Plant material and sample preparation
The study site is a mixed-age ponderosa pine stand located on the eastern side of the Oregon Cascade Range (43°32′ N, 121°41′ W) on private forestland (Crown Pacific Co.). The mean annual precipitation is 645 mm (http:www.orst.eduDepartmentIPPC) and the elevation is 1355 m with soils derived from deep volcanic ash. The stand is a mix of ponderosa pine and lodgepole pine (Pinus contorta) ranging from 15 to 400 years old, and with approximately 250 trees per hectare. The sampled trees were the same trees as used by Pruyn, Gartner & Harmon (2002). Six old-growth trees (> 220 years old) in March 1999, and six mature trees (> 70 years old) and six young trees (> 30 years old) in March 2000 were selected based on their cambial age at breast height (estimated from increment cores), and their health (free of broken tops, stem deformities or disease) (Table 1). After the trees were felled, four heights of the old-growth trees were sampled, at 1 m height (node 220), at nodes 65 (lower third part of the crown), 50 and 15 counting down from the treetop. For the mature trees, only nodes 50 and 15 were sampled, and for the young trees, only node 15. The height to each of these sampled nodes was recorded (Table 2). At each height, and from each tree, discs about 20 cm thick were cut, immediately transported to the laboratory (a 3 h journey) in wet plastic bags, and stored at 5 °C until blocks were prepared within 3 days of felling. A thinner second disc was also cut for use in measuring sapwood and heartwood areas (see below). From each of the thick discs, four blocks were cut: two from the inner sapwood and two from the outer sapwood. Preparation of these specimens made from the blocks followed the procedures outlined by Domec & Gartner (2001). The specimens (about 10 mm radial direction, 10 mm tangential direction, and 130–170 mm axial direction) were split along the grain first with a maul and wedge, and then with a chisel, then stored at 3 °C in clean water which was changed daily until they were used within 3 weeks.
Table 1. Morphological characteristics of young, mature and old-growth ponderosa pine trees sampled from the same site (mean ± SE, n = 6)
|Old-growth trees||33.3 ± 0.4||225 ± 27||62.1 ± 1.9||2390 ± 114|
|Mature trees||12.4 ± 0.9|| 72 ± 5||26.8 ± 1.8|| 375 ± 71|
|Young trees|| 2.9 ± 0.1|| 31 ± 3|| 7.7 ± 0.6|| 55.7 ± 6.2|
Table 2. Morphological characteristics of ponderosa pine trees by height position (node) counting down from the top of the tree, and at the base (about 30 cm; mean ± SE, n = 6)
|Young trees||Node 15|| 1.1 ± 0.2|| 7.7 ± 0.6|| 35.2 ± 5.9||0.12 ± 0.04||100|| 15 ± 1|
|Base|| 0.3||10.1 ± 0.5|| 55.7 ± 4.4||0.14 ± 0.07|| 70 ± 1|| 29 ± 3|
|Mature trees||Node 15|| 8.6 ± 0.8||9.48 ± 0.5|| 55.5 ± 7.4||0.64 ± 0.12||100|| 16 ± 1|
|Node 50|| 1.6 ± 0.2||19.2 ± 0.8|| 192 ± 28|| 9.5 ± 2.9|| 67 ± 1|| 41 ± 2|
|Base|| 0.3||26.8 ± 1.8|| 375 ± 71|| 4.4 ± 2.2|| 71 ± 1|| 54 ± 3|
|Old-growth trees||Node 15||31.0 ± 0.3||6.92 ± 0.2|| 24.1 ± 3.1|| 2.6 ± 0.3||100|| 14 ± 1|
|Node 50||25.5 ± 1.2||22.1 ± 1.1|| 322 ± 61||17.7 ± 3.5|| 66 ± 1|| 41 ± 3|
|Node 65||21.4 ± 1.2||31.8 ± 0.8|| 624 ± 43||53.4 ± 8.3|| 65 ± 1|| 52 ± 1|
|Base|| 0.4 ± 0.1||62.1 ± 1.9||2387 ± 114|| 213 ± 41|| 64 ± 1||116 ± 5|
Hydraulic specific conductivity, vulnerability and relative water content curves
Samples were soaked under a vacuum for 48 h to refill some of the embolized tracheids. Initial specific conductivity (ks(i)) was then measured on a segment as described below. Using Darcy's law, we expressed ks(i) in square metres (m2), which is the direct consequence of separating the viscosity (MPa s) from the water flux (m3 m−2 s) divided by pressure gradient (MPa m−1). Vulnerability curves (VCs) of the tree trunks were constructed using the method described by Domec & Gartner (2001). The method allows calculation of the percentage loss of conductivity (PLC) on segments taken directly from the trunk after moving the sample alternately between the membrane-lined pressure sleeve to measure ks (Spicer & Gartner 1998) and the