Relationship between stem CO2 efflux, stem sap velocity and xylem CO2 concentration in young loblolly pine trees


Chris A. Maier. Fax: +1 919 549 4047; e-mail:


We measured diel patterns of stem surface CO2 efflux (Es, µmol m−2 s−1), sap velocity (vs, mm s−1) and xylem CO2 concetration ([CO2]) (Xs, %) in 8-year-old loblolly pine trees during the spring to determine how vs and Xs influence Es. All trees showed a strong diel hysteresis between Es and stem temperature, where at a given temperature, Es was lower during the day than at night. Diel variations in temperature-independent Es were correlated with vs (R2 = 0.54), such that at maximum vs, Es was reduced between 18 and 40%. However, this correlation may not represent a cause-and-effect relationship. In a subset of trees, vs was artificially reduced by progressively removing the tree canopy. Reducing vs to near zero had no effect on Es and did not change the diel hysteretic response to temperature. Diel Xs tended to decrease with vs and increase with Es, however, in defoliated trees, large increases in Xs, when vs ≈ 0, had no effect on Es. We conclude that at this time of the year, Es is driven primarily by respiration of cambium and phloem tissues and that sap flow and xylem transport of CO2 had no direct influence on Es.


Respiration of above ground woody tissues (stem and branch) comprises 15–25% of forest ecosystem respiration (Ryan et al. 1994, 1996; Xu et al. 2001; Maier et al. 2004). These estimates are based on empirical data where CO2 efflux from the stem (or branch) surface into a chamber is measured with an infrared gas analyzer. This approach assumes that CO2 generated from metabolism of cambium and xylem parenchyma tissues enclosed within the chamber diffuses radially from the stem interior across the cambial sheath to the surface. However, at high transpiration rates, a portion of the respired CO2 in sapwood may be carried upward by the transpiration stream instead of released horizontally through the bark, so that measured CO2 efflux underestimate the actual respiration of the sample section. A number of studies (Negisi 1975, 1978, 1982; Lavigne 1987; Kabubari 1988) found that on warm sunny days, measured stem CO2 efflux (Es) rates were much lower, 25–50%, compared with what would be expected based on temperature alone. Other studies found that the diel relationship between Es and temperature exhibits a hysteresis, where CO2 efflux measurements made at a similar temperature is higher in the late afternoon and evening, when transpiration is low, than in the morning, when transpiration is high (Martin, Teskey & Dougherty 1994; Ryan et al. 1995; Lavigne 1996; Stockfors 2000; Maier 2001; Bosc, De Grandcourt & Loustau 2003). These studies suggest that stem surface CO2 efflux in forest ecosystems may be linked to canopy water use.

Stem CO2 concentrations are high, ranging from 2 to 10% (Hari, Pekka & Korpilahti 1991; Eklund 1993; Teskey & McGuire 2002). Dissolved carbon in the xylem (CO2, H2CO3 and HCO3) is a combination of CO2 derived from respiration of nearby xylem parenchyma and cambium tissues, CO2 imported from respiratory activity of stem and roots lower in the xylem stream, and CO2 taken up in soil water. Given the high xylem CO2 concentration ([CO2]) (Xs), the aqueous transport of carbon in the xylem stream represents a potentially large and poorly understood carbon flux in forest ecosystems. If transport and storage of CO2 in the xylem strongly affects Es under normal field conditions, then the interpretation of stem gas exchange measured with chamber methods becomes equivocal (Martin et al. 1994; Teskey & McGuire 2002). A more complete evaluation of these relationships is needed to understand variation in Es rates.

