Sap flow rates and sapwood density are critical factors in within- and between-tree variation in CO2 efflux from stems of mature Dacrydium cupressinum trees

Authors


Summary

  • • Measurements of CO2 efflux from stems and branches, sap velocity, and respiratory activity of excised wood cores were conducted in Dacrydium cupressinum trees that differed in diameter, age, and canopy emergence. The objective of this study was to determine if consistent linkages exist among respiratory production of CO2 within stems, xylem transport of CO2, and the rate of CO2 diffusing from stem surfaces.
  • • Stem CO2 efflux was depressed during periods of sap flow compared with the efflux rate expected for a given stem temperature and was positively correlated with sapwood density. By contrast, no significant relationships were observed between CO2 efflux and the respiratory activity of wood tissues.
  • • Between 86 and 91% of woody tissue respiration diffused to the atmosphere over a 24-h period. However, at certain times of the day, xylem transport and internal storage of CO2 may account for up to 13–38% and 12–18%, respectively, of woody tissue respiration.
  • • These results demonstrate that differences in sap flow rates and xylem anatomy are critically important for explaining within- and between-tree variation in CO2 efflux from stems.

Introduction

Respiration in stems and branches accounts for 5–42% of total autotrophic respiration in forest ecosystems (Waring & Schlesinger, 1985; Lavigne et al., 1997; Damesin et al., 2002) and 5–15% of gross primary productivity (Meir & Grace, 2002). Despite its importance to tree carbon balances and forest ecosystem carbon cycles, our ability to predict and scale woody tissue respiration is limited. This is attributable, in part, to an inadequate understanding of the physiological processes within stems that regulate CO2 production and/or influence the diffusion of CO2 through stems and branches into the atmosphere.

Rates of respiration are determined largely by temperature (Amthor, 1989) and exponential temperature response functions are typically used to predict temporal variation in woody tissue respiration. However, there is no generally accepted basis for expressing rates of woody tissue respiration or scaling estimates to the forest stand level, although several methods have been utilized. These methods include stem surface area (Kinerson, 1975; Linder & Troeng, 1981), sapwood volume (Ryan, 1990; Ryan et al., 1994), and tissue nitrogen concentration (Ryan, 1991). Also, rates of woody tissue respiration vary widely within trees and between stands, and reliable predictors to account for this variability have been elusive. For instance, large variation in respiration has been found within trees, as respiration rates in the tree crown may be 19–42 times greater than rates at the base of the stem (Sprugel, 1990; Damesin et al., 2002). Additionally, maintenance respiration rates were found to differ by approx. 100–125% between stands of Abies balsamea (Lavigne et al., 1996). Lastly, deviations from the expected relationship between respiration and temperature are regularly observed in woody tissues, as measurements of CO2 efflux often exhibit a pronounced diel hysteresis in which different efflux rates are observed at the same temperature at different times of the day (Ryan et al., 1995; Lavigne et al., 1996).

Diffusion of respiratory CO2 from stems and branches to the atmosphere may be strongly influenced by the movement of sap in xylem conduits. Xylem sap contains dissolved gases, including CO2, and therefore provides a mechanism for the internal movement of CO2 within trees (Negisi, 1979; Stringer & Kimmerer, 1993). The CO2 concentration of gas in xylem tissues ranges between 2 and 12% (Eklund, 1990, 1993; Teskey & McGuire, 2002) and may serve as either a source or a sink for the CO2 that diffuses through stems and branches to the atmosphere. Xylem transport of respiratory CO2 from lower to upper regions of trees may contribute to both the diel hysteresis in the relationship between stem CO2 efflux and temperature and the high apparent rates of respiration from branches. McGuire & Teskey (2004) concluded that CO2 produced by woody tissue respiration diffuses through the bark to the atmosphere (CO2 efflux), is transported upwards by flowing sap (xylem transport flux), or is temporarily stored in the stem as the result of transient changes in the dissolved CO2 concentration of sap (storage flux). These authors determined that 15–75% of respiratory CO2 produced within woody stems may be retranslocated by the xylem stream during periods of transpiration while, at night, nearly 100% of respiratory CO2 diffuses into the atmosphere (McGuire & Teskey, 2004). In this study, measurements of CO2 efflux rates from stems and branches and sap velocity were made in contrasting Dacrydium cupressinum (rimu) trees. The objectives were (1) to derive estimates of the respiration rate of woody tissues, the amount of respiratory CO2 transported within the xylem stream, and internal storage of dissolved CO2, and (2) to determine if patterns in stem CO2 efflux, xylem transport, and CO2 storage can be observed in trees that vary considerably in diameter, age, and canopy emergence.

Additionally, we sought to compare measurements of CO2 efflux from stems and branches with the respiratory activity of underlying tissues and with characteristics of sapwood, such as average sap flux density and sapwood density, that may influence the diffusion of respiratory CO2 to the atmosphere. Descriptions of the radial trend in respiratory activity of wood with depth, i.e. from the bark to the pith, indicate that respiration is highest in the inner bark (composed of living phloem and cambial tissues) and declines with increasing depth into the sapwood (Møller & Müller, 1938; Goodwin & Goddard, 1940; Shain & Mackay, 1973; Pruyn et al., 2002a, 2002b, 2003). Respiration of woody tissues has also been found to increase with tree height in Pinus ponderosa (Pruyn et al., 2002a) and Pseudotsuga menziesii (Pruyn et al., 2002b), consistent with measurements of high CO2 efflux in upper boles and canopy branches. We hypothesized that variation in CO2 efflux between stems and branches is proportional to differences in the respiratory activity of wood cores excised from these woody tissues. In particular, respiration in the inner bark is likely to influence CO2 efflux because of its high metabolic activity and proximity to the stem surface. To our knowledge, this study serves as the first published attempt to compare CO2 efflux from stem surfaces directly with an estimate of the relative CO2 production.

By comparing CO2 efflux from trees differing greatly in size, age, and canopy emergence with estimates of respiratory activity in wood tissues and measurements of sap flow, we anticipated finding consistent linkages between respiratory production of CO2 within stems, xylem transport of CO2, and the actual CO2 diffusing from tree stems. Reliable patterns would greatly improve our ability to model woody tissue respiration from stems and branches at the stand level and to predict differences both between trees and between stands.

Materials and Methods

Site description

The study site was situated within a mixed conifer-angiosperm forest located at Okarito Forest, Westland, New Zealand (lat 43.2°S, long 170.3°E, 50 m above sea level). Mean annual temperature is 11.3°C, with a small range between winter and summer of 8.6°C, and low air saturation deficit. The forest is located on terrace outwashes from glacial moraines formed ∼20 000 yr ago. Annual rainfall is high, approx. 3400 mm, and evenly distributed throughout the year. As a result of the abundant rainfall, the soils are frequently saturated, highly leached, and extremely acidic (pH 3.8–4.4 to a depth of 500 mm). The soil nitrogen concentration at an adjacent similar site was 633 µmol g−1 and the soil-extractable phosphorus concentration was 12 µmol g−1 (Richardson et al., 2004). The soils have a high organic matter content (approx. 30%), low permeability, and low porosity.

