Within-stem oxygen concentration and sap flow in four temperate tree species: does long-lived xylem parenchyma experience hypoxia?

Authors


Rachel Spicer. Fax: + 1617.496.5854; e-mail: spicer@oeb.harvard.edu

ABSTRACT

Oxygen levels as low as 1–5% (gaseous mole fraction) occur in secondary xylem, but it is not known if there is a consistent pattern of decline in O2 from the cambium toward the pith, or whether parenchyma cells experience hypoxic conditions deep within the stem. We developed a system for repeated in situ measurement of O2 at different depths within stems of Acer rubrum, Fraxinus americana, Tsuga canadensis, and Quercus rubra. In summer during active transpiration, O2 declined from the cambium toward the heartwood boundary in F. americana, T. canadensis and Q. rubra, but remained constant in A. rubrum. Average sapwood O2 was about 10%, with the lowest values observed in the innermost sapwood around 3–5%. Before spring leaf flush, O2 content in the outer sapwood was reduced in Q. rubra and T. canadensis relative to summer, and was occasionally lower than in the inner sapwood. Sapwood respiration in T. canadensis was constant above 5% O2, but reduced by about 65% at 1% O2. In F. americana, sapwood respiration was constant above 10% O2 but reduced by 25% at 5% O2, and by 75% at 1% O2, the most extreme inhibition observed. However, when prolonged (72 h) exposure to 1%, 5% and 10% O2 was followed by re-equilibration to 10% O2, no inhibition was found. Given the minor (and reversible) effect of low O2 on parenchyma metabolism at levels common in the inner sapwood, it is unlikely that O2 content severely limits parenchyma respiration or leads to parenchyma cell death during sapwood senescence. Within-stem O2 levels may instead be most relevant to metabolism in the cambial zone and phloem, for which sapwood could serve as a significant source of O2.

INTRODUCTION

The gas composition inside large woody stems differs significantly from that of the ambient atmosphere because of cellular respiration in the xylem, phloem and cambium, and the slow rate of gas diffusion through stem tissue. Secondary xylem is typically low in O2 and enriched in CO2, both of which show seasonal and diurnal variation. Reported values for O2 within the xylem during the growing season range from 1 to 5% (mole fraction expressed as a percent) (Eklund 1993, 2000; Gansert, Burgdorf & Losch 2001; Mancuso & Marras 2003) to 12–19% (del Hierro et al. 2002; Pruyn, Gartner & Harmon 2002b), with variation due to differences in time of year, species and ages studied, and position within the stem. Diurnal variation in O2 within xylem is also significant, but small, with reports ranging from 2 to 5% (that is, %O2 at the same position may fluctuate between two and five percentage points daily; Gansert et al. 2001; del Hierro et al. 2002; Mancuso & Marras 2003). This variation is presumably caused by the net effect of changes in parenchyma respiration with temperature and changes in the rate of O2 supply by the transpiration stream (e.g. del Hierro et al. 2002). Although radial diffusion of O2 through the bark occurs, its contribution to the total supply of O2 relative to that of the transpiration stream is likely smaller than once thought (Mancuso & Marras 2003).

Secondary xylem contains 5–35% parenchyma by volume depending on species (Panshin & de Zeeuw 1980), with conifers ranging from about 5–8% and angiosperms from 10 to 35%. These cells live for anywhere from two to 150 years, also depending on species, and may exist at depths of 20 cm or more within the stem. It is the death of xylem parenchyma in the innermost sapwood that defines and drives the formation of heartwood (i.e. heartwood no longer contains living parenchyma cells; Chattaway 1952; Frey-Wyssling & Bosshard 1959), and there is often a decline in sapwood respiration from the cambium toward the sapwood/heartwood boundary (Goodwin & Goddard 1940; Pruyn, Gartner & Harmon 2002a; Pruyn et al. 2002b). Most measurements of xylem O2 have been made in small stems or at shallow radial depths (∼ 1 cm), and it is not known how O2 levels change across the sapwood in large stems, or how relevant these levels are to parenchyma physiology.

