The Accumulated Response of Deciduous Liquidambar formosana Hance and Evergreen Cyclobalanopsis glauca Thunb. Seedlings to Simulated Nitrogen Additions

Nitrogen depositions in the Yangtze River Delta have is thought to shift the coexistence of mixed evergreen and deciduous species. In this study, the seedlings of the dominant evergreen species Cyclobalanopsis glauca Thunb. and the deciduous species Liquidambar formosana Hance from the Yangtze River Delta were chosen to test their responses to simulated N additions using an ecophysiological approach. N was added to the tree canopy at rates of 0 (CK), 25 kg N ha−1 year−1 (N25), and 50 kg N ha−1 year−1 (N50). The leaf N content per mass (N m, by 44.03 and 49.46%) and total leaf chlorophyll content (Chl, by 72.15 and 63.63%) were enhanced for both species, and C. glauca but not L. formosana tended to allocate more N to Chl per leaf area (with a higher slope). The enhanced N availability and Chl promoted the apparent quantum yield (AQY) significantly by 15.38 and 43.90% for L. formosana and C. glauca, respectively. Hydraulically, the increase in sapwood density (ρ) for L. formosana was almost double that of C. glauca. Synchronous improved sapwood specific hydraulic conductivity (K S, by 37.5%) for C. glauca induced a significant reduction in stomatal conductance (g s) (p < 0.05) in the N50 treatments, which is in contrast to the weak varied g s accompanied by a 59.49% increase in K S for L. formosana. As a result, the elevated maximum photosynthesis (A max) of 12.19% for L. formosana in combination with the increase in the total leaf area (indicated by a 37.82% increase in the leaf area ratio-leaf area divided by total aboveground biomass) ultimately yielded a 34.34% enhancement of total biomass. In contrast, the A max and total biomass were weakly promoted for C. glauca. The reason for these distinct responses may be attributed to the lower water potential at 50% of conductivity lost (P 50) for C. glauca, which enables higher hydraulic safety at the cost of a weak increase in Amax due to the stomatal limitation in response to elevated N availability. Altogether, our results indicate that the deciduous L. formosana would be more susceptible to elevated N availability even if both species received similar N allocation.

