Introduction

The atmospheric [CO2] rose dramatically in the previous century (from ~280 ppm in the 19th century to ~396 ppm in 2013) and is currently increasing at c. 0.4% year−1, due mainly to the combustion of fossil fuels and deforestation (IPCC 2014; WMO 2014). Atmospheric [CO2] is predicted to rise by a further 50% by 2050 (Meehl et al. 2007). Many plant species increase their photosynthesis and growth under elevated [CO2] when other environmental resources do not limit these changes. Knowledge of long-term enhancing effects of [CO2] on plant growth is still limited today (e.g. Norby and Zak 2011; Koike et al. 2015). There is much evidence of decreased photosynthetic activity under elevated [CO2], and this correlates with less RuBisCo protein and a decline in leaf N concentration (Tissue et al. 1993; Rogers et al. 1996; Zhang et al. 2006; Madhu and Hatfield 2014). The V cmax and/or the J max are reduced in plants undergoing photosynthetic down-regulation at elevated [CO2] (Sage 1994; Onoda et al. 2005; Norby and Zak 2011; Koike et al. 2015). The value of V cmax limits the photosynthetic rate at low [CO2], and J max limits the photosynthetic rate at high [CO2]. Changes in the ratio J max/V cmax affect the photosynthetic rate; a decrease in J max/V cmax slows the photosynthetic rate at elevated [CO2]. The reduction in V cmax is usually ascribed to lower amounts of RuBisCo or low RuBisCo activity (Sage et al. 1989; Jacob et al. 1995; Tissue et al. 1999; Onoda et al. 2005). Negative photosynthetic response often occurs at elevated [CO2], especially because of decline in leaf N content that often occurs under conditions of low soil N availability (Norby et al. 1986; Choi et al. 2005b, 2009; Sanz-Sáez et al. 2010; Watanabe et al. 2010).

In contrast, photosynthesis and growth of some plants in a symbiotic relationship with mycorrhizae do not decrease under elevated [CO2] even where there is low soil N availability. In such situations there may be greater carbon allocation to roots, and consequently increased mycorrhizal symbiosis or activity (Choi et al. 2005a, b, 2009; Smith and Read 2008). In particular, N2-fixing trees respond more strongly to elevated atmospheric [CO2] than other species. These N2-fixing trees are able to increase both C assimilation and plant growth under elevated [CO2], because of their ability to maintain a high internal nutrient supply, furthermore, symbiotic N2 fixation should increase as atmospheric [CO2] increases (Norby et al. 1986; Norby 1987; Saxe et al. 1998; Evans et al. 2000; Schortemeyer et al. 2002; Feng et al. 2004; Nguyen et al. 2006; Sanz-Sáez et al. 2010; Hoosbeek et al. 2011). Various studies suggest that total N2 fixation and total plant N increase in several N2-fixing trees grown at CO2 concentrations greater than the ambient value (Vogel et al. 1997; Atkin et al. 1999; Feng et al. 2004; Sanz-Sáez et al. 2010). Of these, Robinia pseudoacacia L. is the best known (Olesniewicz and Thomas 1999; Schortemeyer et al. 2002; Feng et al. 2004). R. pseudoacacia is native to North America and is now planted worldwide as a pioneer during the early stage of succession because of its adaptability to environmental stresses, high photosynthetic rate and rapid growth (Kurokochi et al. 2010). It is also known to be an invasive tree species and has a high N2-fixing capacity (Barrett et al. 1990; Feldhake 2001; Lee et al. 2004; Rice et al. 2004; Karaki et al. 2012; Watanabe et al. 2014). This species was introduced into Japan in 1873 as a revegetation plant and has been planted in degraded areas and roadsides for greening (Maekawa and Nakagoshi 1997; Kurokochi et al. 2010). In forest ecosystems, R. pseudoacacia can enhance soil N concentration and N cycling by producing N-rich litter (Rice et al. 2004; Malcolm et al. 2008; Lopez et al. 2014). The effect of elevated atmospheric [CO2] on its physiological responses and on N allocation patterns in leaves is not well known, however, and its N2-fixing capacity is not always clear (Koike et al. 2007, 2015).