Increasing evidence suggests that the transpiration stream is a major source of O2 in the xylem (Gansert et al. 2001; del Hierro et al. 2002; Mancuso & Marras 2003). However, in many species there are regions of secondary xylem that contain living cells but no longer conduct water. Sapwood contains living parenchyma cells by definition (Panshin & de Zeeuw 1980), but in many trees only the outermost sapwood is active in water transport (Phillips, Oren & Zimmermann 1996; James et al. 2002). In conifers, in which the innermost sapwood remains conductive, water transport may cease shortly before parenchyma cell death during heartwood formation (Nobuchi & Harada 1983). Oxygen has been shown to decline to about 6% near the sapwood/heartwood boundary in Pseudotsuga menziesii (Pruyn et al. 2002b), but it is not known if hypoxic or anoxic conditions occur in the innermost sapwood of other species, or if xylem parenchyma respiration is inhibited at O2 levels common within stems. Despite the current gaps in our understanding of the relationship between within-stem O2 and parenchyma physiology, it has been suggested that O2 depletion is responsible for the decline in respiration and ultimate death of xylem parenchyma cells during the transition from sapwood to heartwood (Panshin & de Zeeuw 1980). One study also suggests that anoxia, generated by flushing with either CO2 or N2, can accelerate the rate of heartwood formation (Eklund & Klintborg 2000).

The main objective of this study was to determine whether or not O2 levels in the innermost sapwood, in which parenchyma cells die during the formation of heartwood, decline to levels low enough to either severely limit respiration or cause cell death. A novel technique is described that allows for repeated ‘point’ measurements of O2 within the xylem of multiple individuals over a period of one year or more. In situ O2 levels at different radial depths are reported for four species representing a diverse range of xylem anatomies and sapwood longevities. Secondary objectives were to relate within-stem spatial patterns of cell death and sap flow to observed O2 levels, and to determine the sensitivity of parenchyma respiration to low O2.

MATERIALS AND METHODS

Plants and field site

Six individuals each of Fraxinus americana, Acer rubrum, Quercus rubra, and Tsuga canadensis were randomly selected from a five hectare tract of natural forest in Harvard Forest, Petersham, MA, USA (42.5°N lat, 72°W long, 220 m elev). Mean temperatures at Harvard Forest are 20 °C in July, −7 °C in January, with 110 cm annual precipitation distributed evenly throughout the year. Trees were cored with a 5-mm borer at 1.4 m above-ground along two radii to determine the locations and ages of sapwood and heartwood (Table 1). All six trees were used for measurement of O2 content at different radial depths. Three of the six trees per species were used for sap flow measurements. For sapwood respiration measurements, large diameter (12 mm) cores were extracted from 20 additional, randomly selected trees each of T. canadensis and F. americana.

Table 1.  Diameter outside bark, age, and number of years and width of sapwood (SW; based on vital staining of live cells with triphenyl-tetrazolium chloride) for study trees
 d.b.h. (cm)Age (year)SW rings (year)SW width (cm)
  1. All measurements were made at 1.4 m height. Mean (± SE) shown for n = 6 trees per species.

Fraxinus americana34.7 ± 2.568 ± 545 ± 27.9 ± 0.9
Acer rubrum33.7 ± 1.749 ± 338 ± 310.1 ± 1.0
Tsuga canadensis39.2 ± 4.060 ± 225 ± 25.8 ± 0.8
Quercus rubra33.1 ± 5.857 ± 68 ± 11.8 ± 0.3

Oxygen probe tube construction and installation

Metal tube assemblies were permanently installed in each of the 24 trees (six per species) used for O2 measurements. Stainless steel tube assemblies (Fig. 1) allowed for a small (about 0.2 cm3) gas headspace within the tree to be accessed periodically with a needle-tipped fibre-optic O2 sensing probe through a layered silicone rubber septum. A gas tight seal was formed by compressing the septum against an aluminium spacer with a smaller diameter tube, which served as a guide for the O2 probe during measurements (i.e. the probe was threaded through the inner guide tube to pierce the septum). Tubes ranged in length from 5 to 18 cm, depending on the depth of insertion into the tree. The septum could be replaced and resealed after repeated use. Because it is necessary to correct all O2 measurements for temperature, a thermocouple junction was sealed in the innermost end of the tube assembly, adjacent to the headspace, with the leads extending to a thermocouple connector outside the tree.

