Height-related trends in leaf xylem anatomy and shoot hydraulic characteristics in a tall conifer: safety versus efficiency in water transport


Author for correspondence
David R. Woodruff
Tel:+1 541 750 7494
Fax:+1 541 750 7329
Email: david.woodruff@oregonstate.edu


  • • Hydraulic vulnerability of Douglas-fir (Pseudotsuga menziesii) branchlets decreases with height, allowing shoots at greater height to maintain hydraulic conductance (Kshoot) at more negative leaf water potentials (Ψl).
  • • To determine the basis for this trend shoot hydraulic and tracheid anatomical properties of foliage from the tops of Douglas-fir trees were analysed along a height gradient from 5 to 55 m.
  • • Values of Ψl at which Kshoot was substantially reduced, declined with height by 0.012 Mpa m−1. Maximum Kshoot was reduced by 0.082 mmol m−2 MPa−1 s−1 for every 1 m increase in height. Total tracheid lumen area per needle cross-section, hydraulic mean diameter of leaf tracheid lumens, total number of tracheids per needle cross-section and leaf tracheid length decreased with height by 18.4 µm2 m−1, 0.029 µm m−1, 0.42 m−1 and 5.3 µm m−1, respectively. Tracheid thickness-to-span ratio (tw/b)2 increased with height by 1.04 × 10−3 m−1 and pit number per tracheid decreased with height by 0.07 m−1.
  • • Leaf anatomical adjustments that enhanced the ability to cope with vertical gradients of increasing xylem tension were attained at the expense of reduced water transport capacity and efficiency, possibly contributing to height-related decline in growth of Douglas fir.


Growth and aboveground biomass accumulation follow a common pattern as tree size increases, with productivity peaking when leaf area reaches its maximum and then declining as tree age and size increase (Ryan & Waring, 1992). Age- and size-related declines in forest productivity are major considerations in setting the rotational age of commercial forests, and relate to issues of carbon storage, since changes in forest structure can influence large-scale biomass accumulation. Despite the ecological and practical significance of the ontogenetic decline in tree growth, the mechanisms responsible for it are not well understood (Ryan et al., 2006). However, available evidence suggests that ontogenetic trends in growth are mainly a function of tree size (height) rather than age (Koch et al., 2004; Woodruff et al., 2004; Bond et al., 2007; Mencuccini et al., 2007).

Height-related changes in leaf function may have an impact on tree growth and forest productivity because leaf stomata are responsible for maximizing photosynthetic carbon gain while simultaneously dealing with the antagonistic task of constraining transpirational water loss to avoid damaging levels of dehydration. The gravitational component of water potential leads to a 0.01 MPa increase in xylem sap tension per meter increase in height, which substantially reduces leaf water potential (Ψl) near the tops of tall trees. Frictional resistance during transpiration leads to an additional path-length dependent reduction in Ψl (Bauerle et al., 1999). In the absence of osmotic adjustment, the turgor of leaf cells will decrease in direct proportion with Ψl along height and path length gradients. Cell volume increase during growth can be tightly coupled with cell turgor pressure (Lockhart, 1965). In addition to cell expansion, a range of growth-related processes including cell division are sensitive to turgor pressure (Boyer, 1968; Kirkham et al., 1972; Hsiao et al., 1976; Gould & Measures, 1977). Given the relationship between height and Ψl, and the dependence of cell division and expansion upon turgor, a causal relationship between tree height and anatomical properties that influence foliar physiology seems likely. Leaves typically comprise a substantially smaller portion of the path length in plant vascular systems than stems, yet they represent a disproportionately large fraction of the whole-plant hydraulic resistance (Yang & Tyree, 1994; Nardini & Salleo, 2000). Height-related changes in leaf anatomy that affect foliar water transport efficiency may thus help to explain observed size- or age-dependent reductions in forest productivity.

Although leaves comprise the terminal portion of the vascular system and are thus likely to experience more negative water potentials than other plant organs, they are typically more vulnerable to embolism than stems (Brodribb & Holbrook, 2003; Bucci et al., 2003; Woodruff et al., 2007). In some cases, however, leaves have been found to be less vulnerable to embolism than stems (Sack & Holbrook, 2006). Given their critical functions in gas exchange and their high vulnerability to loss of hydraulic conductance, leaves may exhibit adaptive morphological characteristics that maximize efficiency, minimize vulnerability or both. Recent work has shown a correlation between tree height and hydraulic safety of Douglas-fir (Pseudotsuga menziesii) shoots (Woodruff et al., 2007), allowing shoots at greater height to maintain hydraulic conductance (Kshoot) and stomatal opening at more negative values of Ψl. This height-related trend may allow taller trees to continue to photosynthesize during periods of greater water stress, but previous work on stems and roots of Douglas-fir (Domec et al., 2006) implies that adaptations to minimize hydraulic vulnerability may also involve trade-offs that reduce shoot water transport capacity.

