Elevated CO₂ Decouples the Short-Term Thermal Responses of Mesophyll and Stomatal Conductance, Enhancing Carbon Gain and Water Use at Moderate Heat but Failing at Extreme Temperatures

Abstract

Rising atmospheric CO₂ concentration and global warming jointly affect plant photosynthesis and water use, but how elevated CO₂ regulates photosynthetic short-term thermal response and its implications for leaf carbon-water balance remain poorly understood. We investigated the temperature responses (25–45 °C) of net photosynthetic rate (A_n), intrinsic water use efficiency (iWUE), stomatal conductance (g_s), and mesophyll conductance (g_m) in Liriodendron tulipifera L., a fast-growing and economically important hardwood species sensitive to heat stress, under five CO₂ concentrations (150–800 μmol mol⁻¹), and quantified the partitioning of stomatal, mesophyll, and biochemical limitations. Elevated CO₂ increased the optimal temperature (T_opt) of A_n, iWUE, and g_m, but decreased that of g_s, creating a divergent thermal response pattern that decoupled carbon acquisition from water loss. The upward shift in the T_opt of g_m enhanced CO₂ supply to chloroplast carboxylation sites at high temperatures, sustaining higher carboxylation rates, indicating that improved mesophyll diffusion, rather than biochemical capacity, drove photosynthetic short-term thermal response. The downward shift in T_opt of g_s aligned with optimal stomatal behavior theory, promoting earlier stomatal closure to conserve water, which together with enhanced mesophyll diffusion resulted in a net improvement in water use efficiency below 40 °C. At 45 °C, however, mesophyll limitation (l_m) became dominant (~70% of total limitation), indicating that elevated CO₂ cannot prevent photosynthetic breakdown under extreme heat, and the water-saving benefit was lost. We suggest that terrestrial biosphere models incorporating separate temperature dependencies of g_m and g_s would improve predictions of tree carbon-water balance under future climates.

1. Introduction

Global warming and rising atmospheric CO₂ concentrations represent two concurrent facets of climate change that impose profound pressures on plant carbon gain and water relations [1,2]. By the end of this century, under the very high-emission scenario (SSP5-8.5), global mean temperature is projected to rise by 3.3–5.7 °C, with CO₂ levels reaching approximately 1135 ppm [3]. For plants, elevated CO₂ offers a potential “fertilization effect” that enhances photosynthesis, yet this benefit is often counteracted by the negative impacts of heat stress on photosynthetic biochemistry and stomatal regulation [4,5]. Understanding how rising temperatures interact with CO₂ concentration to shape the coordinated responses of carbon assimilation and transpiration is therefore essential for predicting plant performance, forest productivity, and ecosystem resilience in future climates.

The diffusion of CO₂ from the atmosphere to the chloroplast carboxylation sites involves two major conductance components in series: stomatal conductance (g_s), which regulates CO₂ entry through the stomatal pore, and mesophyll conductance (g_m), which governs CO₂ transport from the substomatal cavity to the site of Rubisco carboxylation within the mesophyll [6,7,8,9,10]. While g_s has long been recognized as a primary limiter of photosynthesis, growing evidence indicates that g_m imposes a comparable and sometimes larger limitation, reducing the CO₂ concentration at Rubisco by 20–40% relative to the substomatal value [11,12]. Moreover, g_m is not a fixed physical property but a dynamic physiological trait, influenced by carbonic anhydrase (CA) activity, aquaporin permeability, and membrane fluidity [11,13,14,15]. Consequently, both g_s and g_m contribute to the overall diffusional limitation of photosynthesis, and their coordinated regulation is critical for optimizing the trade-off between CO₂ supply and water loss under fluctuating environments.

The response of photosynthesis to temperature is characteristically nonlinear, with net photosynthetic rate (A_n) rising as temperature increases from cool to moderate levels due to faster enzymatic kinetics, but declining sharply when temperatures exceed the thermal optimum (T_opt), largely because of Rubisco deactivation and reduced electron transport capacity [5,16,17]. This temperature dependence extends beyond biochemistry to the two major diffusional components, g_s and g_m. Both g_s and g_m generally increase with rising temperature within the moderate range, facilitating greater CO₂ supply to match the increased carboxylation demand [18,19]. However, as temperature continues to rise beyond the optimum, both conductances decline, g_s primarily through stomatal closure driven by increased leaf-to-air vapor pressure deficit (VPD) [20,21], and g_m through physical and biochemical changes in the mesophyll such as altered membrane permeability, CA activity, and cell wall properties [22]. Because A_n, g_s, and g_m all peak within a similar temperature range, their temperature responses are often treated as broadly covarying, with maximum photosynthetic rates achieved only when both stomatal and mesophyll CO₂ supply remain sufficient to sustain carboxylation [23,24]. This coordination is intuitively appealing, as carbon assimilation requires an uninterrupted CO₂ supply chain from the atmosphere to Rubisco; if any link, whether g_s or g_m, fails prematurely at high temperature, photosynthesis will become CO₂-limited.

Elevated CO₂, however, may fundamentally alter this coordinated thermal landscape. At the biochemical level, increased CO₂ availability as a substrate for Rubisco could enhance carboxylation rates at high temperatures, potentially raising the thermal optimum of photosynthesis [25,26]. Nevertheless, the temperature response of Rubisco carboxylation capacity (i.e., the maximum carboxylation rate-V_cmax) is expected to be relatively conservative, as the catalytic properties of Rubisco are constrained by its evolutionary origin and structural stability [5,25]. At the diffusional level, because elevated CO₂ steepens the CO₂ gradient from the atmosphere to the leaf interior, optimal stomatal behavior theory predicts that plants should down-regulate g_s and close stomata earlier under high CO₂ to minimize water loss [27,28,29,30,31]. At the mesophyll level, emerging evidence suggests that elevated CO₂ may also enhance the thermal stability of g_m, possibly through up-regulation of CA activity or adjustments in aquaporin expression that sustain CO₂ transport across the mesophyll at higher temperatures [15,32]. Whether these opposing influences, enhanced carbon supply and mesophyll diffusion versus earlier stomatal closure, lead to a coordinated or divergent shift in the temperature optima of g_m, g_s remains unknown. Furthermore, the potential divergence in g_m and g_s responses, if it occurs, would likely alter the partitioning among stomatal, mesophyll, and biochemical limitations across temperatures; yet how elevated CO₂ reshapes this partitioning remains unexplored. In addition, heat stress can directly impair photosystem II photochemistry and electron transport [16,26], but whether elevated CO₂ alleviates such photodamage is also uncertain.

