Diverging Responses of Two Subtropical Tree Species ( Schima superba and Cunninghamia lanceolata ) to Heat Waves

: The frequency and intensity of heat waves (HWs) has increased in subtropical regions in recent years. The mechanism underlying the HW response of subtropical trees remains unclear. In this study, we conducted an experiment with broad-leaved Schima superba ( S. superba ) and coniferous Cunninghamia lanceolata ( C. lanceolata ) seedlings to examine HW (5-day long) e ﬀ ects on stem water transport, leaf water use e ﬃ ciency (WUE), morphology and growth, and to elucidate di ﬀ erences in the responses of both species. Our results indicated that HWs can signiﬁcantly reduce hydraulic conductivity in both species. C. lanceolata experienced signiﬁcant xylem embolism, with the percentage loss of conductivity (PLC) increasing by 40%, while S. superba showed a non-signiﬁcant increase in PLC ( + 25%). Furthermore, HW also caused a reduction in photosynthesis rates ( A n ), but transpiration rates ( T r) increased on the 5th day of the HW, together leading to a signiﬁcant decrease in leaf WUE. From diurnal dynamics, we observed that the HW caused signiﬁcant decrease of S. superba A n only in the morning, but nearly the all day for C. lanceolata . During the morning, with a high vapor pressure deﬁcit (VPD) environment, the HW increased T r, which contributed a lot to latently cooling the foliage. In comparing the two tree species, we found that HW e ﬀ ects on S. superba were mostly short-term, with leaf senescence but limited or no xylem embolism. The surviving S. superba recovered rapidly, forming new branches and leaves, aided by their extensive root systems. For C. lanceolata , continued seedling growth initially but with subsequent xylem embolism and withering of shoots, led to stunted recovery and regrowth. In conclusion, apart from the direct thermal impacts caused by HW, drought stress was the main cause of signiﬁcant negative e ﬀ ects on plant water transport and the photosynthetic system. Furthermore, S. superba and C. lanceolata showed clearly di ﬀ erent responses to HW, which implies that the response mechanisms of broad-leaved and coniferous tree species to climate change can di ﬀ er.

usually grow in a hot and humid environment, which indicates a better tolerance to hot weather but less tolerance to drought stress, which are contradictory. Thus, research to increase our understanding of the adaptive mechanisms of S. superba and C. lanceolata to HWs is necessary and timely.
Finally, we designed and carried out pot experiments to characterize the adaptive mechanisms of a broad-leaved tree species (S. superba) and a coniferous tree species (C. lanceolata) in response to HWs in southern China. We imposed an experimental HW to measure the HW effect on both plant water transport and water use processes. We hypothesise that (1) HW will cause a variation in stem hydraulic characteristics, specifically the presence of xylem embolism; (2) HW will also affect the photosynthetic system, including a variation in photosynthesis (A n ), transpiration rates (Tr) and water use efficiency (WUE); (3) Such variations on the hydraulic and photosynthetic systems will further impact the plant leaf nutrient content, biomass and morphology; and (4) finally, broad-leaf and coniferous trees species with different responses to HW also indicates different tolerance strategies.

Experimental Platform
The experimental site was located within a seedling nursery at the Fujian Academy of Forestry Sciences, Fuzhou, Fujian, China (26 • 08 53 ' N, 119 • 17 03 ' E, 63 m a.s.l.), which has a subtropical maritime monsoon climate. Mean annual precipitation is 1600 mm, with 56% falling in the rainy season from March to June. The highest recorded maximum daily precipitation is 131.7 mm and mean annual potential evaporation is 1323 mm. Mean annual temperature is 18.9 • C and monthly mean temperature ranges from 8.9 • C in January to 27.8 • C in July.
Two major subtropical tree species were chosen: one broad-leaved tree species (Schima superba) and one coniferous tree species (Cunninghamia lanceolata). All seedlings were planted in the seedling nursery two years prior to the experiment, with healthy seedlings of comparable size carefully transplanted into pots half a year before the start of the experiment. Three-year-old seedlings are considered the key stage for local forestry work. The average height of C. lanceolata seedlings was 36.1 cm with an average diameter of 3.4 cm at ground level, while the average height and ground-level diameter of S. superba seedlings was 44.4 cm and 6.0 cm, respectively. Aside from their relatively fast growth rate, seedlings are usually transplanted from the nursery to the field at this age. The pots we used were circular, with a 44 cm diameter and a height of 30 cm, which fully simulated the planting distribution during plantation cultivate, and also ensured that the pot internal spaces provided ample space for seedling root growth to avoid limitations in terms of nutrients, water, or space. We used a red loam soil (total organic carbon = 47.3 g kg −1 , total nitrogen = 3.2 g kg −1 and pH = 4.9) collected at Wuyi Mountain at 500 m a.s.l., with a dry soil weight of 20 kg per pot. The pots were buried horizontally in the soil at ground level to avoid excessive soil warming via the sides of the pots and to ensure a soil environment similar to field conditions. In total, 48 seedlings (for each species) were grown in separate pots, evenly divided into 2 treatments (control and HW) with 3 replicate plots, so that each plot contained 16 pots. Eight seedlings (4 for control and 4 for HW treatment) were harvested on the first day after the HW to measure the stem hydraulic conductivity. The other 8 seedlings were kept at ambient conditions until the last day of the experiment for measuring the biomass and leaf nutrition parameters.