Recently, there has been a renewed interest in measuring the origin and fate of carbon in the xylem sap and determining what effect this carbon flux may have on the measurement of Es (Stringer & Kimmerer 1993; Martin et al. 1994; Kaipiainen et al. 1998; Edwards & Wullschleger 2000; Clinton, Maier & Sullivan 2001; Teskey & McGuire 2002, 2005; McGuire & Teskey 2002, 2004; Bowman et al. 2005). Teskey & McGuire (2002) measured sap flow rate and Xs in large trees of several species (Quercus alba, Liriodendron tulipefera and Pinus taeda) and found that diel patterns were opposed, suggesting that transpiration may significantly affect stem [CO2] and thus the driving force for radial diffusion of CO2 in stem tissue. Teskey & McGuire (2005) further demonstrated in hardwood saplings, by artificially manipulating Xs, that stem surface CO2 efflux was directly related to internal [CO2]. McGuire & Teskey (2004) proposed a mass balance approach for estimating stem respiration that accounted for the rates of xylem CO2 inputs, outputs and storage and surface CO2 efflux. They found that the diel flux of respired CO2 within the stem followed different pathways dependent on sap flow rate. At night, when sap flow rates were low, stem surface CO2 efflux accounted for 74–93% of total stem respiration (i.e. the total from all sources); but during the day, when sap flow rates were high, surface CO2 efflux accounted for only 23–72% of estimated total stem respiration. Bowman et al. (2005) found similar results in several Dacrydium cupressinum trees. However, a consistent and measurable relationship between Es and stem sap flow is far from universal. Clinton et al. (2001) found a negative relationship between apparent stem respiration and sap velocity (vs) in large yellow poplar trees, while others found either no relationship (Carey, Delucia & Ball 1996; unpublished observations, Edwards & Wullschleger 2000) or a positive correlation (Levy et al. 1999). In the Levy et al. (1999) study, increases in apparent stem respiration with vs were attributed to transport of CO2 from the roots, which were assumed to be in equilibrium with high soil pCO2.

In this study, we examined the relationship between stem surface CO2 efflux rate (i.e. apparent stem respiration), vs and Xs in stems of 8-year-old loblolly pine trees over a 2 week period. Half of the experimental trees had received optimum nutrition from annual fertilization since planting, while the other half grew in the native nutrient-poor soil. Fertilization had significantly increased tree height, diameter and leaf area relative to non-fertilized controls. Fertilized trees are likely to have a different wood hydraulic architecture (Tyree & Ewers 1991), as well as differing patterns of water uptake (Ewers, Oren & Sperry 2000), canopy conductance, stand transpiration (Ewers et al. 2001) and rates of maintenance respiration (Maier et al. 1998). In some of the trees, we artificially altered vs through a stepwise reduction in canopy leaf area (Pataki, Oren & Phillips 1998). The objectives were to: (1) determine if there is a diel relationship between stem vs and stem surface CO2 efflux; (2) determine if there is a diel relationship between stem vs and Xs; and (3) determine if there is a relationship between Xs and stem surface CO2 efflux.


Site description

The study was conducted in an 8-year-old loblolly pine plantation located at the SETRES II GxE-QTL study site in Scotland County, NC, U.S.A (McKeand et al. 2000). The soil is a Wakulla series characterized as a sandy, siliceous, thermic Psammentic Hapludult (sand to > 43 m), which is very infertile, somewhat excessively drained, with a water holding capacity of 10–12 cm in a 2 m profile. The site receives an average annual precipitation of 1200 mm distributed evenly throughout the year. Annual temperature averaged 17 °C, with a seasonal average of 26 °C in summer and 9 °C in winter. Greenhouse-grown seedlings were planted in November 1993 after the removal of the existing 10-year-old loblolly pine. Five full-sib families of Atlantic coastal plain and Texas origin were planted in 100 tree plots. Our measurements were confined to a non-fertilized and fertilized plot of one Atlantic coastal plain family (9–1046). The site average leaf area index for the non-fertilized and fertilized plots were 1.19 and 2.91 m2 m−2, respectively, in October 2000 (Francisco Flores, North Carolina State University, personal communication).


We selected six trees in the non-fertilized and fertilized plots (12 trees total). Tree height, stem diameter and number of branches were measured (Table 1). Branch and canopy foliage biomass was estimated using site-specific regression equations (Tim Albaugh, North Carolina State University, personal communication). Average (± SE) stem diameter and estimated canopy biomass were 8.2 ± 0.5 cm and 1725 ± 704 kg in non-fertilized trees and 10.6 ± 0.2 cm and 2658 ± 129 kg in fertilized trees. Es, stem temperature and vs were monitored continuously for 11 d. On the last 3 d, Xs was measured in a subset of trees. Instantaneous photosynthesis and stomatal conductance (Licor 6400, Li-Cor, Inc., Lincoln, NE, USA) were measured at the beginning, middle and end of the experiments on upper canopy 1-year-old foliage on all trees during the morning hours (0900–1100 h).