double-ended pressure chamber for causing embolism using air injection (Sperry & Saliendra 1994). We measured ks(i) using filtered (0.22 µm) water adjusted with HCl to pH 2 in order to prevent microbial growth, and using a hydraulic pressure head of 2.79 kPa. Efflux was collected in a 1-mL-graduated micropipette (0.01 mL graduation). We recorded the time required for the meniscus to cross five consecutive graduation marks, and only used the values that were steady. Samples were subjected to several air pressures ranging from 0.5 to 6.0 MPa until more than 95 PLC was reached. The temperature of the solution (to calculate water viscosity), the fresh mass (Mf), the volume (Vf) and the length of each sample were recorded before and after each hydraulic conductivity measurement (see below). After each ks measurement, we removed about 1 mm of wood from each end of the stems to have new surfaces for each experiment.
Vulnerability curves were fitted by the least squares method based on a sigmoidal function:
where PLC is the percentage loss of conductivity [(ks(i) − ks(Ψ))/ks(i)], the parameter a1 is an indicator of the slope of the linear part of the vulnerability curve and b1 is the pressure at which 50 PLC occurred. The actual slope (s = a1 · 25) of the linear part of the vulnerability curve and the pressures at 12 PLC (Ψ12 = 2/a1 + b1) and at 88 PLC (Ψ88 = −2/a1 + b1) were determined from the fitted curves (Domec & Gartner 2001). The value Ψ12, termed the air entry point (Sparks & Black 1999), is an estimate of the xylem tension at which the runaway cavitation and embolism begin when the resistance to air entry of pit membranes within the conducting xylem is overcome (Sperry & Tyree 1988). Of course, Ψ12 is only a linear approximation of the true air entry point, which from the VCs starts very close to Ψ = 0, but it provides a very useful value to compare among curves. Likewise, Ψ88 is the full embolism point, interpreted as approximating the actual tension of the xylem before it becomes non-conductive (Domec & Gartner 2001).
To estimate the change in relative water content (RWC) associated with embolism, we determined RWC initially and after each applied pressure by recording the sample's volume (Vf; cm3, measuring water displacement on a balance using the Archimedean principle), the fresh mass (Mf; g) and the length. Following final pressurization, we recorded the dry mass (Md; g) and by using the length, back-calculated the dry mass and then calculated RWC assuming a density of dry cell wall material of 1.53 g cm−3 (Siau 1984):
Xylem water deficit or the loss of RWC (100 − %RWC) over the range of applied pressure was fit using the following sigmoidal function:
where Max is the maximum loss of RWC possible, a2 is an indicator of the slope of the linear part of the curve and b2 is the pressure at which Max/2 loss of RWC occurred (ΨMax/2).
Water storage capacity can be defined as the amount of water withdrawn from the stem at a given water potential relative to zero water potential (Holbrook 1995). For the same volume of water, RWC varies depending on wood density (Eqn 2; Domec & Gartner 2002). We expressed water storage capacity as the change in RWC per unit change in applied pressure, which allows us to compare differences in storage capacities that are not only related to a difference in total water volume but also to tissue density. One cannot accurately estimate capillary water changes by using the pressure chamber because positive pressure cannot displace capillary water. The maximum capillary water storage will be released at a pressure potential close to zero, and would stop before −0.5 MPa (Tyree & Yang 1990; Holbrook 1995). Therefore, to avoid underestimating capillary water at low applied pressure, volumetric RWC-based capacitance, defined as CRWC = dRWC/d (Edwards & Jarvis 1982) was computed for a range in each dehydration curve from 0.5 to 2.0 MPa (C0.5−2). This range corresponds to the natural range of water potential encountered by these trees (Hubbard, Bond & Ryan 1999; Ryan et al. 2000; authors’ field observations).