Figure 1.

Oxygen tube assembly. A removable inner guide tube compresses a silicone septum against the aluminum spacer to form a gas-tight seal, allowing the headspace to come into equilibrium with the local internal gas composition in the stem. Oxygen measurements are made by threading the needle-tipped O2 probe through the inner guide tube and penetrating the septum. The aluminium spacer is glued in place with a thermocouple junction exposed to the gas headspace and wire leads running to a connector outside of the tree. Small bolts were used to tension the inner guide tube against the stationary outer tube to compress the septum, which could be removed and replaced after repeated use.

Tubes were installed in all trees in April 2002 before the start of cambial activity. Holes 9 mm in diameter (0.5 mm smaller than the tube diameter) were drilled to three or four radial depths per tree (Table 2), all at 1.4 m height, spaced evenly about the circumference of the tree and placed randomly with respect to compass direction. Tubes were installed in the heartwood and innermost and outermost sapwood in all four species, and midway between the two sapwood positions in F. americana and A. rubrum, both of which have relatively wide sapwood. Tubes were coated outside along their length with all-weather epoxy and hammered into the drilled holes to the desired depths, leaving about 3 cm of the tube extending out of the tree. A final bead of epoxy was placed in a circle around the point of insertion against the bark. When not in use, the opening to each inner guide tube was plugged with a small rubber stopper and the entire tube assembly extending from the tree was enclosed in a 3-mil plastic bag for protection from rain and animals.

Table 2.  Maximum depth (cm inside cambium) of O2 probe tube assemblies (mean ± SE; n = 6 trees)
 OuterMiddleInnerHW
  1. Inner positions were placed approximately 5 mm outside the sapwood/heartwood (SW/HW) boundary, with middle positions placed equidistant between the outer and inner positions in species with wide sapwood, and in one individual of T. canadensisa. Sap flow probes were installed to the same maximum depths except for the outer position, where probes were installed to 1 cm, so that the entire heated/sensing region was in the sapwood.

Fraxinus americana0.53.8 ± 0.57.0 ± 0.910.7 ± 0.9
Acer rubrum0.55.0 ± 0.59.0 ± 1.012.6 ± 0.5
Tsuga canadensis0.53.8a4.5 ± 0.610.6 ± 1.1
Quercus rubra0.51.3 ± 0.38.4 ± 1.5

Oxygen measurements

Oxygen measurements were made on all trees during June 2002, when leaves were fully expanded and there was active transpiration, and April 2003, before leaf flush. Oxygen was measured in the headspace behind the septum with a needle-tipped (21 gauge) fibre-optic probe and spectrometer-based sensing system (Ocean Optics, Inc., Dunedin, FL, USA). The probe was calibrated in the laboratory with humidified gas standards, and temperature calibrated using gas standards incubated in a water bath over the range of 15–35 °C. The spectrometer was carried in the field in an insulated cooler and connected to a laptop computer. Readings of ambient air O2 (corrected for temperature) between within-stem O2 readings allowed for frequent checks of the calibration. Within-stem O2 readings stabilized within 15–20 s of insertion through the septum, and repeated insertions gave readings within ± 0.1% (mole fraction; e.g. between 7.8 and 8.0% O2). Temperature within the headspace was recorded for each O2 measurement for later corrections.

Each tree was measured for O2 on multiple (3–5) days, typically between 1000 and 1500 h, and the mean over several days was taken. Diurnal variation was estimated both by making point measurements every 5 h on the same tree, and by continuously monitoring a single position within the sapwood. Both measurements suggested that maximum diurnal variation was about 3% (i.e. the same position might read between 6 and 9% over a 24-h period) in regions of active sap flow, which agrees with other published reports (Gansert et al. 2001; del Hierro et al. 2002; Mancuso & Marras 2003). Tests for gas leakage around the tube were performed in the laboratory on fresh stem segments, sealed at both ends with wax, with a PVC chamber sealed against the bark and around the tube through which 100% O2 was continually flushed. There was no evidence of increased rates of diffusion of O2 into the stem (as measured with the fibre-optic O2 probe) as a result of O2 flushing.