Table 1.  Tree characteristics in April 2001 for the non-fertilized (NF) and fertilized (F) trees located at SETRES II, Scotland County, NC, USA
PlotTreeTreatmentd.b.h. (cm)Height (cm)Number of branchesFoliage biomass (g)
  1. The column labelled ‘Treatment’ refers to whether the canopies of trees were removed [cut (C)] or left intact [uncut (UC)].

  2. d.b.h., diameter at breast height.


Es (µmol CO2 m−2 s−1) measurements were made using an automated, multichamber sampling system (Butnor, Johnsen & Maier 2005) that consisted of stem chambers, an infrared gas analyzer (EGM-2, PP Systems, Amesbury, MA, USA) and a series of solenoids that sequentially measured stem chambers. The system had an open flow-through design where CO2 efflux was estimated as the difference between the CO2 concentration entering and exiting the chamber. Chambers were constructed of Teflon film that surrounded the tree stem 1 m above the ground. The Teflon film was fastened to the stem using collars of closed-cell foam and double-sided tape. Air was distributed to and sampled from the chamber using diffuser rings positioned at the top and bottom of the chamber. Chamber lengths were 25 cm and chamber volume ranged from 0.00179 to 0.00269 m3, depending on stem diameter. All chambers were leak tested prior to use. Airflow to the chambers was fixed at 0.00225 m3 min−1. Each chamber was measured for 6 min to assure stable CO2 measurements. The last minute of each cycle was retained for calculation of surface CO2 efflux rates. A complete cycle through all of the chambers, including a null chamber, was completed in 42 min, which equals approximately 34 observations for each chamber per day. All chambers were continuously flushed with ambient air (0.00225 m3 min−1) when chambers were not measured. Simultaneous measurements of chamber air and stem cambium temperature (3 mm) were made using copper/constantan thermocouples. Immediately after the experiments were completed, stem diameter at the top and bottom of the chambers were measured with digital calipers. Stem surface area inside the chamber was estimated from the average of the two diameter measurements and includes the bark.

Stem vs (mm s−1) was measured using custom-made 30-mm-long thermal dissipation sap velocity probes (Granier 1985, 1987). Briefly, paired probes were inserted radially into the tree such that the probes were approximately 5 cm apart vertically. For each tree, two probes were installed on opposite sides (north and south) of the stem just below the stem chamber. vs was measured every 10 s and these values were averaged every 15 min. In our trees, essentially all of the xylem was hydroactive; however, while we only measured the outer 3 cm, the probes measured the previous 2 years of growth and captured the majority of stem sap flow in these trees (Ewers & Oren 2000).

Xs (%) was measured in situ on four trees using CO2 microelectrodes (Model MI-720; Microelectrodes, Inc., Bedford, NH, USA). We followed methods described by McGuire & Teskey (2002) and Teskey & McGuire (2002). Briefly, electrodes were calibrated with humidified compressed CO2 gas at 2, 5 and 10% concentrations. Because the electrodes are temperature sensitive, a temperature correction was applied (McGuire & Teskey 2002). To measure Xs, a small hole 10 mm in diameter and 7–10 mm deep was drilled through the bark into the xylem, 20–25 cm below the stem chamber. The tip of a 5-cm-long low-density polyethylene tube was inserted into the hole, and the outside edge sealed to the tree with putty adhesive. A microelectrode was then inserted into the polyethylene tube such that the electrode tip did not make contact with xylem. Adhesive putty was used to seal the body of the microelectrode to the polyethylene tube. Four probes were installed, one each in four trees (two non-fertilized and two fertilized).


We examined the relationship between Es, vs and Xs using two different approaches. In the first experiment, we compared hourly average measurements of Es and vs for three trees in the non-fertilized (trees 1, 3 and 6) and fertilized (trees 4, 5 and 6) plots (Table 1). In young loblolly pine trees, the response of Es to diel changes in stem temperature typically exhibits a hysteresis, where at a similar temperature Es is higher at night than during the day (Maier 2001) (Fig. 1). We assumed a priori that this diel hysteresis was a function of vs Therefore, to remove potential effects of vs, only night-time (2300–0500 h) Es measurements were used to model Es. vs during this time was always less than 0.01 mm s−1. Night-time Es was modelled as a function of temperature by fitting the data to the exponential equation:

Figure 1.

