Stomatal Regulation of Photosynthesis in Apple Leaves: Evidence for Different Water-use Strategies between Two Cultivars

Abstract

Background and Aims

Leaf responses to environmental conditions have been frequently described in fruit trees, but differences among cultivars have received little attention. This study shows that parameters of Farquhar's photosynthesis and Jarvis' stomatal conductance models differed between two apple cultivars, and examines the consequences of these differences for leaf water use efficiency.

Methods

Leaf stomatal conductance (g_sw), net CO₂ assimilation rate (A_n), respiration (R_d) and transpiration (E) were measured during summer in 8-year-old ‘Braeburn’ and ‘Fuji’ apple trees under well-watered field conditions. Parameters of Farquhar's and Jarvis' models were estimated, evaluated and then compared between cultivars. Leaf carbon isotope discrimination (Δ¹³C) was measured at the end of the growing season.

Key Results

A single positive relationship was established between V_Cmax (maximum carboxylation rate) and N_a (leaf nitrogen concentration per unit area), and between J_max (maximum light-driven electron transport rate) and N_a. A higher leaf R_d was observed in ‘Fuji’. The g_sw responded similarly to increasing irradiance and leaf temperature in both cultivars. g_sw responded to lower vapour pressure deficit in ‘Fuji’ than in ‘Braeburn’. Maximal conductance (g_swmax) was significantly smaller and A_n was more limited by g_sw in ‘Braeburn’ than ‘Fuji’. Lower g_sw, E and higher intrinsic water use efficiency were shown in ‘Braeburn’ and confirmed by smaller leaf Δ¹³C compared with ‘Fuji’ leaves.

Conclusions

The use of functional model parameters allowed comparison of the two cultivars and provided evidence of different water use ‘strategies’: ‘Braeburn’ was more conservative in water use than ‘Fuji’, due to stomatal limitation of A_n, higher intrinsic water use efficiency and lower Δ¹³C. These physiological traits need to be considered in relation to climate adaptation, breeding of new cultivars and horticultural practice.

INTRODUCTION

The responses of fruit tree crops to fluctuating or changing climatic conditions may help in understanding current practical problems in fruit tree management, such as yield variability. The vegetative and reproductive growth of trees depends on assimilate production which is controlled by tree architecture and leaf functions, both modulated by environmental interactions (Flore and Lakso, 1989; Lakso, 1994). At branch scale, Massonnet et al. (2004) showed that two apple cultivars differed in transpiration rate, suggesting that this may result either from variability in branch structure, which affects light interception within the tree crown, or from differences in leaf physiological functions, or from both. Architectural diversity has been characterized among apple cultivars: Lespinasse (1992) and Costes et al. (2003) classified apple cultivars into four groups (types I to IV) based on branching and fruiting patterns. Massonnet (2004) showed that two group IV apple cultivars (‘Fuji’ and a new hybrid ‘X3305’) have a spatial leaf distribution conferring greater light interception by the canopy than two group III cultivars (‘Braeburn’ and ‘Ariane’). The present study addresses the question of leaf functional differences between two of these cultivars, ‘Fuji’ and ‘Braeburn’.

Stomatal conductance (g_sw) and net CO₂ assimilation rate (A_n) in C₃ fruit species depend upon conditions such as solar irradiance (Marini and Sowers, 1990; Francesconi et al., 1997), leaf temperature (Seeley and Kammereck, 1977; Berry and Bjorkman, 1980), vapour pressure deficit (Watson et al., 1978; Fanjul and Jones, 1982), soil and plant water status (Schulze, 1986) and mineral nutrition. The capacity of cultivars to adjust to environmental variations, and their water use strategy in particular, is affected by differences within a species in leaf biochemistry (e.g. carboxylation rate of Rubisco, electron transport rate) and stomatal responses. Among apple cultivars, stomatal regulation has been related to tree vigour; rapid growth was generally linked to high g_sw (Atkinson et al., 2000; Li et al., 2002). Oren et al. (1999) showed that stomatal sensitivity to leaf-to-air vapour pressure deficit (VPD) varies both within and between species. In the short term and at leaf scale, stomatal movements control the trade-off between A_n and transpiration (E), and hence water use efficiency (WUE) which is the ratio between carbon gain and water loss (Farquhar et al., 1989). The intrinsic water use efficiency (IWUE), which was defined by Comstock and Ehleringer (1992) as the ratio of A_n rate to g_sw, is less dependent upon instantaneous environmental conditions (air temperature and relative humidity) than WUE, and is under tight genetic control in many tree species (Brendel et al., 2002, 2007; Casasoli et al., 2006). WUE can be estimated by carbon isotope discrimination (Δ¹³C), an integrative variable, which is linearly and negatively correlated with WUE in many C₃ species (Farquhar and Richards, 1984; Guehl et al., 1995).