iNTRODUcTiON Evergreen and deciduous broad-leaved tree species can coexist across a variety of landscapes around the globe and play important roles in forest structure and functions (Wang et al., 2007;Kikuzawa et al., 2013;Ouédraogo et al., 2016). In these coexistent ecosystems, they usually differ in ecological habits, which contributes to the explanation of the mechanisms by which these two groups coexist (Xie et al., 2012;Devi and Garkoti, 2013;Álvarez-Yépiz et al., 2017). In recent decades, the fertilization effect of nitrogen deposition in forest ecosystems has received increasing attention (Wang and Feng, 2005;Zhang, 2006) and has been recognized as a threat to plant diversity in these mixed forests (Bobbink et al., 2010;Hietz et al., 2011;Liu et al., 2013a;Lu et al., 2014). As reported, the global nitrogen deposition in the end of last century was already exceeded 25 kg −1 ha −1 yr −1 (Binkley et al., 2004), and will be doubled in the end of this decade (Deutsch and Weber, 2012). However, the underlying mechanisms have rarely been discussed in previous studies (Takashima et al., 2004;Palmroth et al., 2014). Evergreen and deciduous species usually specialize their habitats and survival strategies for the tradeoff between photosynthetic capacity and persistence (Takashima et al., 2004;Shipley et al., 2006;Curtis and Ackerly, 2008). This specialization could enable the deciduous species with higher demands for nitrogen acquisition to photosynthesize more efficiently and allow evergreen trees to invest more N in the durable leaves that can persist through disturbances (Pringle et al., 2011). The existing evidence across 231 evergreen species and 110 deciduous species has already shown higher N enrichment for the latter due to the added N (Xia and Wan, 2008). Thus, we could expect that the deciduous species would also accumulate more leaf N than the evergreen species in the coexistent ecosystems. In addition, the evergreen and deciduous species also differ in the fraction of N investment in photosynthesis. Evidence has indicated that the photosynthetic N use efficiency (PNUE) is much lower for evergreen species than for deciduous species, which was primarily ascribed to the smaller fraction of nitrogen allocated to the photosynthetic apparatus, such as chlorophyll content, in evergreen species (Takashima et al., 2004;Hu et al., 2008). Comparably higher nitrogen content and specific leaf area (SLA) for the deciduous species enable their stronger morphology and photosynthetic plasticity (Hu et al., 2008). Therefore, this evidence supports the increased opportunity for deciduous species to take advantage of high N conditions (Takashima et al., 2004).
However, uncertainties still exist because disturbances such as seasonal drought could negate the fertilization effect for deciduous species by removing the ability to utilize elevated N (Wang et al., 2012a;Liu et al., 2013b). The underlying physiological reasons may be attributed to the different hydraulic tolerances in response to drought (Pivovaroff et al., 2016). The deciduous species possessed hydraulic architecture typical of drought-sensitive plants, i.e., low wood density, wider xylem vessels, higher sapwood-specific hydraulic conductivity (K S ), and high vulnerability to drought-induced embolism (Choat et al., 2004;Kröber et al., 2015). In contrast, the evergreen species had lower K S and leaf specific conductivity but were less susceptible to embolism (Choat et al., 2004;Chen et al., 2008). These differences would lead to stronger stomatal limitations for the deciduous species in dry conditions (Brodribb et al., 2002;Zhang et al., 2013;Kröber and Bruelheide, 2014). In addition, due to the fertilization effect, elevated photosynthesis leads to fast growth as well as decreased wood density and increased stem hydraulic conductivity (Pivovaroff et al., 2016), which potentially enhances the vulnerability of deciduous species in response to seasonal drought (Bauer et al., 2001;Wang et al., 2012b).
As a developing country during past decades, China has also experienced the most severe atmospheric nitrogen deposition in the world due to the rise of anthropogenic nitrogen release (Liu et al., 2013b;Lu et al., 2018). In particular, the Yangtze River Delta region, where urbanization is highly developed, is crucial for nitrogen deposition (Liu et al., 2013b), since the nitrogen deposition rates reached an level of 38.4 kg ha−1 yr−1 in the end of last century (Zhou et al., 2001). In this study, we focused on the morphological and physiological responses of deciduous and evergreen species to the abundance of nitrogen to investigate whether the former could take advantage of the fertilization effect more fully due to seasonal drought disturbance. We hypothesized that: 1. The resource acquisition and utilization strategies of deciduous species will make them more susceptible to the added N, thus leading to a higher increase in leaf N and chlorophyll content than observed in the evergreen species. 2. Meanwhile, the increased stem conductivity and decreased wood density of deciduous species will lead to stronger stomatal limitation, which will neutralize the increase in leaf photosynthesis as well as the overall biomass.
Understanding the effects of N deposition on plant physiological processes for deciduous and evergreen species may illuminate the mechanisms behind overall forest responses to global changes.

