Photosynthesis or persistence: nitrogen allocation in leaves of evergreen and deciduous Quercus species

ABSTRACT

Photosynthetic nitrogen use efficiency (PNUE, photosynthetic capacity per unit leaf nitrogen) is one of the most important factors for the interspecific variation in photosynthetic capacity. PNUE was analysed in two evergreen and two deciduous species of the genus Quercus. PNUE was lower in evergreen than in deciduous species, which was primarily ascribed to a smaller fraction of nitrogen allocated to the photosynthetic apparatus in evergreen species. Leaf nitrogen was further analysed into proteins in the water-soluble, the detergent-soluble, and the detergent-insoluble fractions. It was assumed that the detergent-insoluble protein represented the cell wall proteins. The fraction of nitrogen allocated to the detergent-insoluble protein was greater in evergreen than in deciduous leaves. Thus the smaller allocation of nitrogen to the photosynthetic apparatus in evergreen species was associated with the greater allocation to cell walls. Across species, the fraction of nitrogen in detergent-insoluble proteins was positively correlated with leaf mass per area, whereas that in the photosynthetic proteins was negatively correlated. There may be a trade-off in nitrogen partitioning between components pertaining to productivity (photosynthetic proteins) and those pertaining to persistence (structural proteins). This trade-off may result in the convergence of leaf traits, where species with a longer leaf life-span have a greater leaf mass per area, lower photosynthetic capacity, and lower PNUE regardless of life form, phyllogeny, and biome.

INTRODUCTION

Within a species, leaf nitrogen is a determinant of photosynthetic capacity. There is a strong, positive correlation between photosynthetic capacity and the nitrogen content per unit leaf area. This may reflect the fact that about half of leaf nitrogen is allocated to the photosynthetic apparatus (Evans 1989; Evans & Seemann 1989). However, dependence of photosynthetic capacity on the nitrogen content varies considerably among species (Field & Mooney 1986; Evans 1989). Since the leaf nitrogen content has no clear trend among species (Reich, Walters & Ellsworth 1991; Reich et al. 1999), photosynthetic nitrogen use efficiency (PNUE, photosynthetic capacity per unit nitrogen) may be the most important factor for the interspecific difference in photosynthetic capacity.

Many authors have studied PNUE as an inherent trait of species to be related to their ecological characteristics. A lower PNUE tends to be found in evergreen than in deciduous species (Field & Mooney 1986; Reich et al. 1995), in stress-tolerant species (Chazdon & Field 1987; Poorter, Remkes & Lambers 1990), in late-successional species (Ellsworth & Reich 1996), in species that inhabit higher altitudes (Westbeek et al. 1999; Hikosaka et al. 2002), and in species with a long leaf life-span (Reich et al. 1991).

In studies of physiological mechanisms for the interspecific variation in PNUE, ribulose-1,5-bisphosphate carboxylase (RuBPCase) has been focused upon because it is a key enzyme of photosynthesis (Farquhar, von Caemmerer & Berry 1980) and requires a large amount of nitrogen in leaves (15–30% of leaf nitrogen; Evans 1989; Evans & Seemann 1989). It has been shown that species with a higher PNUE allocate more nitrogen to RuBPCase (Hikosaka et al. 1998; Poorter & Evans 1998; Westbeek et al. 1999; Ripullone et al. 2003). Thus nitrogen allocation to photosynthetic proteins is one of responsible factors for interspecific variation in PNUE.

A question arises, then: why do some species allocate less nitrogen to photosynthetic proteins to have low PNUE? As nitrogen is an element that limits plant growth in many natural and agricultural ecosystems, the ecological significance of low PNUE is a puzzling question. It has been hypothesized that low PNUE species compensate for their low productivity by a long leaf life-span (Small 1972; Berendse & Aerts 1987; Aerts & Chapin 2000). To persist for a long time, leaves may need to be physically tough (Reich et al. 1991; Wright & Cannon 2001). Cell walls play an important role in mechanical toughness of plant tissues. It is known that cell walls accumulate a significant amount of nitrogenous compounds up to 10% of cell wall materials (Lamport & Northcote 1960; Lamport 1965; Reiter 1998). We may hypothesize that species with longer leaf life-span invest more nitrogen in cell walls to increase leaf toughness at the expense of PNUE.