Heat Wave Simulation
Open top chambers (OTCs) with two electric heaters were used to simulate HWs (Figure 1), with each plot being covered by an OTC (a total of 6 OTC). The OTCs were rectangular in shape, with a length of 4.8 m, a width of 2.2 m and a height of 2.2 m, and were constructed using 28 mm steel tubes. During the HW treatment, the OTCs were covered with a polyethylene mesh cloth. Two industrial electric heaters (20 cm × 15 cm × 15 cm, 3500 W each, BGE, Germany) were installed inside each HW OTC at a height of 2.0 m, with a fixed angle of 20 • to prevent hot wind blowing directly on the plants. Four fans providing axial flow were suspended on both sides of the heater to homogenise the air inside the OTC. Thermal control devices were used to prevent the temperature in the OTC from exceeding the local air temperature (45 • C) for comparable subtropical areas. To minimise the impact of the OTC, control plots were also covered by OTCs, but without the heating devices. To quantify the abiotic effects of HW simulation, the air temperature at a 5 cm height (Ta), soil temperature at a 5 cm depth (Ts), and soil volumetric water content (SWC) were monitored during the experiment. The temperature was measured by button type temperature recorders (IBUTTON, DS1922L, China). Each plot had 6 IBUTTONs, 3 for Ta and 3 for Ts, with 18 IBUTTONs used in total. SWC was measured daily at 10:00 a.m. during the 15 days before, during and after the HW by Hydro sense-II sensors with 12-cm rods (Campbell Scientific, Logan, UT, USA). During the HW treatment, the OTCs were covered with a polyethylene mesh cloth. Two industrial electric heaters (20 cm × 15 cm × 15 cm, 3500 W each, BGE, Germany) were installed inside each HW OTC at a height of 2.0 m, with a fixed angle of 20° to prevent hot wind blowing directly on the plants. Four fans providing axial flow were suspended on both sides of the heater to homogenise the air inside the OTC. Thermal control devices were used to prevent the temperature in the OTC from exceeding the local air temperature (45 °C) for comparable subtropical areas. To minimise the impact of the OTC, control plots were also covered by OTCs, but without the heating devices. To quantify the abiotic effects of HW simulation, the air temperature at a 5 cm height (Ta), soil temperature at a 5 cm depth (Ts), and soil volumetric water content (SWC) were monitored during the experiment. The temperature was measured by button type temperature recorders (IBUTTON, DS1922L, China). Each plot had 6 IBUTTONs, 3 for Ta and 3 for Ts, with 18 IBUTTONs used in total. SWC was measured daily at 10:00 am during the 15 days before, during and after the HW by Hydro sense-II sensors with 12-cm rods (Campbell Scientific, Logan, UT, USA). In line with the local climate history, HWs were applied in August, the month with the second highest HW frequency, to reduce the risk of natural HW impact, continuous dry sunny days were selected for HW application because natural HWs are usually dry. The HW treatment lasted for 5 days, and the polyethylene mesh cloth did not fully insulate the OTC, with outside temperatures also influencing inside air temperatures. A pilot experiment showed that the canopy temperature increased by c. 6-10 °C during the day and by c. 4 °C at night. After the HWs, the OTCs were removed, and a shade cloth was installed to avoid any natural HW impacts. There was no irrigation during the In line with the local climate history, HWs were applied in August, the month with the second highest HW frequency, to reduce the risk of natural HW impact, continuous dry sunny days were selected for HW application because natural HWs are usually dry. The HW treatment lasted for 5 days, and the polyethylene mesh cloth did not fully insulate the OTC, with outside temperatures also influencing inside air temperatures. A pilot experiment showed that the canopy temperature increased by c. 6-10 • C during the day and by c. 4 • C at night. After the HWs, the OTCs were removed, and a shade cloth was installed to avoid any natural HW impacts. There was no irrigation during the HW period for both control and HW treatments (the OTS is water proof), but nearly 3 L of water was immediately supplied in each pot when the OTC was removed, and the SWC was maintained between 40% and 50% during the rest of the experiment. The experiment was divided into three periods: "pre HW" (B1-B5), "HW" (H1-H5) and "post HW" (A1-A30). The environmental conditions observed inside the OTCs were in line with realistic summer patterns and HW events ( Figure 2). Before the HW simulation (B5-B1), all the environmental factors (Ta, Ts and SWC) were similar between treatments, indicating that all the environmental variations came from our HW simulation method. When the HW simulation began, the Ta in the HW groups significantly (p < 0.001) increased by 9.9 • C on average, from 24.1 • C to 33.9 • C, while the Ts increased by 6.2 • C, from 24.1 • C to 30.3 • C (p < 0.001). The SWC in the HW groups decreased significantly, with a daily rate of SWC reduction (3.2% day −1 ) more than 3 times higher under HW treatment than that in the control group (1.0% day −1 ). The daily average significantly decreased by more than 12%, from 41.5% to 28.7% (p = 0.001). After the HW, the Ta in the HW group immediately returned to the control level, while the Ts remained slightly higher than the control level until the second day. The SWC in the HW group quickly recovered to the same level as the control because of sufficient irrigation.

