Functional coordination between leaf gas exchange and vulnerability to xylem cavitation in temperate forest trees

Authors


Hafiz Maherali. Fax: + 1 519 7671656; e-mail: maherali@uoguelph.ca

ABSTRACT

We examined functional coordination among stem and root vulnerability to xylem cavitation, plant water transport characteristics and leaf traits in 14 co-occurring temperate tree species. Relationships were evaluated using both traditional cross-species correlations and phylogenetically independent contrast (PIC) correlations. For stems, the xylem tension at which 50% of hydraulic conductivity was lost (Ψ50) was positively associated (< 0.001) with specific conductivity (KS) and with mean hydraulically weighted xylem conduit diameter (Dh-w), but was only marginally (= 0.06) associated with leaf specific conductivity (KL). The PIC correlation for each of these relationships, however, was not statistically significant. There was also no relationship between root Ψ50 and root KS in either cross-species or PIC analysis. Photosynthetic rate (A) and stomatal conductance (gs) were strongly and positively correlated with root Ψ50 in the cross-species analysis (< 0.001), a relationship that was robust to phylogenetic correction (< 0.01). A and gs were also positively correlated with stem Ψ50 in the cross-species analysis (= 0.02 and 0.10, respectively). However, only A was associated with stem Ψ50 in the PIC analysis (P = 0.04). Although the relationship between vulnerability to cavitation and xylem conductivity traits within specific organs (i.e. stems and roots) was weak, the strong correlation between gs and root Ψ50 across species suggests that there is a trade-off between vulnerability to cavitation and water transport capacity at the whole-plant level. Our results were therefore consistent with the expectation of coordination between vulnerability to xylem cavitation and the regulation of stomatal conductance, and highlight the potential physiological and evolutionary significance of root hydraulic properties in controlling interspecific variation in leaf function.

INTRODUCTION

Water supply to leaves depends on maintaining an intact water column in the xylem from roots to shoots. Because this hydraulic pathway is under increasing tension during transpiration, it is vulnerable to cavitation through air seeding, which occurs when air bubbles are aspirated into water-filled conduits (Zimmermann 1983; Tyree & Sperry 1989). Each time cavitation occurs, the resulting vapour-filled conduit no longer carries water, causing a decrease in xylem hydraulic conductivity (Tyree & Ewers 1991). The vulnerability of plants to xylem cavitation is often correlated with variation in moisture availability, and recent evidence indicates that this relationship is adaptive across a broad range of taxonomic groups (Maherali, Pockman & Jackson 2004).

However, our understanding of how interspecific variation in the vulnerability of xylem to cavitation is associated with variation in other plant functional traits that control carbon and water balance is incomplete. For example, resistance to water stress-induced cavitation via air seeding in a large majority of angiosperms depends on the surface tension of the meniscus in each pore of the pit membrane that connects adjacent conduits (Jarbeau, Ewers & Davis 1995). Increased resistance to cavitation should therefore be accompanied by decreased pore hydraulic conductivity and, ultimately, stem hydraulic conductivity (Zimmermann 1983; Sperry & Hacke 2004). Nevertheless, evidence for a trade-off between stem cavitation resistance and xylem water transport across species is inconsistent (e.g. Maherali et al. 2004). The direct link between sustained water transport and leaf transpiration also suggests that there should be functional coordination between vulnerability to xylem cavitation and the regulation of stomatal conductance (gs) across species (Sperry & Pockman 1993; Sparks & Black 1999; Brodribb et al. 2003). In addition, the observation that tissue construction costs are associated with vulnerability to xylem cavitation across species (Hacke et al. 2001) suggests that xylem cavitation could be associated with traits that affect plant carbon uptake. Although there is evidence of functional coordination between hydraulic conductivity (KH) and gas exchange across species (Brodribb & Feild 2000; Meinzer 2002), few studies have explicitly considered the link between cavitation resistance and suites of leaf physiological traits that control CO2 and H2O fluxes across species.

Many insights about the functional significance of xylem cavitation across species have been derived from studies on stems (Tyree & Ewers 1991; Pockman & Sperry 2000; Maherali et al. 2004). However, root xylem properties generally differ from stems and may be more tightly linked to the control of whole plant water transport (Jackson, Sperry & Dawson 2000). For example, roots typically have wider xylem conduits, and in consequence, higher segment hydraulic conductivity per unit cross-sectional area than stems (e.g. McElrone et al. 2004). In addition, roots are generally more vulnerable to xylem cavitation than stems, suggesting that they may be the weakest link along the soil-to-leaf water transport pathway (Kavanagh et al. 1999; Martínez-Vilalta et al. 2002; McElrone et al. 2004). Therefore, incorporating information on the hydraulic properties of roots in addition to that of stems in comparative studies may help identify adaptive relationships between xylem function and other physiological traits.

We examined the relationship between resistance to xylem cavitation and other physiological traits using a group of diverse, co-occurring tree species commonly found in the temperate forest of eastern North America. We determined: (1) if there was a trade-off between vulnerability to xylem cavitation and xylem water transport traits; (2) whether vulnerability to xylem cavitation was correlated with traits associated with leaf carbon and water vapour fluxes; and (3) if the strength of these relationships differed between stems and roots. Because a significant proportion of species at our Duke Forest study site were from the genus Quercus, we also took the opportunity to examine functional associations within this clade by incorporating additional oak species from a nearby temperate sand hill savannah ecosystem. To determine the influence of shared ancestry on the strength of trait relationships and to account for the statistical non-independence of species, we calculated correlations with and without phylogenetic information (e.g. Maherali et al. 2004).

