmental conditions in the growth chamber when leaf gas exchange and water potential were assessed
Day time (h)Irradiance (µmol m−2 s−1)[CO2] (µmol mol−1)Temperature (°C)Relative air humidity (%)VPD (kPa)
  1. VPD, air vapour pressure deficit.

1830–0630  038022·5900·3

Leaf gas-exchange and water potential measurements

Gas-exchange measurements were conducted on the days when controlled changes in VPD were imposed. They were performed using a portable photosynthesis system (IRGA, CIRAS-1, PP-Systems, Hoddesdon, UK) operating in an open mode and fitted with a Parkinson leaf cuvette to measure photosynthetically active radiation (Ip, µmol m−2 s−1) and chamber air temperature (°C). Leaf area-based rates of CO2 assimilation (A, µmol m−2 s−1) and transpiration (E, mmol m−2 s−1), stomatal conductance of water vapour (gs, mol m−2 s−1) and intercellular CO2 concentration (Ci, µmol mol−1) were obtained using the equations of Caemmerer & Farquhar (1981). Gas-exchange measurements were assessed on all plants for each VPD on two randomly chosen, fully expanded leaves per plant. The IRGA was kept in the growth chamber during the day and the plants were left to acclimate to each new atmospheric condition for about 90 min before gas-exchange measurements were performed. The response of A to Ip was tested, and showed that irradiance in the growth chamber (350–500 µmol m−2 s−1) was saturating for A (data not shown).

Soil water content (Θ, m3 m−3) in each pot was estimated weekly using Theta probes on the day of gas-exchange measurements, i.e. 1 day after watering the control plants. Predawn leaf water potential (Ψwp) measurements (Scholander et al. 1965) were made on the days of gas-exchange measurements, before the lights were switched on. Ψwp is an estimation of the average water potential of the soil surrounding the root system (Ritchie & Hinckley 1975). A daily time-course of Ψw was established at the beginning of the experiment to determine the VPD that best corresponded to the minimum leaf water potential (Ψwm) for each species. Thereafter, to avoid excessive leaf removal only Ψwp and Ψwm were estimated on the days of gas-exchange measurement on one or two leaves per plant.


In the well-watered treatment, Θ was 0·32 ± 0·01 m3 m−3 during the experiment. In the drought treatment, Θ fell to 0·01 m3 m−3 for Eperua and Diplotropis, and to 0·05 for Virola (Fig. 1). Leaves of Virola and Diplotropis plants wilted and were shed at Θ ≈ 0·07 and 0·04 m3 m−3, respectively. Leaves of Eperua remained attached throughout. There was a negative effect of soil drought on diameter or height increment for Diplotropis, but not for the two other species (Table 2). For Virola height growth proceeds by successive flushes, which did not occur during the experiment.

Figure 1.

Relationships between predawn or minimum leaf water potential (Ψw) and volumetric soil water content (Θ) in the three species studied during a soil-drying cycle. Large symbols: mean values in well-watered control plants.

Table 2.  Growth of well-watered (control) or droughted (dry) seedlings of three tropical rainforest species growing in controlled environmental conditions before (d = 0) or at the end (d = 77) of a drying cycle
SpeciesTreatmentDayStem diameter (mm)Diameter increment (mm)Plant height (m)Height increment (m)
  1. For a given column, different letters indicate statistical difference (anova followed by Fisher’s least significant difference test; P < 0·05).

DiplotropisControl 0 7·8 ± 1·03·0 ± 0·3c0·8 ± 0·10·2 ± 0·0c
  7710·8 ± 1·1 1·0 ± 0·1 
 Dry 0 8·4 ± 0·70·6 ± 0·2a0·8 ± 0·00·0 ± 0·0a
  77 9·0 ± 0·5 0·8 ± 0·0 
EperuaControl 010·7 ± 1·72·1 ± 0·4b1·1 ± 0·20·1 ± 0·0b
  7712·8 ± 1·9 1·2 ± 0·2 
 Dry 010·5 ± 0·81·4 ± 0·2b0·9 ± 0·20·1 ± 0·0b
  7711·9 ± 0·9 1·0 ± 0·2 
VirolaControl 010·7 ± 0·91·7 ± 0·1b0·4 ± 0·20·0 ± 0·0a
  7712·5 ± 0·8 0·4 ± 0·1 
 Dry 012·1 ± 0·81·0 ± 0·6ab0·4 ± 0·10·0 ± 0·0a
  7713·1 ± 0·3 0·5 ± 0·1 

