Effects of size and microclimate on whole-tree water use and hydraulic regulation in Schima superba trees

Site description

The experiments were conducted in a S. superba plantation (23°10′N, 113°21′E, 41.4 m alt) that was planted in the mid-1980s and located in the ecological observation station of the South China Botanical Garden, Chinese Academy of Sciences, Guangzhou. The study area had a subtropical monsoon climate, with an annual mean temperature of 21.8 °C, and mean temperature of 32.7  °C in the hottest month of July, and 9.8 °C in the coldest month of January. Annual mean rainfall was 1750 mm, with April to September tending to be wet, and October to March of the next year tending to be dry. In 2010, the annual mean rainfall was 2148.4 mm and in 2011, annual mean rainfall was 1421.2 mm (Guangzhou Bureau of Statistics, 2010; Guangzhou Bureau of Statistics, 2011). The soil was loamed with an organic content of 2.3% and total nitrogen content of 0.07%. This S. superba plantation was planted at a density of 603 plants/hm², and the mean height of the stand at the time of the study was 12.7 m. The experimental site covered approximately 2885.6 m².

Tree architecture characteristics

Characteristics of the sampled trees were measured in April 2011 to avoid age effects on transpiration (Table 1) (Zhang et al., 2014). Trees 1–15 were used to estimate transpiration and were divided into four groups based on diameter at breast height (DBH) (Table 2) Trees 16–21 were used to measure leaf water potential and hydraulic conductance, and were classified into three groups according to DBH (Table 2). Trees 17–18 and 21 were used to analyze stomatal conductance. Tree height (H, m) was measured using a clinometer (Suunto Oy, Vantaa, Finland). DBH was measured using a diameter tape. Cores from the stems of five trees outside and close to the plot were taken using an increment core borer. The sapwood depth was measured using a caliper as sapwood and heartwood in S. superba are easy to identify visually. An exponential regression (see Eqs. (3-1), (2) and (3) below) between the sapwood area (A_s, m²) and DBH was established based on the above measures, which was used to calculate the A_s of the 15 trees in the plot. Similarly, leaf area (A_L) was calculated using a log function (see (6) below) established from the five trees outside and close to the plot.

Environmental measurements

We continuously monitored wind speed (WS, m s⁻¹), air temperature (TA, °C), relative humidity (RH, %), and photosynthetically active radiation (PAR, µmol m⁻² s⁻¹) above the S. superba forest canopies from an observation tower set up within the plantation. Soil moisture (SM, m⁻³ m⁻³) was measured 30 cm below the ground surface from October 2010 until December 2011. Wind speed was measured using cup anemometers (AN4; Delta-T Devices, UK). TA and RH were measured using a temperature and humidity sensor (RHT2V-418; Delta-T Devices, UK). PAR was measured using a Li-Cor quantum sensor (LI-190SA; LI-COR, USA). SM was measured using three frequency domain sensors (SM200; Delta-T Devices) at a depth of 30–40 cm, which were set in a triangle around trees 17–21. The output values were measured every 30 s, and 10-min mean values were logged using a DL2e Delta-T logger (DL2e; Delta-T Devices).

The vapor pressure deficit (VPD, kPa) combined with the parameters TA and RH was calculated using the following formula (Campbell & Norman, 1998): (1)${\displaystyle VPD={\mathrm{ae}}^{\left(\frac{bTA}{TA+c}\right)}\left(1-RH\right)}$ Where, a, b, and c are fixed parameters (0.611 kPa, 17.502 [unitless], and 240.97 °C, respectively).

Sap flux density measurements

Sap flux density (J_s) was continuously measured from 2010 until 2011 using a homemade thermal dissipation probe (TDP). The Grainer-type probes were inserted radially at a depth of 20 mm into the stem of 21 trees at 1.3 m aboveground, one pair for each tree. The upper heated probe was supplied with a constant DC current (120 mA). The unheated probe was installed 10–15 cm below the upper sensors. The temperature difference between the paired sensors was recorded using the same logger as environmental measurements. All probes were installed on the north-facing side of the trees and covered with a plastic case, and shielded using a radiation-insulating film in order to minimize the possible effect of direct sunshine.

