Down-Regulation of Photosynthesis to Elevated CO2 and N Fertilization in Understory Fraxinus rhynchophylla Seedlings

(1) Background: Down-regulation of photosynthesis has been commonly reported in elevated CO2 (eCO2) experiments and is accompanied by a reduction of leaf nitrogen (N) concentration. Decreased N concentrations in plant tissues under eCO2 can be attributed to an increase in nonstructural carbohydrate (NSC) and are possibly related to N availability. (2) Methods: To examine whether the reduction of leaf N concentration under eCO2 is related to N availability, we investigated understory Fraxinus rhynchophylla seedlings grown under three different CO2 conditions (ambient, 400 ppm [aCO2]; ambient × 1.4, 560 ppm [eCO21.4]; and ambient × 1.8, 720 ppm [eCO21.8]) and three different N concentrations for 2 years. (3) Results: Leaf and stem biomass did not change under eCO2 conditions, whereas leaf production and stem and branch biomass were increased by N fertilization. Unlike biomass, the light-saturated photosynthetic rate and photosynthetic N-use efficiency (PNUE) increased under eCO2 conditions. However, leaf N, Rubisco, and chlorophyll decreased under eCO2 conditions in both N-fertilized and unfertilized treatments. Contrary to the previous studies, leaf NSC decreased under eCO2 conditions. Unlike leaf N concentration, N concentration of the stem under eCO2 conditions was higher than that under ambient CO2 (4). Conclusions: Leaf N concentration was not reduced by NSC under eCO2 conditions in the understory, and unlike other organs, leaf N concentration might be reduced due to increased PNUE.


Introduction
The primary effect of rising atmospheric carbon dioxide (CO 2 ) concentration on plants is an increase in the photosynthetic rate. C 3 plants show significantly stimulated carbon assimilation and enhanced growth under elevated CO 2 (eCO 2 ) conditions [1][2][3]; however, many studies claim that plants cannot sustain the stimulation of biomass accumulation, which would result under long-term CO 2 fumigation [2,4]. The uncertainties in the sustainability of enhanced productivity and carbon (C) storage under elevated CO 2 (eCO 2 ) conditions are because of different nutrient conditions [5,6]. For example, productivity enhancement decreased with time as nitrogen (N) availability decreased at the Oak Ridge FACE site [4]. A suboptimal N supply can lead to the down-regulation of photosynthesis [7]. Down-regulation of photosynthesis is commonly reported in eCO 2 experiments. This phenomenon involves a reduction in the maximum carboxylation rate (V Cmax ) and maximum electron transport rate driving ribulose biphosphate (RuBP) regeneration (J max ). It is associated with reduced leaf N concentration [3,8] because a large fraction of leaf N concentration is invested in the photosynthetic apparatus [9]. Many researchers have suggested that accumulated nonstructural carbohydrates (NSC) dilute leaf N concentration to balance sink-source organs [10]. Reduced leaf N concentration can be attributed to dilution by additional photosynthates, such as NSC, resulting from a "sink-source imbalance," i.e., an incapacity to form a sufficient "sink" [10,11]. "Sink organs" are plant parts that use photosynthates in storage, construction, and respiration. Plants with large sink organs exhibit little down-regulation despite growing under eCO 2 conditions because the NSC can be translocated to the sink organs. In a study on poplar, free-air CO 2 enrichment (POP-FACE), down-regulation of V Cmax occurred only in Populus alba, which had the smallest sink (diameter growth) among the three poplar clones [12].
Down-regulation of photosynthesis occurs because accumulated NSC suppress the gene expression of Rubisco [10,13], and whereas NSC accumulation is accelerated under eCO 2 conditions and low N availability [14]. Previous studies have suggested that photosynthesis down-regulation is mitigated under eCO 2 conditions at high N concentration [3,15]. Rubisco suppression under eCO 2 conditions is associated with leaf N reallocation from a non-limiting source to other limiting photosynthetic components, such as the light-harvesting complex [9,16]. This change in N allocation can provide an opportunity to increase photosynthetic N use efficiency (PNUE), which is defined as the net C assimilation rate per unit leaf N [8] and is associated with the enhanced efficiency of Rubisco due to photorespiration suppression under eCO 2 conditions [17,18]. However, to the best of our knowledge, there has been no investigation of changes in leaf N allocation in woody plants under different N availability conditions under eCO 2 conditions. In contrast, N reduction also occurs in other sink organs apart from leaves [19][20][21]. For example, the N concentration in the edible parts of major food crops decreased by 9%-15% [20], whereas that aboveground in wheat decreased by 23% [21]. N reduction in sink organs can be regarded as a dilution of NSC accumulation [19]. In a previous meta-analysis, the average concentration of soluble sugars and starch in aboveground woody parts significantly increased by 38.6% and 9.9%, respectively. In roots, starch (9.8%) concentration significantly increased under elevated CO 2 conditions, whereas soluble sugar concentration decreased by 6.7% but not significantly [22]. The effect of eCO 2 on leaf N dilution can be reduced by N fertilization [20,23]; thus, it can be speculated whether NSC changes can be attributed to N concentration variations in other sink organs under different CO 2 and N conditions.
Most previous studies have focused on the effects of elevated CO 2 on dominant overstory trees (Souza et al., 2010), whereas few studies have investigated the effects of interactions of CO 2 and nutrients on understory tree physiological characteristics [24][25][26][27], although the understory contribution to the forest carbon balance is substantial [28,29]. To investigate how the combination of N fertilization and eCO 2 administered to seedlings in the understory affects the N concentrations of sink and source organs and the downregulation of photosynthesis, we exposed understory Fraxinus rhynchophylla seedlings to three different CO 2