MATERIALS AND METHODS

Study site and plant material

Our study was conducted in the Durham Division of Duke Forest, located near Durham, NC (36°01′N, 78°59′W, elevation 150 m). The climate is warm temperate, with a mean annual temperature of 15 °C and average monthly maxima and minima of 26 °C (July) and −2.3 °C (January), respectively (1971–2000 record; Southeast Regional Climate Center, http://cirrus.dnr.state.sc.us). The mean annual precipitation is 1220 mm, with approximately half this amount (627 mm) received during the growing season (April–September). The site was an 80–100-year-old second-growth mixed hardwood conifer forest upon well-drained loamy soils with a predominantly flat topography. We sampled 14 common canopy and subcanopy tree species from nine seed plant families and a variety of wood types (Table 1). Stems from two additional temperate oak species were sampled from a site typical of the sandhills ecosystem in the south-eastern USA located in the US Army's Fort Bragg military base (35°10′17′′N, 79°22′56′′W). This site is characterized by deep (≈ 4 m) sandy soils and experiences similar weather conditions to the Duke Forest site. The site is managed as a longleaf pine savannah with periodic burning every 3 years that maintains a Pinus palustris canopy with various oak species, of which Quercus laevis and Quercus stellata are the most abundant, in the subcanopy.

Table 1.  Species names, family, wood type and location for the 16 tree species from the Piedmont (Duke Forest) and Sandhills (Fort Bragg) of North Carolina used in the study
SpeciesFamilyWood typeLocation
  1. All angiosperms are winter deciduous and all conifers are evergreen.

  2. The phylogenetic relationship among species is shown in Fig. 1.

  3. RP, ring-porous; sRP, semi-ring porous; DP, diffuse porous; NP, non-porous.

Angiosperms
 Acer rubrum L.AceraceaeDPDuke Forest
 Cornus florida L.CornaceaeDPDuke Forest
 Oxydendrum arboreum (L) DCEricaceaeDPDuke Forest
 Quercus alba L.FagaceaeRPDuke Forest
 Quercus falcata MichauxFagaceaeRPDuke Forest
 Quercus laevis WalterFagaceaeRPFort Bragg
 Quercus nigra L.FagaceaeRPDuke Forest
 Quercus phellos L.FagaceaeRPDuke Forest
 Quercus rubra L.FagaceaeRPDuke Forest
 Quercus stellata Wang.FagaceaeRPFort Bragg
 Cercis canadensis L.FabaceaesRPDuke Forest
 Liquidambar styraciflua L.HamamelidaceaeDPDuke Forest
 Nyssa sylvatica MarshallNyssaceaeDPDuke Forest
Conifers
 Juniperus virginiana L.CupressaceaeNPDuke Forest
 Pinus taeda L.PinaceaeNPDuke Forest
 Pinus echinata MillerPinaceaeNPDuke Forest

Leaf gas exchange

To facilitate access for gas exchange measurements and to ensure that all samples experienced similar light environments, we sampled saplings or young trees 2–4 m tall that occurred either in canopy gaps created by natural disturbance or near clearings associated with forest roads. We measured photosynthetic CO2 assimilation (A) and stomatal conductance (gs) to water vapour with an open gas exchange system (LI-6400, Li-Cor Inc., Lincoln, NE, USA) on 2–3 sun-exposed leaves of 9–12 individuals for each species on July 2000, after the leaves had expanded fully. To ensure that leaf gas exchange values reflected the maximum capacities of leaves, we attempted to make measurements under non-limiting conditions. Therefore, individuals were sampled before midday (930 and 1200 h EST). Red-blue light-emitting diodes maintained incident irradiance at saturating levels (1800 µmol m−2 s−1) and a Peltier cooling module maintained leaf temperature between 25 and 30 °C. Although the relative humidity was not controlled directly, vapour pressure deficit (VPD) varied between 1.5 and 2.0 kPa and approximated ambient conditions. To calculate gs, we used a boundary layer conductance of 1.42 mol m−2 s−1, which was calculated on the basis of leaf area and fan speed using the energy balance algorithms of the LI-6400. Following enclosure in the leaf cuvette, leaves reached steady-state values (e.g. when the coefficients of variation of CO2 and H2O within the chamber were < 0.25%) within 5 min. To assess the trade-off between CO2 uptake and water loss, we calculated instantaneous water-use efficiency (WUE) as A/gs. Gas exchange was expressed on a one-sided leaf area basis for all species. Angiosperm leaves were larger than the cuvette, so the default area of 6 cm2 was used for all calculations. For Pinus taeda, we measured the widest diameter of each needle in the cuvette and calculated projected leaf area as the product of the length of the cuvette (3 cm) and the aggregate diameter of all needles. Juniperus virginiana leaf samples were excised after gas exchange data collection and projected leaf area was measured with a LI-3100A leaf area meter (Li-Cor, Inc.).