Site Descriptions
The study was conducted in the Dashu Garden in Jinhua, Zhejiang Province (119.79°E, 29.16°N), which is characterized by a subtropical monsoon climate with an average annual temperature of 17.3°C and an average annual precipitation of 1,300-1,600 mm. The elevation was 163 m and the soil was typical yellow soil. Located at the margin of temperate and subtropical zones, the vegetation is mostly covered by evergreen and deciduous broad-leaved mixed forests. The dry season starts in September and continues into March of the next year (Figure 1). In January 2017, seedlings of two dominant species, Cyclobalanopsis glauca Thunb. (evergreen broad-leaved) and Liquidambar formosana Hance (deciduous broad-leaved) were planted in holes (50 cm in diameter and 50 cm in depth) to carry out the simulated nitrogen addition experiments. In order to manipulate the original growth habitats of the two species more closely and ensure the water uptake from the deep soil, the typical mountainous red-yellow soil was moved from the secondary bare land in North Mountain, which was 1 km away from our site, and fully mixed into planting substrate in the hole.
Current N deposition rate in tropical forests of southern China is reported to range from 15 to 73 g N m −2 a −1 , while the total deposition in our site was 2.69 g N m −2 a −1 (Xu et al., 2015). To simulate natural nitrogen deposition that might occur in the future, three treatments were designed, including the control group (CK), 25 gN m −2 a −1 (N25), and 50 gN m −2 a −1 (N50) (nitrogen from atmospheric deposition was not included); each had five replicates. Each replicate consisted of 10 individuals of 2-year-old seedlings. The experiment was started in April 2017. The NH 4 NO 3 solutions were evenly sprayed from the canopy at the middle and the end of each month until October 2018. During the period of defoliation for L. formosana (November 2017 to March 2018), the NH 4 NO 3 solution was sprayed onto the surface of the soil. In order to exclude the influence of changed water -regime, no extra water was input to the plots except for the rain. Weeding was carried out during the experiment to minimize the interference caused by other factors. After the N treatment was stopped, the leaves of both species were in good condition, indicating that the leaves were not significantly damaged.

gas Exchange and leaf Water Potential
The gas exchange measurements were carried out from 9:30 to 11:30 on typical sunny days from October 11 to October 20, 2018. The photosynthetic light response curve was measured in the fully developed mature leaves (n=15 for each treatment) using a portable photosynthesis system (LiCOR 6800, LiCOR Biosciences, Lincoln, NE, USA). The light intensity intervals were set as 0, 100, 200, 400, 600, 800, 1,000, 1,200, 1,500 μmol·m −2 ·s −1 , and the temperature was 25°C. The flow rate was set at 500 μmol s −1 , and the CO 2 concentration was set at 500 μmol·mol −1 . We use a nonorthogonal hyperbolic model to obtain the light response curve parameters: A is the CO 2 gas exchange rate (μmol CO 2 m −2 s −1 ), I is the instantaneous light intensity (μmol m −2 s −1 ), apparent quantum yield (AQY) is the apparent quantum yield, A max,g is the assimilation rate under saturating light (μmol CO 2 m −2 s −1 ), θ is the curvature, and R d is the dark respiration rate (the CO 2 exchange rate at I=0, μmol CO 2 m −2 s −1 ). We also recorded stomatal conductance (g s, mol m −2 s −1 ) and intercellular CO 2 concentrations (C i , μmol m −2 s −1 ).
Daily courses of leaf instantaneous net photosynthetic rate (A i , μmol CO 2 m −2 s −1 )), g s , and transpiration rate (E t , μmol m −2 s −1 ) were measured at 2 h intervals from 5:00 am to 17:00 pm during October 22 to October 25 on 5 seedlings of each N treatment. Meanwhile, leaf water potential (Ψ L , Mpa) of the detached shoots was synchronously measured with a pressure chamber (PMS, Albany, OR, USA) after the instantaneous gas exchange measurements.