However, to our knowledge, empirical evidence directly testing this divergent shift within a single experimental framework, using simultaneous measurements of A_n, g_m, and g_s across a wide temperature range under elevated CO₂, remains remarkably scarce. This knowledge gap is especially critical for temperate forests, which experience broad seasonal and diurnal temperature fluctuations and are projected to face increasingly frequent heatwaves under future climate scenarios [33,34]. Most previous studies have focused separately on photosynthetic heat tolerance or stomatal responses, without quantifying how the temperature optima of g_m and g_s shift relative to each other under CO₂ enrichment, nor how the partitioning of stomatal, mesophyll, and biochemical limitations shifts with temperature under elevated CO₂. In contrast, the present study provides the first simultaneous measurements of A_n, g_s, and g_m across a wide temperature range under multiple CO₂ concentrations, allowing us to directly test whether elevated CO₂ induces a divergent shift in the T_opt of g_m versus g_s [6,11].

As one of the most important commercial hardwood species in eastern North America, Liriodendron tulipifera L. is highly valued for its rapid growth and high-quality timber, and it also exhibits exceptionally high carbon sequestration capacity [35,36,37]. Moreover, as a mesic species, it is particularly sensitive to heat stress, making it a suitable indicator for assessing the impacts of climate change on forest ecosystems [36,38]. Understanding how elevated CO₂ and temperature affect its photosynthetic physiology therefore has both ecological and practical significance. In this study, we used Liriodendron tulipifera L. to systematically investigate the effects of elevated CO₂ on photosynthetic heat acclimation and CO₂ diffusion processes under controlled CO₂ concentration and temperature conditions. The specific objectives of this study were: (1) to quantify the individual temperature responses of A_n, g_s, and g_m under both ambient and elevated CO₂ concentrations; (2) to determine whether elevated CO₂ induces a divergent shift in the T_opt of g_m versus g_s; and (3) to evaluate the consequences of such divergent responses for leaf-level carbon-water balance across a thermally relevant range. By integrating simultaneous measurements of gas exchange and photosynthetic CO₂ diffusion, we aim to provide a mechanistic understanding of how rising CO₂ may reconfigure the thermal coordination between CO₂ supply and demand, and to improve current predictions of tree physiological performance under future climate scenarios.

2. Results

2.1. The Temperature Response of Photosynthetic Parameters Under Different CO₂ Concentrations

To characterize the temperature responses of photosynthetic parameters under different CO₂ concentrations, we first examined the patterns of A_n, intrinsic water use efficiency (iWUE), g_s, g_m, and V_cmax across the temperature gradient. As shown in Figure 1A, A_n increased with rising CO₂ concentration. Across all CO₂ treatments, A_n first increased and then decreased with increasing temperature, and its optimal temperature shifted upward with elevated CO₂ (see Table 1 for specific T_opt values). iWUE exhibited a unimodal response to temperature under elevated CO₂ (>400 μmol mol⁻¹), peaking at 35–40 °C and decreasing markedly thereafter. In contrast, under low CO₂ (<400 μmol mol⁻¹), iWUE declined monotonically with increasing temperature. Elevated CO₂ partially mitigated the negative effects of heat stress on iWUE; however, this protective effect was lost at temperatures exceeding 40 °C (Figure 1B). The response of g_s to CO₂ differed from that of A_n (Figure 1C). g_s first increased and then decreased with rising temperature; however, when temperature exceeded 40 °C, g_s increased again. Below 40 °C, the T_opt of g_s decreased with increasing CO₂ concentration (Table 1). Moreover, the response of g_s to CO₂ concentration varied with temperature. Below 35 °C, g_m decreased with increasing CO₂ concentration; above 35 °C, the effect of CO₂ on g_m diminished. Under 150 and 300 μmol mol⁻¹ CO₂, g_m declined slowly and then rapidly with increasing temperature. In contrast, under 500, 650, and 800 μmol mol⁻¹ CO₂, g_m first increased and then decreased with temperature (Figure 1D). The optimal temperature of g_m increased with rising CO₂ concentration. V_cmax showed a clear temperature-dependent pattern, with values rising from 50 μmol m⁻² s⁻¹ at 25 °C to a maximum of 140 μmol m⁻² s⁻¹ at approximately 40 °C, followed by a sharp decline as temperature continued to increase (Figure 2).

Figure 1.

Responses of photosynthetic parameters to temperature across a range of CO₂ concentrations (150, 300, 500, 650, and 800 μmol mol⁻¹), measured at photosynthetically active radiation (PAR) of 1200 μmol m⁻² s⁻¹. (A) Net photosynthetic rate (A_n); (B) intrinsic water use efficiency (iWUE, A_n/g_s); (C) stomatal conductance (g_s); (D) mesophyll conductance (g_m). Solid lines are GAMM-fitted curves for different parameters; solid dots are observed means for each temperature-C_a combination; vertical bars are standard errors (SE, n = 3). Hollow circles in A_n, iWUE, g_s, and g_m mark GAMM-estimated optimal temperatures (T_opt), with 95% confidence intervals estimated by parametric bootstrap (Table 1); colors denote C_a concentrations.

Table 1.

Optimal temperatures (T_opt) for net photosynthetic rate (A_n), intrinsic water use efficiency (iWUE), stomatal conductance (g_s), and mesophyll conductance (g_m) under different CO₂ concentrations (C_a) at a photosynthetically Active Radiation (PAR) of 1200 μmol m⁻² s⁻¹. T_opt for g_s was estimated over the 25–40 °C range, because g_s increased again above 40 °C; T_opt for A_n, iWUE, and g_m was estimated over the full 25–45 °C range. Values are means with 95% confidence intervals in parentheses. * represents that T_opt could not be determined within the measured temperature range.