Sap flow measurements

Granier sap flow sensors of variable lengths were constructed following a modification of the methods of James et al. (2002). Briefly, 12-mm-long segments of 2.4 mm diameter aluminium tubing were crimped onto the ends of lengths of PEEK tubing, so that 10 mm served as the heated/sensing length. A 0.8-m length of resistance wire (120 Ω total resistance) was used to generate 0.15 W of power per probe by supplying a constant current of 35 mA with a voltage regulator and potentiometer.

Probes were installed in June 2002 in three of the six trees used for O2 measurements, with the maximum (innermost) radial depth matching the maximum depth of the O2 probe guide tubes (Table 2). Probes were installed at the same depth and height as each O2 tube, but between 5 and 10 cm away horizontally. Constant-current heater circuits were powered by deep cycle batteries in the field, and data was collected with a Campbell datalogger/multiplexer every 10 or 15 min. Because of the total number of probes used and distances between trees in the field, not all probes were monitored at the same time. The standard Granier calibration (Granier 1987) was used to convert the temperature difference between the heated and unheated probes to sap flux density (g m−2 s−1) assuming a constant velocity along the 1 cm sensor length. This calibration has been validated (Clearwater et al. 1999) and tested both for this probe design (James et al. 2002) and for the species included in this study (Catovsky, Holbrook & Bazzaz 2002). Unheated data was collected for several days for each probe after installation to check for intrinsic diurnal trends in temperature difference. In all cases the unheated diurnal variation was less than 1% of the diurnal variation for heated probes.

Sapwood tissue respiration

The effect of low O2 on sapwood respiration was tested by measuring rates of O2 consumption by fresh sapwood following equilibration to known O2 concentrations. Four large increment cores (12 mm diameter) were extracted from each of 10 individuals of F. americana and T. canadensis in July 2004. These species were chosen to include a conifer and angiosperm with significant declines in O2 across the sapwood. Segments about 1 cm3 were cut from the outer sapwood (excluding the cambial zone), weighed, wrapped in moist cheesecloth, and placed in open glass vials (6 mL). One of each of the four cores per tree was randomly assigned to a gas treatment: 1, 5, 10 and 15% O2 (balance N2). Open vials were placed in disposable plastic glove bags and flushed repeatedly (five or six times) with the treatment gas, sealed, and placed in cold (4 °C) storage. Vials were incubated under the treatment gas for 36 h with flushes at 12 and 24 h. Prior tests had shown that 36 h was sufficient to fully equilibrate the samples so that their internal gas composition matched that of treatment gas, while cold storage suppressed respiration and extended the length of time respiration rates (after return to 20 °C) remained stable. At the end of the equilibration period, the vials were sealed (following a final flush) with aluminium crimp-top lids with a central butyl rubber septa. Sealed vials were then transferred to a water bath where they were incubated at 20 °C throughout the measurement period.

Oxygen was measured in each vial every 4 to 6 h for 24–36 h by penetrating the septum with a needle-tipped O2 probe, which was calibrated with humidified standards at 20 °C. The slope of the linear portion of the curve (typically the first 24 h) was taken as the rate of O2 consumption and converted to mol O2 h−1 by calculating the total volume of gas in the vial. Samples were then removed from vials, measured for fresh volume by volume displacement, dried in an oven at 105 °C for 36 h, and weighed. Respiration was expressed on a per fresh tissue volume basis (i.e. mol O2 cm−3 h−1). Similar to other published results (Pruyn et al. 2002a,b) tests showed that respiration was unchanged following 2 weeks of cold storage. Fresh weight, fresh volume and dry weight were used to calculate the total fresh tissue volume occupied by air, water and cell wall material by assuming a constant density for oven-dried solid cell wall material of 1.5 g cm−3 (Panshin & de Zeeuw 1980).

