Xylem hydraulic safety margins in woody plants: coordination of stomatal control of xylem tension with hydraulic capacitance

Authors


*Correspondence author. E-mail: fmeinzer@fs.fed.us

Summary

1. The xylem pressure inducing 50% loss of hydraulic conductivity due to embolism (P50) is widely used for comparisons of xylem vulnerability among species and across aridity gradients. However, despite its utility as an index of resistance to catastrophic xylem failure under extreme drought, P50 may have no special physiological relevance in the context of stomatal regulation of daily minimum xylem pressure and avoidance of hydraulic failure under non-extreme conditions. Moreover, few studies of hydraulic architecture have accounted for the buffering influence of tissue hydraulic capacitance on daily fluctuations in xylem pressure in intact plants.

2. We used data from 104 coniferous and angiosperm species representing a range of woody growth forms and habitat types to evaluate trends in three alternative xylem hydraulic safety margins based on features of their stem xylem vulnerability curves and regulation of daily minimum stem water potential (Ψstem min) under non-extreme conditions: (i) Ψstem min − P50, (ii) Ψstem min − Pe, the difference between Ψstem min and the threshold xylem pressure at which loss of conductivity begins to increase rapidly (Pe) and (iii) Pe − P50, an estimate of the steepness of the vulnerability curve between Pe and P50. Additionally, we assessed relationships between xylem capacitance, species-specific set-points for daily minimum stem water potential and hydraulic safety margins in a subset of species for which relevant data were available.

3. The three types of hydraulic safety margin defined increased with decreasing species-specific set-points for Ψstem min, suggesting a diminishing role of stem capacitance in slowing fluctuations in xylem pressure as Ψstem min became more negative. The trends in hydraulic safety were similar among coniferous and angiosperm species native to diverse habitat types.

4. Our results suggest that here is a continuum of relative reliance on different mechanisms that confer hydraulic safety under dynamic conditions. Species with low capacitance and denser wood experience greater daily maximum xylem tension and appear to rely primarily on xylem structural features to avoid embolism, whereas in species with high capacitance and low wood density avoidance of embolism appears to be achieved primarily via reliance on transient release of stored water to constrain transpiration-induced fluctuations in xylem tension.

Introduction

In a broad sense, stomata regulate transpiration to avoid damaging levels of leaf dehydration and to avoid hydraulic failure in the xylem of the stems to which the leaves are attached. As a consequence, stomatal control of vapour phase conductance is closely coordinated with dynamic changes in the hydraulic properties of the water transport pathway upstream (Meinzer & Grantz 1990; Cochard et al. 2000; Hubbard et al. 2001; Mencuccini 2003). Thus, under non-extreme conditions stomatal regulation of leaf water potential (ΨL) presumably constrains xylem pressure in stems to a range that does not result in excessive loss of stem conductivity from embolism. However, the hydraulic resistance of leaves is substantial (Sack & Holbrook 2006) and dynamic over a daily time scale (Bucci et al. 2003; Brodribb & Holbrook 2004a; Woodruff et al. 2007), resulting in a substantial and variable transpiration-induced disequilibrium between ΨL and stem xylem pressure (Begg & Turner 1970; Ritchie & Hinckley 1971). Therefore, the bulk Ψ of transpiring leaves is often difficult to relate to the magnitude of xylem pressure in the adjacent stem and its impact on hydraulic function. In order to understand the potential impact of stomatal behaviour on stem hydraulics of field-grown plants, stem water potential (Ψstem) must be estimated using techniques such as measurement of Ψ of non-transpiring leaves or shoot tips (Melcher et al. 2001; Bucci et al. 2004a) or stem psychrometry (Dixon & Tyree 1984; Scholz et al. 2007). Although published measurements of ΨL are abundant, estimates of Ψstem in transpiring plants are scarce, making it difficult to draw general conclusions about relationships between stomatal behaviour and stem hydraulic function.

