Divergent patterns of photosynthetic phosphorus-use efficiency versus nitrogen-use efficiency of tree leaves along nutrient-availability gradients


*Correspondence author. E-mail: amane@ecology.kyoto-u.ac.jp


1.  Nitrogen (N) and phosphorus (P) are essential nutrients for photosynthetic carbon assimilation and most frequently limit primary productivity in terrestrial ecosystems. Efficient use of those nutrients is important for plants growing in nutrient-poor environments.

2.  We investigated the pattern of photosynthetic phosphorus-use efficiency (PPUE) in comparison with photosynthetic nitrogen-use efficiency (PNUE) along gradients of P and N availability across biomes with 340 tree and shrub species. We used both total soil N and P concentration and foliar N/P ratios for indicating nutrient-availability gradients.

3.  Photosynthetic phosphorus-use efficiency increased with greater leaf mass per area (LMA) toward decreasing P availability. By contrast, PNUE decreased with greater LMA towards decreasing N and P availability.

4.  The increase in PPUE with decreasing P availability was caused by the net effects of a relatively greater reduction in foliar P concentration and a relatively constant photosynthetic carbon assimilation rate. The decrease in PNUE with decreasing N availability was caused by the effects of a reduction in photosynthetic carbon assimilation rate with greater LMA.

5.Synthesis. Our results suggest that higher PPUE may be an effective leaf-level adaptation to P-poor soils, especially in tropical tree species. Future research should focus on the difference between PPUE and PNUE in relation to leaf economics, physiology and strategy.


Nitrogen (N) and phosphorus (P) are the nutrients most frequently limiting primary productivity in terrestrial ecosystems (Vitousek & Howarth 1991; Elser et al. 2007). Efficient use of these nutrients is believed to contribute to fitness of plants (Aerts & Chapin 2000) and is thought to influence the nutrient cycling and productivity of terrestrial ecosystems (Vitousek 2004). The rate of net photosynthesis accomplished per unit N in a leaf, termed as instantaneous photosynthetic N-use efficiency (PNUE), has been considered an important plant functional trait to characterize species in relation to their leaf economics, physiology and strategy (see Hikosaka 2004 for review). At the global scale, both foliar N and P concentration as well as leaf lifespan (LLS), leaf mass per area (LMA), dark respiration and photosynthetic capacity are the core leaf economics traits (Wright et al. 2004). Both foliar N and P concentration tightly correlate with each other (Niklas et al. 2005; Kerkhoff et al. 2006). Thus, more complete and global scale knowledge of both PNUE and PPUE (photosynthetic P-use efficiency), which are the ratios of the rate of photosynthetic carbon assimilation to foliar N and P content, respectively, is critical to ecological explanations of plant functional traits and the function of ecosystems.

The physiological relationship between N content and photosynthetic capacity in a leaf, which is expressed as PNUE, is relatively well understood. Numerous studies have established that the maximum photosynthetic carbon assimilation rate (hereafter, Amax) positively correlates with foliar N concentrations (Field & Mooney 1986; Evans 1989; Wright et al. 2004). This is due to the fact that the majority of foliar N is integral to the proteins of the photosynthetic machinery, especially ribulose-1,5-bisphosphate carboxylase oxygenase (Rubisco) (Field & Mooney 1986; Poorter & Evans 1998). Studies within a site and ranging across continents and growth forms have shown that PNUE has a negative correlation with LMA and LLS (Reich et al. 1991, 1992; Hikosaka 2004). Wright et al. (2005) have demonstrated that, at the global scale, PNUE increases towards higher mass-based Amax (Amass) and higher mass-based foliar N concentration (Nmass). Physiologically, a lower PNUE with increasing LMA is caused by decreasing photosynthetic carbon assimilation rates via higher resistance to CO2 diffusion (Parkhurst 1994; Hanba, Miyazawa & Terashima 1999) and smaller N partitioning into Rubisco against larger N partitioning into cell walls (Poorter & Evans 1998; Onoda, Hikosaka & Hirose 2004; Takashima, Hikosaka & Hirose 2004). As plants cannot simultaneously maximize PNUE and leaf toughness (i.e. larger LMA), there is a trade-off between photosynthesis and persistence. In nutrient-poor environments, stress-tolerant species have leaves with lower Amass and larger LMA, thus causing a lower PNUE (Reich et al. 1994; Ellsworth & Reich 1996). Although a lower PNUE seems to contradict conservative nutrient use, the reduction of N-use efficiency with a lower Amass is compensated by longer LLS with larger LMA (Small 1972; Hiremath 2000; Cordell et al. 2001a).

