This study was conducted at the Dinghushan Biosphere Reserve (DBR, 23°09’N–23°11’N, 112°30’E–112°33’E) in southern China, which exhibits a typical tropical monsoon climate. The mean annual temperature is approximately 21°C, and the relative humidity averages 80% throughout the year. The mean annual precipitation is approximately 1900 mm; nearly 80% of the rainfall occurs during the wet season (April–September) and 20% during the dry season (October–March) (Liu et al., 2017; Wu et al., 2020a). There are three major tropical forest types in DBR, including the monsoon evergreen broadleaf forest (MEBMF, 30 m altitude above sea level), coniferous and broadleaf mixed forest (CBMF, 300 m altitude above sea level) and mountain evergreen broadleaf forest (MEBF, 600 m altitude above sea level). During 1954–2009, the mean annual temperature at the DBR has increased by approximately 1°C (Zhou et al., 2011).

The net assimilation rate-intercellular CO₂ concentration (A-C_i) curves were generated on fully expanded leaves from three or four seedlings per chamber per species (n = 3–4) between 0900 and 1400 h on clear days in August 2019 and 2020 (last 2 years), using a portable open path gas exchange system (Licor-6400, Li-Cor, Lincoln, NE, United States) equipped with a leaf chamber fluorometer (6400-40). A-C_i curves were generated at saturating PAR of 1500 μmol m^–2 s^–1, leaf temperature (25°C) and ambient relative humidity inside the leaf chamber (∼60%). A-C_i curves were generated using leaf chamber CO₂ values of (in order): 400, 200, 100, 50, 400, 400, 600, 800, 1200, 1400, 1600, and 1800 μmol mol^–1. The maximum rate of photosynthetic Rubisco carboxylation (V_max, μmol m^–2 s^–1) and apparent maximum rate of photosynthetic electron transport (J_max, μmol m^–2 s^–1) were estimated from A-C_i curves.

Leaf light-saturated photosynthesis (A_sat, μmol m^–2 s^–1), stomatal conductance (g_s, mol m^–2 s^–1) and transpiration rate (E, mmol m^–2 s^–1) were measured in August 2018, 2019 and 2020 (last 3 years), at saturating PAR of 1500 μmol m^–2 s^–1, CO₂ of 400 μmol mol^–1 and leaf temperature (25 °C). Instantaneous water-use efficiency (iWUE, μmol m mol^–1) was calculated as A_sat/g_s.

For gas exchange measurements, opposite leaves were harvested for chemical analyses in August 2019 and 2020 (last 2 years). Five leaf discs (6 mm diameter) from one individual per treatment per chamber were cut and dipped into 5 mL of 80% acetone to determine the chlorophyll concentrations [Chl_(a+b), μg cm^–2] (Arnon, 1949).

Some leaves were used to determine leaf area (LA, cm²), with petioles removed, using a leaf area meter (Li-3000A; Li-Cor, Lincoln, NE, United States). The leaves were oven-dried for one week at 65°C, and dry biomass (LD, g) was determined. Leaf mass area (LMA, g cm^–2) was calculated as LD/LA. Oven-dried foliar samples were ground to fine powder in a ball mill. Leaf mass-based nitrogen (N) concentration (N_mass, g kg^–1) was measured using the Kjeldahl method (Bremner and Mulvaney, 1982). Leaf mass-based phosphorus (P) concentration (P_mass, g kg^–1) was measured photometrically after samples were digested with H₂SO₄-H₂O₂ (Anderson and Ingram, 1989). Leaf area-based N concentration (N_a, mg cm^–2) was calculated as N_mass × LMA, and leaf area-based P concentration (P_a, mg cm^–2) was calculated as P_mass × LMA.

Three leaves of each species from randomly selected individuals per chamber were measured for stomatal anatomical traits in June 2019 and 2020 (last 2 years), using image analysis software (OPTPro 2012 4.0, Optec XTS20, Chongqing Optec Instrument, China). Stomatal density (SD), guard cell width (W, μm) and length (L, μm) were also measured using the image analysis software. Stomatal size (SS, μm²) was defined as L * 2 * W (Franks and Beerling, 2009; Wu et al., 2018).