Hydraulic conductivity and Sap Wood Density
The current year branches were cut off (30 cm length) from each sampled tree (n=15). All branches were covered with black plastic bags to prevent water loss before they were immediately transported to the laboratory to determine the sapwood hydraulic conductivity (K S , g cm MPa −1 min −1 cm −2 ) with a high pressure flow meter (HPFM Gen3; Dynamx Corp., Elkhart, Indiana, USA). The specific experimental operations are as follows: First, all the samples were cut off underwater at approximately 5 cm from the base. Then, the tree bark was removed approximately 3 cm from the sample ends. Emboli were removed from samples by vacuum infiltration under solution consisting of 0.22 μm filtered and degassed, distilled water for 8 h before being connected to HPFM under purified water (Barnard et al., 2011). Under the quasi-steady-state mode, deionized water purified by the Water Purification System (Milli-Q Advantage, Merk Millipore, Germany) was degassed and injected into the branches through HPFM at a pressure of 0.5 MPa until a stable flow rate appeared (approximately 5 to 10 min) to obtain the maximum whole-shoot level hydraulic conductance (K h , g cm MPa −1 min −1 ). After the measurements, the base diameter (d, mm) was measured with a vernier caliper to obtain the cross-sectional area (A S, cm 2 ) of the base of the shoot, and the maximum sapwood hydraulic conductivity K S (g cm MPa −1 min −1 cm 2 ) was calculated as: Five segments of 3 cm length for each branch were collected to measure the sap wood density (ρ, g cm −3 ). The fresh volume (V, cm 3 ) of these wood segments was determined gravimetrically by water displacement according to Archimedes' principle after removing the bark and phloem (Osazuwa-Peters and Zanne, 2011). All samples were then oven-dried at 105°C for 24 h to get the dry weight (G, g). The sapwood density was calculated as ρ = G/V. Stem xylem vulnerability was measured using the centrifuge method (Cochard et al., 2005;Sperry et al., 2012). After determining the maximum K S , a centrifugal machine (Sorvall RC-5C; Thermo Fisher Scientific, Waltham, MA, USA) equipped with a custom rotor (Alder et al., 1997) that was able to spin the stem segments was used to induce negative xylem pressure and induce cavitation. Hydraulic conductivity was measured between each induced pressure, and the percent loss of conductivity under a certain negative pressure (PLC i , %) was calculated as: where K hi (g cm MPa −1 min −1 ) and K max (g cm MPa −1 min −1 ) refer to the K h at certain negative pressures and maximum K h . Vulnerability curves were constructed by plotting pressure versus PLC i and fitting a Weibull model (Pammenter and Van der Willigen, 1998): According to this equation, water potential at 50% of conductivity lost (P 50 , MPa −1 ) was determined for each species.

The leaf Economic Traits
One hundred foliage round pieces were collected for 10-20 mature leaves with a hole puncher (10 mm in diameter) for each species in each treatment and were placed in the oven in an envelope for 24 h. The foliage rounds were weighed to estimate the (cm 2 g −1 ). The remaining leaves in another envelope B. Five to 10 leaves were taken from the remaining leaves of each shoot and cut into slices with a width of 0.5 mm (0.2-0.5 g). The leaf chlorophyll of these samples was extracted with a mixed solution (acetone:ethanol = 2:1) for approximately 24 h until they turned white. Then, a spectrophotometer was used to measure the absorbance of the supernatant liquid. The calculation of the total chlorophyll content (Chl (mg m −2 ), chlorophyll a + chlorophyll b) followed the description in Zhang et al. (2017). All the leaves, including those measured for leaf chlorophyll and SLA, on each shoot were then collected and dried in the oven at 65°C for 24 h and weighed for leaf dry mass (M L , g). The multiples of M L and the SLA for each shoot were used obtain the total leaf area (A L, m 2 ), which was used to calculate the ratio of leaf area to sapwood area (A L /A S, m 2 cm −2 ).
The dried leaves were ground into powder and analyzed for N content as a mixed sample by an elemental analyzer (EA Flash 1112; Thermo Fisher Scientific). The nitrogen content per unit area (N a , g m −2 ) and the nitrogen content per unit mass were calculated (N m , mg g −1 ).

The Whole Tree Biomass
To estimate the whole tree biomass, 15 sample trees were selected from each group and harvested to obtained the above-ground and underground total biomass (except fine roots) by calculating the total biomass of the leaves and the root-shoot ratio of each sample tree. Then, we calculated the leaf mass fraction (LMF, %) and the leaf area ratio (LAR, m 2 g −1 ), of which total the leaf area is the product of SLA and the total biomass.

Data Analysis and Processing
Analysis of variance (ANOVA) was conducted on all the traits measured above with SPSS 25 (version 25.0, IBM SPSS Inc., Chicago, USA) to examine the treatment effects (n = 3) for different species. The relationship between the SLA, Chl, A max , and g s values and N a and N m across the two species and three treatments were also fitted to analyze the convergence of leaf functional traits due to the elevated N availability. We also regressed A max and K S with g s for the different species in each treatment to verify the variation of the coordination relationship between shoots and leaves in response to N addition. In these analyses, individual sample points (or leaves) were used. Where a significant relation between any two variables was found (p < 0.05), we re-analyzed the data with analysis of covariance (ANCOVA) allowing the slope and intercept to vary among treatments. Therefore, in our analysis, a single curve indicates that the treatments or tree species have the same fitted relationship (i.e., the interaction items in the ANCOVA analysis are not statistically significant, p > 0.05).