During photosynthetic carbon assimilation, P is required for the production of ATP from ADP, for the production and export of triose-P and for ribulose-1,5-bisphosphate (RuBP) regeneration (Geiger & Servaites 1994). P deficiency in a leaf causes a reduction of Amax through reduced thylakoidal (Conroy et al. 1986; Lauer et al. 1989) and stromal processes (Sivak & Walker 1986; Brooks, Woo & Wong 1988). Therefore, it is physiologically reasonable to regard PPUE as important a leaf trait as PNUE. While a huge number of plant species grow in highly weathered P-deficient environments such as ancient soils and the tropics (Lambers et al. 2008), few studies have compared PPUE along nutrient-availability gradients. Cordell et al. (2001a,b) reported that there was no difference in PPUE of Metrosideros polymorpha (Myrtaceae) between N-limited and P-limited Hawaiian forests. However, Denton et al. (2007) indicated that Banksia species (Proteaceae), which occurred on severely P-impoverished soils in Australia, displayed relatively high PPUE. Lovelock et al. (2007) found that PPUE of mangrove leaves increased with decreasing latitude. Therefore, we hypothesize that with the decline of P availability plants have leaves with higher PPUE and larger LMA, and that PPUE increases with lower Amass and lower mass-based foliar P concentration (Pmass). It must be noted that these hypotheses are quite different from aforementioned patterns of PNUE.

The objective of this paper is to investigate the above hypotheses along P-availability gradients across biomes. We tested the hypotheses by examining the relationships between tree leaf traits and nutrient-availability gradients. We used both total soil N and P concentration and foliar N/P ratios for indicating nutrient-availability gradients. Total soil P concentration is considered a direct index of P availability, but has not been widely analysed, while foliar N/P ratio is an indirect index, but is more widely available. Leaf traits and soil variables were compiled from our own work, published data and unpublished sources. We ask the following research questions. (i) Does the hypothesized negative relationship between PPUE and P availability hold in using both total soil P and foliar N/P ratio? (ii) Does PNUE decrease with both decreasing N and P availability in line with previous studies? (iii) Does PPUE increase towards the lower Amass, lower Pmass and higher LMA, whereas PNUE increases towards the higher Amass, higher Nmass and smaller LMA?