The resistance of xylem to loss of function from cavitation and embolism is often assessed by generating hydraulic vulnerability curves describing the relationship between xylem pressure and loss of conductivity (e.g. Fig. 1). Vulnerability curves typically have a sigmoid shape with loss of conductivity initially increasing gradually as xylem pressure decreases followed by an abrupt transition to a much steeper, nearly linear phase, ending with a more gradual phase as loss of conductivity approaches 100%. The xylem pressure corresponding to 50% loss of conductivity (P50) is widely used for comparisons of xylem vulnerability across species, within different parts of the same individual and within species across environmental gradients (e.g. Melcher et al. 2001; Maherali et al. 2004, 2006). Another important, but less widely used point on the vulnerability curve is the threshold xylem pressure at which loss of conductivity begins to increase rapidly, often referred to as the air entry pressure (Pe). The air entry pressure can be estimated from the x-intercept of a tangent (Fig. 1, dotted line) drawn through the midpoint of a sigmoid function fitted to the vulnerability curve data (Domec & Gartner 2001). Although specific values of P50 are widely reported, values of Pe are much less frequently given and often have to be estimated from published graphs of vulnerability curves.

Figure 1.

 Typical xylem vulnerability curve showing the relationship between the percent loss of hydraulic conductivity and xylem pressure. The xylem pressures corresponding to 50% loss of conductivity (P50) and the air entry threshold (Pe, see text) are shown. The horizontal shaded area corresponds to Pe − P50, a measure of the steepness of the vulnerability curve with larger values of Pe − P50 representing a more gradual rise.

Despite the utility of P50 as an index of resistance to catastrophic xylem failure under extreme drought, it may have no special physiological relevance in the context of stomatal regulation of xylem tension under most conditions. There is increasing evidence that under non-extreme conditions stomatal control of transpiration limits maximum xylem tension to a value close to Pe for stem tissue to which the leaves are attached (Sparks & Black 1999; Brodribb et al. 2003; Domec et al. 2008). Runaway embolism should become increasingly likely once xylem pressure falls below Pe, reaching the steep portion of the vulnerability curve because tissue hydraulic capacitance (C) is largely depleted (Meinzer et al. 2008a) and no longer sufficient to adequately buffer fluctuations in tension induced by rapid changes in transpiration (see below). Thus it may be unlikely that stomatal regulation allows xylem pressure to fall substantially below species-specific values of Pe, except during periods of extreme drought that ultimately lead to branch dieback and shedding (Sperry et al. 1993; Sparks & Black 1999; Nardini & Salleo 2000). Both isohydric and anisohydric species would face a similar risk of runaway embolism as xylem pressure falls below Pe unless the slopes of the steeper portion of vulnerability curves systematically differ among species with these two modes of regulation of Ψ.

Although considerable attention has been paid to xylem safety and efficiency as components of hydraulic architecture, C is an additional and somewhat neglected component of hydraulic architecture that merits further study. Capacitance of plant tissues can play an important role in slowing changes in xylem pressure following transpiration-induced changes in xylem water flux (Cowan 1972; Phillips et al. 1997, 2004; Goldstein et al. 1998; Meinzer et al. 2004, 2008a; Scholz et al. 2007; Höltta et al. 2009). Using an Ohm’s law analogy, the buffering effect of C on fluctuations in xylem pressure can be quantified in terms of the time constant (τ), or time required for xylem pressure to undergo 63% of its total change following a change in water flux. In practice, τ can be calculated as the product of hydraulic resistance and capacitance (× C) between a reference point within the soil and a reference point within the plant. The time constant defined in this manner is thus a dynamic rather than static property of the hydraulic pathway and is expected to fluctuate over the course of a day as changes in tissue hydration cause C and possibly R to vary (Meinzer et al. 2003, 2008a; Scholz et al. 2007). Given that time constants for stomatal responses to changing environmental variables range from a few minutes to 10 min or more (Ceulemans et al. 1989; Jarvis et al. 1999; Allen & Pearcy 2000; Powles et al. 2006), C through its impact on τ for transpiration-induced fluctuations in xylem pressure may be critical for avoiding sharp decreases in xylem pressure and runaway embolism following transient increases in transpiration. Of course, sharp increases in R would increase τ, but at the cost of catastrophic xylem failure and ensuing lethal levels of dehydration. Estimates of τ for whole plants and plant parts are scarce, but available data for branches and whole trees span a range from about 5 min to >2 h (Phillips et al. 1997, 2004; Meinzer et al. 2004).