In the present study, we examined nitrogen allocation in evergreen and deciduous species that belong to the genus Quercus. First we studied nitrogen allocation to photosynthetic proteins as has been done in previous studies (e.g. Hikosaka et al. 1998; Poorter & Evans 1998). Next we devised a new approach, in which leaf proteins were divided into three fractions; water-soluble, detergent-soluble, and detergent-insoluble. The water-soluble fraction includes soluble enzymes in stroma and cytosol (Evans & Seemann 1989). The detergent-soluble fraction includes membrane-associated proteins (Evans & Seemann 1989). We assumed the detergent-insoluble fraction to represent cell wall proteins, which are further assumed to contribute to leaf mechanical toughness. We tested the hypothesis that evergreen species have a lower PNUE with a larger content of cell wall proteins than deciduous species.

MATERIALS AND METHODS

The genus Quercus includes both evergreen and deciduous species. They are important tree species in temperate forests in Asia (Miyawaki 1987). We used two evergreen species, Quercus acuta and Q. glauca, and two deciduous species, Q. serrata and Q. crispula. All four are tall trees that can dominate in temperate forests. Deciduous species extend to higher latitudes than evergreen species but they coexist in ecotonal areas between the warm and the cool temperate region in Japan (Kurosawa, Tateishi & Kajita 1995).

Seeds were collected in autumn 2000 at the Botanical Garden of Tohoku University (35°15′ N, 140°51′ E) except for Q. crispula that was collected in a forest in Aomori Prefecture (40°30′ N, 140°56′ E). They were stored in a refrigerator until germination next spring (2001). Plants were grown at the experimental garden of Tohoku University (35°15′ N, 140°51′ E). Each plant was transplanted into a 1 L pot filled with washed river sand. Two growth irradiances and two nutrient availabilities were applied to obtain a variation in the leaf nitrogen content; high- and low-light were 90% (under a transparent plastic sheet) and 30% (plus neutral shading with shade cloth) of full sunlight, respectively, and high- and low-nutrient availability were created by adding 20 mL of the commercial nutrient solution (Hyponex, N : P : K = 5 : 10 : 5; Murakami-bussan, Kamigori, Japan) that contained 35 and 3.5 mm nitrogen, respectively.

Photosynthetic rates were determined for fully expanded young leaves with an open gas exchange system (Li-6400; LiCor, Lincoln, NE, USA). In August, when the photosynthetic capacity was highest, photosynthetic rates were measured at a leaf temperature of 25 °C, photosynthetic photon flux density of 2000 µmol m⁻² s⁻¹, and vapour pressure deficit of less than 1 kPa. First, photosynthetic rates were determined at air CO₂ partial pressure of 36 Pa (regarded as photosynthetic capacity) and then at various air CO₂ partial pressures.

From leaves used for photosynthetic measurements, leaf discs, 1 cm in diameter, were punched out excluding midrib. Three of them were dried at 70 °C for more than 48 h and used for determination of leaf mass per area (LMA) and nitrogen content (NC analyser; Shimadzu, Kyoto, Japan). Other  discs  were  frozen  in  liquid  nitrogen  and  stored  at −80 °C.

Contents of chlorophyll (chl) and RuBPCase were determined from the frozen leaf discs. One leaf disc was powdered in liquid nitrogen in a mortar with a pestle and homogenized in 1 mL of 100 mm Na-phosphate buffer (pH 7.5) with 0.4 m sorbitol, 2 mm MgCl₂, 10 mm NaCl, 5 mm iodo acetate, 1% polyvinylpyrroridone, 5 mm phenylmethyl sulfonyl fluoride, and 5 mm dithiothreitol. The homogenate was filtered with 20 µm mesh. The chl concentration in the filtrate was determined with 80% acetone (Porra, Thompson & Kriedemann 1989). The RuBPCase concentration was determined according to Hikosaka et al. (1998). The filtrate was applied to sodium dodecyl sulfate (SDS; a detergent) polyachrylamide gel electrophoresis. The gel was stained with Coomassie Brilliant Blue R-250. The band of the large subunit of RuBPCase was extracted with formamide for spectrophotometric determination of RuBPCase. Calibration curves were obtained with RuBPCase purified from Spinacia oleracea.