Materials and methods

Measurement of leaf traits

We measured Amax and foliar P and N concentrations of various tree species in tropical montane forests of Mt. Kinabalu, Borneo (4095 m a.s.l., 6°5′ N, 116°33′ E). Amax was measured using a portable open-system infrared gas analyzer (LI6400, Li-Cor, Inc., Lincoln, NE, USA). Measurements were carried out between 8:00 and 12:00 h on five fully expanded healthy sun-exposed leaves at 2–4 m height on each of four to five trees per species. Data were collected with the irradiance, CO2 concentration, chamber temperature and vapour pressure difference between the leaf and air inside the cuvette of the leaf chamber adjusted to c. 1000 μmol m−2 s−1, 400 μmol mol−1, 25 ± 1 °C and 0.5–1.0 kPa, respectively. Leaves, which were measured for CO2-assimilation rates, were collected, wiped, punched to form 10-mm diameter disks for LMA determination, dried at 70–80 °C for 72 h to constant weight and measured for oven-dried mass. The LMA was calculated as oven-dried mass divided by area. Leaf samples were ground after removing their petioles and main veins. For measuring foliar P concentrations, we digested c. 0.20 g of ground material in 5 mL of concentrated H2SO4 and 2 mL of 30% H2O2 at 400 °C for 2–3 h. After cooling to 100 °C, 30% H2O2 was added dropwise until the solution cleared to a pale yellow colour. Digestion was repeated until the solution became clear. The digests were diluted, filtered through Whatman 2 filter paper, and topped up to 50 mL with deionized water. The concentration of P in the digest was determined using an inductively coupled plasma atomic emission spectrometer (ICPS-7510, Shimadzu Co., Kyoto, Japan). The N concentration was measured using a CN analyzer (JM 1000CN, J-Science Lab Co., Kyoto, Japan). PNUE and PPUE were calculated as Amass divided by Nmass or Pmass (μmol CO2 mol nutrient−1 s−1).

N- and P-availability gradients

The most frequently used approach to evaluate N and P availability in soils has been the extraction of soils in the laboratory with solutions designed to measure each of the operationally defined nutrient fractions (e.g. Hedley sequential extractions for P availability, Johnson, Frizano & Vann 2003). However, there is little consistency in evaluating the available forms (particularly labile inorganic P) in forest soils across studies. Therefore, we used total soil N and P concentrations and foliar N/P ratios as proxies for nutrient-availability gradients. The size of labile P pools can be proportional to total soil P (Richardson et al. 2004). Therefore, we used total soil P as a proxy for P availability. By contrast, total soil (mostly organic) N may not correlate with the availability of inorganic N to plants (Vitousek & Howarth 1991; Lambers et al. 2008). We also included total soil N in our analysis for comparison with total soil P. Both N and P measures for the upper 30 cm of soils were used and expressed on a mass basis in this study, because of the lack of information on soil bulk density required to obtain area basis estimates. Foliar N/P ratios have been used to indicate the potential soil P or N limitation to plant growth and terrestrial net primary productivity (Koerselman & Meuleman 1996; Aerts & Chapin 2000). On the other hand, several studies have indicated that high variation in foliar N/P ratios exists among tree species of tropical forests within a study site (Townsend et al. 2007; Hättenschwiler et al. 2008). Therefore, foliar N/P ratios are influenced by site-specific nutrient availability as well as species-specific differences in physiology and life history. However, foliar N/P ratios increase from the poles towards the equator, mostly reflecting a geographical gradient of soil age and weathering intensity (Hedin 2004; McGroddy, Daufresne & Hedin 2004; Reich & Oleksyn 2004). Most recently, it has been demonstrated that foliar N/P ratios increased toward decreasing P availability within a forest and across biomes (Richardson, Allen & Doherty 2008; Ordoñez et al. 2009). Consequently, we use foliar N/P ratios as a surrogate for P limitation over N limitation. Empirically, greater foliar N/P ratios reflect lower availability of P in soils and lower foliar N/P ratios reflect lower availability of N in soils.

Data sets

We compiled a data set of leaf traits and soil nutrients based on our own data, published data and unpublished sources (our own data and a list of references are available in Appendix S1 and S2, respectively). We used the data from published studies only when we were able to calculate the combinations of foliar N/P ratios and both PNUE and PPUE, or when we were able to obtain measures of soil fertility and PNUE or PPUE at the same site. We used data from natural field studies (no plantation forestry land) and also did not use data from secondary forests, because the purpose in this study was to investigate natural vegetation along natural soil gradients. A total of 402 cases, consisting of 340 species, were included in the data set (some species were replicated in multiple sites). Sites spanned from Svalbard (80°N) to New Zealand (43°S) and included boreal, temperate and tropical localities. Species included broad- and needle-leaved trees and shrubs.