Topt (°C)	Ca (μmol·mol−1)
	150	300	500	650	800
An (μmol·m−2·s−1)	25.0 (25.0–26.4) *	28.8 (26.7–30.9)	33.4 (31.8–35.1)	34.6 (32.9–36.3)	35.2 (33.2–37.2)
gs (mol·m−2·s−1)	40.0 (38.8–40.0) *	29.9 (28.4–32.3)	30.0 (29.0–31.4)	28.8 (26.2–30.1)	25.0 (25.0–26.5) *
gm (mol·m−2·s−1)	25.0 (25.0–26.4) *	25.0 (25.0–26.4) *	32.7 (31.8–33.7)	34.3 (33.4–35.2)	35.1 (34.2–36.0)
iWUE (μmol CO2·mol−1 H2O)	25.0 (25.0–26.4) *	25.0 (25.0–26.4) *	36.2 (35.2–37.3)	36.7 (35.5–38.0)	36.5 (35.3–37.7)

Figure 2.

Temperature response of maximum carboxylation rate (V_cmax). Boxplots display the interquartile range (IQR, 25th–75th percentile, white boxes) and median line. Red diamond markers represent mean values of three replicates. Statistical significance among temperature treatments was assessed using one-way analysis of variance (ANOVA), with the significance threshold set at p < 0.05. Sample size for each temperature: n = 3. Different lowercase letters (a–d) above the boxes indicate significant differences among temperature treatments based on post-hoc multiple comparison tests; treatments sharing the same letter are not significantly different.

The temperature response of ratio of mesophyll to stomatal conductance (g_m/g_s) followed a similar pattern to that of g_m, except that the response curves under 500, 650, and 800 μmol mol⁻¹ CO₂ largely overlapped (Figure 3A). Ratio of chloroplastic to ambient CO₂ concentration (C_c/C_a) decreased with increasing CO₂ concentration. Below 30 °C, C_c/C_a remained relatively stable with temperature; it declined between 30 and 40 °C and increased again above 40 °C (Figure 3B). Under high CO₂ concentrations (500, 650, and 800 μmol mol⁻¹), ratio of intercellular to ambient CO₂ concentration (C_i/C_a) first decreased and then increased with temperature; under low CO₂ concentrations (150 and 300 μmol mol⁻¹), it showed a monotonic increasing trend with temperature (Figure 3C).

Figure 3.

Temperature responses of CO₂-diffusion-related ratios across a range of CO₂ concentrations (150, 300, 500, 650, and 800 μmol mol⁻¹), measured at a PAR of 1200 μmol m⁻² s⁻¹. (A) ratio of mesophyll to stomatal conductance (g_m/g_s); (B) ratio of chloroplastic to ambient CO₂ concentration (C_c/C_a); (C) ratio of intercellular to ambient CO₂ concentration (C_i/C_a). Solid lines are GAMM-fitted curves; solid dots are observed means for each temperature-C_a combination; vertical bars are SE (n = 3); different colors denote different CO₂ concentrations as indicated in the legend. All GAMM smooth terms were tested for significance, with the significance threshold was p < 0.05.

To further evaluate the effects of temperature, CO₂ concentration, and their interaction on photosynthetic traits, we conducted two-way ANOVA for each parameter. Temperature had significant effects on most parameters (p < 0.05), while CO₂ concentration significantly affected A_n, g_m, iWUE, g_m/g_s, C_c/C_a, and C_i/C_a (p < 0.05). Significant temperature × CO₂ interactions were observed for A_n, g_m, iWUE, g_m/g_s, C_c/C_a, and C_i/C_a (p < 0.05), indicating that the response of these traits to temperature was modulated by CO₂ concentration (Table S2).

Model diagnostics confirmed that the GAMM adequately described the temperature responses. Because a separate GAMM was fitted for each combination of photosynthetic parameter and CO₂ treatment, we present a representative example. The residual diagnostics for the GAMM fitted to A_n under Ca = 150 μmol mol⁻¹ are shown in Figure S2. No obvious systematic pattern was observed between residuals and fitted values, and the points in the Q-Q plot generally followed the diagonal line, indicating that the model residuals approximately satisfied the assumptions of normality and homoscedasticity.

2.2. Photosynthetic Limitation Analyses

Under low CO₂ concentrations (150 and 300 μmol mol⁻¹), stomatal limitation (l_s) remained relatively stable or declined slightly with increasing temperature, but decreased sharply when temperature exceeded 40 °C. Under high CO₂ concentrations (500, 650, and 800 μmol mol⁻¹), l_s first increased with temperature, leveled off around 35 °C, and then declined rapidly above 40 °C. Mesophyll limitation (l_m) increased consistently with rising temperature across all CO₂ treatments, reaching its maximum at 45 °C. Biochemical limitation (l_b) decreased gradually with increasing temperature. However, above 40 °C, l_b either leveled off or increased to varying degrees, with the most pronounced increase observed under low CO₂ concentrations (Figure 4).

Figure 4.

Temperature-driven shifts in the relative contributions of photosynthetic limitation components (l_b: biochemical limitation; l_m: mesophyll limitation; l_s: stomatal limitation) under different CO₂ concentrations at a PAR of 1200 μmol m⁻² s⁻¹. (A) Relative limitation proportions at 150 μmol mol⁻¹ CO₂; (B) at 300 μmol mol⁻¹ CO₂; (C) at 500 μmol mol⁻¹ CO₂; (D) at 650 μmol mol⁻¹ CO₂; (E) at 800 μmol mol⁻¹ CO₂. The y-axis represents the relative contribution (%) of each limitation component, with the sum of l_b + l_m + l_s = 100%. Vertical error bars represent standard errors (SE, n = 3).

2.3. Temperature Responses of Chlorophyll Fluorescence Parameters Under Different CO₂ Concentrations

Under different CO₂ concentrations, electron transport rate (J_a) initially increased and then decreased with rising temperature, peaking at 35–40 °C, with overall enhancement under elevated CO₂ (Figure 5A). Non-photochemical quenching (NPQ) generally declined with increasing temperature, reaching its minimum at 35–40 °C before rising again across all CO₂ concentrations (Figure 5B). The low-CO₂ treatment exhibited consistently higher NPQ levels, whereas the high-CO₂ treatment showed a relatively greater decline. Fraction of open PSII reaction center (q_L) continuously increased with rising temperature and reached its maximum at 40 °C (Figure 5C); at 45 °C, q_L decreased markedly across all CO₂ concentrations, while elevated CO₂ generally increased q_L. With rising temperature, maximum photochemical quantum efficiency of PSII (F_v/F_m) exhibited a continuous declining trend (Figure 5D): the decrease was relatively gradual from 25 to 35 °C, but accelerated notably after 35 °C, reaching its lowest value at 45 °C. It should be noted that F_v/F_m values at non−25 °C temperatures were estimated using published correction equations, as Minimum (F_o) and maximum (F_m) chlorophyll fluorescence were directly measured only at 25 °C. Furthermore, the 45 °C values were obtained after sequential exposure to lower temperatures; thus, the observed responses at this extreme temperature likely reflect cumulative heat stress rather than the direct effect of 45 °C per se.

