Long-term water stress leads to acclimation of drought sensitivity of photosynthetic capacity in xeric but not riparian Eucalyptus species

Abstract

Background and Aims Experimental drought is well documented to induce a decline in photosynthetic capacity. However, if given time to acclimate to low water availability, the photosynthetic responses of plants to low soil moisture content may differ from those found in short-term experiments. This study aims to test whether plants acclimate to long-term water stress by modifying the functional relationships between photosynthetic traits and water stress, and whether species of contrasting habitat differ in their degree of acclimation.

Methods Three Eucalyptus taxa from xeric and riparian habitats were compared with regard to their gas exchange responses under short- and long-term drought. Photosynthetic parameters were measured after 2 and 4 months of watering treatments, namely field capacity or partial drought. At 4 months, all plants were watered to field capacity, then watering was stopped. Further measurements were made during the subsequent ‘drying-down’, continuing until stomata were closed.

Key Results Two months of partial drought consistently reduced assimilation rate, stomatal sensitivity parameters (g₁), apparent maximum Rubisco activity ($V_{\text{cmax}}^{\prime }$) and maximum electron transport rate ($J_{\text{max}}^{\prime }$). Eucalyptus occidentalis from the xeric habitat showed the smallest decline in $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$; however, after 4 months, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ had recovered. Species differed in their degree of $V_{\text{cmax}}^{\prime }$ acclimation. Eucalyptus occidentalis showed significant acclimation of the pre-dawn leaf water potential at which the $V_{\text{cmax}}^{\prime }$ and ‘true’ V_cmax (accounting for mesophyll conductance) declined most steeply during drying-down.

Conclusions The findings indicate carbon loss under prolonged drought could be over-estimated without accounting for acclimation. In particular, (1) species from contrasting habitats differed in the magnitude of V′_cmax reduction in short-term drought; (2) long-term drought allowed the possibility of acclimation, such that V′_cmax reduction was mitigated; (3) xeric species showed a greater degree of V′_cmax acclimation; and (4) photosynthetic acclimation involves hydraulic adjustments to reduce water loss while maintaining photosynthesis.

INTRODUCTION

Drought duration and intensity are predicted to increase in future in some regions, particularly in Mediterranean and subtropical climates (IPCC, 2014). The mechanisms underlying plant response to water stress are different for different time scales of stress (Maseda and Fernández, 2006). Short-term water stress reduces plant photosynthesis (A) through limitation of stomatal conductance (g_s) and/or mesophyll conductance to CO₂ (g_m) (Bota et al., 2004; Flexas et al., 2004, 2006, 2012; Grassi and Magnani, 2005; Egea et al., 2011; Galmés et al., 2013; Zhou et al., 2013, 2014), reduction of the maximum carboxylation rate (V_cmax) (Kanechi et al., 1996; Castrillo et al., 2001; Parry et al., 2002; Tezara, 2002; Zhou et al., 2013, 2014) and/or reduction of the maximum electron transport rate (J_max) (Tezara et al., 1999; Thimmanaik et al., 2002; Zhou et al., 2014). Zhou et al. (2014) reported that short-term water stress led to concurrent stomatal, mesophyll and biochemical limitations on photosynthesis, while species from contrasting hydro-climates differed in the drought sensitivity of each process – reflecting inherent differences in drought tolerance.

During longer-term water stress in long-lived plants such as trees, acclimation can take place, rendering the xylem less vulnerable to cavitation. Acclimation to long-term drought can include adjustments at different levels, including leaf physiology, anatomy, morphology, chemical composition, xylem hydraulics, growth and/or carbon partitioning among organs (Maseda and Fernández, 2006; Limousin et al., 2010a, b; Martin-StPaul et al., 2012, 2013). However, much of our knowledge about plant drought responses comes from short-term studies, and thus does not include acclimation responses (Cano et al., 2014). In particular, process-based models of vegetation function typically predict drought effects on trees based on the drought-induced diffusional and biochemical limitations to photosynthesis found in short-term experiments (e.g. Verhoef and Egea, 2014), meaning that they may overestimate the long-term drought impacts if acclimation of these responses occurs.

Although many studies demonstrate acclimation to long-term water stress (e.g. Nogués and Alegre, 2002; Guarnaschelli et al., 2006; Rodriguez-Calcerrada et al., 2011; Ogle et al., 2015), relatively few have examined whether acclimation includes a modification of the functional relationships between photosynthetic traits and water stress, and those few that have, disagree. Limousin et al. (2010b) and Misson et al. (2010) compared the responses of Quercus ilex exposed to different manipulative precipitation regimes for several years in the field, and found high sensitivity of g_s, g_m, V_cmax and J_max to decreased pre-dawn leaf water potential but the functional relationships were not affected by the drought treatment. In contrast, Martin-StPaul et al. (2012) compared the responses of three populations of Q. ilex in sites differing in mean annual rainfall, and found steeper declines of g_s, g_m, V_cmax and J_max as pre-dawn leaf water potential declined in the wettest site compared with the drier sites. Limousin et al. (2013) also found acclimation of the functional responses in a long-term precipitation manipulation experiment in Pinus edulis and Juniperus monosperma.

Another important but largely unexplored question is whether species of contrasting habitat differ in their degree of acclimation in photosynthetic traits during long-term water stress. Cano et al. (2014) reported that long-term water stress led to decreased mesophyll limitation in a xeric species, but not mesic species. A synthesis study by Choat et al. (2012) showed that vulnerability to drought is equally high in plants from both mesic and xeric ecosystems, despite soil water availability being lower in xeric ecosystems – indicating that there must be important differences between the responses of mesic and xeric species to prolonged drought.