Mean values were calculated for each trait for each species at a site if replicated measurements were carried out for the same species. Here, Amax refers to photosynthetic rates measured on young but fully expanded leaves under high light and low water stress conditions and at higher than ambient CO2 concentration. Where traits were reported separately for sun and shade leaves, sun-leaf data were used. Where data were presented separately for recently matured and for old leaves, data for recently matured leaves were used.

Data analyses

When we tested the relationship among leaf traits and nutrient availability, values were base 10 log-transformed in order to attain approximate normality and homogeneity of residuals. When we tested the relationship of PNUE and PPUE versus leaf traits, we used standardized major axis (SMA) slope-fitting techniques. The SMA slopes were tested by (S) MATR version 2.0. (Falster, Warton & Wright 2006). Other statistic analyses were performed using the JMP version 6.0.0 (SAS Institute Inc., Cary, NC, USA).


Relationships between leaf traits and nutrient availability

Photosynthetic phosphorus-use efficiency increased significantly with lower P availability (R2 = 0.15 for total soil P, R2 = 0.53 for foliar N/P ratios, < 0.0001 in both cases) (Fig. 1). In contrast, PNUE showed a peak at the middle values of N/P ratios (i.e. PNUE declined towards both lower N availability and lower P availability) (R2 = 0.07, P < 0.0001). There was no significant relationship between PNUE and total soil N (> 0.1).

Figure 1.

 The relationships between leaf traits [foliar N/P ratio, photosynthetic nitrogen-use efficiency (PNUE) and photosynthetic phosphorus-use efficiency (PPUE)] and nutrient status (total soil P, total soil N and foliar N/P ratio). Summary for regression statistics is shown in Appendix S3.

There was a consistent and significant decline of Nmass and Pmass with decreasing N- and decreasing P-availability gradients with the exception of the relationship between Nmass and total soil N (Fig. 2). Amass significantly and weakly declined towards both lower and greater foliar N/P ratios (R2 = 0.06, < 0.0001), although there was no significant relationship between Amass and total soil N and total soil P (> 0.1 for both cases). LMA significantly increased with decreasing nutrient-availability gradients for all proxies (= 0.003 for total soil N, < 0.0001 for total soil P, < 0.0001 for foliar N/P ratios).

Figure 2.

 The relationships between leaf traits [mass-based maximum photosynthetic carbon assimilation rate (Amass), leaf nitrogen concentration per unit mass (Nmass), leaf phosphorus concentration per unit mass (Pmass) and leaf mass per area (LMA)] and nutrient status (total soil P, total soil N and foliar N/P ratio). Summary for regression statistics is shown in Appendix S3.

Relationships among leaf traits

Amass positively correlated with Nmass (R2 = 0.46, P < 0.0001) and Pmass (R2 = 0.19, < 0.0001) (Fig. 3). This pattern was consistent with other published data (e.g. Wright et al. 2004). The intercept of Amass with Nmass was significantly negative (−9.4 nmol CO2 g−1 s−1; 95% CI, −15.6 –−3.3), while the intercept of Amass with Pmass was significantly positive (47.6 nmol CO2 g−1 s−1; 95% CI, 41.9–53.3) (Fig. 3). PNUE had a significant and strong positive relationship with Amass (R2 = 0.69, < 0.0001) and a significant but weak positive relationship with Nmass (R2 = 0.02, = 0.03) (Table 1). On the other hand, PPUE significantly and positively correlated with Amass (R2 = 0.38, < 0.0001) and negatively correlated with Pmass (R2 = 0.22, < 0.0001). The SMA slope between PNUE and Amass was significantly flatter than 1 (95% CI, 071–0.80), and that between PNUE and Nmass was significantly stepper than 1 (95% CI, 1.22–1.49) (Table 1). In contrast, the SMA slope between PPUE and Amass did not differ from 1 (95% CI, 0.93–1.10) and that between PPUE and Pmass was also almost isometric and negative (95% CI, −1.25 –−1.04).

Figure 3.

