3.1. Leaf Water Potential
There were significant differences between species in minimum daily water potential. Overall,
Q. rubra had lower leaf water potential during the day (
Table 1, effect of Species) than
A. rubrum. However, the overall direct effects of soil moisture or VPD on minimum leaf water potential were not significant. This was not because soil moisture or VPD had no effect, but, as indicated by the significance of the interaction terms this was driven by the fact that there were significantly different response to soil moisture and VPD between the two species (
Table 1, effects of Species × SM and Species × VPD). These interactions are apparent in
Figure 1 (upper panels, Species × SM) and
Figure 2 (Species × VPD), and the significance of the interaction terms in these statistical tests confirm that the regression curves that describe the response of leaf water potential to VPD and soil moisture are significantly different (
Figure 1 and
Figure 2). Our hypotheses as to the differences in hydrologic strategy of the two species that drive these differences are discussed in
Section 3.2. The results were consistent across all four combinations of sources of environmental data—from both shallow (0–30 cm) and deeper (0–60 cm) soil and both within and above canopy atmospheric conditions indicating that the results are not sensitive to the arbitrary choice of sampling height/depth of the environmental conditions. The only exception to this was the interaction between VPD and SM that is significant only in the shallow soil and within canopy air. This interaction was driven by the fact that within canopy VPD had a consistent overall effect (in both species, combined) of decreasing leaf water potential only when the shallow soil was dry.
Because the mid-day leaf water potentials of tree individuals were significantly influenced by species but not by canopy dominance, data were split according to species and linear regressions were performed to determine the influence of soil water potential and VPD on minimum (mid-day) and maximum (predawn) leaf water potentials (
Figure 1 and
Figure 2). Predawn leaf water potential of
Q. rubra was not significantly influenced by soil moisture (linear regression:
p = 0.213,
R2 = 0.050;
Figure 1C). For
A. rubrum, however, there was a positive correlation between predawn leaf water potential values and soil moisture (linear regression:
p = 0.0015,
R2 = 0.298;
Figure 1D). Midday leaf water potentials for
Q. rubra were positively correlated with volumetric soil water content (regression:
p = 0.0003,
R2 = 0.348;
Figure 1A), while for
A. rubrum this relationship was non-significant (linear regression:
p = 0.09,
R2 = 0.097;
Figure 1B).
Table 1.
The results of mixed linear models testing the significance of different variables and interactions affecting minimal leaf water potential in twelve trees of two species (Q. rubra and A. rubrum). Vapor pressure deficit (VPD), soil moisture (SM) and species were included as effective variables. Sample size in each test is n = 67. Soil moisture was measured over two depth intervals: 0–30 cm and 0–60 cm. VPD was measured at two heights: at 46 m above ground (above canopy) and at 3 m above ground (sub-canopy). For both soil moisture and VPD, data at different levels where highly correlated. Four models were run using pair-wise combinations of the two depths and heights. The p-values for each output are listed. Statistical significance was assumed at p < 0.05. (NS) means not significant. The product sign denotes the interaction term of the variables tested for in the mixed linear model.
Table 1.
The results of mixed linear models testing the significance of different variables and interactions affecting minimal leaf water potential in twelve trees of two species (Q. rubra and A. rubrum). Vapor pressure deficit (VPD), soil moisture (SM) and species were included as effective variables. Sample size in each test is n = 67. Soil moisture was measured over two depth intervals: 0–30 cm and 0–60 cm. VPD was measured at two heights: at 46 m above ground (above canopy) and at 3 m above ground (sub-canopy). For both soil moisture and VPD, data at different levels where highly correlated. Four models were run using pair-wise combinations of the two depths and heights. The p-values for each output are listed. Statistical significance was assumed at p < 0.05. (NS) means not significant. The product sign denotes the interaction term of the variables tested for in the mixed linear model.
| | VPD, SM combinations |
---|
| VPD measurements height | 46 m | 46 m | 3 m | 3 m |
| SM integration depth | 0–30 cm | 0–60 cm | 0–30 cm | 0–60 cm |
Effects on minimum daily leaf water potential | Species | <0.001 | 0.009 | <0.001 | 0.012 |
SM | 0.638 (NS) | 0.461 (NS) | 0.128 (NS) | 0.180 (NS) |
VPD | 0.873 (NS) | 0.448 (NS) | 0.157 (NS) | 0.129 (NS) |
Species × SM | 0.001 | 0.027 | 0.003 | 0.041 |
| Species × VPD | <0.001 | <0.001 | <0.001 | <0.001 |
| SM × VPD | 0.434 (NS) | 0.244 (NS) | 0.037 | 0.052 (NS) |
Figure 1.
