Ülo Niinemets. Fax: 00372-7-366021; e-mail: email@example.com
Leaf age-dependent changes in structure, nitrogen content, internal mesophyll diffusion conductance (gm), the capacity for photosynthetic electron transport (Jmax) and the maximum carboxylase activity of Rubisco (Vcmax) were investigated in mature non-senescent leaves of Laurus nobilis L., Olea europea L. and Quercus ilex L. to test the hypothesis that the relative significance of biochemical and diffusion limitations of photosynthesis changes with leaf age. The leaf life-span was up to 3 years in L. nobilis and O. europea and 6 years in Q. ilex. Increases in leaf age resulted in enhanced leaf dry mass per unit area (MA), larger leaf dry to fresh mass ratio, and lower nitrogen contents per dry mass (NM) in all species, and lower nitrogen contents per area (NA) in L. nobilis and Q. ilex. Older leaves had lower gm, Jmax and Vcmax. Due to the age-dependent increase in MA, mass-based gm, Jmax and Vcmax declined more strongly (7- to 10-fold) with age than area-based (5- to 7-fold) characteristics. Diffusion conductance was positively associated with foliage photosynthetic potentials. However, this correlation was curvilinear, leading to lower ratio of chloroplastic to internal CO2 concentration (Cc/Ci) and larger drawdown of CO2 from leaf internal air space to chloroplasts (ΔC) in older leaves with lower gm. Overall the age-dependent decreases in photosynthetic potentials were associated with decreases in NM and in the fraction of N in photosynthetic proteins, whereas decreases in gm were associated with increases in MA and the fraction of cell walls. These age-dependent modifications altered the functional scaling of foliage photosynthetic potentials with MA, NM, and NA. The species primarily differed in the rate of age-dependent modifications in foliage structural and functional characteristics, but also in the degree of age-dependent changes in various variables. Stomatal openness was weakly associated with leaf age, but due to species differences in stomatal openness, the distribution of total diffusion limitation between stomata and mesophyll varied among species. These data collectively demonstrate that in Mediterranean evergreens, structural limitations of photosynthesis strongly interact with biochemical limitations. Age-dependent changes in gm and photosynthetic capacities do not occur in a co-ordinated manner in these species such that mesophyll diffusion constraints curb photosynthesis more in older than in younger leaves.
Mediterranean evergreen broad-leaved species can support large leaf area indices of up to 8 m2 m−2 due to extended leaf longevity (Sala et al. 1994; Rambal 2001) that allows the plants to amortize the cost of foliage construction over several growing seasons. The fraction of leaf area attributable to leaves that are more than 1-year-old is often 0.4–0.6 of the total (Sala i Serra 1992), demonstrating that older leaves form a major fraction of whole canopy leaf area in Mediterranean species.
However, foliage photosynthetic potentials per dry mass and photosynthetic nitrogen use efficiency strongly decline with increasing leaf age in Mediterranean species as well (e.g. Niinemets et al. 2004). This age-dependent decline in photosynthetic potentials is associated with enhanced cell wall lignification and total amount of cell walls (Damesin, Rambal & Joffre 1998). Thicker and more strongly lignified cell walls improve the resistance of foliage to drought (e.g. Niinemets 2001), but these modifications may lead to decreases in diffusion conductance from the outer surface of cell walls to the carboxylation sites in chloroplasts (gm) with increasing leaf age (Evans & Loreto 2000). Decreases in gm may potentially importantly affect leaf photosynthesis rates at any given foliar nitrogen content, and provide an explanation for lower leaf photosynthetic capacities and nitrogen use efficiencies in older leaves.
Because foliage photosynthetic capacity and gm are often strongly correlated (Loreto et al. 1992; Evans & von Caemmerer 1996; Evans & Loreto 2000; Loreto, Centritto & Chartzoulakis 2003; Warren et al. 2003; Flexas et al. 2004; Grassi & Magnani 2005), it has been suggested that they change in a co-ordinated manner such that the overall limitation of photosynthesis by gm is similar in leaves with varying photosynthetic capacity and architecture (Evans & Loreto 2000). Such a strong co-ordination between photosynthetic capacity and gm advocates against the hypothesis of stronger mesophyll diffusion limitations in older leaves. However, a review analysis including a large number of species with differing foliage internal structure demonstrated that the ratio of chloroplastic to internal air space CO2 concentrations increases with increasing gm (Niinemets & Sack 2005), indicating that lower gm values are often compatible with stronger internal diffusion limitations. Several studies have demonstrated a non-linear scaling of photosynthetic capacity with gm (DeLucia, Whitehead & Clearwater 2003; Loreto et al. 2003; Flexas et al. 2004) further suggesting that the co-ordination between diffusion limitations and photosynthetic potentials is incomplete. Especially in Mediterranean species that appear to retranslocate foliage nitrogen to a limited extent, but in which the investments in cell walls strongly increase during leaf ageing, capacities of photosynthesis and gm may largely change independently.
The current study was designed to test the hypotheses that the mesophyll diffusion conductance decreases with increasing leaf age in Mediterranean evergreen species, and that the diffusion conductance limits foliage carbon gain more strongly in older leaves. We further assessed the importance of the age-dependent decrease in foliage photosynthetic potentials due to decreases in foliage nitrogen content and changes in nitrogen investments in photosynthetic machinery, and evaluated the effect of age-dependent changes on the generality of leaf photosynthesis versus nitrogen relationship, which is a common tool to indirectly derive foliage photosynthesis potentials (Wright et al. 2004).