 The relationship between mass-based photosynthetic carbon assimilation rate (Amass) and leaf phosphorus concentration per unit mass (Pmass) (upper) and between Amass and leaf nitrogen concentration per unit mass (Nmass) (lower) for tree species of various biomes. Regression relationships and statistics: Amass = 40.4 × Pmass + 47.6, R2 = 0.19, < 0.0001; Amass = 5.95 × Nmass–9.4, R2 = 0.46, < 0.0001.

Table 1.   Standardized major axis slopes (SMA) (95% confidence intervals in parentheses), coefficient of determination (R2) and P values between photosynthetic nutrient-use efficiency [photosynthetic nitrogen-use efficiency (PNUE) and photosynthetic phosphorus-use efficiency (PPUE)] and leaf traits [mass-based maximum photosynthetic carbon assimilation rate (Amass), leaf nitrogen concentration per unit mass (Nmass), and leaf phosphorus concentration per unit mass (Pmass)] of trees across various biomes
Amass0.75 (0.71, 0.80)0.69<0.0011.01 (0.93, 1.10)0.38<0.001
Nmass1.35 (1.22, 1.49)0.020.0061.77 (1.59, 1.96)0.010.03
Pmass0.86 (0.77, 0.95)0.0040.22−1.14 (−1.25, −1.04)0.22<0.001

There was a consistent and negative correlation between LMA and Amass, Nmass, Pmass, PNUE and PPUE (R2 = 0.12–0.56, < 0.0001 for all cases) (Fig. 4). The correlation between PPUE and LMA was relatively weaker than that between PNUE and LMA (R2 = 0.12, 0.28, respectively, < 0.0001 for both cases).

Figure 4.

 The relationships between leaf mass per area (LMA) and mass-based maximum photosynthetic carbon assimilation rate (Amass) (Amass = 3.96–0.998 × LMA, R2 = 0.56, < 0.0001), leaf nitrogen concentration per unit mass (Nmass) (Nmass = 2.20–0.477 ×  LMA, R2 = 0.39, < 0.0001), leaf phosphorus concentration per unit mass (Pmass) (Pmass = 1.16–0.573 ×  LMA, R2 = 0.22, < 0.0001), photosynthetic nitrogen-use efficiency (PNUE) (PNUE = 2.96–0.544 ×  LMA, R2 = 0.28, < 0.0001) and photosynthetic phosphorus-use efficiency (PPUE) (PPUE = 4.41–0.478 × LMA, R2 = 0.12, < 0.0001).


As expected, our analyses clearly show that in trees PPUE increases with decreasing P availability. PPUE distinctly increased with the decline of P availability as indicated by both total soil P and foliar N/P ratios. There may be an autocorrelation problem between PPUE and foliar N/P ratios, because PPUE is a function of foliar P concentrations. However, PPUE distinctly increased also with decreasing total soil P, which was independent of the measurement of PPUE. Total soil P includes both plant-available and recalcitrant forms (Johnson, Frizano & Vann 2003); however, the concentration of available P fractions is generally proportional to total P (Richardson et al. 2004), and total soil P used in our analysis is therefore considered a valid proxy. In spite of the possibility of potential autocorrelation between PPUE and foliar N/P ratios, the foliar N/P ratio is in effect a valid proxy of P availability. This is supported by the pattern that plants retain disproportionately greater Amass at low Pmass than at low Nmass (Fig. 3). A previous study reported that pioneer species had higher PPUE than climax species within a study site (Raaimakers et al. 1995). However, increasing PPUE in our analysis strongly reflected soil P availability, rather than the difference in plant life history, because we compared PPUE among climax species across biomes.