In Eperua, midday leaf water potential (Ψwm) was markedly less (−1·8 MPa) than predawn (Ψwp, −0·4 MPa) in the well-watered plants (Fig. 1). Initially during the drying cycle (down to Θ = 0·07 m3 m−3), the difference Ψwm − Ψwp increased slightly, mostly due to Ψwm decreasing slightly. Thereafter Ψwp decreased and Ψwm − Ψwp was zero only for Θ of ≈0·03 m3 m−3. Well-watered Diplotropis had lower Ψwm (−1·4 MPa) than Ψwp (−0·5 MPa) (Fig. 1). In the droughted plants, Ψwm − Ψwp decreased with decreasing Θ, approaching zero when Ψwp started to decrease (Θ = 0·09 m3 m−3). During this phase Ψwm increased to −1·0 MPa. Thereafter, both Ψwp and Ψwm decreased sharply. In Virola there was no difference in Ψwp or Ψwm between well-watered and droughted plants as long as leaves were not wilted (Fig. 1). Ψwm (mean −0·4 MPa) was only slightly more negative than Ψwp (mean −0·3 MPa), irrespective of Θ. Drought induced more negative Ψwp and Ψwm in Virola only when leaves wilted (from Θ = 0·07 m3 m−3), with no transition between high and extremely low water potentials (Fig. 1).

The response of leaf gas exchange to soil drought differed among species (Fig. 2). In Eperua, A and gs remained stable down to Θ = 0·10 m3 m−3, then decreased, reaching zero at Θ < 0·04 m3 m−3. In Diplotropis, gs was sensitive to decreasing Θ, with gs starting to decrease from Θ ≈ 0·20 m3 m−3. Complete stomatal closure occurred at Θ = 0·04 m3 m−3. In Virola, both gs and A were more sensitive to soil drought than in the other two species, with complete cessation of leaf-gas exchange at Θ = 0·08 m3 m−3. The greater sensitivity of gs to soil drought in Virola was associated with decreasing Ci during the drying cycle (Fig. 2). In contrast, in Eperua and Diplotropis there was a tendency for Ci to increase in dry soil.

Figure 2.

Relationships between CO2 assimilation rate (A), stomatal conductance of water vapour (gs) or intercellular CO2 concentration (Ci) and volumetric soil water content (Θ) during a soil-drying cycle. Large symbols: mean values in well-watered control plants; vertical bars: ±1 SEM unless it lies within the symbols.

For the well-watered plants, A in Diplotropis always exceeded that in Eperua and Virola (Fig. 3). E and gs were greater in Diplotropis, intermediate in Virola, and least in Eperua. Ci was least in Eperua compared with Diplotropis and Virola. In the well-watered treatment, gs and A decreased strongly and linearly in all species in relation to increasing ν. The absolute sensitivity of gs to increasing ν, calculated as the difference in gs between the lowest (0·8 kPa) and the highest (2·4 kPa) ν, was greater in Diplotropis (0·057 mol m−2 s−1) than in Virola (0·034 mol m−2 s−1) and Eperua (0·027 mol m−2 s−1). Transpiration rates (E) of Eperua and Virola remained stable over the ν range, whereas E decreased in Diplotropis above ν = 1·7 kPa. In Diplotropis and Virola increasing ν resulted in slightly decreasing Ci, whereas Ci remained stable in Eperua. The response of leaf water potential to ν differed among species (Fig. 4). In Eperua, Ψw decreased between 0·7 and 1·5 kPa, down to −1·7 MPa, and slightly decreased thereafter down to −1·9 MPa when ν reached 2·6 kPa. In Diplotropis, Ψw also decreased with increasing ν, but to less negative values than Eperua (−1·4 MPa), and remained stable when ν was higher than 1·5 kPa. In contrast, Virola had similar Ψw all over the ν range (−0·4 MPa), close to predawn values (−0·3 MPa).

