The experiment was conducted at the Zhanggutai National Desertification Control Experimental Station located at the southern edge of the Horqin Sandyland area, Liaoning Province, China (122°22′E, 42°43′N and 226.5 m a.s.l.) during 2013-2018. The climate is semi-arid and continental monsoon. Over the last 30 years, the mean annual air temperature was 7.9°C, the mean frost-free period was 155 days, the mean annual pan evaporation was 1553 mm, and the mean annual precipitation was 475 mm. Approximately, 92% of this precipitation fell from May to September (the main growing season). The soil had an aeolian sand texture, consisting of 84% sand particles (> 0.05 mm), 9% silt particles (0.05–0.002 mm) and 7% clay particles (<0.002 mm). The soil bulk density was 1.61 g⋅cm^–3, the capillary porosity was 33% and the soil was barren with a mean organic matter content of 0.65 g⋅kg^–1 in the upper 1.0 m soil layer (Dang et al., 2019a).

We selected two types of monocultural plantation forests that both served as shelterbelts, MP and CP, as sample forests in this study. The two species were planted in the same year, both reaching approximately 46 years of age in 2013. The stand densities were approximately 400−450 stems⋅ha^–1.

CP is a unique coniferous species in China (Jiang et al., 2002; Cai et al., 2020). Natural CP forests represent an important forest type in the warm/temperate deciduous forest regions of China (approximately 31°−43°N, 103°20′−124°45′E) (Cai et al., 2020). In this species’ area of natural distribution, the multiple-year annual average temperature is 1−2°C, the recorded highest annual average temperature is 14°C, the average lowest temperature (January) ranges between −20°C and -4°C, the annual precipitation is between 400 and 800 mm, and the altitude ranges from 400 to 1000 m a.s.l. The soil types are mainly cinnamon soil, brown soil and gray-cinnamon soil. The diameter growth rate of natural CP trees in forests shows a significant negative correlation with latitude over its entire distribution area and significant positive correlation with average temperature and precipitation in January but does not exhibit any significant correlations with either longitude or altitude.

MP is one of the regional varieties of Scots pine, found naturally in Russia, Mongolia and the Chinese Da Xiggan Ling Mountains and Hulunbuir Sand (approximately 47°35′–53°33′N, 118°58′–127°10′E) (Zhao and Li, 1963). The annual average temperature in this species’ natural distribution area is −2.5 to −2°C, the average lowest temperature (January) is −30 to −24°C, the average annual precipitation is between 325-600 mm and the altitude is between 600−400 m a.s.l. The soil types are mainly gray soil and sandy soil. The MP trees in natural forests are long-lived and reach maturity at approximately 80-90 years (Zhao and Li, 1963).

Two experimental plots, each enclosed by a 15 m × 15 m fence, were each established in the middle of approximately 30 ha sample forest, one of CP and one of MP. According to the recommendations for the numbers of sample trees for sap flow measurements (Köstner et al., 1996; Kume et al., 2010), we selected eight sample trees with similar sizes in each plot. The diameter at breast height (DBH) averaged approximately 24.1 ± 2.31 cm (mean ± SE) in the MP plot, which was significantly higher than the mean DBH value of 20.8 ± 1.3 cm in the CP plot (p < 0.01). More detailed information on the sample trees can be seen in Table 1. Measurements for the meteorological variables and soil moistures had continued since 2013, but the sap flow measurements were performed only during the growing season in 4 years, i.e., 2014, 2015, 2017, and 2018.

Crown widths were measured east-to-west and south-to-north and are listed in Table 1. These were used to calculate the crown projected areas of the sample trees by applying the following relationships based on to our stand investigation:

For MP:

For CP:

where, A_c (cm²) is the crown projected area and DBH (cm) is the DBH (∼1.3 m).

Micrometeorological factors, including radiation, air temperature (T_a), relative humidity (RH), wind speed and precipitation, were measured by an automatic weather station (AR5, Avalon Scientific, Inc., NJ, United States), located in an open area approximately 50 m away from the experimental plots. Variables were recorded every 10 min using a data logger and subsequently averaged (or summed) to generate hourly and daily values. Hourly vapor pressure deficit (VPD, kPa) was calculated based on T_a and RH (Campbell and Norman, 1998).