After performing gas exchange measurements, we measured shoot water potential (Ψ) with a pressure chamber (Plant Moisture Status Instrument Company, Corvallis, OR, USA). In addition, individual leaves used for gas exchange were excised and placed in plastic bags containing moist paper towels. For angiosperms, individual leaf size was subsequently measured with the LI-3100A. For P. taeda, individual fascicles were separated to allow needles to lay flat on the conveyer belt of the LI-3100A, facilitating projected area measurement. For J. virginiana, leaf size was defined as the projected area of scales on the most distal twig on a branch. Projected leaf area on these scales was measured as described above. All leaves were dried to constant mass in a forced convection oven at 65 °C for 48 h and were weighed. Specific leaf area (SLA) was calculated by dividing leaf size by leaf mass. Leaves were subsequently ground to powder with a Crescent Wig-L-Bug (Crescent Dental, Lyons, IL, USA). Powder samples were assessed for C and N content (%) using a CE Instruments NC 2100 elemental analyser (ThermoQuest Italia, Milan, Italy). Integrated WUE was determined on six of these powdered samples per species by measuring carbon isotope discrimination (δ13C; Farquhar, Ehleringer & Hubick 1989) with a Finnigan MAT Delta Plus XL continuous flow mass spectrometer (Finnigan MAT GmbH, Bremen, Germany).

Vulnerability to xylem cavitation

We measured the vulnerability of xylem to cavitation on stem and root segments of 5–7 of the same individuals of each species previously sampled for gas exchange. We collected 0.5–1.0-m long sun-exposed shoots and 30–70-cm long roots from 20–50-cm soil depths during August and September of 2001 (stems) and 2003 (roots). Stem and root samples were typically 2–5 years old, as determined by growth ring counts of cross sections. To minimize dehydration, all tissues were immediately placed in plastic bags containing moist paper towels. In the laboratory, all stem samples were recut under water to a length of 14 cm, and the cut ends were trimmed with a razor blade. Root samples were recut to longer lengths (between 24 and 59 cm) in order to minimize the refilling of cavitated vessels during measurements, particularly for Quercus. Although we have no information on vessel length distributions for our species, previous studies on diffuse porous angiosperms indicated that the average vessel length for similar aged tissue was < 4 cm (Sperry, Hacke & Wheeler 2005). Average vessel length is typically longer in ring-porous species (e.g. ≈ 13 cm in Vitis vinifera,Sperry et al. 2005), suggesting that some vessels may have been open in Quercus stems.

KH was measured as described by Sperry, Donnelly & Tyree (1988) in an air-conditioned laboratory (20 °C). Segments were cleared of air emboli by perfusing them at high pressure with filtered (0.2 µm) distilled water for 15–20 min at 100 kPa. A hydrostatic pressure of 1.5–2 kPa was used to measure volume flow rate (Q, kg/s), which was calculated by collecting efflux continuously with a vial placed on a 0.0001 g balance connected to a computer. Hydraulic conductivity (KH; kg m MPa−1 s−1) was expressed as the volume flow rate divided by the pressure gradient [Q / (dP / dx)]. Specific conductivity (KS; kg m−1 MPa−1 s−1) was calculated by dividing KH by the segment's cross-sectional xylem area. Leaf specific conductivity (KL; kg m−1 MPa−1 s−1) was calculated by dividing KH by the leaf area distal to the measured segment. Leaf area distal to the measured stem segment was measured on fresh tissue using a LI-3100A leaf area meter (Li-Cor, Inc.).

Vulnerability to drought-induced xylem cavitation was measured as the reduction in KH of a stem or root segment as a function of decreasing xylem tension (a vulnerability curve; Sperry et al. 1988). For stem segments of all species and root segments of J. virginiana, vulnerability curves were constructed using the centrifuge technique (Alder et al. 1997). Segments were spun on an axis for 5 min in a custom designed rotor (Alder et al. 1997) at a specific rotational velocity to produce xylem tensions ranging from −0.5 to −10 MPa in 0.5–2 MPa increments. KH was measured on each stem following each successive spin in the centrifuge. Preliminary experiments indicated that for many species (especially ring-porous Quercus), an initial spin at 1 MPa produced large (70–90%) losses in KH. As a result, it was possible that flushing stems prior to constructing vulnerability curves refilled non-functional conduits (called cavitation fatigue; Hacke, Sperry & Pittermann 2000) or that this initial exposure to low negative pressures caused open vessels to cavitate. Regardless of the process operating, all stems were spun at low speed (−0.5 MPa) after flushing to allow functional conduits to remain filled while embolizing non-functional conduits and open vessels (Hacke et al. 2000). In addition, only water transport measurements made after this initial spin were used in calculations to prevent non-functional conduits and open vessels from contributing to estimates of KH, KS and KL.

Although the use of pure water as a perfusing solution can underestimate KH because of hydrogels in the pit membrane (Zwieniecki, Melcher & Holbrook 2001), we used pure water rather than a solution of KCl for our measurements. The effect of KCl on KH varies considerably across angiosperms (Boyce et al. 2004) and even among individuals of a single species (Zwieniecki et al. 2001). This variable effect may be associated with differences in the ionic concentrations of xylem sap across individuals and species. As a result, the use of a constant KCl solution in our interspecific comparisons may not correct the bias associated with using pure water. Nevertheless, the use of pure water instead of KCl as our perfusing solution may have reduced KH in our study species and may have altered patterns of interspecific variation in hydraulic traits. We note, however, that KCl and pure water have produced similar vulnerability curves in other angiosperm species (Davis et al. 2002). In addition, previous studies have detected trade-offs between vulnerability to cavitation and KH despite using non-KCl solutions (e.g. Sperry & Pockman 1993; Hacke et al. 2000; Martínez-Vilalta et al. 2002).