Substantial intraspecific variation has been found in leaf form and function (Sprugel et al., 1996; Grassi & Bagnaresi, 2001) and even within individuals along height or light gradients (Niinemets et al., 1999; Koike et al., 2001; Woodruff et al., 2004). Despite the great diversity in leaf structural and physiological traits, leaf and shoot hydraulic properties appear to be coordinated with physiological function in a consistent manner across species. Recent research has begun to more fully describe the connections between leaf anatomical characteristics and leaf physiological function such as correlations between leaf architecture and transport efficiency (Aasamaa et al., 2005; Sack & Frole, 2006); as well as leaf architecture and gas exchange (Salleo et al., 2001; Sack et al., 2003; Brodribb et al., 2005, 2007). Xylem conduit properties such as length, diameter, wall thickness and pit abundance have been analysed as determinants of hydraulic conductivity and resistance to embolism of wood (Hacke et al., 2001; Pitterman et al., 2006; Sack & Holbrook, 2006; Sperry et al., 2006). However, much less is known about the influence of these anatomical characteristics upon hydraulic efficiency and safety of leaf xylem. Specifically, we know of no other research that has investigated the impact of tree height upon these leaf anatomical characteristics and associated shoot physiological attributes.

The goal of this study was to determine the extent to which height may influence key anatomical features related to leaf hydraulic efficiency and tension-induced vulnerability in a coniferous tree, and to examine the extent to which the observed anatomical patterns are associated with adaptive physiological advantages. We examined leaf xylem anatomical properties likely to be associated with height-related trends in hydraulic function, including characteristics related to cell expansion such as hydraulic mean diameter of leaf tracheid lumens (Dh), leaf tracheid length (Tl) and tracheid wall thickness-to-span ratio (tw/b)2. Analyses also included number of tracheids per needle cross section (T#), which is dependent upon cell division; plus total lumen area per needle cross-section (LAt), a characteristic influenced by both cell expansion and cell division. Sampling was conducted exclusively from fully sun-exposed branches near the tops of trees of different height classes in order to rule out the potentially confounding influence of factors such as irradiance, relative humidity and branch length upon height-related trends in leaf hydraulic architecture and anatomy.

Materials and Methods


Five separate stands, each containing Douglas-fir (P. menziesii (Mirb.) Franco) trees of a different height class, were located within 3.1 km of each other in the Wind River Basin of southwestern Washington, USA. Samples were collected within 1–5 m of the tops of the trees at mean sampling heights of 5.0, 12.7, 18.3, 34.5 and 55.0 m. All samples were obtained from branches in fully sun-exposed locations. Access to treetops in the 55-m sampling height class was facilitated by a 75-m-tall construction tower crane at the Wind River Canopy Crane Research Facility (WRCCRF, Carson WA, USA). Tree tops in all other height classes were accessed by nonspur climbing.

The Pacific maritime climate of the region is characterized by wet winters and dry summers. Mean annual precipitation in the region is c. 2.2 m, much of which falls as snow, with a dry season from June through September. The mean annual temperature is 8.7°C with means of 0°C in January and 17.5°C in July. The soils are well-drained and of volcanic origin (Shaw et al., 2004). Very low precipitation between June and September (c. 119 mm) typically leads to drought conditions in the upper portion of the soil profile. Soil water however, remains accessible to Douglas-fir roots at depths greater than c. 1 m throughout the summer dry period (Warren et al., 2005; Meinzer et al., 2007).