We adopted the standardized precipitation evapotranspiration index (SPEI) to describe the atmospheric drought severity (Vicente-Serrano et al., 2010; Beguería et al., 2014). We calculated the SPEI in the R package SPEI¹ based on precipitation and temperature data over a 12-month time scale during 2004-2018. We used SPEI to categorize dry and wet gradings according to the standards (Chen and Sun, 2015).

We measured the volumetric soil water content (θ, cm³⋅cm^–3) at 20-cm intervals in the upper 1.0 m soil layer at three locations in each experimental plot with ECH₂O EC-5 probes (METER Group, Inc., Pullman, WA, United States). The data were collected at 10 min intervals and averaged to hourly or daily scales. The sensor readings were site-specific and were calibrated using the following formula based on the soil-core method: θ = 0.99421 × θ_sensor + 0.00128 (R² _adj = 0.94, n = 202, p < 0.001). We calculated the relative extractable soil water (REW), which is defined as the quotient of the actual extractable water to the maximum extractable water, to describe the relative soil moisture conditions at the site (Granier et al., 1999):

where, θ_fc (%) is the field capacity (17.5% in the 0.0–1.0 m soil depth based on field observations) and θ_min is the minimum soil moisture during the experimental period (2.3%). θ¯ is the mean of soil moisture values from the corresponding soil layers (%). We adopted the threshold value of REW = 0.4 recommended in several reports to define soil water stress (Granier et al., 1999; Bernier et al., 2002; Poyatos et al., 2005), which corresponds to the θ value of 0.086 cm³⋅cm^–3 at our site. A more detailed threshold value of REW was later deduced from the model based on our field measurements.

Sap flux density (J_s, cm⋅s^–1) in the outer 3 cm width xylem layer was measured continuously using thermal dissipation sensors (Dynamax Inc., Houston. TX. United States). We installed the sensors 1.3 m above the soil level on the north sides of the stems. The distance between the two probes of each sensor was 0.04 m (see the specifications in the brochure)². The upper needle of a probe was heated with a constant power of 0.2 W. The sensors were shielded with reflective foil that extended 1.0 m below and 0.5 m above to minimize effects from incident radiation. We sealed the foil with the stem above the installation to prevent ingress of raindrops and stem-flow water. The temperature difference between the two probes was measured at 10-min intervals and recorded every hour using SQ2040 data loggers (Grant Instruments Ltd, Cambridge, United Kingdom). The measurements were recorded during an entire growing season each year. At the end of the growing season, we removed all probes from the trees and reinstalled them at the beginning of the next growing season (in early April) to minimize possible signal dampening (Moore et al., 2010). J_s was calculated using Granier’s original equation (Granier, 1987):

where, ΔT is the measured temperature difference between the heated and reference needles. ΔT₀ is the maximum ΔT when the sap flux density is close to zero, which is determined over approximately 10 consecutive measuring days, using a linear regression (Lu et al., 2004).

The total sap flow through the section of trunk instrumented was considered to be equal to the total transpiration from the canopy (Köstner et al., 1996). The canopy transpiration intensity (T_r, mm⋅day^–1) was calculated based on the measured sap flux density (J_s, cm⋅s^–1), the sapwood area (A_s, cm²) at the instrumented section, and the projected area of the crown (A_c, m²):

where, n is the number of sample trees (n = 8 for each of the two species in the study). J_s,i,j is the measured sap flux density in the outer 3 cm width of xylem of the jth-tree at the ith-hour. A_s,j is the sapwood area of the jth-tree calculated by Eq. (7) or Eq. (8) based on the DBH values. A_c,j is the projected crown area of the jth-tree calculated by Eq. (1) or Eq. (2) based on the DBH values.

For MP:

For CP:

where, A_s (cm²) is the sapwood area and DBH (cm) is the DBH (∼1.3 m).