The large size of conduits in roots (McElrone et al. 2004) prevented us from constructing vulnerability curves with the centrifuge method because attaching samples to the hydrostatic measuring apparatus refilled embolized conduits. Therefore, vulnerability curves for roots were constructed using the air-injection method (Cochard, Cruiziat & Tyree 1992; Sperry & Saliendra 1994) in which xylem cavitation was induced by successively increasing positive air pressures on a segment inside a double-ended pressure chamber. Previous studies indicate that the centrifuge and air-injection methods yield very similar vulnerability curves (Pockman, Sperry & O’Leary 1995). Each root segment was notched (0.5–1 mm deep) with a razor blade to provide an entry point for air, and the bark at the proximal and distal ends was removed. The root segment was then inserted into a pressure chamber with both ends protruding. The root ends were connected to the hydraulic measurement apparatus and KH was measured as described previously. The pressure inside the chamber was then raised to 0.5 MPa and this pressure was maintained for 15 min. After pressurization, the segment was allowed to equilibrate for 10 min at low pressure (0.1 MPa) followed by 5 more minutes at 0 Mpa, and KH was measured again. This process was repeated at progressively higher pressures in 0.5–1.0 MPa increments to generate a vulnerability curve for each segment. To maintain consistency with vulnerability curves constructed with the centrifuge method, KH after pressurization at 0.5 MPa was taken as the reference value.

Percent loss in conductivity (PLC) following each spin in the centrifuge or pressurization of the chamber was calculated as PLC = 100*[(Kmax − KH)/Kmax], where KH is the hydraulic conductivity of the segment measured after each chamber pressurization and Kmax is the hydraulic conductivity after the initial 0.5 MPa spin or pressurization. Vulnerability curves were fit with an exponential sigmoid equation (Pammenter & Van Der Willigen 1998):

image(1)

where Ψ is the negative of the injection pressure, a is a measure of the degree that conductivity responds to injection pressure or tension (curve slope) and b represents the Ψ at which a 50% loss in conductivity occurs (Ψ50 or curve displacement along the x-axis). Coefficients a and b were estimated using the non-linear regression procedure in Systat 8.0 (SPSS, Evanston, IL, USA).

Xylem conduit measurements

Xylem conduit diameter was measured on stem segments used for hydraulic measurements. Stem cross sections (20 µm thick) were made with a sliding microtome (American Optical Co., Buffalo, NY, USA) and stained with toluidine blue. The cross sections were viewed with a light microscope interfaced with a Nikon CoolPix 950 digital camera (Nikon Inc., Melville, NY, USA). The captured images were downloaded onto a computer for image analysis using NIH Image 1.58 (U.S. National Institutes of Health; http://rsb.info.nih.gov/nih-image/). We measured radial strips of cells on sectors spaced at 90° intervals in the outermost portion (2–5 growth rings) of each cross section. Up to 300 conduits were measured for each stem segment. Hydraulic diameter (Dh) of lumens was calculated as Dh = 2xy / [x + y] for tracheids and Dh = (2x2y2 / [x2 + y2])1/2 for vessels, where x and y are the short and long sides of the conduit, respectively (Lewis & Boose 1995). To determine the functional significance of conduit size distribution within each segment, we calculated the hydraulically weighted mean diameter (Dh-w) for each segment as Σ Dh5 / Σ Dh4 (Pockman & Sperry 2000).

Statistical analyses

Because our primary interest was in examining which traits are functionally integrated with xylem vulnerability to cavitation, correlations were examined with root and stem Ψ50 as the anchor traits (i.e. Schwilk & Ackerly 2001). One limitation of solely using cross-species comparisons to determine functional integration of physiological traits is that these correlations may be biased by the similarity of closely related species (Felsenstein 1985; Harvey & Pagel 1991). This shared evolutionary history prevents individual species from being statistically independent data points (Harvey & Pagel 1991). To account for shared evolutionary history, we calculated phylogenetically independent contrasts (PICs, Felsenstein 1985), which are determined for each speciation event in the phylogeny (e.g. as the difference in trait values between sister taxa along each branch; Felsenstein 1985; Garland, Harvey & Ives 1992). This method identifies adaptive relationships because the strength and sign of the contrast correlation indicates whether evolutionary shifts in a trait are associated with changes in another trait or ecological variable (Pagel 1993; Ackerly & Reich 1999).

A fully bifurcated phylogenetic tree (Fig. 1) incorporating taxa sampled at our study site was derived from published sources using methods and assumptions described in Maherali et al. (2004). Relationships between traits were evaluated using traditional cross-species correlations and correlations among PICs. Pearson correlations (R) for the cross-species analysis were calculated using SPSS 10.0 (SPSS, Evanston, IL, USA). PICs and Pearson correlations among PICs were calculated using the Phenotypic Diversity Analysis Programs (PDAP; http://www.biology.ucr.edu/people/faculty/Garland/PDAP.html; Garland, Midford & Ives 1999; Garland & Ives 2000) from within the Mesquite system for phylogenetic computing (Maddison & Maddison 2004). All data were log-transformed prior to analyses. Because trees from various sources were combined to produce our phylogeny, it was impossible to incorporate into our analysis information on phylogenetic branch lengths, which indicate the number of evolutionary changes along each ancestor–descendant pathway (Harvey & Pagel 1991). In the absence of such information, we assumed that branch lengths are equal to calculate PIC correlations, a conservative assumption that minimizes Type I error rate (Ackerly 2000). We note that because our study species were sampled non-randomly and disjunctly across seed plant families, ancestral reconstructions based on parsimony are likely to be biased, and this bias will influence the strength and statistical significance of PIC correlations. However, this limitation must be balanced against inflated Type I error rates of non-phylogenetically corrected cross-species correlations (Garland et al. 1992; Ackerly 2000). To be conservative and to facilitate interpretation, we present both standard and PIC correlations.