Effects of N Additions on leaf Economics
As shown in Figure 2, the SLA variation was well explained by the N m of each species along the treatments (p < 0.05 for all the relations based on the ANCOVA analysis). Of note, the data in N25 and N50 shared the same relations for L. formosana. The mean SLA for L. formosana in the highest N treatments (N50) was 184.66 cm 2 g −1 , which was 35.99% higher than the unfertilized treatment CK (p < 0.01, Table 1). Similarly, the mean SLA for C. glauca increased 34.72% in the N50 treatments compared with the CK treatments (p < 0.01, Table  1). In addition, the ratio of leaf area to sapwood area (A L /A S ) was also enhanced by 26.06% for L. formosana but not for C. glauca ( Table 1).
The fertilization significantly increased the mean leaf N m (p < 0.01, Table 1) and N a (p < 0.05, Table 1, except for L. formosana), which further enhanced the Chl in the N50 treatments by 72.15 and 63.63% for L. formosana and C. glauca respectively. The mean N m of L. formosana leaves increased 44.03% from 16.92 mg g −1 in the CK treatment to 24.37 mg g −1 in the highest fertilized N50 treatments. The corresponding increase in C. glauca was 49.46% (from 9.24 to 13.81 mg g −1 ). The difference in N a between CK and N50 of L. formosana was not significant (p > 0.05), but it was 32.10% for C. glauca.
Differences in N a across treatments and species were clearly presented by the relations between Chl and leaf N (Figure 3). For each treatment, the concentration of Chl closely responded to the variations of N a for the L. formosana with different slopes (p < 0.05, ANCOVA), which ranged from the lowest of 31.68 in the CK treatment to the highest of 116.83 in the N50 treatment. The relationship was consistent among treatments for C. glauca (p < 0.01, ANCOVA). On an area basis, the ratio of Chl to N a increased 40 and 22.2% for L. formosana and C. glauca from CK to N50, indicating the higher allocation of newly acquired N to photosynthetic functions for the deciduous L. formosana.

Variation in Photosynthetic capacity and Water Transport capacity
According to the light response curve, C i decreased by 11.64 and 42.04% for L. formosana and C. glauca, respectively. Meanwhile, AQY increased by 15.38 and 43.90%, respectively. The g s decreased by 37.5% in the N50 treatments for C. glauca but was not significant for L. formosana (only 8.33%, p > 0.05) ( Table  1). Among the treatments, the A max for L. formosana increased 12.19% from CK to N50, whereas this enhancement was not observed in C. glauca. Qualitatively, the treatment means of A max were found to increase with SLA and N m for L. formosana but were not significant for C. glauca, while the g s was reduced by the enhancement of SLA and N m only for C. glauca (p < 0.05, Figure  4). In addition, the relationship between A max and g s was altered in the N50 (Figure 5A), where the A max for L. formosana tended to be more sensitive to the increase in g s when g s < 0.15 mol m −2 s −1 , and A max for C. glauca tended to rapidly decrease when g s > 0.06 mol m −2 s −1 (Figure 5A).
In agreement with the enhanced A max (albeit not significantly so for C. glauca), the water transport capacity per sapwood area (K S ) increased by 59.49 and 39.18% for L. formosana and C. glauca, respectively. Surprisingly, the improved water transport capacity in the N50 treatments tended to maintain higher and lower g s sensitivity for L. formosana and C. glauca, respectively ( Figure 5B).

