The experiment was conducted at the National Natural Reserve of the Wuyi Mountains (27°48.11′–28°00.35′N, 117°39.30′–117°55.47′E) in southeastern China. The reserve is in the humid warm subtropics and has a mean annual precipitation of 2583 mm and a mean annual temperature of 14.2°C. The average temperatures in July (summer) and December (winter) are 23.8 and 3.6°C, respectively. The air temperature decreases by 0.45 and 0.56°C with every 100-m increase in elevation in summer and winter, respectively. P. taiwanensis is distributed >1100 m a.s.l. We therefore established five sites along an elevational gradient: E1, E2, E3, E4, and E5 at 1200, 1400, 1600, 1800, and 2000 m, respectively. The soil N concentrations did not vary significantly with elevation. In contrast to soil N concentrations, the soil P concentrations increased significantly with elevation, from 0.19 ± 0.01 mg g^–1 (mean ± standard error, SE) at E1 to 0.43 ± 0.02 mg g^–1 (mean ± standard error, SE) at E5 (Lyu et al., 2021).

We selected the mature individuals about 30–50 years old. Furthermore, to remove the biological influence of tree age on decreasing growth at higher elevation, we selected current-year branch (without apparent leaf area loss) and collected fully mature needles to measure the carbon flux. We established three 20 × 20 m plots at each elevation. Three trees were selected in each plot. Three branches with tips at the outer edge of the crown were randomly selected for each tree in summer (July) and winter (December) in 2017. A total of 90 branches (five elevations × three plots × three trees × two seasons) were selected.

Fully mature needles were collected from each branch selected (without apparent loss of foliar area). Photosynthetic and respiration rates were measured using an LI-6800 portable photosynthesis system (LI-COR, Lincoln, United States), and temperature response curves were developed based on measurements at 17, 22, 27, 32, and 37°C in summer and 5, 10, 15, 20, 25 and 30°C in winter. The light level in the leaf chamber was maintained at 2,000 μmol m^–2 s^–1, air flow was set at 300–500 μmol s^–1, and CO₂ concentration was set at 400 μmol mol^–1. Net assimilation rate was measured from 09:00 to 12:00. The rate of dark respiration (R_d) was measured using needles shaded with a black cloth for 1 h. R_d under these conditions is stable in detached leaves for several hours or longer (Reich et al., 2016). Measurements were made in July (summer) and December (winter) from 1200 to 2000 m. All plants were measured in one elevation over 3–5 days.

Q₁₀ of the temperature-response function for each leaf, and the respiration rate at a standard measurement temperature of 25°C (R₂₅), were calculated using the temperature-response equations proposed by Slot et al. (2013, 2014b) and Reich et al. (2016):

where, a and b are, respectively the intercept and the slope of the response curve. Q₁₀ values were calculated from the slope of these equations as:

R₂₅ was calculated for each of the 5–7 set cuvette temperatures of each leaf as:

where, R₂₅ is dark respiration measured at a leaf temperature of 25°C, and a, b, and c are coefficients that describe the response of the natural log of respiration to temperature.

The photosynthetic thermal optimum for each leaf measured was estimated using nonlinear regression of the data for photosynthetic thermal response:

where, A_(T) is the measured net rate of CO₂ assimilation (μmol m^–2 s^–1) at foliar temperature T, b is a parameter for the spread of the parabola (Battaglia et al., 1996), T_opt is the optimal temperature for photosynthesis, and A_opt is the rate of photosynthesis at T_opt.

The foliar values were averaged. Mixed-effects analyses of variance (ANOVAs) were used to compare T_opt, A_opt, and parameter b. The influence of elevation on T_opt, A_opt, R₂₅, and Q₁₀ was analyzed using LSD tests and multivariate analyses of variance (multiple-comparisons ANOVAs) using the agricolae package in R version 3.4.4. These variables were assessed using IBM SPSS Statistics V.22.0 (International Business Machines Corporation, Armonk, United States). The level of significance for testing slope heterogeneity was P < 0.05 (i.e., slope heterogeneity was rejected if P > 0.05). An LSD test and a t-test were used to analyze the variance. The data for elevation and season did not differ significantly when α was examined to identify common scaling exponents using the standardized major-axis package in R.

The allometric relationships between T_opt and A_opt were described after log 10-transformation. A scaling approach consisted of y = βx^α (Eq. 4), where y and x are T_opt and A_op, respectively, β is the normalization constant (intercept), and α is the scaling exponent (slope). The equation describes an isometric relationship when α = 1 and an allometric relationship when α ≠ 1. Eq. 4 was log₁₀-transformed to log₁₀ y = log10 (β) + α log₁₀ x and then fitted using model II standardized major-axis regression of the “smatr” package (Warton et al., 2006). A common scaling exponent was calculated when the scaling exponents did not differ significantly (P > 0.05) among the groups.

The relationship between temperature and photosynthetic rate can generally be described with a parabolic curve, in which the rate increases before reaching the optimal temperature and then decreases (Battaglia et al., 1996; Thuiller et al., 2005; Walker et al., 2006). Our findings were consistent with this relationship; P. taiwanensis had a higher photosynthetic rate under T_opt conditions. The ranges of T_opt in our study were 19.25–23.6 and 10.68–17.63°C in summer and winter, respectively, and the average temperatures in summer and winter at our experimental site were 23.8 and 3.6°C, respectively (Lyu et al., 2021). These conditions indicate that rising global temperatures (of 1.1–6.4°C by 2100) (Intergovernmental Panel on Climate Change [IPCC], 2013) could increase the photosynthetic rate in P. taiwanensis, especially in winter. Optimal thermal acclimation could ensure the maximum absorption of CO₂ by plants and reduce CO₂ concentration in the atmosphere (Sendall et al., 2015).