Figure 1.

A phylogenetic tree showing the relationship among the species used in this study.

RESULTS

The vulnerability of roots and stems to xylem cavitation varied widely across species (Figs 2 & 3). The most resistant species in the community, J. virginiana, reached 50% stem cavitation (Ψ50) at a water potential that was nearly eightfold more negative than that of the most vulnerable species, Quercus falcata. In general, stems of Quercus were more vulnerable to cavitation than those of other species. Even the Quercus species occurring on well-drained sandy soils at the Fort Bragg site were more vulnerable to cavitation than most of the non-Quercus Duke Forest species. With the exception of Cercis canadensis, Quercus species also had roots that were the most vulnerable to cavitation. In all species, roots were more vulnerable to xylem cavitation than stems (Figs 2 & 3). In addition, the KS of roots was on average 5.6 times higher than that of stems (paired t-test, P < 0.01, Table 2).

Figure 2.

Curves of the vulnerability of xylem to cavitation for stems (filled circles and solid lines) and roots (empty circles and dashed lines) in seven oak species from Duke Forest and Fort Bragg, NC. Although means (± 1 SE) are presented, curves were fit using all data. The xylem tension causing 50% loss in hydraulic conductivity (Ψ50) for each species is shown at the bottom of each panel.

Figure 3.

Curves of the vulnerability of xylem to cavitation for stems (filled circles and solid lines) and roots (empty circles and dashed lines) in nine deciduous and conifer species from Duke Forest, NC. Although means (± 1 SE) are presented, curves were fit using all data. The xylem tension causing 50% loss in hydraulic conductivity (Ψ50) for each species is shown at the bottom of each panel.

Table 2.  Shoot water potential (MPa) during gas exchange measurements, and stem KS(kg m−1 MPa−1 s−1), stem KL(kg m−1 MPa−1 s−1) and root KS(kg m−1 MPa−1 s−1) for each species in the study
SpeciesWater potential (± 1 SE)Stem KSStem KL (× 10−4)Root KS
  1. No water potential data were collected for Q. laevis, Q. stellata and P. echinata.

  2. No root data were collected for Q. laevis and Q. stellata.

Angiosperms
 Acer rubrum−1.30 ± 0.110.55 ± 0.091.20 ± 0.210.90 ± 0.23
 Cornus florida−0.81 ± 0.070.15 ± 0.020.22 ± 0.021.62 ± 0.52
 Oxydendrum arboreum−1.24 ± 0.110.32 ± 0.060.56 ± 0.111.48 ± 0.66
 Quercus alba−1.14 ± 0.191.37 ± 0.182.02 ± 0.2515.4 ± 5.98
 Quercus falcata−1.92 ± 0.071.49 ± 0.582.78 ± 0.794.82 ± 1.86
 Quercus laevis0.49 ± 0.270.84 ± 0.12
 Quercus nigra−1.54 ± 0.261.19 ± 0.202.56 ± 0.482.66 ± 0.89
 Quercus phellos−1.97 ± 0.161.35 ± 0.362.44 ± 0.5811.1 ± 8.23
 Quercus rubra−1.30 ± 0.091.33 ± 0.302.26 ± 0.472.42 ± 0.82
 Quercus stellata0.81 ± 0.251.83 ± 0.49
 Cercis canadensis−1.41 ± 0.090.25 ± 0.070.38 ± 0.113.23 ± 0.87
 Liquidambar styraciflua−1.17 ± 0.100.70 ± 0.111.45 ± 0.292.07 ± 0.94
 Nyssa sylvatica−1.01 ± 0.090.18 ± 0.020.29 ± 0.042.14 ± 0.62
Conifers
 Juniperus virginiana−1.10 ± 0.070.25 ± 0.041.38 ± 0.151.69 ± 1.21
 Pinus taeda−1.57 ± 0.040.32 ± 0.020.99 ± 0.153.82 ± 1.17
 Pinus echinata0.26 ± 0.042.12 ± 0.430.91 ± 0.32

Based on rank order, species with low stem Ψ50 also had low root Ψ50(Fig. 4). However, we observed that differences between stem and root Ψ50 increased as overall vulnerability to xylem cavitation decreased. For example, species with vulnerable xylem, such as Quercus and Nyssa sylvatica, had similar stem and root Ψ50, whereas species with resistant xylem, such as Cornus florida and J. virginiana, had stems with much more negative stem Ψ50 than root Ψ50 (Fig. 4).

Figure 4.

The xylem tension causing 50% loss in hydraulic conductivity (Ψ50) for stems plotted against the same parameter for roots for 14 co-occurring species in Duke Forest.