Leaf hydraulic conductance and vulnerability

The maximum hydraulic conductance of terminal shoots (Kshoot) and shoot hydraulic vulnerability to embolism were determined. These characteristics represent the capacity of foliated shoots to transport water and their ability to avoid loss of hydraulic function as water stress increases. Kshoot was measured as a proxy for leaf hydraulic conductance (Kleaf) because of the difficulty in measuring hydraulic conductance on individual Douglas-fir needles. We believe Kshoot to be a reliable proxy for Kleaf given that leaves represent the majority of the hydraulic resistance to water flow in shoots and a considerable fraction of the entire hydraulic resistance within whole plants (Yang & Tyree, 1994; Nardini & Salleo, 2000; Sack & Holbrook, 2006). The methods used were adapted from Brodribb & Holbrook (2003), and involve the use of the following equation based on an analogy between rehydrating a shoot and recharging a capacitor:

Kshoot = C ln(Ψof)/t, (Eqn 1)

(Kshoot = shoot hydraulic conductance (mmol m−2 s−1 MPa−1); C = capacitance; Ψo = leaf water potential before partial rehydration; Ψf = leaf water potential after partial rehydration; t = duration of rehydration). Branches c. 30–50 cm long were collected from trees early in the morning before significant transpirational water loss and were placed in plastic bags with moist paper towels and stored in the dark in a refrigerator. Measurements of leaf water potential were conducted over the next 3 d on excised twigs (c. 10–15 cm long) for initial values (Ψo), and for final values after a period of rehydration of t seconds (Ψf). Deionized water used for rehydration of Kshoot samples was partly degassed by subjecting it to a vacuum for a minimum of 2 h before transferring it into a separate container. Samples were dehydrated to a range of Ψo before Kshoot measurements in order to obtain full vulnerability curves. Branches were left to equilibrate in the dark in sealed plastic bags for at least 30 min following periods of dehydration. Water temperature was between 21°C and 23°C, and the photosynthetic photon flux density at the foliage was maintained at c. 1000 µmol m−2 s−1 during Kshoot measurements. Exposure time of foliage to full irradiance was typically less than 5 min. Rehydration times during Kshoot measurements ranged between 20 s and 290 s. All Kshoot samples, regardless of height class, bore only 2 yr of foliage (current and previous year).

We subjected some Kshoot samples to variable periods of ‘prehydration’ before initial measurements of Ψl in order to obtain a wider range of initial values of Ψl and Kshoot. Prehydration times ranged from one to 12 h. The methods described by Brodribb & Holbrook (2003) were used to estimate C. Briefly, the Ψl corresponding to turgor loss was estimated as the inflection point of the graph of Ψl vs relative water content (RWC). The slope of the curve before, and following turgor loss provided C in terms of RWC (Crwc) for pre-turgor loss and post-turgor loss, respectively. Crwc was then multiplied by the saturated mass of water in the shoots and then divided by leaf area in order to provide a value of C expressed in absolute terms and normalized by leaf area. Mean pre-turgor loss C values were 0.67, 0.52, 0.51, 0.55 and 0.47 mol m−2 MPa−1 for foliage from the 5.0, 12.7, 18.3, 34.5 and 55.0 m sampling heights, respectively. Post-turgor loss C values were 2.65, 1.18, 1.31, 1.75, and 1.61 mol m−2 MPa−1 and turgor loss points were −2.22, −3.25, −2.94, −2.96 and −3.05 MPa for the same sampling heights, respectively. Samples for Kshoot and pressure–volume curve measurements were collected from three separate trees for each sample height (except 34.5 m, for which there were two trees; and 55.0 m, for which there were four trees) at 10 dates: 13 July, 21 July, 21 October 2005; 28 June, 1 July, 23 August, 21 September, 25 September, and 5 December 2006; and 24 January 2007. All foliage was sampled after hardening of current year foliage to minimize confounding effects from intra-annual variability. For two of the heights (5 m and 55 m), foliage was collected and analysed on multiple occasions throughout the study and no seasonal trends were evident in Kshoot.