In this paper, we test three general hypotheses on the long-term water stress impacts on photosynthesis of contrasting tree species: (1) long-term water stress should allow the possibility of acclimation, such that the diffusional and/or biochemical limitations found in short-term water stress are mitigated; (2) relative to species from riparian habitats, species from xeric habitats should show a greater degree of acclimation; and (3) plants would respond to water stress by controlling stomatal openness in the shorter term, whereas longer-term acclimation to water stress would involve hydraulic adjustments to reduce water loss while maintaining photosynthesis. We tested these hypotheses in a common glasshouse experiment with three congeneric (Eucalyptus) taxa originating from riparian and xeric habitats. Plants were maintained under either field capacity (FC) or partial drought treatment, and gas exchange was monitored after shorter- and longer-term treatment. After the longer-term treatment, we watered all plants to FC and then withheld water and compared their stomatal, mesophyll and biochemical responses during the ‘drying down’ process. We investigated whether the plants acclimate to the partial drought treatment. We then investigated whether the plants could better tolerate water stress during the drying-down process after the longer-term water stress, and how the contrasting taxa differed in their degree of photosynthetic acclimation to water stress. We also assessed the hydraulic adjustments after the longer-term water stress.

MATERIALS AND METHODS

Choice of plant taxa

Three evergreen taxa were selected from the widely distributed Australian genus Eucalyptus. The selected taxa were E. occidentalis from south-western Australia (seed source: Spa Bundaleer, Western Australia; 33·19°S, 138·33°E), E. camaldulensis subsp. subcinerea from central Australia (seed source: Arthur Creek, Northern Territory; 22·40°S, 136·38°E) and E. camaldulensis subsp. camaldulensis from south-eastern Australia (seed source: Barmah SF, New South Wales; 35·50°S, 145·07°E). Although all three plant taxa occur in warm and dry environments, E. camaldulensis subsp. camaldulensis and E. camaldulensis subsp. subcinerea are considered to be mesic species because they are strongly riparian plant taxa which only occur along rivers. Past studies have reported that E. camaldulensis is a phreatophyte with strong sensitivity to soil moisture content (White et al., 2000; Merchant et al., 2007).

Plant material, growth conditions and experimental design

Seeds of the three Eucalyptus taxa were collected directly from their natural habitats by the Australian Tree Seed Centre at Canberra and were germinated in May 2012 at Macquarie University. At 3 months, the seedlings were transplanted into 90 -L pots containing 80 kg of loamy soil (collected from the Robertson area in New South Wales, Australia), evenly mixed with slow-release fertilizer. The plants were grown in the open air with regular watering and full sunshine for 3 months to allow natural establishment. At 6 months (December 2012), the plants were placed in a glasshouse under a 25°C/18 °C diurnal temperature cycle and maintained in a moist condition (100 % FC). Plants were under the natural photoperiod, and received natural light which was attenuated 40 % by the glasshouse roof. Humidity in the glasshouse was not monitored during this experiment, but previous experiments under similar conditions found an average vapour pressure deficit of 1·3 kPa with no significant differences among glasshouses (J. W. Kelly et al., unpubl. res.). Beginning in February 2013, plants of each taxon were subjected to one of two treatments – full watering (100 % FC) or moderate partial drought (70 % FC). The 70 % FC was used to induce moderate drought, following Atwell et al. (2007) who observed a significant 40 % reduction of leaf and stem growth in Eucalyptus tereticornis when grown at this level of soil moisture availability under similar growth conditions. Three plants of each taxon were randomly assigned to each treatment. Soil water content was maintained based on pot weighting. The soil surface was covered with gravel to minimize water loss. Pots were randomly located in the glasshouse, and randomly relocated twice a week.

Measurements of light-saturated net CO₂ assimilation rate (A_sat), g_s, V_cmax and J_max were conducted at 2 months of treatment (April 2013). Measurements of these parameters plus steady-state fluorescence and maximum fluorescence were conducted at 4 months of treatment (June 2013). After 4 months, all pots were watered to 100 % FC. Then watering was ceased, and plants were subjected to drying-down until stomatal conductance was close to zero. Daily measurements of pre-dawn leaf water potential, leaf gas exchange, CO₂ response curves and chlorophyll fluorescence were conducted throughout the drying-down process. The dry-down process took 18–25 d.

Pre-dawn leaf water potential

Pre-dawn leaf water potential (Ψ_pd) was adopted as the consistent measure of soil moisture across plants. Ψ_pd is the best measure of plant water availability because it integrates soil water potential over the root zone (Schulze and Hall, 1982). In particular: (1) Ψ_pd is not influenced by daytime transpiration, while daytime leaf water potential depends strongly on transpiration as well as soil water status; (2) Ψ_pd is independent of differences in rooting depth and soil water access; and (3) unlike volumetric soil moisture content, Ψ_pd is independent of soil texture. Ψ_pd was measured using a pressure chamber (PMS 1000, PMS Instruments, Corvallis, OR, USA). All measurements were completed before sunrise. Two leaves per sapling were sampled. When the observed difference between the two leaves was >0·2 MPa, a third leaf was measured.