The ability of parenchyma cells to recover from prolonged hypoxia was tested in a separate experiment using identical methods with the following exceptions. Outer sapwood segments from an additional 10 individuals each of F. americana and T. canadensis (three cores per tree, each randomly assigned to one of three treatments) were first incubated at 1, 5 and 10% O2 for 72 h at room temperature to allow for normal respiration under prolonged exposure to the treatment gases. All samples were then transferred to 10% O2 and allowed to re-equilibrate in cold storage (4 °C) for 36 h. Vials were then sealed, incubated at 20 °C in a water bath, and respiration rates were measured (all starting at 10% O2) as described above.

Statistical analyses

Tissue respiration data were tested in an ancova using the surface area of each sample as a covariate, due to the effect of surface area on rates of diffusion. The difference between species in the relative proportions of tissue occupied by cell wall, air and water was tested in an anova. In both cases, tree within species was used as a blocking factor. The effects of radial position, season, and their interaction on in situ O2 content were tested within each species in an anova with tree as a blocking factor. A mixed model analysis, which would account for the hierarchical structure in the O2 data (radial position within tree, tree within species, and repeated measures between seasons), was not performed due to the number of missing values (i.e. for any given radial position, only four of the six trees may be represented, and these trees likely differ between seasons), and the fact that species differed in the total number of radial positions. In all cases, significant main effects were compared with a Fisher LSD correction for multiple comparisons; significant interaction terms were compared using contrasts and Bonferroni-corrected P-values to limit the experiment-wise error rate (Sokal & Rohlf 1995).

RESULTS

Oxygen contents

Oxygen levels generally declined from the outer toward the inner sapwood in Q. rubra and F. americana, but remained constant in A. rubrum(Fig. 2). The inner sapwood had significantly lower O2 than the outer sapwood in both seasons in F. americana and Q. rubra (Fisher adjusted P-values = 0.03 and 0.06, respectively) but only in the summer in T. canadensis (adjusted contrast P-value < 0.005). The effect of radial position depended on the season only in T. canadensis (i.e. the radial position × season interaction was significant; P-value = 0.04). The lowest O2 levels recorded in the innermost sapwood were about 3%, and these were observed in F. americana, T. canadensis, and Q. rubra, although rarely in each species.

Figure 2.

Oxygen content (mole fraction expressed as percentage) at different radial depths within the stem during summer (leaves fully expanded) and spring (before leaf flush); mean ± SE, n = 6 trees. Shaded region represents heartwood as determined by vital staining with triphenyl-tetrazolium chloride. Oxygen was measured in situ with a fibre optic needle-tipped and spectrometer through permanently installed tubes.

Oxygen levels measured before leaf flush in the spring were significantly lower than those measured in the summer throughout the sapwood in Q. rubra (P-value < 0.0001), but only in the outermost sapwood in T. canadensis (adjusted contrast P-value < 0.005), and were unchanged in the heartwood. In contrast, spring O2 levels were higher than in summer in A. rubrum (P-value = 0.003) with no effect of radial position (P-value = 0.5). In Fraxinus, only radial position had a significant effect on O2 (P-value = 0.04), with higher O2 in the outer sapwood relative to both inner sapwood and heartwood (Fisher-adjusted P-values = 0.03 and 0.01, respectively).

Sap flow

Species varied considerably in the extent to which the inner sapwood was functional in water transport (Fig. 3). A. rubrum, T. canadensis, and Q. rubra all had significant flow in the innermost sapwood (31 ± 3, 37 ± 7 and 25 ± 6% of outer flow, respectively; mean ± SE across 6 d, n = 3 trees). In A. rubrum the middle sapwood was nearly as conductive as the outer sapwood (87 ± 6% of outer flow rates), whereas in F. americana only the outer sapwood was conductive (<1% of outer flow in both middle and inner positions). In no case was heartwood conductive.

Figure 3.

Sap flux density (g m−2 s−1) at different radial depths within stems of Acer rubrum, Tsuga candensis, Quercus rubrum, and Fraxinus americana. Two individuals are shown for each species. Sap flow was measured at the same radial depths as O2 measurements using modified Granier sensors with 1-cm-long heated/sensing tips. Outer positions were in the outermost 1 cm of sapwood; inner positions were 5 mm from the sapwood/heartwood boundary; middle (mid) was half way between inner and outer; heartwood (hw) was half way between the sapwood/heartwood boundary and pith.