Quilot et al. (2002) and Tardieu (2003) suggested that ecophysiological models could be valuable tools for studying complex processes, i.e. interrelated physiological functions, particularly for comparing cultivars. Each genotype can be represented by a set of response parameters for a given range of environmental conditions (Tardieu, 2003). Thus, we hypothesized that parameters from ecophysiological models could be applied to apple cultivars and used to characterize the differences in leaf function between them.

The objective of this study was to identify the physiological traits that are cultivar-dependent in apple trees. To this end stomatal conductance, photosynthesis and dark respiration were compared in ‘Fuji’ and ‘Braeburn’. The comparison was based on (a) parameters of the empirical stomatal conductance model of Jarvis (1976) and maximal stomatal conductance; and (b) parameters of the biochemical photosynthesis model of Farquhar et al. (1980). The relationship between leaf A_n and g_sw, E and Δ¹³C was also examined to assess water use strategies of these apple cultivars.

MATERIALS AND METHODS

Plant material

Two apple cultivars belonging to two different architectural types of the Lespinasse's classification (Lespinasse, 1992) were studied: ‘Braeburn’ (type III), and ‘Fuji’ (type IV), grafted on dwarfing M9 rootstock. Three representative 8-year-old trees of ‘Braeburn’ and of ‘Fuji’ planted in December 1994, 6 m × 1·8 m apart, in a north–south orientation and trained using the Solaxe system (Lauri and Lespinasse, 2000) were used. The experimental plot was carefully irrigated using a microjet system monitored by tensiometers to avoid soil water deficits. All trees were unpruned, and the crop loads adjusted by chemical and manual thinning. The annual fruit yields were between 25 kg and 35 kg per individual tree. All experiments were carried out at INRA experimental station (Mauguio, located near Montpellier, South of France) during the 2002 growing season.

Leaf physiological traits

For all the measurements described, leaves were sampled at random within tree crowns at an external position for sunlit leaves and an internal position for shaded leaves. The sample details of each experiment are specified below.

Stomatal conductance

Stomatal responses to PPF (photosynthetic photon flux, μmol m⁻² s⁻¹), T_l (leaf temperature, °C) and VPD (leaf to air water vapour pressure deficit, kPa) were measured using a portable infrared photosynthesis system (LI-6400; Li-Cor, Inc., Lincoln, NE, USA). Stomatal responses were measured between early July and early August on different samples of four sunlit leaves, simultaneously in the two cultivars. Each set of measurements was obtained by varying only one environmental parameter, the others being set at standard conditions, i.e. PPF = 1500 µmol m⁻² s⁻¹, T_l = 25 °C, VPD ≤ 1·5 kPa. CO₂ concentration of the air at the leaf surface (C_a) was maintained at 35 Pa during these measurements. Each g_sw value was recorded after equilibration for at least 20 min (steady-state condition) as described by Le Roux et al. (1999). Leaves used for the measurements were located in comparable positions for all cultivars, thus minimizing uncontrolled effects resulting from the presence of fruits near the measured leaf, the length of the bearing shoot or the leaf position on that shoot.

Stomatal responses to PPF (1500, 1000, 600, 400, 200, 100, 50, 0 µmol m⁻² s⁻¹) were obtained by varying the intensity of a red/blue LED light source (LI-6400-02B; Li-Cor, Inc.). Stomatal responses to T_l [20 °C, 25 °C, 30 °C (±1 °C)] were measured early in the morning with leaf chamber temperature regulated by integrated Peltier coolers that allowed the temperature to be controlled to within ±6 °C of the air temperature. Stomatal responses to increasing VPD (1·0, 1·5, 2·0, 2·5, 3·0 and 3·5 kPa) were obtained by regulating the relative humidity of the air in the LI 6400 cuvette chamber either by scrubbing a fraction of the inlet air through a desiccating column or by humidifying it through a series of 25-L cans containing water.

The stomatal response to the different environmental factors was expressed as a fraction of the highest g_sw value reached for each leaf (Le Roux et al., 1999).

Maximal stomatal conductance (g_swmax, mol m⁻² s⁻¹) was estimated on an independent sample of 12–17 sunlit or shaded leaves of each cultivar under the standardized environmental conditions previously described, simultaneously with the measurements of stomatal response to the environment.

Leaf photosynthesis and respiration

Net CO₂ assimilation rate (A_n) in relation to internal concentration of CO₂ (C_i) was measured in situ during the same period (early July) with the LI-6400 system, by changing C_a, in the following order: 0, 180, 2·5, 150, 5, 100, 7·5, 60, 10, 40, 15, 20 and 30 Pa. The alternating high and low C_a values avoided leaf saturation by assimilates resulting from persistently high CO₂. All measurements were made under standard environmental conditions (T_l = 25 °C, PPF = 1500 µmol m⁻² s⁻¹, VPD ≤ 1·5 kPa). Eight sunlit and shaded leaves were measured in both cultivars.