MATERIALS AND METHODS
The study was conducted from October to November 2004 in Biotopo di Marocche, Drena, Trentino, Italy (45°58′ N, 10°56′ E), altitude approximately 90 m. The site has Mediterranean climate with average winter temperature of 5.4 °C, average summer temperature of 22.7 °C, annual average temperature of 14 °C and annual precipitation of 920 mm. The study was started after the onset of autumn rains to allow the plant water relations to recover from summer drought, in particular, to achieve high stomatal conductance during gas-exchange measurements. The experiments were finished before the advent of nights with freezing temperatures. The current-year leaves were 7–8 months old by the time of the study.
For all species, we selected two free-standing mature trees. The trees were 11 m tall in Quercus ilex, 4–6 m tall in Olea europea and 3–4 m tall in Laurus nobilis. Peripheral branches with entire foliated length were taken from the lower exposed crown at 0700–0800 h. The branches were immediately re-cut under water, the cut ends were kept in water and the branches were transported to the laboratory. All harvested branches were exposed to more than 75% of full sunlight in their natural position in the canopy.
Foliage photosynthetic measurements
In the laboratory, the branches were preconditioned at room temperature and at dim light of 50–100 µmol m−2 s−1 during 8 h light period for 2–3 d to stabilize leaf gas-exchange characteristics. The branches were re-cut under water every day during the stabilization period. Preliminary experiments demonstrated that while the stomatal conductance was relatively high also immediately after branch sampling, the stomatal conductance rapidly declined during the photosynthesis measurements. In contrast, stomatal conductances were high and stable in preconditioned branches.
Leaf net assimilation versus internal CO2 (A–Ci) response curves together with the effective quantum yields of photosystem II were measured with a Li-Cor LI-6400 portable photosynthesis system equipped with a LI-6400-40 Leaf Chamber Fluorometer (Li-Cor, Inc., Lincoln, NE, USA). Leaf temperature was set at 25 °C, quantum flux density at 1000 µmol m−2 s−1 (20% blue LED, 80% red LED) and vapour pressure deficit at 1.1 kPa. To achieve maximum stomatal openness at the beginning of experiments, leaves were kept at 50 µmol mol−1 of ambient CO2 before the start of an A–Ci curve (Centritto, Loreto & Chartzoulakis 2003). At this low CO2 concentration, the maximum stomatal conductances were obtained 20 min to 1.5 h after leaf enclosure in the chamber. Species-specific estimates of maximum stomatal conductance to water vapour observed during the measurements were 0.17 mol m−2 s−1 for Q. ilex, 0.05 mol m−2 s−1 for L. nobilis, and 0.13 mol m−2 s−1 for O. europea. After full stomatal opening, the net assimilation rates were measured at eight ambient CO2 concentrations between 50 and 2500 µmol mol−1. The experimental leaf was kept at every CO2 concentration until steady-state rates of net assimilation were achieved. Then the steady-state value was recorded, and the effective quantum yield of photosystem II determined by applying a 1-s pulse of white light of 10 000 µmol m−2 s−1 to close all photosystem II centres (Fm′ determination).
After the completion of an A–Ci curve, the light was switched off, and leaf respiration rate was measured immediately after stabilization of chamber temperature and ambient CO2 concentrations (typically 1–2 min after switching off the light).
In each branch, gas-exchange characteristics of every leaf age class were measured. Altogether we analysed three branches (total number of leaves measured = 14) in Q. ilex and in O. europea (12 leaves) and two branches in L. nobilis (six leaves).
Calculation of the rate of photosynthetic electron transport from fluorescence
Immediately after leaf gas-exchange and fluorescence measurements, leaf reflectance and transmittance were estimated with a Taylor-type integrated sphere (Li-Cor 1800; Li-Cor, Inc.) using a Field Spec Pro spectroradiometer (Analytical Spectral Devices, Inc., Boulder, CO, USA). From these measurements, leaf absorptance for every wave-length in the photosynthetically active spectrum, θ(λ), was determined. The amount of absorbed quantum flux density (Qabs) was further calculated from the values of θ(λ), the spectra of blue and red LEDs of LI-6400-40 Leaf Chamber Fluorometer light source (Li-Cor Inc. 2002), and the total quantum flux density of blue and red LEDs. For the specific spectral composition of the Li-Cor light source, the average (± SE) values of leaf absorptance were 0.9337 ± 0.0017 for Q. ilex, 0.9358 ± 0.0017 for L. nobilis, and 0.9314 ± 0.0024 for O. europea.
where ΦPSII is the effective quantum yield that is calculated from the maximum fluorescence yield in light-adapted state, Fm′, and steady-state fluorescence yield in light, F, as (Fm′ − F)/Fm′.
Determination of mesophyll diffusion conductance (gm) using combined gas-exchange/chlorophyll fluorescence approach
Because only 2 cm2 of leaf area is enclosed in the LI-6400-40 Leaf Chamber Fluorometer cuvette, A and Ci values were corrected for diffusion according to the recommendations of Li-Cor Inc. (2002) before further calculations. A diffusion correction term (k) of 0.445 µmol s−1 was used.