More than half of the N absorbed in roots goes into photosynthetic proteins in the leaves: into LHC, RuBisCo for CO2 fixation, proteins involved in electron transport, and other proteins (Evans and Terashima 1987; Kitaoka and Koike 2004). RuBisCo is the main enzyme involved in photosynthesis and is the most plentiful protein in the planet, accounting for approximately 50% of the soluble protein in most leaves (Hopkins and Hüner 2009). Although carboxylation reaction of RuBisCo carboxylation reaction, approximately three times faster than the oxygenation reaction in the air, RuBisCo is not saturated at current ambient [CO2] (Buchanan et al. 2015). Photosynthesis under elevated [CO2] is limited usually by the availability of inorganic phosphate (Pi) in leaves or by the electron transport capacity, and RuBisCo is substantially down-regulated (Farquhar et al. 1980; Sharkey 1985; Sage 1990; Buchanan et al. 2015). Plants grown under elevated [CO2] are likely to have greater N-use efficiency and less N allocated to RuBisCo (Makino et al. 2000; Koike et al. 2015). As a result, elevated CO2 has negative effects not only on V cmax and RuBisCo activity but also on the amounts of RuBisCo and soluble protein (Makino et al. 2000; Pérez et al. 2011). The N2-fixing activity of R. pseudoacacia is expected to increase under elevated [CO2], and its N allocation to RuBisCo may increase.

We focus in this study on the relation between physiological activities in R. pseudoacacia, LMA, leaf N content, acetylene reduction activity of Frankia sp., RuBisCo content and N allocation patterns in leaves at ambient and elevated [CO2]. We first investigated the effect of [CO2] during four growing seasons upon leaf biometric changes of R. pseudoacacia in connection with acetylene reduction activity of Frankia sp., the photosynthetic nitrogen use efficiency (PNUE), and water use efficiency (WUE) at each [CO2]. Second, we looked into the photosynthetic and physiological parameters relating to RuBisCo content and leaf N content grown under ambient and elevated [CO2]. Third, we examined the RuBisCo, Chl, leaf N content and the N allocation patterns in the leaves of R. pseudoacacia grown at ambient and elevated [CO2].

Materials and methods

Study site

This study was conducted at the FACE system in Hokkaido University’s Sapporo Experimental Forest, in northern Japan (43°06′N, 141°20′E). The FACE system consists of six 5 m high circular experimental plots, having a diameter of 6 m. This arrangement is based on the system at Davos, Switzerland (Hättenschwiler et al. 2002). Elevated [CO2] in three FACE rings was 500 ± 50 µmol mol−1 for 65% of the time, and 500 ± 100 µmol mol−1 for 88% for the time, which is adequate performance (Takagi et al. 2004; Eguchi et al. 2008a, b). After dawn, CO2 fumigation was stopped when irradiance was less than 60 µmol m−2 s−1, which is below the light compensation point of the study species (Eguchi et al. 2008b). During the growing seasons, from May to late November, CO2 was applied in these three FACE rings. The three control rings were left at ambient [CO2] (ca. 370 µmol mol−1). Long-term (1981–2010) annual precipitation in this area is 1006.5 mm year−1, and the average temperature is 8.9 °C (Japan Meteorological Agency 2016). The soil type at the FACE was fertile brown forest soil, which is native to Sapporo Experimental Forest and is distributed widely throughout Japan.

Plant materials

One-year-old R. pseudoacacia L. (the black locust) was planted in 2003 at ambient [CO2] (about 370 ppm as of 2003) as control and at elevated [CO2] (500 µmol mol−1, as has been estimated for the year 2040 by the IPCC), in the FACE system. Thirty R. pseudoacacia trees were planted randomly at six FACE rings: fifteen trees in elevated rings and fifteen trees in control rings, all approximately 30 cm tall. After four growing seasons under elevated [CO2] or ambient [CO2], we selected twelve R. pseudoacacia trees, six from the three elevated [CO2] rings and six from the three control rings. Sampling and measurements on the trees took place on clear rainless days. Leaves grown at each concentration of CO2 were collected randomly at a height of between 120 and 150 cm for measurement of physiological characters, chemical analysis (N, chl and RuBisCo content) and morphological measurements (e.g. LMA). These leaves were taken from sun-exposed branches of six trees from elevated [CO2] and six from ambient [CO2] in early September.