An example of the diel pattern of hysteresis between measured stem surface CO2 efflux (Es) and cambium temperature (Tc), where at a given temperature, Es was lower during the day than at night. Response patterns from a non-fertilized tree are shown for day of year (DOY) 110 near the beginning of the experiment and DOY 117 after > 90% of the canopy had been removed. Corresponding measurements for three times are shown.

Es = β0 e(kTc),(1)

where Es is measured stem CO2 efflux (µmol m−2 s−1), β0 is CO2 efflux at 0 °C, k is the temperature coefficient and Tc is measured cambium temperature. Nonlinear regression (PROC NLIN, SAS Institute, Cary, NC, USA) was used to estimate β0 and k in Eqn 1. Model performance was examined graphically by comparing predicted and observed values of Es and by calculating the percent root mean square error (%RMSE), a measure of model precision, and the percent absolute deviation (%AD), an estimate of model accuracy (Maier 2001). A t-statistic was used to test for differences between non-fertilized and fertilized plots. Equation 1 was then used to predict diel patterns of Es (Ep). Residual Es (Er), the difference between Es and Ep, represents variation assumed to a result of xylem transport of CO2. We hypothesized that the diel pattern of Er and the ratio Er/Ep would be correlated with vs.

In the second experiment, we examined how artificially reducing vs, through a progressive removal of canopy leaf area, affected the diel patterns Es and Xs. In trees with low leaf area, transpiration is proportional to leaf area (Cienciala & Lindroth 1995; Sala, Smith & Devitt 1996), and abrupt reductions in leaf area can reduce canopy transpiration and vs (Oren et al. 1999). In this experiment, the canopies of three of the six trees in the non-fertilized (trees 2, 4 and 5) and fertilized (trees 1, 2 and 3) plots (Table 1) were removed in thirds (Cut treatment). The canopy of the cut tree was divided vertically into three levels based on an equal number of branches. Canopy removal was done equally from each level based on branch foliage biomass (Table 1). After an initial period, to establish individual tree Es and vs behaviour (≈ 48 h), one-third of the canopy leaf area in the cut trees was removed by removing branches. Measurements continued for 3 d, and then another third of foliage biomass was removed followed by another 4 d of measurements after which the remaining foliage was removed except for a single 1-year-old branch at the top of the canopy. Branch removal was completed by 1000 h on the day of treatment.

Diel patterns of vs and Es were compared graphically between uncut and cut trees. To facilitate comparisons, maximum daily rates of vs and Es data were normalized to the maximum rates measured on DOY (day of year) 109 before the cutting treatments began. Changes in the normalized maximum rates of vs and Es resulting from the cutting treatment were compared using repeated measures analysis of variance (anova) (PROC MIXED, SAS Institute, Cary, NC, USA), utilizing an autoregressive covariance structure. We hypothesized that during the daylight hours maximum vs would decrease and maximum Es would increase in cut trees relative to uncut trees. On the last 2 d of the experiment, we compared diel patterns of vs, Es and Xs. We hypothesized that Xs would increase in cut trees relative to uncut trees because respired CO2 in the xylem would accumulate when vs is low in cut trees, and that increases in Xs in cut trees would cause a concomitant increase in Es.


Diel patterns

Stem cambium temperatures ranged from 0–30 °C over the 11 d of measurements (Fig. 2a). Stem temperatures peaked in early afternoon on most days. Es had a strong diel pattern that was well correlated with cambium temperature (Fig. 2b), however, maximum daily Es always occurred after maximum stem temperature (35–210 min) (Fig. 2a & b) creating a diel hysteresis. For example on DOY 110, Es was lower during the morning and early afternoon than at night when measured at a similar temperature (Fig. 1). The magnitude of the daily hysteretic response varied among trees and within a tree over time.

Figure 2.

An example of the diel response patterns of (a) cambium temperature (Tc), (b) measured (Es), predicted (Ep) and residual (Er) stem surface CO2 efflux, and (c) sap velocity (vs) for a fertilized uncut tree. Residual respiration is the difference between Es and Ep. Predicted stem surface CO2 efflux (Ep) was estimated using temperature response curves developed from nighttime Es measurements when vs < 0.01 mm s−1. Inset: Diel patterns of the parameters for 1 d. DOY, day of year.