Respiration (R_d) was estimated simultaneously over the same period (early July) by measuring CO₂ production rate at the end of the night, before sunrise (0600–0800 h in solar time), for eight sunlit and shaded leaves.

Leaf nitrogen concentration

All leaves measured for stomatal conductance or photosynthesis were immediately collected and placed in a cool box. Leaf area (A_l, m²) was recorded with a flatbed scanner (CanoScan 4400F, Canon) coupled to the Optimas software (V6·5; Media Cybernetics, Silver Spring, MD, USA). Leaves were then rapidly frozen in liquid nitrogen and stored in a freezer at −26 °C before freeze drying (Heto VR1 CT110, Denmark) for dry mass (W, g) determination. Leaf nitrogen content (N) was measured by catharometry after dry ignition (Agronomy Laboratory Cirad Montpellier, France), and leaf nitrogen concentration expressed per unit area (N_a, g m⁻²).

Leaf carbon isotope discrimination

The primary growth of 73–81 long shoots per cultivar was monitored at weekly intervals from April to June. Coloured labels were positioned below the apex of each growing shoot, indicating the leaf positioned below the label, the period of its expansion during the 2002 growing season. In October, three different sets of ten sunlit leaves per cultivar were collected on the basis of leaf expansion date (3 April, 17 April and 3 May) from three trees per cultivar. These samples were placed immediately on ice, frozen in liquid nitrogen and lyophilized. Leaves were crushed and homogenized using a centrifugal crusher. Leaf discrimination for ¹³C (Δ) was calculated as proposed in Farquhar et al. (1989):

1

where δ_a and δ_l are the carbon isotopic compositions of air and leaf, respectively, expressed in δ units (‰) relative to the international Pee Dee Belemnite (PDB) standard. Measurements of δ_l were obtained on a flow isotope ratio mass spectrometer (Delta S; Finnigan MAT, Bremen, Germany) and a constant value of −8 ‰ was used for δ_a as described by Farquhar et al. (1989).

Estimation of model parameters

Leaf stomatal conductance model

Jarvis' empirical model (Jarvis, 1976), modified by Stewart (1988), describes stomatal conductance as a combination of independent relative functions (varying from 0 to 1) characterizing leaf response to PPF, T_l, VPD, C_a and Ψ (bulk soil water potential):

2

In this equation, g_swmax (mol m⁻² s⁻¹) is expressed as a linear function of leaf nitrogen concentration per unit area (N_a), as proposed by Le Roux et al. (1999), and f₁ to f₅ are empirical functions whose best fit values for these parameters, i.e. those presenting the closest correlation, were estimated by linear or non-linear least squares regression using SigmaPlot software (2001, SPSS Inc., Chicago, IL, USA). Relative g_sw (dimensionless) is drawn from eqn (2) as the ratio of g_sw to g_swmax.

To simplify the Jarvis-Stewart model, the functions f₄ and f₅ which describe the responses of relative g_sw to C_a and to Ψ, respectively, were neglected. Indeed, C_a variations in the orchard during the day are negligible in comparison to other sources of variation. Regarding the f₅ function, it was considered that soil water potential showed little variation in the irrigated conditions used.

Leaf photosynthesis model

The parameters of Farquhar's model (Farquhar et al., 1980), modified by Le Roux et al. (1999), were estimated in shaded and sunlit leaves of the two cultivars, assuming that mesophyll CO₂ conductance (g_i) was infinite. This assumption has been questioned (Evans and von Caemmerer, 1996; Ethier and Livingston, 2004; Warren and Dreyer, 2006) because g_i generally limits photosynthesis and leads to the underestimation of V_Cmax (maximum carboxylation rate). Estimating g_i is still a matter of controversy, and it was decided not to take this parameter into account. All values of V_Cmax presented are therefore ‘apparent’ values that underestimate the real values.

Apparent V_Cmax and maximum electron transport rate (J_max) were estimated from A_n–C_i curves, and dark respiration rate (R_d) from direct field measurements (see above). The best fit by non-linear least squares regression (SAS macro; P. Montpied, EEF, INRA Champenoux, pers. comm.) resulted from measurements made with C_a at <30 Pa for V_Cmax and >50 Pa for J_max. Four to seven leaf data sets were retained for the determination of V_Cmax and J_max, and six or seven for R_d. Farquhar's parameters were expressed in relation to the individual leaf N_a value, this variable being closely correlated to intercepted irradiance (DeJong and Doyle, 1985).