The internal mesophyll diffusion conductance from the substomatal cavities to chloroplasts was estimated according to variable electron transport rate method of Harley et al. (1992) as
where Rd is the rate of non-photorespiratory respiration, and Γ* is the hypothetical CO2 compensation point in the absence of Rd. The values of gm were calculated for measurements of net assimilation rate at internal CO2 concentrations of 100–350 µmol mol−1, and the average value of gm was determined for each leaf. It has been demonstrated that over this Ci range, the values of gm are stable, and the values of gm are relatively insensitive to minor errors in Γ*, Rd and A (Harley et al. 1992). Because the measurements were started when steady-state stomatal conductances were achieved, the values of stomatal conductance were generally stable during these measurements at relatively low ambient CO2 concentrations. Whenever time-dependent decreases in stomatal conductance were observed during these measurements, the data were discarded. The average coefficient of variation (100 × (standard deviation)/(average of all gm values for a given leaf)), of single-leaf gm estimates was 8.4% (range 0.12–28%) in Q. ilex, 5.9% (1–10%) in L. nobilis and 5.0% (1–13%) in O. europea.
In the calculations of gm, we used the standard Rubisco kinetic characteristics determined in vitro (Jordan & Ogren 1984; Niinemets & Tenhunen 1997) to derive a value of Γ* at 25 °C (45.3 µmol mol−1). We preferred the in vitro Rubisco characteristics, as the previous in vivo determinations of Γ* are likely underestimates due to assuming an infinite gm in the calculations (Ethier & Livingston 2004). In fact, the true Γ* is equal to Γ*(apparent) + Rd/gm (von Caemmerer et al. 1994; Piel et al. 2002), implying that an independent value of gm is needed to determine the values of true Γ* from in vivo estimations. Nevertheless, the values of gm determined over Ci range of 100–350 µmol mol−1 were only moderately affected by alternative estimates of Γ*, e.g. in vivoΓ* = 42.9 µmol mol−1 at 25 °C according to Bernacchi et al. (2001), and our main conclusions on the significant age effects on gm and gm-dependent CO2 drawdown in chloroplasts were qualitatively not affected by using different estimates of Γ*.
The rate of non-photorespiratory respiration continuing in the light (Rd) was taken as half of the rate of respiration measured in the dark as reported in previous studies (Villar, Held & Merino 1995; Piel et al. 2002) An alternative estimate of Rd was derived from the A–Ci curves as in Grassi & Magnani (2005). Both estimates of Rd were strongly correlated (r2 = 0.84, P < 0.001 for all data), and the regression equation was Rd(derived from dark measurements) = 1.01(Rd from A–Ci curves), demonstrating that a realistic estimate was employed in gm calculations.
Estimation of mesophyll diffusion conductance from A–Ci curves
We use the A–Ci curve analysis method suggested by Ethier & Livingston (2004) to derive an alternative estimate of gm. This method is based on the circumstance that for a set of fixed biochemical characteristics of Rubisco, the shape of A versus Ci response curves deviates from the rectangular hyperbola model when gm is not infinite (Ethier & Livingston 2004). This leads to underestimation of maximum carboxylase activity of Rubisco (Vcmax) when gm is neglected.
The curve-fitting procedure suggested by Ethier & Livingston (2004) combines the equations describing the dependence of A on Ci and chloroplastic CO2 concentration (Cc). For Rubisco-limited conditions, the net assimilation rate is given as
where Kc is the Michaelis–Menten constant for CO2 and Ko for oxygen and O is oxygen concentration. Internal and chloroplastic CO2 concentrations are related as
We used the same Rubisco data as for Γ* to fix Kc and Ko (Jordan & Ogren 1984; Niinemets & Tenhunen 1997), and Rd was taken as half of the rate of respiration measured in the dark as in Eqn 2. Equation 5 was further fitted to Rubisco-limited region of A versus Ci response curves (Ci < 220 µmol mol−1) using an iterative method that minimized the sum of squares of [A(predicted) − A(measured)].
The values of gm derived from combined gas-exchange/fluorescence approach (Eqn 2) and from novel fitting of A–Ci curves (Eqn 5) were similar with the average difference of 8.8% (0.2–24.9%) between the two different methods. Furthermore, the alternative estimates of gm were strongly correlated, and the slope of this relationship approached one, and intercept zero (Fig. 1). Because of the strong correlation, all statistical relationships were qualitatively the same with both estimates of gm (Eqns 2 and 5). Therefore, in the Results section, we demonstrate only the statistical relationships with gm from the combined gas-exchange/fluorescence approach (Eqn 2).
Calculation of maximum Rubisco carboxylase activity (Vcmax) and the capacity for photosynthetic electron transport (Jmax) from A–Cc curves
Curve fitting by Eqn 5 also provided values of Vcmax, and the chlorophyll fluorescence estimates of electron transport (Eqn 1) at highest CO2 concentrations provided values of the capacity for photosynthetic electron transport (Jmax). Further estimates of Vcmax, Rd, and Jmax were derived from fitting A–Cc curves. The A–Cc curves were obtained from A–Ci curves according to Eqn 4 and using the values of gm determined from Eqn 2. The A–Cc curve fitting was conducted as in Niinemets et al. (1999).