Study Site
The study was conducted using an OTC at the National Institute of Forest Science (37 • 15 04 N, 136 • 57 59 E) in South Korea. The mean annual temperature is 13.0 • C, and the annual precipitation is 1271.2 mm (19 May 2020, https://data.kma.go.kr/). The OTC experiment utilized three decagonal chambers (10 m in diameter and 7 m in height) with different atmospheric CO 2 concentrations: ambient, 400 ppm (aCO 2 ); ambient × 1.4, 560 ppm (eCO 2 1.4); and ambient × 1.8, 720 ppm (one OTC per treatment). Two eCO 2 concentrations have been predicted for 2050 and 2070 by the IPCC [30]. A 0.15-mm-thick polyolefin film with a light transmittance of approximately 88% and chemical and water resistance formed the outer covering material. The CO 2 exposure device consisted of 16 cylindrical discharge polyvinyl chloride pipes in each chamber (height: 1.5 m, diameter: 10 cm), vertically connected to the pipe, thus exposing the plants to CO 2 . The CO 2 concentrations in the chambers were measured using an infrared gas analyzer (ZRH type, Fuji Electric System Co. Ltd., Tokyo, Japan), and the CO 2 concentrations inside were controlled by mixing pure liquefied CO 2 from 8:00 a.m. to 6:00 p.m. [31]. The overstory canopy comprised six temperate species. Three seedlings of 4-year-old Pinus densiflora Siebold & Zucc and 2-year-old Fraxinus rhynchophylla HANCE, Quercus acutissima, Acer pseudosieboldianum (Pax) Kom., Crataegus pinnatifida for. pinnatifida, and Sorbus alnifolia Siebold & Zucc Koch were planted in September 2009. The density within the OTC was 2547 trees ha −1 . The leaf area index was 0.28 ± 0.06, 0.34 ± 0.07, and 0.46 ± 0.07 m 2 m −2 under aCO 2 , eCO 2 1.4, and eCO 2 1.8, respectively, using leaf litter collection methods [32] in 2017 (p = 0.166).
Seedlings of 2-year-old F. rhynchophylla were transplanted into 10 L pots filled with a 1:1:1:1 mixture of cocopeat: peat moss: vermiculite: pearlite and exposed to three different CO 2 treatments using the OTCs in March 2019. Plants were fertilized monthly from May to July for 2 years with NH 4 NO 3, as follows: unfertilized treatment (N1), 5.6 gN m −2 yr −1 (N2), or 11.2 gN m −2 yr −1 (N3) [33]. There were five replicate seedlings in each N fertilization system and a total of 15 replicate seedlings in each CO 2 treatment (5 replications × 3 N concentrations). Furthermore, F. rhynchophylla is a deciduous temperate species; the average height of the seedlings was 59.64 ± 0.76 cm, and the diameter was 9.23 ± 0.29 cm in 2019.
Three leaves were collected (three leaves × five replicates × three N treatments in each treatment), including one for leaf gas exchange and two leaves nearby in July 2019 and July and September 2020. To measure NSC, the N per unit area (N area , g m −2 ), N per unit mass (N mass, g m −2 ), Rubisco (g m −2 ), and chlorophyll (g m −2 ), leaf discs with 1-cm diameter were punched out, excluding the midrib, and stored in a liquid nitrogen tank (−196 • C). The dried samples at 70 • C were powdered using a homogenizer (FastPrep-24, MP Biomedicals, Solon, OH), and leaf N concentration was measured using a CHNS-Analyzer Flash EA 1112 (Thermo Electron Corporation, Massachusetts, USA) at NICEM, Seoul National University. PNUE (µmol N -1 s −1 ) was calculated as the ratio of A max to N area .
The seedlings were harvested and sectioned into leaves, stems, branches, and roots in September 2020 to evaluate biomass allocation to determine their dry weights, and the root, stem, and stem and root N mass were estimated using the same methods as those for leaf N concentrations.