Our principal objectives in this study were to evaluate various measures of xylem hydraulic safety margins across a broad range of woody plant species and to further assess interactions between woody tissue C and hydraulic safety in a subset of nine tropical tree species for which relevant data were available (Meinzer et al. 2008a). Three alternative hydraulic safety margins were defined: the difference between minimum stem water potential (Ψstem min) and P50, the difference between Ψstem min and Pe and the difference between Pe and P50, an estimate of the steepness of the vulnerability curve between Pe and P50 (Fig. 1). In order to identify key points for stomatal limitation of minimum xylem pressure when stomatal control was still fully effective, data collected only under non-extreme conditions for a given species and habitat type were included. Based on a previous study of tropical trees (Meinzer et al. 2008a), we hypothesized that safety margins would increase with decreasing Ψstem min and that this pattern would be associated with a declining contribution of C to buffering of transpiration-induced fluctuations in xylem pressure.

Materials and methods

Data on xylem vulnerability to embolism, stem capacitance and Ψstem min of woody species were obtained from the literature. An extensive, but not exhaustive, search of the literature yielded partial or complete data sets for 104 woody species representing a wide range of xylem anatomy and phylogeny (see Table S1 in Supporting Information). Only data collected from field-grown plants were considered. Data obtained from experiments involving imposition of severe drought were not included. The following additional criteria for incorporation of data were applied: Three measures of xylem vulnerability to embolism were estimated from standard hydraulic vulnerability curves consisting of plots of percent loss of hydraulic conductivity (PLC) vs. stem xylem pressure (Fig. 1), or from tables of data listing key points along vulnerability curves. The three measures of vulnerability were (i) the xylem pressure corresponding to 50% loss of hydraulic conductivity (P50), (ii) the air entry threshold (Pe) corresponding to about 12% loss of conductivity wherein species with pronounced sigmoid vulnerability curves PLC begins to rise sharply with declining xylem pressure (Domec & Gartner 2001) and (iii) Pe − P50, a measure of how steeply PLC rises once Pe has been reached. Larger values of Pe − P50 imply a more gradual rise of PLC once xylem pressure has fallen below Pe. Stem sapwood capacitance expressed per unit sapwood volume (kg m−3 MPa−1) for nine co-occurring tropical tree species was obtained from Meinzer et al. (2008a).

Daily Ψstem min was obtained from studies in which Ψ was determined with a pressure chamber on covered, non-transpiring leaves (Begg & Turner 1970; Bucci et al. 2004a) and from studies in which Ψ was measured on excised previously transpiring leafy twigs (Table S1). The covered leaf method is preferable for estimating Ψstem because the hydraulic resistance of leaves often results in substantial transpiration-induced disequilibrium between stem and leaf Ψ. However, the occurrence of daily minimum plant Ψ is typically associated with marked stomatal limitation of transpiration, which would constrain the degree of disequilibrium between leaf and stem Ψ.

In three co-occurring tropical tree species for which published data were available (Meinzer et al. 2008a), relative values of stem C and hydraulic resistance (R) in relation to Ψstem were used to estimate relative time constants (τ = × C) for changes in xylem pressure following changes in xylem water flux. Relative rather than absolute values of τ were used because the data necessary to express both C and R in units that would yield τ in time units only were not available. The purpose of this exercise was to determine whether P50 and Ψstem min of different species occurred at similar points along the trajectory of τ as Ψstem declined.