Respiration and tissue air/water content

The effect of O2 level on sapwood respiration was strongly dependent on species (i.e. there was a significant species × treatment interaction, P-value < 0.0001), although respiration in F. americana and T. canadensis was unchanged at O2 levels above 10 and 5% O2, respectively (Fig. 4). In F. americana, respiration was reduced by less than one-quarter between 10 and 5% O2, and by about one-third between 5 and 1% O2. In T. canadensis, only 1% O2 significantly reduced respiration relative to the maximum rate, but to a lesser extent than in F. americana. Sapwood respiration following 72 h of exposure to 1 and 5% O2, and then a period of equilibration to 10% O2, was unchanged in both species relative to the 10% O2 control (Table 3). In no case did exposure to 1% O2 for up to 7 d result in significant parenchyma death, as judged by vital staining with triphenyl-tetrazolium chloride.

Figure 4.

Respiration (mol O2 cm−3 h−1 × 10−7) of excised sapwood tissue of Tsuga canadensis and Fraxinus americana following equilibration to four known O2 concentrations. Within each species, bars with the same lowercase letters were not significantly different (adjusted P-values > 0.05; contrasts of interaction means from ancova).

Table 3.  Sapwood tissue respiration (mol O2 cm−3 h−1 × 10−7) measured at 10% O2 following a 72-h incubation at 20 °C in 1, 5 and 10% O2 (mean ± SE; n = 10 trees)
 1% O25% O210% O2
  1. There were no significant differences between gas treatments, nor was there a significant gas ×species interaction (anova with tree as block).

Fraxinus americana5.0 ± 0.35.5 ± 0.45.5 ± 0.3
Tsuga canadensis3.6 ± 0.23.9 ± 0.33.4 ± 0.2

The two species differed significantly in the proportions of fresh volume occupied by water, cell wall and air (Fig. 5; adjusted P-value < 0.0001 in all cases). Sapwood of F. americana was composed of roughly equal proportions of water, air and cell wall material, with air occupying about 31 ± 1% (mean ± SE) volume, whereas T. canadensis tissue was only about 11 ± 1% air.

Figure 5.

Proportion of fresh tissue volume occupied by water, cell wall, and air in Tsuga canadensis and Fraxinus americana. Within each tissue volume component (e.g. air), bars with the same lowercase letters indicate that the two species were not significantly different (Fisher-adjusted P-values < 0.0001 for all significant differences, anova).

DISCUSSION

Oxygen levels as low as 1–5% (gaseous mole fraction) have been recorded in sapwood, but most of these measurements have been made in young, small-diameter trees (Eklund, Little & Riding 1998; Mancuso & Marras 2003) or at shallow radial depths (e.g. 5–10 mm beneath the cambium; Eklund 1990, 1993; del Hierro et al. 2002). Increasing evidence suggests that the transpiration stream is a major source of O2 in secondary xylem (Gansert et al. 2001; del Hierro et al. 2002; Gansert 2003; Mancuso & Marras 2003). Given that the cessation of water transport typically precedes the death of parenchyma cells during heartwood formation by one to many years, it is possible that parenchyma cells deep within stems experience hypoxic or even anoxic conditions. We have found that although O2 levels are often reduced in the inner sapwood, they are not low enough to cause cell death, and would be unlikely to cause a biologically significant reduction in respiration rate except at the lowest O2 levels. In four temperate species that differ widely in anatomy and sapwood longevity, average sapwood O2 was about 10% with the lowest values observed in the innermost sapwood around 3–5%. Sapwood tissue respiration was unaffected by a drop from 15 to 10% O2 in F. americana, and by a drop from 15 to 5% in T. canadensis. When measured at 5% O2, respiration in F. americana was only reduced to 75% of the maximum rate. At 1% O2, a value lower than any actually observed in the innermost sapwood, respiration was reduced to just 35 and 25% of the maximum rate for T. canadensis and F. americana, respectively. Respiration was able to recover to normal rates when measured at 10% O2 following 72 h at 1 and 5% O2, and parenchyma remained alive for up to 7 d under 1% O2. Given the relatively minor effect of reduced O2 on parenchyma metabolism at O2 levels that appear to be common in the innermost sapwood, and the moderate effect found at the most extreme levels (whose occurrence is likely rare and transient), it is unlikely that O2 content plays a role in the death of parenchyma cells during sapwood senescence.