Sap flux density (J_s, g m⁻² s⁻¹) was weighted using the following equation: (2)${\displaystyle J_s=\lambda \left(40\ast J_{0-40}+\left(d-40\right)\ast J_{40}\right)∕d\left(d>40\right)or{J}_{0-40}\left(d⩽40\right)}$ where, J₀₋₄₀ (Sap flux density at 0–40 mm) is calculated using the empirical equation proposed by Granier (1987), d is sapwood thickness, and λ are correction factors of 1.164, 1.128, and 1.096 for Ranks 1, and 3, respectively. J₄₀ (sap flux density at ⩾40 mm) was calculated following Mei et al. (2010b): (2-1)${\displaystyle J_{40}=0.45\times J_{0-40}}$ Whole-tree transpiration (F_s, g d⁻¹) was calculated as: (3)${\displaystyle F_s=\sum \left(J_{0-40}\times A_{0-40}+J_{40}\times A_{40}\right)\times t}$ where, t is 600 s (data were averaged and stored every 10 min in the logger), A₀₋₄₀ and A₄₀ are the sapwood areas in the outer xylem (0–40 mm) and inner xylem (⩾40 mm), respectively, which were calculated as: (3-1)${\displaystyle A_{0-40}=0.166•{\left(DBH\right)}^{1.336}}$ (3-2)${\displaystyle A_{40}=A_s-A_{0-40}}$ where A_s is the total sapwood area (m²), calculated using the regression equation relating it to DBH (m) as follows: (3-3)${\displaystyle A_s=0.465•{\left(DBH\right)}^{1.794}.}$ Monthly water use (∑F_s, g mon⁻¹) was calculated as: (4)${\displaystyle \sum F_s=30\times F_s}$ Monthly canopy transpiration (∑E_L, g m⁻² mon⁻¹) was calculated as: (5)${\displaystyle \sum E_L=30\times E_L=30\times \frac{F_s}{A_L}}$ Where, E_L is canopy transpiration (g m⁻² d⁻¹), A_L is the total leaf area (m²), which was calculated using a regression equation relating it to DBH (m) as follows (Schäfer, Oren & Tenhunen, 2000): (6)${\displaystyle log{A}_L=1.672•logDBH+3.199}$ Annual stand transpiration (E_s, mm y⁻¹) was calculated from ∑∑F_s per ground area over 1 year. ΔE_L was the difference between predawn and midday E_L

Leaf water potential and stomatal conductance

The branches sampling procedure was conducted on trees 16–21 over 5–7 continuous sunny days in five time periods, Oct 2010, Jan, Apr, Jul, and Oct 2011. Leaf water potential (ψ_L, MPa) was measured using a portable pressure chamber in two ways: at 1-hour intervals from 05:00–06:00 to 20:00 in Oct 2010, and point measuring at 05:00–06:00, 13:00, and 20:00 in the other time periods. This minimized the destructive sampling of tree branches. Predawn leaf water potential (ψ_pd, MPa) and midday leaf water potential (ψ_md, MPa) referred to ψ_L at 05:00–06:00 and 13:00, respectively. Iso/anisohydraulic regimes during the time periods were determined using the linear framework of (Martinez-Vilalta et al., 2014): (7)${\displaystyle {\psi }_{md}=\Lambda +\sigma {•\psi }_{pd}}$ Where Λ is the intercept of the relationship, and σ is the slope. A σ = 0 implies strict isohydry, σ = 1 implies strict anisohydry, σ1 implies extreme anisohydry and 0 < σ < 1 implies partial isohydry.

Instantaneous stomatal conductance (g_s, m s⁻¹) was simultaneously measured on trees 17–18 and 21 using a Li-6400 Portable Photosynthesis System, at the same time intervals and periods as ψ_L.Δg_s represented the difference between predawn and midday g_s. All measurements were taken three times.