We collected soils (0–10 cm) in the wet season (June) from 2015 to 2019. Soil cores (inner diameter: 3.5 cm) from each layer of each chamber were combined, homogenized. There were three replicates per treatments for each forest type. A soil subsample was air-dried before determination of pH, total N (TN, g kg^–1), total P (TP, g kg^–1), soil available N concentration (AN, mg kg^–1) and soil available P concentration (AP, mg kg^–1). Soil pH was measured at a soil-to-water ratio of 1:2.5. After digestion with concentrated sulfuric acid, TN was determined by the Kjeldahl method (Bremner and Mulvaney, 1982) and TP by an ultraviolet spectrophotometer (Lambda650/250, PE, United States) (Anderson and Ingram, 1989). After soil was extracted with 1 M KCl solution, AN was determined colorimetrically (Liu et al., 2017). AP was extracted with 0.03 M NH₄F and 0.025 M HCl and measured by inductively coupled plasma optical emission spectrometer (ICP-OES) (Optima 2000 DV, Perkin Elmer, United States) (Bray and Kurtz, 1945).

The stem basal diameter (D, mm) and height (H, cm) of all seedlings per species in each chamber were measured in June (wet season) and December (dry season) from 2015 to 2019. The stem volume (V, cm³) of all seedlings per species in each chamber was also calculated as 3.14 * (D/2) ^ 2 * H.

Data were assessed using Kolmogorov-Smironv test for normality and Levene’s test for homogeneity of variance prior to statistical analysis. When the data did not conform to the assumption of normality and homogeneity of variances, they were logarithmically transformed. A one-way ANOVA was used to evaluate warming effect on T_air, T_soil, RH, SVWC, VPD, PAR and rainfall in the wet and dry seasons in the three forest types. Significant differences in foliar morphological, physiological traits and soil properties among the treatments were analyzed using a one-way ANOVA, followed by Tukey’s multiple comparison test. A repeated measures general liner model was used to evaluate the effects of warming and time on D²H for the four tree species. Data were analyzed using SPSS 24.0 (SPSS Inc., Chicago, IL, United States). The mixed liner model was used to analyze the main effects of environmental variables on the responses of photosynthetic capacity (J_max and V_max), gas exchange (A_sat, g_s, E and iWUE), biochemistry (concentrations of Chl_(a+b), N_a and P_a) and stomatal morphology (SD and SS). Differences were considered to be statistically significant at P < 0.05. Data were analyzed using R (R 3.5.3). We calculated the response ratio (RR) of AN and AP with warming (mean value measured in the warmed environment divided by the mean value measured in the control environment).

The relative importance of stomatal traits (g_s, SD and SS), biochemistry [concentrations of Chl_(a+b), N_a and P_a], concentrations of soil nutrient availability (AN and AP) and soil types (lateritic soil and yellow soil) in explaining variations in photosynthesis (J_max, V_max and A_sat), was determined using variation partitioning analysis (VPA) and redundancy analysis (RDA) using the “varpart” function in the “Vegan” package in R (R 3.5.3).

Under translocation warming, T_air in CBMF at 30 m was 0.86 ± 0.07°C (wet season), 0.52 ± 0.02°C (dry season) and 0.68 ± 0.05°C (annual) higher compared to temperatures at 300 m, but they were not significantly different (Figure 2A). T_soil in CBMF at 30 m was 1.35 ± 0.01°C (wet season) higher compared to temperature at 300 m, and although it was higher (0.68 ± 0.01°C) in the dry season, it was not significantly different (Figure 2B). T_air in MEBF at 300 m was 1.01 ± 0.01°C (wet season) and 0.95 ± 0.06°C (annual) higher, and at 30 m was 1.87 ± 0.08°C (wet season) and 1.63 ± 0.08°C (annual) higher compared to temperatures at 600 m. T_air trended higher at 300 m (0.78 ± 0.10°C) and 30 m (1.3 ± 0.08°C) compared to temperature at 600 m in the dry season, but it was not significantly different (Figure 2C). T_soil in MEBF at 300 m was 0.78 ± 0.01°C (wet season) and 1.46 ± 0.01°C (dry season) higher, and at 30 m was 2.22 ± 0.01°C (wet season) and 2.50 ± 0.01°C (dry season) higher compared to temperatures at 600 m (Figure 2D). In MEBF and CBMF, the average monthly soil SVWC was significantly lower at 30 m compared to 600 and 300 m, in the wet and dry seasons (P < 0.05, Table 1).