Figure 5.

Temperature responses of chlorophyll fluorescence parameters under different CO₂ concentrations (150, 300, 500, 650, and 800 μmol mol⁻¹), measured at a PAR of 1200 μmol m⁻² s⁻¹. (A) Electron transport rate (Jₐ); (B) non-photochemical quenching (NPQ); (C) fraction of open PSII reaction center (q_L); (D) maximum photochemical quantum efficiency of PSII (F_v/F_m). Vertical error bars represent standard errors (SE, n = 3). Different colors denote different CO₂ concentrations as indicated in the legend. Statistical significance among temperature treatments was assessed using two-way analysis of variance (ANOVA), with the significance threshold was p < 0.05.

2.4. Multivariate Relationships Among Photosynthetic Physiological Traits

Correlation analysis revealed distinct associations among the measured photosynthetic traits (Figure 6). A_n was significantly correlated with J_a (R² = 0.69), iWUE (R² = 0.48), F_v/F_m (R² = 0.39), q_L (R² = 0.33), V_cmax (R² = 0.16), and g_m (R² = 0.13), indicating that photosynthetic performance was associated with variation in electron transport, intrinsic water use efficiency, photochemical characteristics, biochemical capacity, and mesophyll conductance. In contrast, A_n was not significantly correlated with g_s (R² = 0.00), l_b (R² = 0.00). The lack of a significant correlation between A_n and g_s may partly reflect the integration of all temperature treatments, because the temperature responses of stomatal conductance and photosynthetic assimilation were not necessarily synchronized across the full temperature range. Among the remaining traits, significant correlations were observed between q_L and both J_a (R² = 0.68) and V_cmax (R² = 0.59), between q_L and NPQ (R² = 0.54), between g_m and both F_v/F_m and l_m (R² = 0.48), and between J_a and V_cmax (R² = 0.52). The PCA further summarized the multivariate response patterns across temperature and CO₂ treatments (Figure 7). PC1 and PC2 explained 47.1% and 19.4% of the total variance, respectively, together accounting for 66.5% of the overall variation. The temperature treatments showed a clear gradient primarily along PC1, with the 45 °C treatment clearly separated from the 25–40 °C treatments, whereas the CO₂ treatments exhibited largely overlapping distributions (Figure 7A,B). This indicates that temperature, rather than CO₂ concentration, was the dominant axis of multivariate variation in photosynthetic traits. The loading plot showed that PC1 was mainly loaded by J_a, q_L, F_v/F_m, A_n, and iWUE, whereas NPQ loaded in the opposite direction (Figure 7C), consistent with the correlation patterns described above. These loadings indicate coordinated contributions of carbon assimilation, CO₂ diffusion, and photochemical parameters to the observed multivariate response.

Figure 6.

Correlation heatmap of photosynthetic parameters. Values within the boxes represent coefficients of determination (R²), and the color gradient indicates the strength of R², with higher R² values shown in darker colors, indicating stronger correlations. ***, **, and * denote significance at p < 0.001, p < 0.01, and p < 0.05, respectively. A_n: net photosynthetic rate (μmol m⁻² s⁻¹); g_s: stomatal conductance (mol m⁻² s⁻¹); g_m: mesophyll conductance (mol m⁻² s⁻¹); iWUE: intrinsic water use efficiency (μmol CO₂·mol⁻¹ H₂O); V_cmax: Maximum carboxylation rate of Rubisco (μmol m⁻² s⁻¹); J_a: Apparent electron transport rate (μmol m⁻² s⁻¹); NPQ: Non-photochemical quenching; q_L: Fraction of open PSII reaction centers; F_v/F_m: Maximum quantum efficiency of PSII; l_s: Stomatal limitation; l_m: Mesophyll limitation; l_b: Biochemical limitation.

Figure 7.

Principal component analysis (PCA) of photosynthetic and photochemical parameters across temperature and CO₂ treatments. (A) Score plot colored by temperature (25, 30, 35, 40, and 45 °C). (B) Score plot colored by CO₂ concentration (150, 300, 500, 650, and 800 μmol mol⁻¹). (C) Loading plot showing the contribution of each parameter to PC1 and PC2; the color gradient indicates each variable’s contribution (%) to the principal components, and arrows denote the direction and magnitude of the loadings. Ellipses in (A,B) represent 95% confidence regions for each group. PC1 and PC2 explained 47.1% and 19.4% of the total variance, respectively. A_n: net photosynthetic rate (μmol m⁻² s⁻¹); g_s: stomatal conductance (mol m⁻² s⁻¹); g_m: mesophyll conductance (mol m⁻² s⁻¹); iWUE: intrinsic water use efficiency (μmol CO₂·mol⁻¹ H₂O); V_cmax: Maximum carboxylation rate of Rubisco (μmol m⁻² s⁻¹); J_a: Apparent electron transport rate (μmol m⁻² s⁻¹); NPQ: Non-photochemical quenching; q_L: Fraction of open PSII reaction centers; F_v/F_m: Maximum quantum efficiency of PSII; l_s: Stomatal limitation; l_m: Mesophyll limitation; l_b: Biochemical limitation; C_c/C_a: ratio of chloroplastic to ambient CO₂ concentration; C_i/C_a: ratio of intercellular to ambient CO₂ concentration.

