Internal conductance under different light conditions along the plant profile of Ethiopian mustard (Brassica carinata A. Brown.).

This study focused on the internal conductance (g(i)) along the plant profile of Ethiopian mustard under two light conditions: (i) light from the top only (I1); (ii) light from the top integrated by supplementary lateral light along the whole plant profile (I2). Lateral light strongly increased the productivity (e.g. +104% of seed oil) and net photosynthesis (A). The latter appeared more driven by g(i) (r=0.78**) than by stomatal conductance (g(s)) (r=0.51*). Importantly, irradiance also considerably shortened the time from leaf appearance to senescence, which means that corresponding leaves in I1 and I2 had different ages. Therefore, since leaf age and irradiance have counteracting effects on g(i), I1 sometimes showed higher g(i) values than I2. With respect to irradiance, leaf age had clearly higher effects on g(i), which radically declined from the top to the basal leaves, even under constant light conditions. The internal conductance caused a significant drawdown of CO(2) from the sub-stomatal cavity (C(i)) to the site of carboxylation (C(c)) that, in turn, led to a substantial underestimation of V(cmax) calculated using the A/C(i) model. Again, the trends of g(i) and g(s) were not consistent along the plant profile, and so the ratio between stomatal and internal limitations to A changed from top to bottom leaves, accordingly. This study suggests that g(i) may be a valuable trait for increasing photosynthetic capacity and productivity; nonetheless, it suggests caution in selecting leaves for high g(i), as the latter can considerably change along the plant profile due to leaf age and irradiance effects.


Introduction
Ethiopian (E.) mustard (Brassica carinata) is an interesting C 3 oil crop evolved through the inter-specific hybridization between B. nigra and B. oleracea (Hemingway, 1995). Because of the high erucic acid content of its seeds, it has recently attracted a growing interest in a number of nonfood applications such as biodiesel, bio-polymers, lubricants, soaps, and surfactants (Becker, 1999). In addition, the residual defatted meal has some precursors of biologically active compounds (glucosinolate) which could be used as soil amendments for plant defence (FAIR-CT96-1946 Project report; Anon., 2000). Due to its higher tolerance of drought (Cardone et al., 2003) and the lower tendency to pod-shattering, E. mustard could be very competitive against rapeseed under unfavourable environmental conditions. Furthermore, it produces a higher biomass than rapeseed that could eventually be processed into electricity and/or heat.
Studying internal conductance (g i ) is of enormous importance in crops like E. mustard which are envisaged to be grown in environments where high water use efficiency is imperative. There is, in fact, evidence that increasing g i may improve photosynthesis and water use efficiency (Warren and Adams, 2006), while the increase in stomatal conductance (g s ) generally involves significant transpiration costs that crops might be unable to support. This was emphasized in a number of studies (Evans and Vellen, 1996;Lauteri et al., 1997;Flexas et al., 2008) showing consistent and positive correspondence between water use efficiency and the g i /g s ratio.
There is increasing evidence that g i can significantly decrease the CO 2 concentration at the Rubisco site (for a review, see Ethier and Livingston, 2004) to an extent depending on water shortage, salinity, leaf age, mineral nutrition, etc. (Bernacchi et al., 2002;Long and Bernacchi, 2003). The irradiance can significantly modify the leaf anatomy and mesophyll structure (Nobel, 1991;Parkhurst, 1994) and thus it can be envisaged to affect also g i . Nonetheless, as far as is known (for a review, see Flexas et al., 2008), only eight studies out of over 130 articles on g i documented the relationships between irradiance and g i (Lloyd et al., 1992;Hanba et al., 2002;Piel et al., 2002;Gorton et al., 2003;Warren et al., 2003Warren et al., , 2007Laisk et al., 2005), and, of these, only three involved herbaceous crops (Laisk et al., 2005;Flexas et al., 2007a;Evans et al., 2008). Moreover, most of these experiments used natural light gradients or examined instantaneous responses to irradiance (Evans et al., 2008). By contrast, the present study is a manipulative mid-term acclimation experiment, which should provide greater power to disentangle the various determinants of g i . Specifically, the experiment focused on the influence of irradiance upon g i along the plant profile.