In 2018, due to the great decrease in battery power capacity, insufficiency of sensor power occurred on some days in two independently observed plots, resulting in effective data being obtained only during the daytime when the sunlight rose to relatively strong levels. Days when these data on any one plot were incomplete were excluded when calculating daily T_r for comparisons between tree species.

We calculated canopy stomatal conductance (G_s, cm⋅s^–1) from canopy transpiration (T_r) on the crown projected area basis and VPD using the simplification of the inversion of the Penman-Monteith model (Monteith and Unsworth, 1990). G_s was calculated on an hourly basis:

where, λ is the latent heat of vaporization of water (MJ⋅kg^–1), γ is the psychrometric constant (kPa⋅°C^–1); ρ is the density of air (kg⋅m^–3) and c_p is the specific heat of air (MJ⋅kg^–1⋅°C^–1).

We defined the growing season for these pines at the site as being for the 6-month period from May 1 to October 30. We calculated G_s only when the daytime VPD was greater than 0.6 kPa (Ewers and Oren, 2000). Daytime was defined as when solar radiation exceeded 50 W⋅m^–2 (May and June). Considering the incompleteness of recordings of sap flow on some days in 2018, the maximum of hourly sap flux density in a day (J_s–max) was use to describe the xylem sap flow capacity, while only the days with complete recordings of sap flow were selected to calculate canopy transpiration. Average values of J_s, T_r, and G_s for sample trees were compared between tree species at different timescales using repeated one-way ANOVA at p < 0.05 or p < 0.01 significance level. Relationships between the variables studied were evaluated using correction and simple and nonlinear regression analyses. The daily T_r values of the two pine trees were linearly fitted monthly, and the slope k was derived. The relationships in which the slope k declined with the decreasing of REW were fitted with an exponential equation. The relationships between G_s and VPD were analyzed using boundary-line analysis performed with quantile regression in the statistical R package Quantreg³. We adopted Lohammar’s function (Oren et al., 1999) and the 95th quantile to reflect the boundary-line relationship between Gs and VPD. All statistical analyses were conducted with OriginPro (Version 2021, OriginLan Inc., Northampton, MA, United States).

During the 6-year period from 2013 to 2018, the annual precipitation in 2013 and in 2016 exceeded the 30-year average by approximately 17 and 31%, respectively, but in 2014, 2015, 2017, and 2018, it accounted for 81, 86, 71, and 93% of the 30-year average, respectively. These results indicate 4 years of relative drought from the perspective of annual precipitation. However, more specifically, the drought year sequence seems to have been interrupted by wet years, thus dividing the drought year sequence into two 2-year-long sets. The SPEI values for these 6 years were between −0.26 and 0.35, so their inter-annual variation was essentially consistent with the trend for the annual precipitation (Figure 1A). However, these 6 years belonged to the ‘near normal’ category except for 2016, which belonged to the ‘moderately wet’ category based on the SPEI.

The relative extractable soil water within the 0–1.0 m depth layer (REW) for both sites during the 6-year period varied with annual precipitation, indicating the direct effects of precipitation on soil moisture. In 2014, the REW for the MP plot was significantly lower than that for CP, 0.28 to 0.31 (p = 0.04), but in 2015, the REW in MP (0.34) was significantly higher than that in CP (0.30) (p = 0.03), indicating a small decrease in soil moisture in CP, while there was a 12% increase in precipitation during the 2014-2015 period. In 2017, the REW for the MP plot averaged 0.38, which was significantly lower than the value of 0.41 for the CP plot (p = 0.01). In 2018, the REW levels in both the MP and CP plots increased above 0.43 with a 38% increase in precipitation (Figure 1B). Although the total amount of precipitation in 2017 was the lowest in the 4-year period, the soil moisture at the two sites was better than that during the 2014−2015 period due to the greater precipitation in 2016 (623.6 mm).