Values were determined for the Ψl at which 20, 50 and 80% loss of hydraulic conductance occurred (Ψ20, Ψ50 and Ψ80, respectively). These values of Ψl were obtained from a sigmoid regression curve fitted to mean Kshoot versus Ψ data. For example, Ψ50 represents the leaf water potential at 50 percent of the regression curve's maximum value on the y-axis (Fig. 1). Following our method used in previous work (Woodruff et al., 2007) to identify an objective and functionally relevant measure of shoot hydraulic vulnerability an analysis was used that was similar to that originally proposed for description of vulnerability curves for wood by Domec & Gartner (2001). This method provided a means for designating the point at which Kshoot had declined to an initial minimum value (Ψl at minimum Kshoot). For each height class, data were grouped by water potential ranges (0 to −0.5 MPa, < −0.5 to −0.75 MPa, < −0.75 to −1.0 MPa, < −1.0 to −1.25 MPa, < −1.25 to −1.5 MPa, < −1.5 to −1.75 MPa, < −1.75 to −2.0 MPa, < −2.0 to −2.25 MPa, < −2.25 to −2.5 MPa, < −2.25 to −2.5 MPa, < −2.5 to −2.75 MPa and < −2.75 to −3.0 MPa) to compute the corresponding mean Kshoot values (n = 2–28 branches per height class) over each Ψ range (Fig. 1). The value of Kshoot representing the midpoint between the y-intercept of the sigmoid function and 0 (dotted line in Fig. 1) was selected for estimating the slope of the portion of the sigmoid curve containing the relatively rapid and nearly linear decline in Kshoot. We took the derivative of the sigmoid function to calculate its instantaneous slope at the Kshoot midpoint and used the x-intercept of the resulting tangent (dashed line in Fig. 1) as an objective estimate of Ψl at minimum Kshoot. Maximum Kshoot (Kshoot-max) was estimated from the mean of the Kshoot values obtained from the most hydrated samples in each height class (Ψl = 0 to –0.5 MPa). Kshoot-max, Ψl at minimum Kshoot and Ψ50 data were analysed using regression analysis. Error bars for Kshoot-max represent standard errors of branches (n = 7–28).

Figure 1.

Example of a typical relationship between mean hydraulic conductance (Kshoot) and leaf water potential (Ψl) illustrating the method used to estimate the Ψl at which Kshoot has declined to its initial minimum value. Data are from Douglas-fir (Pseudotsuga menziesii) 34.5 m height class. The horizontal dotted line represents the midpoint of the y-range of the sigmoid curve (50% loss of Kshoot). The instantaneous slope at this point yielded a tangent (dashed line) whose intercept with the x-axis was designated as the value of Ψl at which Kshoot had reached its initial minimum value. Bars, ± SE; n = 2–28 branches per height class.

Pressure–volume analyses (Tyree & Hammel, 1972) were conducted on branchlets c. 10-cm-long. These samples were excised from branch samples that were collected early in the morning before significant transpirational water loss, sealed in plastic bags with moist paper to prevent desiccation and then stored in a refrigerator within 1–4 h of excision. Pressure–volume curves were initiated by first determining the fresh mass of the twig, and then measuring Ψl with a pressure chamber (PMS Instrument Company, Corvallis, OR, USA). Alternate determinations of fresh mass and Ψl were repeated during slow dehydration of the twig on the laboratory bench until values of Ψl exceeded the measuring range of the pressure chamber (−4.0 MPa). The inverse of water potential was plotted against relative water content to create a pressure–volume curve. For normalizing C on a leaf area basis, one-sided leaf areas of the branchlets were obtained with a scanner and imagej version 1.27 image analysis software (Abramoff et al., 2004).

Leaf xylem anatomical characteristics

We analysed leaf xylem anatomical properties likely to be associated with height-related trends in hydraulic function. These included total tracheid lumen area per needle cross section (LAt), hydraulic mean diameter of leaf tracheid lumens (Dh), number of tracheids per needle cross section (T#), leaf tracheid length (Tl), pit number per tracheid (Pit#) and tracheid wall thickness-to-span ratio (tw/b)2. Macerations of isolated needle xylem were prepared for analyses of tracheid length and pit number per tracheid. Cross-sections of needles were made by hand-sectioning of fresh tissue for analyses of all other anatomical characteristics. For macerations samples were each submerged in c. 10 ml of a solution of 15 g sodium chlorite dissolved in 250 ml of distilled water. Ten drops of acetic acid were added to each test tube and the solutions were heated at 90°C for at least 24 h. The macerated cells were rinsed, stained with toluidine blue, and mounted on slides. Each needle cross-section was analysed with an image analysis system consisting of a compound microscope and video camera. Anatomical characteristics obtained from macerations were obtained from between six and eight branches per height class, and anatomical characteristics obtained from hand sectioning represent between six and nine branches per height class. All anatomical measurements were obtained from foliage produced during 2006. Images were obtained using ×20 or ×40 objective lenses with total magnifications of ×200 and ×400. Data were pooled per height class and subjected to regression analysis.