Total Nonstructural Carbohydrates
Three discs of leaves and 15-20 mg of root and stem were ground, and the ground samples were used to measure the NSC (sum of the soluble sugars and starch). The soluble sugars were extracted with 1.5 mL of 80% (v/v) ethanol in a water bath at 80 • C for 30 min. After 10 min of centrifugation at 14,000× g, the soluble sugar concentration was determined colorimetrically at 490 nm using the phenol-sulfuric method [37]. Starch was extracted from the remaining pellets after measuring the sugar concentrations. The pellet was incubated in 2.5 mL sodium acetate buffer (0.2 M) in a 100 • C water bath for 1 h. After cooling to room temperature, 2 mL of the buffer and 1 mL of amyloglucosidase (0.5% by weight, Sigma A9229-1G) were added. The mixed samples were incubated overnight in a water bath at 55 • C. After 10 min of centrifugation at 14,000× g, the starch concentration was determined colorimetrically at 490 nm using the phenol-sulfuric method.
Chlorophyll was extracted using the dimethyl sulfoxide (DMSO) method [39]. The two discs were incubated in a brown bottle containing 5 mL of DMSO at 65 • C for 6 h in a water bath. The samples were estimated at two wavelengths, 649 and 665 nm, using a spectrophotometer (Optizen 2120 UV, Mecasys, Korea). The chlorophyll N content was derived using the DMSO equation [40] as follows: Total chlorophyll content (µg mL −1 ) = 21.44 A 649 + 5.97 A 665 (1)

Statistical Analysis
A three-way repeated-measures ANOVA was performed using the R program (ver. 3.3.2; R Core Team, 2016) for A max , V Cmax , J max , leaf N mass , leaf N area , Rubisco, PNUE, NSC, chlorophyll, and chlorophyll:Rubisco. Three-way repeated measures of ANOVA were performed with the fixed factors "CO 2 treatment" and "nitrogen" and the random factor "period." Two-way ANOVA was conducted on N mass , NSC and biomass of stem and root, and total leaf production to compare the CO 2 and N effects within a period. In repeated-measures ANOVA and two-way ANOVA, when the main factors of CO 2 and N and interaction with CO 2 treatment and N treatment were significant, Tukey's HSD test was performed to compare the CO 2 and N effects.

Whole Plant Biomass Allocation
There was a significant increase in the aboveground biomass with increased N fertilization, but no effect of CO 2 treatment was observed ( Figure 1). Leaf production under N3 condition (8.34 ± 0.83 g) was 43.3% and 61.7% higher than that under N1 (5.82 ± 0.62 g, p = 0.060) and N2 (5.16 ± 0.72 g, p = 0.018) conditions, respectively ( Figure 1a). Stem production also increased with N fertilization, with that under N3 condition being 50. 2% higher than that under N1 condition (24.00 ± 3.12 g, p = 0.011, Figure 1b). Unlike the aboveground biomass, roots showed no differences with the N treatments ( Figure 1c). measures ANOVA and two-way ANOVA, when the main factors of CO2 and N and in teraction with CO2 treatment and N treatment were significant, Tukey's HSD test wa performed to compare the CO2 and N effects.

Whole Plant Biomass Allocation
There was a significant increase in the aboveground biomass with increased N fert lization, but no effect of CO2 treatment was observed ( Figure 1). Leaf production unde N3 condition (8.34 ± 0.83 g) was 43.3% and 61.7% higher than that under N1 (5.82 ± 0.62 g p = 0.060) and N2 (5.16 ± 0.72 g, p = 0.018) conditions, respectively ( Figure 1a). Stem pro duction also increased with N fertilization, with that under N3 condition being 50. 2% higher than that under N1 condition (24.00 ± 3.12 g, p = 0.011, Figure 1b). Unlike the above ground biomass, roots showed no differences with the N treatments ( Figure 1c). . Results are presented as mean + SE and summary of a two-way ANOVA. In summary of a two-way ANOVA, CO2 represents CO2 treatment and N represents nitrogen treatment. The X axis represents N fertilization and CO2 treatments: aCO2 (white bar), eCO21.4 (grey bar) and eCO21.8 (black bar). The significance of the primary effects and their interactions are shown with the abbreviations. n.s., non-significant; *, p < 0.05 (two-way ANOVA, CO2 × N). There were differences between N treatments in dry weight of leaf and stem. Different capital letters indicate significant differences among N treatment and lowercase letters indicate significant differences among CO2 treatments (Tukey's test, p < 0.05).