Figure 3.

Relationships between CO2 assimilation rate (A), transpiration rate (E), stomatal conductance of water vapour (gs) or intercellular CO2 concentration (Ci) and leaf-to-air vapour pressure difference (ν) in well-watered conditions (Θ ≈ 0·32 m3 m−3). The absolute sensitivity of gs to increasing ν is represented for the three species by hooks. Vertical bars: ±1 SEM.

Figure 4.

Relationships between leaf water potential (Ψw) and leaf-to-air water vapour pressure difference (ν) in well-watered conditions (Θ ≈ 0·32 m3 m−3). Vertical bars: ±1 SEM.


Interspecific differences

For the well-watered plants and under small VPD, the ranking of the three species based on their Ci (Fig. 3) confirmed our expectations (Bonal et al. 2000a). Eperua displayed smaller Ci (i.e. larger A/gs) than Virola and Diplotropis. Bonal et al. (2000a) observed that extremely low A/gs in tropical trees (very negative δ13C) was associated not with small A, but rather with large gs. In this study the difference in Ci between Eperua and Virola or Diplotropis was associated with a smaller maximum gs– and not with a larger A– in Eperua (Fig. 3).

The ranking of species found here in seedlings for gs was in accord with their δ13C ranking in the field (Bonal et al. 2000a; Bonal et al. 2000b). Whether this association of traits holds for all rainforest tree species remains an open question.

Anisohydric species are characterized by leaf water potential (Ψw) suppress, which decreases markedly with increasing atmospheric evaporative demand during the day and with increasing soil drought. In contrast, isohydric species maintain Ψw constant during the day, at a value that does not depend on soil water status until plants are close to death (Stocker 1956; Tardieu & Simonneau 1998). While in Eperua stomatal closure occurred from Θ ≈ 0·1 m3 m−3, Diplotropis and, particularly, Virola were characterized by higher thresholds for the onset of stomatal closure (Fig. 2). Eperua had characteristics of stomatal control (Fig. 2) consistent with the anisohydry (Jones 1998; Stocker 1956; Tardieu & Simonneau 1998). Diplotropis had an isohydric behaviour (Jones 1998; Stocker 1956; Tardieu & Simonneau 1998), at least at the beginning of the drying cycle, as Ψwm did not decrease – and Ψwp − Ψwm did not increase – over a large range of Θ (down to <0·1 m3 m−3) (Fig. 1). Such stomatal control involves an interaction between hydraulic (leaf water status) and root-to-shoot chemical signals (Tardieu & Simonneau 1998). In drying soil, chemical signals induce stomatal closure in isohydric species (Davies & Zhang 1991; Tardieu, Zhang & Gowing 1993) and stomata are more sensitive to these signals when transpiration rate – which depends on leaf-to-air vapour pressure difference (ν) – is high, or when Ψw is low (Tardieu & Simonneau 1998). The sensitivity of gs to Θ in Diplotropis (Fig. 2) was even accompanied by a slight increase in Ψwm, with Θ decreasing to 0·1 m3 m−3 (Fig. 1), which accords with observations made by Bonal et al. (2000a) under field conditions.