Physiological trends with height

Timed rehydration measurements showed that Kshoot declined sigmoidally with Ψl. A logistic three-parameter sigmoid function (y = a/(1 + (x/x0)b) yielded r2 values ranging from 0.98 to 0.99 for the dependence of mean Kshoot on Ψl (Fig. 2). Kshoot-max was used as a measure of hydraulic efficiency and Ψl at minimum Kshoot and Ψ50 were used as measures of hydraulic vulnerability. Kshoot-max decreased with increasing height (P = 0.035) from 7.2 mmol m−2 s−1 MPa−1 at 5 m, to 3.2 mmol m−2 s−1 MPa−1 at 55.0 m (Fig. 3a). Foliage from the tops of taller trees was more resistant to loss of hydraulic conductance than foliage from shorter trees regardless of the index of hydraulic vulnerability employed (Fig. 3b). Values of Ψl at minimum Kshoot and Ψ50 decreased linearly with increasing height at a rate of 0.012 Mpa m−1 (P = 0.020) and −0.009 Mpa m−1 (P = 0.0088), respectively. Similar results were obtained for height-related trends in Ψ80 and Ψ20 (P = 0.021 and 0.0078, respectively). There was a positive correlation between Kshoot-max and Ψl at minimum Kshoot (P = 0.0047) and between Kshoot-max and Ψ50 (P = 0.009), showing that increased resistance to tension-induced loss of Kshoot was associated with reduced maximum water transport capacity (Fig. 4).

Figure 2.

Shoot hydraulic conductance (Kshoot) in relation to leaf water potential (Ψl) for Douglas-fir (Pseudotsuga menziesii) foliage samples obtained within 1–5 m of the tops of the trees at mean sampling heights of (a) 55.0 m, (b) 34.5 m, (c) 18.3 m, (d) 12.7 m and (e) 5.0 m. A logistic three-parameter sigmoid function (y = a/(1 +  (x/x0)b) was fit to Kshoot in relation to Ψl.

Figure 3.

(a) Maximum hydraulic conductance (Kshoot-max) (bars, ± SE; n = 7–28 branches per height class) and (b) different measures of shoot hydraulic vulnerability versus height in Douglas-fir (Pseudotsuga menziesii). Values of Ψ20, Ψ50, Ψ80 and Ψl at minimum Kshoot were obtained from vulnerability curves as illustrated in Fig. 1.

Figure 4.

Maximum hydraulic conductance (Kshoot-max) in relation to shoot hydraulic vulnerability measured as leaf water potential (Ψl) at minimum Kshoot and Ψ50 in Douglas-fir (Pseudotsuga menziesii).

Anatomical trends with height

Quantitative analyses of leaf tracheid characteristics revealed height-related trends in anatomy (Fig. 5). Observed trends were consistent with the pronounced decline in Kshoot-max with increasing height. LAt declined with height by 19 µm2 m−1 (P = 0.019). Dh declined with increasing height by 0.031 µm m−1 (P = 0.0016). T# decreased with height by 0.43 m−1 (P = 0.05). Tl declined with increasing height by 5.2 µm m−1, although the trend was marginally significant (P = 0.053). Pit# decreased with height by 0.07 m−1 (P = 0.03) and (tw/b)2 increased with increasing height by 0.001 m−1 but the trend was not significant (P = 0.085).

Figure 5.

Leaf tracheid anatomical characteristics along a height gradient sampled from the upper 1–5 m of Douglas fir (Pseudotsuga menziesii) trees of different height classes. (a) Total lumen area, (b) hydraulic mean diameter, (c) tracheid number, (d) tracheid length, (e) pit number and (f) thickness to span ratio (tw/b)2. Bars, ± SE; n = 6–9 branches for (a–c, f); n = 6–8 branches for (d, e).

Relationships between anatomy and physiology

Correlations of Kshoot-max, Ψl at minimum Kshoot and Ψ50 with leaf tracheid anatomical characteristics were evaluated in order to elucidate potential causal relationships between leaf structural and shoot hydraulic attributes. Many of the leaf tracheid anatomical properties analysed were significantly correlated with either Kshoot-max, Ψl at minimum Kshoot, Ψ50, or all three (Table 1). LAt was correlated with hydraulic efficiency (P = 0.018), with Ψl at minimum Kshoot (P = 0.025) and Ψ50 (P = 0.0066). For every 100 µm2 increase in LAt, there was an increase in hydraulic efficiency such that Kshoot-max increased by 0.43 mmol m−2 MPa−1 s−1, and an increase in hydraulic vulnerability such that Ψl at minimum Kshoot increased by 0.06 MPa, and Ψ50 increased by 0.05 MPa. Kshoot-max, Ψl at minimum Kshoot and Ψ50 were also strongly correlated with changes in Dh (P = 0.014, P = 0.0082 and P = 0.004, respectively). On average, a 1 µm increase in Dh was correlated with an increase in Kshoot-max of 2.9 mmol m−2 MPa−1 s−1, an increase in Ψl at minimum Kshoot of 0.40 MPa and an increase in Ψ50 of 0.31 MPa. T# was significantly correlated with Ψl at minimum Kshoot (P = 0.044) and with Ψ50 (P = 0.015), but marginally so with hydraulic efficiency (P = 0.057). Tl had no significant correlation with either hydraulic efficiency or vulnerability (P = 0.29 for Kshoot-max, P = 0.18 for Ψl at minimum Kshoot and P = 0.13 for Ψ50). Pit# was significantly correlated with Kshoot-max (P = 0.014), with Ψl at minimum Kshoot (P = 0.0096), and with Ψ50 (P = 0.0021). (tw/b)2 was significantly correlated with Kshoot-max (P = 0.028), but not significantly correlated with either Ψl at minimum Kshoot or Ψ50 (P = 0.071 and P = 0.094, respectively).