Photosynthetic Characteristics and Nonstructural Carbohydrates
Elevated CO2 exposure increased the light-saturated photosynthetic rate (Amax), an there was no N fertilization effect or interaction effect between N and CO2 treatments (Ta ble 1). The average Amax was 20.1% and 31.0% higher under eCO21.4 (p = 0.036) and eCO21. (p < 0.001) conditions than under aCO2 condition, respectively, whereas Amax showed a increasing tendency under eCO2 conditions compared with that under aCO2 condition i all N treatments, although the difference was not significant ( Table 2). Unlike Amax, down regulation of other photosynthetic characteristics occurred, including Jmax, leaf N concen tration, Rubisco content, and chlorophyll content ( Table 1). The average Jmax unde eCO21.8 was 20.9% and 18.0% lower than that under eCO21.4 condition (p = 0.002) an under aCO2 condition (p = 0.008), respectively, and Jmax under eCO21.8 condition decrease by 32.4% compared with that under aCO2 condition in N3 ( Table 2, p = 0.012). Similarly the average VCmax under eCO21.8 showed the lowest value and was lower than that unde aCO2 in all N treatments, but there were only marginal differences in VCmax among CO treatments (p = 0.083, Table 2). Leaf Nmass decreased under eCO2 conditions and increase with N fertilization, but there was no interaction between N and CO2 treatments (Tabl 1). The average leaf Nmass decreased by 9.4% and 15.3% under eCO21.4 (p = 0.035) an . Results are presented as mean + SE and summary of a two-way ANOVA. In summary of a two-way ANOVA, CO 2 represents CO 2 treatment and N represents nitrogen treatment. The X axis represents N fertilization and CO 2 treatments: aCO 2 (white bar), eCO 2 1.4 (grey bar) and eCO 2 1.8 (black bar). The significance of the primary effects and their interactions are shown with the abbreviations. n.s., non-significant; *, p < 0.05 (two-way ANOVA, CO 2 × N). There were differences between N treatments in dry weight of leaf and stem. Different capital letters indicate significant differences among N treatment and lowercase letters indicate significant differences among CO 2 treatments (Tukey's test, p < 0.05).