In situ O2 measurements in woody stems may instead be most relevant to cellular metabolism in the outermost region of the stem. Our measurements suggest that in spring, when the cambium is active and secondary xylem and phloem begin to differentiate, O2 levels can be drawn down through a combination of increased consumption and reduced supply (i.e. no active transpiration). We saw a marked decrease in O2 in the spring (before leaf flush, relative to summer) in T. canadensis and Q. rubra, both of which had active cambia at the time of measurement (based on appearance and bark slippage). In contrast, A. rubrum, a species in which xylem production begins several weeks after leaf flush (Suzuki, Yoda & Suzuki 1996; Lavigne, Little & Riding 2004), had slightly higher O2 levels throughout the stem in spring relative to summer. The timing of xylem production in particular is related to stem respiration (Lavigne et al. 2004) because the cambium produces far more xylem cells than phloem cells. This difference in the rate of cell production between xylem and phloem may be greatest in spring, at least in conifers (Wilson & Howard 1968). The decrease in O2 in spring for T. canadensis was such that the lowest O2 levels within the stem occasionally (in some individuals) occurred in the outer sapwood.

Drawdown of O2 in the outermost sapwood is the result of O2 consumption by respiring xylem parenchyma, but could also include consumption by cambial tissue and phloem, in which respiration is markedly (2–10x) higher due to the greater proportion of living cells (Goodwin & Goddard 1940; Pruyn et al. 2002a,b). Several of our outermost tubes were pushed outward within the stem through the action of cambial activity during the second (2003) spring so that the headspace in the tube was in the cambial zone or phloem, and these positions produced some of the lowest values observed (approximately 1% O2; these were excluded from the analysis). This is in keeping with observations of a positive relationship between O2 levels within the cambium and cambial activity (as measured by cell number; Eklund et al. 1998), and of hypoxic conditions developing within the phloem (van Dongen et al. 2003). It has also been suggested that the vascular cambium may become anaerobic and undergo fermentative metabolism (Kimmerer & Stringer 1988). In Q. rubra, in the absence of transpiration in spring, the greatest reduction in O2 was in the inner sapwood. This cannot be explained by cambial activity, but could be the result of locally increased sapwood respiration before bud break. There is some evidence that respiration increases during the dormant season in the innermost sapwood (Shain & MacKay 1973). The extent to which sapwood serves as an O2 source for the cambial zone and phloem will depend on the diffusive properties of the different tissues. The cambium has historically been viewed as a barrier to diffusion of gas into the stem (e.g. Hook & Brown 1972), but if the phloem and cambium are strongly hypoxic, then O2 permeability of the cambium may be more relevant to diffusion from the xylem to the phloem, across the cambium.

The combined spatial patterns of O2 content and sap flow during active transpiration also underscore the importance of gas diffusion into and within the stem. In the summer, T. canadensis and F. americana show similar rates of O2 decline toward the sapwood/heartwood boundary, despite very different flow rates in the inner sapwood: in T. canadensis flow is only reduced by about 60% (relative to outer sapwood), whereas in F. americana there is essentially no flow except in the outermost annual rings. Although sap flow measurements were not concurrent with O2 measurements (they are separated by about 1 month), there is unlikely to be any change in the relative difference in flow between radial positions within the stem during that time. In ring-porous species, the outermost annual ring is functional before leaf expansion (Suzuki et al. 1996), whereas in diffuse-porous and conifer species, the contribution of the outermost annual ring will be small over a 1-cm distance (the probe heated/sensing length). Maintenance of such high levels of O2 in the absence of flow in F. americana is not the result of a lower respiration rate, as tissue respiration is higher in F. americana than in T. canadensis, even in the inner sapwood (unpublished results for inner position). Instead, a higher air fraction in F. americana tissue is likely to facilitate movement of gases within the stem. Given that diffusion of gases is 10 000 times faster in gas than in liquid, the volume fraction of gas will have a profound effect on the diffusive properties of woody tissue.