Table 1.  Per cent of variance (r2) and significance values for shoot hydraulic conductance (Kshoot) and vulnerability and leaf xylem anatomical characteristics of Douglas-fir (Pseudotsuga menziesii)
Leaf xylem anatomical characteristicHydraulic efficiency (Kshoot-max)Hydraulic vulnerability
50)l at minimum Kshoot)
  • LAt, total lumen area per leaf cross section; Dh, hydraulic mean diameter of leaf tracheid lumens; T#, total number of tracheids per leaf cross section; Tl, leaf tracheid length; Pit#, number of pits per tracheid; (tw/b)2, ratio of thickness of tracheid wall to span of tracheid lumen; Ψl, leaf water potential; Ψ50, 50% loss of hydraulic conductance.

  • (+), Positive correlation; (−), negative correlation.

  • *,**

    , P ≤ 0.05 and P ≤ 0.01, respectively.

LAt (µm2)0.88* (+)0.94** (+)0.85* (+)
Dh (µm)0.90* (+)0.96** (+)0.93** (+)
T#0.75 (+)0.89* (+)0.79* (+)
Tl (µm)0.35 (+)0.58 (+)0.50 (+)
Pit#0.90* (+)0.97** (+)0.92** (+)
(tw/b)20.84* (−)0.66 (−)0.72 (−)


Our results indicate that there were opposing height-related trends in shoot hydraulic efficiency and safety in Douglas-fir, which were correlated with several leaf tracheid anatomical characteristics. The observed trends in leaf anatomy were consistent with likely impacts of reduced turgor on cell expansion and division with increased tree height (Koch et al., 2004; Woodruff et al., 2004). The decline in Dh and Tl with height is consistent with a reduction in the turgor-driven cell expansion associated with increased gravitational and path length resistance. Given that less cell expansion also implies a relative increase in available wall material per unit cross-sectional area, an increase in (tw/b)2 suggests both reduced lumen diameter and greater wall thickness, both resulting from a reduction in turgor-driven cell expansion. Reduced T# with height-associated decline in turgor is consistent with previous work which has shown a causal link between turgor and cell division (Boyer, 1968; Kirkham et al., 1972). The reduction in LAt represents a combination of both a decline in Dh and T#. Aside from the possibility of a correlation between overall tracheid size and the number of pits per tracheid, a direct connection between turgor during tracheid expansion and Pit# is unclear.

These results suggest that the effects of tree height upon leaf cell development lead to enhanced ability to avoid water stress-induced embolism at the expense of reduced water transport capacity and efficiency. Moreover, localized height-related hydraulic restrictions on water uptake by expanding cells in terminal shoots may have a negative synergistic impact on turgor-limited tissue expansion. The reduction in cell expansion limits hydraulic conductance due to the effect of reduced lumen diameter upon hydraulic resistance, and reduced hydraulic conductance may in turn limit potential cell expansion because it represents a hydraulic limitation imposed upon the system that supplies the water necessary for foliar cell expansion. During the relatively short period of shoot expansion in the late spring, osmotic adjustment in Douglas-fir is insufficient to compensate for the vertical gradient of increasing tension (Woodruff et al., 2004; Meinzer et al., 2008). Thus, if water entry into growing tissue does not keep pace with continuous cell wall relaxation, turgor will be further reduced, resulting in a substantial water potential disequilibrium between adjacent regions of growing and nongrowing tissue (Boyer et al., 1985; Nonami & Boyer, 1989, 1990).

