Differential development of chloroplasts and mesophyll surface area in C. sieboldii
Both Smes and proportion of the intercellular air spaces attained their maximum values at around FLE in C. sieboldii (Figs 5 & 8). This indicates that the expansion of mesophyll cells and the epidermal cells was nearly completed at around FLE.
Chloroplast development in C. sieboldii was much slower than that in P. vulgaris, although leaf area expansion of these species proceeded similarly (Figs 2 & 4). In P. vulgaris, the number of chloroplasts on unit leaf area declined with leaf area expansion, which was also reported for Solanum tuberosum L. (Šesták 1985). Thus, substantial materials (i.e. carbon and nitrogen) are used for chloroplast development at the very early stages of leaf area expansion in P. vulgaris and S. tuberosum (normal greening species). On the other hand, in C. sieboldii, Sc increased much later than Smes as indicated by the increase in Sc/Smes with leaf age (Fig. 7). Such differential development of Smes and Sc in C. sieboldii would be important in understanding delayed greening. The low values of Sc/Smes of young expanding leaves might be effective in protecting the young chloroplasts from stresses from herbivory because such leaves have a large amount of structural compounds (Kursar & Coley 1992a,b).
Limitation of photosynthetic rate by respiration rate, and stomatal and internal CO2 transfer conductances in C. sieboldii
In the expanding leaves of C. sieboldii, the potential photosynthetic rate (Apg) was limited by high respiration rate (Fig. 13b). As the leaf tissues were still immature, energy for constructing these tissues would be mainly provided by respiration (Lambers, Szaniawski & de Visser 1983). Small intercellular air spaces (Fig. 8b) of the young leaves would apparently contribute to lowering the internal CO2 transfer conductance. However, gi was not a major factor limiting the rate of photosynthesis in the expanding leaves. The high rday of the young expanding leaves contributed to maintain high Ci and partly obviated the need for CO2 diffusion from the atmosphere to the chloroplasts (Miranda et al. 1981).
Stomatal conductance to water vapour (gs) attained the maximum values at around FLE both in P. vulgaris and C. sieboldii (Fig. 3c). The difference in the period taken for stomatal development would be much smaller than the difference in the period for chloroplasts development. After FLE, gs declined in P. vulgaris, whereas it stayed at a constant level in C. sieboldii. For many herbaceous species, it has been reported that gs attains the maximal value at around FLE and declines thereafter (Šesták 1985). Although the data are limited, the trend of changes in gs observed in C. sieboldii could be generally the case in the delayed greening species. In T. cacao, gs increased sharply with leaf area expansion and attained the maximum near FLE (Abo-Hamed, Collin & Hardwick 1983). In Brachystegia spiciformis, another delayed greening species, gs did not increase greatly in comparison with the marked increase in net photosynthetic rate after FLE (Choinski & Johnson 1993). The limitation of potential photosynthetic rate by stomatal conductance (Ls) increased after FLE in C. sieboldii (Fig. 13c). This was because gs did not increase greatly after FLE, although the potential photosynthetic rate increased.
There was a positive relationship between An and gi (Fig. 10). Using the on-line carbon isotope discrimination technique for gi estimation, Hanba also found a positive relationship between the two factors throughout the leaf emerging to senescing in Alnus and Acer (Dr Y.-T. Hanba, Okayama University; personal communication). The gi values estimated in the present study agreed very well to those obtained by the on-line discrimination method for the same species (Hanba et al. 1999).
Makino, Mae & Ohira (1983) reported that specific activity of fully activated Rubisco was almost constant from young to senescing leaves in rice. They also claimed that the activation state could not be a major factor responsible for the changes in photosynthetic rate during leaf ontogeny. Thus, the ontogenetic differences in the specific activity and the activation state of the present gi estimation were not taken into account in evaluating the present gi estimation. On the other hand, we made sensitivity analyses. When different specific activities of 18 and 32 mol mol−1 s−1 were used for the estimation of gi, the trend did not change markedly (Fig. 11). On the other hand, when the low specific activity of 15 mol mol−1 s−1 reported for rice (Makino, Mae & Ohira 1988) was used, some gi values were estimated to be negative and deviated from the values obtained by the on-line method. Judging from the agreement to the values obtained by the on-line method and the results of the sensitivity analyses, the trend of changes in gi obtained in the present study would be fairly robust. The gi values estimated with a combination of gas exchange and carbon isotope fractionation were reported for mature leaves of evergreen trees such as Citrus limon, Macadamia integrifolia (Syvertsen et al. 1995), Quercus glauca, Quercus Phillyraeoides and Camellia japonica (Hanba et al. 1999). These values were similar to the present values calculated for mature leaves of C. sieboldii (leaf age 20–35 d after FLE).