The values of Vcmax estimated by two methods were strongly related (for all data, r2 = 0.82, P < 0.001, Vcmax[from Eqn 5) = 1.026(Vcmax from A–Cc curves)], and the average difference between the two estimates was 12.5% (0–32%). The capacities of photosynthetic electron transport measured by fluorescence (Eqn 1) and derived from A–Cc curves were also strongly linked (for all data pooled, r2 = 0.92, P < 0.001, Jmax[from Eqn 1 = 1.027(Jmax from A–Cc curves)]. Because of these strong correlations, all statistical relationships with leaf age and foliage characteristics were qualitatively the same for different Vcmax and Jmax estimates (data not shown), and we report only the correlations with Vcmax and Jmax derived from A–Cc curves.
For measurements with stable stomatal conductance (s. above), we calculated the average values of A, Rd, stomatal conductance to water vapour and Ci measured at ambient CO2 concentrations of 200–400 µmol mol−1 (average ambient CO2 concentration 330–340 µmol mol−1). These values were used to derive the average Cc according to Eqn 4, average Cc/Ci ratio and average CO2 drawdown (ΔC = Ci − Cc). Because photosynthesis responds linearly to CO2 over the concentration range chosen, and Eqn 4 is linear with respect to both Ci and A, average Cc determined from average Ci and A is equivalent with average Cc determined from individual combinations of A and Ci.
Chemical and structural analyses
Foliage carbon and nitrogen percentages were estimated with a Perkin Elmer series II CHNS/O Analyzer 2400 (Perkin Elmer Life and Analytical Sciences, Inc., Boston, MA, USA). Leaves were scanned with a resolution of 300 dpi, and leaf area was estimated with UTHSCSA Imagetool 2.00alpha (C. Donald Wilcox, S. Brent Dove, W. Doss McDavid and David B. Greer, Department of Dental Diagnostic Science, The University of Texas Health Science Center, San Antonio, TX, USA; http://ddsdx.uthscsa.edu).
The leaves used for gas-exchange measurements were immediately weighted after measurements to determine the fresh mass. They were then oven-dried at 70 °C for at least 48 h, and the dry mass was determined. The remaining leaves on a given branch were separated between age classes, and weighed before and after oven-drying as the sample leaves.
Apparent nitrogen investments in photosynthetic machinery
From the values of Jmax and Vcmax, we calculated the apparent nitrogen fractions in the rate-limiting proteins of photosynthetic electron transport, PB (Niinemets & Tenhunen 1997)
and in Rubisco (PR):
where Jmc is the capacity for photosynthetic electron transport per unit cytochrome f (mol e– (mol cyt f)−1 s−1 (Jmc = 156 mol e– (mol cyt f)−1 s−1 at 25 °C) (Niinemets & Tenhunen 1997), and Vcr is the maximum rate of ribulose-1,5-bisphosphate carboxylation per unit Rubisco protein (Vcr = 20.5 µmol CO2 (g Rubisco)−1 s−1 at 25 °C) (Niinemets & Tenhunen 1997). The scaling coefficients 8.06 and 6.25 are fixed based on the stoichiometry of rate-limiting proteins and nitrogen content of proteins. Equations 6 and 7 assume that the values of Jmc and Vcr are essentially conserved among C3 plants. The actual nitrogen fractions of N in photosynthetic apparatus may vary due to a certain species-dependent variability in Rubisco specific activity and the rate of electron transport per unit cytochrome f (Niinemets & Tenhunen 1997; Spreitzer 1999). Nevertheless, within a species, the apparent nitrogen fractions given by Eqns 6 and 7 are proportional to the fraction of N in rate-limiting proteins.
Distribution of leaves with varying age and age-dependent modifications in leaf structure and nitrogen content
In Quercus ilex, six leaf age classes (0.7- to 5.7-year-old) were present, whereas L. nobilis and O. europea supported three leaf age classes (0.7- to 2.7-year-old). The fractional contribution of current year leaves to total branch foliage dry mass varied from 0.5 to 0.8 (Fig. 2a), demonstrating that older leaves comprised an important fraction of total branch foliage biomass.
Leaf dry mass per unit area increased (MA, Fig. 2b) and nitrogen content per dry mass (NM, Fig. 2c) decreased with increasing leaf age in all species. Nitrogen content per area (NA) decreased with increasing leaf age in Q. ilex and L. nobilis, but it was independent of age in O. europea (Fig. 2d). Thus, the relative decrease in NM exceeded the increase in MA in Q. ilex and L. nobilis (NA = MANM), whereas the relative change in MA and NM was similar in O. europea.
There was a certain branch-to-branch variability in foliage chemical and structural variables. Therefore, the explained variances in the regressions with leaf age were generally larger for the structural and chemical variables standardized with respect to the values of current-year leaves (Fig. 2b inset). As an exception, the relationship between standardized NA and leaf age was not significant in Q. ilex (r2 = 0.12, P > 0.2).