Vulnerability to loss of hydraulic conductance in Douglas-fir shoots was substantially greater than that observed in shoots of the tropical conifer Podocarpus grayii (Brodribb & Holbrook, 2005) and in the leaves of four temperate pine species (Cochard et al., 2004) but similar to that found in leaves for a number of tropical angiosperm species (Brodribb & Holbrook, 2003; Bucci et al., 2003), for several pteridophytes and gymnosperms (Brodribb & Holbrook, 2006) and for Douglas-fir shoots along a vertical gradient within an individual forest canopy (Woodruff et al., 2007). Maximum values of Kshoot measured in Douglas fir were similar to those reported for other gymnosperm species (Brodribb & Holbrook, 2005, 2006) and for Douglas fir along a vertical gradient within a single forest canopy (Woodruff et al., 2007).

Contrary to studies indicating that conduit length accounts for a substantial portion of the variation in conducting efficiency of wood (Pothier et al., 1989; Domec et al., 2006; Sperry et al., 2006) tracheid length was not a significant determinant of shoot hydraulic conductance. The strongest anatomical correlates of Kshoot-max were Pit#, LAt, and Dh. The observed significant relationship between Dh and transport efficiency is not unexpected given that even small changes in conduit diameter lead to major changes in transport efficiency according to the Hagen–Poiseuille equation:

image(Eqn 2)

(Kt (m4 MPa−1 s−1) is theoretical conductance ρ is the density of water (5.55 × 107 mmol m−3) and η is the viscosity of water (1.002 × 10−9 MPa s at 20°C). An independent analysis of vertical trends in shoot hydraulic conductance was performed by calculating a theoretical leaf area-specific conductance (Kleaf-theoretical) using a modified version of the Hagen–Poiseuille equation, and using T# (per needle) and leaf area of a single needle (Al):

image(Eqn 3)

Leaf area measurements were conducted on foliage at each sampling height and these values were used to normalize Kleaf-theoretical by leaf area. Leaf area was found to scale inversely with height by 0.24 mm2 m−1. Despite the trend in leaf area with height, Kshoot-max correlated strongly with leaf-specific values of Kleaf-theoretical (P = 0.006; Fig. 6). Note that although Kshoot-max and Kleaf-theoretical are both conductance values, they are not directly comparable because they have different units. The comparison between Kshoot-max and Kleaf-theoretical is still noteworthy, however, because the strong positive correlation between the two provides additional support for a causal relationship between key leaf xylem anatomical properties (in this case LAt and Dh) and shoot hydraulic function. In the case of a direct comparison with identical units the values of Kshoot-max would likely be substantially lower than Kleaf-theoretical due to deviations in actual conduits from idealized tubes resulting from features in tracheid lumens that affect water flow such as warts and bordered pits (Domec et al., 2006).

Figure 6.

The relationship between theoretical leaf area-specific conductance (Kleaf-theoretical) and measured maximum hydraulic conductance (Kshoot-max) in Douglas fir (Pseudotsuga menziesii). Bars, ± SE; n = 6–9 branches for Kleaf-theoretical; n = 7–28 branches for Kshoot max.

Although the values of hydraulic efficiency and vulnerability were comparable to those found in Douglas-fir at the same site in a previous study (Woodruff et al., 2007), the trend in both of these with height was more consistent in the current study (Fig. 3). It is important to note that although the 55-m sampling locations in this study and in Woodruff et al. (2007) were equivalent, all other sampling locations were different. In the previous study, samples were taken from three locations within a height gradient of a single old growth forest canopy, and at one location at the tops of smaller trees in a nearby stand. In the current study sampling was limited to the shorter branches at the tops of sun-exposed trees of different heights to eliminate any confounding effects from variable levels of irradiance, relative humidity or branch length upon height-related trends in leaf hydraulic architecture and anatomy. Branch length is a potentially critical factor because of the trend in branch length with depth within a forest canopy (K. Bible, unpublished) and because of the substantially higher levels of resistance to water transport in branches compared to boles (Domec & Gartner, 2002). The increased hydraulic resistance associated with the greater branch length found lower in the canopy is likely to limit water availability to the attached foliage. Branch length is thus a potentially confounding variable with height in its impact upon water availability to leaves.