3. Discussion

3.1. Elevated CO₂ Divergently Shifts the Thermal Optima of g_m and g_s, Enhancing Photosynthetic Heat Tolerance While Improving Intrinsic Water Use Efficiency

Our results showed that elevated CO₂ significantly increased the T_opt of A_n and g_m, but decreased that of g_s, creating a divergent thermal response pattern in which carbon acquisition and water loss became thermally decoupled (Figure 1). The upward shift in the T_opt of g_m allowed higher CO₂ diffusion capacity to be maintained at elevated temperatures, increasing the CO₂ concentration at the chloroplast carboxylation sites. This increase enhanced substrate availability for Rubisco carboxylation while suppressing photorespiratory competition, thereby sustaining a higher carboxylation rate at elevated temperatures [4,5]. In contrast to the pronounced upward shift in the T_opt of g_m, the T_opt of V_cmax is expected to remain largely invariant with short-term CO₂ elevation. This contrasting pattern suggests that the enhanced thermal tolerance of photosynthesis under elevated CO₂ was driven primarily by improved mesophyll CO₂ diffusion capacity, rather than by changes in biochemical capacity per se. As the CO₂ diffusion capacity within the mesophyll, g_m determines the efficiency of CO₂ transport from intercellular airspaces to the chloroplast carboxylation sites, and its variation is governed by multiple factors including cell wall thickness, liquid-phase diffusion pathways, carbonic anhydrase activity, and aquaporin regulation [6,39,40]. Previous studies have shown that elevated CO₂ can induce adjustments in mesophyll anatomy and aquaporin expression, thereby enhancing CO₂ supply to chloroplasts [11,41]. Thus, the elevated T_opt of g_m observed in our study likely reflects an enhanced short-term thermal response of the mesophyll diffusion system under elevated CO₂. The close correspondence between the T_opt shifts of A_n and g_m further supports the interpretation that improved CO₂ supply to the chloroplasts, rather than changes in carboxylation capacity per se, drove the enhanced thermal tolerance of photosynthesis.

In stark contrast to the rightward shift of g_m, the T_opt of g_s shifted leftward under elevated CO₂ (Table 1). This finding is consistent with the classic optimal stomatal behavior theory [27], which posits that stomata are regulated to maximize the ratio of carbon assimilation to transpirational water loss. Under elevated CO₂, the steeper CO₂ gradient from the atmosphere to the leaf interior allows the same A_n to be achieved at a lower g_s [30,42]. When extended to temperature responses, this optimization framework predicts that the temperature at which g_s optimization occurs should also shift downward: under high CO₂, plants can achieve their optimal iWUE at lower temperatures, allowing them to close stomata earlier and conserve water while relying on enhanced g_m to compensate for reduced stomatal CO₂ supply [27,43,44,45]. Our results provide direct experimental support for this prediction, demonstrating that elevated CO₂ not only reduces the absolute magnitude of g_s but also lowers the thermal threshold for stomatal closure.

The opposing temperature responses of g_m and g_s observed here have important implications for leaf-level carbon-water balance. By shifting T_opt of g_m upward (enhancing CO₂ supply to chloroplasts at high temperatures) and T_opt of g_s downward (promoting earlier stomatal closure to conserve water), elevated CO₂ effectively decouples the temperature dependencies of carbon acquisition and water loss. The correlation analysis provided additional support for the coordinated variation among photosynthetic carbon assimilation, electron transport, and related physiological processes. A_n was significantly correlated with J_a, iWUE, F_v/F_m, q_L, V_cmax, and g_m, indicating that photosynthetic carbon assimilation was associated with changes in electron transport, intrinsic water-use efficiency, photochemical characteristics, biochemical capacity, and mesophyll conductance (Figure 6). In contrast, A_n was not significantly correlated with g_s or l_b. The absence of a significant correlation between A_n and g_s may partly reflect the integration of all temperature treatments, because the temperature responses of stomatal conductance and photosynthetic assimilation were not necessarily synchronized across the full temperature range. Other significant correlations were observed among q_L, J_a, V_cmax, and NPQ, as well as between g_m and F_v/F_m, l_m, and J_a and V_cmax. These relationships suggest that photosynthetic performance was associated with coordinated changes in CO₂ diffusion, biochemical capacity, electron transport, and photochemical regulation. These relationships are consistent with the temperature-response patterns observed in the GAMM analyses and limitation analysis, further supporting the interpretation that coordinated changes in CO₂ diffusion and biochemical processes contribute to the temperature dependence of photosynthesis. This decoupling resulted in a net enhancement of iWUE under elevated CO₂, consistent with previous studies that have reported improved iWUE under CO₂ enrichment [46,47]. Notably, this benefit was most pronounced at moderately elevated temperatures (30–40 °C), precisely the range where plants are most likely to experience heat stress under future climate scenarios.

3.2. Elevated CO₂ Fails to Alleviate Photosynthetic Breakdown at 45 °C Due to Mesophyll Diffusion and Photochemical Impairment

Although elevated CO₂ raised the thermal optima of both A_n and g_m, its mitigating effect on extreme heat stress was clearly limited. At 45 °C, we observed a rapid decline in l_s, whereas l_m emerged as the dominant constraint, accounting for approximately 70% of the total limitation (Figure 4). This interpretation is further supported by the C_c/C_a data (Figure 3B), which showed that the CO₂ concentration at the chloroplastic carboxylation site declined sharply at 45 °C, even under elevated CO₂, indicating that CO₂ transport to Rubisco was severely impeded regardless of external CO₂ supply. These findings are consistent with previous studies showing that under severe heat stress, the capacity for internal CO₂ diffusion toward chloroplasts becomes the key constraining process for photosynthesis [18], and they extend these earlier observations by demonstrating that this shift from stomatal to mesophyll limitation occurs across all CO₂ conditions. The underlying mechanisms are well documented: extreme high temperatures can reduce g_m by destabilizing mesophyll cell membranes, decreasing aquaporin activity, and compromising chloroplast structural integrity, all of which restrict CO₂ access to carboxylation sites [6,18,48]; in addition, the activity of carbonic anhydrase, which facilitates CO₂ transport across the mesophyll, is temperature-sensitive and may decline under extreme heat. These temperature-induced changes in mesophyll function were severe enough to override any potential benefit from elevated CO₂, indicating that CO₂ enrichment cannot prevent the fundamental breakdown of mesophyll diffusion at extreme temperatures.

The sharp decline in photosynthetic capacity at 45 °C was accompanied by a pronounced decrease in estimated F_v/F_m (Figure 5D), suggesting possible PSII impairment. This apparent photochemical limitation, together with reduced q_L and elevated NPQ, implies that the photosynthetic apparatus was functionally constrained at extreme temperatures. The failure of CO₂ enrichment to sustain photosynthesis at 45 °C thus likely reflects photochemical capacity limitations rather than CO₂ supply alone. However, because the F_v/F_m values at non-25 °C temperatures were derived from published equations rather than measured directly in this study, these interpretations remain indicative and await direct experimental confirmation under the actual temperature regimes.