The significance of relationships between Ψstem and Pe, P50 and Pe − P50 and between P50 and Pe was evaluated using one-way anova. Differences in the dependence of P50 on Pe among three groups of species provisionally identified post hoc were assessed by pair-wise ancova after it was determined that the regressions were heterogeneous (F = 12·5, P < 0·0001).

Results

Observations compiled for nine coniferous and 28 angiosperm species showed that daily Ψstem min measured when soil water deficits were non-extreme was tightly coordinated with both Pe and P50 in a similar manner across species and across both conifers and angiosperms (P < 0·0001, Fig. 2). The relationship between Pe and Ψstem min was close to the 1:1 line over a relatively wide range of species values of Ψstem min. For species having Ψstem min more negative than about −3 MPa, Pe began to decline noticeably more quickly than Ψstem min. In the species with the least negative values of Ψstem min, Pe was less negative than Ψstem min. In contrast, P50 was always more negative than Ψstem min, and decreased more than 2·5 times faster than Ψstem min with declining species-specific values of Ψstem min.

Figure 2.

 Relationship between stem xylem vulnerability properties and daily minimum stem water potential (Ψstem min) for several coniferous and angiosperm species. Pe is the air entry pressure and P50 is the pressure causing 50% loss of hydraulic conductivity (see Fig. 1). Dashed lines represent the 1:1 relationships. Species and references are listed in Table S1.

The hydraulic safety margin expressed as Ψstem min − P50 was always positive and increased about 10-fold over the range of Ψstem min observed (Fig. 3a). The safety margin expressed as Ψstem min − Pe was negative or near zero for species with Ψstem min less negative than about −2·5 MPa and increased to about 4 MPa in species with the lowest values of Ψstem min observed (Fig. 3b). A third safety margin, Pe − P50, based on species-specific xylem vulnerability curve characteristics alone, increased from about 1 to 4 MPa as Ψstem min declined from about −0·5 to −4·5 MPa (P < 0·0001, Fig. 3c). Greater values of Pe − P50 indicate a more gradual loss of hydraulic conductivity as xylem pressure falls below Pe.

Figure 3.

 Relationships between three types of hydraulic safety margins and daily minimum stem water potential (Ψstem min). (a) The difference between Ψstem min and the xylem pressure causing 50% loss of conductivity (P50), (b) the difference between Ψstem min and the air entry threshold (Pe) and (c) Pe − P50. Horizontal dashed lines represent safety margins of zero. Species and references are listed in Table S1.

Xylem vulnerability curve data were found for more species than were data for Ψstem. Variation in P50 was linearly related to that of Pe, but three distinct groups of species appeared to emerge based on trajectories of regressions fitted to relationships between Pe and P50 (P < 0·0001 for all regressions, Fig. 4). Among 16 species of conifers and 47 species of angiosperms native to wet, mesic and seasonally dry habitats, the slope of the relationship between P50 and Pe was 1·42, indicating that the slope of the steeper portion of the vulnerability curve became more gradual as Pe declined (Fig. 4a). In two other groups of species, the slopes of the relationships between P50 and Pe were not significantly different from 1·0, indicating that the difference between Pe and P50 was constant over the entire range of Pe observed (Fig. 4b,c). In 12 members of Cupressaceae (e.g. Juniperus spp.) native to semi-arid and arid zones and 14 angiosperm species native to Mediterranean climate zones, P50 was about 4·5 MPa more negative than Pe over a broad range of Pe (Fig. 4b). Among 15 vesseless angiosperm species, P50 was only about 1 MPa more negative than Pe (Fig. 4c). The slopes of the regressions in Fig. 4b,c were both significantly different from that in Fig. 4a (P ≤ 0·01) and they differed significantly in their offset from the 1:1 line (P < 0·0001).

Figure 4.