The water-soluble, SDS-soluble, and SDS-insoluble fractions were isolated from another leaf disc. The leaf disc was homogenized in 1 mL of the phosphate buffer as mentioned above. The mortar was washed with 3 mL of the phosphate  buffer,  which  was  added  to  the  homogenate. The  homogenate  was  centrifuged  at  15 000 g  for  30 min and the supernatant was regarded as the water-soluble fraction. The phosphate buffer that contained 3% SDS was added to the pellet and heated at 90 °C for 5 min. The mixture was centrifuged at 4500 g for 10 min. This procedure was repeated four times. The supernatants obtained through this process were collected (SDS-soluble fraction). The final pellet was regarded as the SDS-insoluble fraction. Soluble proteins were precipitated with 10% (water-soluble) or 20% (SDS-soluble) trichloroacetic acid (TCA) and washed with ethanol. After hydrolysis of the precipitated proteins by 0.316 mmol Ba(OH)₂ with 200 µL water in an autoclave (120 °C, 0.12 MPa) for 15 min, protein content in each fraction was determined with the ninnhydrin method (McGrath 1972). Calibration curves were made with bovine serum albumin.

The photosynthetic apparatus was divided into three categories: (1) RuBPCase; (2) bioenergetics (other Calvin cycle enzymes, ATP synthase, and electron carriers); and (3) light-harvesting (photosystem I and II) (Hikosaka & Terashima 1995; Niinemets & Tenhunen 1997). Nitrogen in RuBPCase (N_r) was calculated assuming that nitrogen concentration in RuBPCase is 16% (Hikosaka & Terashima 1995). Nitrogen in bioenergetics (N_b) was estimated from gas exchange characteristics. The maximum rate of electron transport in chloroplasts (J_max) was determined from CO₂ response curve of photosynthesis according to a biochemical model of photosynthesis (Farquhar et al. 1980). We assumed that nitrogen in bioenergetics is proportional to J_max,  where  the  ratio  of J_max  to  the  cytochrome f  content is 156 µmol mol⁻¹ s⁻¹ (Niinemets & Tenhunen 1997) and nitrogen in bioenergetics per unit cytochrome f is 9.53 mol mmol⁻¹ (Hikosaka & Terashima 1995). Nitrogen in light  harvesting  (N_h)  was  calculated  assuming  37.1 mol mol⁻¹ chl (Evans & Seemann 1989). Nitrogen in the water-soluble, SDS-soluble, and SDS-insoluble fractions was estimated assuming 16% nitrogen in proteins.

RESULTS

In each species, photosynthetic capacity (P_max, the light-saturated rate of photosynthesis at 25 °C and 36 Pa CO₂ partial pressure) was strongly correlated with the leaf nitrogen content per unit area (N_L) (Fig. 1a). These correlations were not affected by growth conditions in each species (P > 0.05, Ancova). They were not different within each leaf habit (deciduous or evergreen) but were different between different leaf habits: evergreen species had a significantly lower P_max at a given N_L than deciduous species (Table 1). Photosynthetic nitrogen use efficiency (P_max/N_L) was higher in deciduous (mean and standard deviation were 135 ± 17 µmol mol⁻¹ s⁻¹ for Q. serrata and 134 ± 16 for Q. crispula) than in evergreen species (83 ± 23 for Q. acuta and 85 ± 9 for Q. glauca).

Figure 1. Photosynthetic capacity as a function of the content of leaf nitrogen (a) and ribulose-1,5-bisphosphate carboxylase (RuBPCase) (b) per unit leaf area in Quercus acuta (•), Q. glauca (▪), Q. serrata (○), and Q. crispula (□).