Leaf gas exchange, carbon response curve, chlorophyll fuorescence and mesophyll conductance

Leaf gas exchange measurements were performed on current-year, fully expanded sun-exposed leaves, using a portable photosynthesis system (LI-6400, Li-Cor Inc., Lincoln, NE, USA) equipped with an LI-6400-40 Leaf Chamber Fluorometer. Before each measurement, the leaf was acclimated in the chamber for 20–30 min to achieve stable gas exchange, with leaf temperature maintained at 25 °C, reference CO₂ concentration at 400 µmol CO₂ (mol air)⁻¹, and a saturating photosynthetic photon flux density (Q) of 1800 µmol photons m⁻² s⁻¹. Vapour pressure deficit (D) was held as constant as possible during the measurement (D  = 1·5 kPa). After the leaf acclimated to the cuvette environment, A_sat and g_s were measured. Leaf CO₂ uptake (A) versus intercellular CO₂ concentration (C_i) curves were then determined with the cuvette reference CO₂ concentration set as follows: 300, 200, 150, 100, 50, 400, 400, 600, 800, 1000, 1400 and 2000 µmol CO₂ (mol air)⁻¹. The leaf was allowed to equilibrate for at least 3 min at each C_i step. After completion of these measurements, the light was switched off for 3 min and then leaf respiration rate was measured at the ambient CO₂ concentration as the approximation of dark respiration. A and C_i values at each step were corrected for CO₂ diffusion leaks with a diffusion correction term (k) of 0·445 µmol m⁻² s⁻¹, following the manufacturer’s recommendation (Li-Cor).

At each measurement step, steady-state fluorescence (F_s) and maximum fluorescence ($F_{\text{m}}^{\prime }$) were measured during a light-saturating pulse, allowing calculation of the photochemical efficiency of photosystem II (PSII) as Φ_PSII = ($F_{\text{m}}^{\prime }$ − F_s)/$F_{\text{m}}^{\prime }$. The rate of photosynthetic electron transport from fluorescence (J_ETR) was then calculated following Krau and Edwards (1992), as J_ETR = 0·5Φ_PSII·α·Q, where 0·5 is a factor accounting for the light distribution between the two photosystems and α is leaf absorptance (assumed to be 0·85–0·88 in the LI-6400 calculations).

Mesophyll conductance was quantified following the variable electron transport rate method of Harley et al. (1992):

(1)

where the value of Γ^* was taken from Bernacchi et al. (2002), and the rate of non-photorespiratory respiration continuing in the light (R_d) was taken as half the rate of respiration measured in the dark (Niinemets et al., 2005). Thereafter, g_m was quantified for every step of the carbon response curves, and then used to calculate the CO₂ concentration at the chloroplast (C_c) as follows:

$$C_{\text{c}}=C_{\text{i}}–(A/g_{\text{m}}).$$ (2)

Estimation of $V_{\text{cmax}}^{\prime }$, $J_{\text{max}}^{\prime }$, V_cmax and J_max

Estimation of $V_{\text{cmax}}^{\prime }$, $J_{\text{max}}^{\prime }$, V_cmax and J_max (the terms without primes being ‘true’ values, accounting for g_m) followed methods described in detail by Zhou et al. (2014). In brief, the values were quantified from CO₂ response curves using the leaf photosynthesis model of Farquhar et al. (1980), based on the curve fitting routine was that introduced by Domingues et al. (2010) using the least-squares fitting method in the ‘R’ environment (R Development Core Team, 2010).

Conduit anatomy, Huber values, sapwood-specific and leaf-specific hydraulic conductivity

One current-year branch 25 cm long from the distal apex was collected from each plant for measurements of xylem conduit anatomy. A cross-section at 25 cm from the apex was made by hand, stained with methylene blue and mounted in water for immediate microphotography. The microscope (Olympus BX53, Olympus America Inc.) was interfaced with a digital camera at ×4 magnification to record the whole sapwood area, and at ×20 magnification to record the conduit diameter. Both measurements were made using ImageJ image analysis software (http://rsb.info.nih.gov/ij/; ImageJ 1.48; US National Institutes of Health, Bethesda, MB, USA). Huber values (HVs) were calculated as the ratio between the cross-sectional sapwood area and the leaf area supplied. The hydraulically weighted vessel diameter was calculated for each plant to determine the relative contribution of the conduit to hydraulic conductivity. Calculation of hydraulically weighted diameter and sapwood-specific hydraulic conductivity (K_S, kg m⁻¹ s⁻¹MPa⁻¹) was performed as described by Lewis and Boose (1995). Leaf-specific conductivity (K_L) was calculated as the product of K_S and HV.

Leaf mass per area, intrinsic water use efficiency, leaf nitrogen content, height and basal diameter

After 4 months of treatment, three current-year fully expanded sun-exposed leaves of each plant were collected, scanned and measured with the ImageJ image analysis software. The leaves were then oven-dried at 60 °C for at least 48 h and weighed to calculate leaf mass per area (LMA, kg m⁻²). The dried leaf tissue was analysed for stable carbon isotope composition (δ¹³C, ‰) and nitrogen content on a mass basis (N_mass, %) as described by Mitchell et al. (2008). δ¹³C provides a time-integrated measure of intrinsic water-use efficiency over the period in which the leaf carbon is assimilated. Nitrogen content on an area basis (N_area, g m⁻²) was calculated as the product of N_mass and LMA. The height and trunk basal diameter of each plant were also measured.