Canopy conductance and whole-tree hydraulic conductance

Canopy conductance (g_c, m s⁻¹) was calculated by inverting the following equation (Lu et al., 2003): (8)${\displaystyle \lambda E_c=\frac{\delta R_n+K_t\rho C_p\mathrm{V\; PD}{g}_a}{\lambda \left[\delta +\gamma \left(1+g_a∕g_c\right)\right]}}$ where, λ is the latent heat of vaporization (2.39 MJ kg⁻¹), E_c (mm h⁻¹) is estimated from whole-tree transpiration divided by the projected canopy area of 3.96 m², δ is the slope of the relationship between saturated vapor pressure and temperature (kPa °C⁻¹), R_n isthe net radiation above the forest canopy (MJ m⁻² h⁻¹), K_t is a unit conversion equal to 3,600 s h⁻¹, ρ is air density (1.29 kg m⁻³), C_p is the specific heat of air (1.013 MJ kg⁻¹ °C ⁻¹), γ is the psychometric constant (0.066 kPa °C⁻¹), and g_a isaerodynamic conductance (m s⁻¹), which was calculated using the method in Delzon et al. (2004).

Whole-tree sapwood-specific hydraulic conductance K_p and leaf-specific hydraulic conductance K_L (g m⁻² s⁻¹ Mpa⁻¹) were indirectly estimated following Cochard, Bréda & Granier (1996): (9)${\displaystyle K_p=J_d∕\left({\psi }_{pd}-{\psi }_{md}\right)\text{or}{K}_L=J_l∕\left({\psi }_{pd}-{\psi }_{md}\right)}$ where, J_d and J_l are the difference in sap flux density of sapwood and leaf area between predawn and midday (g m⁻² s⁻¹), respectively. We ignored time lags between sap flux density and transpiration (Zhao et al., 2016).

Data analysis

January and July represented the dry and wet seasons, respectively, and October 2010 and 2011 represented annual differences in south China, respectively. All statistical analyses were performed using SPSS 15.0 (SPSS, Chicago, IL, USA) and Origin 8.1 (OriginLab, Northampton, MA, USA). The paired-samples t-test was used to analyze the seasonal and annual differences in K_p and K_L. Major factors affecting E_L and F_s were determined using multilinear regression. All data were standardized before modeling. The strengths of relations were evaluated using the absolute values of the coefficients of separate models.

Effects of tree size on daily sap flux density, whole-tree and canopy transpiration

J_s, F_s, and E_L varied month by month, although all were at their maximums in the wet season and at their minimums in the dry season. The inter-class daily maximum of J_s followed weakly their size in 2010 and 2011 (Figs. 1A–1B). The total occurring rate of J_s ranking of Ranks 1 and 4 mapping their size ranking were 33.3% and 25.0%, respectively, which were higher than 4.2% and 12.5% of trees ranked 2 and 3 during the 2 years. Moreover, the separate mapped rate for Ranks 1-4 in 2010 were higher than that in 2011 except Rank 2, i.e., 41.7%, 0%, 16.7% and 33.3% compared with 25.0%, 8.3%, 8.3% and 16.7%, in turn. The total and separate mapped rates on E_L for the four Ranks were similar to that of J_s (Figs. 1C–1D). In contrast, the mapped rates on F_s increased up to 100% and 100% in Ranks 1 and 4, respectively, and up to 58.3% and 58.3% in Ranks 2 and 3, respectively (Figs. 1E–1F). Meanwhile, the separate mapped rates for all the four Ranks in 2010 were no different from that in 2011. F_s was evidently more close to present the assumption of the larger the tree, the greater the transpiration among the three parameters.