Mesophyll diffusion conductance and photosynthetic potentials in relation to age
Mesophyll diffusion conductance (gm) significantly modified the photosynthesis versus CO2 response curves in all cases (Fig. 3 for sample graphs). gm scaled negatively with leaf age in all species (Figs 3, 4a & b). The capacity for photosynthetic electron transport determined from A–Cc curves (Jmax) and the maximum carboxylase activity of Rubisco (Vcmax) per unit area was negatively associated with age in Q. ilex and L. nobilis but not in O. europea (Fig. 4c & e). As a result of the strong increase of MA with leaf age, the mass-based characteristics (area-based values divided by MA) decreased more strongly with leaf age than area-based characteristics (cf. Fig. 4a, c, e & 4b, d, f). The photosynthetic capacities per unit dry mass were negatively related to age in all species (Fig. 4d & f).
Area-based non-photorespiratory respiration rates at light were independent of leaf age in all species (r2 = 0.12, P > 0.2 for Q. ilex, r2 = 0.47, P > 0.1 for L. nobilis and r2 = 0.00, P > 0.9 for O. europea).
All these age-dependent changes were qualitatively identical with the characteristics standardized with respect to the current-year leaf values (data not shown).
The photosynthetic potentials were generally strongly linked with the mesophyll diffusion conductance both on an area (for Jmax: Fig. 5a; for Vcmax: r2 = 0.78, P < 0.001 for Q. ilex, r2 = 0.89, P < 0.01 for L. nobilis and r2 = 0.17, P > 0.2 for O. europea) and a mass basis (for Jmax: Fig. 5b; for Vcmax: r2 = 0.94, P < 0.001 for Q. ilex, r2 = 0.91, P < 0.005 for L. nobilis and r2 = 0.47, P < 0.03 for O. europea) with the exception of area-based values of O. europea. Nevertheless, the latter data apparently fitted the broad interspecific relations between gm and photosynthetic potentials (Fig. 5a).
Mesophyll conductance limits photosynthesis more in older leaves
Average stomatal conductance at an average ambient CO2 concentration of 335 µmol mol−1 was negatively associated with leaf age in Q. ilex(Fig. 6a), but there was a weak positive correlation between leaf age and internal CO2 concentration (Ci) in this species (Fig. 6b). Neither the stomatal conductance (Fig. 6a) nor the internal CO2 concentration (Fig. 6b) were related to leaf age in the other two species (Fig. 6a & b). Thus, stomatal constraints of photosynthesis were only weakly associated with leaf age. Average stomatal conductance and net assimilation rate at an average ambient CO2 concentration of 335 µmol mol−1 were strongly correlated in all species.
Despite the correlations between photosynthetic potentials and mesophyll diffusion conductance (Fig. 5), the ratio of chloroplastic (Cc) to internal CO2 concentration (Cc/Ci) scaled positively with mesophyll diffusion conductance (Fig. 7a), and the CO2 drawdown from the internal air space to chloroplasts (ΔC = Ci − Cc) was larger in leaves with lower diffusion conductance (Fig. 7b). The correlations were generally similar for mass based gm, except for O. europea, in which the mass-based gm explained a larger fraction of variance in Cc/Ci versus gm (r2 = 0.80, P < 0.001) and in ΔC versus gm (r2 = 0.68, P < 0.01). Overall, these correlations implied that Cc/Ci ratio scaled negatively (Fig. 7c) and ΔC (Fig. 7d) positively with leaf age.
Species differences in CO2 drawdown
At low mesophyll diffusion conductances of 0.02–0.04 mol m−2 s−1, the CO2 drawdown was larger in Q. ilex than in the other species (Fig. 7b. P < 0.001 according to anova). Greater CO2 drawdown in Q. ilex than in the other species at common gm (Fig. 7b) was associated with larger net assimilation rates at common ambient CO2 concentration, Vcmax and Jmax. As the result of this, the major drop in CO2 concentration from the ambient air to chloroplasts occurred between the diffusion pathway from substomatal cavities to chloroplasts in Q. ilex.
Larger net assimilation rate in Q. ilex was the outcome of a greater Ci(average ± SE = 211.7 ± 4.4 µmol mol−1) at a common ambient CO2 concentration of 335 µmol mol−1 than in the other species (155 ± 5 µmol mol−1 in L. nobilis and 166 ± 6 µmol mol−1 in O. europea, means are significantly lower from the average Ci in Q. ilex, P < 0.001 according to anova; Fig. 6b).
Correlations between leaf structure and mesophyll diffusion conductance
Mesophyll diffusion conductances per unit area (r2 = 0.67, P < 0.001 for Q. ilex, r2 = 0.68, P < 0.05 for L. nobilis and r2 = 0.38, P < 0.05 for O. europea) and dry mass (r2 = 0.77, P < 0.001 for Q. ilex, r2 = 0.73, P < 0.02 for L. nobilis and r2 = 0.60, P < 0.01 for O. europea) were negatively associated with leaf dry mass per unit area (MA). Analogous correlations were also observed between gm per unit area and mass and leaf dry to fresh mass ratio (FD, data not shown). As the result of these correlations, CO2 drawdown from the internal air space to chloroplasts increased with increasing MA(Fig. 8a) and FD (Fig. 8d), whereas Cc/Ci ratio scaled negatively with both MA and FD(data not shown).
Implications for photosynthesis versus nitrogen relationships
The apparent fraction of leaf nitrogen in bioenergetics (Eqn 6, Fig. 9a) and in Rubisco (Eqn 7, Fig. 9b) was negatively associated with leaf age in Q. ilex, and the declining trend was also apparent in L. nobilis, but not in O. europea (Fig. 9a & b).