Photosynthetic rate

Curves of the net photosynthetic rate versus intercellular CO2 concentration (P N/C i) were examined for leaves that were exposed to the sun, using an open gas exchange system (LI-6400, Li-Cor, Lincoln, NE, USA) with a light-emitting diode (LED) light source (6400-02B, Li-Cor) between 09:00 and 13:00 h, when photosynthetic activity and stomatal conductance were stable to changes in environmental conditions. The measurement conditions were: leaf temperature, 25 °C; relative humidity, 70%; photosynthetic photon flux density (PPFD), 1200 μmol m−2 s−1, which is considered to be optimal for RuBisCo activity (Larcher 2003; Choi et al. 2005a, 2006; Kitao et al. 2009). The O2 concentration in the leaf chamber was normal 21% and the water vapour deficit (VPD) in the leaf chamber was about 1.2 kPa.

The intercellular CO2 concentration (C i) response curve of the net photosynthetic rate (P N), known as the P N /C i curve, was measured with a CO2 mixer for the sequence of concentrations 370, 300, 220, 140, 60, 140, 220, 300, 370, 500, 700, 900, 1200 and 1500 μmol mol−1 in the control specimens, and the sequence 500, 370, 220, 140, 60, 220, 300, 370, 500, 700, 900, 1200 and 1500 μmol mol−1 for R. pseudoacacia grown in the FACE; measurements were recorded after equilibration to a steady state had taken place (coefficient of variation <2%).

The P N /C i curve was fitted using the model of Farquhar et al. (1980). From the P N /C i curves, V cmax was calculated by fitting the following equation to the initial slope the P N /C i curves (C i < 300 μmol m−2 s−1) with the least square method with Kaleida Graph™ (Ver 4.11, Synergy Software, PA, USA) (Farquhar et al. 1980; von Caemmerer and Farquhar 1981; Sharkey 1985).

$$P_{\text{c}} = V_{\text{cmax}} \frac{{C_{\text{i}} - \varGamma_{*} }}{{C_{\text{i}} + K_{\text{c}} \left( {1 + \frac{O}{{K_{\text{o}} }}} \right)}} - R_{\text{d}} ,$$
(1)

where P c is the photosynthetic rate limited by RuBisCo activity, C i is intercellular concentration of CO2 (μmol m−2 s−1), Г* is the CO2 compensation point (Pa), R d denotes day respiration (μmol m−2 s−1). O is the intercellular O2 partial pressure and K c and K o are Michaelis–Menten constants of RuBisCo activity for CO2 and O2, respectively.

J max was calculated from Farquhar et al. (1980) and Harley et al. (1992) to a near-plateau of P N /C i curves (C i > 600 μmol m−2 s−1) as follows:

$$P_{\text{j}} = J_{\hbox{max} } \frac{{C_{\text{i}} - \varGamma_{*} }}{{4 \ (C_{\text{i}} + 2\varGamma_{*} )}} - R_{\text{d}} ,$$
(2)

where P j is the electron transport-limited gross carboxylation rate. We applied 40.4 Pa, 24.8 kPa, and 3.69 Pa for K c, K o, and Γ*, respectively (von Caemmerer et al. 1994).

The PNUE was calculated as the maximum mass-based net CO2 assimilation rate (P max) divided by leaf N content per unit leaf area (Narea) (PNUE = P max/Narea; μmol m−2 s−1/g m−2). WUE was taken as the ratio of P N to the transpiration rate at PPFD saturation and constant water VPD (Field et al. 1983; Choi et al. 2014).

After measurement of the photosynthetic rate, six disk samples (each 1 cm2) were taken from the measured leaf for determination of the LMA (=1/SLA; specific leaf area), N content and Chl. The remaining leaves were placed within 60 s into a deep freeze (−85 °C), prior to measurement of the RuBisCo content.