Comparison of model parameters between the cultivars

All statistical analyses were performed using Statistica 6 software (StatSoft, Inc., Tulsa, OK, USA). The cultivar effect was computed by a one-way ANOVA after ensuring that the data were normally distributed and that the variances were homogeneous. When the parameters also depended on another variable, the cultivar effect was determined by a covariance analysis (ANCOVA), using the continuous variable as covariate. Mean values for both cultivars were compared by the Newman–Keuls test for parametric analyses (more than ten replicates) or the Mann–Whitney test for non-parametric analyses (less than ten replicates).

Evaluation of model predictions

According to the similarity or dissimilarity of cultivar response to environmental variables, single or cultivar-specific sets of parameters were adopted for each model. Given that independent measurements were used to parameterize Jarvis' and Farquhar's models, the quality of model outputs was tested by cross-validation between the two data sets. Leaf gas exchange measurements made to parameterize Farquhar's model were used to evaluate the quality of Jarvis' model. For this purpose, the environmental conditions prevailing during these measurements (PPF, T_l and VPD) were used as inputs into the Jarvis equation to calculate g_sw values. Reciprocally, the independent measurements made to parameterize Jarvis' model and the resulting g_sw estimations were used to evaluate the quality of Farquhar's model, inferring C_i values from the equation drawn from Harley et al. (1992):

3

The quality of these estimations, by comparison with measured values, was determined by two parameters: root mean square error (RMSE) and bias (b). These were calculated as follows:

4

5

where X_s,i are the estimated values, X_m,i the measured values and n the number of observations.

RESULTS

Stomatal responses to environmental variables

‘Braeburn’ and ‘Fuji’ cultivars showed a similar stomatal response to PPF when fitted by an equilateral hyperbola (Fig. 1A). Non-parametric statistical analysis detected a significant difference (P = 0·034) between the two cultivars only at 100 µmol m⁻² s⁻¹. Relative g_sw was at a maximum when PPF exceeded 1000 µmol m⁻² s⁻¹.

Fig. 1.

Response of stomatal conductance, g_sw, normalized by the maximum stomatal conductance value, g_swmax, to (A) irradiance (PPF), (B) leaf temperature (T_l) and (C) leaf-to-air water vapour pressure deficit (VPD) in ‘Fuji’ and ‘Braeburn’ apple cultivars. In (A), [g_sw/g_swmax = (a_P PPF + b_P)/(c_P PPF + d_P)] fitting parameters: ‘Fuji’: a_P = 66·10⁻⁴, b_P = 0·716, c_P = 66·10⁻⁴, d_P = 1·716 (R² = 0·93); ‘Braeburn’: a_P = 90·10⁻⁴, b_P = 0·856, c_P = 90·10⁻⁴, d_P = 1·856 (R² = 0·94). In (B), (g_sw/g_swmax = a_T T_l² + b_T T_l + c_T) fitting parameters: ‘Fuji’: a_T = –49·10⁻⁴, b_T = 0·28, c_T = –2·94 (R² = 0·57); ‘Braeburn’: a_T = –39·10⁻⁴, b_T = 0·22, c_T = –2·12 (R² = 0·61). In (C), the threshold VPD value ensuring g_sw = g_swmax is noted. Equation beyond this value (g_sw/g_swmax = a_D VPD + b_D) uses fitting parameters: ‘Fuji’: a_D = –0·17, b_D = 1·23 (R² = 0·89); ‘Braeburn’: a_D = –0·21, b_D = 1·39 (R² = 0·75). Results of the Mann–Whitney test: in (A) P = 0·034 at PPF = 100 µmol m⁻² s⁻¹ but P > 0·05 at the other PPF levels; in (B) P > 0·05 at three temperatures; in (C) P = 0·029 between the threshold values. The ANCOVA procedure using VPD as the covariate detected no significant differences (P > 0·05) between the cultivars. Responses to PPF, T_l and VPD were recorded under standard conditions (see Materials and Methods).

The response of relative g_sw to leaf temperature (T_l) was represented by a second order polynomial in ‘Fuji’ and ‘Braeburn’ (Fig. 1B). For both cultivars, the optimal temperature for g_sw was close to 29 °C, without significant differences (P > 0·05) between the two cultivars whatever T_l.

The response of g_sw to VPD displayed two stages in both cultivars (Fig. 1C): at low VPD, relative g_sw was around 1, but it decreased linearly as VPD increased beyond a given threshold. The VPD threshold was significantly higher in ‘Braeburn’ (1·84 kPa) than in ‘Fuji’ (1·38 kPa) according to the Mann–Whitney test (P = 0·028). The rate of g_sw decrease that resulted from VPD increase was slightly faster in ‘Braeburn’ than in ‘Fuji’, although the slopes were not significantly different (P > 0·10; Fig. 1C).