Seasonal patterns of rank water use and whole-tree transpiration

The ∑F_s and ∑E_L of trees ranked 1–4 generally showed a low-high-low trend, which followed the change in rainfall over the dry season (one to three months)-wet season (four to nine months)-dry season (10–12 months) (Figs. 2A–2B). The mean annual ∑F_s for Ranks 1, 2, 3, and 4 were 1301.28, 472.69, 469.28, and 135.34 g mon⁻¹ in the wet season, and 924.25, 348.05, 345.53, and 106.65 g mon⁻¹ in the dry season. The mean annual ∑E_L for Ranks 1, 2, 3, and 4 were 9.75, 5.57, 8.21, and 5.54 g m⁻² mon⁻¹ in the wet season, and 6.92, 4.10, 6.04, and 4.37 g m⁻² mon⁻¹ in the dry season. ∑F_s and ∑E_L of all tree ranks in the wet season were averagely 1.37 and 1.35 times higher than in the dry season during 2010 and 2011. That is, water use in the fast growth period was almost always more than 50%.

Annual patterns of stand water use

The total water use for Ranks 1, 2, 3, and 4 were 312.8, 185.9, and 126.9 mm in the wet season, and 262.4, 145.9 and 116.5 mm in the dry season in 2010. The mean daily E_s was 0.86 mm d ⁻¹ in 2010 and 0.72 mm d⁻¹ in 2011. Moreover, F_s of all ranks declined to a certain extent, i.e., Ranks 1 and 2 declined from 40.2 and 15.8 g d⁻¹ in 2010 to33.0 and 11.2 g d⁻¹ in 2011, respectively, while Ranks 3 and 4 declined from 13.9 and 4.0 g d ⁻¹ in 2010 to 12.9 and 3.9 g d⁻¹ in 2011, respectively.

The stand ∑F_s was determined mainly by soil moisture (SM) and rainfall, with the effect of SM greater than that of rainfall (Table 3). Of the five factors (PAR, VPD, SM, WS, and rainfall), only SM was significant in the linear regression. Of the remaining four factors, rainfall was more significant than WS in the linear regression. Of PAR, VPD, and WS, no significant relations were observed. However, ∑E_L was determined mainly by SM, PAR, and rainfall, with the effect of SM >PAR >Rainfall >WS (Table 3). No relations were observed between ∑E_L, VPD, and WS.

Leaf water potential

Significant seasonal differences were observed in the ψ_md − ψ_pd for Ranks 1, 2 and 3 (Table 4). Daily ψ_pd and ψ_md ranged from −0.45 to −0.20 MPa and −0.97 to −0.61 MPa in Jan with SM at 23.39∼23.91 m⁻³ m⁻³, and −0.25 to −0.16 MPa and −1.47 to −0.68 MPa in Jul with SM at 34.57∼35.03 m⁻³ m⁻³ in 2011 (Fig. 3A). However, no significant annual differences were found in the ψ_md − ψ_pd for the three ranks (Table 4). Daily ψ_pd ranged from −0.42 to −0.18 MPa and −0.40 to −0.20 MPa, and the ψ_md ranged from −1.27 to −0.35 MPa and −1.33 to −0.56 MPa in the Oct 2010 and 2011, respectively.

The maximum of ψ_L (actual water potential at predawn) was inversely proportional to the DBH (n = 6, R² = 0.80, p < 0.05), whereas no relation was found between the minimum ψ_L (actual water potential at midday) and DBH. That is, bigger trees had more water available in the rhizosphere. However, the strongest potential capacity to access soil water may not have been related to tree size at the time of measuring. According to the model relating ψ_pd and ψ_md proposed by Martinez-Vilalta et al. (2014), extreme strict anisohydry (σ > 1) was only found in Rank 3 in Jan and Rank 2 in Apr in 2011 (n = 10, R² = 0.54 and 0.48, p < 0.05).

Stomatal conductance and canopy conductance

Daytime hourly mean g_s (trees 17, 18 and 21) ranged from 0.011 to 0.151 m s⁻¹. A linear relation between g_s and PAR was observed in all five time periods. In contrast, a linear relation between g_s andVPD, SM, and WS was only observed in Oct, Apr and Jul, Jan and Oct in 2011, respectively.

Canopy conductance varied seasonally (much higher in the wet season than in the dry season), which was in agreement with Kumagai et al. (2004) and Barradas et al. (2005). The g_c for the whole plantation ranged from 0.00058–0.034 mm s⁻¹ in the wet season and 0.000092–0.070 mm s⁻¹ in the dry season. Individually, the g_c of trees ranked 1–4 ranged from 0–0.00069 mm s⁻¹. Stomatal conductance of Rank 1 trees was positively related to g_c in each of the five time periods (Table 5).