Site description
In 2007, 24 plants of E. mustard, variety CT 204, were grown in 0.21-m-diameter plastic pots (two plants per 6.6 l pot) in a 232.5 m (2 m height) controlled-environment room where relative humidity, temperature, and photoperiod could be regulated and monitored. Light was provided by six groups of 12 independently controllable neon tubes (Philips Master TL-D 58W/840). Pots were weighed twice a week and watered immediately after, thus maintaining the soil moisture constantly close to the field water capacity, the latter being determined gravimetrically. The soil substrate was a mixture (2:1, v/v) of sand and neutral peat (46% C; 0.8% N; pH 6). The Ruakura solution, a nutrient solution especially designed for growing plants in sandy soils (Smith et al., 1983), was applied at a dose of 150 ml kg À1 of soil once a month or once a week during wintertime and after the spring regrowth, respectively.
At the start of light treatment, all plants had at least 10 visible pods and 80% of open pollinated flowers. Plants were randomly disposed in two separate rows. The distance between two succeeding plants was maintained close to that in the field (about 4 cm). Daylength was 16 h, while light/ dark temperatures (°C) and relative humidity (RH%) were 25/15 and 50/80, respectively. The environmental parameters inside the room were constantly monitored by a micrometeorological station (lMetos, Pessl Instr., Weiz, Austria) and, at the same time, the ambient CO 2 was measured (IRGA-WM-4 plus datalogger; PP-Systems, Hertfordshire, UK). This concentration was used to set the [CO 2 ] in the air entering the cuvette during the gas-exchange measurements.
The growth room was modified in order to provide, from one side, light from the top only (I1), and from the other, light from two sides, i.e. light from the top plus supplementary lateral light through horizontal neon tubes (I2). A plastic wall permeable to air flux was placed between the two treatments in order to protect I1 from the lateral light of I2. For each treatment, photosynthetic photon flux density (PPFD) was measured vertically along the plant profile, and horizontally throughout the plant, by a 0.75-mlong portable light sensor system with 30 PAR sensors covered by a diffuser (EMS 7/L Canopy Analyser; S.W & W.S. Burrage, Kent, UK) (Fig. 1).

Growth and production
Phenological stages were determined every other week on four randomly selected plants in each treatment. To distinguish different growth stages, the CETIOM phenological scale (CETIOM, 2003) was used, which is an alphanumeric code tailored for rapeseed, representing seven main growing stages and a number of sub-stages classifying the plant growth from emergence to seed physiological maturity. In addition, plant height and the number of leaves, siliqua, and seeds, as well as their areas, were measured using allometric models specially created for this experiment (data not shown). Briefly, the allometric models for leaves and siliqua were obtained from the linear relationship between their actual areas as measured by an area meter (Li-3000; LI-COR, Nebraska, USA), and the estimated areas, which were calculated from the rectangles subtending them. The seed area was calculated assuming the seed shape is a sphere.
In all plants, three arbitrary and equally dimensioned (0.3 m height) plant layers (L1-L3) were characterized on the base of their productivity parameters (biomass, siliqua, seed, and oil production) and physiological traits. The seed oil was analysed according to the Soxhlet's method (Soxhlet, 1879). To extract oil completely, 0.5 g seed flour was mixed with Na-sulphate anhydrous and quartz sand then treated in the Soxhlet device for 6 h in a solution of 200 ml n-hexane. Because of the low seed production in each single layer, the amount of flour was slightly lower than indicated in the official Soxhlet method. For this reason, the Soxhlet device was first subjected to a test run in order to correct systematic errors due to the small samples.

Physiological parameters
Leaf gas-exchange measurements were performed by a portable open-path system (CIRAS-2; PP-Systems). During measurements the air flow entering the chamber was 250 cm 3 min À1 , RH ;60%, leaf temperature 2460.5°C (mean 6SD), the latter calculated using the energy balance method. Leaf-to-air vapour pressure deficit was between 0.8 kPa and 1.0 kPa.