Table 1.  Regression and correlation coefficients

	Slope	Intercept	Correlation coefficient (r)
Difference in regression coefficients among four species was tested with the analysis of covariance (Ancova) according to Sokal & Rohlf (1981). Significance for intercepts was tested only when there was no significant difference for slopes. When there was a significant difference among four species, significance between two species (shown as alphabets) was tested according to Rice (1989) (Sequential Bonferroni test, α < 0.05) except for the relationship between water-soluble nitrogen and leaf nitrogen content (P < 0.05). Significance for intercepts was tested only when there was no significant difference for slopes.
Pmax–NL
Among four species	P < 0.05
Q.acuta	109 a	−2.00 a	0.85
Q.glauca	109 a	−2.35 a	0.79
Q.serrata	173 a	−2.96 b	0.92
Q.crispula	172 a	−3.18 b	0.92
Pmax–RuBPCase content
Among four species	P < 0.01
Q.acuta	4.61 a	1.88 a	0.84
Q.glauca	4.37 a	2.13 a	0.75
Q.serrata	8.35 b	−1.75 b	0.96
Q.crispula	7.61 ab	−0.36 b	0.89
Nr–NL
Among four species	ns	P < 0.001
Q.acuta	0.231	−0.0060 a	0.91
Q.glauca	0.219	−0.0052 a	0.81
Q.serrata	0.234	−0.0015 b	0.95
Q.crispula	0.231	−0.0018 b	0.93
Ne–NL
Among four species	ns	P < 0.001
Q.acuta	0.0905	−0.0043 a	0.90
Q.glauca	0.0819	−0.0036 a	0.95
Q.serrata	0.0908	−0.0023 b	0.89
Q.crispula	0.0912	−0.0022 b	0.96
Nh–NL in high light
Among four species	ns	P < 0.001
Q.acuta	0.110	−0.0021 a	0.88
Q.glauca	0.134	−0.0047 a	0.92
Q.serrata	0.124	−0.0012 b	0.95
Q.crispula	0.099	−0.0009 b	0.92
Nh–NL in low light
Among four species	ns	P < 0.001
Q.acuta	0.126	0.0004 a	0.95
Q.glauca	0.120	0.0011 a	0.76
Q.serrata	0.133	0.0024 b	0.91
Q.crispula	0.141	0.0015 b	0.90
Water soluble N–NL
Among four species	ns	P < 0.01
Q.acuta	0.399	−0.0078 a	0.62
Q.glauca	0.272	0.0035 ab	0.32
Q.serrata	0.319	0.0042 bc	0.84
Q.crispula	0.400	−0.0024 c	0.88
SDS-soluble N–NL
Among four species	ns	ns
Q.acuta	0.142	0.0109	0.40
Q.glauca	0.024	0.0235	0.04
Q.serrata	0.293	0.0010	0.74
Q.crispula	0.230	0.0037	0.69
SDS-insoluble N–NL
Among four species	ns	P < 0.001
Q.acuta	−0.0278	0.0154 a	−0.42
Q.glauca	−0.0555	0.0180 a	−0.49
Q.serrata	−0.0233	0.0078 b	−0.62
Q.crispula	−0.0137	0.0072 b	−0.27
Total protein N–NL
Among four species	ns	ns
Q.acuta	0.513	0.0184	0.62
Q.glauca	0.240	0.0473	0.20
Q.serrata	0.589	0.0141	0.88
Q.crispula	0.616	0.0079	0.85

P_max was strongly correlated with the RuBPCase content in each species (Fig. 1b) as well as with N_L. Evergreen species had a significantly lower P_max especially at higher RuBPCase contents (Table 1). However, the difference between evergreen and deciduous species was smaller in P_max–RuBPCase than in P_max–N_L relationships (Fig. 1), which indicates that allocation of nitrogen to RuBPCase is a primary factor for the difference in the P_max–N_L relationship.

Figure 2 shows nitrogen contents in photosynthetic proteins. In each species, the content of RuBPCase nitrogen (N_r) was positively correlated with N_L irrespective of growth conditions (Ancova, P > 0.05) (Fig. 2a). N_r was significantly higher in deciduous than in evergreen species when compared at a common N_L (Table 1). Similar tendencies were observed for the relationship between the nitrogen content in bioenergetics (N_b) and N_L (Fig. 2b; Table 1). The relationship between nitrogen in light harvesting (N_h) and N_L differed depending on growth irradiance: leaves grown at the low irradiance had a higher content of chl and thus N_h (Fig. 2c & d). When compared at the same growth irradiance, deciduous species had a significantly higher N_h at a given N_L (Table 1). Owing to the higher N_h, the nitrogen content in the photosynthetic apparatus (N_r + N_b + N_h) was slightly higher in leaves grown at the low irradiance (data not shown).

Figure 2. The content of nitrogen allocated to RuBPCase (a), bioenergetics (b), and light-harvesting in high-light grown leaves (c) and in low-light grown leaves (d). Symbols are as in Fig. 1.

When N_b and N_h (high and low growth irradiance separately) were plotted against N_r, there was no significant difference in the regression line among species (Anova, P > 0.05; data not shown), suggesting that four species had a similar nitrogen allocation within the photosynthetic apparatus.