Stem Hydraulic Conductivity
Stem hydraulic conductivity was measured on the first post HW day (A1) ( Table 1). Main stems of c. 15-cm long, each from a different individual, were collected from 5 cm above the soil in the early morning. After excision, the cut ends of the branches were immediately re-cut under water to remove embolized tracheid (2 cm removed). The stem segment, selected for hydraulic conductivity measurements, and the downstream parts with leaves were subsequently labelled. Remaining under water throughout this entire procedure, the labelled stem segments were then immediately transported to the laboratory where 10-cm-long stem segments were cut from the middle part of the originally sampled segments for determining hydraulic conductivity (K h ). After smoothing both ends with a sharp razor blade, the stem segments were connected to a tubing apparatus. The perfusion solution used for hydraulic conductivity measurements was 10 mmol·L −1 oxalic acid solution, degassed and filtered through a 0.2 µm pore. We used a hydraulic head of 50 cm to generate hydrostatic pressure by which a steady flow of the solution was passed through the stem segments. Stem K h (g·m·s −1 MPa −1 ) was calculated as follows: where J v is the flow rate of solution through the segment, p is the hydrostatic pressure generated by the hydraulic head and L is the length of the segment. After the determination of K h , the degree of stem xylem embolism was estimated by comparing native hydraulic conductivity and maximum hydraulic conductivity after embolisms were removed, which is the percentage loss of conductivity (PLC). We then connected the stem to a tubing apparatus and flushed it with filtered (0.2 µm) oxalic acid solution at 100 kPa for 15 min to eliminate any existing embolisms and subsequently determined the maximum hydraulic conductivity (K max ) by applying the same method as for K h . PLC was then calculated as follows: Then, the specific conductivity (K s , g m s −1 MPa −1 ), leaf specific conductivity (K l , g m s −1 MPa −1 ) and Huber value (H v ) were calculated as follows: where A w is the sapwood area (cm 2 ), and LA is the total leaf area (cm 2 ) terminal to the segment. Table 1. Time series on measuring of stem hydraulic conductivity, leaf gas exchange, biomass, leaf C/N content and plant appearance during the whole experiment. The experiment was divided into three periods: "pre HW" (B1-B5), "HW" (H1-H5) and "post HW" (A1-A30).

Pre Heat Wave
Heat

Leaf Gas Exchange
Photosynthetic measurements were conducted during and after the HW treatment to determine the effects of HW on foliar gas exchange. To assess diurnal changes, measurements were conducted before (B1), during (H3 and H5), and after the HW (A3, A5 and A7) during the morning (8:00-10:00 a.m.) (Table 1). Moreover, the diurnal dynamics of A n , Tr and WUE were determined every two hours on day H5 from 8:00 a.m. to 16:00 p.m. Mature fully expanded leaves were chosen randomly then marked from each individual (48 leaves in total), and leaf temperature (T leaf ), leaf vapor pressure deficit (VPD), stomatal conductance (g s ), net photosynthesis rate (A n ) and transpiration rate (Tr) were measured with a portable photosynthesis system (LI-6800, LI-COR Biosciences, Lincoln, NE, USA). During measurements, the photosynthetic photon flux was maintained at 1500 µmol·m −2 ·s −1 to ensure that light-saturated photosynthesis rates were reached for the two study species. The ambient CO 2 concentration was maintained at 400 µmol·mol −1 , and the airflow was auto-set in the leaf chamber without temperature and water controls. We ensured that the broad leaf area (green part) fully filled the leaf chamber, while needle leaves were arranged together to ensure complete cover of the leaf chamber (6 cm 2 circle). The leaf water use efficiency (WUE) was calculated as WUE = A n /Tr.

