The samples sites are located in the Wuyishan National Park (117°24’13”–117°59’19”E, 27°31’20”–27°55’49”N) on the border of Fujian and Jiangxi Province. The total area of Wuyishan National Park is approximately 1,280 km², and the Jiangxi area accounts for 278.57 km². The region has a subtropical monsoon climate, with high temperatures and rain in July and pleasantly moist in January. The average annual temperature in the park is approximately 17°C–19°C, and the average annual precipitation is 1,684–1,780 mm. The highest peak of Wuyi Mountain is 2,160.8 m above sea level and the highest peak in the southeast of the Chinese mainland. There are developed complete altitudinal belt spectra in vegetation along the different elevations in Wuyi Mountain. The forest communities including from P. edulis forest and evergreen forest in low elevations, coniferous and broad-leaved mixed forest and deciduous forest in medium elevations, to dwarf forest and mountainous steppe in high elevations.

Five typical bamboo species (P. edulis, I. tessellatus, O. oedogonatum, Yushania hirticaulis, and Y. wuyishanensis) were examined along different elevations (see Sun et al., 2017). Owing to the size differences among the five species, two plot sizes were used. Specifically, three 10 m × 10 m plots were established for P. edulis at elevations 840, 1,040, and 1,240 m a.s.l.; three 10 m × 10 m plots were established for O. oedogonatum at elevations 1,100, 1,200, and 1,400 m a.s.l.; three 5 m × 5 m plots were established for I. tessellatu at elevations 1,040, 1,440, and 1,840 m a.s.l.; and three 5 m × 5 m plots were established for Y. hirticaulis (1,800 m a.s.l.) and Y. wuyishanensis at an elevation of 2,100 m a.s.l. All plots were located at least 20 m apart.

In the summer of 2017, three individuals of 3-year-old bamboo with average height and DBH were selected from each plot between 9:00 and 12:00 a.m. Overall, nine branches in each altitude gradient (three branches for each individual and the total number of samples is 99) were collected from canopies and immediately placed in water to reduce water loss. Embolisms were removed by recutting branch ends under water (Yoder et al., 1994; Mori et al., 2010; Michaletz et al., 2016). The LI-6800 portable photosynthesis system (LI-COR, Lincoln, NE, USA) was used to measure P_n. We randomly selected and tested 5 to 10 leaves (no visible signs of herbivory or disease) and recorded the mean values of P_n for each branch. The light intensity of the leaf chamber is set to 1,600 μmol m^-2 s^-1, the temperature of the leaf chamber is set to 25°C, and the humidity is 55%. The carbon dioxide concentration is set to 400 µmol mol^-1, and the flow rate is set to 500 mmol s^-1.

When the photosynthetic and respiration measurement was done, the leaves were brought back to the laboratory and scanned by an Epson V19 scanner (Epson V19, Epson, Suwa, Japan). The length and width of leaves were calculated by Image J software (National Institute of Health, Bethesda, ML, USA). The W/L is calculated as the ratio of width to length. Finally, the leaf samples were placed in the oven at 105°C for 1 h and then dried at 75°C to constant and weighed. The LMA was calculated by leaf weight and area. In addition, the dried bamboo leaves were crushed by a ball mill and screened through the 100-mesh nylon mesh and then sealed in a sealed bag for measurements. The leaf nitrogen concentration in leaves (leaf N) were measured with a Germany Vario EL III Element Analyzer, and phosphorus concentration (leaf P) was determined by a continuous flow analyzer (San ++, SKALAR, Netherlands).

The SPSS19.0 software was used to analyze the mean and standard deviation of the leaf functional traits of each bamboo species, and Pearson correlation analysis analyzed the correlations between traits. The relationships between bamboo leaf functional traits were best fit by the mathematical equation log (y) = log (β) + αlog (x), where β is the normalization constant and α is the scaling exponent. Model Type II regression was used to determine the numerical values of β and α using the (Standardised) Major Axis Estimation package ‘smatr’ version 4.0.0 in R software (R Core Team, 2020; Warton et al., 2012). The data from species showing no statistically significant differences in the numerical values of the two regression parameters were pooled to determine a common scaling exponent using the standardized major axis package in R (Warton et al., 2006; Warton et al., 2012). The significance level for testing slope heterogeneity was P < 0.05 (e.g., slope heterogeneity was rejected when P > 0.05). When α > 1 or α < 1, there is an allometric relationship between Y and M, while, when α = 1 or -1, there is an isometric scaling relationship.

A structural equation model was used to create an empirical model for predicting how morphology and nutrient affect the bamboo leaf carbon capture and economic trait variations. The model ran through SPSS AMOS 22.0 (SPSS. Inc. Chicago. IL, USA). The SEM was simplified and evaluated using maximum-likelihood chi-square tests (Grace et al., 2007). The CMIN/DF (the ratio of the chi-square test value and the degrees of freedom) was between 0.0 and 2.0, and CFI ≥ 0.90 or root mean square error of approximation (RMSEA) < 0.08; the model was considered appropriate in this study.