Maximal stomatal conductance

A positive linear relationship was fitted between maximal stomatal conductance (g_swmax) and leaf nitrogen concentration per leaf area (N_a) for each cultivar (Fig. 2). The covariance analysis with N_a as the covariate showed a significant difference (P = 0·005) in mean g_swmax with larger values in ‘Fuji’ than in'Braeburn', although the slopes were similar (0·074 and 0·062, respectively). Moreover, the range of g_swmax variations was different between cultivars: maximal g_swmax was significantly larger in ‘Fuji’ (0·36 mol m⁻² s⁻¹) than in ‘Braeburn’ (0·23 mol m⁻² s⁻¹; Fig. 2).

Fig. 2.

Relationships between maximum stomatal conductance, g_swmax, and leaf nitrogen per unit area, N_a, determined in ‘Fuji’ and ‘Braeburn’ apple cultivars. In the linear equation (g_swmax = a_N N_a + b_N) the fitting parameters are: ‘Fuji’: a_N = 0·074, b_N = –0·020 (R² = 0·74); ‘Braeburn’: a_N = 0·062, b_N = –0·038 (R² = 0·77). The ANCOVA with N_a as the covariate detected significant differences (P = 0·005) between the cultivars. Mean g_swmax values followed by different letters are significantly different: ‘Fuji’, 0·19^a, ‘Braeburn’, 0·12^b.

Leaf photosynthesis and respiration

A unique positive and linear relationship between V_Cmax and N_a, and between J_max and N_a, was observed in both cultivars, as shown by the ANCOVA analyses (Fig. 3A, B). Of the two variables combined in N_a, i.e. nitrogen concentration on a mass basis (N_w) and leaf mass to area ratio (W_a), W_a played the most important role in J_max and V_Cmax variations. Indeed, the correlation was higher between V_Cmax, J_max and W_a (R² = 0·56 and 0·54, respectively) than between V_Cmax, J_max and N_w (R² = 0·23 and 0·29, respectively; data not shown). The mean J_max : V_Cmax ratio was 1·2 at 25 °C.

Fig. 3.

Relationships between (A) maximum carboxylation rate, V_Cmax, (B) maximum electron transport rate, J_max, and (C) dark respiration rate, R_d, and leaf nitrogen per unit area, N_a, in ‘Fuji’ and ‘Braeburn’ apple cultivars. In (A), linear equation (V_Cmax = a_V N_a + b_V) and fitting parameters a_V = 137·3, b_V = –96·94 (R² = 0·77). In (B), linear equation (J_max = a_J N_a + b_J) and fitting parameters: a_J = 136·4, b_J = –70·98 (R² = 0·76). In (C), linear equation (R_d = a_R N_a + b_R) and fitting parameters: ‘Fuji’: a_R = –0·23, b_R = –0·78 (R² = 0·79); ‘Braeburn’: a_R = –0·27, b_R = –0·37 (R² = 0·67). The covariance analysis using N_a as covariate detected no significant differences between the cultivars for V_Cmax and J_max, but significant differences (P = 0·007) for R_d. The R_d means followed by different letters are significantly different: ‘Fuji’, –1·31^a, ‘Braeburn’, –0·96^b. Conditions for the estimation of V_Cmax, J_max and R_d are given in Materials and Methods.

The relationship between leaf respiration (R_d) and leaf nitrogen per mass area (N_a) differed significantly, with ‘Fuji’ greater R_d than ‘Braeburn’ (ANCOVA analysis, P = 0·009; Fig. 3C).

Evaluation of model predictions

The g_sw values predicted by Jarvis' model satisfactorily matched the measured values in both cultivars, as shown by the RMSE (0·04) and bias (–0·008; Fig. 4A). The different points were close to the 1:1 line, particularly for ‘Braeburn’. Some of the highest g_sw values were underestimated by the model for ‘Fuji’.

Fig. 4.

Comparison of (A) stomatal conductance and (B) photosynthetic rate measured in situ on leaves of ‘Fuji’ and ‘Braeburn’ apple cultivars, and simulated by the Jarvis and Farquhar sub-models, respectively. The root mean square error (RMSE), the bias and the 1:1 line are indicated.

Predictions of leaf photosynthesis using Farquhar's model resulted in a fairly good match between estimated and measured values for all cultivars as shown by the RMSE (2·2) and bias (1·13; Fig. 4B). However, the model overestimated (2-fold) some of the highest values for leaf photosynthesis, particularly at high PPF (>600 µmol m⁻² s⁻¹, data not shown).