During the last 5 years of the IR warming experiment, T_air in MEBMF in the warming treatment was 1.59 ± 0.01°C (wet season), 2.12 ± 0.05°C (dry season) and 1.82 ± 0.03°C (annual) higher compared to temperatures in the control plots (Figure 2E). T_soil in the IR warming treatment was 1.06 ± 0.02°C (dry season) higher compared to temperature in the control plots, and although it trended higher (0.38 ± 0.01°C) in the wet season, it was not significantly different (Figure 2F). Infra-red warming treatments significantly reduced the average monthly air RH, and increased VPD, of MEBMF in the wet and dry seasons (P < 0.05, Table 1).

In CBMF, 0.68°C of translocation warming induced higher g_s for S. superba, but no effects on stomatal density and smaller stomatal size for M. breviflora and S. rehderianum (Figures 3A–C). The 0.68°C translocation warming treatment reduced transpiration of M. breviflora but enhanced that of S. superba (Figure 3D). Declining iWUE was observed in S. rehderianum and S. superba (Figure 3E). In MEBF, g_s of M. breviflora was increased by 0.95°C of translocation warming, but that was reduced by 1.63°C of translocation warming. S. rehderianum and S. superba had higher g_s under 0.95°C and 1.63°C of translocation warming (Figure 3F). Warming did not affect stomatal density for the three tree species (Figure 3G). Stomatal size of S. rehderianum was decreased by 1.63°C translocation warming (Figure 3H). Transpiration of M. breviflora was increased by 0.95°C translocation warming, but reduced by 1.63°C translocation warming. S. rehderianum and S. superba had higher transpiration under 0.95°C and 1.63°C translocation warming (Figure 3I). iWUE for the three tree species were not affected by warming (Figure 3J).

In MEBMF, g_s of M. breviflora, S. rehderianum and S. superba were reduced by the 1.82°C IR warming treatment (Figure 3K). Schima superba exhibited reduced stomatal density under 1.82°C of IR warming (Figure 3L). The 1.82°C IR warming treatment enhanced stomatal size for M. breviflora, but decreased that of S. rehderianum (Figure 3M). Decreasing transpiration had been observed in M. breviflora and S. rehderianum under the 1.82°C IR warming treatment (Figure 3N), but iWUE of M. breviflora was enhanced by 1.82°C of IR warming (Figure 3O).

In CBMF, S. rehderianum had higher Chl_(a+b) concentration in the 0.68°C translocation warming treatment (Figure 4A). N_a for the three tree species were not affected by translocation warming (Figure 4B). P_a of S. superba was enhanced by 0.68°C of translocation warming (Figure 4C). In MEBF, the concentrations in Chl_(a+b), N_a and P_a for M. breviflora, S. rehderianum and S. superba were not affected by warming (Figures 4D–F). In MEBMF, warming exerted no effects on the concentrations in Chl_(a+b) and N_a for the three tree species (Figures 4G,H). In MEBMF, P_a of S. superba was decreased by 1.82°C of IR warming (Figure 4I).

In CBMF, J_max and V_max of S. superba in the 0.68°C translocation warming treatment were significantly increased by 106.2% and 43.6%, respectively (Figures 5A,B). The 0.68 °C translocation warming treatment did not affect J_max/V_max for M. breviflora, S. rehderianum and S. superba (Figure 5C). The 0.68°C translocation warming treatment decreased A_sat of M. breviflora by 26.9%, but increased that of S. superba by 33.4% (Figure 5D). In MEBF, J_max for the three tree species were not influenced by warming (Figure 5E). V_max of S. superba under the 1.63°C translocation warming treatment was enhanced by 56.1% (Figure 5F). The 1.63°C translocation warming treatment decreased J_max/V_max for S. superba (Figure 5G). The 0.95°C translocation warming treatment enhanced A_sat of M. breviflora by 43.9%, but the 1.63°C translocation warming treatment decreased A_sat by 17.4%. The 0.95°C and 1.63°C translocation warming treatments enhanced A_sat of S. rehderianum and S. superba by approximately 40.0% (Figure 5H).