N, Chl and RuBisCo content in leaves

Leaf N content was determined with a N–C analyzer (NC-900, Shimadzu, Kyoto, Japan), and Chl content in the leaves was extracted using dimethyl sulfoxide (DMSO) as according to Barnes et al. (1992) and Shinano et al. (1996). Chl (a and b) content was measured using a double-beam spectrophotometer (UV mini-1240, Shimadzu, Kyoto, Japan), and was calculated as according to Barnes et al. (1992).

$${\text{Chl }}a\;(\mu {\text{g}}\;{\text{mg}}^{ - 1} ) = (14.85 \times A_{665} - 5.14 \times A_{648} ) \times (a/b)$$
(3)
$${\text{Chl}}\;b \, (\mu {\text{g}}\;{\text{mg}}^{ - 1} ) = (25.48 \times A_{648} - 7.36 \times A_{665} ) \times (a/b)$$
(4)

A 665: absorbance at 665 nm, A 648: absorbance at 648 nm, a: DMSO solution (mL), b: dry mass of leaves (mg1).

The RuBisCo content was determined by SDS-PAGE (Makino et al. 1985; Hikosaka 1996; Kitaoka and Koike 2004). Disks (4 cm2) were punched out of the frozen leaves, and were homogenized in 100 mM sodium phosphate buffer (pH 7.5) containing 0.4 M sorbitol, 2 mM MgCl2, 10 mM NaCl, 5 mM iodoacetate acid, 1% (w/v) β-mercaptoethanol and 2% (w/v) polyvinylpyrrolidone, together with acid-washed quartz sand. The homogenate was centrifuged at 10,000g for 1 min at 0–4 °C. The resulting supernatant fluid was treated with SDS solution for 1 min at 100 °C. The gel was stained with Coomassie Brilliant Blue R-250 (CBB), and the RuBisCo large subunit was extracted with formamide, so as to facilitate spectrophotometric determination of RuBisCo. The RuBisCo content was determined spectrophotometrically by formamide extraction of the CBB Blue R-250-stained bands corresponding to the large and small subunits of RuBisCo separated by SDS–PAGE using calibration curves made with RuBisCo purified from R. pseudoacacia leaves. A calibration curve was made with bovine albumin (Sigma-Aldrich, St. Louis, MO).

N allocation patterns in the leaves

Allocation of N to NL (LHC, PSI and PSII, even though the PSI and PSII were found as one unit of megacomplex; Yokono et al. 2015), to NE (electron transport proteins except NL plus the N content of carbon cycle proteins except NR) and to NR (leaf N allocation to RuBisCo), as components of the total N concentration, was calculated using the formula of Kitaoka and Koike (2004):

$${\text{N}}_{\text{L}} = \left[ {\text{Chl}} \right] \times 37.1 \times 14,$$
(5)

where [Chl] is the calculated Chl a + b concentration (mmol m−2), 37.1 is mol N per mol Chl, and 14 is the molecular weight of N (Evans and Seemann 1989).

$${\text{N}}_{\text{E}} = J_{\hbox{max} } /156 \times 9.53 \times {\text{LMA }} \times {\text{Nm,}}$$
(6)

where 156 is the ratio of J max to the cytochrome f content (mmol electrons (mol cyt f)−1 s−1) at 25 °C, 9.53 is N in bioenergetics per unit cytochrome f (mol mmol−1), LMA is the leaf dry mass per unit leaf area (g m−2), and Nm is the N content per unit leaf mass (mg g−1) (Hikosaka and Terashima 1995; Niinemets and Tenhunen 1997).

$${\text{N}}_{\text{R}} = {\text{RuBisCo }}({\text{g}}\;{\text{m}}^{ - 2} ) \times 0.204,$$
(7)

where 0.204 is N content rate in the RuBisCo. NR is calculated by the method of Evans and Seemann (1989).

Acetylene reduction activity

The N fixation rate in R. pseudoacacia was determined by acetylene reduction assay (ARA) (Schöllhorn and Burris 1967; Hardy et al. 1968; Rivera-Ortiz and Burris 1975; Shah et al. 1975; Peterson and Burris 1976). Approximately 2.5 g of nodule cluster with subtending fine roots were harvested from the same trees used for photosynthetic rate measurement, and was placed immediately in the syringe. Then, 10% (v/v) of the air in the syringe was removed and replaced with an equal volume of acetylene (C2H2) (99.9%). The nodules were maintained at ambient soil temperature (as measured by a digital thermometer) by placing the incubation syringe within the forest floor prior to assay. After 1, 5, 10 and 30 min, 10 ml of gas was removed from the syringe and injected into a gas chromatograph to measure the production of ethylene from the acetylene (G-3000, Hitachi, Tokyo, Japan).