Jmax per unit area was positively related to nitrogen content per unit area (NA) only in Q. ilex(Fig. 10a), whereas the correlation between Jmax per unit dry mass and nitrogen content per unit dry mass (NM) was significant in Q. ilex and L. nobilis (Fig. 10b). Although the range of variation in NM was limited, the data of O. europea apparently also fitted the Jmax/mass versus NM relationship. For all data pooled, linear regressions yielded r2 = 0.59 (P < 0.001) for Jmax/mass versus NM, and r2 = 0.25 (P < 0.005) for Jmax/area versus NA. The correlations were analogous with Vcmax(data not shown).
For all data pooled, NM scaled negatively with MA(r2 = 0.38, P < 0.001). For pooled data, MA was negatively associated with Jmax/mass (r2 = 0.48, P < 0.001) and Vcmax/mass (r2 = 0.52, P < 0.001), but not with Jmax/area (r2 = 0.07, P > 0.1) and Vcmax/area (r2 = 0.11, P = 0.07).
Contribution of older leaves to total tree foliage in Mediterranean evergreen species
We observed that the fraction of total foliage biomass in older leaf age-classes is between 0.2 and 0.5 in the three Mediterranean species studied (Fig. 2a). Previous work has shown that current-year leaves may even comprise less than half of total tree foliage (Sala i Serra 1992), underscoring the overall significance of older leaves in Mediterranean evergreen species. In slowly growing canopies, such as those in water-stressed Mediterranean environments, the fractional contribution of older leaves may be particularly large (Niinemets & Lukjanova 2003b). This is because the fraction of leaves with certain age is a function of both leaf longevity and the rate of new shoot production. Provided all shoots produced are of the same length and with the same number of leaves per unit shoot length, the number of leaves with age Λ relative to the current year leaf number, N(Λ), is (Niinemets & Lukjanova 2003b)
where Rb is the number of shoots produced in current year relative to the number of shoots produced in the previous year (bifurcation ratio), ΛC denotes the age of current-year leaves, and S(Λ) is the leaf mortality function that characterizes the fraction of remaining leaves with age Λ. Apart from reductions in growth, average leaf age further increases in stressed plants due to enhanced leaf longevity (Kayama, Sasa & Koike 2002; Wright & Westoby 2002).
In non-senescent leaves of evergreen conifers, foliar nitrogen contents per dry mass (NM) generally continuously decrease with increasing leaf age with the values of NM varying more than two-fold between the youngest and oldest leaves (e.g. Niinemets & Lukjanova 2003a). In evergreen sclerophyllous Mediterranean oaks, several studies have demonstrated a relative time-independence of leaf N contents per dry mass (de Lillis & Fontanella 1992; Rapp et al. 1992; Robert et al. 1996; Niinemets et al. 2004). However, only 0- to 2-year-old leaves have been sampled in these studies. In our study, NM also changed by only approximately 20% between current-year and 2-year-old leaves in Q. ilex (Fig. 2c), whereas the corresponding changes were approximately 30% in O. europea and 100% in L. nobilis (Fig. 2c). Thus, these data confirm the relatively slow decline in NM in Q. ilex and O. europea, but also highlight the faster rate of change in L. nobilis.
Nitrogen contents per unit area (NA) were invariable in O. europea, and changed only approximately 5% between the leaf age of 0–2 years in Q. ilex and 40% in L. nobilis (Fig. 2d), demonstrating that the primary explanation for age-dependent decreases in NM was the accumulation of cell-wall components and corresponding dilution of foliar nitrogen concentration rather than net N retranslocation (NM = NA/MA).
Changes in internal diffusion conductance (gm) and photosynthetic potentials (Jmax and Vcmax) with leaf age
The range of gm values in current year leaves of 0.05–0.1 mol m−2 s−1 (Figs 3a & 4a) agrees well with previously observed values of gm of 0.06–0.12 mol m−2 s−1 in current-year leaves of O. europea (Centritto et al. 2003) and Q. ilex (Loreto et al. 1992). These values are relatively low compared with the gm values of 0.15–0.8 mol m−2 s−1 obtained for herbs or temperate deciduous trees (Evans & Loreto 2000; Flexas et al. 2004; Terashima, Araya & Miyazawa 2005). Even the initial values were relatively low, there was further a large age-dependent decline in gm to values of 0.015–0.04 mol m−2 s−1 in all species (Figs 3b & 4a). Furthermore, the mass-based estimates of gm varied five- to seven-fold between the youngest and oldest leaves within species, and altogether 10-fold for all data pooled (Fig. 4b), highlighting the strong effect of leaf age on internal diffusion conductance.
Because the utility of optical methods of photosynthesis measurements depends of the distribution of light and photosynthetic capacity within the leaf (Evans & Vogelmann 2003; Evans et al. 2004), it may be argued that the age-dependent decline in gm represents an artefact of the combined chlorophyll fluorescence/gas exchange technique. However, the values of gm determined by an alternative method that uses the changes in the shape of the A–Ci curves to evaluate gm (Eqn 5, Ethier & Livingston 2004) provided essentially the same estimates of gm as the combined fluorescence/gas exchange approach (Fig. 1). This demonstrates that our observations of the age-dependent declines in gm are robust and not biased by the analysis technique.