Responses of leaf water potential values (ψL) to soil water potential at the top 30 cm for 6 Q. rubra and 6 A. rubrum individuals. Significant regressions are shown in dashed lines. Soil water potential is in (MPa) and listed on reversed logarithmic scale with highly negative values (dry soil) on the right hand side of the scale. Mid-day leaf water potentials for: (A) Q. rubra; and (B) A. rubrum. Predawn leaf water potentials for: (C) Q. rubra; and (D) A. rubrum.
Figure 1.
Responses of leaf water potential values (ψL) to soil water potential at the top 30 cm for 6 Q. rubra and 6 A. rubrum individuals. Significant regressions are shown in dashed lines. Soil water potential is in (MPa) and listed on reversed logarithmic scale with highly negative values (dry soil) on the right hand side of the scale. Mid-day leaf water potentials for: (A) Q. rubra; and (B) A. rubrum. Predawn leaf water potentials for: (C) Q. rubra; and (D) A. rubrum.
Figure 2.
Relationship between mid-day leaf water potentials of Q. rubra and A. rubrum and sub-canopy vapor pressure deficit. A. rubrum shows very mild response to VPD while Q. rubra’s response slope to VPD is significantly steeper.
Figure 2.
Relationship between mid-day leaf water potentials of Q. rubra and A. rubrum and sub-canopy vapor pressure deficit. A. rubrum shows very mild response to VPD while Q. rubra’s response slope to VPD is significantly steeper.
Mid-day (minimum) leaf water potential values were positively correlated with both above-canopy VPD (measured at 46 m) and inside-canopy VPD (measured at 3 m above ground) for both
Q. rubra (linear regression; 46 m VPD:
p < 0.0001,
R2 = 0.74; 3 m VPD:
p < 0.0001,
R2 = 0.76) and
A. rubrum (linear regression; 46 m VPD:
p < 0.0001,
R2 = 0.35; 3 m VPD:
p < 0.001,
R2 = 0.30) (
Figure 2). In general,
Q. rubra exhibited a much larger range of minimum leaf water potentials (−2.06 to −0.275 MPa) during day-time hours, as compared to
A. rubrum (−0.6 to −0.14 MPa). While both species are significantly affected by VPD, regression slope between mid-day leaf water potentials and VPD in
Q. rubra was significantly steeper than
A. rubrum.
3.2. Discussion
The results indicate contrasting water use strategies for
Q. rubra and
A. rubrum. Changing soil moisture conditions did not induce a significant response in the pre-dawn leaf water potential values of
Q. rubra. This suggests that
Q. rubra may be able to accesses a sustained soil water source in deeper soil layers and recharge its stem storage overnight, so that its water potential remains stable even when the top soil layers are very dry. Observations of soil water potential in our site confirm that possibility (
Figure 3 and
Figure 4). The variation in soil moisture and the fraction of time that the soil was drier than −0.5 MPa (−5 bar) are far smaller in the soil at depth between 1–2 m (
Figure 3D,E) than in the shallower soil layers (≤60 cm) (
Figure 3A–C). By correlating pre-dawn water potentials with the soil moisture in each layer separately (rather than the mean for the top 30 cm), we found that pre-dawn leaf water potential in
A. rubrum were correlated with soil water potential in each depth up to 60 cm (15, 30 and 60 cm) while those of
Q. rubra were not correlated with soil water potential at any depth, including the shallow layers. Similar observations were reported by Bréda
et al. [
32], which found that predawn water potentials were not sensitive to soil moisture for members of the
Quercus genus. However, that finding was limited above a 40% threshold of plant extractable water, much higher than in our site. Thus,
Q. rubra may be above the threshold necessary to induce a response to soil moisture deficits.
Q. rubra and
Quercus species in general have been found to be deeply rooted tree species [
33,
34] and thus should have access to a larger reservoir of soil water within the profile than shallow rooted species. It has been shown that water is increasingly taken up from deeper soil horizons during periods of drought in the
Quercus genus [
32,
35].
Figure 3.
Distribution of soil water potential at different depths at our site during the experiment (summer 2011). The normalized histogram (probability density function, PDF) of half-hourly measurements of soil water potentials (in MPa) is plotted. Each depth is a result from 8 (A–D) or 4 (E,F) soil moisture probes at different locations. Dashed lines illustrate the 10th and 90th percentiles of the soil moisture distribution.
Figure 3.
Distribution of soil water potential at different depths at our site during the experiment (summer 2011). The normalized histogram (probability density function, PDF) of half-hourly measurements of soil water potentials (in MPa) is plotted. Each depth is a result from 8 (A–D) or 4 (E,F) soil moisture probes at different locations. Dashed lines illustrate the 10th and 90th percentiles of the soil moisture distribution.