Midday leaf water potential (Ψwm) in Virola remained remarkably stable over the drying cycle, at least as long as leaves did not dehydrate, so Ψwm did not depend on soil water content (Fig. 1). A second peculiar trait in Virola was the high stomatal sensitivity to soil drought (Fig. 2). These results confirm those obtained for 15-year-old trees of V. surinamensis growing in a plantation in French Guiana (Bonal et al. 2000a). Such a pattern is also at least partly typical of isohydric stomatal control (Jones 1998; Stocker 1956; Tardieu & Simonneau 1998). However, Virola’s leaf water potential did not conform completely with that of previously described isohydric species. Over the drying cycle, Ψwp remained high and stable, close to Ψwp (Fig. 1), at least as long as plants were not dehydrated. Furthermore, under well-watered (Fig. 4) or soil drought conditions (data not shown), Ψwm did not decrease with increasing ν, even though transpiration rates were high (Fig. 3). Similar results were found by Reekie & Bazzaz (1989) for three tropical gap-tree species (Cecropia obtusifolia, Myriocarpa longipes, Piper auritum) and by Bonal et al. (2000a) for V. surinamensis.Bonal et al. (2000b) found that Ψwm and Ψwp were similar and exceeded −0·5 MPa in several Myristicaceae canopy trees species in French Guiana (Virola sebifera, Iryanthera hostmanii, Iryanthera sagotiana, Osteophloeum platyspermum).

These results led us to hypothesize the existence of a homeostatic regulation of Ψw in Virola. The underlying mechanisms have not been explored and remain unclear, but stomatal control maintaining a ‘set-point’ leaf water potential can be invoked (Comstock & Mencuccini 1998; Jones 1998). For Virola, this set point appears to be very high, and stomatal closure occurs early in the drying cycle, the plants avoiding significant water loss during soil drought and maintaining high Ψw. This behaviour does not rule out possible effects of root-to-shoot chemical signals on stomata (Davies & Zhang 1991; Tardieu et al. 1993), but tends to show that for Virola such signals may not be the primary control on stomatal conductance (Comstock & Mencuccini 1998). Bonal et al. (2000a) showed that small differences between Ψwm and Ψwp in V. surinamensis were associated with a large hydraulic conductivity of the whole plant (Lp). Large Lp, combined with stomatal sensitivity to soil drought (Fig. 2), may allow Virola to avoid xylem embolism under pronounced soil drought (Tyree & Ewers 1996; Tyree & Sperry 1988).

Interspecific differences in the sensitivity of gs to ν were also observed (Fig. 3). As with soil drought, the stomatal sensitivity to atmospheric drought was lower in Eperua than in the two other species. In Eperua, increasing ν above 1·5 kPa did not affect E (Fig. 3), whereas Ψw decreased (Fig. 4). This apparent inconsistency found in the present pot experiment could reflect the effect of a mild soil-water deficit around the root system, as discussed by Tardieu & Simonneau (1998).

Ecological implications

Our results clearly point to the fact that in the tropical rainforest of French Guiana, the seedlings of co-existing species have very contrasting leaf gas-exchange regulation patterns, and specially stomatal regulation, in response to drought. These results allow us to infer some conclusions from an ecological point of view. Among the three species studied, the gas exchange of Eperua seedlings was least sensitive to either soil or atmospheric drought, and Eperua appeared to be tolerant of these environmental conditions. This species will then present a positive CO2 assimilation balance under moderate-to-severe drought conditions, and will remain competitive in a large range of environmental conditions. In contrast, Diplotropis seedlings will dramatically close stomata under drought. Although such a response will allow a favourable tree water status to be maintained, it will induce severe restriction in the CO2 assimilation balance when drought conditions become durable. This might result in the reduction of this species’s competitivity and its ability to establish under unfavourable environmental conditions. This effect will be even more pronounced for Virola. Regarding the dynamics of species distributions, it might be expected that species such as Diplotropis and, especially, Virola would be very sensitive to long-lasting droughts such as those that occurred over the Guiana plateau during the Quaternary (Prance 1982). In contrast, the distribution of species such as Eperua might have been less affected by these changes. Our results, and the work of Bonal et al. (2000a), clearly underline the influence of soil humidity conditions on the dynamics of the floristic composition of the tropical rainforest of French Guiana under climatic changes.


The excellent technical collaboration of Pascal Imbert for the growth chamber utilization is acknowledged. D. Bonal was supported by a grant from INRA and GIS-Silvolab (French Guiana).

Received 20 October 2000; accepted 5 March 2001