Stomatal sensitivity to water stress

The g₁ parameter (kPa^−0·5) was introduced by Medlyn et al. (2011) to represent the stomatal behaviour. Medlyn et al. (2011) showed that a stomatal optimality hypothesis results in a simple theoretical model of very similar form to the widely used empirical stomatal models (Ball et al., 1987; Collatz et al., 1991; Leuning, 1995; Arneth et al., 2002):

$$g_s\approx g_0+\mathrm{1·6}\hspace{0.17em}(1+\frac{g_1}{\sqrt{D}})\frac{A}{C_{\text{a}}}$$ (3)

where C_a is the atmospheric CO₂ concentration at the leaf surface (µmol mol⁻¹) and g₀ is the leaf water vapour conductance when photosynthesis is zero (mol H₂O m⁻² s⁻¹). The derivation of the model by Medlyn et al. (2011) provides an interpretation for g₁ as being inversely proportional to the marginal carbon cost of water. An alternative derivation of the same expression and further empirical support were provided by Prentice et al. (2014). The g₁ parameter has proved to be a useful, experimentally determined measure of stomatal sensitivity across climates and plant functional types (PFTs) (Héroult et al., 2013; Zhou et al., 2013, 2014). Its response to water stress is consistent with the theoretical analysis by Mäkelä et al. (1996) suggesting that the marginal water cost of carbon gain should decline exponentially with decreasing soil moisture availability, and the rate of decline with soil moisture should increase with the probability of rain (Zhou et al., 2013, 2014).

We estimated g₁ for each pre-dawn leaf water potential from measurements of A, g_s, C_a and D by re-arranging eqn (3). The parameter g₀ is not part of the optimization and was assigned the value 0·001 mol m⁻² s⁻¹.

Analytical model for the function of stomatal, mesophyll and biochemical responses during the drying-down process

We used equations introduced by Zhou et al. (2013, 2014) to analyse the response functions of g₁, g_m, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ to Ψ_pd during the drying-down process. An exponential decrease of g₁ and g_m with declining Ψ_pd was fitted to each set of observations:

$$g_{\text{1}}=g_{\text{1}}*\text{exp}[b_{\text{1}}({\text{Y}}_{\text{pd}}+0.\text{3})]$$ (4)

$$g_{\text{m}}=g_{\text{m}}*\text{exp}[b_{\text{2}}({\text{Y}}_{\text{pd}}+0.\text{3})]$$ (5)

where g₁*, b₁, g_m* and b₂ are fitted parameters: g₁* is the g₁ value at Ψ_pd = −0·3 MPa, and b₁ represents the sensitivity of g₁ to Ψ_pd. g_m* is the g_m value at Ψ_pd = −0·3 MPa, and b₂ represents the sensitivity of g_m to Ψ_pd. Species adopting different water use strategies are expected to differ in their g₁ sensitivity (b₁) and g_m sensitivity (b₂) to water stress. The responses of $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ were quantified using the logistic function (Tuzet et al., 2003):

(6)

We also quantified all parameters defining the drought response of V_cmax. The function ƒ(Ψ_PD) accounts for the relative effect of water stress on V_cmax, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$. The form of this function allows a relatively flat response of V_cmax, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ under moist conditions, followed by a steep decline, with a flattening again (towards zero) under the driest conditions. K is the value of ƒ(Ψ_pd) under moist conditions. S_f is a sensitivity parameter indicating the steepness of the decline, while Ψ_f is a reference value indicating the water potential at which K decreases to half of its maximum value. Species adopting different water use strategies might be expected to differ in the sensitivity of V_cmax, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ to water stress (S_fV, S_fV′, and S_fJ′) and reference water potential (Ψ_fV, Ψ_fV′ and Ψ_fJ′).

Statistical analyses

The analysis of variance package anova() in R was used to assess treatment effects, and interactions between species and treatment and between treatment and duration. The non-linear least-squares package nls() in R was used to find initial values (least-squares estimates) of the parameters of the exponential functions for responses of g₁ and g_m (g₁*, b₁, g_m* and b₂); the alternative non-linear least-squares package nls2() was used to find initial values of the parameters of the logistic functions for responses of V_cmax, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ (V_cmax*, $V_{\text{cmax}}^{\prime }$*, $J_{\text{max}}^{\prime }$*, S_fV, S_fV′, S_fJ′, Ψ_fV, Ψ_fV′ and Ψ_fJ′). These initial values were then input into the maximum-likelihood estimation package bblme() in R to yield best estimates and standard errors for each parameter. The package glht() was used to conduct multi-comparison analysis on response curves of each parameter across the three contrasting Eucalyptus taxa in the drying-down process after 4 months of treatments. Principal components analysis (PCA) was conducted on the best estimates of parameters to investigate the correlations among the key traits defining the longer-term drought responses of g₁, g_m, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ of three contrasting taxa.

RESULTS

Effect of partial drought on A_sat, g_s, g₁, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ after 2 and 4 months

The partial drought consistently reduced the light-saturated CO₂ assimilation rate (A_sat) after 2 months (P < 0·001) and 4 months (P < 0·01). There was a significant interaction between species and treatment (P < 0·01; Table 1), indicating that partial drought had a greater impact on photosynthesis in the riparian E. camaldulensis than the xeric E. occidentalis (Fig. 1A). There was also a significant interaction between treatment and time (P < 0·05; Table 1), indicating that in both riparian and xeric species, the effect of water stress on photosynthesis recovered over time. Compared with E. camaldulensis subsp. camaldulensis and E. camaldulensis subsp. subcinerea, E. occidentalis from the xeric habitat showed the smallest decline in A_sat after 2 months (Fig. 1A). In E. occidentalis, A_sat recovered completely after 4 months, such that there was no difference in A_sat between the droughted and well-watered plants (Fig. 1A). The partial drought significantly reduced g_s in all taxa after 2 months (P < 0·01) and 4 months (P < 0·001) (Fig. 1B), and reduced g₁ after 4 months (P < 0·01) (Fig. 2A).