In the cross-species analysis, stem Ψ50 increased significantly with specific conductivity (KS; Fig. 5a, P < 0.001) and with mean hydraulically weighted conduit diameter (Dh-w; Fig. 5c, P < 0.001), but was only marginally associated with leaf specific conductivity (KL; Fig. 5b, P = 0.06). However, the phylogenetically independent contrast (PIC, Table 3) correlation for each of these relationships was not statistically significant (Fig. 5a–c, insets). The absence of an evolutionary correlation across taxa is consistent with the observation that data points among more closely related species tended to be clustered on all three plots of the relationship between hydraulic efficiency and Ψ50 (Fig. 5a–c). In addition, there was no relationship between root Ψ50 and root KS for both cross-species and PIC correlations (Table 3). Within Quercus, Ψ50 increased significantly with increasing KL(Fig. 6b, P = 0.05) but was not associated with either KS (Fig. 6a) or Dh-w (Fig. 6c). Although significant PIC correlations suggested that there were evolutionary correlations between Ψ50 and KS (Fig. 6a, inset, P = 0.05) and Ψ50 and KL (Fig. 6b, inset, P = 0.01), these relationships were driven by a single contrast between Q. laevis and Q. falcata and therefore may not be biologically meaningful within the genus.

Figure 5.

Specific conductivity (KS; a), leaf specific conductivity (KL; b), and hydraulically weighted conduit diameter (Dh-w; c) expressed as functions of xylem tension causing 50% loss in hydraulic conductivity (Ψ50) for stems of 14 co-occurring tree species from Duke Forest. Note that negative xylem tension data were converted to positive values for log10 transformation. Therefore, the sign of the correlation coefficients are negative, even though they describe positive correlations. Plots of phylogenetically independent contrasts for each pair of traits and the corresponding phylogenetically independent contrast (PIC) correlation coefficients are shown in the insets. †= 0.10, ***< 0.001

Table 3.  Magnitude and statistical significance of Pearson correlations on cross species and phylogenetically independent contrast (PIC) data sets comprising 14 co-occurring tree species in Duke Forest, Durham, NC
Trait or contrastCross speciesPIC
Stem Ψ50Root Ψ50Stem Ψ50Root Ψ50
  • *

    = 0.10,

  • **

    < 0.05,

  • ***

    < 0.01,

  • ****

    < 0.001.

  • Negative xylem tension data were converted to positive values for log10 transformation. Therefore, the sign of the correlation coefficients are negative, even though they describe positive correlations.

  • KS, specific conductivity; KL, leaf specific conductivity; Dh-w, hydraulically weighted mean conduit diameter; AL/AS, leaf: xylem area ratio; A, photosynthetic rate; gs, stomatal conductance; SLA, specific leaf area.

Root Ψ500.83****0.62**
Root KS−0.43−0.02
Stem KS−0.77****−0.43
Stem KL−0.52*−0.30
Stem Dh-w−0.80****−0.45
AL/AS−0.36−0.41−0.09−0.17
A−0.65**−0.90****−0.51**−0.92****
gs−0.47*−0.80****−0.11−0.70***
δ13C−0.31−0.52*0.002−0.32
%N−0.35−0.330.060.10
Leaf size−0.48−0.72***−0.17−0.62**
SLA−0.16−0.440.04−0.43
Figure 6.

Specific conductivity (KS; a), leaf specific conductivity (KL; b), and hydraulically weighted conduit diameter (Dh-w; c) expressed as functions of xylem tension at 50% cavitation (Ψ50) for stems of seven Quercus species. Note that negative xylem tension data were converted to positive values for log10 transformation. Therefore, the sign of the correlation coefficients are negative, even though they describe positive correlations. Plots of phylogenetically independent contrasts for each pair of traits and the corresponding phylogenetically independent contrast (PIC) correlation coefficients are shown in the insets. The statistically significant relationship (b) is driven by a single species (Quercus laevis). *P ≤ 0.05

There were important associations between Ψ50 and leaf level traits, particularly for roots. In the cross-species analysis, both A and gs increased with increasing stem Ψ50 (Fig. 7a & b, P = 0.02 and 0.10, respectively). However, in the PIC analysis, there was only an association between A and stem Ψ50 (= 0.04; Fig. 7a inset) while there was no relationship between gs and stem Ψ50 (Fig. 7b inset). Both the cross-species and PIC correlations for these relationships were sensitive to the inclusion of conifers in the analysis. For example, when only angiosperms were included, A was weakly correlated with stem Ψ50 (= −0.47, P = 0.15; PIC = −0.18, P = 0.30). In contrast, A and gs increased strongly with increasing root Ψ50 in both the cross-species (< 0.001; Fig. 7c and d) and PIC analyses (< 0.01; Fig. 7c and d, inset). These relationships were also strong within the angiosperms (for A versus root Ψ50: R = 0.82, P = 0.002, PIC = −0.79, P = 0.002; for gs versus root Ψ50: R = −0.64, P = 0.04, PIC = −0.58, P = 0.03).

Figure 7.

Photosynthetic rate (A), and stomatal conductance (gs), expressed as functions of stem (a, b) and root (c, d) xylem tension at 50% cavitation for 13 co-occurring tree species in Duke Forest. Gas exchange was not measured on P. echinata. Note that negative xylem tension data were converted to positive values for log transformation. Therefore, the sign of the correlation coefficients are negative, even though they describe positive correlations. Plots of phylogenetically independent contrasts for each pair of traits and the corresponding phylogenetically independent contrast (PIC) correlation coefficients are shown in the insets. †= 0.10, *< 0.05, **< 0.01, ***< 0.001.