Stem vs increased rapidly during the morning and reached maximum values between 1000 and 1200 h then rapidly declined in the afternoon (Fig. 2c and inset). An exception to this pattern was observed on DOY 115, which was rainy and cool, when maximum vs was measured in late afternoon. During the night, vs was less than 0.01 mm s−1. Peak sap velocity generally occurred 5–8 h before maximum Es.

Experiment 1

To remove potential effects of vs on Es, temperature response curves were developed using only nighttime measurements when vs was less than 0.01 mm s−1. Night-time Es was well correlated (R2 = 0.90–0.96, Eqn 1) with stem cambium temperature (Table 2). Predictions based on the equation showed good agreement with observed values. The %RMSE averaged 12.3% of mean night-time Es. All of the models exhibited a mean %AD of less than 14% and indicates that the models accurately predicted night-time Es over the time period measured. There was no significant difference between non-fertilized and fertilized trees in basal CO2 efflux rate (β0; non-fertilized: 1.22 ± 0.08SE; fertilized: 1.37 ± 0.04; P = 0.12) or the temperature coefficient (k; non-fertilized: 0.073 ± 0.001; fertilized: 0.071 ± 0.001; P = 0.77).

Table 2.  Parameter estimates and fit statistics for Equation 1
  • Equation 1 was fitted to stem surface CO2 efflux (Es) measured at night, between 2300 and 0500 h, when sap velocity (vs) was less than 0.01 mm s−1.

  • a

    n is the number of observations.

  • b

    Percent root mean square error. inline image

  • c

    Percent absolute deviation. inline image

Non-fertilized11.530.0750.9420912.6 9.5
31.060.0700.9620612.2 8.9
61.180.0700.96206 9.9 7.6

Using the parameters in Table 2 with Eqn 1, we compared predicted Ep with observed Es (Fig. 2b and inset). During the day, Ep was always greater than measured Es, with the largest differences occurring around midday during periods of rapid change in stem temperature and vs. Er, the difference between Ep and Es, generally decreased with increasing vs (Fig. 2b and inset). The negative ratio of Er/Ep, a relative measure of reduced Es, was negatively correlated with vs (Fig. 3) and there was no difference between non-fertilized and fertilized trees in the slope of this relationship. These data suggest that during the day high vs could potentially reduce Es up to 40% of that predicted on temperature alone.

Figure 3.

The relationship between the negative ratio of residual and predicted stem surface CO2 efflux (Er/Ep) and sap velocity (vs). Predicted stem surface CO2 efflux (Ep) was estimated using temperature response curves developed from nighttime Es measurements when vs < 0.01 mm s−1. Residual respiration is the difference between Es and Ep. Er/Ep is expressed as a negative ratio to illustrate the potential reduction in Es at high vs Each point is the average of three trees. y = −0.026 − 4.526x R2 = 0.54.

Experiment 2

In this experiment, we examined how artificially changing vs through a progressive removal of canopy leaf area affected Es. Foliage was removed in thirds on DOY 110, 113 and 117 (arrows, Fig. 4). To aid in making comparisons between cut and uncut trees, vs and Es were normalized to the maximum rates measured on DOY 109 before the branch removal treatment began (Fig. 5). There were no significant differences in vs or Es between non-fertilized and fertilized trees (Table 3). Removal of one-third to two-thirds of the canopy leaf area had only small effects on vs and Es; and non-fertilized and fertilized trees behaved differently (Figs 4 & 5, Table 3). In the non-fertilized trees, maximum daily vs was reduced ≈ 20% following the first cutting (DOY 110) when roughly one-third of the canopy was removed (Fig. 5). Removal of the second third of the canopy (DOY 113) had no further effect on the magnitude of this response. In fertilized trees, the first cutting treatment had no effect on vs, but vs was significantly less in the cut trees 3 d following the second cutting treatment (Fig. 5). After removal of most of the canopy (DOY 117), vs was reduced 80–90% of that in uncut trees in both non-fertilized and fertilized trees. The lack of a large decrease in vs following abrupt changes in leaf area was likely due to stomatal compensation. Stomatal conductance significantly increased in foliage of cut trees after the second cutting treatment (Table 4). Net photosynthesis tended to increase in cut trees following branch removal, but this difference was only significant for DOY 114.