Hydraulic conductance and canopy conductance

Daily K_p and K_L did not vary with tree size. A proportional relationship between K_p or K_L and DBH was observed in Oct 2011 (n = 6, R² = 0.55 and 0.64, p < 0.1). Moreover, both the maximums and minimums of K_p and K_L occurred in Rank 2 and 3 trees.

There were significant differences in K_p and K_L for Rank 2 and 3 trees between 2010 and 2011 (Table 4). The K_p and K_L for Rank 2 trees in Oct 2010 was 70.69 and 86.57 g m⁻² s⁻¹ MPa⁻¹, respectively. The K_p and K_L for Rank 3 trees in Oct 2010 was 1.92 and 2.23 g m⁻² s⁻¹ MPa⁻¹, respectively. The K_p and K_L for Rank 2 trees in Oct 2011 was 44.37 and 36.40 g m⁻² s⁻¹ MPa⁻¹, respectively. The K_p and K_L for Rank 3 trees in Oct 2011was 1.20 and 0.95 g m⁻² s⁻¹ MPa⁻¹, respectively

There were significant seasonal differences for Rank 1 and 2 trees in the 2 years (Table 4). The mean daily K_p and K_L for Rank 1 trees in the wet season was 39.33 and 23.40 g m⁻² s⁻¹ MPa⁻¹, respectively. The mean daily K_p and K_L for Rank 2 trees in the wet season was 0.032 and 0.013 g m⁻² s⁻¹ MPa⁻¹, respectively. The mean daily K_p and K_L for Rank 1 trees in the dry season was 57.84 and 44.37 g m⁻² s⁻¹ MPa⁻¹, respectively. The mean daily K_p and K_L for Rank 2 trees in the dry season was 0.016 and 0.012 g m⁻² s⁻¹ MPa⁻¹, respectively.

In addition, ΔE_L was well related to Δg_s and K_L in Oct 2010 and 2011, and related to K_L in Jan and Jul 2011, respectively (Table 3). Daily ΔE_L, Δg_s and K_L ranged from 0.65–10.02 g m⁻² d⁻¹, 0.01–0.20 m s⁻¹, and 0.0082–0.051 g m⁻² s⁻¹ Mpa⁻¹, respectively.

Scaling up water use from the individual to the stand

F_s showed stronger relation with the ranks than J_s and E_L. There are fewer studies comparing this parameter, but more interspecies studies comparing daily mean J_s. Horna et al. (2011) compared J_s of seven species with no size classes in July and December between 2007 and 2008, but found no significant differences. Under a common setting criterion (i.e., depth of the xylem, azimuth), J_s showed high interspecific differences (Ewers et al., 2002; O’Brien, Oberbauer & Clark, 2004; Daley & Phillips, 2006; Horna et al., 2011). Even within species, J_s may vary with tree social position (Ambrose et al., 2010), cultural type (Kunert et al., 2012), local condition (Granier et al., 1990; Motzer et al., 2005), age (Oguntunde & Oguntuase, 2007), and tree-size parameters (Loustau et al., 1996; Ewers et al., 2002; Kume et al., 2010). Therefore, future research should sample a large number of trees to improve the accuracy (Oishi, Oren & Stoy, 2008). Many studies have provided evidence of the mechanism of this relation. Some studies suggested that variability of J_s is related to changes in a single environmental factor (e.g., Granier et al., 1990), while others support the influence of multiple environmental factors (Bovard et al., 2005; Daley & Phillips, 2006). Many studies that used F_s instead of J_s might have concluded different sap flow measurements, which also likely amplified individual tree differences (Roberts, Vertessy & Grayson, 2001; Tausend, Meinzer & Goldstein, 2000; Motzer et al., 2005; Yunusa et al., 2010; Zeppel et al., 2010). We found that F_s showed “the larger the tree, the greater the transpiration” more effectively than J_s.E_L takes into account the importance of leaf area. Some studies have found that E_L is an important determinant of tree water use (Myers et al., 1996; Radersma, Ong & Coe, 2006). In our study, clear differences in E_L were also found between ranks, i.e., the leaf-area of Rank 1 was 1.48 times that of Rank 3, but transpired 51.6% more water in Jul 2011.