 Relationships between the xylem pressure causing 50% loss of stem hydraulic conductivity (P50) and the air entry pressure (Pe) for several coniferous and angiosperm species. The dashed lines represent 1:1 relationships. Species and references are listed in Table S1.

Minimum Ψstem showed a strong positive correlation (P = 0·0002) with sapwood capacitance among nine tropical tree species (Fig. 5a). Minimum daily Ψstem was about 1 MPa less negative in the species with the highest C (410 kg m−3 MPa−1) than in the species with the lowest C (83 kg m−3 MPa−1), consistent with the buffering effect on xylem pressure of transient withdrawal of water from internal storage. Among the same nine species, the difference between Pe and P50 decreased in a linear fashion (P < 0·0001) with increasing sapwood C (Fig. 5b), indicating that the slope of the steeper portion of the xylem vulnerability curve was more abrupt in species with higher C.

Figure 5.

 Relationship between stem sapwood capacitance and (a) daily minimum stem water potential (Ψstem min) and (b) the stem hydraulic safety margin defined as the difference between the air entry pressure and the pressure causing 50% loss of conductivity (Pe − P50) for upper canopy branches of nine tropical tree species (Table S1).

The dependence of relative C and relative hydraulic resistance on Ψstem was characterized for three tropical tree species with high, intermediate and low values of sapwood C and daily Ψstem min (Fig. 6). As expected, relative C initially declined sharply as Ψstem decreased followed by an asymptotic approach to a minimum value of C, whereas the initial increase in relative R associated with embolism was gradual followed by a steep rise. Regardless of species-specific values of Ψstem min and C, Ψstem min occurred near or just after the transition from a steep to gradual decline in C as Ψstem declined, and was associated with a relatively narrow range of increase in relative resistance (30–39%). Overall, the patterns in Fig. 6 implied that the relative time constant (× C) for a response in xylem tension following a change in xylem water flux initially declined sharply as Ψstem decreased followed by a gradual approach to a minimum value of τ (Fig. 7). Across three species representing a broad range of sapwood C, Ψstem min was consistently associated with the transition from a steep to a gradual, nearly linear decline in τ as Ψstem decreased. Consistent with the data shown in Fig. 3a, the difference between the Ψstem min and P50 increased with decreasing species-specific values of sapwood C and minimum Ψstem (Fig. 7).

Figure 6.

 The dependence of relative capacitance and relative hydraulic resistance on stem water potential for three co-occurring tropical tree species. The arrows indicate mean daily minimum stem water potentials (Ψstem min). Absolute values of capacitance (C) are shown.

Figure 7.

 Calculated relative time constants for changes in stem water potential in response to changes in xylem water flux for three co-occurring tropical tree species. Time constants were estimated as the product of resistance and capacitance (× C) from the data shown in Fig. 6. Vertical lines represent daily minimum stem water potential (Ψstem min) and the stem xylem pressure corresponding to 50% loss of hydraulic conductivity (P50). Horizontal dashed lines represent tangents drawn through the nearly linear portions of the curves at low values of stem water potential.

Discussion

Even under non-extreme conditions of adequate soil moisture and moderate vapour pressure deficit, the potential risk of catastrophic xylem failure is high, unless stomatal behaviour is tightly coordinated and integrated with the properties of the hydraulic pathway upstream from the leaves as they change over the course of the day. We found a notable degree of convergence in trends of xylem hydraulic safety over a broad range of Ψstem min among 37 coniferous and angiosperm species native to diverse habitat types. The three types of hydraulic safety margin defined increased with decreasing species-specific values of Ψstem min. This pattern probably reflects a diminishing role of C in slowing fluctuations in Ψstem as Ψstem becomes more negative. Nevertheless, stomatal control interacts with C to determine species-specific set-points for Ψstem min, and therefore hydraulic safety margins. Overall, the patterns observed were consistent with earlier suggestions that although coordination between stomatal behaviour and stem hydraulic capacity generally avoids catastrophic hydraulic failure, tolerance of a certain degree of embolism may enhance productivity under some conditions, especially if loss of stem hydraulic function is reversible over a relatively short time scale (Jones & Sutherland 1991; Vogt 2001; Brodribb & Holbrook 2004b; Sperry 2004).