Photosynthetic Characteristics and Nonstructural Carbohydrates
Elevated CO 2 exposure increased the light-saturated photosynthetic rate (A max ), and there was no N fertilization effect or interaction effect between N and CO 2 treatments ( Table 1). The average A max was 20.1% and 31.0% higher under eCO 2 1.4 (p = 0.036) and eCO 2 1.8 (p < 0.001) conditions than under aCO 2 condition, respectively, whereas A max showed an increasing tendency under eCO 2 conditions compared with that under aCO 2 condition in all N treatments, although the difference was not significant ( Table 2). Unlike A max , down-regulation of other photosynthetic characteristics occurred, including J max , leaf N concentration, Rubisco content, and chlorophyll content ( Table 1). The average J max under eCO 2 1.8 was 20.9% and 18.0% lower than that under eCO 2 1.4 condition (p = 0.002) and under aCO 2 condition (p = 0.008), respectively, and J max under eCO 2 1.8 condition decreased by 32.4% compared with that under aCO 2 condition in N3 ( Table 2, p = 0.012). Similarly, the average V Cmax under eCO 2 1.8 showed the lowest value and was lower than that under aCO 2 in all N treatments, but there were only marginal differences in V Cmax among CO 2 treatments (p = 0.083, Table 2). Leaf N mass decreased under eCO 2 conditions and increased with N fertilization, but there was no interaction between N and CO 2 treatments ( Table 1). The average leaf N mass decreased by 9.4% and 15.3% under eCO 2 1.4 (p = 0.035) and eCO 2 1.8 (p < 0.001) conditions compared with that under aCO 2 condition, respectively (Table 2). In addition, N mass showed an interaction effect between CO 2 and time period (Table 1); thus, the differences in leaf N mass decreased in September 2020 (Figure 2a). The reduction in leaf N mass under eCO 2 1.4 decreased from 21.0% (aCO 2 : 2.14 ± 0.08% vs. eCO 2 1.4: 1.69 ± 0.09%) in 2019 to no significant decrease in July and September 2020; similarly, the reduction in leaf N mass under eCO 2 1.8 compared with that under aCO 2 was significant in July, 2020 (aCO 2 : 1.71 ± 0.06% vs. eCO 2 1.8: 1.27 ± 0.08%, −25.5%) but became non-significant in September 2020. For N treatment, N mass showed incremental changes under both N2 and N3 treatments (N2: 1.69 ± 0.06%, +17.6%, N3: 1.68 ± 0.06%, +17.1%) compared with those under N1 (1.44 ± 0.07%, maximum P = 0.002). Similar to N mass , N area also decreased under eCO 2 conditions ( Table 1). The average N area was 15.1% and 24.0% lower under eCO 2 1.4 (p = 0.001) and eCO 2 1.8 (p < 0.001) conditions than under aCO 2 condition, respectively (Table 2). Unlike N mass , N area did not show an N-fertilization effect. In addition, there was no interaction between N and CO 2 interaction ( ambient × 1.8: eCO21.8) and three different N treatments (N1: unfertilized, N2: N1 + 5.6 gN m −2 yr −1 and N3: N1 + 11.2 gN m −2 yr −1 ).  To examine the dilution hypothesis and determine whether N fertilization could mitigate the down-regulation of photosynthesis, NSC were quantified in different CO2 and N treatments. NSC differed depending on the CO2 treatments and time periods (Table 1). In contrast to previous results of NSC under eCO2 conditions, the average NSC was 44.5% lower under eCO21.4 (p < 0.001) than under aCO2, and the average NSC under eCO21.8 tended to be lower than that under aCO2 but not significantly (p = 0.061). Moreover, the reduction in NSC under eCO2 conditions was noticeably greater in N3 than in the other N treatments ( Table 2). The NSC in September (28.21 ± 1.51 g m −2 ) were higher than those in July (2019: 21.58 ± 0.94 g m −2 , 2020: 19.78 ± 1.02 g m −2 , maximum p < 0.001). In addition, NSC showed a correlation between CO2 concentrations and time period (Table 1). There was no reduction in NSC under eCO2 in July 2019 and 2020, but NSC were reduced by 36.5% and 24.1% under eCO21.4 (p < 0.001) and eCO21.8 (p < 0.001) relative to those under aCO2 (35.40 ± 3.19 g m −2 ) in September 2020 (Figure 2c).  Results are presented as mean + SE in each CO 2 treatment: aCO 2 (white circle), eCO 2 1.4 (grey circle) and eCO 2 1.8 (black circle). The X axis represents period (19-July: July 2019, 20-July: July 2020 and 20-Sep: September 2020). There were interactions between CO 2 × Period. Different letters indicate significant differences among CO 2 treatments in each period and ns indicate non-significant (Tukey's test, p < 0.05).  Rubisco content decreased under eCO 2 conditions, but there was no N fertilization effect ( Table 1). The average Rubisco content decreased with increasing CO 2 concentration (maximum p = 0.040) and Rubisco content under eCO 2 1.8 condition decreased by 39.0% compared with that under aCO 2 condition in N3 (Table 2, p < 0.001). In addition, Rubisco contents showed an interaction between CO 2 and time period (Table 1). Rubisco did not differ with CO 2 concentrations in 2019 but decreases under eCO 2 condition in 2020 (Figure 2b). Rubisco content was 37.0% and 36.5% lower under eCO 2 1.8 than under aCO 2 in July (aCO 2 : 2.38 ± 0.15 g m −2 , p = 0.014) and September (aCO 2 : 1.66 ± 0.20 g m −2 , p < 0.001), respectively.