The inter-seasonal dynamics of daily REW in each year of the 4-year period showed similar patterns at both the MP and CP sites (Figure 2). Calculations indicate that the drought days for REW < 0.4 at the CP site were 138, 132, 96, and 90 days in 2014, 2015, 2017, and 2018, these accounting for 75, 72, 52, and 49%, respectively, of the total days in the 6-month growing season (May 1 to October 31). The drought days for REW < 0.4 in MP were 145, 111, 115, and 88 days, in 2014, 2015, 2017, and 2018, accounting for 79, 60, 63, and 48%, respectively, of the total days in the 6-month growing season. Specifically, in 2014 and 2015, the number of days with REW < 0.2 accounted for more than 33% of the growing season at the CP site and approximately 30% at the MP site. The minimum daily REW during the 4-year period at the MP site appeared in 2014 with a value of 0.07, while the daily REW at the same time at the CP site was 0.12. The minimum daily average REW during the 4-year period in CP was 0.06 in 2015, with the daily average REW in MP at that time being 0.08 (Figure 2).

The average hourly J_s–max during the entire growing season over the 4 years averaged 6.21 ± 1.72 cm⋅h^–1 for MP, with a maximum of 11.18 cm⋅h^–1. The average was significantly higher than the average of 5.60 ± 4.86 cm⋅h^–1 for CP (p < 0.05), with a maximum of 10.57 cm⋅h^–1. There were significant differences in the J_s–max for MP over the 4 years of 2014, 2015, 2017, and 2018, with values of 6.00 ± 2.31, 5.93 ± 2.44, 6.66 ± 2.58, and 6.23 ± 1.76 cm⋅h^–1, respectively (p < 0.05). Similarly, significant differences were found for CP, with average hourly J_s–max values of 5.61 ± 3.15, 5.12 ± 3.04, 6.07 ± 2.76, and 5.61 ± 1.98 cm⋅h^–1, in 2014, 2015, 2017, and 2018, respectively (p < 0.05). The average hourly J_s–max in trees of both MP and CP varied between years, with the maximum in 2017 and the minimum in 2015. The average hourly J_s–max for MP each year was higher than that for CP. However, the difference was significant (p < 0.05) only in 2015 (Figure 3).

The average daily T_r of the sampled trees during the entire growing season in the 4-year trial period averaged 0.84 ± 0.36 mm⋅d^–1, with a maximum of 1.72 mm⋅d^–1 for MP. The average is slightly higher than the average daily T_r value of 0.79 ± 0.43 mm⋅d^–1 for CP (p = 0.07), with a maximum of 1.77 mm⋅d^–1. There were no significant differences in the daily canopy transpiration between years during the four-season study (p = 0.29), with the average daily T_r values being approximately 0.85 ± 0.32, 0.87 ± 0.41, 0.79 ± 0.36, and 0.81 ± 0.28 mm⋅d^–1 for MP, for the seasons 2014, 2015, 2017, and 2018, respectively. Meanwhile, the differences were very significant for CP (p < 0.01), with daily average T_r values of approximately 0.89 ± 0.44, 0.75 ± 0.44, 0.72 ± 0.42, and 0.80 ± 0.35 mm⋅d^–1 in 2014, 2015, 2017, and 2018, respectively, with a maximum in 2014 and a minimum in 2017 (Figure 4A).

Generally, the canopy stomatal conductance during each growing season in the four-season study in MP was slightly higher than that in CP (p = 0.07), with the average daily G_s in the two species being approximately 0.22 ± 0.07 and 0.20 ± 0.11 cm⋅s^–1, respectively. Additionally, there were significant differences in the inter-annual variation in G_s between years (p = 0.01) in MP, with daily averages of approximately 0.22 ± 0.06, 0.22 ± 0.09, 0.19 ± 0.06, and 0.24 ± 0.09 cm⋅s^–1 in the four seasons, respectively. The highest value was in 2018, and the lowest was in 2017. More significant inter-seasonal variation patterns were found in CP (p < 0.01), with average daily values in G_s being approximately 0.24 ± 0.11, 0.19 ± 0.11, 0.16 ± 0.10, and 0.26 ± 0.12 cm⋅s^–1. The average daily value for G_s in MP was significantly higher than that in CP only in 2015 and in 2017, while it was slightly lower in 2014 and 2018 (Figure 4B).