The seasonal decline in water use with leaf area has been shown in many studies. This has been explained as preventing canopy desiccation (Dye, 1996; Farrington et al., 1994; Hutley, O’Grady & Eamus, 2000). In this study, tree water use only accounted for 9.9% and 13.9% of rainfall in the 2010 and 2011 wet seasons, respectively, and 48.4% and 31.2% of rainfall in the 2010 and 2011 dry seasons, respectively. However, we only showed a decline in tree water use in the dry season after ignoring leaf growth. It showed the effects of external environmental factors more effectively than current approaches. Llorens et al. (2010) also showed that transpiration decreased significantly during the drier summers of 1998 and 2000, compared with the wetter summer of 1997.

Many studies have reported that SM determines transpiration (Jiao et al., 2015; Gazal et al., 2006; Chang et al., 2014; Zhao & Liu, 2010). Some studies have observed a decline in transpiration with decreasing SM. Gartner et al. (2009) found that birch and Norway spruce trees reduced their transpiration in response to drought. In this study, tree water use and canopy transpiration were significantly affected by SM at all daily and monthly scales. The decreasing transpiration could be partly attributed to the decrease in SM in the 2011 dry season, when rainfall decreased by 727.2 mm, compared with the 2010 wet season, although there were no SM data for 2010. Other studies have shown that rainfall influences tree transpiration when soil water in the upper profile is insufficient (O’Grady, Eamus & Hutley, 1999). Even in the afternoon, tree transpiration may vary before and after a rain event (Wang et al., 2017). Nevertheless, this effect did not show a simple proportional relation. Huang & Zhang (2016) reported that 0–5-mm precipitation increased transpiration, while >5mm precipitation decreased transpiration in two xerophytic shrubs. Conversely, water use was weakly related to rainfall, indicating that the trees strongly depended on groundwater (Morris, Mann & Collopy, 1998). The inconsistent effect of rainfall and SM was possibly related to root depth. Zhao & Liu (2010) showed that soil water content at 10–20-cm depth depended significantly on rainfall. It is likely that plants depend on water uptake up to 50 cm soil depth (VanSplunder et al., 1996; Lagergren & Lindroth, 2002). Our SM measurements at 30–40-cm depth determined the decrease in S. superba transpiration in the dry year (2011), which was possibly attributable to the combined effects of SM and rainfall. This was also shown in the decrease of 16.3% in E_s between the two years, which was lower than 33.8% of precipitation.

The total water use of S. superba in 2010 and 2011 was approximately 14.6% and 18.5% of the average annual rainfall, respectively. The low level of water consumption was possible attributed to its growth near over-mature stage. From studies of broad-leaved plantations in south China, the total annual mean daily E_s of 0.79 mm d⁻¹ of S. superba (30–35a) was higher than the 0.59 mm d⁻¹ of Acacia mangium (19a) at the Heshan experimental station in Guangdong Province (22textdegree40′N, 112°54′E, 226 m alt), and lower than the 1.01 mm d⁻¹ of Eucalyptus urophylla ×E. grandis (4–5a) at the Huangmian Forest Farm in Liuzhou, Guangxi Province (24°45.8′N, 109°53.6′E, 80 m alt), and 1.48 and 1.53 mm d⁻¹ of E. urophylla at Hetou (21°05′N, 109°54′E, 25 m alt) and Jijia (20°54′N, 109°52′E, 70 m alt), respectively, in Leizhou Peninsula, Gouangdong Province (Morris et al., 2004; Ma et al., 2008; Zhu et al., 2015). As they have similar latitude/longitude and climate background, water use of plantations poses no threat to the ground water balance, irrespective of the mix of exotic and native species, although there are differences (Myers et al., 1996).