In three layers (L1-L3) of three labelled plants per treatment (n¼3), the two main parameters representing photosynthetic capacity, i.e. the maximum carboxylation rate (V cmax ) and the electron transport rate (J max ), were determined at 7 and 21 d after the start of treatment (DAT) (36 A/C i curves in total) at a constant PPFD of 1500 lmol photons m À2 s À1 provided by an LED light unit above the cuvette. V cmax and J max were calculated using the biochemical model proposed by Farquhar et al. (1980), which represents the mechanistic relationship between net photosynthesis (A) and the intercellular CO 2 concentration (C i ). Briefly, the model relies on the concept that A is co-limited by three processes: (i) the activity and kinetics of Rubisco (W c ); (ii) the regeneration of ribulose-1,5-bisphosphate (RuBP) that is supported by the electron transport chain (W j ); (iii) the availability of triose phosphates for the Calvin cycle (W p ): where C* is the CO 2 compensation point in the absence of dark respiration; R d is the day respiration, i.e. the mitochondrial respiration during illumination. In this experiment, the apparent C* (C a *) and R d were determined in the three plant layers (L1-L3) of the I1 plants using the method described by Laisk (1977), i.e. the determination of the intercept of four A/C i curves generated at different low PPFDs (from 100 to 500 lmol photons m À2 s À1 ) for C i values below 15 Pa. To calculate the true C*, C a * was corrected according to von Caemmerer and Evans (1991) where g i is the internal conductance. Since C* and R d were not significantly different along the plant profile, their average values were taken for the subsequent calculations: C* and R d were 35.863.2 lmol CO 2 mol À1 and 0.8060.11 lmol CO 2 m À2 s À1 (mean 6SD), respectively. These values were also used for I2 as previous studies showed that both C* and C a * were not significantly affected by irradiance level (Piel et al., 2002;Warren et al., 2007).
Each point of the A/C i curve took about 4 min. Measurements started at C a , the [CO 2 ] in the incoming air, of about 400 lmol mol À1 . First, C a was progressively reduced to 200 and 100 lmol mol À1 and then gradually increased up to about 2000 lmol mol À1 across the series of 400, 500, 600, 800, 1000, and 1600 lmol mol À1 using a CO 2 mass flow controller (generally 10 data points for each A/C i curve). Each data point was recorded after equilibration to a steady state (CV <2% was considered tolerable). The software Photosyn Assistant (Dandee Sci., Scotland, UK) was used to fit the model and calculate the photosynthetic parameters. Before running the model, it was assumed that the threshold between W c and W j was 20-25 Pa (Wullschleger, 1993;Manter and Kerrigan, 2004).
Finally, chlorophyll index (SPAD-502 chlorophyll meter; Minolta), which is widely known to be related to leaf nitrogen content (Debaeke et al., 2006), was measured in the leaves used for gas-exchange analysis.

Soluble sugars extraction, photosynthetic pigments, and carbon isotope composition
Two leaf discs (15 mm diameter) were collected about 4 h after each A/C i curve: one disc was used to determine the relative water content, i.e. the ratio of tissue fresh weight minus dry weight to tissue turgid weight minus dry weight, the specific leaf area (SLA, leaf area/dry mass) and photosynthetic pigments [chlorophylls a and b (Chla and b) and carotenoids (Car)]; the other disc was used for the extraction of leaf soluble sugars and for the determination of the carbon isotope composition (d 13 C). Briefly, the soluble sugars were first dried at 40°C in a ventilated stove, then re-diluted with 8 ml water, and finally centrifuged for 20 min at 3000 g. The solution was sequentially filtered with two ionic-exchange resins, DOWEX-50 (H + ) and DOWEX-1 (Cl -), in order to separate amino acids and organic acids from soluble sugars (Brugnoli et al., 1988). Successively, 2 ml sugar solution was treated with a colouring DNS solution (3.5-dinitrosalicylic acid) in order to quantify the extracted sugars through the absorbance at k 530 nm (Summer, 1921). The rest of the solution was maintained at -80°C and then analysed for d 13 C. The latter was determined through the combustion of 1 g sample in an elemental analyser (model NA 1500; Carlo Erba, Milan, Italy), coupled to a dual-inlet mass spectrometer (model SIRA II; GV-Instruments, Middlewich, UK). d 13 C of the sample was compared with that of a working standard CO 2 calibrated against the international standard Vienna-Pee Dee Belemnite. The possible fractionation during combustion was controlled by testing a standard sucrose (Sigma Chemicals, USA) with a known d 13 C (-25.0960.06&). The carbon isotope discrimination (D&) was then calculated as: where d a and d p are the carbon isotope composition of source air and plant material, respectively, relative to Vienna-Pee Dee Belemnite.