The statistics on the daily T_r of the two pine trees during the whole measurement period as well as their relationships are given in Figure 5. In general, the daily T_r in MP varied positively with the daily T_r in CP during the four growing seasons. When REW > 0.4, the average daily T_r was approximately 1.03 ± 0.37 mm⋅d^–1 in CP, which was significantly higher than that in MP (p < 0.01), averaging 0.86 ± 0.33 mm⋅d^–1. However, when REW was in the range of 0.2–0.4, T_r in CP was lower than that in MP (p = 0.01), with daily T_r values of 0.72 ± 0.41 mm⋅d^–1 compared with 0.81 ± 0.36 mm⋅d^–1. In particular, when REW < 0.2, the daily T_r in CP was significantly lower than that in MP (p < 0.01), with values of 0.47 ± 0.31 mm⋅d^–1 compared with 0.83 ± 0.34 mm⋅d^–1. The slopes of the linear regressions decreased with decreasing REW levels (Figure 5).

Detailed descriptions of the relationships for daily T_r between MP and CP at the monthly scale are presented in Figure 6. Although the daily T_r in CP was slightly higher than that in MP under water-favorable conditions, the opposite was true when soil moisture was lower. For example, the daily T_r in CP was lower than that in MP in October 2014 when the REW values at the two sites were low (REW = 0.16–0.17). Similar cases occurred in July 2015 when REW was 0.17 at the MP sites and approximately 0.14 at the CP sites. Again, in June 2017, the REW at the two sites was approximately 0.30–0.31 (Figure 6). Analyses show that the slope (k) of the linear regressions at the monthly scale decreased exponentially with declining REW (Figure 7). The daily T_r in CP was higher than T_r in CP with k > 1 given that at REW above 0.34.

In each growing season, we selected two sunny days with differing but typical levels of soil moisture to explore the possible differences between the two species in diurnal patterns of T_r in response to drought. On the two selected days in 2014, during which REW had dropped from approximately 0.51–0.54 to 0.08–0.12, the hourly T_r decreased significantly (p < 0.01) in the CP plot, with the peak value of T_r decreasing by 62%, while it decreased only slightly (p = 0.16) in the MP stand, with the peak value of T_r declining by only 13% (Figure 8A). On the 2 days in 2015, when REW had declined from a relatively high value of approximately 0.39–0.49 to a very low value of approximately 0.07–0.08, a similar significant reduction in hourly T_r was found in both CP and MP trees (p < 0.01); the peak T_r values fell by 92% and 58%, respectively (Figure 8B). However, there were only slight reductions on the 2 days in 2017, when REW dropped from 0.69 to 0.30 in both MP (p = 0.95) and CP (p = 0.07) plots. Meanwhile, the peak values of T_r dropped by 10% and 41%, respectively (Figure 8C). A nearly identical pattern occurred on the 2 days in 2018, when REW dropped from approximately 0.45–0.46 to 0.24, with the hourly T_r value differing insignificantly for both MP (p = 0.53) and CP (p = 0.15) trees, and the peak T_r values fell by 20% and 36%, respectively (Figure 8D). We conclude that the peak value of T_r on any day was typically higher in CP than in MP plot when soil moisture was not under drought status (i.e., REW > 0.4). This was the case in each of the 4 years of our study. Conversely, when soil moisture was under drought (REW ≤ 0.4), the daily peak value of T_r was lower for CP than for MP.

On the daily time scale, G_s of trees in both CP and MP generally decreased with the decline in REW each year. However, G_s responded to the decline of REW more quickly in CP than in MP in each year according to the slopes of linear regressions at the daily scale, indicating a tighter regulation in response to drought in CP than in MP, which was usually considered isohydric behavior (Figure 9). The daily G_s showed a poor relationship with REW in MP according to the coefficients of determination from linear regression in each year except for in 2018.