Day respiration rate calculated at the intersection of Γ* on the A–Ci curves is underestimated in the presence of internal CO2 transfer conductance (von Caemmerer et al. 1994). Therefore, the values of rday estimated in this study might be lower than actual day respiration rates. To estimate the maximum extent to which Apg values are affected by respiration rate, we also calculated Apg using dark respiration rate instead of rday. Dark respiration rate was always greater than day respiration rate as has been reported (Brooks & Farquhar 1985; Villar, Held & Merino 1994). The ratio of Ls or Li calculated using dark respiration rate to those calculated using rday ranged from 0·87 to 0·99. Thus, the calculated limitations by gs or gi did not differ markedly even though we might underestimate the day respiration rates. Effects of the overestimation on Lr could be considerable because the ratio of Lr calculated using dark respiration rate to those calculated using rday ranged from 1·1 to 1·9. However, the trend of Lr during leaf development did not change greatly.
We expected that the increase in An after FLE in delayed greening species would be partly caused by an increase in gi. We observed a dramatic increase in Sc after full leaf area expansion (Fig. 4d). The value of gi would be proportional to Sc if other conditions are unchanged (Evans et al. 1994). However, gi did not increase proportionally to the increase in Sc (compare Figs 4d & 11a). Moreover, gi even decreased at the later stages. The internal CO2 transfer conductance is separated into two components, gas-phase conductance (intercellular air space) and liquid-phase conductance (cell wall, cell membrane, cytosol, chloroplast envelope and stroma). The CO2 transfer conductance through the air-phase component would decrease after FLE because the increase in mesophyll thickness would result in a longer CO2 diffusion pathway from the substomatal cavities to the chloroplasts (Fig. 8a). The CO2 transfer conductance through the air-phase component was calculated from the mesophyll thickness, the fraction of mesophyll cells occupied by the intercellular air spaces and diffusion coefficient of CO2 in air according to Nobel (1991). The values obtained were very large; 1·6 ± 0·12 mol m−2 s−1 1(mean ± SD; n = 4) and 1·4 ± 0·23 mol m−2 s−1(mean ± SD; n = 8), at and after FLE, respectively. Corresponding resistances were only 3·5 and 5·6% of the total internal resistance to CO2 diffusion. Thus, the decrease in gi at the later stages should be attributed to a decrease in liquid-phase conductance. The increase in mesophyll cell wall thickness would contribute to the decrease in gi through increasing distance in the liquid phase after FLE (Fig. 8c). Judging from the electron micrographs, other factors such as thickness of the cell membrane, cytosol, and chloroplast envelope did not appear to change after FLE. The increase in thickness of chloroplast stroma should not be a major factor leading to low CO2 transfer conductance when activity of carbonic anhydrase in the chloroplast is high (Cowan 1986). Actually, in the case of mature leaves of oak, carbonic anhydrase activity is considerable (Gillon & Yakir 2000). As the properties of cell wall, cell membrane, cytosol and chloroplast envelope can change with leaf age and will affect liquid-phase conductance, further studies are needed to elucidate the responsible factors.
In C. sieboldii leaves after FLE, the increase in Rubisco content was faster than the increase in Sc (compare Figs 4d & 9). Therefore, Rubisco content per chloroplast surface area would increase toward the later stages of leaf development. Such high Rubisco content would lower Cc/Ci in mature leaves because of the increase in carboxylation rate per chloroplast surface area.
The increase in mesophyll cell wall thickness and high Rubisco content per chloroplast surface area would be mainly responsible for the decrease in Cc/Ci after FLE (Fig. 13d). The decrease in Cc was also observed in senescing leaves of wheat by Loreto et al. (1994), and it was suggested that gi might limit photosynthetic rate in the course of leaf senescence.
If a leaf is to acquire its photosynthetic functions quickly, available materials should be preferentially invested in chloroplasts to raise photosynthetic capacity. Furthermore, the rapid chloroplast development leads to high Sc, which is essential to efficient photosynthetic CO2 transfer (Evans et al. 1994; Hanba et al. 1999). However, the leaves of the evergreen broad-leaved tree had low Sc/Smes when they were young (Fig. 7) and mesophyll cell walls thicken after FLE (Fig. 8c), which suggests that a considerable share of the available materials for leaf construction was preferentially invested in cell wall construction (structural compounds) throughout the leaf development. In the C. sieboldii saplings used in this study, the leaf longevity was about two years. Such long-lived leaves would be valuable enough to be protected by thick cell walls at the expense of an efficient CO2 transfer and quick construction of leaf photosynthetic functions.