Internal conductance and C c As given by Brugnoli et al. (1988) and further developed by Lauteri et al. (1997), g i was calculated through the discrepancy between D determined in soluble sugars (D s ) and the expected D (D i ) calculated using the Farquhar model (Farquhar et al., 1982): where, b is the discrimination associated with carboxylation reactions; e s , the fractionation occurring during the dissolution of CO 2 ; a 1 , the discrimination caused by the diffusion of CO 2 in the liquid phase; f, the discrimination during photorespiration. Details on values and references were reported elsewhere (Monti et al., 2006). The concentration of CO 2 at the site of carboxylation (C c ) was then obtained as C c ¼C i ÀA/g i . Once C c was calculated, V cmax and J cmax were recalculated using the A/C c model (V cmax_Cc ; J cmax_Cc ).
Stomatal and internal limitations to photosynthesis were calculated as proposed by Jones (1985). In short, the drawdown of CO 2 from the leaf surface to the site of carboxylation is a function of stomatal and internal conductance, i.e.
C c ¼ C a ÀðA=g s ÞÀðA=g i Þ Assuming that mitochondrial respiration is not significantly changing during the experimental period, the relationship between A and V cmax can be expressed (Farquhar et al., 1980) as: where A max is the light-saturated net photosynthesis. As dC c can also be expressed as: Combining the last three equations gives the following: Therefore, the stomatal (l s ) and internal (l i ) limitations can be calculated as: where g tot is the total conductance to CO 2 (1/g tot ¼1/g s + 1/g i ).

Statistical analysis
All data were subjected to statistical analysis according to the general linear model for repeated measurements performed by SYSTAT 10.2 (Systat Software, Inc., Chicago, IL, USA). The Bartlett's test (P <0.05) was used to test the homogeneity of the variance. The LSD Fisher's test (P <0.05) was applied to separate the significantly different means. The statistical significance of the correlation coefficients was tested according to Pearson's correlation test (P <0.05). To solve the area under a curve (see Fig. 1), the trapezoidal rule was used.

Light effect on growth and production
Increasing the light distribution along the plant profile visibly modified the cycle length and productivity: I2 took 14 d less than I1 to complete the cycle (Fig. 2) and it produced 47%, 140%, and 61% more biomass, siliqua, and seeds, respectively (Fig. 3). In addition, I2 had a higher oil content (%) than I1, with a consequent double oil yield (Fig.  3). Similarly, the size of siliqua and seed was significantly affected by the irradiance: the total siliqua and seed areas were 71 and 31 cm 2 higher in I2 than I1. The leaf area (LA) and SLA were conversely not influenced by the irradiance decreasing rapidly with seed maturation. Nonetheless, LA greatly changed along the plant profile: in L1, it was 2-and 5-fold as high as in L2 and L3, respectively. SLA was lower in L3 (25 m 2 kg À1 ) and similar in L1 and L2 (37 m 2 kg À1 , on average). The relative water content averaged 0.8560.09 and was not influenced by irradiance or by leaf position along the plant profile. The oil content (%) was also modified by light intensity: 25% higher oil content was found in I2 compared with I1 (Fig. 3).

