The sap flow velocity is measured by TDP. A set of thermal dissipation probes (AV-3665R, Rainroot Scientific Ltd., China) was installed in each tree, and the distance between the two probes is 10 cm. The basic principle is that two probes composed of thermal elements are inserted into the sapwood, and the probe inserted on top is heated using a constant current, while the probe below serves as a control. As the heat transfer from the sapwood increases with the stem flow velocity, more heat is removed from the heating probe, resulting in a temperature difference. The temperature difference between the two probes is greatest when the stem flow velocity is zero or very small (ΔTmax). 20 mm probes are used, powered by a 12 V rechargeable battery.

The installation was scheduled for July 2021. The probes were installed at 1.3 m above ground in the trees. First, the rough bark of the sample wood was scraped off at the probe mounting point, and then two vertical holes were drilled with a drill of a specific specification into which the probes were inserted. After the probes were inserted, they were clamped with a foam block and wrapped with insulating and antiradiation material after being fixed with tape. Finally, the probes were sealed with tape to prevent rainwater from entering. The whole stem flow velocity measurement system, including the stem flow meter, was composed of the TDP feedback line and data collector, which was used to collect and record the sap flow data automatically with a sampling frequency of 60 s and a data collection interval of 10 min. The sap flow velocity was calculated by Granier’s original formula (Granier, 1985) as follows:

Where F_d is the sap flow velocity (cm³⋅cm^–2⋅S^–1); F_s is the sap flow (cm³⋅S^–1); ΔTmax is the maximum temperature difference of the probes when the sap flow velocity is zero or very small. The weakest sap flow velocity of each day generally occurs in the early morning hours, and the ΔT_max is determined using the maximum temperature difference over a 3-day period in this study; ΔT is the temperature difference when sap flow velocity is occurring; A_s is the sapwood area of the sample (cm²).

The cumulative sap flow volume S (cm³) is:

Where n is the number of collected samples; F_si is the sap flow (cm³⋅S^–1) at the i-th collection; Δt is the sampling interval time (s).

Results obtained with continuously heated TDP probes showed that the natural thermal gradient had a direct effect on the sap flow measurement with a large potential error, and the natural thermal gradient of more than 0.2°C was not negligible (Do and Rocheteau, 2002). Before conducting the experiment, we referred to the results of Ma (2020) for the sap flow measurement of the same tree species in the region and found that their natural thermal gradients were all below 0.2°C. Therefore, we neglected the effect of the natural thermal gradient in our experiment.

The transpiration is measured by LLE. Water levels of barrel-planted P. tomentosa and S. babylonica are recorded under continuous sunny, windless weather using a marsupial bottle, which is a bottle made of special glass (5 cm inner diameter, 5.6 cm outer diameter, 100 cm height). The upper end of the bottle has a vent, the lower end contains the vent at 8 cm above the ground and the water inlet at 3 cm from the ground, and the water inlet and the bottom of the barrel of the two drainage ports are connected with polyvinyl chloride (PVC) pipe to the end of the door valve to prevent the water from entering the barrel when filling the bottle. When filling the bottle, the door valve is closed, the upper exhaust port is opened, the air flows, and the atmospheric pressure is stable. When the water level reaches approximately 90 cm, the upper exhaust port is blocked, the door valve is opened, and the vacuum state is maintained in the bottle. When the water in the barrel reaches saturation, the height of the falling water level is recorded. The recording time was 6:00–20:00, and the water level change was recorded every half hour. To avoid the effect of sunlight on the bottle, it was wrapped with radiation-proof film. The difference in water level height between two consecutive records of the bottle multiplied by the bottom area is the transpiration of P. tomentosa and S. babylonica in half an hour (Figure 1). The marsupial bottle has the ability to control the water level and automatic replenishment, which allows the water level inside the planting container to be maintained at an appropriate level. Based on the water balance principle, the amount of water consumed in the marsupial bottle is equal to the amount of water consumed by the tree in the well-packed planting container.

A soil moisture monitor (TH302, Tang Hua Ltd., China) was used to monitor the change in soil moisture transport. Its working principle is to use probes to monitor the moisture in the soil in real time and then reflect the change in soil moisture content. The soil moisture monitoring probes were inserted into the two species, P. tomentosa and S. babylonica. The probes were inserted at soil depths of 10, 20, and 30 cm and were covered with soil. The other end of the probes were connected to a universal data collector, and the data collector was connected by Bluetooth to THHelper software on a cell phone. The data were recorded and stored at a time interval of 5 min.

The change in diameter at breast height (DBH) was monitored using a trunk radial change recorder (DBL60, ICT International, Australia), which has a high resolution and is capable of measuring small changes in the stem to micrometer (μm) level accuracy. The principle of the operation is to monitor the stem change in real time using the relationship between the pressure and the sensor due to the radial variation in the trunk that exerts pressure on the fixed probe. The trunk radial change recorders were installed at the stem of each tree species (1.5 m above ground level) before the start of the formal experiment and were set to record and store data once every 30 min.