In contrast,
A. rubrum shows a significant positive correlation between pre-dawn leaf water potential values and soil moisture levels. This indicates that the amount of water recharge
A. rubrum can accomplish is linked to the amount of water in the shallow soil layers, and that members of this species do not fully recharge overnight unless soil water is sufficiently high [
36].
A. rubrum was found to be a shallow rooted species with most root biomass concentrated in upper soil layers [
34].
Both
Q. rubra and
A. rubrum show a positive correlation between mid-day leaf water potential and vapor pressure deficit (
Figure 1). This is consistent with many studies, including Turner
et al. [
37] in which it was found that increasing VPD leads to decreases in leaf water potential and gas exchange, but also leads to an increase in transpiration rate in woody species. However, the specific response to VPD was different between the two species. Leaf water potentials of
Q. rubra showed a more pronounced response to VPD than those of
A. rubrum, as seen in the steeper slope of the regression line. As explained by McDowell
et al. [
38] anisohydric plants allow leaf water potential to drop when the soil or the air are dryer, while isohydric plants maintain a relatively constant leaf water potential by reducing stomatal conductance as soil and air conditions dry. Our leaf water potential observations confirm that
Q. rubra presents an anisohydric strategy with larger diurnal variation and overall lower mean values of leaf water potential. Our observations that mid-day water potential in
A. rubrum was not correlated with soil moisture confirm the predictions of Meinzer
et al. [
21], which used sap-flow observations to analyze the hydraulic strategies of trees in a forest plot in the Eastern US.
A. rubrum in our site presented a characteristic isohydric strategy by maintaining relatively constant and high leaf water potential even when soil moisture and VPD conditions were stressing.
Figure 4.
Time series of soil water potential at different depths. The period of the experiment, characterized by a 2-week sequence of low precipitation and consequent drying of the soil column is marked by vertical dashed lines. The soil layers between 100 and 200 cm deep show the lowest variation and never reached below −0.5 MPa during the experiment period.
Figure 4.
Time series of soil water potential at different depths. The period of the experiment, characterized by a 2-week sequence of low precipitation and consequent drying of the soil column is marked by vertical dashed lines. The soil layers between 100 and 200 cm deep show the lowest variation and never reached below −0.5 MPa during the experiment period.
We hypothesize that by reducing stomatal conductance
A. rubrum is able to keep leaf water potentials high, even when VPD and soil moisture are restrictive. This is despite the fact that
A. rubrum is not fully recharged with water in most mornings. This is consistent with the results of Oren & Pataki [
13], who found that
A. rubrum had higher stomatal sensitivity to VPD than co-occurring
Q. Alba and attributed the difference to the differences in xylem properties between diffuse- and ring-porous species. In contrast, the ring-porous
Q. rubra exhibits a risk-prone strategy. Its larger ring-porous vessels are more conductive and therefore can continue providing water at adequate rates even when VPD is high. As leaf water potential data suggest, this is done without down-regulating of the stomatal conductance, which implies a higher risk of hydraulic failure. The hypothesis of strong stomatal regulation as a mechanism of maintaining isohydric leaf water potential in
A. rubrum is in contrast with those of Bush
et al. [
9]. They found that diffuse porous species (including
Acer platanoides) showed increased transpiration with increasing VPD, indicating that down regulation of stomatal conductance did not take effect at high VPD values. However, their experiment was set in an irrigated site with soil water content between 20%–30% (of garden top-soil) while our experiment site was set in deep sandy soil and soil water content was between 4%–9%. It is highly possible that for the range of soil moisture conditions tested by Bush
et al. [
9] the trees did not reach the level of stress that demanded down regulation of stomatal conductance. Consistent with our results, Ni
et al. [
39] showed that saplings of different Oak species maintain non-negligible transpiration rates in soil water potentials below −0.5 MPa while the stomatal conductance of
Acer saccharurn at such low water potential approaches zero.
Deep roots can provide sufficient night-time recharge to full capacity and, possibly, additional water supply during the day. By allowing transpiration at maximal rates,
Q. rubra leaf water potential is reduced to low levels and may potentially lead to increased risk of widespread cavitation.
A. rubrum, which as a diffuse porous species is potentially more sensitive to xylem cavitation than ring porous species, may however reduce cavitation risk by closing stomata at less negative leaf water potentials [
40]. Our results indicate that
Q. rubra and
A. rubrum take different ends of the safety-efficiency tradeoff [
41] with the ring porous species optimizing at the efficiency end of the scale. Taneda & Sperry [
18] showed that although the ring-porous species
Quercus gambelii was highly vulnerable to cavitation during the day, it was able to refill embolized vessels overnight. That study also indicated that
Q. gambelii accessed a much deeper pool of soil water compared to the co-occurring
Acer grandidentatum.