Taken together, our results suggest that here is a continuum of relative reliance on different mechanisms that confer hydraulic safety under dynamic conditions. Species with low capacitance and denser wood experience greater daily maximum xylem tension and appear to rely primarily on xylem structural features to avoid embolism, whereas in species with high capacitance and low wood density, avoidance of embolism appears to be achieved primarily via reliance on transient release of stored water to constrain fluctuations in xylem tension.

Stomatal control of stem xylem pressure

Minimum Ψstem was close to Pe over a wide range of Ψstem min from about −0·5 to −3 MPa (Fig. 2). This finding is consistent with earlier reports that when soil water availability is adequate, stomata constrain Ψstem to minimum values at or less negative than Pe (Sparks & Black 1999; Brodribb et al. 2003; Domec et al. 2008). In species with Pe below about −4 MPa, stomatal regulation of Ψstem was more conservative resulting in a safety margin (Ψstem min − Pe) of nearly 4 MPa in the species with the lowest values of Pe (cf. Figs 2a and 3b). Stomatal regulation of stem xylem pressure at values substantially more negative than Pe implies an increased risk of runaway embolism as xylem pressure begins to traverse the steep portion of the sigmoid hydraulic vulnerability curve where tissue C has effectively been exhausted, causing a sharp reduction in the time constant for changes in xylem pressure in response to changes in transpiration. The effectiveness of stomata in constraining stem xylem pressure diminishes as the time constant for flow-induced changes in xylem pressure decreases relative to that for reductions in stomatal conductance. The trend towards higher values of Ψstem min − Pe in species with lower Ψstem min is consistent with this interpretation because the relative role of C in buffering changes in Ψstem diminishes as Ψstem declines (Figs 5 and 7). The specific mechanisms by which stomata respond to and control ΨL to result in species-specific values of Ψstem min are not known. Regardless of the mechanisms, the coordination of leaf and stem Ψ set-points optimizes reliance on hydraulic capacitance of stem tissue (Meinzer et al. 2008a).

The data presented for daily Ψstem min represent pressure chamber measurements on non-transpiring leaves attached to transpiring shoots as well as conventional measurements on transpiring leaves or leafy shoots. However, the hydraulic resistance of leaves is substantial, leading to pronounced uncoupling or disequilibrium between leaf and stem Ψ in transpiring shoots (Begg & Turner 1970; Meinzer et al. 2001; Melcher et al. 2001; Bucci et al. 2004a). Thus, a violation of our assumption that ΨL is approximately equal to Ψstem when ΨL is at its daily minimum value and stomata have partly closed (see Materials and methods) would yield estimates of Ψstem min that are too negative. If actual species values of Ψstem min were less negative than those shown in some cases, the overall trend of increasing hydraulic safety margins with declining Ψstem min would remain the same, but stomatal regulation of Ψstem in relation to Pe and other key points for stem hydraulic function would appear to be even more conservative. Nevertheless, most of the data for Ψstem min≥−2 MPa were obtained from non-transpiring leaves (Table S1) reinforcing the inference that the safety margin defined as Ψstem min − Pe was actually negative for values of Ψstem min≥−2 MPa. The impact of stomatal behaviour on the hydraulic function of stems to which leaves are attached cannot be assessed fully without knowledge of Ψstem, but explicit attempts to estimate Ψstem using techniques such as pressure chamber measurements on non-transpiring leaves or stem psychrometry are rather scarce in the literature. Studies aimed at describing the integration of stomatal and stem hydraulic function in intact plants should incorporate methods designed to estimate the Ψ of the stems supporting the leaves.