Instantaneous WUE, calculated as A/gs, was not correlated with Ψ50 in either stems or roots. In addition, A and gs were not associated with measures of KH (KS and KL) in stems (data not shown, R = 0.35, P = 0.25). Leaf size was not associated with stem Ψ50 but was positively correlated with root Ψ50 (< 0.01, Table 3), a relationship that was robust to phylogenetic correction (= 0.01; Table 3) but not statistically significant when conifers were removed from the analysis (= −0.31, P = 0.35, PIC = −0.23, P = 0.25). Other leaf traits of functional significance such as integrated WUE (measured as δ13C), leaf nitrogen or SLA were either marginally associated (e.g. δ13C) or not associated with stem or root Ψ50 (Table 3).

DISCUSSION

Vulnerability to xylem cavitation and water transport

Studies within single communities have provided evidence that there is a cross-species correlation, and thus a trade-off, between vulnerability to xylem cavitation and xylem transport capacity (Pockman & Sperry 2000; Martínez-Vilalta et al. 2002). On the surface, our results are consistent with these previous studies. We found considerable interspecific variation on how vulnerable stems and roots were to xylem cavitation for a group of co-occurring woody species (Table 2). There were also large differences across species in the efficiency of xylem transport, as measured by KS and KL. However, even though all three measures of hydraulic efficiency (KS, KL and mean Dh-w) increased with Ψ50 across 14 co-occurring species, there was no relationship between these parameters when phylogenetic information was included (Fig. 5a–c, insets). These results are consistent with a broader analysis of stems of 167 woody plant species across a variety of biomes (Maherali et al. 2004). Similarly, there was no relationship between the vulnerability of roots to cavitation and root KS for these species for both cross-species and PIC correlations (Table 3). We conclude therefore that there was weak evidence for a trade-off between Ψ50 and the efficiency of xylem transport within stems or roots for the species in this temperate forest community.

Resistance to water stress-induced cavitation via air seeding in a large majority of angiosperms depends on the surface tension of the meniscus in each of the pores of the pit membrane that connects adjacent conduits (Tyree & Sperry 1989). As a result, an increase in resistance to cavitation must be accompanied by a presumably costly decrease in pore KH (Sperry & Hacke 2004). The absence of an evolutionary correlation between Ψ50 and xylem transport efficiency, despite a trade-off between air seeding pressure and pit conductivity, may be associated with differences in other xylem traits across our taxonomic sample (Fig. 1). For example, increased vessel length and diameter both increase KH and may compensate for hydraulic limitations in the pit membrane (Comstock & Sperry 2000; Sperry & Hacke 2004).

Because the ability to detect trade-offs between vulnerability to cavitation and xylem water transport capacity may be confounded by large differences in xylem vessel length and diameter across taxa, comparative studies on closely related species with similar xylem anatomy may yield insights on the evolution of this physiological trade-off. However, our comparisons of Ψ50 and water transport capacity within a single genus, Quercus, were equivocal. Of the three indices of hydraulic efficiency, only KL was correlated with vulnerability to cavitation in the cross-species analysis (Fig. 6). However, the statistical significance of this correlation was driven by the presence of a single species, Q. laevis. Similarly, the significant PIC correlations between KH (both KS and KL) and Ψ50 were driven by a single contrast between Q. laevis and Q. falcata. The sensitivity of these relationships to a single data point suggests that there may not be an evolutionary trade-off between Ψ50 and xylem water transport capacity in Quercus stems.

Roots were more vulnerable to xylem cavitation than stems (Figs 2–4), a result that is consistent with previous studies (Kavanagh et al. 1999; Jackson et al. 2000; Martínez-Vilalta et al. 2002; McElrone et al. 2004). Interestingly, the magnitude of the difference between stem Ψ50 and root Ψ50 appeared to be correlated with overall vulnerability to xylem cavitation. For example, species that were relatively vulnerable to cavitation, such as Quercus, had stem Ψ50 and root Ψ50 values that differed only modestly (an average of 25%), whereas resistant species such as Oxydendrum arboreum and Cornus florida had stem xylem that was more than twice as resistant to cavitation than root xylem. These differences may be species specific or reflect some level of acclimation by roots of all species to the relatively high soil moisture levels at the Duke Forest site (H. Maherali, personal observation). For example, it is possible that vulnerability to cavitation in roots is more closely calibrated to the less negative water potentials the roots experience relative to stems. Although there is some evidence that cavitation resistance in roots is phenotypically plastic (as reviewed in Sperry et al. 2002), little is known about the degree to which phenotypic plasticity influences vulnerability to cavitation in roots versus stems.

Root KS was higher than stem KS in all species (Table 2), a result that has been reported in other studies (e.g. Kavanagh et al. 1999; Martínez-Vilalta et al. 2002; McElrone et al. 2004). Higher KS in roots also suggests that conduit diameter was greater in roots than in stems. Like differences in Ψ50, systematically larger vessels in roots relative to stems may be associated with less negative water potential present in the root zone. If roots experience limited cavitation, then the maintenance of high KS would decrease the overall hydraulic resistance of the root system, suggesting that root tissue may be specialized for water uptake and transport. It is also possible that roots are more specialized for water transport than stems because biomechanical stresses on xylem for canopy support and wind resistance would be less severe in roots than stems (McElrone et al. 2004).