During the test period, solar radiation (RA, W⋅m^–2), relative air humidity (RH, %), atmospheric temperature (TA, °C), wind speed (WS, m⋅s^–1), rainfall (Rain, mm) were automatically monitored by a weather station (RR-9170, Rainroot Scientific Ltd., China) located approximately 5 m from the sample site. The sampling interval was 10 min for recording and storage. The vapor pressure deficit (VPD, kPa) was calculated by the formula (Zhao et al., 2017):

At the end of the test, the trunk core was drilled at a height of 1.3 m using a growth cone, and then the drilled core was dyed to observe the areas of sapwood and heartwood distribution, and the sapwood thickness was measured and recorded statistically to determine the sapwood area (Dang et al., 2020).

Figure 2 shows that the daily changes in sap flow velocity of P. tomentosa and S. babylonica under typical sunny, cloudy and rainy conditions showed an increasing and then decreasing pattern, and the sap flow velocity was sunny > cloudy > rainy. The sap flow velocity of P. tomentosa showed a “single-peak” pattern of change on sunny days, a “double-peak” pattern on cloudy days, and fluctuated very little on rainy days. The sap flow velocity of S. babylonica showed a “single-peak” pattern on sunny and cloudy days, and it was also very small on rainy days.

The correlations between the sap flow velocity of P. tomentosa and S. babylonica and meteorological factors were analyzed on sunny, cloudy and rainy days (Table 2), and the results showed that the sap flow velocity of P. tomentosa was highly significantly and positively correlated with TA, RA, and VPD, highly and negatively correlated with RH on sunny, cloudy and rainy days, the overall correlation between sap flow velocity and meteorological factors was weak on rainy days. The sap flow velocity of S. babylonica was highly significantly and positively correlated with TA, RA, WS, and VPD, highly significantly and negatively correlated with RH on sunny and cloudy days, and it was also weakly correlated with each meteorological factor on rainy days.

To further clarify the main meteorological factors affecting P. tomentosa and S. babylonica under different weather conditions, stepwise regression analysis was used to establish regression models with TA, RH, RA, WS, and VPD, and the factors entering the models were the main influencing factors. The regression equations of sap flow velocity and meteorological factors on sunny, cloudy and rainy days were as follows:

P. tomentosa (sunny days) = 1.137VPD+0.001RA+ 0.015RH - 1.916 R² = 0.926

P. tomentosa (cloudy days) = 0.014WS+0.037TA+0.001RA - 0.009RH+0.13 R² = 0.967

P. tomentosa (rainy days) = 0.031TA+0.743VPD-0.001WS - 0.379 R² = 0.580

S. babylonica (sunny days) = 0.497VPD+0.013RH- 0.025TA - 0.746 R² = 0.954

S. babylonica (cloudy days) = 0.140VPD - 0.092 R² = 0.899

S. babylonica (rainy days) = 0.001WS+0.062RH+0.040TA - 0.001RA+2.049VPD - 6.864 R² = 0.863

From the above Equations, it can be seen that the main influencing factors of sap flow velocity of P. tomentosa on sunny days were VPD, RA, and RH; on cloudy days they were WS, TA, RA, and RH; on rainy days they were TA, VPD, and WS; the main influencing factors of sap flow velocity of S. babylonica on sunny days were VPD, RH, and TA; on cloudy days they were VPD; and on rainy days they were WS, RH, TA, RA, and VPD.

As seen in Figure 3, the sap flow velocity and DBH changes of both P. tomentosa and S. babylonica showed opposite trends at the daily scale. During the daytime, with increasing temperature and solar radiation, transpiration of the trees was enhanced, and the sap flow velocity was accelerated. The water stored in the trunk during the previous night might be utilized preferentially, so the water in the trunk was reduced continuously, leading to the shrinking of the stem, and the sap flow velocity increased continuously at this stage. Later, as the temperature and solar radiation decreased, transpiration also decreased gradually, the sap flow velocity began to show a decreasing trend, the stem began to expand slowly until the temperature and solar radiation increased, the sap flow velocity increased, and the stem contracted again.

As seen in Table 3, the DBH microvariation and TA, RH, RA, and VPD of P. tomentosa and S. babylonica were all highly significantly correlated. Both were highly significantly negatively correlated with TA, WS, and VPD and highly significantly positively correlated with RH.