A common age-dependent modification in leaf functioning in non-senescent leaves of evergreen species is the decrease in leaf photosynthetic potentials (Field 1983; Ishida et al. 1999; Niinemets et al. 2004). In our study, the capacities for photosynthetic electron transport (Jmax) and the maximum carboxylase activity of Rubisco (Vcmax) varied five-fold on an area basis for all data pooled (Fig. 4c & e), and seven-fold on a mass basis (Fig. 4d & f). Such important modifications in foliage functioning are commonly not included in scaling of photosynthesis from leaf to canopy scale in Mediterranean environments (Sala & Tenhunen 1996; Falge et al. 1997; Rambal et al. 2003). Given the importance of older leaves (Fig. 2a) and extensive age-dependent changes in foliage functioning, we argue that age effects need consideration in scaling photosynthesis rates from single leaves to higher hierarchical levels.
The correlation between mesophyll conductance and photosynthetic capacity is not absolute
There are strong positive relationships between foliage photosynthetic capacity and gm (von Caemmerer & Evans 1991; Loreto et al. 1992, 2003; Evans & von Caemmerer 1996; Evans & Loreto 2000; Warren et al. 2003; Flexas et al. 2004; Grassi & Magnani 2005). These correlations have been interpreted as indicative of a co-ordination between photosynthetic capacity and gm such that the photosynthesis limitation due to gm is similar in leaves with different architecture and photosynthetic capacity (Evans & Loreto 2000). Although we observed strong correlations between the photosynthetic capacity and gm, the correlations were curvilinear rather than linear (Fig. 5), implying that gm varied relatively more than photosynthetic potentials. As the result of this curvilinearity, the ratio of chloroplastic to internal CO2 concentrations (Fig. 7a & c) increased and the CO2 drawdown from the internal air space to chloroplasts (Fig. 7b & d) decreased with increasing gm, overall indicating that gm exerted a greater control over net assimilation rates in older leaves (Fig. 7c & d). Given that stomatal constraints were only weakly age-dependent and non-photorespiratory respiration rates were independent of age, the age-dependent modifications in gm were the primary factor affecting the realized net photosynthesis rates in different-aged leaves with common photosynthetic potentials.
A similar curvilinearity of photosynthetic capacity versus gm relations has been observed for current-year leaves of various O. europea cultivars (Loreto et al. 2003), in New Zealand native conifers (DeLucia et al. 2003) and for a set of Mediterranean and herbaceous Mediterranean species (Flexas et al. 2004), suggesting that the non-coordinated changes in mesophyll diffusion conductance and photosynthetic potentials may be more general. In fact, Niinemets & Sack (2005) recalculated Cc/Ci ratios from a series of published studies for a large number of species and found a strong positive scaling of Cc/Ci ratio and gm. As in our study, Niinemets & Sack (2005) found that mass-based values of gm were more strongly correlated with Cc/Ci ratio. Stronger correlations with mass-based gm possibly reflect the circumstance that Ci is an average value for leaf internal gas-phase volume, and Cc is the average value for total chloroplast volume that both scale with mass rather than with area. Overall, these data collectively suggest that decreases in mesophyll diffusion conductance generally do result in greater diffusion limitations of photosynthesis.
Why did gm vary in leaves with different age?
Liquid-phase conductance for CO2 diffusion scales positively with chloroplast to total leaf surface area (Evans et al. 1994; Evans & von Caemmerer 1996; Hanba et al. 2001). As leaf photosynthetic capacity decreases in older leaves (Fig. 4c–f), so does the number of chloroplasts, resulting in a decline in gm. This relationship probably provides the mechanistic explanation for the relationships between gm and leaf photosynthetic potentials.
Apart from chloroplast area, gm varies significantly with leaf structural variables. In particular, gm often decreases with increasing MA (Syvertsen et al. 1995; Kogami et al. 2001; Terashima et al. 2005). In contrast, some studies have found a positive correlation between MA and gm (Hanba, Miyazawa & Terashima 1999; Evans & Loreto 2000; Piel et al. 2002). This discrepancy among the studies is because increases in MA may either reflect accumulation of cell wall compounds with concomitant increases in tissue density, or be associated with greater number of mesophyll layers and accordingly greater chloroplast to total leaf surface area ratios, for example, during sun-shade acclimation (e.g. Niinemets, Valladares & Ceulemans 2003). In our study, the age-dependent decline in gm, and increases in CO2 drawdown from the internal air space to chloroplasts were associated with larger MA and leaf dry to fresh mass ratio (Fig. 8). As MA was negatively related to nitrogen concentration, age-dependent increases in MA reflected the increase in the fraction of cell walls in the leaves rather than changes in mesophyll layers. In young developing leaves (leaf age 0–40 d) of broad-leaved evergreen trees, it has been shown that cell wall thickness increases with increasing leaf age, resulting in lower Cc/Ci ratio in older leaves (Miyazawa & Terashima 2001; Miyazawa, Makino & Terashima 2003). Damesin et al. (1998) have further demonstrated that in Mediterranean oaks, there are continuous increases in cell wall lignification and total amount of cell walls with leaf age. We suggest that these changes provide the primary explanation for the negative scaling of gm with MA and leaf dry to fresh mass ratio in our study (Fig. 8).