One notable exception was the increase in g_s when temperature exceeded 40 °C (Figure 1C). This anomalous rise, observed across all CO₂ treatments, likely reflects stomatal dysfunction under extreme heat rather than an adaptive response, as membrane destabilization and impaired guard cell metabolism may lead to passive stomatal opening [49]. Alternatively, the elevated VPD at 40–45 °C may have driven passive hydraulic opening, as stomata lose the ability to maintain turgor-driven closure under extreme evaporative demand [50]. Although relative humidity was held constant in our experimental design, the concomitant rise in VPD with increasing temperature (from 1.37 kPa at 25 °C to 3.32 kPa at 45 °C; Table S1) may have contributed to the observed decline in g_s, particularly at temperatures above 35 °C. Thus, the temperature response of g_s reported here reflects the combined effects of temperature and VPD, rather than temperature alone. This interpretation is supported by the concurrent decline in A_n and g_m at 45 °C, indicating that the g_s increase did not translate into enhanced photosynthesis [51].

Nevertheless, it is important to acknowledge that the 45 °C responses presented here may have been influenced by the cumulative thermal history of the leaves. Because measurements were conducted following sequential exposure to 25, 30, 35, and 40 °C, the observed decline at 45 °C could partly reflect progressive heat damage that had already been initiated at lower temperatures, rather than the instantaneous effect of 45 °C alone. This stepwise warming regime, however, realistically mimics natural heatwave events, during which plants experience gradual temperature increases over consecutive days rather than abrupt thermal shocks.

In addition, we acknowledge that anatomical and morphological traits, such as stomatal density and aperture, leaf area, leaf thickness, and specific leaf area, would provide valuable complementary evidence for the mechanistic understanding of plant responses to elevated CO₂ and temperature. However, given that the present experiment was short-term in nature, such structural traits were unlikely to change appreciably during the measurement period, and they were therefore not assessed in this study. Future long-term temperature response experiments incorporating these traits would further strengthen our understanding of the structural underpinnings of photosynthetic acclimation under elevated CO₂ and temperature.

4. Materials and Methods

4.1. Plant Species and Growth Sites

A controlled-environment experiment was conducted at Oak Ridge National Laboratory (ORNL, Oak Ridge, TN, USA; 35°54′ N, 84°20′ W), using Liriodendron tulipifera L. as the material. Two-year-old bare-root saplings were initially established in 6.2 L containers filled with a standard commercial substrate (Sun Gro Horticulture, Agawam, MA, USA) and cultivated outdoors under naturally varying light, temperature, and humidity conditions. Following a ten-month establishment period, saplings exhibiting uniform growth and vigor were selected and repotted into 35-L containers with the identical growth medium. Prior to measurements, all plants underwent a 30-day acclimation period inside a programmable walk-in growth chamber (Model BDW80, Conviron, Winnipeg, MB, Canada). Plants received regular irrigation and were supplemented with a complete 20-10-20 NPK fertilizer (Southern AG, Palmetto, FL, USA) at routine intervals. All experimental measurements were performed in July 2017 across five temperature levels: 25, 30, 35, 40, and 45 °C. The same fully expanded leaves were used for measurements across all five temperature levels. At each target temperature, the chamber was adjusted accordingly, and leaves were allowed a minimum equilibration period of two hours before any recordings were taken. Upon completion of measurements at a particular temperature, the chamber environment was restored to the acclimation baseline. The five temperature treatments were conducted in the order 25, 30, 35, 40, and 45 °C on separate days, allowing plants to recover overnight between successive temperature exposures. Recovery was confirmed by checking that the steady-state photosynthesis returned to its baseline level each morning before the next day’s measurements, ensuring that the photosynthetic apparatus had recovered from the previous day’s heat exposure prior to each new temperature treatment.

4.2. Concurrent Measurements of Gas Exchange and Chlorophyll Fluorescence

Gas-exchange and chlorophyll fluorescence measurements were conducted using a LI-6800 portable photosynthesis system (LI-COR Inc., Lincoln, NE, USA) equipped with a multiphase flash fluorometer chamber (6800-01A, LI-COR Inc., Lincoln, NE, USA). Measurements were performed on three individual saplings per treatment, with one fully expanded leaf selected from each sapling, serving as three biological replicates. RH was held constant at 60% with a flow rate of 500 mL min⁻¹. Prior to CO₂-response curve measurements, leaves were light-adapted at the saturating light intensity until steady-state photosynthesis was achieved, a process typically requiring about 20 min. CO₂-response curves were subsequently measured under light intensity of 1200 μmol m⁻² s⁻¹ across 12 CO₂ concentrations (400, 300, 200, 150, 100, 50, 500, 650, 800, 1000, 1200, and 1500 μmol mol⁻¹). The time interval between measurements at different CO₂ concentrations was restricted to 120–240 s. Gas exchange parameters (e.g., A_n and g_s), light-adapted stead-state (F_s) and maximum ($F_{\mathrm{m}}^{\prime }$) chlorophyll fluorescence was obtained from CO₂ response curves for the same leaves. Stomatal conductance to CO₂ (g_s) was calculated from water vapor conductance (g_sw) using the conversion factor 1.56 (g_s = g_sw/1.56).

For dark-adapted chlorophyll fluorescence parameters, plants were placed in complete darkness overnight by extinguishing the growth chamber light source prior to measurements. F_o and F_m were determined only at 25 °C. F_o and F_m at other temperatures (30, 35, 40, and 45 °C) were estimated based on the temperature responses of F_o and F_m reported by Pospíšil et al. (1998) [52], which covered all target temperatures (20–45 °C). Note that F_o and F_m were measured only at 400 μmol mol⁻¹, not across the range of CO₂ concentrations.

4.3. Derivation of Chlorophyll Fluorescence Parameters

The actual PSII photochemical efficiency (Φ_PSII) was determined following the method of Genty et al. (1989) [53]:

$${\Phi }_{\mathrm{P}\mathrm{S}\mathrm{I}\mathrm{I}}={\displaystyle \frac{F_m^{\prime }-F_s}{F_m^{\prime }}},$$ (1)

$F_{\mathrm{m}}^{\prime }$ is Maximum fluorescence in the light-adapted state, F_s is Steady-state fluorescence in the light-adapted state.