Daily Dynamics of gas Exchange and Water Relations
The light was increased on a daily basis before 9:00, and g s and E t were enhanced in the highest N treatments for both species (Figure 6). Notably, even g s for both species also experienced a rapid reduction at 11:00, which may be attributed to the lowest Ψ L at this time, while the recovery of the g s after the replenishment of the Ψ L at 13:00 only occurred for L. formosana, which also maintained higher E T at the same time (Figure 6). In contrast, the g s and E t for C. glauca persistently decreased until FigURE 2 | The specific leaf area (SLA) as samples of ten leaves averaged for each individual of Cyclobalanopsis glauca Thunb. and Liquidambar formosana selected for gas exchange measurements as a function of leaf N per unit mass (N m ). Solid lines fit linearly to the CK data, broken lines fit linearly to the N25 data, and dotted line fit linearly to the N50 data. Of note, the data of N25 and N50 share the same relations for L. formosana. the end of the experiments. As a result, the E t of L. formosana in the N50 treatments was apparently enhanced compared to that in the CK treatment, thus leading to excessive water loss (decreased lowest Ψ L ). The reduced E t in the N50 from 11:00 to 17:00 for L. formosana maintained the water status compared to the CK treatments (stable lowest Ψ L ). These differences enable L. formosana to support higher photosynthesis in the N50 treatments, at the cost of exacerbated water stress. In contrast, the A i that was elevated by the increased g s before 9:00 for C. glauca tended to be neutralized after 11:00 in the N50, with the benefits of stable water status (Figure 6).

Variation in Plant Biomass
In agreement with the enhanced leaf N m , the total biomass of L. formosana individuals increased by 34.34% in the N50 treatment (Figure 7). In contrast, the growth of C. glauca was not affected by the enhanced N availability. In addition, the N concentration of plant biomass was found to increase by 86.57% (L. formosana) and 62.22% (C. glauca) in the N50 treatments. Similarly, the highest N addition rate increased the fraction of leaves to LAR (leaf area divided by total aboveground biomass) by 37.82% for L. formosana and 44.15% for C. glauca. Notably, the LMF (leaf mass/total aboveground biomass) did not change for either species, which was consistent with the increased leaf SLA across the two species (Table 1).

Hydraulic changes
The ρ decreased by 21.56% (L. formosana) and 12.95% (C. glauca) between the CK treatments and the N50, indicating the reduced mechanical strength for both species due to the enhanced N availability. When the two species were compared, we found that the P 50 was much lower for C. glauca, which indicated higher resistance to cavitation (Figure 8). L. formosana displayed an enhanced xylem vulnerability to cavitation (increased P 50 ) in the highest N treatments, which is in contrast to C. glauca, with weak variations.