Stomatal and hydraulic regulation of transpiration from the individual to the stand

Regulations of g_s included physical factors, such as ABA, PH value, and flagellin (Small & Maxwell, 1939; Zhang, He & Assmann, 2008), and environmental factors, such as VPD, light intensity, air humidity, and atmospheric CO₂ concentration (Jarvis, 1976; Aasamaa & Sõber , 2011; Monteith, 1995). Johnson et al. (2001) reported that photosynthetic photon flux density (PPFD) was most correlated with g_s in Acer saccharum. Monteith (1995) examined three phases of diurnal g_s, corresponding to diurnal VPD, and found that low VPD increased the minimal transpiration limited by stomata (phase C), and larger values of VPD caused a large decline in g_s (phase B), while minimal medium values of VPD increased transpiration (phase A). Although g_s correlated well with VPD only in Oct 2011, our results followed Monteith (1995).

Extreme anisohydry was found in S. superba, and that seemed to be a dynamic mechanism and varied with ranks and periods. However, it is doubtful that S. superba, which grew in South China, showed extreme anisohydry. Martinez-Vilalta et al. (2014) found only five species showed extreme anisohydry (σ > 1) in 102 species. Moreover, they noted that the phenomenon may occur in phreatophytes and drought-deciduous species. In our study, ψ_md–ψ_pd was approximately maintained seasonally constant but ψ_pd was correlated with soil water availability, which be in consistent with an isohydrodynamic behavior proposed by Franks, Drake & Froend (2007). They suggested that the behavior was linked to a combined hydraulic regulation with stomatal control and plant hydraulic conductance. We also found there were annual and seasonal coordinated regulations with stoma and hydraulic conductance on canopy transpiration (Table 3). However, it was difficult to explain the difference between Oct 2010 and Jul 2011 only linking SM and rainfall, because the latter had higher SM and rainfall. It is still suggested that a higher SM in the dry season (compared with Jan and Oct 2011) partially resulted in stomatal inverse control on leaf water potential (Fig. 3B), even though the inverse control was not in agreement with ψ_L positively controlling g_s reported by Comstock & Mencuccini (1998), Williams & Araujo (2002) and Ripullone et al. (2007).

Actually, instantaneous g_s did not effectively represent the bulk surface conductance unless it was sampled from different canopy layers. Stomatal conductance from the same canopy layer could be scaled up to calculate canopy stomatous conductance in our study (Table 5). It was merely too low values of g_c in our study compared to those reported for Qinhai spruce (0.3–51.3 mm s⁻¹; (Chang et al., 2014)), Scots pine stands (13–28 mm s⁻¹; Granier et al., 1996; Dolman et al., 1998; Sturm et al., 1998), Norway spruce stands (10–13 mm⁻¹; Lu et al., 1995; Alsheimer et al., 1998; Cienciala et al., 1998), and Fagus sylvatica (3.3–18.5 mm⁻¹; Magnani et al., 1998). However, our results concurred with the ranges reported for Prunus armeniaca (0.0012–0.0024 mm s⁻¹; Barradas et al., 2005), and a mixed stand composed mainly of Eucalyptus crebra and Callitris glaucophylla (maximum 0.0083 mm s⁻¹; Whitley et al., 2009), and a Pinus canariensis forest (0.0033 mm s⁻¹; Kučera et al., 2017). The reported ranges of values vary greatly, not only by species (Köstner et al., 1992), age (Forrester, Collopy & Morris, 2010), temporal scales (Bernier et al., 2006), soil, root and canopy components (Morris et al., 2004), but also by different models of calculation using the same method with different parameters (e.g., Morris, Mann & Collopy, 1998; Zeppel & Eamus, 2008). In contrast, the inter-class daily conductance was smaller than that of whole trees, and the highest average conductance was 0.00069 mm s⁻¹. Moreover, the maximum g_c always occurred in Rank 3 trees, as there were too few trees in some of the ranks, especially Rank 1. Thus, the results could not confirm that the larger trees determined the whole conductance (Martin et al., 1997).