Final Harvest
Thirty days after the end of the HW (A30) (Table 1), the seedlings were harvested and then sorted into different categories (by species and by plant tissue (leaves, stems and roots)), oven-dried at 65 • C for 3 days and weighed to measure the biomass. Dried leaves were pulverized with a ball mill, and approximately 4 mg of the sample was collected and wrapped into a tin cup to determine the carbon and nitrogen content with a C/N elemental analyser (IRMS, Isoprime 100, UK) for measuring the leaf total nitrogen content (TN) and total carbon content (TC), and the C/N ratio was calculated.

Data Analysis and Statistics
The average value of each plot (containing 4 pots) was used for replications (n = 3) before the analysis. Paired T-tests were applied to examine the effects of HWs on the environment (Ta, Ts and SWC). Multivariate analysis (two-way ANOVA) was used to examine the effects of the HW, the species and their interaction on the hydraulic characteristics (K h , K max , PLC, K s , K l and H V ) and biomass (leaf, stem, aboveground, underground and total) of the two tree species. Repeated measures analysis was used to examine daily or temporal variability of the photosynthetic system (T leaf , VPD, g s , A n , T r and WUE) and leaf nutrient parameters (TC, TN and C/N ratio) during different stages (days) for different species when combined with the HW treatment. Independent T-tests were applied to examine short-and long-term HW effects on biomass, leaf C/N content, the photosynthesis system and seedling hydraulic characteristics of S. superba and C. lanceolata. All statistical analyses were conducted using SPSS 22.0 for Windows (SPSS Inc., Chicago, Illinois, USA).

HW Effects on Plant Hydraulic Characteristics
The hydraulic characteristics of both S. superba and C. lanceolata were significantly affected by the HW treatment, although interspecific differences were apparent ( Table 2). An HW caused the K h of S. superba to decrease significantly by 52% from 0.032 g·m −1 s −1 ·MPa −1 (p = 0.046), while the K max decreased by 24% from 0.035 g·m −1 s −1 MPa −1 (p = 0.069) ( Figure 3). In comparison, the K h of C. lanceolata decreased significantly by 35% from 0.041·g m −1 s −1 MPa −1 (p = 0.034), and the K max decreased by 51% from 0.046 g m −1 s −1 MPa −1 (p = 0.054) (Figure 3a,b). Therefore, under the HW treatment, the PLC of S. superba increased by 25% (from 10.3%), which was, however, not significantly different from that of the control group (p = 0.206), whereas the PLC of C. lanceolata was significantly increased by 40% from 12.5% (p = 0.006) (Figure 3c). Although HWs caused no significant differences in H V for the two species (Figure 3f), they did significantly decrease the K s and K l of C. lanceolata, but not significantly in S. superba, where the K S of S. superba decreased by 40% from 0.195 g m −1 s −1 MPa −1 and the K l decreased by 47%, while the K S of C. lanceolata decreased by 66% from 0.407 g·m −1 s −1 ·MPa −1 and the K l decreased by 75% (Figure 3d,e).    Table 2. Outcome of the multivariate ANOVA analysis (p-values given) on the effects of heat wave (HW), species (S) and their interactions on hydraulic characteristics (hydraulic conductivity (K h ), max hydraulic conductivity (K max ), percentage loss of conductivity (PLC), specific conductivity (K s ), leaf specific conductivity (K l ) and Huber value (H V )) and biomass (leaf, stem, aboveground, underground, total and root/shoot) after heat wave effect. p-values of <0.05 are indicated in bold.  Figure 2. The major microclimatic variables before, during and after the 5-day heat wave period. Control: C, Heat Wave: HW. Ta, air temperature (a), Ts, soil temperature (b), SWC, soil water content (c). B1-B5, days before the heat wave, H1-H5, days during the heat wave, A1-A5, days after the heat wave. The bar chart to the right indicates the change in environmental conditions during the heat wave (5 day means), and the p-values are the result of paired T-tests