In MEBMF, 1.82°C of IR warming significantly reduced J_max and V_max of M. breviflora by 34.4% and 26.1%, respectively (Figures 5I,J). Lower J_max/V_max for S. superba was observed in MEBMF (Figure 5K). A_sat of M. breviflora, S. rehderianum and S. superba in the 1.82°C IR warming treatment were decreased by 36.3%, 23.1%, and 23.8%, respectively (Figure 5L).

The soil type of CBMF and MEBMF was lateritic soil, and that of MEBF was yellow soil. The concentrations of TN and TP, and pH for CBMF, MEBF and MEBMF were not significantly influenced by warming (Table 2). In CBMF, AN was significantly increased by 0.68°C of translocation warming, while AP showed decreasing trends to warming (Supplementary Figure 1A). In contrast, the 0.95°C and 1.63°C translocation warming treatments in MEBF significantly decreased AN but had no significant effects on AP (Supplementary Figure 1B). In MEBMF, AN and AP were not significantly affected by IR warming (Supplementary Figure 1C).

In the translocation warming treatment, lower A_sat for M. breviflora was mainly driven by higher T_soil and lower SVWC. Stomatal size of M. breviflora was significantly decreased by higher T_air. Higher T_air significantly increased g_s for S. rehderianum, but decreased its stomatal size. The enhancements in photosynthetic capacity and gas exchange for S. superba were significantly driven by higher T_air and T_soil (Table 3, P < 0.05).

In the 1.82°C IR warming treatment, lower RH significantly decreased gas exchange for M. breviflora. Stomatal size of M. breviflora was significantly increased by higher T_air and lower RH. g_s of S. rehderianum and S. superba, and stomatal size of S. rehderianum were mainly reduced by increasing T_air. Transpiration of S. rehderianum was decreased by higher T_air and VPD. The reduction in P_a for S. superba was driven by T_air, RH and VPD. Stomatal density of S. superba was significantly decreased by declining RH and increasing VPD (Table 4, P < 0.05).

The change in photosynthesis trait for M. breviflora between CBMF and MEBF was mainly correlated with soil type, stomatal traits and nutrient availability, and their ratios were 43%, 26%, and 16%, respectively (Supplementary Figure 2A). The soil type explained the 10% changes in photosynthesis trait for S. rehderianum between CBMF and MEBF (Supplementary Figure 2B). Biochemistry was main factor inducing the shift in photosynthesis trait for S. superba between CBMF and MEBF (Supplementary Figure 2C).

In CBMF, the 0.68°C translocation warming treatment reduced stem volume of M. breviflora and S. rehderianum by 6.2% and 6.3%, respectively (Figures 6A,B). The stem volume of S. superba was increased by 117.8% (Figure 6C). In MEBF, the 1.63°C translocation warming treatment significantly decreased the stem volume of M. breviflora by 54.9% (Figure 6D). The 0.95°C and 1.63°C translocation warming treatments significantly increased the stem volume of S. rehderianum in MEBF by 26.5% and 106.5%, respectively (Figure 6E). The 0.95°C and 1.63°C translocation warming treatment significantly enhanced stem volume in MEBF by 33.5% and 119.7% (Figure 6F).

In MEBMF, the 1.82°C IR warming treatment reduced stem volume of S. rehderianum by 42.8% (Figure 6H). The 1.82°C IR warming treatment had no significant effect on the stem volume of M. breviflora and S. superba (Figures 6G,I).

Translocation experiments utilize natural temperature gradients to increase T_air and T_soil (Dunne et al., 2004; Salinas et al., 2011), but may indirectly reduce soil moisture, leading to lower SVWC (Hu et al., 2016; Fang et al., 2020). Although some studies have analyzed the impact of few environmental variables on biological traits in response to translocation treatments, few have investigated the full suite of physical environmental variables and subsequent impacts on biological traits (Budge et al., 2011; Li et al., 2016; Soderberg et al., 2020; Lie et al., 2021).

The IR heaters directly increased T_air and T_soil in the dry (but not wet) season in MEBMF. The IR heaters also dried the air, leading to lower RH and higher VPD, as has been observed in other studies, with observations of some impacts on the plants (Amthor et al., 2010; de Boeck et al., 2012). However, most studies using IR heaters did not address the potential impacts of the underlying indirect effects of reduced RH and increased VPD, although it was biological relevant (Aronson and McNulty, 2009). Overall, the main difference between the two warming methods was that translocation warming indirectly dried the soil, while IR warming indirectly dried the air.