Table 1.

P-values of comparison across three Eucalyptus taxa exposed to two watering treatments (100 % versus 70 % FC) on components of leaf gas exchange (leaf net photosynthesis at saturating light A_sat; stomatal conductance g_s; stomatal sensitivity parameter g₁; apparent Rubisco activity $V_{\text{cmax}}^{\prime }$; apparent maximum electron transport rate $J_{\text{max}}^{\prime }$; ratio of $J_{\text{max}}^{\prime }$ to $V_{\text{cmax}}^{\prime }$) measured after 2 and 4 months of treatment, and on hydraulic traits, growth status and leaf nitrogen content (carbon-isotope composition δ¹³C; leaf-specific conductivity K_L; sapwood-specific conductivity K_S; Huber value HV; height; basal diameter; leaf mass per area LMA; leaf nitrogen content on an area basis N_area) measured after 4 months of treatment

	Asat	gs	g1	$V_{\text{cmax}}^{\prime }$	$J_{\text{max}}^{\prime }$	$J_{\text{max}}^{\prime }$/$V_{\text{cmax}}^{\prime }$	δ13C	KL	KS	HV	Height	Basal diameter	LMA	Narea
Species	0·001	<0·001	0·029	0·37	0·05	0·13	0·69	0·001	0·001	0·007	0·49	0·32	0·01	0·32
Treatment	<0·001	<0·001	0·01	<0·001	0·002	0·11	0·02	0·004	0·91	<0·001	0·093	<0·001	0·12	0·98
Time	0·004	0·59	0·074	0·022	0·013	<0·001	NA	NA	NA	NA	NA	NA	NA	NA
Species × Treatment	0·002	0·16	0·86	0·049	0·28	0·23	0·81	0·18	0·67	0·27	0·88	0·94	0·95	0·96
Species × Time	0·084	0·15	0·59	0·70	0·94	0·74	NA	NA	NA	NA	NA	NA	NA	NA
Treatment × Time	0·045	0·84	0·40	0·043	0·48	0·034	NA	NA	NA	NA	NA	NA	NA	NA
Species × Treatment × Time	0·50	0·70	0·86	0·80	0·52	0·78	NA	NA	NA	NA	NA	NA	NA	NA

Data are shown in Figs 1–3 (means ± s.e.; n = 3). NA, not applicable.

Fig. 1.

(A) Leaf net photosynthesis at saturating light (A_sat) and (B) stomatal conductance (g_s) of three Eucalyptus taxa exposed to watering treatments of 100 % field capacity (FC) (Ψ_pd = –0·3 ± 0·1 MPa) and 70 % FC (Ψ_pd = –0·5 ± 0·1 MPa), measured after 2 and 4 months of treatment. Values are means ± s.e. (n = 3).

Fig. 2.

Components of leaf gas exchange in three Eucalyptus taxa exposed to watering treatments of 100 % FC (Ψ_pd = –0·3 ± 0·1 MPa) and 70 % FC (Ψ_pd = –0·5 ± 0·1 MPa), after 2 and 4 months of treatment. (A) Stomatal sensitivity parameter g₁; (B) apparent Rubisco activity $V_{\text{cmax}}^{\prime }$; (C) apparent maximum electron transport rate $J_{\text{max}}^{\prime }$; (D) ratio of $J_{\text{max}}^{\prime }$ to $V_{\text{cmax}}^{\prime }$. Values are means ± s.e. (n = 3).

$V_{\text{cmax}}^{\prime }$ was significantly reduced by the partial drought after 2 months (P < 0·001) but not after 4 months. There were significant interactions between species and treatment (P < 0·05; Table 1) and between treatment and time (P < 0·05; Table 1) (Fig. 2B). Compared with the two riparian taxa, E. occidentalis showed the smallest decline in $V_{\text{cmax}}^{\prime }$ after 2 months, and the $V_{\text{cmax}}^{\prime }$ recovered after 4 months (Fig. 2B). The partial drought significantly reduced $J_{\text{max}}^{\prime }$ (P < 0·01) and increased the ratio between $J_{\text{max}}^{\prime }$ and $V_{\text{cmax}}^{\prime }$ (P < 0·05) after 2 months, while there was no significant reduction after 4 months (Fig. 2C, D).

Effect of partial drought on δ¹³C, K_L, K_S, HV, height, basal diameter, LMA and N_area after 4 months

Compared with plants of the three taxa kept under 100 % FC (Ψ_pd = –0·3 ± 0·1 MPa) for 4 months, plants kept under 70 % FC (Ψ_pd = –0·5 ± 0·1 MPa) showed less negative δ¹³C (P < 0·05) (Fig. 3A; Table 1), larger K_L (P < 0·01) (Fig. 3B; Table 1), larger HV (P < 0·001) (Fig. 3D; Table 1), lower height (marginally significant, P < 0·1) (Fig. 3E; Table 1) and lower basal diameter (P < 0·001) (Fig. 3F; Table 1). There was no effect of treatment on K_S (Fig. 3C; Table 1), LMA (Fig. 3G; Table 1), N_weight and N_area (Fig. 3H; Table 1) or hydraulically weighted vessel diameter (Supplementary Data Fig. S1). These long-term responses to partial drought did not differ significantly among species.