The native lowland forests of New Zealand, such as this field site, are characterized by large, emergent conifer (Podocarpaceae) trees with a dense, mixed understory (Ogden & Stewart, 1995). The study plot (50 m × 50 m) is dominated by 200–400-yr-old rimu (Dacrydium cupressinum Lamb) trees with a mean canopy height of 20 m. Other common tree species include kamahi (Weinmannia racemosa), Westland quintinia (Quintinia acutifolia), miro (Podocarpus ferruginea), and southern rata (Metrosideros umbellata). The basal area distribution on the study plot was 73%D. cupressinum, 12%W. racemosa, 10%Q. acutifolia, and 5% other species.

Measurements of CO2 efflux from stems and branches

Measurements of CO2 efflux rates were made on nine D. cupressinum stems, ranging in diameter from 0.18 to 0.67 m, for up to 160 h in January 2002. Measurements were also made on eight canopy branches with diameters between 0.04 and 0.14 m for 36 h. Branches were selected from two trees accessible from the 22-m permanent tower located at the study site. Clear polycarbonate gas-exchange chambers were attached to the selected stems at ∼1.3 m above the ground. The chambers were half-cylindrical in shape, enclosed 0.025 m2 of stem surface, and were equipped with small 24-V fans (Model D241L-24VDC; Micronel, Vista, CA, USA) to adequately mix the air within the chamber. The chambers used for branch measurements were smaller (∼0.003 m2 of branch area) and opaque. Also, the small size of the branch chambers eliminated the need for fans to thoroughly mix the air volume in the chambers.

The bark of D. cupressinum is smooth and typically did not require scraping of loose bark or fissures to achieve an adequate seal between the chambers and the bark. Closed-cell neoprene foam was fitted to the chamber edges and served as a gasket between the tree and the chambers. In stems, caulking cord (Mortite, Inc., Kankakee, IL, USA) was applied around chamber edges to seal small gaps that typically occurred between the foam gasket and the tree. In branches, foam gaskets were not used and the caulking cord was applied directly between the chambers and the bark surface. The chambers were then secured tightly to stems and branches using ratchet straps or hose clamps. Corticular photosynthesis does not occur in D. cupressinum stems; dissection of the inner bark tissue indicated that photosynthetic chlorenchyma tissue is absent.

The difference in CO2 concentration between air entering and leaving chambers attached to trees was measured with an infrared gas analyzer (Model LI-6262; Li-Cor Inc., Lincoln, NE, USA) in an open flow system. A datalogger and relay driver (Models CR23X and SDM-CD16AC; Campbell Scientific Inc., Logan, UT, USA) controlled a series of solenoid valves (Model MEBH-3-0,9-AW-QS-3; Festo AG & Co., Esslingen-Berkheim, Germany) that determined the flow of compressed air (430–470 µmol mol−1[CO2]) through the measurement chambers. Air flow to the chambers was maintained at 0.5 l min−1 by a flow controller (Model SR-10; Sierra Instruments, Monterrey, CA, USA), but only during measurements. As a result, CO2 accumulated in the chambers between measurements and was then flushed out during the next measurement period. Measurements were only considered valid if the CO2 efflux rate was stable (< 3% variation) for the last 60 s of the measurement period. This criterion prevented artificially high rates of CO2 efflux caused by leaky chambers and/or the slow flushing of accumulated CO2. The sampling duration for each chamber was 12 min (5 min for branches) with all stems sampled once during each 132-min measurement cycle (55-min measurement cycles for branches). A measurement cycle consisted of nine stem (or branch) sampling periods and tests of the zero and span calibrations of the gas analyzer. Sapwood temperatures were measured using constantan-chromega thermocouples placed 15 mm beneath the bark surface and adjacent to the sampling chamber.

The temperature response of CO2 efflux (EA) was determined using a modified Arrhenius function as adopted by Turnbull et al. (2003), where CO2 efflux at a given sapwood temperature is described as:

image(Eqn 1 )

[Eo, CO2 efflux rate (µmol CO2 m2 s−1) at the base temperature To (here 288 K or 15°C); Ta, stem sapwood temperature (K); Rg, gas constant (8.314 J mol−1 K−1); Ao, a parameter related to the energy of activation (kJ mol−1 K−1) which describes the shape of the temperature response]. Nonlinear curve fitting for temperature response curves was conducted with SigmaPlot 2001 (SPSS, Inc., Chicago, IL, USA). To facilitate comparisons with other studies, Q10 parameters were calculated (the ratio of CO2 efflux at 25°C divided by CO2 efflux at 15°C).

Measurements of sap flux density

Sap flux density (m3 H2O m−2 sapwood h−1) was measured concurrently with CO2 efflux utilizing the thermal dissipation technique in five rimu stems (Granier, 1985, 1987). Two probes were inserted into the stems to a depth of 20 or 40 mm, with one probe located 40 mm above the other, at a height of 2.5 m above the ground. Sap velocity was related to the temperature difference between the two probes, measured using a thermocouple located midway along each probe every 10 s, and half-hourly averages were recorded on a datalogger (Model CR21X; Campbell Scientific, Inc., Logan, UT, USA). A more detailed description of the methods utilized for measuring sap flux density at the Okarito Forest plot can be found in Barbour & Whitehead (2003).

Estimating woody tissue respiration and internal fluxes of CO2

Measurements of stem CO2 efflux and sap flux density were used to derive estimates of the respiration rate of the underlying woody tissues, the amount of respiratory CO2 transported within the transpiration stream, and the CO2 flux resulting from internal storage of CO2 for five D. cupressinum trees ranging in diameter from 0.27 to 0.61 m. This was accomplished using a modified version of the theoretical framework outlined by McGuire & Teskey (2004):

image(Eqn 2 )

[β (m2/m3), a tree-specific parameter related to the ratio of stem surface area to sapwood volume; RS (µmol CO2 m−3 s−1), CO2 produced by woody tissue respiration per unit sapwood volume; EA (µmol CO2 m−2 s−1), CO2 efflux from the bark to the atmosphere per unit stem surface area; FT (µmol CO2 m−3 s−1), xylem transport of CO2 per unit volume; ΔS (µmol CO2 m−3 s−1), storage of CO2 in the stem per unit volume associated with transient changes in the dissolved CO2 concentration of sap]. ΔS is not the quantity of CO2 stored in the stem; rather it is the rate of change of this stored CO2, i.e. the rate at which CO2 dissolves in or escapes from the xylem sap (McGuire & Teskey, 2004).