The leaf P of P. edulis and O. oedogonatum was the highest at low elevation, with a value of 1.71 g/kg, while the leaf P of I. tessellatus leaves was the highest at high elevation, with the values of 1.46 g/kg. Except for leaf P, no significant differences have been found in the W/L, leaf N, LMA, and P_n of P. edulis among the three elevations. The leaf N and LMA of O. oedogonatum were significantly higher at low elevations, but the leaf N of I. tessellatus was the highest at high elevations, and there was no significant difference in LMA between the three elevations. The P_n was the highest at high elevation for O. oedogonatum and I. tessellatus. The W/L, leaf N, leaf P, LMA, and P_n in the leaves of Y. wuyishanensis were significantly higher than those of Y. hirticaulis (T-test, P < 0.05).

There are significant positive correlations between W/L and leaf P, LMA versus P_n, leaf N and leaf P vs. P_n, leaf P and LMA vs. P_n, respectively, in bamboo species. The leaf P of different bamboos showed an isometric scaling relationship concerning W/L with α = 1.0, leaf N vs. P_n, and leaf P vs. P_n showed an allometric scaling relationship with exponents of 0.58 and 0.73, respectively. Meanwhile, leaf P and leaf N showed an allometric scaling relationship with an exponent of 0.80 (Figure 1). These results indicated that there was a significant trade-off between leaf P and W/L, leaf P and P_n, leaf N and P_n, and leaf N and leaf P.

The structural equation model showed that elevation influenced LMA by regulating the W/L of bamboo leaves, and leaf nitrogen and phosphorus concentration drove the changes in photosynthetic characteristics. This model had a high goodness of fit (χ²/df = 1.73, CFI = 0.97, RMSEA = 0.086). The altitude, W/L, LMA, and leaf N and P could together explain 44% of the P_n variation in five bamboo species (Figure 2A). Among them, the standard total effect value of leaf P is 70.0%, leaf N is 38.5%, and W/L is 23.6% (Figure 2B). These results indicated that the leaf W/L rather than LMA mainly (<0.001) affects the leaf N and P and thus drive the changes of P_n among five bamboos at the different elevations.

The leaf P of P. edulis and O. oedogonatum decreased significantly with increasing elevations. It is consistent with the result that Huang et al. (2020) found the leaf P of P. edulis leaves decreased with increasing elevations. Thus, our results show the limitation of phosphorus to P. edulis and O. oedogonatum growth enhanced with the altitude increasing. A potential cause for such a low leaf P may be the temperature decreased with increasing elevations, which inhibited the soil phosphorus mineralization ability (Zhang et al., 2022). The second reason is bamboo renowned for its fast growth rate (Song et al., 2017), and fast-growing organisms need to increase demand for P content that constituted ribosomes, ATP, and RuBisCO (Reich and Oleksyn, 2004; Li et al., 2021c). Owing to environmental stresses increased with increased elevations (i.e., low temperatures and nutrient loss), the photosynthetic ability and growth rates will decrease to some extent, together resulting in a decrease in leaf P storage. In contrast, the leaf P of I. tessellatus showed higher at the high-elevation site than at either the low or middle elevation (Table 1), which concurred with Han et al.’s (2005) report that leaf P increased with the mean annual temperature across China. Interestingly, previous studies demonstrated that precipitation plays a crucial role in limiting the distribution of Moso bamboo rather than the temperature across mainland in China (Shi et al., 2020). Thus, not only climate factors but also the difference in vegetation types, soil physical traits, and chemical properties affected the leaf P and may exert huge differences (van de Weg et al., 2009). Likewise, the leaf N of I. tessellatus and O. oedogonatum showed opposite trends with elevation changes (Table 1). The controversial exerts herein indicate that leaf N may not be sensitive to temperature changes. Indeed, Hong et al. (2014) reported that no relationship has been found between the leaf N concentration and the annual average temperature. These results suggest that the elevation gradient has a stronger regulation on the leaf functional traits of O. oedogonatum and I. tessellatus than P. edulis. On the other hand, in the understory habitat, the responses of leaf N and P to the elevation gradient were significantly different between O. oedogonatum and I. tessellatus (Table 1), indicating that light availability is another factor that affects these traits. Furthermore, another work suggests that the variation of phylogenetics was the main factor that affected the leaf N concentrations rather than temperature (He et al., 2006a). Generally, leaf P and N are among the most crucial nutrients and limit photosynthesis ability in terrestrial ecosystems (Elser et al., 2007). Our results indicated that P_n was the highest at high elevations for I. tessellatus might benefit by having a higher nutrient concentration (leaf P the leaf N) than lower and medium elevations (Table 1).