Relationship between stomatal conductance and photosynthesis

The range of g_sw variations for sunlit or shaded leaves was narrower in ‘Braeburn’ (between 0·05 and 0·23 mol m⁻² s⁻¹) than in ‘Fuji’ (between 0·05 and 0·36 mol m⁻² s⁻¹; Fig. 5), in accordance with the lower g_swmax found in this cultivar. ‘Braeburn’ also had lower maximal leaf A_n values (13·1 µmol CO₂ m⁻² s⁻¹) than ‘Fuji’ (15·2 µmol m⁻² s⁻¹, Fig. 5). Differences between cultivars were confirmed by independent measurements made during the same growing season using a Licor LI-6200 system (data not shown). Specific relationships between A_n and g_sw were adjusted by a non-linear fit in ‘Fuji’ and by a linear fit in ‘Braeburn’ (Fig. 5). Analysis with g_sw as the covariate yielded significant differences in A_n between the cultivars (P = 0·007), with lower CO₂ net assimilation rates in ‘Braeburn’.

Fig. 5.

Relationships between net assimilation, A_n, and stomatal conductance, g_sw, in ‘Fuji’ and ‘Braeburn’ apple cultivars. Equation for ‘Fuji’: [A_n = (a_A b_A g_sw)/(a_A + b_A g_sw) + c_A]; equation for ‘Braeburn’: (A_n = a_A g_sw + b_A). Fitting parameters: ‘Fuji’: a_A = 18·685, b_A = 4·671, c_A = –0·183 (R² = 0·90); ‘Braeburn’: a_A = 43·835, b_A = 2·611 (R² = 0·82). The covariance analysis, using g_sw as the covariate, revealed significant differences (P < 0·05) between the cultivars for A_n. The mean A_n values followed by different letters were significantly different: ‘Fuji’, 10·06^a, ‘Braeburn’, 8·84^b. A_n and g_sw were measured in sunlit and shaded leaves (n = 12 and 17 in ‘Braeburn’ and ‘Fuji’, respectively) under standardized environmental conditions: PPF = 1500 µmol m⁻² s⁻¹, T_l = 25 °C and VPD < 1·5 kPa.

Leaf ecophysiological traits under optimized conditions

The data sets stemming from leaf responses to environmental variables were pooled together, restricting the range of irradiance and air humidity variations to conditions which determine optimal (or sub-optimal) g_sw: 600 ≤ PPF ≤ 1500 µmol m⁻² s⁻¹ and 0·90 kPa ≤ VPD ≤ 2·1 kPa. On this basis, the average g_sw value of sunlit leaves was 25 % smaller in ‘Braeburn’ than in ‘Fuji’ (P = 0·0036; Table 1), and the average A_n value 8 % smaller in the first cultivar (n.s.). Averaged E was 17 % lower, and WUE 5 % higher, when ‘Braeburn’ sunlit leaves were compared with equivalent ‘Fuji’ leaves, but these differences were not significant (P > 0·10). In contrast, the IWUE of ‘Braeburn’ leaves was 15 % greater than in ‘Fuji’ (P < 0·001).

Table 1.

Averaged (±s.d.; n = 24) ecophysiological traits of sunlit leaves in two apple cultivars

	An (μmol CO2 m−2 s−1)	gsw (mol H2O m−2 s−1)	E (mmol H2O m−2 s−1)	WUE (mmol CO2 mol−1 H2O)	IWUE (μmol CO2 mol−1 H2O)
Braeburn	10·895 ± 1·775	0·166 ± 0·051b	2·034 ± 0·989	6·132 ± 2·001	68·941 ± 14·210a
Fuji	11·784 ± 2·874	0·223 ± 0·074a	2·458 ± 1·126	5·325 ± 1·492	55·328 ± 9·281b
Cultivar effect	n.s.	**	n.s.	n.s.	***

Measurements were performed during July and August 2002 with the following conditions: CO₂ = 35 Pa; 600 ≤ PPF ≤ 1500 µmol m⁻² s⁻¹; 20 °C ≤ T_l ≤ 30 °C; 0·90 kPa ≤ VPD ≤ 2·1 kPa.

Values compared by one-way ANOVA procedure for determination of the cultivar effect. **P < 0·01; ***P < 0·001; n.s., not significant. Values followed by different letters are significantly different according to Mann-Whitney test.

Carbon isotope discrimination of sunlit leaves

Smaller carbon isotope discrimination (Δ¹³C) occurred in leaves that expanded in May rather than in April for both cultivars (Fig. 6). ‘Fuji’ leaves displayed a higher Δ¹³C than ‘Braeburn’ leaves irrespective of when they expanded. Non-parametric statistics detected a significant difference between ‘Fuji’ and ‘Braeburn’ during April (P = 0·025; Fig. 6).

Fig. 6.

Carbon isotope discrimination, Δ¹³C, for sunlit apple leaves which expanded at start, middle and end of extension growth period in ‘Fuji’ and ‘Braeburn’ apple cultivars. All leaves were sampled in October for isotopic analysis. Δ¹³C of the two cultivars were statistically compared using the Mann–Whitney test. P-values are indicated where the differences are significant and NS where they are not. Values followed by different letters are significantly different.