Light effects on g i
Since the light regime substantially modified the cycle length, the corresponding leaves in I1 and I2 had different ages. Therefore, the differences in g i between I1 and I2, within each layer, probably reflected cumulative effects of leaf age and irradiance. Nonetheless, a reliable estimate of the leaf age effects on g i could be obtained in I2, since these plants maintained a constant light level along the profile (Fig. 1). Similarly, the effects of the leaf age on g i were quantified through the comparison of the same leaf at 7 and 21 DAT.
The internal conductance drastically declined from top to basal leaves irrespective of irradiance level (Fig. 4). Therefore, leaf age effects are likely to have prevailed over irradiance effects in changing g i . Nevertheless, the influence of irradiance upon g i was not negligible, as testified by the significant interaction between irradiance and plant layer. The reason behind this interaction was that g i was significantly higher in L3 of I2 with respect to I1, while the opposite occurred in L1. By contrast, there were no significant differences in g i between I1 and I2, in L2 (Fig.  4). The interaction between irradiance and plant layer was also significant 14 d later when, however, it was no longer possible to measure L1 due to the advanced senescence of the basal leaves. Besides, in L2, g i was higher in I1, while in L3 it not longer differed between I1 and I2 (Fig. 4).

Photosynthesis and g i
Internal conductance had a significant impact on A, which appeared more driven by g i than by g s (Fig. 5). Overall, g i was fairly consistent with A, being higher in L3 and then gradually decreasing downward (Table 1). Moreover, since A declined downward more rapidly in I2 than in I1, the leaf-age effect on A was likely to be more prevalent than that of irradiance. A result which is consistent with the effects of leaf age on g i ; however, unlike for g i , I2 showed higher A values than I1 also in L2 (Table 1).
The leaf Chl and Car contents were also rather consistent with A. Specifically, Chla, Chlb, and Car did not change between L3 and L2, while it declined drastically in L1 (Table 2). Conversely, no significant effects of the irradiance were observed on Chla, Chlb, and Car. Similarly, SPAD values were only slightly influenced by the irradiance; however, unlike photosynthetic pigments, SPAD values  radically decreased downward, thus revealing a likely decline in leaf N-content parallel to leaf ageing (Table 2).
Stomatal conductance (g s ) was weakly influenced by the irradiance, and it appeared significantly related to g i (Fig.  5). Nonetheless, unlike g i , g s remained rather constant between L3 and L2 in I2. More than that, g i decreased more markedly than g s from the apex to the basal leaves, with the result that the ratio between g i and g s significantly changed with the position of the leaf and, to a lesser extent, with the irradiance (Fig. 6). Specifically, in L3, g i was about three times larger than g s both in I1 and I2, while in L2 g i was much larger than g s only in I2. In the basal leaves, g i and g s were nearly equivalent (Fig. 6). The diffusive limitations to A through stomata (l s ) averaged 21% and they were significantly higher than internal limitation (l m ) (Fig. 6). Overall, l s values were higher in the upper layer and then decreased downward, contrary to that observed for l m .
The latter, however, was not negligible, ranging from 9% to 15% in L3 and L1, respectively. The irradiance effects on diffusive limitations to CO 2 were generally negligible; the only significant effect was in L2 where l m was significantly higher in I2 than I1, both on DAT 7 and DAT 21 (Fig. 6).
The drawdown of CO 2 from the atmosphere to the carboxylation site was significantly influenced by g i . C c resulted in being 7-9% lower than C i in L2 and L3, and 12-14% lower in L1 (Table 1). This led to a significant underestimation of V cmax as calculated by the conventional A/C i model, and, to a lesser extent, of J max as well (Table 1). Importantly, the correlation between V cmax and J max significantly improved (r¼0.90**) once the two parameters were recalculated using the A/C c model.

Discussion
This study revealed that light distribution along the plant profile can enormously affect biomass, seed and oil yields, and, to a lesser extent, the fatty acid composition. Therefore, plant density and its architecture should be taken into account by agronomists and breeders in establishing the optimal growing conditions towards maximizing the productivity of this crop.