Fig. 3.

Hydraulic traits, growth status and leaf nitrogen content of three Eucalyptus taxa exposed to watering treatments of 100 % FC (Ψ_pd = –0·3 ± 0·1 MPa) and 70 % FC (Ψ_pd = –0·5 ± 0·1 MPa), measured after 4 months of treatment. (A) Carbon-isotope composition δ¹³C; (B) leaf-specific conductivity K_L; (C) sapwood-specific conductivity K_S; (D) Huber value (sapwood area per unit leaf area); (E) height; (F) basal diameter; (G) leaf mass per area; (H) leaf nitrogen content on an area basis. Values are means ± s.e. (n = 3).

Response of g₁, g_m, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ in the drying-down process after 4 months

In the drying-down process after 4 months of treatments, all plants showed considerable decline of A_sat, g_s, g₁, g_m, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ as water availability declined (Figs 4 and 5). There was no effect of growth treatment (100 % FC vs. 70 % FC) on the response of g₁, g_m and $J_{\text{max}}^{\prime }$ to declining Ψ_PD (Fig. 5A, B, D; Table 2). There was a significant effect of growth treatment on the response curve of $V_{\text{cmax}}^{\prime }$ to declining Ψ_PD in xeric E. occidentalis but not in either subspecies of riparian E. camaldulensis (Fig. 5C; Table 2). Compared with plants under 100 % FC, E. occidentalis plants from the 70 % FC treatment had significantly more negative Ψ_fV′ (the water potential at which $V_{\text{cmax}}^{\prime }$ decreases to half of its maximum value) (P < 0·01; Table 2). Similarly, there was a difference between these contrasting taxa when comparing the effect of growth treatment on the response curve of V_cmax, which was consistent with the contrasting responses of $V_{\text{cmax}}^{\prime }$ (Supplementary Data Fig. S2; Table 2). Eucalyptus occidentalis plants from 70 % FC had more negative Ψ_fV (marginally significant, P = 0·07) than plants from 100 % FC (Table 2). Estimated parameter values for each Eucalyptus taxon from each treatment are given in Table 2.

Fig. 4.

(A) Light-saturated CO₂ assimilation rate (A_sat) and (B) stomatal conductance as a function of pre-dawn leaf water potential during a drying cycle following 4 months of treatment at 100 % FC (solid squares) or 70 % FC (open squares). Dark green: E. camaldulensis subsp. camaldulensis; light green: E. camaldulensis subsp. subcinerea; red: E. occidentalis.

Fig. 5.

Components of leaf gas exchange as a function of pre-dawn leaf water potential during a drying cycle following 4 months of treatment at 100 % FC (solid squares for raw data; solid lines for fitted curves) or 70 % FC (open squares for raw data; solid lines for fitted curves). (A) Stomatal sensitivity parameter g₁; (M) mesophyll conductance g_m; (C) apparent Rubisco activity $V_{\text{cmax}}^{\prime }$; (D) apparent maximum electron transport rate $J_{\text{max}}^{\prime }$. Dark green: E. camaldulensis subsp. camaldulensis; light green: E. camaldulensis subsp. subcinerea; red: E. occidentalis.

Table 2.

Fitted values of 13 traits defining the drought responses of stomatal and non-stomatal components of three Eucalyptus taxa during the drying-down process after 4 months of treatment

Species and treatment	g1*	b1	$V_{\text{cmax}}^{\prime }$*	SfV′	ΨfV′	gm*	b2	Vcmax*	SfV	ΨfV	$J_{\text{max}}^{\prime }$*	SfJ′	ΨfJ′
100 % FC
E. occidentalis	3·68	0·97	107	1·82	−2·28	0·32	0·55	134	1·94	−2·69	148	1·83	−2·92
E. camaldulensis subsp. subcinerea	3·54	0·31	113	0·65	−1·28	0·26	0·51	126	0·63	−2·16	142	2·61	−1·48
E. camaldulensis subsp. camaldulensis	3·04	0·84	106	1·9	−1·85	0·28	0·5	127	2·71	−1·93	134	1·57	−2·15
70 % FC
E. occidentalis	3·57	0·48	118	1·42	−3·35	0·34	0·39	142	2·46	−3·46	154	1·21	−3·67
E. camaldulensis subsp. subcinerea	3·89	0·69	107	1·01	−2·66	0·24	0·32	138	2·46	−2·43	138	2·45	−2·49
E. camaldulensis subsp. camaldulensis	2·52	0·78	94·8	10·3	−2·16	0·24	0·43	125	9·97	−2·14	136	2·79	−1·97

The 13 traits are: g₁* and b₁ (eqn 4), g_m* and b₂ (eqn 5), $V_{\text{cmax}}^{\prime }$* ($V_{\text{cmax}}^{\prime }$ estimated at Ψ_pd = 0), S_fV′ (sensitivity of $V_{\text{cmax}}^{\prime }$), Ψ_fV′, V_cmax* (V_cmax estimated at Ψ_pd = 0), S_fV (sensitivity of V_cmax), Ψ_fV, $J_{\text{max}}^{\prime }$* ($J_{\text{max}}^{\prime }$ estimated at Ψ_pd = 0), S_fJ′ (sensitivity of $J_{\text{max}}^{\prime }$), Ψ_fJ′ (eqn 6). Fitted functional relationships are shown in Fig. 5.