Statistical analysis

To determine the effects of the FACE system, an analysis of variance (ANOVA) was carried out by StatView 5.0 (version 5.0, SAS Institute Inc., Cary, NC). Results are stated as mean ± SD.

Results

Photosynthetic responses

The net photosynthetic rate/intercellular CO2 concentration curves at light saturation are shown in Fig. 1, and the value of the photosynthetic parameters of R. pseudoacacia grown at ambient and elevated [CO2] are set out in Table 1. Elevated [CO2] affected the net photosynthetic rate, A growth, A max, V cmax, and J max. The A growth was significantly higher (8%) for plant materials grown at elevated [CO2] relative to controls (P < 0.01; see Fig. 1; Table 1). In contrast, under enhanced [CO2] and light saturation (A max), the photosynthetic rate of R. pseudoacacia grown at ambient [CO2] was about 32 µmol m−2 s−1, and at elevated [CO2] was 24 µmol m−2 s−1 (27% less; P < 0.001). Moreover, V cmax (−15%) and J max (−19%) were significantly lower for materials grown at elevated [CO2] (P < 0.001).

Fig. 1
figure 1

Response of the net photosynthetic rate to intercellular CO2 concentration for R. pseudoacacia grown at ambient (370 μmol m−2 s−1) and elevated (500 μmol m−2 s−1) [CO2]. Values are mean ± SD for individual trees in three rings in each CO2 treatment

Full size image
Table 1 A growth, A max, V cmax and J max of R. pseudoacacia grown at ambient (370 μmol mol−1) and elevated (500 μmol mol−1) [CO2], and relative changes between the treatments
Full size table

RuBisCo content in leaves

Figure 2 shows the relation between the RuBisCo content and the V cmax in leaves of R. pseudoacacia grown at ambient and elevated [CO2]. The RuBisCo content and V cmax in leaves grown at elevated [CO2] was significantly less than ambient [CO2] (P < 0.001). Strong correlation between V cmax and RuBisCo content was observed across CO2 treatments, moreover (R 2 = 0.79).

Fig. 2
figure 2

Relation between the RuBisCo content and the V cmax in leaves for R. pseudoacacia grown at ambient [CO2] (370 μmol mol−1, open squares) and elevated [CO2] (500 μmol mol−1, closed triangles). The significance of the CO2 treatment effect is indicated by the P value. CO2 treatment had significant effect on V cmax (P = 0.0004) and RuBisCo content (P = 0.00009). The difference was tested by ANOVA

Full size image

Leaf physiological variables and leaf structural variables

Table 2 shows the leaf N content, C/N ratio, Chl a + b, Chl a/b, PNUE, WUE and LMA in current year leaves of R. pseudoacacia grown at ambient and elevated [CO2]. The leaf N content of R. pseudoacacia grown at ambient [CO2] was 2.87 g m−2, and at elevated [CO2] it was 2.24 g m−2. The leaf N content at elevated [CO2] was significantly less (by 23%) than at ambient [CO2] (P < 0.05). The C/N ratio and PNUE were sharply lower at elevated [CO2], but the difference was not significant. The Chl a + b in leaves grown at ambient and elevated [CO2] were, respectively, 0.26 and 0.24 g m−2; the difference was not significant. The Chl a/b was slightly higher in leaves grown at elevated [CO2] than at ambient [CO2], but the difference was not significant. The WUE in leaves was almost identical at elevated and ambient [CO2]. We observed large differences in LMA. The LMA in leaves grown at ambient [CO2] was 145.5 g m−2, whereas at elevated [CO2] it was 107.0 g m−2. The LMA at elevated [CO2] was significantly less (by 27%) than at ambient [CO2] (P < 0.01).