HW Effects on Leaf Gas Exchange and WUE
Differences in daily T leaf , VPD, g s , A n , Tr and WUE between the HW and control groups were used to measure the resistance and resilience of S. superba and C. lanceolata seedlings to HW stress (Table 3). HW led to a significant increase in T leaf during the HW period (H3 and H5) (p < 0.01), while the T leaf on day H3 was higher than on day H5, while the leaf VPD was highest at H5 for both species (Figure 4a,b,g,h). In line with higher VPD values on that day, Tr was significantly increased during day H5 (p < 0.05), to a similar extent for both species (Figure 4e,k), as there was no significant variation of g s (Figure 4c,i). Significant species-specific variation only occurred for the A n and WUE (p < 0.01). For the broadleaved species S. superba, A n was significantly (p < 0.01) affected by HW only at H5, with a reduction of more than 55% (Figure 4d), but the significant effect of HW on A n disappeared three days later. In comparison, for the coniferous species C. lanceolata, the significantly reduced A n occurred earlier and lasted longer than that observed in S. superba: HW caused a significant reduction in A n of 41% and 51% during the HW period (H3 and H5), and A n was still 49% lower 3 days after the end of the HW treatment (Figure 4j). Although the WUE of each species responded differently to HW, the only significant difference was recorded for H5 for both species (Figure 4f,l).   HWs also significantly affected the diurnal dynamics of plant leaf gas exchange on the last day of the HW (Table 4). On the 5th day of HW, T leaf was significantly higher in the HW group than in the controls for both species (Figure 5a,g). The VPD variation followed the change in temperature and radiation (Figure 5b,h), with the highest T leaf and VPD occurring at noon (12:00 a.m.). The peak A n and WUE of S. superba and C. lanceolata both occurred at 8:00, and the most significant negative effects of the HWs were also recorded at this time (p < 0.001). Additionally, HWs also significantly increased the Tr at this time of day (p < 0.05). At 10:00 am, the decrease in A n had become insignificant, but the differences in Tr were still significant for both species (p < 0.001), associated with a significantly increased g s (p < 0.05). Species-specific differences regarding the response of the g s , A n , Tr and WUE to HWs started appearing from noon. For S. superba, there were no significant differences between the HW and control groups after this time, but for C. lanceolata, significant negative effects (p < 0.01) of HWs on A n and g s were observed at 12:00 a.m. and 14:00 pm, and the significant difference lasted until 16:00 p.m. for A n , while g s was not significant at this time of day (Figure 5i,j). This was also coupled with a significant reduction in WUE (Figure 5l) at 12:00 a.m. and 16:00 p.m. and an increased Tr at 16:00 p.m. (Figure 5k).

HW Effects on Plant Leaf Nutrients and Biomass
The HW caused a negative effect on the plant photosynthetic system, in turn causing significant variation in leaf nutrient contents and also affecting nutrient accumulation in the 30 days following the HW. Leaf TN was significantly affected by HWs, and interspecific and temporal differences were Level of significance are * p < 0.05; ** p < 0.01; *** p < 0.001 and ns, non-significant.

HW Effects on Plant Leaf Nutrients and Biomass
The HW caused a negative effect on the plant photosynthetic system, in turn causing significant variation in leaf nutrient contents and also affecting nutrient accumulation in the 30 days following the HW. Leaf TN was significantly affected by HWs, and interspecific and temporal differences were also apparent (p < 0.05), while HW had an insignificant effect on the leaf TC and C/N ratio with some interspecific differences ( Table 5). The HW effect on S. superba leaf nutrients was most notable on the first day following the HW (A1), with a significantly higher TN (p = 0.035) and lower C/N ratio (p = 0.018) in the HW group; these significant differences disappeared 30 days later (Figure 6a-c). In contrast, the HW effect on C. lanceolata was insignificant during A1, while significant increases in leaf TN (p = 0.016), insignificant increases in leaf TC (p = 0.076), and a decrease in the C/N ratio (p = 0.019) were observed 30 days after the HW (Figure 5d-f).   We found significant species-specific effects of HWs on plant biomass, especially belowground biomass, at 30 days after the end of the HW treatment (A30) (p = 0.013, Table 2). In the control treatments, the biomass of C. lanceolata (40.3 g) was much larger than that of S. superba (3.4 g), but the root:shoot ratio of S. superba (0.71) was higher than that of C. lanceolata (0.50) (  We found significant species-specific effects of HWs on plant biomass, especially belowground biomass, at 30 days after the end of the HW treatment (A30) (p = 0.013, Table 2). In the control treatments, the biomass of C. lanceolata (40.3 g) was much larger than that of S. superba (3.4 g), but the root:shoot ratio of S. superba (0.71) was higher than that of C. lanceolata (0.50) ( Table 6). HWs did not significantly affect the aboveground biomass of C. lanceolata (p = 0.711), with increases of leaf biomass and decreases of stem biomass seen, as well as a decrease in the belowground biomass by 21% (4.3 g, p = 0.074). For S. superba, the HW insignificantly stimulated the aboveground biomass by 62% (2.1 g, p = 0.083), coupled with a 49% increase of leaf biomass, 74% increase of stem biomass, and a 107% increase in the belowground biomass (2.6 g, p = 0.083). Moreover, HWs increased the root:shoot ratio of S. superba (0.90) but reduced this ratio in C. lanceolata (0.36). In general, HW caused no significant (p = 0.964) effect on the total biomass of C. lanceolata, (60.5 g in control treatment and 60.0 g in HW treatment) but a significant (p = 0.008) increase in the S. superba total biomass by 80%, from 5.8 g to 10.5 g.