NPQ was determined according to Bilger and Björkman (1990) [54]:

$$\mathrm{N}\mathrm{P}\mathrm{Q}={\displaystyle \frac{F_m-F_m^{\prime }}{F_m^{\prime }}},$$ (2)

The quantum yields of regulated (Φ_NPQ) and non-regulated (Φ_f,D) energy dissipation in PSII were calculated following Hendrickson et al. (2004) [55]:

$${\Phi }_{\mathrm{N}\mathrm{P}\mathrm{Q}}={\displaystyle \frac{F_s}{{\mathrm{F}}_{\mathrm{m}}^{\prime }}}-{\displaystyle \frac{F_s}{F_m}},$$ (3)

$${\Phi }_{\mathrm{f},\mathrm{D}}={\displaystyle \frac{F_s}{F_m}},$$ (4)

q_L was calculated according to Kramer et al. (2004) [56]:

$$q_L={\displaystyle \frac{F_m^{\prime }-F_s}{F_m^{\prime }-F_o^{\prime }}}\times {\displaystyle \frac{F_o^{\prime }}{F_s}},$$ (5)

$F_o^{\prime }$ is Minimum fluorescence in the light-adapted state.

J_a was calculated as:

$$J_a={\Phi }_{\mathrm{P}\mathrm{S}\mathrm{I}\mathrm{I}}\times \mathrm{P}\mathrm{A}\mathrm{R}\times \alpha \times \beta ,$$ (6)

where PAR denotes the photosynthetically active radiation, α is the leaf absorptance coefficient, and β is the partition coefficient of absorbed light energy between PSI and PSII. In this study, values of α = 0.84 [57,58] and β = 0.5 [59] were used consistently for all treatments. We acknowledge that the use of constant α and β values across the entire temperature range represents a pragmatic simplification. However, given the lack of direct, environment-specific calibrations for each treatment, this approach provides a necessary and widely accepted first-order estimate.

4.4. Estimation of Mesophyll Conductance with Variable J Method

g_m was estimated using the Variable J method described by Harley et al. (1992) [60]:

$$g_m={\displaystyle \frac{A_n}{C_i-\frac{{\mathsf{\Gamma }}^{\ast }(J_a+8\left(A_n+R_d\right))}{J_a-4(A_n+R_d)}}},$$ (7)

where C_i is the intercellular CO₂ concentration, R_d is half of the measured dark respiration rate (R_n, R_d = R_n/2) [61,62,63]. R_n was measured after dark adaptation using the LI-6800 system at 25 °C. The temperature dependence of R_d was corrected according to Bernacchi et al. (2001) [64]:

$$R_d=\mathrm{E}xp\left(18.72-{\displaystyle \frac{46390}{8.314\ast \left(273.15+T_{leaf}\right)}}\right)$$

${\Gamma }^{\ast }$ is the CO₂ compensation point in the absence of mitochondrial respiration. The temperature dependence of ${\Gamma }^{\ast }$ was corrected according to Bernacchi et al. (2002) [18]:

$${\Gamma }^{\ast }=\mathrm{E}xp\left(13.49-{\displaystyle \frac{24460}{8.314\ast \left(273.15+T_{leaf}\right)}}\right),$$ (8)

where T_leaf is the leaf temperature (°C).

The CO₂ concentration at the chloroplast carboxylation site (C_c) was calculated as:

$$C_c=C_i-{\displaystyle \frac{A_n}{g_m}},$$ (9)

Following the criteria of Harley et al. (1992) [60], we verified that all dC_c/dA_n values remained between 10 and 50, which lends confidence to the validity of our derived g_m.

4.5. Estimation of Mesophyll Conductance and Maximum Carboxylation Rate Using Curve-Fitting Approach

We estimated g_m and V_cmax using the curve-fitting method described by Sharkey et al. (2007) [65] and Sharkey (2016) [66], which exploits the curvature of the A_n-C_i response curve induced by finite g_m. The parameters were obtained by non-linear regression minimizing the sum of squared residuals. To validate the reliability of the estimated g_m values, we compared the variable J method against the A_n/C_i curve-fitting approach. Both methods were applied to the same dataset, and the results were highly consistent (R² = 0.834, p < 0.001; Figure S1). Accordingly, only the g_m values derived from the variable J method are reported herein.

4.6. Relative Limitation Analyses

Relative limitation analysis includes l_s, l_m, and l_b. In this study, we followed the approach of Grassi and Magnani (2005) [67] to calculate these components using the following equations:

$$l_s={\displaystyle \frac{\frac{g_{tot}}{g_s}\times \frac{\partial A_n}{\partial C_c}}{g_{\mathit{tot}}+\frac{\partial A_n}{\partial C_c}}}$$ (10)

$$l_m={\displaystyle \frac{\frac{g_{tot}}{g_m}\times \frac{\partial A_n}{\partial C_c}}{g_{tot}+\frac{\partial A_n}{\partial C_c}}}$$ (11)

$$l_b={\displaystyle \frac{g_{tot}}{g_{tot}+\frac{\partial A_n}{\partial C_c}}}$$ (12)

where g_tot is the total CO₂ conductance from ambient air to the chloroplast (1/g_tot = 1/g_s + 1/g_m). The slope ∂A_n/∂C_c was calculated as the first derivative of the A_n/C_c curve over the C_c range of 50–100 μmol mol⁻¹ for each temperature-CO₂ combination using linear regression [68].

4.7. Data Analyses and Statistics

All statistical analyses and graphics were performed in R version 4.4.2. For graphical presentation, five representative CO₂ concentrations (150, 300, 500, 650, and 800 μmol mol⁻¹) were selected from the full measurement series to clearly illustrate the temperature response trends across sub-ambient to elevated CO₂ levels. For each treatment, means and standard errors (SE) were calculated from three biological replicates, each from an individual sapling.