T_opt was higher in summer than winter along the gradient and decreased significantly as elevation increased (Figure 1A). This suggests that temperature acclimation would further modify the temperature optimum in response to seasonal changes of P. taiwanensis. Elevation significantly affected T_opt. Species growing near their warmer, lower elevational limits, where boundaries are partly determined by thermal limitations (Berry and Bjorkman, 1980; Tjoelker et al., 1998, 2008; Davis and Shaw, 2001), Gunderson et al., 2010) or increased levels of competition, may be constrained in their potential to acclimate to warming (Reich et al., 2015). In contrast, species growing near their colder, higher elevational limits may respond more strongly to environmental change. This finding provides further evidence that species have capacities to acclimate relative to changing temperatures. Under a future warming scenario P. taiwanensis will move from lower to higher altitudes, probably ceding its dominance at lower altitudes but expanding to higher altitudes such is being observed in several sites along the world for other forest species (Peñuelas and Boada, 2003; Peñuelas et al., 2007).

A_opt in winter did not differ significantly along the elevational gradient, and A_opt was correlated with T_opt in summer, but not in winter. The scaling slopes of T_opt and A_opt did not differ significantly between summer and winter, but the normalization constants varied significantly, perhaps because low temperatures limit photosynthesis in alpine species in winter (Sendall et al., 2015; Lyu et al., 2021). Many studies (Smith and Dukes, 2013; Sendall et al., 2015; Smith et al., 2015) have reported that the photosynthetic rate of leaves are affected by low temperatures. Photosynthesis can also be strongly influenced by environmental factors such as light and water (Körner, 1998; Valladares et al., 2014), which may also account for the lack of significant differences in winter maximum photosynthetic rate among the elevations.

Respiration rates is generally assumed to double with 10°C temperature increase; that is, it has a Q₁₀ (the proportional increase in respiration rates with 10°C warming) of 2.0 (Slot et al., 2014a). Many studies (Stockfors and Linder, 1998; Atkin and Tjoelker, 2003; Tjoelker et al., 2008) have found that Q₁₀ decreased with increasing temperature, inconsistent with our study. Q₁₀ in our study, however, was decreased with decreasing temperature in summer. The mean Q₁₀ value was higher in winter than summer.

The respiration rate of plant leaves is extremely sensitive to changes in temperature over short timescales (several minutes); Ecosystems and plant environments, however, are controlled and regulated by the long-term threshold of the temperature of the environment but also may be affected by their own plant growth and development, including changes in foliar morphology, matrix, and nutrient status.

Q₁₀ decreased significantly with increasing elevation in summer, indicating that the respiratory sensitivity of P. taiwanensis leaves decreased significantly with decreasing temperature. This finding is consistent with a previous study on the foliar NSC concentrations where the rate decreased significantly as elevation increased (Lyu et al., 2021). P. taiwanensis is insensitive to low temperatures, which is beneficial for increasing the storage of carbohydrates, thus providing effective resource use for developing mechanisms to acclimate to high levels of stress.

As plants become less sensitive to environmental changes over time (i.e., they acclimate), the initial response can represent the instantaneous characteristics of plants. Our results indicated that P. taiwanensis could acclimate to environments with low temperatures by reducing its instantaneous sensitivity to temperature. It could thereby obtain the minimum amount of carbon necessary for survival, which could be an important strategy of carbon metabolism for survival at alpine treelines.

Climatic warming may increase plant respiration, increasing the release of CO₂ from terrestrial ecosystems and further increasing atmospheric warming. The respiratory response to temperature in our study increased with temperature at the five elevations in both summer and winter. R₂₅, however, was highest at E5 (2,000 m) and decreased toward E1 (1,200 m), with 42.7 and 21.83% decreases between E5 and E1 in summer and winter, respectively, indicating that the rate of respiration decreased with increasing temperature. Stress due to high temperatures can lead to respiratory acclimation and thereby reduce respiration (Reich et al., 2016). Plant respiration always increases in response to a short-term increase in temperature, but responses can vary over the long term (Teskey and Will, 1999; Smith and Dukes, 2013; Reich et al., 2016). A plant that has experienced warmer temperatures will typically have a rate of respiration at a given temperature lower than a plant that has experienced cooler temperatures (Slot et al., 2014a; Wang et al., 2020).

As plant respiration responds positively to temperature, a warming world may result in additional respiratory CO₂ release, and hence further atmospheric warming (Atkin et al., 2006; Reich et al., 2016). In our study, R₂₅ increased significantly with elevation and was higher in winter than summer at each elevation indicating that the warming can decrease of respiration.

Furthermore, Lyu et al. (2021) found that the respiration rates for P. taiwanensis increased with elevation in summer. It indicate that respiration rates of P. taiwanensis can acclimate to altered temperatures and weakening the positive feedback of plant respiration to rising global air temperature. Thus the increase in respiration rates of P. taiwanensis in response to climatic warming may be smaller than predicted and thus may not increase atmospheric CO₂ concentrations. The populations acclimated to lower altitude thus to high temperatures have lower R_d and thus a clear acclimation capacity to decrease R_d when temperatures rise permanently and the population has had time enough to acclimate by reducing R_d. Lombardozzi et al. (2015) and Sendall et al. (2015) suggested that foliar respiratory acclimation globally may have a larger ameliorating impact than expected on CO₂ losses with rising temperatures under climate change. Such amelioration would be even larger if stems and roots acclimated similarly to leaves, which require further research (Reich et al., 2016).