A bottom-up comparison of leaves under two light conditions was the original goal of this study. Nonetheless, irradiance had an unexpected strong influence upon the plant growth that caused an early maturation of I2 plants. Consequently, the difference between corresponding leaves of I1 and I2 were probably the consequence of the cumulative effects of the irradiance and leaf age. Anyway, since I2 plants were subjected to constant light intensities along the whole profile, the ageing effects on g i could be reasonably estimated along the profile of I2 plants or, possibly, by measuring the same leaf at 7 d and 21 d after the treatment started.
Generally, g i increases until the leaf is fully expanded and then decreases parallel to the leaf senescing (Hanba et al., 2002;Ethier et al., 2006). Similarly, the present results show Fig. 5. Relationship between net assimilation rate (A) and stomatal (g s , inset figure) or internal (g i ) conductance. r, Correlation coefficient; * and **, statistical significance of r for P <0.05 and <0.01, respectively (Pearson's correlation test). Each point represents a single leaf. Table 1. Net assimilation rate (A), CO 2 concentration at the substomatal cavity (C i ) and at the chloroplasts (C c, ) and maximum carboxylation (V cmax_Cc ), and electron transport (J max_Cc ) rates The two latter parameters were calculated on C c basis (A/C c curve). The underestimation (%) of V cmax and J max if calculated using the A/C i curve is given in parenthesis. Different letters indicate vertical significant differences (P <0.05, LSD Fishers's test) within each DAT (days after the start of treatment; n¼3).

DAT
Light that under I2 conditions, g i strongly decreased (from 0.527 to 0.074 mol CO 2 m À2 s À1 in L3 and L1, respectively), thus supporting the evidence of likely important effects of leaf age on g i (Warren, 2008a, and references therein). Moreover, since the downward decrease in g i occurred to a similar extent in I1 and I2, leaf age clearly seemed to have more prevalent effects on g i than irradiance. Nevertheless, the effect of the irradiance on g i should not be ignored since in L3, where the leaves were young and likely to be more active, g i changed significantly between I1 and I2, and this caused a diverse overestimation of the CO 2 concentration at the carboxylation site between I1 and I2. Similar to g i , A declined from top to bottom leaves; however, unlike g i , I2 showed higher A values than I1 also in L2, and therefore, irradiance can be supposed to have a higher influence on g i than A.
In spite of a considerable change in g i and A, C i and C c remained fairly constant along the plant profile (C i /C c ratio of 1.0860.14 SD), irrespective of the light treatment. It derives that C c was never significantly related to A; when A changed, g i and g s scaled accordingly and so C c remained constant. A fact that would indicate, presumably, a feedforward mechanism on the diffusive stomatal limitation to CO 2 through the internal CO 2 concentration. The drawdown from C i to C c (on average 30 lmol mol À1 ) was pretty low if compared with the average value of 226 species calculated by Warren (2008a). Nonetheless, it was not negligible as it determined a substantial underestimation of V cmax and, to a lesser extent, of J max as calculated by the conventional A/C i model. Specifically, V cmax and J max calculated on a C c basis were on average 1.24 (r¼0.91**) and 1.08 (r¼0.67**) higher than V cmax and J max calculated on the C i basis, respectively. These two coefficients were slightly lower (V cmax ) than or very similar (J max ) to those calculated (1.62 and 1.08, respectively) in a recent review (Warren, 2008a). Moreover, the correlation between V cmax and J max increased when the A/C c model was used instead of the A/C i model. As also ascertained in other herbaceous crops (Monti et al., 2006), not g i taking into account may lead to the decrease in A to be attributed to biochemical constraints, while diffusive limitations could be the major cause.
The sum of stomatal and non-stomatal limitations was about 30% with little variation along the plant profile. Nevertheless, the diffusive limitations to CO 2 were modified by the significant interaction between leaf age and irradiance. As a result, the relative importance of stomatal and non-stomatal limitations changed accordingly. Specifically, in the two upper layers, the diffusive limitations to CO 2 in mesophyll were about half those in stomata, while in the bottom leaves (L1) they were nearly equal or even higher in the mesophyll (Fig. 6).