Hydraulic safety margins

The air-seeding pressure, P50, is widely used as a comparative index of xylem hydraulic safety. Physiologically, however, attainment of P50 indicates that nearly catastrophic hydraulic failure has already occurred and that the risk of further runaway embolism is acute because xylem pressure is operating along the steepest portion of the vulnerability curve (Tyree & Sperry 1988). Thus, the relevance of P50 as an indicator of resistance to hydraulic failure may be restricted to episodic extreme drought conditions under which it becomes physically impossible for stomata to constrain xylem pressure above or around Pe. These conditions are expected to lead to branch dieback and shedding, and in some cases death of the entire plant, which would ultimately limit species distributions along aridity gradients. In the present study we have identified three alternative hydraulic safety margins referenced to species-specific values of daily Ψstem min set by stomatal control of transpiration (Fig. 3). All three types of safety margins increased with declining species-specific values of Ψstem min. Interestingly, the safety margin defined as Ψstem min − Pe was negative in species having values of Ψstem min down to about −2 MPa, implying that branches of these species may regularly experience a substantial loss of their conductivity even when soil water availability is not severely restricted. Consistent with this, native loss of conductivity was about 36% among 20 co-occurring tropical tree species having mean daily values of Ψstem min between −0·7 and −1·5 MPa (Santiago et al. 2004; Meinzer et al. 2008a). Even higher levels of native embolism were found in two co-occurring Mediterranean Quercus species (Tognetti et al. 1998). The extent to which daily losses of conductivity may be reversed overnight is unclear because branch samples in the studies cited were collected after the onset of transpiration in the morning. However, results of other studies suggest that a certain fraction of stem embolism may be readily reversible over diel and seasonal cycles (e.g. Zwieniecki & Holbrook 1998; Melcher et al. 2001; Vogt 2001) and that in leaves embolism reversal may be even more rapid with recovery of hydraulic conductance occurring during the afternoon while xylem pressure is still well below zero (Bucci et al. 2003; Brodribb & Holbrook 2004a; Woodruff et al. 2007).

In contrast to Ψstem min − Pe, Ψstem min − P50 was always positive and increased by a factor of about 10 over the observed range of Ψstem min. The absence of negative values of Ψstem min − P50 reinforces the notion that stomatal regulation normally prevents stem xylem pressure from traversing the steep portion of the vulnerability curve. Increasing values of the safety margin defined as Pe − P50 indicate a progressively more gradual slope of the vulnerability curve once the air entry threshold has been crossed, which can be regarded as a second line of defence against catastrophic xylem failure once stomatal regulation can no longer prevent xylem pressure from entering the steeper portion of the vulnerability curve. The specific xylem structural features responsible for the observed behaviour of Pe − P50 are not known, but could certainly be examined. Most likely they relate to the incidence of safety features in the population of cells contributing to embolism avoidance, which could include not only tracheids and vessels, but also parenchyma and fibre cells.

It is interesting that when all vulnerability curve data were pooled, two types of relationships between Pe and P50 emerged, one in which there was an increasing departure from the 1:1 relationship with declining Pe (Fig. 4a), and one in which there was a constant offset from the 1:1 line (Fig. 4b,c). The patterns observed suggest that phylogeny and evolutionary adaptations to certain habitat types were important determinants of relationships between Pe and P50. Among 15 vesseless angiosperms, Pe − P50 remained nearly constant at about 1 MPa over a 3·5 MPa range of Pe, whereas Pe − P50 was about 4·5 MPa over an 8 MPa range of Pe in 28 species of Cupressaceae and angiosperms native largely to arid and Mediterranean climates zones. In other conifers and angiosperms native to more mesic climate zones Pe − P50 increased with declining Pe. The species represented in Fig. 4 do not entirely overlap with those in Fig. 3 because published data on Ψstem were not available for many of the species in Fig. 4. The overall trends in hydraulic safety margins were consistent with a diminishing role of C as Ψstem declined (Fig. 5), increasing the likelihood that stomatal responses would not be rapid enough to avoid catastrophic hydraulic failure unless safety margins increased or were consistently large.