Figure 3 shows allocation of nitrogen to three protein fractions. The content of water-soluble proteins was positively correlated with N_L across different growth conditions (Fig. 3a). Deciduous species had significantly higher contents of water-soluble proteins than evergreen species when compared at the same N_L (Table 1). The content of SDS-soluble proteins were also correlated with N_L (Fig. 3b) but there was no significant difference among the four species (Table 1). The content of SDS-insoluble proteins was relatively stable against N_L in each species (Fig. 3c) with evergreen species having a significantly higher content of SDS-insoluble proteins (Table 1). No significant difference was found in total protein (water-soluble + SDS-soluble + SDS-insoluble) between species (Fig. 3d). The total protein accounted for 71–77% of leaf nitrogen.

Figure 3. The content of protein nitrogen in the water-soluble (a), the SDS-soluble (b), and the SDS-insoluble fraction, and the content of total protein nitrogen (d). Symbols are as in Fig. 1.

Figure 4 shows allocation and use of nitrogen as a function of LMA. Evergreen leaves had a larger LMA and a lower PNUE (P_max/N_L) (Fig. 4a). Across evergreen and deciduous species, the fraction of leaf nitrogen allocated to RuBPCase decreased with increasing LMA (Fig. 4b) while that to SDS-insoluble proteins increased (Fig. 4c).

Figure 4. Photosynthetic nitrogen use efficiency (photosynthetic capacity per unit nitrogen) (a), the fraction of RuBPCase nitrogen (b), and the fraction of the SDS-insoluble protein nitrogen (c) as a function of leaf mass per area. Symbols are as in Fig. 1.

Figure 5 summarizes nitrogen allocation in leaves of each species. Results obtained from leaves grown under high light and high nutrient conditions are shown. ‘Other protein nitrogen’ was calculated as nitrogen in the water- and SDS-soluble protein minus nitrogen in the photosynthetic apparatus. ‘Other nitrogen’ was calculated as the residual. Nitrogen allocated to the photosynthetic apparatus was smaller in evergreen species (about 30%) than in deciduous species (40%) and the difference was partly counterbalanced by greater allocation of nitrogen to SDS-insoluble proteins.

Figure 5. Nitrogen allocation in leaves grown at high light and high nitrogen availability. Mean of three to four leaves.

DISCUSSION

Partitioning of leaf nitrogen

PNUE was lower in evergreen than in deciduous species (Figs 1a & 4a), which is in accord with earlier studies (Field & Mooney 1986; Reich et al. 1995; Ripullone et al. 2003). Lower PNUE in evergreen species was associated with a smaller allocation of nitrogen to the photosynthetic apparatus (Fig. 2). Similar trends have been observed across species with different PNUE (Hikosaka et al. 1998; Poorter & Evans 1998; Westbeek et al. 1999; Ripullone et al. 2003). Thus the present result confirms the generality of the contribution of nitrogen allocation to the interspecific difference in PNUE.

We divided leaf proteins into three fractions (Fig. 3). The water-soluble fraction has been termed as ‘soluble protein’ in previous studies, of which about one-half was represented by RuBPCase (Terashima & Evans 1988; Hikosaka & Terashima 1996). This was also the case in the present study (Figs 2a & 3a). The amount of the water-soluble protein was greater in deciduous species (Fig. 3a; Table 1). The SDS-soluble fraction was presumed to mostly consist of membrane-associated proteins such as thylakoid components (Evans & Seemann 1989). This fraction was not significantly different between deciduous and evergreen species (Fig. 3b). Then the smaller content of photosynthetic proteins in evergreen species (Fig. 2) suggests that evergreens invest more nitrogen in the SDS-soluble proteins that are not involved in photosynthesis.

A remarkable difference was found in SDS-insoluble proteins: evergreen species allocated two-fold more nitrogen to SDS-insoluble proteins than deciduous species (Fig. 3c). It is surprising that some leaves of evergreen species invest nitrogen into the SDS-insoluble proteins in an amount that is comparable with that invested into RuBPCase (Fig. 4). Since water and SDS remove soluble and membrane-associated proteins, we assumed the SDS-insoluble fraction to represent cell wall proteins that are tightly bound to cell walls (Reiter 1998). The amount of cell wall proteins presented here would be minimal because the SDS treatment might remove proteins that were weakly associated with cell walls. These results suggest that leaves invest a considerable amount of nitrogen in cell walls and that evergreen species allocate more nitrogen to cell walls than deciduous species, which necessarily decreases their PNUE (Fig. 4).