Table 2 Leaf N content, C/N ratio, Chl a + b, Chl a/b, PNUE, WUE and LMA of R. pseudoacacia grown at ambient (370 μmol mol−1) and elevated (500 μmol·mol−1) [CO2], and relative changes between the treatments
Full size table

Table 3 shows the RuBisCo/N, Chl/N, V cmax/N, J max/N and J max/V cmax of R. pseudoacacia grown at ambient or elevated [CO2], and relative changes between the treatments. The RuBisCo/N of R. pseudoacacia grown at ambient [CO2] was 1.37/g m−2, and at elevated [CO2] it was 1.02/g m−2. The RuBisCo/N at elevated [CO2] was significantly less (by 26%) than at ambient [CO2] (P < 0.05). The value of V cmax/N and J max/N were sharply higher at elevated [CO2], but the difference was not significant. However, Chl/N was significantly greater (by 22%) at elevated [CO2] than at ambient [CO2] (P < 0.05). The ratio of the capacities of RuBP regeneration to RuBP carboxylation (J max/V cmax) was almost identical at elevated and ambient [CO2].

Table 3 RuBisCo/N, Chl/N, V cmax/N, J max/N and J max/V cmax of R. pseudoacacia grown at ambient (370 μmol mol−1) and elevated (500 μmol mol−1) [CO2], and relative changes between the treatments
Full size table

Acetylene reduction activity

Figure 3 shows the acetylene reduction activity of a N-fixer that is symbiotic with R. pseudoacacia, under ambient and elevated [CO2]. The acetylene reduction activity of Frankia sp. grown at elevated [CO2] was significantly greater than for that grown at ambient [CO2] (P < 0.01). The acetylene reduction activity of Frankia sp. grown at ambient [CO2] was 12.2 µmol C2H4 g−1 nodule h−1, and the acetylene reduction activity of Frankia sp. grown at elevated [CO2] was 15.0 µmol C2H4 g−1 nodule h−1.

Fig. 3
figure 3

Acetylene reduction activity of Frankia sp., symbiotic with R. pseudoacacia, grown at ambient [CO2] (370 μmol mol−1) and elevated [CO2] (500 μmol mol−1). The black bars are means of 24 measurements (open circles) for each [CO2] treatment. The significance of the CO2 treatment effect is indicated by the P value. The difference was tested by ANOVA

Full size image

N allocation patterns in the leaves

Figure 4 shows the N allocation patterns in the leaves of R. pseudoacacia grown at ambient and elevated [CO2]. The N allocation patterns in leaves of R. pseudoacacia grown at elevated [CO2] differed from that at ambient [CO2], especially to RuBisCo and remaining N. For NR there was a slight decrease in R. pseudoacacia grown at elevated [CO2] relative to ambient [CO2], from an allocation of approximately 27.6% N to 20.8%. Remaining N was slightly greater in R. pseudoacacia grown at elevated [CO2] relative to ambient [CO2], from an allocation of approximately 63.8–69.3%. R. pseudoacacia grown at ambient [CO2] allocated 2.7% N to NE and 5.9% N to NL; at elevated [CO2] these figures were 2.7 and 7.2%.

Fig. 4
figure 4

N allocation patterns in the leaves (%) of Robinia pseudoacacia grown at ambient [CO2] (370 μmol mol−1) and elevated [CO2] (500 μmol mol−1). The diagonally hatched box denotes N allocated to RuBisCo (NR); the black box denotes N allocated to electron transport proteins (NE); the dark-gray box denotes N allocated to LHC, PSI and PSII (NL); the white interior box denotes others. Values are mean ± SD for individual six trees in three rings in each CO2 treatment

Full size image

Discussion

During the fourth year of exposure of R. pseudoacacia to elevated [CO2] in the FACE system, we observed changes in the photosynthesis parameters and in N allocation patterns in the leaves, as well as changes to leaf structural and biochemical characteristics and acetylene reduction activity of Frankia sp. symbiotic with R. pseudoacacia. Previous studies determined that long-term exposure to elevated [CO2] in controlled environments results in a decline in photosynthetic capacity, associated with reduced levels of RuBisCo and organic N per unit leaf area (Choi et al. 2005a, b; Koike et al. 2007; Lambers et al. 2008; Choi et al. 2009; Robinson et al. 2012). We also observed a photosynthetic down-regulation of R. pseudoacacia in response to elevated [CO2]. This may be associated with reduction of N and RuBisCo content, and also changed N allocation patterns in the leaves, corresponding to conditions giving rise to down-regulation (Tables 1, 2, 3; Figs. 1, 2, 3, 4).