HW Effects on Plant Appearance
The HW effects on the plant hydraulic and photosynthetic systems gave rise to differences in plant and leaf appearance (Figure 7). The edges of the S. superba leaves were partially withered, reducing the effective photosynthetic leaf area (Figure 7b,c,e). This phenomenon was most apparent in the top leaves of the plants, with 70% of the seedlings showing similar symptoms. Moreover, more than 75% of S. superba individuals sprouted new branches at the soil surface and developed new leaves after the end of the HW (data were estimated by observation) (Figure 7a-d). C. lanceolata leaves also had a withered appearance that first manifested in the distal branches and then gradually occurred on multiple branches (Figure 7f-k). The withered leaves located at the top of the branch did not recover 30 days after the HW. than 75% of S. superba individuals sprouted new branches at the soil surface and developed new leaves after the end of the HW (data were estimated by observation) (Figure 7a-d). C. lanceolata leaves also had a withered appearance that first manifested in the distal branches and then gradually occurred on multiple branches (Figure 7f-k). The withered leaves located at the top of the branch did not recover 30 days after the HW.

Discussion
Thermal damage and drought stress are widely considered to be the main environmental stresses associated with HWs. These factors both influence plant water transport processes and the photosynthetic system, in turn affecting the plant's WUE. In this study, we provide direct evidence that HWs will disturb the plant water transport system because of xylem embolism. Although gradual recovery might happen after the HW, immediate HW impacts led to persistent differences in plant leaf C and N contents, biomass and morphology weeks after the end of the event. We also observed distinct differences between the subtropical broad-leaved S. superba and the conifer C. lanceolata, regarding both direct HW stress and recovery processes.

HWs Caused a Negative Effect on Plant Water Transmission System by Xylem Embolism
The imposed HW led to a reduction in the K h and an increase in the PLC in the stems, which was especially apparent in C. lanceolata (Figure 3). This indicates xylem embolism, which is a phenomenon that commonly occurs in tree species of arid and semi-arid regions, caused by a combination of soil water stress and high VPD [45,46]. Studies on hydraulic functioning have mostly focused on the drought thus far [46,47], while there is a lack of studies that consider concurrent dry and hot conditions, the combination of which is often observed during HWs. During HW periods, extreme high temperatures also lead to increased evaporative demand from the air. These factors combine to increase water stress in plants, which is manifested as increased xylem tension. Most studies so far indicated that xylem water potential tends to decrease during drought, especially when plants desiccate, which will lead to increasing xylem tension, causing water column rupture whereby external microbubbles can enter the vessel, blocking the transport of water [47][48][49]. Typically, these consequences include a rapid cessation of photosynthetic CO 2 assimilation, loss of canopy evaporative cooling through transpiration and greater probability of photodamage over a short time scale [50]. Over longer time scales, low photosynthetic rates associated with drought-induced stomatal closure can lead to depletion of non-structural carbohydrate pools, which interferes with translocation of sugars through the phloem and the production of chemical defense compounds needed to prevent herbivory and disease [23,51]. All this theory hinges on the premise of drought, but when drought is accompanied by high temperature, the impact may be different. On the one hand, besides hydraulic failure causing cessation and low photosynthetic rates, thermal stress also leads to the same results [9]. This double stress can accelerate this depletion of non-structural carbohydrate pools. On the other hand, responses such as reduced transpiration cooling or stomatal closure are less likely to occur in a high temperature environment [52]. This is because thermal stress can be avoided by sustaining high transpiration to latently cool leaves below the air temperature or through biochemical adaptation such as heat-shock protein generation to increase thermal tolerance [52]. Our experiment also detected a significantly higher Tr combined with a high VPD during the HW period (Figure 4), especially at the hottest time of day ( Figure 5). Therefore, combined with the low soil water content under HWs (Figure 2), we assumed that the HW-induced xylem embolism in this experiment was primarily caused by the depletion of soil moisture due to the increased consumption of water through plant transpiration and the high VPD associated with high temperatures. Moreover, passive xylem embolism can produce significant morphological damage, as was the case for C. lanceolata in our study, which makes recovery slow [53]. In general, the xylem embolism caused by HW may accelerate drought-only impacts as plants need to find a balance between cooling and water saving, which make xylem embolism more likely to happen during the HW period.
However, many woody plants commonly operate xylem embolism when facing drought stress [48]. Although some plant death is caused by hydraulic failure, many individuals can survive despite a high loss of conductivity [23,54]. Some studies also indicated that under drought conditions, some tree species undergo embolism of large-diameter vessels to keep the small-diameter vessels transporting water normally, thereby reducing water consumption and maintaining water supply to ensure tree survival throughout the dry season [55]. When the drought abates, plants can refill embolised xylem conduits via spontaneous bubble dissolution once xylem pressure nears atmospheric values, or under positive root/stem pressure [56,57]. Trees can also repair the total conducting area following seasonal replacement through regrowth of new xylem [48]. However, the premise is that the loss of xylem hydraulic conductance is not beyond the thresholds, which was considered to be above 60% [55,58,59]. While the xylem embolism observed in the current experiment was below this value, the plants still had a chance of survival after a 5-day long HW. Moreover, this recovery process depends on the degree of damage to the apical and cambial meristematic tissues, the functional status of the remaining hydraulic pathway, the overall health of trees and the plant stocks of water, non-structural carbohydrates and nutrients [46]. In addition, under HW stress, thermal damage can lead to leaf withering (Figure 7), and plants need to repair their photosynthetic apparatus first. Moreover, the significant decrease of leaf C/N of C. lanceolata ( Figure 6) also indicated potential deficiency of non-structural carbohydrates, which would all prolong this recovery process when comparing to a single drought stress.