DiScUSSiON
The Accumulation and Allocation of Extra N In N-saturated ecosystems, the extra N is considered to be weakly accumulated in plant tissues (Curtis and Ackerly, 2008). The subtropical region in China was thought to be N saturated 1 | Treatment means of light-saturated CO 2 exchange rate (A max ), stomatal conductance (g s ), intercellular CO 2 (C i ), apparent quantum yield (AQY), leaf N per unit leaf mass (N m ) and per unit leaf area (N a ), total amount of chlorophyll (Chl) on a mass and area basis, leaf area per mass (SLA), hydraulic conductance (K S ), wood density (ρ), and the ratio of leaf area to sapwood area (A L /A S ).  during the past decades (Lu et al., 2014). However, unexpectedly, the mean leaf N m and N a were obviously enhanced in the N50 treatments for both L. formosana (44.03% for N m ) and C. glauca (49.46 and 32.10%) ( Table 1). Meanwhile, the leaf chlorophyll content was improved synchronously and was linearly related to the leaf N across the treatment and species (Figure 3), which revealed an enhanced ability to receive light energy, as presented in other studies (Xia and Wan, 2008;Tilman and Isbell, 2015; FigURE 4 | The averaged maximum CO 2 exchange rate per unit leaf under light-saturated conditions (A max ) and stomatal conductance for H 2 O (g s ) across each light response curve as a function of specific leaf area (SLA) and leaf N per mass (N m ). Open and closed symbols represent the data of Liquidambar formosana and Cyclobalanopsis glauca leaves, respectively. The data are shown as the means ± SD.
FigURE 5 | A max as a function of mean g s for H 2 O averaged over each light response curve and the K S averaged from per tree supported g s for H 2 O for Liquidambar formosana and Cyclobalanopsis glauca. Open and closed symbols represent L. formosana and C. glauca, respectively. p < 0.01 for linear regressions for each species and treatments. Solid lines fit linearly to CK and N25 data, dotted and broken lines fit linearly to N50 data, and the broken lines fit linearly to N25 data for C. glauca in the right Figure. Panel A: Amax as a function of mean gs, Panel B: gs as a function of mean KS. Wooliver et al., 2016). These results may indicate that the N in the ecosystem was not saturated in this ecosystem, which is contradictory to previous studies conducted in tropical forests (Brookshire et al., 2012;Lu et al., 2018). In addition, the weak changes in N a for L. formosana compared to evergreen C. glauca contradict our hypothesis, even though the former still has higher leaf N (p < 0.01). The reason may be attributed to the increase in A L :A S of 26.06% for L. formosana, but not for C. glauca, and N m increased significantly ( Table 1). The A L :A S that reciprocal to the Huber value implied the supported leaf area by the sap wood per unit area. This implied the tendency of L. formosana to enhance its light acquisition ability to optimize its carbon assimilation (Iversen and Norby, 2008). A study indicated a higher allocation to the photosynthetic apparatus in deciduous species (Takashima et al., 2004), which leads to a weak increase in N allocation per leaf area ( Table 1).
On the basis of area, the allocation of N to Chl was found to vary across the two species (Figure 3). The N a determined Chl was weakly related to the N treatments for C. glauca. Similar results have already been reported in another evergreen species, Eucalyptus cladocalyx (Myrtaceae) (Simon et al., 2010). A max enhanced fraction of Chl in the N in the highest N treatments was found for L. formosana (Figure 3), which is consistent with our hypothesis (Xia and Wan, 2008). The preferential allocation by L. formosana leaves of N to chlorophyll synthesis with increasing N fertilization is quite common in deciduous species (Baltzer and Thomas, 2005;Gradowski and Thomas, 2008;Yin et al., 2009). In fact, this evolutionary shift could also be observed in annual herbaceous plants with a similar resource utilization strategy (Feng et al., 2009;Wang et al., 2012a;Mu et al., 2016) but never in evergreen species (Warren and Adams, 2004), which need to invest the N in mechanisms to conserve water and nutrients and to tolerate water and nutrient stress (Warren et al., 2003).

The Effect of N on Photosynthesis
However, even though both species accumulated N in the leaf biomass and Chl, the elevated N was not translated to photosynthesis for C. glauca (Table 1, Figure 4). The A max for L. formosana was positively promoted by the enhanced N m and SLA (p < 0.01, R 2 = 0.27 and 0.17), which is consistent with the resource-limited photosynthesis for most deciduous species in previous studies (Xia and Wan, 2008;Kröber and Bruelheide, 2014;Wooliver et al., 2016). In contrast, the A max for the evergreen C. glauca did not respond to N accumulation. The significantly increased AQY indicated the enhanced capacity to assimilate CO 2 for both species, which was consistent with the accumulated Chl in the N50 treatment across the two species (Table 1, Figure 3). However, the large proportion of decreased C i for C. glauca (by 42.04%) may imply severe stomatal limitation when the L. formosana was compared (11.64%), thus eliminating the N effect on photosynthesis. Indeed, studies have indicated that the effect of nitrogen on photosynthesis of evergreen species will be enhanced by elevated CO 2 (Bauer et al., 2001). Thus, we can expect that the stomatal-limited CO 2 could explain the weak changes of A max for the evergreen C. glauca.