DISCUSSION

Quality of model parameter estimations

This study describes the parameterization of Jarvis' stomatal conductance model and of Farquhar's photosynthesis model for leaves of two apple cultivars, and compares the estimated parameters. The stomatal behaviour in response to changing PPF and T_l was similar to that commonly found in other species (Leuning et al., 1995; Le Roux et al., 1999), and previously observed in apple (Warrit et al., 1980; Pretorius and Wand, 2003). The response of g_sw to VPD consisted of a two-step process: g_sw was close to maximum at low or moderate VPD and showed a severe decrease after a VPD threshold. This response was similar to that observed by Watson et al. (1978) in apples, and by Mediavilla and Escudero (2004) in different oak species. Dragoni et al. (2004) obtained the same result in ‘Royal Empire’ apple trees, and concluded that at low VPD, stomatal conductance was limited by the feedback control exerted by photosynthetic products, while the stomatal response to VPD became a limiting factor at higher VPDs. In other studies (e.g. Winkel and Rambal, 1990; Jones, 1998; Le Roux et al., 1999), the initial plateau was not observed, but this may be due to measurements being started at VPD values close to the threshold (around 1·5 kPa). The g_swmax values observed in the present study were smaller than those observed in earlier studies with apple trees (Wünsche et al., 2000; Glenn et al., 2001) but comparable with those described under South African climate (Pretorius and Wand, 2003; Gindaba and Wand, 2007). Indeed, the estimation of g_swmax remains difficult. The environmental conditions a priori supposed to maximize g_sw were drawn from the literature (e.g. optimal temperature commonly 25 °C; Pretorius and Wand, 2003), but it was observed a posteriori that some of these values were not fully adequate in the present conditions: for the two cultivars analysed here, optimal temperature was near 29 °C. However, as measurements were performed in similar conditions for both cultivars, the comparison discussed below remains valid.

Parameters of leaf photosynthetic capacity were linearly related to nitrogen concentration per unit leaf area (N_a), as shown in other deciduous tree species, such as walnut (Le Roux et al., 1999) and peach (Le Roux et al., 2001; Walcroft et al., 2002). V_Cmax at 25 °C was larger than in a series of forest tree species (Dreyer et al., 2001). In addition, the J_max : V_Cmax ratio (1·2) was lower than usually reported even if it is subject to marked variability (Leuning, 1997). Reviewing the literature, Wullschleger (1993) calculated a mean J_max : V_Cmax ratio of 1·64 for 109 species including Malus sp. The high nitrogen status of the experimental trees in the present study, indicated by the large N_a values (maximum up to 4 g m⁻², i.e. N_w of 3·4 %; Fig. 2) of sunlit leaves, could account for the high V_Cmax values observed. Indeed, Grassi et al. (2002) showed that V_Cmax values increase and the J_max : V_Cmax ratio decreases with high mineral nutrient supply in Eucalyptus. The excess N is preferentially allocated to Rubisco rather than to the proteins that regulate the rate of electron transport. The consequence is therefore an increase in V_Cmax but not in J_max, thus decreasing the J_max/V_Cmax ratio. The high V_Cmax values estimated here also suggest a large mesophyll conductance to CO₂ in apple leaves; this requires further examination and explicit estimates of mesophyll conductance. Indeed, different studies demonstrated that mesophyll resistance to CO₂ transfer imposes limitation to photosynthesis in many species (Warren and Adams, 2006).

Quality of model predictions

In the present study, Jarvis' model produced a fair prediction of leaf stomatal conductance over the range of environmental conditions which are commonly encountered in a Mediterranean climate. As the model underestimated some g_sw values, this can result either from the simplification of the Jarvis' model considered here (i.e. C_a and Ψ variations neglected) or from underestimation of g_swmax.

Leaf photosynthesis estimated by Farquhar's model was slightly more problematic. Measured values matched predicted values when PPF was below 600 µmol m⁻² s⁻¹, but the model overestimated A_n values at higher irradiance. This was explained by the J_max module in Farquhar's model which was identified as insufficiently limiting at high PPF. This said, the overestimated photosynthesis rate for strongly sunlit leaves could be considered as having little impact on estimates of photosynthesis at larger scales (e.g. the branch) since most of the leaves on a branch do not receive high irradiance because of within-tree shade and leaf blade orientation (Massonnet, 2004).

Cultivar-dependent leaf functions: differences in water use strategy

The experimental plan used in this study, which included two apple cultivars, allowed the cultivar effect to be explored on leaf photosynthetic and stomatal properties, and also on transpiration rate and water use, in the middle of the growing season.