Direct measurements of EA and sap flux density were made in five rimu stems for five consecutive days. During the night, FT and ΔS were assumed to be negligible, resulting in EA being equal to RS. This is in accordance with the observations of McGuire & Teskey (2004) in which ≥ 93% of respiratory CO2 diffused into the atmosphere at night. The night-time temperature response of CO2 efflux, utilizing only data collected between 00:00 and 06:00 h, was then used to predict woody tissue respiration (RS) during the day when xylem transport and storage fluxes of CO2 occur:

image(Eqn 3 )

(To, 15°C; inline image and inline image, parameters derived from the temperature response of EA between 00:00 and 06:00 h). The domain of this model, which was determined by the range of sapwood temperatures measured between 00:00 and 06:00 h, accounted for 56.6–74.4% of the daytime temperature range for which values of RS were estimated. Differences between the predicted rates of RS and the measured rates of EAβ were then calculated, hereafter referred to as RS − EA (µmol CO2 m3 s−1), and assumed to be attributable to xylem transport (FT) and/or storage (ΔS) of CO2. Therefore, positive values of RS − EA indicate that the rate of EAβ was less than the rate of RS predicted from the current sapwood temperature, whereas negative values of RS − EA suggest that EAβ was greater than the rate of RS.

Values of RS − EA were then partitioned into FT and ΔS components. The magnitude of FT was assumed to be dependent solely on the current sap flux density. The proportion of RS −EA attributable to FT was determined by linear regression analyses of the average maximum sap flux density recorded on each measurement day, calculated as the average of sap flux density measurements recorded between 12:00 and 16:00 h on that day, and the concurrent average value of RS − EA. Separate linear regressions were conducted for each of the five sampled trees. Therefore:

image(Eqn 4 )

(ν, sap flux density; α, a tree-specific parameter representing the slope of the linear relationship between ν and RS − EA). The variability in RS − EA that was not explained by sap flux density was assumed to result from dissolved CO2 moving into or out of storage. Therefore, ΔS was calculated as the difference between RS − EA and the rate of FT predicted from the sap flux density. Positive values of ΔS indicate an increase in the dissolved CO2 concentration of xylem sap, whereas negative values represent a decrease in the dissolved CO2 concentration of xylem sap. This method of partitioning RS −EA into FT and ΔS components requires the assumption that ΔS is negligible during times of maximum sap flux density. This assumption is supported by the finding that the CO2 concentration of xylem sap is often well correlated with sap velocity (Teskey & McGuire, 2002). Therefore, both the rate of change of the CO2 concentration of xylem sap and the resulting changes in storage are assumed to be low during the 3–5-h peak in sap velocity typically observed in D. cupressinum.

Wood respiratory potential: radial and vertical profiles and temperature response

Wood cores were extracted from 11 D. cupressinum stems at a height of 1.3 m during January 2003 with a 5-mm increment borer. Three of these trees were also sampled at four heights along the stem including the lower stem (at 1.3 m height), lower canopy (∼1 m above the lowest canopy branch), mid canopy (at the midpoint between the top and bottom of the canopy), and upper canopy (∼1–2 m from the tree top). These three trees differed greatly in diameter (0.18, 0.31 and 0.63 m) and height (17.1, 24.1 and 28 m) and, as a result, sampling points along the stems differed in absolute height. During transportation back to the laboratory, cores were stored in the dark and kept in plastic bags containing a moist paper towel to reduce desiccation of the samples. The excised cores were then cut into sections ∼15 mm in length and stored for 24 h at 4°C. Core segments were cut corresponding to wood tissue types including inner bark (containing phloem and cambial tissues), outer and inner sapwood, sapwood–heartwood boundary, and outer, middle, and inner heartwood. Discrimination of the sapwood–heartwood boundary was determined visually in the field, shortly after the cores were removed, from the change in wood coloration. Outer sapwood was defined as the 15-mm section of sapwood located adjacent to the cambium whereas inner sapwood was the 15 mm of sapwood located adjacent to the sapwood–heartwood boundary. The next 15-mm section of wood was identified as the sapwood–heartwood boundary.

Respiratory activity of the core segments was determined by polarographical measurements of oxygen consumption in a Clark-type liquid-phase oxygen electrode (Model D10; Rank Brothers, Cambridge, UK). Oxygen consumption was assayed at 20°C in 20 mm MES buffer (Azcón-Bieto et al., 1994) that had been equilibrated in ambient air. Before respiration measurements, core segments were incubated at the measurement temperature for 2 h to allow respiratory activity to stabilize. Excised core segments were placed in the electrode cuvette and the depletion of oxygen was recorded for 20 min. Results from these oxygen electrode measurements are presented as the ‘respiratory potential’ of the extracted wood, following Pruyn et al. (2002a), rather than its respiration rate as the electrode cuvette/buffer conditions are significantly different from conditions within the tree stem. As a result, respiratory potential measurements are not directly comparable, in absolute terms, to measurements of CO2 efflux. However, it is hypothesized that the relative differences between trees in respiratory potential and CO2 efflux will be consistent with each other. Cumulative respiratory potentials for the volume of wood under the CO2 efflux chambers were estimated by multiplying the core-based measurements by the volume of wood corresponding to that tissue. The total volume of wood under the chamber was assumed to be a sector from the pith of the stem to its surface.

In order to compare the effects of temperature on respiratory potential with the temperature response of stem CO2 efflux, additional cores were extracted from three rimu trees at two heights (at 1.3 m above the ground and in the upper canopy). The respiratory potential of each extracted core was then measured sequentially at four different temperatures (10, 15, 20, and 25°C). The wood cores were allowed to acclimate to each temperature for 2 h before measurement in the oxygen electrode. The respiratory response of the cores to temperature was described by fitting data to both modified Arrhenius and Q10 functions. Parameters for Arrhenius function are similar to those used for CO2 efflux except that Po (nmol O2 g−1 s−1) is used to describe the rate of respiratory potential at the base temperature of 15°C and Ta is the measurement temperature of the oxygen electrode.

Nonparametric Wilcoxon signed rank tests were used to test for differences in respiratory potential and nitrogen content between wood tissues. Differences in the temperature response parameters (Po, Ao, and Q10) of respiratory potential between wood canopy positions and in comparison to temperature response parameters of CO2 efflux were assessed using paired t-tests. Both analyses were performed using SPSS 13.0 (SPSS, Inc., Chicago, IL, USA) and differences were considered statistically significant if probabilities (P) were < 0.05.

Measurements of wood density, water content, and nitrogen concentration

The oven-dry density, or specific gravity (Panshin & de Zeeuw, 1980), for each core segment was determined as the ratio of dry mass to fresh volume. Fresh volume was calculated by carefully measuring the length and diameter with electronic calipers at four positions around the circumference of each segment. Dry mass was determined after wood samples were oven-dried for 48 h at 75°C. Volumetric wood water content was determined by dividing the difference between fresh and dried wood mass by fresh wood volume for each sample. The nitrogen concentration of the extracted wood cores was obtained on dried and ground wood material using a CNS autoanalyzer (Carlo Erba NA 1500; CE Instruments, Milan, Italy).