Hydraulically limited Plant Biomass Accumulation
The stomatal limitations did occur for C. glauca in the highest N treatments. The g s in the N50 treatments decreased by 37.5% for C. glauca and tended to be less related to the sapwood conductivity ( Figure 5). In contrast, the g s tended to be more and more sensitive to the increase of K S, along with the weak decrease (by 8.33%, p > 0.05) for L. formosana. The K S for both species increased significantly (by 59.49 and 39.18%). We expected that the reduced g s may be attributed to the enhanced water lose on the leaf supported by the increased K S and SLA. In dry conditions, the increased AQY will raise photosynthesis and leaf transpiration under low light conditions, which further lead to excessive water loss and severe stomatal control for the evergreen C. glauca (Figure 6). In addition, the increased SLA also added to the water loss, which elevated the light acquisition ability as well as the transpiration area. However, stomatal limitation did not occur for L. formosana, and a similar increase in SLA, AQY, and K S was observed (Table 1, Figure 6), which was thought to be related to the different water use regulation strategies for the evergreen and deciduous species. Evergreen species tended to rapidly close their stomata in response to excessive water loss to maintain higher leaf water potential in contrast to the deciduous species ( Figure  6), which is known as isohydric-prone versus anisohydric-prone behavior (Kumagai and Porporato, 2012;Klein and Niu, 2014;Siddiq et al., 2017). The first will reduce the risk of damaging xylem cavitation driven by excessive tension in the trees' hydraulic system (Klein and Niu, 2014). However, a consequence of this strategy is that these trees close their stomata in response to even mild water stress-a process that reduces leaf carbon (C) uptake (Siddiq et al., 2017). In our study, the leaf scale photosynthesis was translated into the accumulation of whole tree biomass (by 34.34%) for L. formosana in the N50 treatments but not for C. glauca, even though the N concentration of plant biomass for both species was apparently elevated. These results indicated that the prevailing seasonal drought did not neutralize the fertilization effect for L. formosana but had the inverse effect in C. glauca. It is noted that the increase in biomass observed for L. formosana does not imply certain success in species competition. In fact, even though anisohydric trees could allow their leaf Ψ (Ψ L ) to decrease during drought by sustaining relatively high g s (and thus C assimilation), a greater risk of cavitation in the xylem could ultimately lead to rapid declines in leaf water supply that may affect a range of physiological variables, including photosynthetic capacity and g s (McDowell et al. 2008;Roman et al., 2015). In fact, the P 50 of C. glauca almost doubled that of L. formosana, thus revealing advantages in response to seasonal drought. Furthermore, the elevated N availability tended to enhance the leaf water loss by increasing LAR and weakly changing LMF for both species, while the P 50 was elevated only for L. formosana. Thus, extreme drought events may threaten the survival of L. formosana via hydraulic failure (Roman et al., 2015), especially in the context of elevated N availability.
cONclUSiON In our study, the two species that belong to two different forest types behave distinctly in response to elevated N availability. Both the deciduous L. formosana and the evergreen C. glauca accumulated N in Chl, which led to elevated AQY in the leaves. In combination with the increased SLA and whole tree leaf area, the transpiration demands were promoted for both species. However, the lower compensation of elevated K S could not balance the excessive leaf water loss, which lead to decreased g s for the evergreen C. glauca. In contrast, the elevated A n for the deciduous L. formosana accumulated a 34.34% increase in whole tree biomass. However, due to the anisohydric behavior and less negative P 50 , especially when N was elevated for L. formosana, the competitive relationship between the two species is still inconclusive due to the risk of hydraulic failure in the face of the gradually enhanced seasonal drought in the Yangtze River Delta.

DATA AVAilABiliTY STATEMENT
All datasets generated for this study are included in the article/ supplementary material.

AUTHOR cONTRiBUTiONS
ZZ designed the experiments. YZ, XZ, and ST conducted the experiment, XF, YC, LZ, XL and CW conducted the statistical analyses of data. ZZ and YZ wrote the manuscript. All authors read and approved the final manuscript.

FUNDiNg
The study was supported by the National Nature Science Foundation of China (Grant No. 41701226) and the Zhejiang Province Public Welfare Technology Application Research Project (LGF19C030002). Jinhua Science and Technology Research Project (2019-4-163). Data used in this study were collected by the author and are available from the author upon request.