The hourly G_s generally decreased with the increasing of VPD in both CP and MP (Figure 10). In particular, there were clear boundary lines in G_s that varied with VPD. The boundary was well fitted with a logarithmic function, from which we deduced the parameters and calculated a new variable, G_s–ref, which was defined as the upper value of canopy conductance when the VPD was equal to 1.0 kPa. G_s–ref was 14.2% higher in CP than in MP, indicating greater water-consumer behavior in CP. Another parameter (m), which was used to describe the sensitivity of canopy conductance to VPD, was 13.2% higher in CP than in MP.

In this study, our results indicated that the water use intensity of MP and CP individuals in semi-arid sandy land was at relatively low levels, with canopy transpiration averaging 0.84 and 0.79 mm⋅d^–1, respectively, during the 4-year period based on sap flow measurements. We estimate that these averages amount to transpiration levels of approximately 155 mm and 145 mm transpiration during the entire growing season of each year, which account for approximately 39% and 37% of annual precipitation in the 4 years, respectively. We found that the water use intensity of MP was slightly higher than that of CP individuals of the same age, while the tree sizes were significantly higher for MP than for CP individuals due to the more rapid growth of MP (p < 0.01). We deduce that the overall water use efficiency of MP individuals should be higher than that of CP individuals in sandy habitat. The faster growth in MP in comparison to CP at our site is consistent with many previous reports, which have shown that both the aboveground and belowground biomass are generally higher in MP than in CP of the same age (Zhao and Li, 1963; Zhu et al., 2005). These physiological differences are especially obvious in harsh habitats (Zhu et al., 2005). The fewer fine roots in CP help explain why the water use sensitivity to drought in CP was higher than that in MP. Measurements at the leaf scale show that MP features a higher photosynthetic rate, lower transpiration rate, and higher water use efficiency than CP under favorable soil moisture conditions (Bai et al., 2008; Ding et al., 2011). However, this trend would change under severely dry soil conditions (Liu et al., 2019). Combining these pieces of information, it can be concluded that the drought stress seems to be more harmful to the needles of CP, while it imparts greater damage to the fine roots of MP.

In this study, we compared water consumption of two co-existing pine species by monitoring xylem sap flow and accompanying environmental factors during a 4-year period. In general, both CP and MP in this semi-arid sandy land showed a conservative water consumption and excellent regulation of canopy stomatal conductance, which basically agrees with findings of previous studies (Song et al., 2014, 2020; Wen et al., 2017; Tang et al., 2018; Wu et al., 2018; Dang et al., 2019a). Our results provide evidence that both CP and MP exhibit strong drought resistance, like that of Scots pine, compared with other local tree species (Poyatos et al., 2005, 2008, 2013; Macinnis-Ng et al., 2016; Urban et al., 2019). We thus conclude that CP is similar to MP, and both should be considered water-savers based on their relatively low levels of canopy transpiration. Pines tend to be more homogeneous in their vulnerability to embolism than other conifers (Martinez-Vilalta and Pinol, 2002). The homogeneity of the conducting system of pines may compromise the ability of ecosystems to resist the expected increase in aridity due to climate change.

Isohydric and anisohydric behaviors in species have been widely discussed to describe their water use strategies under water stress (McDowell et al., 2008). In general, isohydric behavior is characterized by decreases in stomatal conductance to maintain nearly constant leaf water potentials (Tardieu and Simonneau, 1998; McDowell et al., 2008; Garcia-Forner et al., 2017). Many species, including Scots pine, have been regarded as isohydric species (Irvine et al., 1998; Llorens et al., 2010; Leo et al., 2013; Urban et al., 2019). Although the concept of fixed hydraulic behavior is being challenged by studies showing that individual plants or species can vary along a continuum spanning isohydric to anisohydric (Ogle et al., 2012; Feng et al., 2019; Guo and Ogle, 2019) and are affected by plant-environment interactions (Hochberg et al., 2018), we think it is a helpful framework concept if the aims are limited to comparisons of water use patterns in co-existing species.