Results

CO2 efflux from stems and branches and interactions with sap flow

Sapwood temperature did not explain all diel variation in CO2 efflux (EA) from woody stems of D. cupressinum (Fig. 1a). At many times during the measurement period, CO2 efflux did not respond to, or declined during, coincident increases in sapwood temperature during the day. However, the response of night-time stem CO2 efflux to variation in sapwood temperature was well described by the modified Arrhenius model (Eqn 3) in all trees (r2 = 0.65–0.87; P < 0.001), as shown in Fig. 1b. The standardized residuals from these response curves exhibited uniform variance across the modeled temperature range, with a slight tendency for greater variance at sapwood temperatures greater than 18°C. Stem CO2 efflux at night (inline image) was found to be 0.43–0.95 µmol CO2 m−2 s−1 at 15°C and the temperature response parameter (inline image) ranged from 53.65 to 91.18 kJ mol−1 K−1. Branch CO2 efflux was between 0.96 and 5.04 µmol CO2 m−2 s−1 at 15°C and was 1.5–8.0 times greater than efflux rates from the stems of the same trees. Stem CO2 efflux rates were not related to diameter; however, linear regression analysis indicated that branch diameter is a strong predictor of branch CO2 efflux rates (y = 0.4114x − 0.7156; r2 = 0.82; P < 0.01).

Figure 1.

The diel variation in stem CO2 efflux and sapwood temperature for two trees (a) and the relationship between stem CO2 efflux at night and sapwood temperature for five rimu (Dacrydium cupressinum) trees (b). The fitted curves in (b) result from the modified Arrhenius equation described by Turnbull et al. (2003) (see Eqn 1). Modeled parameter values for Eo (CO2 efflux rate at 15°C; µmol CO2 m−2 s−1) and Ao (a parameter related to the energy of activation; kJ mol−1 K−1), respectively, were 0.54 and 53.65 for tree 5, r2 = 0.67; 0.76 and 59.90 for tree 12, r2 = 0.86; 0.64 and 64.11 for tree 18, r2 = 0.84; 0.88 and 68.26 for tree 38, r2 = 0.81; 0.95 and 91.18 for tree 39, r2 = 0.85. All nonlinear regressions were highly significant at P < 0.0001.

Analysis of the calculated differences between estimated EAβ and predicted RS (RS − EA) suggested that there is a diel trend in the proportion of woody tissue respiration that diffuses to the atmosphere. RS − EA tended to become more positive, i.e. CO2 efflux was lower than would be expected from the predicted rate of woody tissue respiration, in the late morning (typically between 09:30 and 11:30 h). This increase in the magnitude of RS − EA corresponded to the onset of transpirational sap flux in the D. cupressinum stems, as shown in Fig. 2. RS − EA tended to approach zero, i.e. the measured CO2 efflux was close to the predicted rate of woody tissue respiration, between 00:30 and 03:00 h at night. These times correspond to the cessation of both sap flux and stem re-filling in the D. cupressinum stems. This pattern was observed over several days and in all five of the trees that were sampled for both sap flux density and stem CO2 efflux.

Figure 2.

The daily pattern of sap flux density (solid line) and RS − EA (circles) measured in one Dacrydium cupressinum tree over 5 d. RS − EA was calculated as the difference between the measured rate of CO2 efflux (EAβ, where EA is the CO2 efflux rate and β is a tree-specific parameter related to the ratio of stem surface area to sapwood volume) and the predicted rates of woody tissue respiration (RS).

For each of these trees, the maximum sap flux density recorded on each measurement day was found to be positively correlated with the concurrent value of RS − EA (r2 = 0.54–0.82; P < 0.001), as shown in Fig. 3. These relationships only held when examining variation in sap flux density resulting from day-to-day environmental changes and were not consistent for differences in sap flux density between trees on a given day. The respiratory CO2 transported in the xylem stream (FT) was hypothesized to be proportional to RS − EA at periods of maximum sap flux density. Therefore, the correlation between RS − EA and maximum sap flux density suggests that larger amounts of CO2 produced by woody tissue respiration (RS) were transported in the xylem stream (FT) on days when high rates of maximum sap flux density were observed.

Figure 3.

Variation in the difference between the predicted rate of RS and the measured rate of EAβ (RS − EA) as function of daily maximum sap flux density in five Dacrydium cupressinum trees over 4 d (Rs, CO2 produced by woody tissue respiration; EA, CO2 efflux rate; β, a tree-specific parameter related to the ratio of stem surface area to sapwood volume). Maximum sap flux density represents the average of measurements between 12:00 and 16:00 h on a single day. The solid lines represent least squares regressions. The β parameter values (m2 m−3) for the sampled trees were: tree 5, 28.2; tree 12, 69.4; tree 18, 33.6; tree 38, 28.2; tree 39, 23.7.

In all trees, hysteresis was observed in the relationship between sap flux density and RS − EA, as shown in Fig. 4a and b. This hysteresis is attributable to variation in the storage flux (ΔS) caused by changes in the dissolved CO2 concentration of xylem sap. Between 06:00 and 12:00 h, the magnitude of RS − EA was typically less than would have been predicted from the linear relationship between daily maximum values of RS − EA and sap flux density, indicating that CO2 was being released from storage. By contrast, between 18:00 and 00:00 h, CO2 tended to enter into storage as the magnitude of RS − EA was greater than that predicted from the relationship between daily maximum values of RS − EA and sap flux density.

Figure 4.

(a) A theoretical representation of the xylem transport and CO2 storage components of RS − EA (Rs, CO2 produced by woody tissue respiration; EA, CO2 efflux rate). Values of RS − EA greater than xylem transport, indicated by the solid line, are associated with positive values of ΔS and storage of CO2 in xylem sap (ΔS, storage of CO2 in the stem per unit volume associated with changes in the CO2 concentration of sap). By contrast, values of RS − EA less than xylem transport are associated with negative values of ΔS and release of CO2 from xylem sap. Solid curved arrows represent the diel pattern of hysteresis observed in the relationship between RS − EA and sap flux density. (b) Measurements of sap flux density and RS − EA for a single tree, tree 38, over 3 d. Symbols represent measurements for different times of day. The solid line represents a least squares regression of the relationship between daily maximum sap flux density and RS − EA, also shown in Fig. 3, for this tree.

Across all nine sampled trees, the stem characteristic that best predicted between-tree differences in stem CO2 efflux was the density of the outer sapwood (Fig. 5). No significant linear relationships (P > 0.05) were found between CO2 efflux and tissue nitrogen concentration in inner bark or sapwood tissues, sapwood depth, tree size, cumulative respiratory potential, or respiratory potential in any specific wood tissues (i.e. inner bark, outer or inner sapwood, or sapwood–heartwood boundary). In addition, the temperature responses of stem CO2 efflux and respiratory potential of wood tissues from the lower stem were not consistent with each other, as the values of Ao and Q10 for stem CO2 efflux were significantly lower than those of lower stem respiratory potential (paired t-tests, P < 0.05, Table 1).

Figure 5.

The relationship between the wood density of outer sapwood and the rate of stem CO2 efflux (Eo) normalized to 15°C in eight Dacrydium cupressinum trees. The solid line represents a least squares regression: y = 2.55x + 0.48; r2 = 0.66; P < 0.05.