A container full of soil without trees planted was used as a blank control, and the marsupial bottles continuously supplied water to the P. tomentosa and S. babylonica planted in the containers and the blank control. It can be seen from Figure 4 that the higher the depth of the blank experimental soil layer is, the greater the soil water content, and the soil water content at different depths showed an increasing trend, and the greater the depth of the soil layer is, the more obvious the trend of soil water content change. As soil capillary water was transported upward under the action of matrix potential and gravitational potential, its rising rate gradually decreased with time under continuous water supply. For P. tomentosa and S. babylonica, the lower the depth of the soil layer is, the lower the soil water content. Water was transported from bottom to top, and its soil water content was lower as it was closer to the surface soil. The trend of soil water content changes at three different depths was not the same, which might be caused by the large absorption of surrounding soil water by the roots. The soil water content was highest at depths of 20–30 cm, and the trend of change was not obvious, indicating that there might be a small amount of roots distributed in this layer, while the trend of soil water content at depths of 10–20 cm was obvious and similar to the trend of sap flow velocity, which indicated that the roots of P. tomentosa and S. babylonica may be mainly distributed in this soil layer. The upward trend of the change of soil water content at 0–10 cm depth may be due to sufficient water to make the soil water continuously transported upward, even though the phenomenon of water absorption by the roots existed at this soil layer, but in general it did not affect the continuous rise, and in the process of rising and constantly approaching the soil water content at 10–20 cm, it could finally reach a relative equilibrium state.

The correlations between the sap flow velocity and soil water content at different depths were determined for P. tomentosa and S. babylonica, and the results showed that the sap flow velocity of both species was highly significantly negatively correlated with the soil water content at 10–20 cm (P < 0.01). Figure 5 shows that the sap flow velocity of P. tomentosa and S. babylonica showed an opposite trend with the soil water content at 10–20 cm. As the roots of trees may be widely distributed at this depth, with the increase in transpiration, the trees gradually absorb a large amount of water, and when the root absorption rate is greater than the water transport rate, the soil water content begins to decline; when the transpiration rate gradually decreases, the root absorption is smaller than the soil water transport rate, and then the soil water content gradually increases.

TDP was used to measure the sap flow, and LLE was used to simultaneously measure transpiration. The sap flow measured by TDP of the two species was positively correlated with the transpiration measured by LLE (P < 0.01), showing a consistent pattern of variation, but the transpiration was significantly greater than the sap flow (Figure 6), indicating that the TDP-measured sap flow underestimated the true values.

The transpiration and sap flow of the two species were linearly fitted separately (Figure 7). The sap flow of P. tomentosa and S. babylonica were both smaller than the transpiration, with error values of -60.96% and -63.37%, respectively. Therefore, it is necessary to correct the original Granier’s formulas.

The correction equations were obtained by fitting a power function with the TDP-measured K as the horizontal coordinate and the LLE-measured transpiration rate as the vertical coordinate. The specific correction equations for P. tomentosa and S. babylonica were F_d = 0.0287K^1.236 (R² = 0.941) and F_d = 0.0145K^0.852 (R² = 0.904), respectively. Compared with Granier’s original formula F_d = 0.0119K^1.231, the coefficient α of the correction formula for P. tomentosa (0.0287) is larger than that of Granier (0.0119), and the correction coefficient β (1.236) is similar to that of Granier (1.231); the coefficient α (0.0145) of the correction formula for S. babylonica is larger than that of Granier (0.0119), and the correction coefficient β (0.852) is smaller than that of Granier (1.231). The K values and transpiration rates of P. tomentosa and S. babylonica were integrated to establish the combined correction equation: F_d = 0.0235K^1.080 (R² = 0.957). Compared with Granier’s original formula F_d = 0.0119K^1.231, the coefficients α (0.0235) and β (1.080) of the combined correction formula were both different from the coefficients of Granier’s original formula (Figure 8).

There were significant positive correlations between the sap flow measured by Granier’s original formula, specific correction formulas, combined correction formula, and transpiration measured by LLE of P. tomentosa and S. babylonica (p < 0.01); however, the sap flow calculated by Granier’s original formula was obviously lower than that calculated by the other three methods (Figure 9). Compared with Granier’s original formula, both the specific correction formulas and the combination correction formula were closer to the transpiration value measured by LLE.

The sap flow calculated by Granier’s original formula and specific correction formula and the transpiration measured by LLE of P. tomentosa and S. babylonica were fitted linearly (Figure 10). The results showed that the sap flow calculated by the specific correction formulas was closer to the reference line of 1:1, while that calculated by Granier’s original formula was obviously lower. Compared with those of transpiration, the errors of sap flow calculated by the specific correction formulas were 6.18% and 5.86%, respectively, and the errors were obviously reduced.

As shown in Figure 11, the overall sap flow calculated by the combined correction formula for P. tomentosa and S. babylonica was closer to the 1:1 reference line, with an error of -3.97%. The errors of the combined correction formula for P. tomentosa and S. babylonica were -12.76% and -2.32%, respectively, which was more accurate.

To further verify the accuracy of the specific and combined correction formulas for P. tomentosa and S. babylonica, the validity of the formulas was tested for deviations from the true transpiration rates using the D-index, RMSE, MAE, and MBE (Table 4). The combined correction formula and specific correction formulas of P. tomentosa and S. babylonica were both valid and similar, with the transpiration rates measured by LLE.