In previous studies, pine species such as Scots pine have been shown to conserve water via sensitive stomatal regulation and to maintain relatively constant shoot water potentials under drying conditions compared with other genera such as Larix (Dulamsuren et al., 2009; Urban et al., 2019) and Quercus (Poyatos et al., 2008; Chirino et al., 2011; Morán-López et al., 2014). In this study, we conclude that both CP and MP display typical isohydric behavior by analyzing G_s. Specifically, MP individuals have been shown to exhibit a remarkable ability to maintain relatively constant water potentials by reducing stomatal conductance, which is consistent with findings from previous reports (Dang et al., 2019a,b; Song et al., 2020). It has been reported that CP maintains its leaf water potential at a relatively constant level of approximately −1.7 MPa by markedly reducing its leaf stomatal conductance (Wen et al., 2017; Tang et al., 2018; Wu et al., 2018). Direct measurements at the leaf scale also indicate that both CP and MP may be drought-tolerant tree species with high water potential and delayed dehydration (Tang et al., 2018); their leaves exhibit relatively strong water retention capacity and an ability to maintain turgor (Ding et al., 2011; Liu et al., 2019).

Interspecific variation related to water use patterns was confirmed in our study. Compared with MP, we found significantly higher canopy transpiration and canopy conductance values for CP individuals when soil water was favorable but sharper reductions under drought conditions. This behavior led to relatively lower annual water consumption by CP trees than by MP trees in this semi-arid sandy habitat. The results imply that CP is more sensitive to soil-water conditions (REW) and atmospheric water conditions (VPD) than is MP. These findings provide an explanation for the tighter stomatal regulation in CP than in MP, enabling the former to cope better with low REW and high VPD and to exhibit stronger adaptation in future projected dry environmental conditions. This interpretation is directly supported by certain case study comparisons of stomatal conductance (Quan et al., 2004) and sap-flow (Yan et al., 2018) between these two species.

In this study, our results indicate that CP exhibits a more isohydric behavior than MP by making a direct comparison of G_s from an individual perspective. However, this conclusion does not signify with certainty that the drought resistance of CP is stronger than that of MP. Previous studies at the level of organs such as needles and/or of seedlings indicate that the indexes such as the ratio of bound water to free water of needles, the water content of needles, the predawn leaf water potential, and the relative permeability of plasmalemma of needles are higher in MP than values of these parameters in CP (Bai et al., 2008). These results imply that the needles of MP feature greater drought-tolerant attributes than those of CP, with delayed dehydration to maintain high water potential through biochemical response (Wang et al., 2015). The drought resistance of trees is a complex trait, resulting from morphology, physiological and biochemical response characteristics, cellular photosynthetic organelles, and protoplasm structure, etc. Only by comprehensive studies from different perspectives can we draw more objective conclusions.

Although the fine control of stomatal openness in isohydric behavior has been considered an important mechanism in drought resistance, isohydric species are often considered to suffer from carbon depletion caused by excessive stomatal closure in the course of multi-annual droughts, resulting in long-lasting growth decline and finally tree death (Buras et al., 2018). Alternatively, trees may depend on other nonstomatal mechanisms, such as water absorption by widely distributed roots, to improve drought resistance in coordination (Konôpka et al., 2005).

The ability of roots to access deep soil water has been found to be crucial to a species’ adaptation to drought (Jiang et al., 2020). Although the roots of larger diameter generally account for a greater proportion of root biomass, approximately 92% of the mineral nutrients and 75% of the water that supports growth is absorbed by fine roots in lateral roots rather than by taproots (Makkonen and Helmisaari, 2001; Epron and Osawa, 2017). Both MP and CP are shallow-rooted with nearly 85−90% of roots within the upper 60-cm soil layer (Jiang et al., 2002; Xue, 2003). This shallow-rooted feature enhances the ability of trees to use precipitation opportunistically but is not helpful for resisting severe drought because fine roots would die almost immediately when soil turns dry (Vanguelova and Kennedy, 2007), thus directly leading to a reduction in soil-root water transport (Konôpka et al., 2005). This explains why the shallow-rooted trees can easily suffer from drought stress when drought occurs in MP forests (Song et al., 2015, 2018) or CP forests (Zhou and Shangguan, 2007). In contrast, the water use strategy of MP individuals is more prominently reflected in flourishing roots, but that of CP is more prominently reflected in stomata regulation.