HWs Decreased WUE Due to the Reduction in the Leaf A n or Increase in the Tr
Photosynthesis is a process sensitive to temperature peaks, with both the "light" (electron transport rate) and the "dark" (Calvin cycle) reactions being thermally sensitive, and this is especially true for PSII in the light reaction [60] and Rubisco activase in the dark reaction [61]. This can lead to a variety of potential physiological disorders, including ammonia poisoning, biofilm damage and protein denaturation, eventually leading to plant death [62]. Moreover, because the plant respiration rate is generally higher than the photosynthesis rate under high temperatures, carbon loss exceeds carbon gain [63], which could accelerate the depletion of non-structural carbohydrate pools. In fact, some plants that are unable to sustain high photosynthetic rates during high temperature (>40 • C) may prefer to keep their stomata open (at the expense of WUE) to ensure transpiration cooling to safely pass the HW period. But the premise to satisfy this assumption is that the environment should have sufficient water supply and the water can successfully transport to the leaf tissue [15,53]. In this situation, soil water depletion or xylem embolism causes a reduction in water transport, resulting in less transpiration cooling, with the potential for thermal damage leading directly to negative effects to the photosynthetic system.
In this study, HWs not only decreased the efficiency with which water was transported through the plant but also caused a decrease in WUE on the 5th day of HW due to a reduction in the A n and a simultaneous increase in the Tr (Figure 4). However, from the daily dynamics of A n , Tr and WUE during the same day, we noticed that the contribution to lowering WUE by a decrease in A n and an increase in Tr depended mostly on leaf VPD ( Figure 5). Under low VPD conditions, the decrease in A n was the main reason for the decrease in WUE (8:00 a.m.), but under high VPD conditions, it was mostly an increased Tr that led to the decrease in WUE (10:00 a.m.). We assume that although no irrigation was supplied during the experiment, the naturally formed dew could help foliage cooling via evaporation in the early morning. Later, during the day, trees need to transpire to lower leaf temperatures [52], which is associated with an increased Tr, as seen in our experiment ( Figure 4, 10:00 a.m.). According to the processes described above, 5-day long HW stress caused no permanent foliar damage but temporarily reduced the photosynthetic rate. The dissipated difference in A n and WUE between the HW and the control group 7 days after the HW ended supports this assumption (Figure 4), also suggesting recovery from HW stress.