Water relation strategies

PCA (Fig. 6) showed strong dominance by the first principal component (PC1), which explained 49·2 % of the total variation. PC1 showed a continuum from the most xeric Eucalyptus taxon (E. occidentalis towards the left) to the most mesic taxon (E. camaldulensis subsp. camaldulensis towards the right), characterized by the positive correlation among b₁, S_fV′, S_fJ′, Ψ_fV′ and Ψ_fJ′, and by their negative correlations with g₁*, g_m*, V_cmax*′ and J_max*′ (Fig. 6). Eucalyptus camaldulensis subsp. camaldulensis, found to the right of PC1, was characterized by (1) lower initial g₁, g_m, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ values under moist conditions; (2) a relatively higher rate of decline in g₁, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ in the drying-down process; and (3) the decrease of $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ commencing at a more negative pre-dawn leaf water potential (Fig. 6; Table 2).

Fig. 6.

Principal components analysis of ten drought-response traits, b₁ (sensitivity of g₁) and g₁* (eqn 4), b₂ (sensitivity of g_m) and g_m* (eqn 5), $V_{\text{cmax}}^{\prime }$* ($V_{\text{cmax}}^{\prime }$ estimated at Ψ_pd = 0), S_fV′, Ψ_fV′, $J_{\text{max}}^{\prime }$* ($J_{\text{max}}^{\prime }$ estimated at Ψ_pd = 0), S_fJ′ and Ψ_fJ′ (eqn 6).The first principal component (PC1) showed a continuum of species from more mesic to more xeric, which explained 49·2 % of the total variation. The second principal component (PC2) showed a continuum of plants from 70 to 100 % FC treatment, which explained 24·1 % of the total variation. Trait values are shown in Table 2.

The second principal component (PC2) explained 24·1 % of the total variation. Along PC2, the plants were arranged into two loose groups along another continuum from the Eucalyptus taxa kept under 70 % FC for 4 months (towards the bottom) to the Eucalyptus taxa kept under 100 % FC (towards the top), characterized by the positive correlation with b₂. The Eucalyptus taxa kept under 70 % FC for 4 months were characterized by a lower rate of decline of g_m in the drying-down process. PC1 and PC2 suggested a direction of acclimation from plants kept under 100 % FC to that of 70 % FC in long-term water stress. Compared with plants kept under 100 % FC, plants kept under 70 % FC for 4 months tended to have lower values of b₂ and Ψ_fV′. Parameter values of the three Eucalyptus taxa from the two treatments are shown in Table 2.

DISCUSSION

We investigated whether plant photosynthetic rates and their response to soil moisture stress could acclimate to longer-term low soil moisture availability. This study provides important experimental evidence that the photosynthetic acclimation can occur in longer-term water stress, and highlights the significant acclimation of $V_{\text{cmax}}^{\prime }$ in xeric but not riparian Eucalyptus taxa.

Differential acclimation of photosynthetic responses in contrasting Eucalyptus taxa to long-term water stress

After 2 months of 70 % FC treatment, we observed a significant reduction in A_sat, g_s, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ in all taxa (Figs 1 and 2). However, the effect of partial drought on photosynthesis differed among species, such that E. occidentalis from xeric habitat showed less stomatal and biochemical limitations than the riparian E. camaldulensis at 2 months.

Meanwhile, the effect of partial drought differed at the different time points, being smaller at 4 months than at 2 months. Four months of 70 % FC treatment allowed for acclimation of $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ such that the biochemical limitations found in short-term water stress in all taxa are mitigated (Fig. 2). However, E. occidentalis showed more recovery of $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ than E. camaldulensis after 4 months of water stress (Fig. 2).

In the subsequent drying-down process, all plants consistently showed a progressive increase in stomatal, mesophyll and biochemical limitation with increasing water stress (Figs 4 and 5; Table 2), which is consistent with previous studies in which a drying-down process was imposed without the longer-term water stress (Zhou et al., 2013, 2014). The response functions of g₁, g_m, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ to declining Ψ_PD were compared between plants from 70 and 100 % FC using a multi-comparison analysis. We investigated two possible classes of causes accounting for mitigation of limitation on $V_{\text{cmax}}^{\prime }$ in longer-term water stress: (1) mitigation of limitation on g_m as reported by Galle et al. (2009) and Cano et al. (2014); and (2) mitigation of limitation on V_cmax, which has never been reported before according to our knowledge. The results showed the three contrasting Eucalyptus taxa differed notably in their degree of modification of the functional relationship of $V_{\text{cmax}}^{\prime }$ (and V_cmax) with declining Ψ_PD (Fig. 5; Table 2). When comparing the plants kept under 70 and 100 % FC, E. occidentalis showed a significantly higher degree of $V_{\text{cmax}}^{\prime }$ (and V_cmax) acclimation than the two riparian taxa in the longer-term water stress, by significantly displacing the start of severe limitations on $V_{\text{cmax}}^{\prime }$ (and V_cmax) to lower Ψ_fV′ (and Ψ_fV) (Fig. 5; Table 2). E. occidentalis plants kept under 70 % FC for 4 months – during which the significant acclimation process occurred – showed the ability to continue active photosynthesis down to much lower soil water potential (−3·9 MPa) than plants of the two riparian taxa kept under 70 % FC for four months (Figs 4 and 5).