Contrasting effects occurred on g i along the plant profile because g i is known to decrease with leaf ageing (Bernacchi et al., 2005;Grassi and Magnani, 2005;Flexas et al., 2007b), while it increases with irradiance (Terashima et al.,  2006). Since the leaves aged faster in I2 than in I1, it was not surprising that g i was sometimes higher in I1, as the older I2 leaves were probably less capable of responding promptly to light. Therefore, on the basis of these results, three scenarios along the plant profile could be identified. (i) At the top (L3), the leaves were still young both in I1 and I2. In this layer, g i was mostly influenced by light and, to a much lesser extent, by leaf age. Thus it follows that irradiance effects on g i could be reliably estimated in this layer, since the counteracting effects of leaf age on g i can be assumed to be negligible. In this layer, g i and irradiance were found to be positively related in agreement with other studies (Boardman, 1977;Warren et al., 2003;Evans et al., 2008;Flexas et al., 2008). (ii) In the intermediate layer (L2), I2 had significantly older leaves than I1. The negative age effects offset or prevailed over the positive effects of irradiance on g i , with the result that 21 DAT the internal conductance was higher in I1. (iii) In the basal leaves (L1), the age effects were further evident both in I1 and I2. There is evidence that g i and g s are positively related (Loreto et al., 1992;Centritto et al., 2003;Warren, 2008b). Nonetheless, Warren (2008b) demonstrated that this relationship is not ubiquitous as no relationship was found between g s and g i when g s was modulated by the atmospheric water deficit. The present study, though it showed g s and g i being generally related (r ¼ 0.51*), seems somehow to support the results of Warren (2008b) since g s and g i were sometimes independent in responding to irradiance. Under I2 conditions, g s remained fairly stable from L3 to L2 to then decreased appreciably in L1; conversely, g i declined by 25% from L3 to L2, and then it decreased drastically from L2 to L1 (from 0.335 to 0.074 mol m À2 s À1 ). Consequently, the responsiveness of stomata and mesophyll to light changed along the plant profile and so the g i /g s ratio changed accordingly. The reason behind the different responses of g s and g i to light was not clear. Whether or not the mesophyll can drive g s is still an open matter. Recent studies suggest that stomatal response to light is mostly driven by a signal generated by the mesophyll (Mott et al., 2008). Nonetheless, stomata also responded to red-light, probably through a mechanism regulated by the chloroplasts of the guard cells (Tominaga et al., 2001). As a result, mesophyll should have little or no effect on g s . Furthermore, Messinger et al. (2006) showed that stomatal response to redlight was evident even when C i was maintained constant through regulating the ambient CO 2 . Of course, the fact remains that other mechanisms might influence the relationship between stomata and mesophyll. Therefore, since the light source used in the present experiment (neon tubes) is characterized by a prevailing far-red-to-orange light spectrum, an hypothesis could be that this specific light spectrum, or the still uncertain responsiveness to blue-light of the mesophyll and stomata (Marten et al., 2007;Loreto et al., 2008), was the cause of the differing behaviour of stomata and mesophyll, as guard cells could be more sensitive to farred light than mesophyll receptors.
In summary, four main findings could be underlined: (i) productivity was strongly increased by the lateral light through the rise in photosynthetic rates of basal and intermediate leaves as well as other green organs (Ruuska et al., 2004); (ii) irrespective of irradiance level, g i declined drastically from the top to the bottom leaves thus suggesting a likely prevalent effect of leaf age compared with irradiance-for this reason, the positive effects of irradiance on g i are visible only in the top leaves; (iii) once again, g i led to significant underestimation of V cmax which was higher related to J max once it was recalculated using the A/C c curve; (iv) g i and g s were significantly related and they scaled with the photosynthetic capacity, but the ratio between stomatal and non-stomatal limitations was not constant along the profile.