The effects of low O2 on plant tissue respiration are well-studied in roots and aquatic plants where flooding creates hypoxic conditions, but vascular tissue has received less attention. We found no respiratory inhibition until O2 was reduced to below 10% (F. americana) or 5% (T. canadensis). A ‘critical oxygen pressure’ between 10 and 5% O2 is in agreement with others reported in the literature for excised plant tissue in water-saturated air (Saglio, Raymond & Pradet 1983; Saglio et al. 1984). When incubated in aqueous solution, respiration shows exponential or logarithmic decline from higher O2 levels (i.e. inhibition occurs below 20–30%; Geigenberger et al. 2000), but this is likely due to restricted diffusion through an aqueous medium and an unstirred boundary layer (Saglio et al. 1984). Respiration in T. canadensis was less inhibited by low O2 than in F. americana, which is surprising given that diffusion through F. americana tissue is likely to be faster because of the higher air content (i.e. slowed diffusion of O2 to the active site of cytochrome oxidase means that effectively, respiration is more sensitive to reduced O2). It is possible that respiration in T. canadensis is inherently more robust against low O2 conditions (or was in such a state at the time of sample collection), or that the pre-measurement equilibration time served to acclimate T. canadensis more than F. americana through a shift in metabolic pathways toward O2 conservation (Xia & Saglio 1992; Andrews et al. 1993; Andrews et al. 1994; Drew et al. 1994; Drew 1997; Geigenberger et al. 2000; van Dongen et al. 2003).

The system of permanently installed tubes described here allows in situ‘point’ measurements of gaseous O2 and is well-suited to large sample sizes, field measurements and long-term studies, unlike current systems, which are designed for continuous data collection at a single position within the stem and the capture of diurnal variation (Gansert et al. 2001; del Hierro et al. 2002). Diurnal variation of O2 is likely to be greatest in the outermost sapwood, where variation in both flow rate (driving O2 supply) and temperature (driving respiration and therefore O2 demand) is greatest. Although our measurements were not always made when O2 was at a daily minimum, our overestimation of the minimum O2 levels experienced in the innermost sapwood is likely to be very small. Diurnal variation reported for the outside of the stem during active transpiration is typically two to four percentage points [e.g. diurnal min–max of 2–6% in Olea europaea (Mancuso & Marras 2003) and 3–5% in Betula pendula (Gansert et al. 2001)]. Reduced or negligible sap flow in the innermost sapwood and reduced temperature variation deep within the stem both should minimize the variation in O2 near the sapwood/heartwood boundary.

Reported patterns of sapwood O2 and sap flow suggest that the determinants of spatial patterns of O2 within secondary xylem are complex. Among the four published studies to date reporting concurrent sap flow and O2 measurements, two have found daily O2 maxima to correspond with peak sap flow and stable O2 minima in the absence of flow (del Hierro et al. 2002; Mancuso & Marras 2003), while two have found minima at the end of the light period (Gansert et al. 2001; Gansert 2003). Supply of O2 will be determined by sap flow (and the dissolved O2 concentration in soil water) and diffusion of O2 into and within the stem. Demand will be determined by parenchyma respiration, which is in turn determined by temperature, within-stem gas composition, and possibly by parenchyma age. The effect of low O2 on parenchyma respiration should provide a negative feedback, such that depletion of O2 leads to decreased consumption. It is not known whether xylem parenchyma show acclimation to hypoxic conditions as in other plant tissues, including down-regulation of glycolysis and preferential induction of ethanolic rather than lactic acid fermentation under extreme hypoxia or anoxia (Xia & Saglio 1992; Drew et al. 1994; Drew 1997; Geigenberger et al. 2000; van Dongen et al. 2003). A better understanding of how parenchyma cell age, gas composition (both O2 and CO2), and temperature all affect sapwood respiration will be key to determining within-stem patterns of hypoxia and anoxia.

ACKNOWLEDGMENTS

The authors wish to thank the US Environmental Protection Agency for funding through a Graduate Fellowship, Harvard Forest for the use of resources and facilities, and Brent Berger (NSF REU Program at Harvard Forest) for assistance in the field. Two anonymous reviewers provided helpful comments on the original manuscript.

Ancillary