Photosynthetic characteristics are determined mainly by leaf structure (such as LMA), leaf N content, and the N allocation patterns in the leaves. The LMA underwent various changes with light, atmospheric [CO2], nutrient, temperature and other factors (Poorter et al. 2009). The LMA, which represents the light-intercepting capability of a leaf per unit area, is an index related to many important physiological characteristics (Reich et al. 1999). Changes in leaf morphology and anatomy may occur due to the effects of elevated [CO2] on the leaf carboxylation rate per unit leaf area (Luo et al. 1994; Peterson et al. 1999). In the present study there was a significantly lower LMA and leaf N content of R. Pseudoacacia grown at elevated [CO2] than at ambient [CO2]. There have been similar findings in N-fixing trees and other tree species, in which the LMA decreases in specimens grown at elevated [CO2], and in some herbaceous plants grown in a FACE system (Norby 1996; Peterson et al. 1999; von Caemmerer et al. 2001; Gutiérrez et al. 2009; Liu et al. 2012; Novriyanti et al. 2012; Koike et al. 2015). Other studies found that the LMA increases in specimens grown at elevated [CO2], however, (Staudt et al. 2001; Poorter et al. 2009). The LMA of R. pseudoacacia grown at elevated [CO2], given that there is no restricting root zone in the FACE system and that the plant is undergoing symbiosis with N2 fixative micro-organisms, was significantly lower than at ambient [CO2]. In a review of the FACE system, Koike et al. (2015) stated that N-fixing Alder species, in symbiosis with the N-fixing micro-organism Frankia sp., usually had lower LMA (high SLA; reciprocal of LMA) when grown at elevated than at ambient [CO2], because these N-fixing species are able to act as a sink of photosynthates (Koike 1993; Koike et al. 1997; Tobita et al. 2011; Liu et al. 2012). Fast-growing species mostly have low LMA and lower leaf tissue densities, moreover. This is an important finding, because it was determined in the FACE experiment that this is due mainly to reduced foliage thickness in both leaves and phyllodes of R. pseudoacacia. Also, PNUE and WUE are almost identical at elevated [CO2] and ambient [CO2]. It is likely that the combination of decreasing leaf N content and of LMA with photosynthetic rate led to almost identical in PNUE and WUE at elevated [CO2]. Moreover, we found that the acetylene reduction activity was increased in specimens of R. pseudoacacia grown at elevated [CO2], relative to those grown at ambient [CO2] (Table 2; Fig. 3). Several studies have found that N2-fixing trees (e.g. Alnus glutinosa, A. hirsute, Elaeagnus angustifolia, R. pseudoacacia) grown at elevated [CO2] increase the nitrogenase activity of N2-fixing micro-organisms, or fixed more N, and made their leaf N concentration smaller (Norby 1987; Luo et al. 1994; Koike et al. 1997; Vogel et al. 1997; Olesniewicz and Thomas 1999; Tobita et al. 2005; Millett et al. 2012). Plants grown at elevated [CO2] commonly accumulate carbohydrate in leaves and other tissues as starch, soluble sugars and structural compounds. Decreased leaf N may be due to a combination of dilution from increased carbohydrate and decreased investment in RuBisCo or other enzymes, or the effect of watering (Coleman et al. 1993; Ainsworth and Long 2005).

Also important is that in R. pseudoacacia grown at elevated [CO2], the decreases of J max, V cmax, Chl and RuBisCo content in the leaf corresponded to conditions giving photosynthetic down-regulation found with LMA, and leaf N content decreased (Tables 1, 2; Figs. 1, 2). Values of J max and V cmax were reduced in elevated [CO2], by 19 and 15%, respectively (P < 0.001). Also, A max was 27% less at elevated [CO2] (P < 0.001). Ainsworth and Long (2005) reported in FACE experiments that V cmax was at least 20% less at elevated [CO2], and that leaf N was reallocated from RuBisCo to other functions. These results indicate that the photosynthetic down-regulation, inclusive of V cmax and J max, Chl and RuBisCo content at elevated [CO2], are due to a decrease in leaf N content. RuBisCo is the rate-limiting enzyme in photosynthetic carbon assimilation and photorespiratory oxygen reduction. Photosynthesis at elevated [CO2] is limited by the electron transport capacity (Farquhar et al. 1980) or RuBP regeneration (Onoda et al. 2005). The parameters V cmax and J max are strongly correlated with RuBisCo content, leaf structure (LMA), cytochrome and especially leaf N content (von Caemmerer and Farquhar 1981; Evans 1987; Onoda et al. 2005; Yamori et al. 2010). In the present study, we found a decrease in RuBisCo/N (−26%, P = 0.02) at elevated [CO2] (Table 3). This implies that the reason for the reduction in RuBisCo content is that leaf N was not distributed sufficiently to RuBisCo at elevated [CO2] (Table 3; Figs. 2, 4). Also, V cmax and J max were significantly less at elevated [CO2], with corresponding RuBisCo content. This fact also indicates that the decrease in the J max/V cmax reduces the CO2 stimulation of photosynthesis, and the photosynthetic rate was limited by RuBP regeneration rather than RuBP carboxylation in R. pseudoacacia specimens exposed to elevated [CO2].