In the present study, 71–77% of leaf nitrogen was found to be allocated to proteins (Fig. 3d). This value seems slightly lower than that reported for several Alaskan tundra species (75–80% for Eriophorum vafinatum, Chapin, Shaver & Kedrowski 1986; 75–88% for several species with different life forms, Chapin 1989). Where is nitrogen allocated to other than proteins? Several authors suggested that a significant amount of nitrogen is invested in nucleic acids (8.5%, Evans & Seemann 1989; 15%, Chapin et al. 1986; 5–10%, Chapin 1989). Chapin et al. (1986) showed that several percent of N was invested in free amino acids and lipids. Thus these compounds may explain about 20% of leaf nitrogen. Furthermore, some species are known to accumulate secondary compounds such as alkaloids and cyanogenic glycosides (Burns, Gleadow & Woodrow 2002). However, the accumulation is species-dependent and there seems no report for the species that we studied. It should be noted that the protein content might have been underestimated in the present study. Proteins that are soluble to TCA and ethanol were not determined. It is known that cell wall proteins are difficult to extract completely: 10–20% of cell wall proteins have been suggested to be hardly solubilized by KOH at 25 °C (Lamport 1965). Although hydrolysis by Ba(OH)₂, which we used in the present study, is considered to be more effective for extraction than KOH solution, it is probable that some of the tightly bound proteins was not extracted. Furthermore, the ninhydrin method might underestimate proline and hydroxyproline (Yemm & Cocking 1955), which are rich in cell wall proteins (Lamport 1965; Showalter 1993). These proteins, however, would explain a few percent of leaf nitrogen.

Ecological significance

Nitrogen allocation to the SDS-insoluble fraction was strongly correlated with LMA irrespective of species and growth conditions (Fig. 4c). This is in accord with our recent study showing a positive correlation between the contents of cell wall proteins and LMA across leaves of Polygonium cuspidatum plants germinated at different times (Onoda, Hikosaka & Hirose 2004). Terashima et al. (1995) observed that leaf thickness was similar in evergreen and deciduous Quercus species, while the cell wall thickness of mesophyll cells was greater in an evergreen than in a deciduous Quercus species. Thus the higher LMA in evergreen species is attributable to a higher leaf density resulting from a greater cell wall thickness. Variation in cell wall thickness might have produced the strong correlation between LMA and nitrogen allocation to SDS-insoluble proteins.

Why did evergreen species have a greater amount of cell wall proteins? It has been suggested that the greater LMA contributes to mechanical toughness of leaves (Reich et al. 1991; Wright & Cannon 2001). Mechanical protection is important for maintaining leaves for a long period (Coley, Bryant & Chapin 1985; Reich et al. 1991). For example, Coley (1983) showed that leaf toughness was the trait that was most highly correlated with levels of herbivory across 46 species in a tropical forest. Evergreen Quercus species must make their leaves tough enough to maintain leaves for longer than 2 years (Hikosaka 2004). On the other hand, leaves of deciduous species have a shorter leaf life-span (5–6 months at the site of seed collection). They can allocate less nitrogen for leaf toughness and more for photosynthesis to have a high PNUE (Fig. 4).

The present results may explain a convergence of leaf traits across species. Regardless of life form, phyllogeny, and biome, species with a shorter leaf life-span have a smaller LMA, higher nitrogen concentration per unit leaf mass, higher P_max, and higher PNUE (Reich et al. 1991, 1999; Reich, Walters & Ellsworth 1997; Wright, Reich & Westoby 2001; Wright et al. 2004). Species with a smaller LMA invest more nitrogen in the photosynthetic apparatus and thus have higher growth rates (Poorter et al. 1990), leading to a competitive success at high nutrient availabilities. Species with a greater LMA, on the other hand, invest more nitrogen in cell walls, leading to reduction in the amount of photosynthetic proteins and thus to slow growth rates. However, they have a longer leaf life-span and mean residence time of nitrogen in leaves, which may be advantageous at low nutrient availabilities (Aerts & van der Peijl 1993; Aerts & Chapin 2000). The convergence of leaf traits across species may result from a trade-off in nitrogen partitioning within the leaf: one for photosynthetic capacity and the other for persistence of leaves.

ACKNOWLEDGMENTS

We thank K Sato for kind help in growing plants and Y Onoda and Y Yasumura for comments. This study was supported in part by grants from Japan Ministry of Education, Science, Sports, and Culture.