Differences in the Response of Two Subtropical Tree Species to HWs
We directly derived the degree of xylem embolism from the relative loss of conductivity (PLC) under the HW and found that the broadleaved S. superba had a better embolism resistance and resilience than the coniferous C. lanceolata in a short-term (5 days) single HW (Figure 3). The larger K s and K l of S. superba also suggests that a higher water potential inside its vascular bundles and more leaf and less sapwood areas contributing to water transport may benefit water transmission in vascular bundles [32]. In addition, S. superba had a higher root:shoot ratio than C. lanceolata ( Table 6), indicating that S. superba was able to absorb relatively more soil water (compared to its total mass), which was beneficial to offset HW stress [47,64,65]. Furthermore, the HW event even enhanced the root:shoot ratio of S. superba, while the opposite effect was observed for C. lanceolata. Moreover, the leaf TN concentration in S. superba did not increase in the recovery period, unlike in the control and both variations of C. lanceolata ( Figure 6). Based on these observations, we conclude that harmful HW effects on the water transport system of S. superba were mostly rapid yet short-lived, while longer-lasting (legacy) effects were apparent in C. lanceolata. The differences in leaf morphology may explain why these broadleaved (S. superba) and coniferous tree species (C. lanceolata) responded differently to HWs. The leaves of S. superba were mainly concentrated at the tip of the stem, where their transpiration provides the driving force for stem flow [66]. When the HW occurred, many leaves were damaged by the high temperatures and partly withered at the leaf tips (Figure 7a-e), reducing the effective leaf area and decreasing the water potential in the stem, in turn lowering water transport (Figure 3). The ensuing leaf regeneration combined with the extensive root system [67] led to full recovery and even enhancements in the water transmission capacity [68]. In contrast, the leaves of C. lanceolata were evenly distributed on both sides of the branches (Figure 7f-k), which indicates that the water potential in the xylem gradually increases from the bottom to the tip [69]. Transpiration sustains the hydraulic potential so that the bottom of the xylem is more affected by the number of leaves than the tip area. When plants are unable to access a sufficient water supply, the leaves located at the tip of the branch will wither and die off first, while leaves near the bottom may survive [70]. This gradual withering process actually postpones the time at which HW effects become apparent. A significant variation in C. lanceolata leaf nutrient content was detected only 30 days after HW stress, not immediately after the HW. However, tender shoots and growing points of the C. lanceolata located at the tip of the branch were easily affected by water shortage, and xylem embolism occurred when many old leaves died in the tip area without new leaf growth [71,72]. When the stem is fully embolized and tender shoots die, the branch is unable to continue growing and gradually withers. Large-scale death of C. lanceolata seedlings was observed in forestry production in recent years (pers. comm. with Xiangqing Ma, Fujian Agriculture and Forestry University, Fuzhou, China), which seems to be related to short-term drought.
HWs affected the A n and WUE of both S. superba and C. lanceolata similarly (Figure 4), indicating a consistent response of the photosynthetic system to the extreme conditions [9,73]. However, these measurements were mainly performed on the surviving leaves after the HW, and therefore, considering the difference in the quantity of effective photosynthetic organs may be more meaningful. In S. superba, many leaves showed different degrees of withering beyond repair (Figure 7e), which may be related to a higher vulnerability of broadleaved species to high temperatures than needle-leaved species due to the larger, more atmospherically decoupled surface areas of the latter causing faster warm-up under conditions of high radiation [32]. Unlike the more gradual and slow withering of C. lanceolata leaves, the withering of S. superba leaves was more distinct, with some seedlings having no leaves left. The conditions faced during HWs thus appeared more perilous to S. superba, but if the seedlings survived, fast recovery was possible through the potential of this species to grow new branches and leaves. Importantly, these HW impacts on tree seedlings are likely to influence adult trees in terms of their morphology or growth rates. Increased lateral branch generation will cause trees to have fewer straight features, which will affect the quality of the wood and reduce the tree height. On the other hand, a low-growing tree with thicker side branches is considered to be better adjusted to drought and high temperature stress, which would imply that HW impacts at an early stage may lead to better HW resilience later in life [74]. These theories are difficult to verify, in part due to the difficulties in simulating HW events with adult trees, and because this would require a long observation period.

Conclusions
HW impacts are usually a combination of high temperature and drought effects [3,13]. In the subtropics, water stress seems to play a pivotal role. If the water supply is insufficient, plants will be unable to maintain their leaf temperature below thermal stress thresholds by transpiration, and the ensuing high temperature stress will directly cause leaves to wither and fall. Although both tree species in our study are common in subtropical areas and therefore would be expected to be adapted to a high temperature environment, the imposed HW still led to significant reductions in tree water transport and water use efficiency, although species-specific differences were apparent. We found that the broadleaved S. superba altered its biomass distribution through expansion of the root system, germination of new branches and leaf renewal. Because it avoided xylem embolism and because fresh leaf growth quickly compensated for the leaves lost during the HW, recovery was rapid for this species. The coniferous C. lanceolata maintained the growth of aboveground biomass but decreased the belowground biomass. This may have led to an imbalance between water demand and supply, especially as xylem embolism further aggravated the supply shortage. As the growing points at the branch tip withered and died first, the entire plant growth was affected, hampering the germination of fresh leaves and thus slowing recovery. Generally, HW effects on S. superba seedlings were short-lived, while significant legacy effects were observed in C. lanceolata. Although these results pertain to young trees due to technical limitations and time constraints, our findings are nevertheless useful in forestry production. Upon transplantation of S. superba and C. lanceolata, supplying shade to leaves will increase survival rates during HWs, as this will directly avoid high radiation doses and concurrent higher leaf temperatures and indirectly lower soil evaporation. Moreover, the maintenance of an appropriate water supply seems crucial for C. lanceolata during long-lasting HWs, and proper trimming of withered seedling branches afterwards might accelerate the regeneration of branches and benefit wood quality.