The functional traits of bamboo leaves may not only be regulated by a single environmental factor but are also constrained by the scaling relationships between leaf traits. Previous studies have found that there are extensive resource allocation strategies between leaf economic traits in plants, including leaf N and P, P_n, the respiration rate (R_d), and the leaf life span (LL) and LMA (Wright et al., 2004; Li et al., 2021a). For instance, Reich et al. (2010) suggest that a two-thirds allometric relationship exists between leaf N and P in major global plant communities. Indeed, our results show a significant positive correlation between leaf N and P across five bamboos (Table 2) with the scaling exponents of 0.80 (CIs: 0.68–0.94) (Figure 1). Moreover, it is not significantly different from the exponents of 0.69 (CIs: 0.63–0.79) that were found between leaf N and P in the global data (P_0.69 = 0.06) reported by Reich et al. (2010). The scaling relationship between leaf N and P and P_n was α < 1.0, with the exponents of 0.58 and 0.73, respectively (Figure 1). It indicates that more leaf P than N might be needed to increase the P_n. There may be two reasons. One is that plants tend to have closer trait correlations when facing the pressure of resource acquisition and utilization (Liu et al., 2019). Under the condition of general phosphorus deficiency in a subtropical forest, the photosynthetic carbon acquisition capacity of leaves is limited by the change in leaf phosphorus content, indicating that bamboos have a high demand for the limiting the element phosphorus (Li et al., 2021c; Aribal et al., 2022). Second, the distribution of phosphorus components in leaves leads to the possibility that bamboo allocates more leaf P to the photosynthetic system. The plants usually increase metabolic phosphorus and reduce phospholipid input to enhance photosynthetic phosphorus utilization efficiency adapted to low-phosphorus environments (Hidaka and Kitayama, 2013; Hayes et al., 2018). Previous studies also indicate that bamboo could maintain the normal photosynthesis process and growth under low-phosphorus conditions (long-term nitrogen deposition test) (Song et al., 2020).

As a crucial leaf trait, the LMA can be seen as the leaf investment of plants and closely correlates with the leaf life span (LL), nutrient concentration, and photosynthetic capacity (Westoby et al., 2002). In this study, LMA is positively related to leaf P but not with P_n (Table 2). Moreover, W/L is positively correlated with leaf P and P_n, respectively (Table 2), which indicates that P_n keeps pace with a leaf shape change. Consistent with some prior reports (e.g., Shi et al., 2020; Du et al., 2021), bamboo leaf surface area or specific leaf area mainly relies on the variation of W/L and thus has a closer relationship with carbon capture capacity. Therefore, we see that the positive relationship between P_n and W/L is an important attempt to reveal the links between bamboo physiological traits and leaf shape changes. Our results are consistent with the Lin et al.’s (2020) reports that W/L plays a key role in the studied leaf functional traits in bamboo, particularly in the scaling exponent of leaf dry mass vs. leaf area. Such a scaling relationship is often referred to as “diminishing returns” (Sun et al., 2017), indicating that gains in leaf area do not keep pace with increasing leaf mass investments. Nevertheless, our results found the isometric relationship between leaf P and W/L, which indicated that bamboo leaf shape increased proportionately with phosphorus concentration increase. Further, the allometric scaling relationship has been found between P_n and leaf P. In this context, it is critical to note that P_n gains do not keep pace with increasing leaf P and W/L investments. When these patterns are concerned together, it is reasonable to conclude that W/L should be included in the study of the leaf economic spectrum for bamboo in the future.

Since the adjustment of leaf economic traits is deeply correlated with photosynthesis and productivity, it is necessary to quantitatively analyze the processes of leaf economic traits driven by environmental and intrinsic factors (Li et al., 2021a). The structural equation model has been used to quantitatively analyze the causal relationship between variables, which has been more mature in the field of plant functional trait research (Vile et al., 2006; Yu et al., 2022). This study hypothesized that elevation drives changes in leaf morphological characteristics and nutrient content that affect photosynthesis (Figure 2) as an initial model. Indeed, the data analysis revealed that altitude, morphology trait (LMA and W/L), and nutrient content (leaf N and P) together explain 44% variations of P_n (Figure 2). On the one hand, compared with leaf W/L, the mean value of the LMA ranged from 42.04 g cm^-2 to 59.08 g cm^-2 and the coefficient of variation of LMA among five bamboos is lower (i.e., LMA is 12.26% and W/L is 20.64%). Previous studies have also found that W/L performed a higher variability than LMA (Lin et al., 2020). Therefore, contrary to W/L, the response of LMA to the elevational gradient might not be more sensitive than W/L. These results have also confirmed the “diminishing returns” hypothesis, indicating that there exists a constant allometric relationship between leaf bamboo area (carbon capture) and dry mass (investment) in different environments (Sun et al., 2017; Lin et al., 2020). On the other hand, tropical and subtropical forests are usually phosphorus-limited ecosystems (Hou et al., 2020), and other studies showed that, in habitats with high resource acquisition and utilization pressure, plant functional traits will have closer correlation characteristics (Liu et al., 2019). Therefore, in the structural equation model of this study, the standard direct effect contribution of leaf P to the model is 70.0%, which also proved that bamboo species have the characteristics of high demand for phosphorus concentration in a subtropical forest (Li et al., 2021c; Li et al., 2022b). Furthermore, we highlight that future studies need to focus on the leaf phosphorus concentration and its relationship with leaf shape traits, particularly W/L, which may provide an important view to study bamboo leaf economic traits changes in different environments.