The anatomical characteristics with the highest correlation to shoot vulnerability were Pit#, Dh and LAt. Loss of hydraulic conductivity owing to embolism is believed to be caused by ‘air seedling’ (Zimmerman, 1983; Sperry & Tyree, 1988). Once the tension within a xylem conduit exceeds the capillary forces at the air–water interface in a pit membrane, an air bubble is pulled through a pore in the pit membrane from an adjacent air-filled conduit. The air bubble expands in the conduit resulting in an ‘embolized’ or nonhydraulically functional conduit. The tension required to initiate air seeding is believed to be a function of the pit aperture diameter (Zimmerman, 1983), pit membrane and pit chamber characteristics (Domec et al., 2006), and the frequency of pits per conduit (McCully & Canny, 1988). The very strong relationship between pit number and hydraulic vulnerability (Table 1) provides support for the theory of air seeding as a primary cause of hydraulic failure via embolism, and is consistent with previously observed relationships between xylem anatomy and loss of hydraulic conductivity in wood (Zimmerman, 1983; Hacke et al., 2004; Domec et al., 2006).

Reduced conduit diameter has long been associated with increased resistance to embolism in wood (Sucoff, 1969; Ewers, 1985). The only mechanistic relationship between conduit size and hydraulic vulnerability that has been confirmed by experimentation is the association of freeze–thaw induced embolisms with conduit diameter (Davis et al., 1999). Studies of freezing-induced embolism in conifer wood have shown a substantially reduced susceptibility to this phenomenon than in angiosperm wood as a result of the smaller diameters of tracheids compared with vessels. Whether or not this trend holds true in conifer foliage is yet to be determined. Although shoot samples were not subjected to freeze–thaw treatments, repeated freeze–thaw cycles are common at the study site during the winter months. The potential for freeze–thaw-induced embolism is therefore potentially relevant for interpretation of the adaptive significance of trends in leaf tracheid anatomy in Douglas-fir.

The conduit thickness to span ratio is typically analysed as a measure of vulnerability to implosion as opposed to vulnerability to embolism. It is not expected to have a direct influence upon resistance to embolism, since wall thickness is likely to be more relevant to cell wall implosion than to air seeding. However, Hacke et al. (2001) have shown a constant safety factor between implosion and embolism in both gymnosperms and angiosperms, so (tw/b)2 has been used as an index of vulnerability. Based upon previous cryo-scanning electron microscopy (SEM) analyses of leaf tracheids at varying levels of dehydration, it was concluded that implosion of leaf tracheids in Douglas fir is an uncommon event and that short-term variation in Kshoot are associated with reversible embolism (Woodruff et al., 2007).

Reduced shoot hydraulic conductance is likely to limit stomatal conductance, and therefore photosynthesis, consistent with age and height-related reductions in tree growth. Height-related decline in shoot hydraulic conductance could thus be considered complementary to other hydraulic factors which have been proposed as the basis for height-related decline in productivity, such as increased stomatal closure owing to resistance associated with greater length of hydraulic path and the gravitational potential gradient opposing the ascent of water in taller trees (Yoder et al., 1994; Ryan & Yoder, 1997). The slope of the relationship between tree height and shoot hydraulic vulnerability (−0.012 MPa m−1 for Ψl at minimum Kshoot, Fig. 3b) is slightly more negative than the hydrostatic gradient, providing further evidence of gravity and path length as the driving forces for height-related trends in hydraulic architecture. Other nonvascular leaf anatomical factors could be at play in the relationship between tree height and decline in productivity such as increased extravascular resistance to water transport (Brodribb et al., 2007) or increased occurrence of nonphotosynthetic foliar biomass such as astrosclereids (Apple et al., 2002).

It is noteworthy that although both leaf xylem anatomy and shoot hydraulic properties were correlated with tree height, these were often more highly correlated with each other; providing further evidence of the influence of these foliar anatomical properties upon hydraulic function. The decline in shoot hydraulic vulnerability with increasing tree height partially mitigates the effects of increasing xylem tension associated with gravity and increased path-length resistance, and may allow taller trees to continue to photosynthesize during periods of greater water stress. This advantage could be derived from evolutionary forces selecting for leaf anatomical properties conferring greater resistance to loss of conductivity. The observed vertical trends in xylem anatomical traits governing hydraulic efficiency and safety could also be primarily driven by the height-related decline in turgor during periods of cell expansion and cell division (Kirkham et al., 1972; Woodruff et al., 2004; Meinzer et al., 2008).


This work was supported by the USDA Forest Service, Pacific Northwest Research Station Ecosystem Processes Program. We thank Ken Bible, Mark Creighton, Matt Schroeder and the rest of the staff at the Wind River Canopy Crane Research Facility located within the Wind River Experimental Forest, T.T. Munger Research Natural Area. Thanks also to Manuela Huso for statistical consultation and Kate McCulloh for help with sample collection and advice on microscopy technique.