Based on these instantaneous measurements, the two apple cultivars displayed a similar photosynthetic capacity (V_Cmax and J_max) on the basis of normalized leaf N_a values. In both cultivars, V_Cmax, J_max and R_d increased with N_a, as commonly observed in numerous other species (Reich et al., 1998), including walnut (Le Roux et al., 1999), peach (Le Roux et al., 2001) and mango (Urban et al., 2003). The relationship between R_d and N_a is interpreted to be a consequence of changed leaf structure resulting from different positions within the canopy, and higher maintenance costs in sunlit leaves considering their carbon budget (Mitchell et al., 1999). This is supported by structural modifications between sunlit and shaded leaves observed in these two cultivars (Massonnet, 2004). The thinner palisade tissue in shaded leaves and lower leaf mass to area ratio could account for the diminished respiration rate. The higher leaf R_d in ‘Fuji’ suggests that the potentially larger photosynthetic rate in this cultivar is partly counterbalanced by greater respiratory losses.

The comparison of the two cultivars revealed different stomatal responses. ‘Braeburn’ and ‘Fuji’ differed with respect to maximal stomatal conductance, ‘Fuji’ having significantly higher g_swmax than ‘Braeburn’. The two cultivars showed similar responses to PPF and T_l. The stomatal response to VPD showed some differences between cultivars: a significantly lower VPD threshold value was observed in ‘Fuji’, whereas the rate of relative g_sw decrease beyond this threshold was slightly steeper in ‘Braeburn’.

‘Braeburn’ showed a linear relationship between A_n and g_sw, i.e. no apparent saturation of the photosynthesis process even at the highest observed g_sw. In this case, carbon gain is likely to be limited more by stomatal conductance than by the photosynthetic apparatus itself. Such a linear relation between A_n and g_sw has already been reported in apple (Lakso, 1979) and mango (Urban et al., 2004), but it was shown here that this cannot be generalized. Indeed, ‘Fuji’ showed some non-stomatal limitation of leaf photosynthesis (rectangular hyperbola fit; Fig. 5). With this curvilinear relationship, A_n reaches a plateau at the highest observed g_sw values, suggesting that A_n is limited either by the amount of Rubisco or its activity, or by the rate of electron transport. In this case, the trade-off between CO₂ assimilation and E is modified, lowering the WUE value. Cultivars with a prevailing stomatal limitation generally appear more water-conserving and consequently exhibit a higher WUE than cultivars with limitation by the photosynthetic machinery (Jones, 1985). The present results are in accordance with this interpretation since instantaneous gas exchange measurements on sunlit leaves indicated slightly higher WUE, and significantly higher IWUE, in ‘Braeburn’ than in ‘Fuji’. The higher water use efficiency in ‘Braeburn’ was confirmed when integrated over a much longer time scale (i.e. the whole growing season), by the smaller values of carbon isotope discrimination (Δ¹³C) in its leaves. Farquhar and Richards (1984) pointed out that the isotopic composition reflects the effect of plant water status on photosynthesis, with a linear negative relationship between Δ¹³C values and WUE in various species. In the present study, values of leaf Δ¹³C were very informative in terms of the physiological properties of a cultivar and complemented instantaneous stomatal conductance measurements, because Δ¹³C integrated environmental conditions which varied throughout the growing season. The g_sw values, predicted by the Jarvis model, result from the response to several environmental variables, without consideration of their interaction. This possible imbalance is overcome by using the more integrative variable, Δ¹³C, which was therefore useful for comparing functional capacities between cultivars, in addition to the parameters of the ecophysiological models. As the present comparison considered two apple cultivars, it would be worthwhile to examine leaf ecophysiological traits in more depth, among a larger cultivar range. This was performed in a recent study showing a strong variability of leaf ecophysiological traits (e.g. g_sw, IWUE, A_nmax …) among a recombinant apple F₁ population (Regnard et al., 2007), opening new perspectives for breeding.

CONCLUSIONS

In the present study, the results of both Δ¹³C analysis and instantaneous responses of g_sw to environmental variables indicated different water use strategies in the two apple cultivars analysed. Based on the results, ‘Braeburn’ is more water-conserving than ‘Fuji’. Stomatal responses to environment, and leaf respiration rate, differed between the two cultivars; they also had contrasted crown architectures. As a consequence, combined functional and architectural traits probably play complementary roles in determining the physiological functions of these apple cultivars at the integrated scales of fruiting branches or whole trees. Further examination of a larger range of cultivars is required. The present study also shows that the description of the environmental plasticity of different cultivars is of interest for horticulture in order to determine the potential adaptation of a cultivar to a given climate and consequently the need of finely tuning cultural practices (e.g. irrigation) when the environmental conditions are limiting for growth and productivity.