Vulnerability to xylem cavitation and leaf traits

Our results revealed a strong association between leaf gas exchange and the vulnerability of xylem to cavitation. For example, we observed that A and gs both increased with increasing Ψ50 (Fig. 7a–d). Previous studies have identified roots as the most limiting component of the hydraulic pathway (as reviewed in Meinzer 2002; Sperry et al. 2002). Our results are consistent with this hypothesis; relationships between gas exchange and Ψ50 were stronger for roots than for stems. Moreover, the correlation between root Ψ50 and A or gs was quite robust even with the inclusion of phylogenetic information, suggesting that the functional relationship among these traits is adaptive. In consequence, variation in the vulnerability of roots to cavitation across species may drive interspecific variation in leaf CO2 and H2O fluxes.

The observation that root Ψ50 increased with increasing maximum gs is consistent with a more general trade-off between vulnerability to cavitation and water transport capacity (Fig. 7d). Because the magnitude of gs depends on the hydraulic conductivity of the entire soil–leaf pathway (Sperry & Pockman 1993; Nardini & Salleo 2000; Meinzer 2002), our results suggest that the conducting efficiency versus xylem safety trade-off can be manifested at the level of the whole plant rather than just within specific organs. A relationship between root Ψ50 and maximum gs is also consistent with the expectation of functional coordination between vulnerability to xylem cavitation and the regulation of stomatal conductance across species. For example, species with less negative root Ψ50 might behave as cavitation avoiders through tight regulation of stomatal opening, whereas those species with more negative root Ψ50 could behave as cavitation resisters with relatively weak stomatal regulation (Sparks & Black 1999; Martínez-Vilalta, Sala & Pinol 2004). If stomata act to maintain a constant leaf water potential through hydraulic regulation, stomatal sensitivity to VPD must be positively correlated with the magnitude of maximum gs (Oren et al. 1999). Our results are consistent with this model because species in Duke Forest with high gs, and therefore high stomatal sensitivity to water deficits, also have xylem that is more vulnerable to cavitation.

The correlation between photosynthetic rate and the vulnerability of roots to xylem cavitation may be an indirect effect of coordination between gs and Ψ50, rather than a direct influence of root hydraulic traits on photosynthetic capacity. Several lines of evidence support this conclusion. Firstly, photosynthesis and stomatal conductance are generally correlated both within populations and across species, suggesting that photosynthetic capacity is determined by stomatal limits to CO2 diffusion (Wong, Cowan & Farquhar 1979). Secondly, we observed no relationship between vulnerability to cavitation and other traits associated with carbon uptake, such as N content and SLA (Table 3). Finally, although there was a weak association between root Ψ50 and integrated WUE (as measured by δ13C), variation in carbon isotope discrimination was correlated with gs and not A (data not shown).

Unlike previous studies (as reviewed in Meinzer 2002), we did not observe a significant correlation between KL and leaf gas exchange. It is possible that the relationship between KL and leaf gas exchange was confounded by variation in leaf water potential across species (Table 2). For example, leaf water potential in some oaks (Q. nigra, Q. phellos and Q. falcata) was negative enough to exceed the point at which 50% of stem KH would be lost (Fig. 2). As a result, stomatal closure likely occurred in these species, reducing gas exchange from its maximum. However, Quercus species generally had higher A and gs than other species (Fig. 7a–d), suggesting that variation in leaf water potential may not have affected the relative ranking of gas exchange values among species. An alternate explanation for the absence of a correlation between KL and leaf gas exchange is that stem xylem may have accounted for much less resistance of the hydraulic pathway than that provided by xylem in the petioles and leaf veins. This result has been observed in other studies of deciduous trees (e.g. Brodribb, Holbrook & Gutiérrez 2002; Sack et al. 2003).

CONCLUSIONS

Our findings add to the recent body of evidence documenting an important role for roots in controlling plant water transport (Jackson et al. 2000; Sperry et al. 2002). In particular, we identified a strong link between the vulnerability of root xylem to cavitation and leaf gas exchange across species. As a result, physiological integration between these different organs may represent a trade-off between vulnerability to xylem cavitation and water transport at the whole-plant level and may ultimately reflect correlated evolution between root and leaf traits. Moreover, the relationship between root vulnerability to cavitation and leaf CO2 and H2O fluxes represents a functional link between two major physiological strategies associated with plant performance (Westoby et al. 2002; Ackerly 2004). Further studies of links between hydraulic function and other major axes of plant performance (e.g. leaf life span, seed size, plant architecture) will increase our understanding of how organisms adapt to their environment and will provide a framework for predicting ecological responses to environmental change.

ACKNOWLEDGMENTS

We thank C.W. Cook for assistance in several phases of the project, including plant identification and determination of leaf N. We also thank R.N. Addington for many patient hours spent sectioning samples and measuring the dimensions of xylem conduits. We are grateful to the US Army (Fort Bragg) for access to field sites. We thank C.M. Caruso, H. Freitas, A. Zanne, two anonymous reviewers and members of the Jackson Lab for thoughtful discussions at various stages and for comments on this manuscript. This work was funded by grants from the National Science Foundation, the Mellon Foundation, the Natural Sciences and Engineering Research Council (Canada) and the Portuguese Technology and Science Foundation (BPD/3623/2000 to MCC). This work was also supported by doctoral fellowships from the Portuguese Technology and Science Foundation (PRAXIS XXI/BD/21668/99) and the Gulbenkian Foundation to CFM and a National Science Foundation graduate fellowship to CJW.

Ancillary