The 0.68°C (CBMF) and 1.63°C (MEBF) of translocation warming decreased A_sat for M. breviflora in the wet season, inhibiting its growth. In contrast, translocation warming increased A_sat for S. rehderianum in the wet season in MEBF, and increased J_max, V_max and A_sat for S. superba in the wet season in CBMF and MEBF, leading to more rapid growth. M. breviflora in CBMF exhibited smaller stomatal size in response to 0.68°C of translocation warming, which may have contributed to lower A_sat (Zhu et al., 2020). A_sat for M. breviflora in CBMF and MEBF were mainly reduced by the direct effect of higher T_soil, subsequently leading to lower water uptake from the soil, and lower g_s. Higher T_air directly increased g_s of S. rehderianum in MEBF, leading to increasing A_sat and growth (Dai et al., 2021). However, growth of S. rehderianum in CBMF was lower due to translocation warming. The differential responses of photosynthesis for S. rehderianum in the CBMF and MEBF sites may be attributable to warming impacts on soil types and nutrient availability. S. rehderianum exhibited smaller stomatal size in response to increasing T_air caused by translocation and IR warming, which would be beneficial for lowering transpiration, as has been observed previously (Parkhurst, 1994; Pellizzari et al., 2017; Wu et al., 2020a). J_max, V_max, and A_sat for S. superba were increased by higher T_air and T_soil in both translocation warming sites. T_air at the translocation warming sites might be lower than T_opt of S. superba, which enhanced its photosynthetic capacity and rate due to increasing g_s (Way and Yamori, 2014; Slot and Winter, 2017). Higher T_soil could increase water uptake to promote leaf water supply, thereby increasing g_s for S. superba, resulting in increased photosynthetic capacity and rate (Wu et al., 2012; Harrison et al., 2020; Wu et al., 2020b). Increased T_air induced by translocation warming resulted in higher transpiration for S. superba, which might be used to cool leaves (Drake et al., 2018).

IR warming reduced A_sat for S. rehderianum and S. superba in the wet season in MEBMF. g_s of S. rehderianum and S. superba in the wet season in MEBMF was mainly reduced by the direct effect of higher T_air under 1.82°C of IR warming, when T_air might exceed its T_opt (Taylor et al., 2019), which reduced A_sat. Additionally, S. rehderianum and S. superba under 1.82°C of IR warming might have limited capacity to alter T_opt to acclimate to higher T_air (Carter et al., 2021), resulting in declining g_s and A_sat. Thus, the 1.82°C IR warming treatment significantly decreased growth for S. rehderianum in MEBMF due to declining A_sat.

Translocation warming indirectly reduced SVWC due to higher T_soil. Lower SVWC may decrease soil water availability, thereby increasing stomatal limitation to A_sat for M. breviflora in the wet season, and inhibiting its growth (Zhang et al., 2020). Similar to our results, Reich et al. (2018) found that warming impacts on photosynthesis was dependent on SVWC. In contrast, growth of M. breviflora was enhanced by translocation warming in MEBF, when SVWC was higher, leading to higher A_sat.

IR heaters indirectly reduced RH and increased VPD. The reductions in J_max, V_max and A_sat for M. breviflora in the wet season in MEBMF were attributable to indirect effect of declining RH rather than increasing T_air under 1.82°C of IR warming, which was also related to decreased g_s due to high stomatal sensitivity to dry air (Kitao et al., 2014). Xiao et al. (2020) also found that reduced g_s was driven by declining RH instead of increasing T_air. Higher T_air and lower RH both enhanced stomatal size for M. breviflora, which might be beneficial for increasing transpirational cooling to reduce leaf temperature in warmer environments (Miserere et al., 2021). Stomatal density for S. superba was decreased by the indirect effects of higher RH and lower VPD, which could reduce A_sat and water loss, similar to previous observations (Du et al., 2019; Fanourakis et al., 2020; Miserere et al., 2021). The IR warming treatment reduced leaf P_a concentration for S. superba due to lower transpiration, which further inhibited photosynthesis (Aliniaeifard and van Meeteren, 2013; Leon-Sanchez et al., 2019; Shrestha et al., 2021).