Table 1. Parameter values from fitted temperature–response curves for respiratory potential in inner bark and outer sapwood tissues and stem CO2 efflux of Dacrydium cupressinum trees
Respiratory potentialStem CO2 efflux
Inner barkOuter sapwood
Lower stemUpper canopyLower stemUpper canopy
  1. Temperature–response curves for respiratory potential were calculated at two stem positions (lower stem and upper canopy); the temperature response of stem CO2 efflux was only measured at the lower stem position. The temperature–response of both respiratory potential and stem CO2 efflux was modeled using both a modified Arrhenius function and a Q10 function. Values shown are means (± standard error); n = 3. Within each row, different letters adjacent to listed values indicate statistically significant differences at P < 0.05 according to paired t-tests.

  2. P o, respiratory potential at 20°C; Ao, a parameter related to the energy of activation; Q10, relative change in respiratory potential with a 10°C increase in temperature (15–25°C).

P o (nmol O2 g−1 s−1) 0.29 (0.02)a 0.40 (0.02)b 0.08 (0.01)c 0.15 (0.02)d 
A o (kJ mol−1 K−1)82.3 (15.4)a43.8 (10.1)ab79.4 (22.1)a41.4 (2.1)ab45.5 (1.8)b
Q 10  3.31 (0.63)a 1.88 (0.25)ab 3.33 (0.97)a 1.79 (0.05)ab 1.92 (0.05)b

Estimates of woody tissue respiration and internal fluxes of CO2

The relative magnitudes of the component fluxes (EA, FT and ΔS) of woody tissue respiration (RS) exhibited substantial diel changes (Table 2). CO2 efflux to the atmosphere (EA) comprised greater than 99% of woody tissue respiration between 00:00 and 08:00 h when sap flux was negligible. However, between 16:00 and 20:00 h, EA accounted for only 70.0% of woody tissue respiration on average and as little as 62.5% in one tree. Xylem transport of CO2 (FT) was negligible during the night and, as expected, increased in relative importance with the onset of sap flux 1–2 h after sunrise, reached a maximum of 12.6–37.6% of RS between 12:00 and 16:00 h, and declined thereafter. The movement of CO2 into and out of storage (ΔS) was also found to be an important component of the diel variations in woody tissue respiration. The movement of dissolved CO2 out of xylem sap to the atmosphere (i.e. ΔS was negative) was estimated to be greatest between 08:00 and 12:00 h, accounting for 1.7–15.8% of RS. By contrast, CO2 tended to move into storage during the late afternoon and evening, reaching a peak of 12.5–18.4% of RS between 20:00 and 00:00 h.

Table 2. Average flux densities of stem respiration (RS) and its component fluxes (EA, FT and ΔS), the relative contribution of each component flux to RS, and 24-h totals for RS and its component fluxes
Time (h)Stem temperature (°C)CO2 flux densitiesContribution to RS (%)
R S (µmol CO2 m−3 s−1) F T (µmol CO2 m−3 s−1)ΔS (µmol CO2 m−3 s−1) E A (µmol CO2 m−2 s−1) F T ΔS E A
  1. Component fluxes of RS include stem CO2 efflux (EA), xylem transport (FT), and CO2 storage (ΔS) as outlined in Eqn 2. Values represent averages derived from measurements on five Dacrydium cupressinum trees sampled for 2 d. The minimum and maximum values observed across the five trees for each time interval are shown in parentheses. The average value for the parameter β for these five trees was 36.6 (m2 m−3).

00–0416.732.2 (17.1, 60.9)   0.4 (0.0, 1.8) 0.3 (−0.9, 1.7) 0.888 (0.57, 1.13)   0.6 (0.0, 2.9)  0.7 (−2.8, 5.8)99.2 (94.1, 102.8)
04–0816.631.4 (16.9, 59.5)−0.5 (−3.0, 0.4)−1.3 (−3.1, 0.3) 0.897 (0.60, 1.12)−0.5 (−5.1, 1.5)  −2.8 (−8.8, 0.0)99.4 (98.5, 103.6)
08–1217.735.2 (18.3, 65.5)   3.0 (−1.6, 6.7)−4.7 (−10.1, −1.4) 0.940 (0.65, 1.24)  11.1 (−2.4, 20.7)−12.1 (−15.8, −1.7)98.1 (83.5, 102.4)
12–1619.843.0 (20.8, 81.9)   9.7 (3.4, 16.7)−1.7 (−4.8, 0.5) 0.922 (0.62, 1.26)  24.3 (12.6, 37.6)  −3.5 (−11.0, 1.4)78.4 (73.4, 85.0)
16–2020.848.1 (22.3, 89.0)  10.6 (3.2, 13.4) 5.9 (2.6, 12.0) 0.913 (0.58, 1.24)  18.9 (10.6, 31.0) 11.6 (7.7, 16.2)70.0 (62.5, 74.0)
20–0019.241.3 (20.2, 76.0)   3.0 (0.5, 11.5) 6.2 (3.9, 9.4) 0.915 (0.56, 1.23)   5.1 (1.2, 15.2) 15.8 (12.5, 18.4)81.3 (77.8, 84.8)
24-h total (mmol m−2) 79.3  10.6 (5.8, 13.7)  1.4 (−0.9, 3.8)88.1 (85.8, 91.2)
24-h total (mol m−3)  3.33   0.38 0.07   

E A, FT, and ΔS differed greatly in their relative contributions to woody tissue respiration (RS) over a 24-h period (Table 2). The net contribution of ΔS to woody tissue respiration was found to be small, as diffusion of CO2 out of storage was largely offset by movement of CO2 into storage later in the day. Xylem transport of CO2 was more important to the internal cycling of carbon than the storage flux and accounted for 6–14% of respired CO2, across the five sampled trees, over the course of an entire day. Although FT and ΔS were large components of RS at various times of the day, EA was by far the most important CO2 flux and accounted for 86–91% of RS over a 24-h period.

Between-tree variation in the 24-h totals for FT and ΔS in the five sampled D. cupressinum trees was principally related to the 24-h total woody tissue respiration of the tree (RS) (Fig. 6a). It is important to note that the ΔS values shown in Fig. 6a and b represent mean daily totals for the CO2 associated with storage fluxes both into (positive) and out of (negative) the xylem sap. Despite the consistency of the ratio of ΔS : RS suggested by Fig. 6a, variation in the proportion of RS attributable to ΔS ranged between 5 and 10% and was positively correlated with the volumetric water content of wood, as shown in Fig. 6b.

Figure 6.

(a) The relationship between mean daily woody tissue respiration (RS) and mean daily xylem transport of CO2 (FT; closed circles) and mean daily amount of CO2 associated with both positive and negative storage (ΔS; open circles) fluxes in five Dacrydium cupressinum trees over 5 d. (b) The relationship between wood volumetric water content and the cumulative amount of CO2 associated with both positive and negative storage fluxes as a proportion of total woody tissue respiration (ΔS/RS) over the same time. The solid lines represent least squares regressions. (a) FT, y = 0.139x −0.070; r2 = 0.96; P < 0.01; ΔS, y =  0.122x − 0.088; r2 = 0.93; P < 0.01; (b) y = 16.14x + 0.49; r2 = 0.73; P = 0.05. For both (a) and (b), error bars represent ± standard error.