N absorbed in roots enters photosynthetic proteins in the leaves, specifically LHC proteins, RuBisCo for CO2 fixation, electron transport systems, and other remaining soluble proteins (Kitaoka and Koike 2004). Elevated [CO2] led to increased leaf carbohydrates and less leaf protein when photosynthesis was down-regulated. Photosynthetic down-regulation can occur through sugar-mediated repression of RuBisCo, and is related to reallocation of N away from the photosynthetic apparatus or an effect on photosynthetic enzymes (Dewar et al. 1998; Stitt and Krapp 1999; Makino et al. 2000). Changes in the N allocation patterns within the photosynthetic apparatus are an important mode of acclimation to changes in light, temperature and atmospheric [CO2] (Katahata et al. 2007; Lambers et al. 2008). Models of optimal photosynthesis suggest that N allocation patterns within leaves reflect a trade-off between CO2 fixation and LHC proteins (Evans 1989; Hikosaka and Terashima 1995). Consistent with this hypothesis, allocation of N to RuBisCo was negatively correlated with elevated [CO2]. The NR at elevated [CO2] was between 27.6 and 20.8% in our study. This is because R. pseudoacacia leaves grown at elevated [CO2] allocated N less rapidly to RuBisCo than to other photosynthetic proteins, such as NE and NL, than did specimens grown at ambient [CO2]. Remaining N increased more rapidly than N in other photosynthetic proteins (such as NE and NL) in leaves grown at elevated [CO2] than in leaves grown at ambient [CO2]. Although PSI and PSII were found as one unit of megacomplex, we empirically use the terms PSI and PSII (Yokono et al. 2015). Relatively less N was allocated to RuBisCo in leaves of R. pseudoacacia grown at elevated [CO2], consistent with J max/V cmax ratios generated from P N/C i curves.

These results suggest that leaf N is reallocated from RuBisCo to remaining N at elevated [CO2]. RuBisCo is a soluble protein, whereas the reaction center and LHC proteins are thylakoid membrane proteins. Soluble proteins are resolved more easily than membrane protein (Makino et al. 1988; Katahata et al. 2007). Remobilizing the N from RuBisCo may, therefore, be more efficient than remobilizing it from light-harvesting proteins. These responses in the R. pseudoacacia leaves are related to decreased RuBisCo and leaf N content, resulting in a low carbon gain at elevated atmospheric [CO2], because the same photosynthetic carbon assimilation rate can be maintained with lower RuBisCo content (Hilbert et al. 1991). The A growth was significantly higher in R. pseudoacacia grown at elevated [CO2] than at ambient [CO2], however. We have suggested that R. pseudoacacia allocates more NL (LHC, PSI and PSII) in leaves to improve light harvesting but less NR in leaves, because sufficient CO2 for carbon fixing is available at elevated [CO2]. Moreover, R. pseudoacacia allocated same NE and more N to remaining N at elevated [CO2]. According to these observations, R. pseudoacacia grown at elevated [CO2] is well prepared for sudden changes in environmental conditions, such as light, temperature or atmospheric [CO2], with N stored as soluble remaining N storage in soluble proteins.

Author contribution statement

DSC, HT and YW conducted experiments, and measured the photosynthetic rates and acetylene reduction activities. TK directed the experiment and managed the experimental facilities. DSC and TS analyzed leaf nitrogen allocation and also the various nutrients. DSC, TK and RDG discussed the results and co-wrote this paper.