Generalized additive mixed models (GAMM) were fitted to the response curves of photosynthetic parameters to temperature or CO₂ concentration, from which T_opt for A_n, iWUE, g_s, and g_m were derived. GAMM were fitted using penalized thin-plate regression splines (bs = “tp”) with k = 4. To account for the within-individual correlation arising from repeated measurements on the same leaf across temperatures and CO₂ treatments, replicate identity (Rep) was included as a random intercept. The models were fitted using restricted maximum likelihood (REML), and gamma = 0.5 was used to increase the penalty on model complexity and reduce the risk of overfitting. Model diagnostics were conducted for each GAMM, including scatterplots of residuals versus fitted values and Q-Q plots to evaluate residual distribution and variance structure. Residual normality was assessed using the Shapiro–Wilk test, and no substantial deviations were found. For each fitted GAMM, we predicted responses over a fine temperature grid and defined T_opt as the temperature yielding the maximum predicted value. Because g_s increased again above 40 °C owing to stomatal dysfunction rather than an adaptive response, the T_opt of g_s was estimated over the 25–40 °C range; for A_n, iWUE, and g_m, T_opt was estimated over the full 25–45 °C range. Uncertainty in T_opt was estimated by parametric bootstrap. For each fitted GAMM, we simulated 1000 sets of response values, refitted the model to each simulated dataset, and re-estimated T_opt from the fitted curve. The 2.5th and 97.5th percentiles of the resulting T_opt distribution were taken as the 95% confidence interval.

Subsequent Duncan’s multiple range tests (p < 0.05) were performed to identify specific differences among treatments, including (i) comparisons among CO₂ concentrations within each temperature level, and (ii) comparisons among temperatures within each CO₂ concentration level. The detailed results, with significance letter assignments for each parameter, are presented in Tables S3 and S4, respectively. To evaluate the effects of temperature, CO₂ concentration, and their interaction on photosynthetic traits, two-way ANOVA was performed for each parameter, with temperature and CO₂ concentration as fixed factors.

To further characterize the relationships among physiological traits, correlation analysis was conducted using the measured photosynthetic and chlorophyll fluorescence parameters. Principal component analysis (PCA) was performed to summarize the multivariate variation across temperature and CO₂ treatments.

5. Conclusions and Implications

Our findings reveal that elevated CO₂ shifts the thermal optima of g_m and g_s in opposite directions, decoupling carbon acquisition from water loss. Under elevated CO₂, plants may close stomata earlier to save water while relying on enhanced mesophyll diffusion to sustain carbon gain below 40 °C, but this strategy fails when g_m collapses under extreme heat. The shift from stomatal to mesophyll limitation at 45 °C, along with apparent photochemical limitation, implies that current terrestrial biosphere models may overestimate carbon uptake during heatwaves. Many models fail to account for the CO₂ sensitivity of stomatal conductance parameters, which can lead to significant overestimates of evapotranspiration and carbon uptake under elevated CO₂ [69]. Moreover, despite being recognized as a key limiting factor in photosynthesis, g_m is scarcely considered in climate change research [70]. Models should explicitly account for their separate temperature dependencies, including the nonlinear decline of g_m at high temperatures and temperature-induced declines in electron transport. The distinct responses of g_m and g_s also help resolve uncertainties in photosynthetic acclimation, which may involve different mechanisms under ambient versus elevated CO₂. Finally, the sharp F_v/F_m decline at 45 °C underscores a fundamental limit to CO₂ fertilization: once photochemistry is impaired, extra CO₂ cannot restore carbon gain. Given projections of more frequent heatwaves [33,34], the benefits of elevated CO₂ may be transient under extreme conditions. Future work should prioritize identifying traits for thermotolerance in photochemistry and mesophyll diffusion and explore acclimation potential in forest trees.

Abbreviations

The following abbreviations are used in this manuscript:

An	Net photosynthetic rate
iWUE	Intrinsic water use efficiency
gs	Stomatal conductance
gm	Mesophyll conductance
Topt	Optimal temperature
Vcmax	Maximum carboxylation rate
lm	Mesophyll limitation
ls	Stomatal limitation
lb	Biochemical limitation
Fv/Fm	Photochemical efficiency
CA	Carbonic anhydrase
VPD	Vapor pressure deficit
ΦPSII	PSII photochemical efficiency
NPQ	Non-photochemical quenching
ΦNPQ	Quantum yields of regulated
Φf,D	non-regulated energy dissipation in PSII
qL	Fraction of open reaction centers
Ja	Electron transport rate
Ci	Intercellular CO2 concentration
Rn	Net photosynthetic rate
Rd	Day respiration rate
Cc	Chloroplast carboxylation site
Ca	Ambient CO2 concentrations
PAR	Photosynthetically Active Radiation
gsc	Stomatal conductance to CO2
Fo	Minimum chlorophyll fluorescence
Fm	Maximum chlorophyll fluorescence
$F_m^{\prime }$	Maximum fluorescence in the light-adapted state
Fs	Steady-state fluorescence in the light-adapted state
$F_o^{\prime }$	Minimum fluorescence in the light-adapted state
Tleaf	Leaf temperature
Γ*	CO2 compensation point in the absence of mitochondrial respiration
gm/gs	Ratio of mesophyll to stomatal conductance
Cc/Ca	Ratio of chloroplastic to ambient CO2 concentration
Ci/Ca	Ratio of intercellular to ambient CO2 concentration

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/plants15182889/s1, Figure S1: Correlation analysis between g_m Sharkey and g_m Harley; Table S1: Temperature-dependent changes in vapor pressure deficit (VPD) under the experimental conditions (RH = 60%); Figure S2: Diagnostic plots for the GAMM fitted to the relationship between net photosynthetic rate (A_n) and temperature under 150 μmol mol⁻¹ CO₂; Table S2: Significance analysis of temperature × CO₂ (C_a) interaction; Table S3: Significance letters for each parameter under different C_a concentrations at the same temperature, based on Duncan’s multiple range test; Table S4: Significance letters for each parameter under different temperature concentrations at the same C_a, based on Duncan’s multiple range test.

Author Contributions

R.L. and K.J. were responsible for data analysis and writing the manuscript. J.H. was responsible for design and performance of the research, data analysis and writing the manuscript. L.L. edited the manuscript. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

The data underlying this article will be shared upon reasonable request to the corresponding author.

Conflicts of Interest

The authors declare no conflicts of interest.

Funding Statement

This research was supported by the National Natural Science Foundation of China (Grant No. 32401559), the Natural Science Foundation of Jiangsu Province (Grant No. BK20240669), Jiangsu Young Scientific and Technological Talent Support Program (Grant No. JSTJ-2025-099), and the scientific research foundation for advanced talents from Nanjing Forestry University. J.H. acknowledges the China Scholarship Council (CSC) for supporting her visit to Oak Ridge National Laboratory, where the data presented in this study were collected.