We have analysed these data further by looking at the values from the derivative of Eqn 3, which gives us the slope of the tangent at a given pressure, which, by definition, is the instantaneous water storage capacitance. The derivative of Eqn 3 is:
The instantaneous volumetric RWC-based capacitance was computed at the air entry point Ψ12 (CΨ12) and at a pressure corresponding to the maximum water storage capacity possible within the entire range of applied pressure (Cmax). Cmax occurs when Ψ = b2 and therefore also corresponds to the slope of the linear part of the sigmoidal curve (Domec & Gartner 2001):
Cmax = a2 · Max/4(5)
Wood density and latewood proportion
Wood density values (g cm−3) were determined for each sample that was tested hydraulically:
Density = Md/Vf,(6)
where Md is the oven dry mass (dried at 105 °C), and Vf is the fresh volume (see above). We then used a transverse section made with a microtome and stained with safranin-O to determine the latewood proportion of each sample used for the VCs. We analysed each section (one line scan through each sample) with an image analysis system consisting of a compound microscope, a video camera, and the software package NIH Image (v. 1.60, Rasband 1996). We calculated an average latewood proportion for each sample as the mean of the average latewood proportion of all the growth rings analysed for that sample.
Tree growth rates
Mean annual growth rate in height (cm year−1) for the corresponding number of sapwood rings was estimated by dividing the mean height of growth above each disc by the cambial age (Table 2). Because no heartwood was present at or above node 15, we divided the height to the top of the tree directly by the number of rings. For node 65, the number of rings in sapwood corresponded to the height growth above the next disc (between node 50 and treetop). For the base and node 50, the number of rings in sapwood corresponded to a growth located between nodes 110–230 and nodes 15–50, respectively, so we took a weighted average (by year) between the total annual growth height between these two discs.
Number of rings in heartwood and sapwood as well as heartwood and sapwood areas were determined on the thin disc from each height for each tree. We drew two perpendicular diameters, counted the rings, measured heartwood and total widths, and then averaged values for the two diameters. Sapwood area was taken as the area of whole disc (excluding bark and cambium) minus the area of heartwood. The annual radial growth rate (cm year−1) for each sample (outer and inner) was calculated by dividing the number of growth rings in each sample by the sample width. Using the growth rate of the outer samples, percentage yearly increase in sapwood area (in.Sap in % year−1) was calculated by dividing the difference between the sapwood areas of the last two rings by the total sapwood area, assuming that no sapwood was lost.
Least squares methods were used to fit relationships between hydraulic parameters and applied pressure, and linear and non-linear relationships among hydraulic parameters. Each sample was used as a single replicate that gave a single value of each hydraulic parameter (ks, s, Ψ12, Ψ50, Ψ88, Cmax, CΨ12 and C0.5−2). These hydraulic parameters were compared among locations (radial and height positions) with an analysis of variance (anova). We used a strip-plot randomized complete block design (trees as block) with radial tissue and height position as the strip plot factors. The first task was to investigate the assumptions of normal distribution and constant variance among treatment groups in the residuals. In order to check for normality, we first examined the stem leaf and box-plot of the residuals. It showed a fairly symmetric and non-skewed distribution, which suggested that the residuals were normally distributed. Constant variance over the sample positions was ascertained by examining the plot of the residuals versus the predicted values. The experiment was designed to assess values at both inner and outer sapwood, but for height position we were interested in an estimate of the entire sapwood. Therefore, the effect of height position on the hydraulic parameters was made by weighting the values by the proportion of the total sapwood area occupied by the outer and inner sapwood shells (Table 2). The inner shell represented the area from the third growth ring exterior to the heartwood/sapwood boundary to the middle of the sapwood zone, and the outer shell represented the area from the middle of the sapwood zone to the cambium. All statistical procedures were conducted with Statistical Analysis Systems software (SAS 1997). Least square (LS) means were generated from a PROC MIXED procedure, and multiple comparisons among means were calculated using least square differences (LSD), and the pooled standard error was reported.