For the first time, we report significant acclimation of drought sensitivity of V_cmax in xeric but not riparian Eucalyptus (Figs 2 and 5; Supplementary Data Fig. S2; Table 2). This study adds novel information to previous studies investigating whether plants acclimate to long-term water stress by modifying the functional relationships between photosynthetic traits and water stress (Limousin et al., 2010b; Misson et al., 2010; Martin-StPaul et al., 2012). After long-term water stress during which the acclimation process occurred, the xeric but not riparian Eucalyptus taxa could significantly modify the functional relationship between Rubisco activity and declined Ψ_PD. The inherent differences among the three Eucalyptus taxa of contrasting habitat were not only reflected in their contrasting degree of tolerance in short-term water stress, but also in their contrasting degree of acclimation in long-term water stress. These differences supported our original hypothesis, namely that xeric species would show stronger acclimation to water stress, and probably reflect stronger evolutionary pressures for drought tolerance in dry habitats.

Short-term water stress could lead to the decrease of Rubisco activity, associated with down-regulation of the activation state of the enzyme (Galmés et al., 2013), reduction in Rubisco content and/or soluble protein content (Hanson and Hitz, 1982; Wilson et al., 2000; Xu and Baldocchi, 2003; Grassi et al., 2005; Misson et al., 2006). However, the longer-term water stress might lead to significantly higher protein content and/or protein allocated to Rubisco in leaves of the drought-tolerant taxa than drought-sensitive taxa (Tezara and Lawlor, 1995; Panković et al., 1999), indicating that higher Rubisco content could be one factor in conferring higher drought tolerance and acclimation in xeric species than mesic species. In this study, there was no difference in leaf N_mass or N_area between plants from the two treatments (Fig. 3), but we were unable to further test if the plants kept under 70 % FC had higher nitrogen allocation to Rubisco than that of plants kept under 100 % FC.

We did not find significantly higher g_m in xeric species than mesic species after longer-term water stress as reported by Cano et al. (2014). Species-specific physiology may be of particular importance in comparing their responses to longer-term water stress, leading to varied findings among studies investigating the comparative acclimation in contrasting species (Ogaya and Peñuelas, 2003; Cano et al., 2014). Note that the drought stress treatments applied here differed considerably from those used by Cano et al. (2014). Our low-water treatment was 70 % FC for all taxa, whereas Cano et al. (2014) defined their low-water treatment by stomatal conductance <0·05 mol m⁻² s⁻¹, leading to a lower soil moisture potential in the xeric than the mesic species. Differences in the depth and severity of drought applied, as well as age of plant material, may account for different findings across drought experiments. The current experiment, at 4 months, is shorter than many field-based drought manipulations; however, acclimation is likely to occur more rapidly in seedlings and saplings than in the mature trees typically used in field studies (e.g. Limousin et al., 2010b, 2013). For example, a drought pre-conditioning period of 32 d was sufficient to increase drought survival of Eucalyptus globulus seedlings (Guarnaschelli et al., 2006). It remains to be shown whether the same patterns found in the present study occur in mature Eucalyptus trees in the field, and also for contrasting species of other genera and/or in other biomes.

Hydraulic adjustments after the longer-term water stress

Without sufficient control of water loss during severe drought, xylem embolism and damage could occur and lead to hydraulic failure. However, maintaining water potentials through stomatal closure during severe drought reduces photosynthesis, potentially leading to carbon starvation and mortality (Sala et al., 2012; Adams et al., 2013). Hydraulic down-regulation has been reported to link to drought acclimation of leaf gas exchange in long-term precipitation manipulation experiments (e.g. Limousin et al., 2013; Martin-StPaul et al., 2013). Our results showed an adjustment of hydraulic properties in plants exposed to 4 months of 70 % FC treatment (Figs 2, 3 and 6; Table 2), which would reduce the vulnerability of xylem to cavitation while maintaining photosynthesis during longer-term water stress. The plants kept under 70 % FC for 4 months had stopped growth in height and basal diameter, and significantly increased their sapwood area invested per unit leaf area, yielding higher leaf-specific conductivity. Meanwhile, the plants kept under 70 % FC for 4 months had lower δ¹³C, b₂, Ψ_fV′ and Ψ_fJ′ when compared with plants kept under 100 % FC (Fig. 6; Table 2). This combination of trait changes allowed them to increase the intrinsic water use efficiency, decrease g_m, $V_{\text{cmax}}^{\prime }$ and $J_{\text{max}}^{\prime }$ slowly to maintain more CO₂ supply to the chloroplasts, and also maintain higher photosynthetic capacity and RuBP regeneration capacity down to lower water potential. Prentice et al. (2014) predicted that HV and $V_{\text{cmax}}^{\prime }$ are necessarily linked, which supported our results on the acclimation of carboxylation capacity and the larger HVs in plants kept under 70 % FC for 4 months.

We conclude that the photosynthetic drought responses of plants in the long term can be different from those observed in short-term experiments. For plant physiologists and modellers, this study highlights the importance of considering the acclimation process and its variation with the habitat of species when predicting the long-term drought effect. The findings on intra- and inter-species variation of photosynthetic acclimation in long-term drought could also help restoration ecologists on the selection of species for restoration schemes aiming to increase long-term drought resistance and resilience of forest ecosystems.

SUPPLEMENTARY DATA

Supplementary data are available online at www.aob.oxfordjournals.org. and consist of the following. Figure S1: hydraulically weighted vessel diameter of three taxa at 100 and 70 % FC treatments after 4 months. Figure S2: Rubisco activity V_cmax as a function of pre-dawn leaf water potential during a drying cycle following 4 months of treatment at 100 or 70 % FC.