Radial and vertical trends in respiratory potential

The respiratory potential of extracted inner bark samples (containing phloem and cambial tissues) averaged 0.51 ± 0.03 (mean ± standard error) nmol O2 g−1 s−1 and was significantly greater than that of other wood types (P < 0.005; Fig. 7). Respiratory potential declined rapidly with depth into the stem to values of 0.14 ± 0.04, 0.09 ± 0.03, and 0.12 ± 0.03 nmol O2 g−1 s−1 for outer sapwood, inner sapwood, and the sapwood–heartwood boundary, respectively. No significant differences were observed among wood samples from these inner depths. However, a nonsignificant trend towards slightly increased respiration within the sapwood–heartwood boundary was observed. Volume-based measurements of respiratory potential showed the same trend, with inner bark, outer sapwood, inner sapwood, and sapwood–heartwood boundary samples exhibiting rates of 0.25 ± 0.02, 0.06 ± 0.01, 0.04 ± 0.01, and 0.08 ± 0.02 nmol O2 cm−3 s−1, respectively. Tissue nitrogen concentration was also significantly greater in the inner bark and declined with increasing depth towards the pith (P < 0.05; Fig. 7). However, differences in respiratory potential between wood samples within- and between-trees were only partially explained by tissue nitrogen concentration, as the overall relationship between respiratory potential and nitrogen content was weak (y = 0.99x − 0.27; r2 = 0.39).

Figure 7.

The observed trends in respiratory potential and wood nitrogen (N) concentration with increasing radial depth from the bark surface (open circles, mean respiratory potential; closed circles, mean N content; error bars represent ± standard error; n = 12). The four data points, moving from the bark surface towards the pith, for both respiratory potential and N content correspond to inner bark (IB), outer sapwood (OS), inner sapwood (IS), and the sapwood–heartwood boundary (SH). Respiratory potential and N content values for inner bark were statistically different from remaining wood tissue types at P < 0.05 according to nonparametric Wilcoxon signed rank tests, as indicated by the asterisk adjacent to inner bark data points.

The respiratory potential of inner bark samples was found to vary with stem position, as respiratory potential tended to be highest in the upper canopy (P = 0.05; Fig. 8). As a result of the large variation in respiratory potential between trees, values are presented relative to the maximum respiration potential observed for each individual tree. The respiratory potential of the extracted wood in the lower stem (for inner bark samples) was lower than that in the upper canopy, but was higher than the respiratory potential of wood from the lower or mid canopy. Less intratree variation was observed in the respiratory potential of outer sapwood, as all stem positions were statistically similar (P = 0.07; Fig. 8). This trend was partially explained by tissue nitrogen concentration, as this tended to be higher, although not significantly, in the tree canopy. The higher respiratory potential of wood from the upper canopy compared with the lower stem is consistent with the higher rates of CO2 efflux observed in canopy branches relative to that in lower stems.

Figure 8.

Within-tree variation in respiratory potential for inner bark (grey bars) and outer sapwood (open bars) sampled at four vertical stem positions. Respiratory potential (nmol O2 g−1 s−1) is represented as a percentage of the maximum respiratory potential observed within a given tree. Values shown are means (± standard error); n = 3. Different letters adjacent to bars indicate statistically different values at P < 0.05 according to paired t-tests.

Respiratory potential in both inner bark and outer sapwood increased in response to increasing measurement temperature. Respiration in upper canopy wood tended to be more responsive to temperature (i.e. Ao and Q10 values were higher), but this difference was not statistically significant (P = 0.06; Table 1). Furthermore, Ao and Q10 were similar across tissue types, at both tree heights, as values for inner bark and outer sapwood tissues were not statistically different (P = 0.84; Table 1).

Discussion

Effects of sap flow and storage on stem CO2 efflux

Studies of woody tissue respiration have long acknowledged the likelihood that the transport of respiratory CO2 produced within stems and branches in the xylem stream may influence CO2 efflux from stems and branches (Negisi, 1978; Sprugel & Benecke, 1991). This study indicates that the CO2 efflux from D. cupressinum stems is influenced by both sap flow and storage of CO2 in xylem water and that these interactions occur in trees that vary greatly in size and canopy dominance. In D. cupressinum, we estimate that xylem transport (FT) of CO2 accounts for 10.6% of woody tissue respiration (RT) over a 24-h period and 13–38% of predicted woody tissue respiration during periods of peak transpiration. This is in general agreement with the findings of McGuire & Teskey (2004) that FT accounts for 14–15% of RS, over a 24-h period, in Fagus grandifolia and Liquidambar styraciflua. Other studies have also reported that transpiration acts to transport respiratory CO2 to higher positions in the plant in both seedlings and mature trees (Negisi, 1978; Martin et al., 1994; Teskey & McGuire, 2002; McGuire & Teskey, 2004). Therefore, it is becoming evident that interactions between sap flow and stem CO2 efflux are commonplace in forest trees.

Our estimates of ΔS suggest that CO2 efflux from D. cupressinum stems is also influenced by transient changes in the concentration of dissolved CO2 in xylem water. Our results suggest that dissolved CO2 moved out of storage in the morning and early afternoon, whereas the amount of dissolved CO2 stored in xylem water increased in the late afternoon and evening. This trend is consistent with in situ measurements of the dissolved CO2 concentration of xylem water, in which rapid decreases in CO2 concentration were observed with the onset of sap flux in the morning and increases in CO2 concentration occurred as sap velocity began to decline in the late afternoon (Teskey & McGuire, 2002). During periods of maximum CO2 transfer into and out of xylem sap, these opposing fluxes averaged 15.8 and −12.1% of RS, respectively. The high percentages of total woody tissue respiration attributable to both xylem transport and storage of respiratory CO2 in xylem sap, at certain times of the day, result in uncertainty in the source of stem CO2 efflux as this CO2 may have been respired locally, in lower positions in the stem or roots, or in the soil.

As may have been expected, trees with high respiratory production of CO2 (RS) also tended to have high estimated rates of xylem transport and CO2 storage (Fig. 6a). However, between-tree variation in the proportion of RS comprised by ΔS was determined by the volumetric water content of the stemwood (Fig. 6b). This finding is consistent with the McGuire & Teskey (2004) method of measuring ΔS directly, in which changes in the dissolved CO2 concentration of xylem sap were multiplied by the volumetric water content of the stem segment. In Acer saccharum, sapwood water content was found to be higher in trees larger than 0.2 m in diameter than in smaller trees (Pausch et al., 2000) and there was a weak positive linear relationship (r2 = 0.18) between mass-based water content and tree size in D. cupressinum trees at this study site (Barbour & Whitehead, 2003). Therefore, it is likely that deviations from the expected rate of RS are likely to be greater in large trees because of their increased water content and greater capacity for dissolved CO2 storage.