Nitrogen Chamber
To examine the dilution hypothesis and determine whether N fertilization could mitigate the down-regulation of photosynthesis, NSC were quantified in different CO 2 and N treatments. NSC differed depending on the CO 2 treatments and time periods (Table 1). In contrast to previous results of NSC under eCO 2 conditions, the average NSC was 44.5% lower under eCO 2 1.4 (p < 0.001) than under aCO 2 , and the average NSC under eCO 2 1.8 tended to be lower than that under aCO 2 but not significantly (p = 0.061). Moreover, the reduction in NSC under eCO 2 conditions was noticeably greater in N3 than in the other N treatments ( Table 2). The NSC in September (28.21 ± 1.51 g m −2 ) were higher than those in July (2019: 21.58 ± 0.94 g m −2 , 2020: 19.78 ± 1.02 g m −2 , maximum p < 0.001). In addition, NSC showed a correlation between CO 2 concentrations and time period (Table 1). There was no reduction in NSC under eCO 2 in July 2019 and 2020, but NSC were reduced by 36.5% and 24.1% under eCO 2 1.4 (p < 0.001) and eCO 2 1.8 (p < 0.001) relative to those under aCO 2 (35.40 ± 3.19 g m −2 ) in September 2020 (Figure 2c). Table 1 shows that PNUE also increased with eCO 2 concentration. The average PNUE was higher under eCO 2 1.4 (52.0%, p < 0.001) and eCO 2 1.8 (66.6%, p < 0.001) than under aCO 2 , whereas PNUE increased under eCO 2 1.8 in all N treatments (Table 2). There was a decrease in the chlorophyll content under eCO 2 1.8 compared with that under aCO 2 ( Table 1). The chlorophyll content was 11.8% lower under eCO 2 1.8 than that under aCO 2 (p = 0.006, Table 2). Moreover, the chlorophyll: Rubisco ratio showed no differences between CO 2 treatments (Table 1); thus, there was no leaf N reallocation under eCO 2 conditions and N fertilization treatments.

Whole-Plant Biochemical Characteristics
Stem N mss was increased under eCO 2 conditions, but there were no differences between the N treatments (Figure 3a). Stem N mss under eCO 2 1.8 (0.56 ± 0.05%) increased by 44.9% and 39.7% compared with that under aCO 2 (p = 0.004) and eCO 2 1.4 (0.40 ± 0.03%, p = 0.008), respectively. Although there was no interaction effect between N and CO 2 , stem N mass under eCO 2 1.8 was 74.1% and 83.5% higher than that under aCO 2 (p = 0.036) and eCO 2 1.4 (p = 0.020) in the N3 treatment, respectively. There were no differences in root Nmass between CO 2 and N treatments (Figure 3b). For NSC, there was no effect of CO 2 and N treatments, with an interaction effect of only N × CO 2 on the stem. The NSC under eCO 2 1.4 (230.08 ± 56.50 g) decreased by 76.1% compared with that under aCO 2 (307.99 ± 28.79 g) only in N2 (p = 0.006) (Figure 3c). There were no differences in the root NSC between the CO 2 and nitrogen treatments (Figure 3d). Nmass between CO2 and N treatments (Figure 3b). For NSC, there was no effect of C and N treatments, with an interaction effect of only N × CO2 on the stem. The NSC und eCO2 1.4 (230.08 ± 56.50 g) decreased by 76.1% compared with that under aCO2 (307.9 28.79 g) only in N2 (p = 0.006) (Figure 3c). There were no differences in the root NSC b tween the CO2 and nitrogen treatments (Figure 3d). In the summary of the two-way ANOVA, CO2 represents the CO2 treatment and N represents the nitrogen treatment. The X axis represents N fertilization and CO2 treatments: aCO2 (white bar), eCO21.4 (grey bar) and eCO21.8 (black bar). The significance of the primary effects and their interactions are shown with the abbreviations. n.s, non-significant; *, p < 0.05; **, p < 0.01; ***, p < 0.001 (two-way ANOVA, CO2 × N). Different lowercase letters indicate significant differences among CO2 treatment in each N treatments and capital letters indicate significant differences among N treatment (Tukey's test, p < 0.05).

Biomass under eCO2 and N Fertilization
Biomass enhancement under eCO2 conditions has been reported in many prev studies [41]. However, the N availability and water status determined the incremen net primary production at the eCO2 Duke FACE site [42] and Flakaliden [43]. In the derstory, the FACE experiment across years reported that the aboveground biomas the understory community was 25% greater under eCO2 than under aCO2 [44]. Howe similar to our study, which showed no increase in understory seedling biomass un eCO2 conditions in any plant part (leaf, stem, or root) (Figure 1), 15 tropical spe showed no increase in understory stand biomass (leaf and stem) under eCO2 condit . Results are presented as mean + SE and summary of a two-way ANOVA. In the summary of the two-way ANOVA, CO 2 represents the CO 2 treatment and N represents the nitrogen treatment. The X axis represents N fertilization and CO 2 treatments: aCO 2 (white bar), eCO 2 1.4 (grey bar) and eCO 2 1.8 (black bar). The significance of the primary effects and their interactions are shown with the abbreviations. n.s, non-significant; *, p < 0.05; **, p < 0.01; ***, p < 0.001 (two-way ANOVA, CO 2 × N). Different lowercase letters indicate significant differences among CO 2 treatment in each N treatments and capital letters indicate significant differences among N treatment (Tukey's test, p < 0.05).