Capacitance and hydraulic safety

Stomata regulate xylem pressure under dynamic conditions in which water released into the transpiration stream via discharge of capacitance comes into play. Therefore, C must be considered in attempts to understand how intact plants avoid hydraulic failure under the dynamic conditions that prevail in the field. The buffering influence of C can be viewed as a dynamic component of overall hydraulic safety because it varies with Ψstem over daily and seasonal cycles. In contrast, properties such as Pe and P50 are static over the daily time scale of stomatal regulation. The patterns reported here are consistent with a partial reliance on C to confer a margin of hydraulic safety rather than sole reliance on xylem structural features that govern properties such as Pe and P50. Species-specific values of Ψstem min and C were positively correlated (Fig. 5a). Additionally, previous studies have noted positive correlations between C and P50 and between C and Ψstem min − P50 (Domec & Gartner 2001; Pratt et al. 2007; Meinzer et al. 2008a). Even the slope of the steeper portion of the vulnerability curve (Pe − P50) was steepest in species with the highest C (Fig. 5b). Thus, declining species-specific values of C imply increasing reliance on xylem structural features that confer adequate hydraulic safety margins via changes in features such as Pe, P50 and Pe − P50.

Capacitive release of water via xylem cavitation can contribute to buffering of fluctuations in xylem pressure despite the negative impact of cavitation on hydraulic conductivity (Höltta et al. 2009). However, capacitive release of water from wood does not necessarily imply a trade-off involving reduced conducting efficiency. In conifers, cavitation of latewood tracheids can supply water to the transpiration stream without a significant loss of sapwood hydraulic conductivity because the conductivity of the more cavitation-resistant earlywood tracheids is about an order of magnitude greater than that of latewood tracheids (Domec & Gartner 2001, 2002). In diffuse-porous angiosperms, living axial and ray parenchyma can constitute up to 40–75% of sapwood volume (Panshin & de Zeeuw 1980; Chapotin et al. 2006) making these tissues a potentially important source of C. In several tropical tree species water released via cavitation of vessels accounted for only about 15% of total daily reliance on C (Meinzer et al. 2008a).

When stem tissue is fully hydrated, species with higher values of C are expected to show greater time constants for changes in xylem pressure in response to changes in transpiration. These attributes would partly compensate for relatively long time constants of several minutes for stomatal closure (Ceulemans et al. 1989; Jarvis et al. 1999; Allen & Pearcy 2000; Powles et al. 2006), permitting safety margins based on features such as Pe and P50 to appear small. However, as Ψstem min and Pe are approached, the hydraulic system becomes nearly inelastic because C is essentially exhausted (Figs 6 and 7) causing time constants for transpiration-induced changes in xylem pressure to approach minimum values, thereby making it difficult for stomata to respond quickly enough to dampen abrupt decreases in xylem pressure that could trigger rampant embolism. The inverse relationship between dynamic and static components of hydraulic safety represents a trade-off whose common denominator is likely to be wood density. Several studies have noted inverse relationships between wood density and C (Meinzer et al. 2003, 2008b; Pratt et al. 2007; Scholz et al. 2007), minimum shoot Ψ as well as variation in shoot Ψ (Stratton et al. 2000; Meinzer 2003; Bucci et al. 2004b; Jacobsen et al. 2007, 2008), xylem specific conductivity (Stratton et al. 2000; Bucci et al. 2004b; Meinzer et al. 2008b) and P50 (Hacke et al. 2001; Pratt et al. 2007). However, some studies have not detected a dependence of some of these traits on wood density (e.g. Jacobsen et al. 2007; Meinzer et al. 2008b).

Acknowledgements

This research was supported by National Science Foundation grants IBN 9905012 and IOB 0544470.

Ancillary