Within-tree differences in the amount of the respiratory CO2 transported by sap flow (FT) were well, and positively, correlated with between-day differences in maximum sap flux density resulting from variation in air saturation deficit (Fig. 3). Between trees, the mean daily amount of FT was driven by the respiratory production of CO2 (RS) (Fig. 6a). However, the slopes of the linear relationships in Fig. 3 were not consistent across trees, indicating that the amount of respiratory CO2 transported by a unit volume of xylem sap was not constant across trees. The slope of these relationships was not significantly correlated with the sapwood density or sapwood depth of these trees. This finding indicates that the rate of diffusion of respiratory CO2 into xylem sap is likely to be dependent on many variables, such as factors that affect the capacity of xylem sap to take up respiratory CO2 (such as the dissolved CO2 concentration, pH, and temperature of xylem sap), anatomical characteristics of tracheids (such as cell wall thickness and diameter as this determines the surface area to volume ratio of the cell), and factors that influence the rate of CO2 diffusion to the atmosphere (such as the thickness and permeability of bark to CO2 diffusion). Clearly, further research on the factors limiting CO2 diffusion both into xylem sap and to the atmosphere will be necessary to understand between-tree variation in xylem transport of CO2.

Potential sources of error in modeled estimates of RS, FT, and ΔS

The methods used in this paper to estimate woody tissue respiration (RS), xylem transport of CO2 (FT), and storage of CO2S) from the measured rates of stem CO2 efflux (EA) at night require several assumptions, including (1) inline image and inline image do not exhibit diel variation; (2) FT and ΔS are negligible at night; (3) FT is solely dependent on sap flux density, and (4) ΔS is negligible during periods of maximum sap flux density. These assumptions could not be directly tested in this study and therefore serve as potential sources of error in our estimates of RS, FT, and ΔS. While quantification of the relative importance of these assumptions to the findings of this study was not feasible, it is possible to determine the conditions that would violate these assumptions and result in over- or underestimates of RS, FT, and ΔS in this study.

The first assumption, that inline image and inline image do not differ between day and night or over the study period, is required to use the night-time temperature response of EA to derive daytime estimates of RS. Although previous studies have demonstrated broad seasonal patterns in the temperature-normalized rate of woody tissue respiration (Ryan et al., 1997; Maier, 2001) and its temperature response (Lavigne et al., 1996; Stockfors & Linder, 1998), little is known about variations in these parameters over shorter periods of time. Acclimation of respiration to temperature, over periods of several days, has been demonstrated in both leaves (Atkin et al., 2000; Bolstad et al., 2003) and roots (Bryla et al., 1997; Covey-Crump, 2002), but it is unknown if large woody stems share this capacity to quickly adjust to prevailing temperature conditions. Respiratory acclimation to warmer temperatures during the day would likely result in a decrease in inline image and/or inline image and therefore would cause potentially large overestimates of daytime values of RS and RS − EA.

Our study also assumes that FT and ΔS are negligible at night, such that EA is equal to RS. McGuire & Teskey (2004) demonstrated that FT and ΔS were indeed small (cumulatively less than 10% of RS) at night in Fagus grandifolia and Liquidambar stryraciflua. However, in Platanus occidentalis, FT and ΔS comprised a fairly large proportion of RS at night, 15 and −26%, respectively. This indicates that our assumption of negligible FT and ΔS at night may be not be valid in all tree species. The validity of this assumption may also vary seasonally. Stem tissues pose significant barriers to the diffusion of CO2 from the sapwood to the atmosphere (Eklund & Lavigne, 1995). As a result, CO2 concentrations within tree stems remain high, relative to the atmosphere, during both day and night (3–9%, Teskey & McGuire, 2002) with concentrations returning to atmospheric levels only after the end of the growing season (Eklund, 1990). This suggests that ΔS may have a broad seasonal trend in which accumulation of CO2 in tree stems (+ΔS) occurs at the beginning of the growing season and the release of CO2 (–ΔS) to the atmosphere predominates towards the end of the growing season. Therefore, our assumption of negligible ΔS at night may tend to underestimate RS at the beginning of the growing season and overestimate RS at the end of the growing season.

We also assumed that the rate of FT in the measured D. cupressinum trees was solely dependent on sap flux density and that the CO2 storage flux (ΔS) was negligible during the 3–5-h peak in sap flux density. While FT is mostly dependent on sap flux density, variation in FT has been shown to be caused by changes in the dissolved CO2 concentration of the xylem sap even if sap flux density remains constant (McGuire & Teskey, 2004). Large increases or decreases in the xylem sap CO2 concentration could result in either over- or underestimates, respectively, in our estimates of FT. Similarly, large changes in xylem sap CO2 concentration during the midday peak in sap flux density would result in nonnegligible rates of ΔS. McGuire & Teskey (2004) found that the contribution of ΔS to RS at times of peak sap flux density was typically small (3–13%), but ranged between negligible values (1%) and up to 20% of RS. If the ΔS values during the period of maximum sap flux density in the D. cupressinum trees of this study were in this upper range, the estimates of α of our model and our assessment of FT would be less accurate.

Previous studies have found that stem CO2 efflux exhibits a lagged response to sapwood temperature (Ryan et al., 1995; Stockfors & Linder, 1998) and it has been proposed that heterogeneity of temperature within the stem may explain deviations from the relationship between CO2 efflux and current sapwood temperature, such as diel hysteresis. In D. cupressinum, the relationship between stem CO2 efflux and temperature was not improved by using stem temperatures lagged up to 6 h. However, it is possible that spatial heterogeneity of both temperature and sap flux density may contribute to variations in RS, FT, and ΔS with depth into the tree stem. Consideration of the radial variation in sap flux density may be particularly important for our study, as our sap flux sensors only measured the outer 2 or 4 cm of sapwood and may have therefore overestimated sap flux density, resulting in underestimates of our assessments of both α and FT. However, this error is likely to be small as both sap flux density (Phillips et al., 1996) and respiratory potential should decline with radial depth into trees.

Future research aimed at improving understanding of diel and seasonal variation in the dissolved CO2 of xylem sap and the potential for thermal acclimation of woody tissue respiration over short periods of time would greatly reduce potential sources of error in the model of this study. Furthermore, future investigations aimed at linking the CO2 diffusing from tree stems to the respiratory production of CO2 within stems, xylem CO2 transport, and internal storage of CO2 should consider incorporating spatial heterogeneity in temperature and sap flux density by measuring both stem temperature and sap flux density at several radial depths.

Effects of respiratory potential on stem CO2 efflux

CO2 efflux from woody stems has been found to be related to tree growth rate (Edwards & Hanson, 1996; Lavigne & Ryan, 1997; Maier, 2001) during the growing season and tissue nitrogen concentration (Maier, 2001; Vose & Ryan, 2002) and sapwood volume (Ryan et al., 1995; Maier et al., 1998) during the dormant season. These previous findings led to our initial hypotheses that CO2 efflux from stems should be determined by the local production of CO2 by growth and/or maintenance respiration within stems. In this study, between-tree differences in the local production of CO2 within D. cupressinum stems were also estimated by measuring the respiratory potential of excised wood tissues. However, no significant relationships were found between stem CO2 efflux, or estimated rates of RS, and tissue nitrogen concentration, the respiratory potentials of any specific wood tissues (i.e. inner bark, outer and inner sapwood, or sapwood–heartwood boundary) or the cumulative respiratory potential for the volume of wood under the CO2 sampling chambers. Furthermore, for respiratory potential in inner bark and outer sapwood tissues from the stems, the temperature response parameters, Ao and Q10, were significantly different (P < 0.05; Table 1) from these parameter values for stem CO2 efflux. The lack of agreement between respiratory potential and CO2 efflux further suggests that CO2 efflux from stem surfaces is not solely a function of the respiratory activity of the underlying wood. Possible factors that may contribute to this discrepancy between wood respiratory potential and stem CO2 efflux, in addition to xylem transport and storage of CO2, include between-tree differences in the dissolved CO2 concentration of xylem sap and differences in the permeability of bark to CO2 diffusion.