Biomass under eCO 2 and N Fertilization
Biomass enhancement under eCO 2 conditions has been reported in many previous studies [41]. However, the N availability and water status determined the increment of net primary production at the eCO 2 Duke FACE site [42] and Flakaliden [43]. In the understory, the FACE experiment across years reported that the aboveground biomass of the understory community was 25% greater under eCO 2 than under aCO 2 [44]. However, similar to our study, which showed no increase in understory seedling biomass under eCO 2 conditions in any plant part (leaf, stem, or root) (Figure 1), 15 tropical species showed no increase in understory stand biomass (leaf and stem) under eCO 2 conditions [45]. Moreover, increased seedling biomass under eCO 2 conditions is associated with light availability [46]. However, the pot limitation effect can also limit the biomass incurred by eCO 2 . In addition, F. excelsior and Dactylis glomerata seedlings showed marginal increases under eCO 2. conditions in a pot experiment, similar to our results [47].
N fertilization can cause an increase in net primary production in terrestrial ecosystems [48]; however, understory tree responses to N fertilization were not significant in some studies [49,50]. In addition, N fertilization significantly decreased the aboveground biomass of understory Dicranopteris dichotoma and Lophatherum gracile by 82.1% and 67.2%, respectively, compared with that under unfertilized treatments [51]. However, in our study, leaf and stem biomass were increased by N fertilization without an interaction effect between CO 2 and N ( Figure 1). In summary, unlike overstory biomass, there was no increase in understory biomass under eCO 2 conditions owing to low light availability and the effect of N availability but without an interaction with eCO 2 .

Down-Regulation of Photosynthesis and NSC under eCO2 and N Fertilization
Tables 1 and 2 show reductions in leaf N and Rubisco contents under eCO 2 conditions, without an N fertilization effect, except in case of leaf N mass . A previous meta-analysis showed that down-regulation of photosynthesis was associated with reduced leaf N mass and Rubisco content under eCO 2 conditions [8,52]. Similar to the findings of other studies, leaf N and Rubisco contents also decreased under eCO 2 conditions in understory seedlings (Table 2) in our study; however, the differences in N mass among CO 2 conditions decreased owing to retranslocation in September (Figure 2a). The dilution hypothesis suggests that down-regulation of photosynthesis is because of NSC accumulation resulting from a sink-source imbalance [11,53]. Yin [23] suggested that this sink-source imbalance can be mitigated by N fertilization under eCO 2 conditions ; however, in our study, there were no interaction effects between N and CO 2 on J max , Rubisco content, and NSC, and the differences in Rubisco content and NSC among CO 2 treatments were greater in N3 due to increased Rubisco content under aCO 2 conditions by N fertilization (Tables 1 and 2). Moreover, contrary to the dilution effect hypothesis, a reduction in NSC under eCO 2 1.4 condition occurred in our study (Tables 1 and 2). Some studies have shown no differences in understory NSC under eCO 2 conditions, whereas NSC in a light-sufficient canopy increased under eCO 2 conditions. For example, the NSC of Hedera helix was consistently 60% higher under eCO 2 in the subcanopy; in contrast, leaves in the understory showed no differences in NSC between eCO 2 and aCO 2 conditions [54]. The accumulation of NSC increased noticeably with height in the canopy under eCO 2 conditions in 15 tropical species, suggesting that full solar illumination accelerates sink and source imbalances under eCO 2 conditions [45].
In addition, there was an interaction effect of CO 2 and time period, resulting in a greater decrease in NSC under eCO 2 conditions in September 2020, when leaf senescence started in the overstory and light availability increased in the understory (Table 1, Figure 2c). Therefore, decreased NSC under eCO 2 conditions in the understory was related to decreased leaf N concentration and low light availability, and down-regulation of photosynthesis was not caused by the accumulation of NSC in this study. Some studies have suggested that photosynthesis down-regulation cannot be explained entirely by sink size [55,56]. Bloom, et al. [57] and Wujeska-Klause, Crous, Ghannoum and Ellsworth [17] suggested that lower leaf N concentration under eCO 2 condition than under aCO 2 condition is driven by insufficient reductant from decreased photorespiration, resulting in reduced NO 3 − assimilation compared with root absorption of NO 3 − . Although photosynthesis was downregulated, A max enhancement of understory saplings under eCO 2 conditions occurred for 2 years, regardless of N fertilization (Table 2). Similar to our results, Sefcik, et al. [58] reported that the long-term improvement of A max under eCO 2 conditions was 97% for seedlings in deep shade and 47% for those in moderate shade compared with seedlings grown under CO 2 conditions. Similarly, several previ-ous studies have reported that the impact of eCO 2 increases A max under light-limiting conditions [59][60][61].