Species differences in age-dependent modifications in leaf functioning
Qualitatively similar age-dependent changes in leaf structure, chemical composition, mesophyll diffusion conductance and photosynthetic potentials were observed in all species (Figs 2, 3, 4 & 7) with the exception of nitrogen content and photosynthetic potentials per area in O. europea (Figs 2d, 4c & e). The species mainly differed in the rate of change in foliage structural and functional characteristics with age (cf. L. nobilis versus Q. ilex and O. europea in Figs 2, 3, 4 & 7).
In addition, the CO2 drawdown at common mesophyll conductance and leaf age (Fig. 7b & d) and MA (Fig. 8a) was larger in Quercus ilex than in the other species. This raises the question of why Q. ilex differed from the other species, when gm and photosynthetic potentials fitted the same curvilinear relations in all species (Fig. 4). When Rubisco-limits the assimilation, chloroplastic CO2 concentration corresponding to a certain combination of Vcmax, net assimilation rate (A) and non-photorespiratory respiration rate (Rd) is equal to:
where Km is the effective Michaelis–Menten constant for CO2 (Farquhar, von Caemmerer & Berry 1980). Thus, greater CO2 drawdown in Q. ilex than in the other species at common gm (Fig. 7b) and MA (Fig. 8a) was associated with larger net assimilation rates at a common ambient CO2 concentration, Vcmax and Jmax. Larger A, in turn, was the outcome of a greater Ci at a common ambient CO2 concentration in Q. ilex (Fig. 6b), possibly because of deeper rooting depth and/or more complete recovery from summer water stress. As a result of a lower degree of stomatal limitation, the major drop in CO2 concentration from the ambient air to chloroplasts occurred between the diffusion pathway from substomatal cavities to chloroplasts in this species. Such a dependence of CO2 drawdown on actual net assimilation rates has been previously experimentally shown for a series of species (von Caemmerer & Evans 1991). In the field, the overall significance of gm-limitations strongly depends on actual stomatal openness. As variations in stomatal openness are primarily the function of soil water status, the relative degree of stomatal and mesophyll diffusion limitation significantly varies during the season in communities experiencing summer drought (Grassi & Magnani 2005).
It has been suggested that a certain co-ordination exists between mesophyll diffusion and stomatal conductances such that decreases in stomatal conductance result in reduced gm values (Loreto et al. 2003; Flexas et al. 2004). However, in our study, species differences in stomatal conductance and gm were not apparently linked (Figs 4a & 6a). Furthermore stomatal conductance was independent of age in two species (Fig. 6a), and stomatal limitations were essentially independent of age in all species (Fig. 6b). Thus, in our study, age-dependent changes in gm were not biased by a possible co-ordination of stomatal and internal diffusion conductances.
Implications for scaling of leaf photosynthetic capacities with nitrogen
Age-dependent changes in photosynthetic capacity were associated both with decreases in foliage nitrogen contents (Fig. 2c & d) and with lower fractions of nitrogen in photosynthetic machinery in older leaves (Fig. 9). With increasing cell wall thickness in ageing leaves, a greater fraction of leaf N may be trapped in cell walls (Dyckmans et al. 2002), providing an explanation for an apparently larger fraction of non-photosynthetic N in older leaves. It has been previously reported that the apparent fractions of leaf N in photosynthetic machinery varied approximately two-fold between current- and 1-year-old-leaves in Mediterranean oaks (Niinemets et al. 2004). However, the changes in apparent nitrogen investments in their study probably also reflected age-dependent modifications in gm, which affects Jmax and Vcmax calculations from A–Ci curves, and thereby the derivations of apparent nitrogen fractions (Eqns 6 & 7). When the values of Jmax and Vcmax were derived using chloroplastic CO2 concentrations, the apparent nitrogen investments varied only 20% for the first two leaf age classes in Q. ilex and L. nobilis and were independent of age in O. europea (Fig. 9).
Area-based leaf photosynthetic potentials were less strongly associated with nitrogen than the mass-based photosynthetic potentials (cf. Fig. 10a & b). Low fraction of explained variance in area-based photosynthetic capacity versus nitrogen relations was mainly because the variation in NA primarily resulted from changes in MA that actually scaled negatively with photosynthetic potentials per mass. This triangular network of correlations explains the contrasting scaling of area- and mass-based photosynthetic capacities for current-year leaves of a large range of species (Wright et al. 2004). As our study demonstrates, this functional correlation network also applies to various-aged leaves. However, there was a significant degree of variation also in the correlations of mass-based photosynthetic potentials with N (Fig. 10b). For instance, at a NM of 2%, Jmax/mass varied two-fold. This variation was mainly because of age- and species-dependent modifications in fractional N investments. Thus, the age effects should be taken into account in scaling of foliar photosynthetic potentials with nitrogen.
These data jointly indicate that leaf age exerts a major control on foliage photosynthetic capacities and internal diffusion limitations of photosynthesis, and also affects the functional scaling of foliage photosynthetic potentials with leaf structural variables and nitrogen.
Financial support was provided by the Estonian Science Foundation (grant 5702), the Estonian Ministry of Education and Science (grant 0182468As03) and the Province of Trento, Italy (grant DL3402).