Vertical and radial trends in respiratory potential

While wood respiratory potential did not explain between-tree variation in CO2 efflux, the observed trends suggest that respiratory production of CO2 varies predictably within trees. For example, both respiratory potential and wood nitrogen concentration were highest near the bark surface, in inner bark tissues, then declined significantly with depth towards the pith. This trend of declining respiratory potential has been observed in other tree species (Møller & Müller, 1938; Goodwin & Goddard, 1940; Shain & Mackay, 1973; Pruyn et al., 2002a,b, 2003) and is likely to be attributable to higher proportions of live cells in inner bark tissues compared with sapwood (Stockfors & Linder, 1998).

Within-tree gradients in respiratory potential were also observed with stem position, as respiratory potential for both inner bark and outer sapwood samples was found to be highest in the upper canopy. Similarly, high wood respiratory potentials have been observed in the upper portions of P. ponderosa and P. menzeisii stems (Pruyn et al., 2002a, 2002b). Increased respiratory potentials in the canopy may be a result of the close proximity of carbohydrate sources, increased respiration associated with transporting carbohydrates to and from storage locations in the xylem parenchyma (Sprugel, 1990), and/or increased live cell volume. This trend towards high wood respiratory potential in the upper canopy is consistent with the high rates of CO2 efflux measured in canopy branches in D. cupressinum and in other studies (Sprugel, 1990; Damesin et al., 2002). However, significant interactions between sap flux and CO2 efflux found in this study indicate that the high rates of CO2 efflux from branches and upper stems were likely a result of both the high respiratory potential of underlying wood and the diffusion of respiratory CO2 transported in the transpiration stream to upper parts of the tree.

In contrast to respiratory potential, wood nitrogen concentration did not differ between stem positions. Furthermore, the overall variation in respiratory potential in excised core sections was described weakly by wood nitrogen concentration. The lack of strong agreement between respiratory potential and nitrogen concentration is inconsistent with previous reports of a strong relationship between maintenance respiration and tissue nitrogen concentration (Vose & Ryan, 2002) and may be attributable to low nitrogen concentration in D. cupressinum wood relative to that in other tree species (Carey et al., 1997; Maier, 2001), differences in stored carbohydrates within the wood cores, growth rate, and/or variation in tree age and size.

Effects of sapwood density on stem CO2 efflux

Sapwood density was found to be strongly correlated to the parameter Eo for CO2 efflux in D. cupressinum stems. In light of the effects of xylem transport and CO2 storage on CO2 efflux, it is interesting that sapwood density proved to be the strongest predictor of stem CO2 efflux, as this structural trait has been found to regulate trade-offs among various physiological characteristics that determine tree hydraulic characteristics. For instance, decreased wood density led to increased hydraulic conductivity (Stratton et al., 2000); however, this also increased occurrence of xylem cavitation (Hacke et al., 2001) and leaf turgor loss (Stratton et al., 2000).

Wood density may influence both xylem transport and storage CO2 fluxes in woody stems. For example, wood density been found to be negatively correlated with wood saturated water content (Meinzer, 2003). Furthermore, Barbour & Whitehead (2003) reported that sapwood density explained 94% of variation in average sap velocity in D. cupressinum trees with exposed crowns. Most of the D. cupressinum trees included in our study were in sheltered canopy positions. While this relationship may not be valid in the trees in our study, the findings of Barbour & Whitehead (2003) highlight the importance of sapwood density to stem hydraulic function and provide a possible linkage between sapwood density and xylem transport of CO2.

In addition, Roderick & Berry (2001) have predicted that, in coniferous trees, wood density will be negatively correlated with tracheid diameter. Variation in tracheid geometry associated with changes in wood density may potentially influence the rate of CO2 diffusion into and out of xylem sap. For example, trees with high wood density may possess smaller diameter tracheids that exhibit greater surface area relative to the volume of xylem water in their lumen. This increased proportion of tracheid surface area available for diffusion may facilitate exchange of CO2 between respiring tissues and xylem sap, resulting in increased xylem transport of CO2. In support of this theory, a weak correlation (y = 48.7x+ 3.8; r2 = 0.15) was found between the maximum proportion of RS attributable to FT and sapwood density. Further research on other tree species and growth environments is clearly necessary to determine the mechanism of this relationship between sapwood density and stem CO2 efflux, as well as the potential effects of sapwood density on xylem transport and CO2 storage.

Between-tree variation in stem CO2 efflux in D. cupressinum trees at Okarito Forest did not scale well with commonly used scalars such as stem surface area, sapwood volume, or nitrogen concentration. Similarly, other studies have reported difficulty in deriving scalars that work consistently across stands (Lavigne et al., 1996). This study demonstrates that xylem transport and storage of CO2 and the hydraulic characteristics of stems that determine these fluxes are equally important in predicting stem CO2 efflux. Sapwood density may be linked to xylem transport and CO2 storage, as well as tree growth rates and conditions. For example, variation in sapwood density between conspecific trees may arise from differences in tree age and genotype (Panshin & de Zeeuw, 1980), growth rate (Roderick, 2000), climate (D’Arrigo et al., 1992; Briffa et al., 2004), atmospheric CO2 concentration (Atwell et al., 2003; Kilpeläinen et al., 2003), and competition between neighboring trees (Larocque & Marshall, 1995). Based on the observed linkage between stem CO2 efflux and sapwood density and the potential for sapwood density to influence xylem transport and storage of CO2, we propose that sapwood density may be useful in predicting between-tree differences in CO2 efflux from woody tissues in conifers and that the efficacy of sapwood density as a scalar for CO2 efflux from woody stems and branches should be investigated in other tree species and forest types.

Acknowledgements

The authors appreciatively acknowledge technical assistance at the Okarito Forest site from J. Hunt, G. Rogers, and T. McSeveny. We also thank N. van Gestel for the processing of wood samples for nitrogen concentration. This work was funded by National Science Foundation Grant INT 02-05121 to KLG and DTT, an A.W. Mellon Foundation Grant to KLG, funding from the Foundation for Research, Science and Technology, New Zealand and a Landcare Research Investment Postdoctoral Fellowship to MMB. WPB also gratefully acknowledges graduate support from Columbia University's Center for Environmental Research and Conservation and the National Science Foundation's GK12 program. Timberlands West Coast and the Department of Conservation have provided access and logistical support at the Okarito Forest site. This paper was greatly improved by the thoughtful suggestions and comments of Robert Teskey, Mary Anne McGuire, and two anonymous reviewers.

Ancillary