P limitation of photosynthesis can occur under low temperature (Huner et al. 1993) and high CO2 conditions (Niinemets et al. 1999). However, P limitation can be initiated directly by a reduced P supply in laboratory experiments (Rao & Terry 1989; Kirschbaum & Tompkins 1990; Jacob & Lawlor 1991; Pieters, Paul & Lawlor 2001) or in field experiments in the tropics where soil P supply is much impoverished (Cordell et al. 2001a,b). Therefore, the greater magnitude of variation in Pmass in response to soil P availability is a factor to explain the increasing pattern of PPUE. Additionally, greater LMA had no or a weak effect on reducing PPUE (Fig. 4), whereas PNUE strongly decreased with reducing Amass via greater LMA (Table 1 and Fig. 4). There appear to be physiological mechanisms that maintain Amass in spite of increasing LMA on P-poor soils. When we calculate PPUE, we use the total foliar P concentration irrespective of how foliar P is fractionated among structural, nucleic, metabolic and storage P in foliar tissues. Villar et al. (2006) reported that foliar lipid (structural P) concentration of fast-growing plants with small LMA is greater than that of slow-growing plants with large LMA. Several studies have shown that a relatively large proportion of P is stored in vacuoles, and a small proportion is used for photosynthesis when P availability is improved (Bieleski 1973; Turnbull, Warren & Adams 2007). Fertilization experiments on trees growing on P-poor soils increased their photosynthetic carbon assimilation rate as well as the concentration of Pi in leaves (Thomas, Montagu & Conroy 2006; Turnbull, Warren & Adams 2007). If the increased P is mostly stored in vacuoles or largely allocated to lipids, PPUE will decrease with increasing Pmass, resulting in a negative isometric SMA slope between PPUE and Pmass as our results demonstrate. Conversely, higher PPUE on P-poorer soils can be achieved by decreasing Pmass, of which a greater fraction is allocated to metabolic P than to structural and storage P to maintain Amass.

Photosynthetic nitrogen-use efficiency demonstrated a completely different pattern from that of PPUE. PNUE had a peak at the intermediate foliar N/P ratio (from 16 to 17 on a mass basis), where both Amass and Nmass were highest and LMA was lowest (Fig. 2 and Appendix S3). The reduction of PNUE with greater LMA and with decreasing Amass and Nmass was in line with other studies in nutrient-poor environments (Reich et al. 1994; Ellsworth & Reich 1996). Our results reflect that greater LMA at low N and P availability causes the reduction of PNUE, as earlier studies have shown. Leaves of greater LMA have a reduced PNUE because a greater ratio of foliar N is invested in cell walls. Consequently, the allocation of N for photosynthetic enzymes is reduced, resulting in a reduction of Amass (Poorter & Evans 1998; Onoda, Hikosaka & Hirose 2004; Takashima, Hikosaka & Hirose 2004). Greater LMA also causes a reduction in Amass through another mechanism; greater LMA involves a longer path of CO2 diffusion from stomata to mesophyll cells and chloroplasts and hence reduces Amass (Parkhurst 1994; Hanba, Miyazawa & Terashima 1999). In our analysis, PNUE probably decreased due to these physiological and anatomical changes, because the decline of PNUE was mostly affected by the reduction of Amass (Table 1), which strongly and negatively correlated with LMA (Fig. 4).

We suggest that the raised PPUE is an effective adaptation and an important functional trait in trees growing on P-poor soils, especially in tropical tree species. Furthermore, there was not a negative correlation between PPUE and LMA. This may suggest that plants on P-poor soils can maintain Amass while maintaining leaf longevity unlike plants on N-poor soils. Future research should focus on the difference between PPUE and PNUE in relation to leaf economics, physiology and strategy.


We thank Augusto Franco, Dylan Craven, Ian Wright and Peter Reich for kindly allowing us to use leaf trait data. We are also grateful to the following persons: Jamili Nais, Rimi Repin and Maklarin Lakim of the Sabah Parks for assisting every aspect of our research; Yasuto Fujiki and Peter Akau for their assistance in the fieldwork. This research was supported by a grant-in-aid from the Japanese MESSC (18255003) to K.K. and by a fellowship from JSPS for Young Scientists to A.H.