PNUE and N Allocation under eCO2 Conditions and N Fertilization
Increased PNUE is associated with the enhanced efficiency of Rubisco due to photorespiration suppression under eCO 2 conditions [8]. Similar to the findings of a previous study, PNUE increased under eCO 2 conditions regardless of N fertilization (Tables 1 and 2). In general, increased leaf N concentration was accompanied by decreased PNUE, but the increase in PNUE was sustained under eCO 2 conditions despite a high N supply, similar to our results [62,63]. Davey, Parsons, Atkinson, Wadge and Long [63] suggested that the eCO 2 effect may improve PNUE, independent of leaf N content. Increased PNUE under eCO 2 conditions was associated with the reallocation of leaf N from Rubisco to other photosynthetic apparatuses, such as light harvesting or electron transport enzymes [64]. However, there were no differences in the chlorophyll: Rubisco ratio among the CO 2 treatments (Table 1), and chlorophyll were reduced by the effect of eCO 2 ( Table 2). In contrast to the findings of our study, other studies have noted an increase in chlorophyll under eCO 2 conditions [65]. In addition, a small decrease in Rubisco content via suppression of Rubisco small subunit (RBC) genes led to increased chlorophyll content under eCO 2 conditions [16]. In contrast, the chlorophyll content was significantly reduced under eCO 2 conditions in some seedling experiments [66,67].

Biochemical Changes in Sink Organs under eCO2 and N Fertilization
Unlike the reduction of leaf N mass under eCO 2 conditions, the stem N mass under eCO 2 conditions increased compared with that under aCO 2 conditions, showing noticeable differences under N3 treatment in this study (Figure 3a). Root N mass also showed a similar tendency, but the difference was not significant (Figure 3b). However, the average root N mass under eCO 2 conditions decreased by 7.1% compared with that under aCO 2 conditions in a meta-analysis [68], and stem N mass under eCO 2 conditions was also reduced in 11 species [69]. However, similar to the findings of our study, eCO 2 significantly increased the grain yield of rice crops in middle N and high N concentrations with a reduction of leaf N concentration under eCO 2 conditions, which was attributed to the increased N uptake under eCO 2 conditions being the highest at high N concentration [70].
Taub and Wang [19] suggested that the reduction of N mass in whole plants can be attributed to the accumulation of NSC under eCO 2 conditions. In this study, the NSC of stems showed no differences among CO2 treatments, except for N2, being lower under eCO 2 1.4 than under aCO 2 (Figure 3c). For root NSC, there were no differences between the N and CO 2 treatments (Figure 3d). In the case of Populus cathayana, root starch significantly decreased under eCO 2 conditions compared with that under aCO 2 conditions under low N concentration, and stem starch increased under eCO 2 conditions in both N-fertilized and unfertilized conditions [71]. However, there were no changes in soluble sugar in aboveground woody parts and roots under eCO 2 conditions compared with those under aCO 2 conditions, whereas starch increased by 9.8% under elevated CO 2 conditions in 71 species in a meta-analysis. The NSC of roots and stems varied with species and growth conditions, and we concluded that the N mass of roots and stems was not affected by the NSC of those parts in this study.

Conclusions
The reduction of leaf N and Rubisco contents decreased under eCO 2 conditions in the understory, but N fertilization could not mitigate this decrease. In addition, NSC decreased under eCO 2 conditions; this result contradicts that of previous overstory studies, wherein leaf NSC increased under eCO 2 conditions. The reduction of leaf N and Rubisco content was not caused by NSC accumulation, and decreased NSC was associated with insufficient light availability in the understory. Reduction of leaf N mass could be related to PNUE improvement, resulting from reduced photorespiration driving less availability of reductant in NO 3 − under eCO 2 conditions. Unlike leaf N, stem N increased under eCO 2 conditions, and there was no change in root N. The increase in stem N mass might have been caused by increased N uptake under N fertilization treatment.