Figure 4.

Diel patterns of hourly sap velocity (vs) and stem surface CO2 efflux (Es) for non-fertilized and fertilized trees with intact canopies (uncut) and those where the canopy was progressively removed (cut) in thirds (arrows) over the course of the study. Each point is the average of three trees. DOY, day of year.

Figure 5.

Least squares mean (± SE) of maximum daily sap velocity (vs) and stem surface CO2 efflux (Es) for non-fertilized and fertilized trees. Data are normalized to the maximum values measured on day of year (DOY) 109. Comparisons are made between trees with intact canopies (uncut) and those where the canopy was progressively removed (cut) in thirds over the course of the study. Each point is the average of three trees. An asterisk denotes a significant difference between uncut and cut means at α = 0.05.

Table 3.  Probability values for the effect of fertilization, cutting treatment and time and their interactions on the normalized maximum daily sap velocity (vs) and stem surface CO2 efflux (Es)
Fertilization (F)0.39640.4608
Cut (C)<0.00010.0914
F × C0.25560.1258
Day (D)<0.0001<0.0001
F × D0.2127<0.0001
C × D<0.00010.1538
F × C × D0.17090.4096
Table 4.  Stomatal conductance (gL) and net photosynthesis (Pn) of upper canopy foliage in non-fertilized (NF) and fertilized (F) trees with intact canopies [uncut (UC)] and those where the canopies were progressively removed [cut (C)] over the course of the study
  1. Measurements on day of year (DOY) 108 represent pre-cutting values whereas values on DOY 114 and 117 are after two-thirds of the canopy was removed in cut trees. Values are the mean of three trees. Numbers within parentheses are the SEs of the mean.


In contrast, cutting treatment had little effect on the diel patterns of Es (Fig. 4). There was no significant fertilizer or fertilizer-by-cutting treatment interaction on normalized Es (Table 3). However, in the fertilized trees Es declined 7–15% two days following the first cutting treatment (Fig. 5). This difference was maintained throughout the experiment. There was no apparent response of Es to large changes in vs The pattern of Es in fertilized and non-fertilized trees was similar throughout the experiment even after the final removal of branches when vs in cut trees was ≈ 10% of that in uncut trees (Figs 4 & 5). Furthermore, large reductions in vs following the final cutting treatment had little effect on the diel Es-temperature hysteresis. For example, the magnitude of the hysteresis was similar on days having a similar range in temperature, but a large difference in maximum vs (compare DOY 110 and 117, Fig. 1).

We measured Xs in four trees over the last 2.5 d of the experiment. During this time, Xs ranged from 1 to 8%. Xs changed diurnally reaching a maximum at night and a minimum near noon (Fig. 6). In uncut trees, Xs generally decreased during the day when vs was high and increased at night when vs was low suggesting that sap flow had a strong influence over Xs. A similar pattern was observed in cut trees; however, Xs increased more relative to uncut trees when vs was reduced following the final cutting treatment. Large diel changes in Xs appeared to have little effect on Es. For example in the fertilized trees, Xs in cut trees increased twofold after trees received the last canopy removal (DOY 117), but there was no apparent change in Es. In the non-fertilized trees, Es was similar between uncut and cut trees on day 118 despite a fourfold difference in Xs. These data suggest that Es in these trees was not influenced by CO2 transported in the xylem stream.

Figure 6.

Comparison of stem surface CO2 efflux (Es), cambium temperature (Tc), sap velocity (vs) and xylem CO2 concentration (Xs) for a tree with an intact canopy (uncut) and one with the canopy removed (cut) in the (a) non-fertilized and (b) fertilized plots. Measurements are for the last 2 d of the study. The final cutting treatment was completed by 1000 h on day of year (DOY) 117.


Our trees showed the typical diel counterclockwise hysteresis between Es and stem temperature (Fig. 1) reported for trees in other studies (Ryan et al. 1995; Lavigne 1996; Stockfors 2000; Maier 2001; Damesin et al. 2002; Bosc et al. 2003). We used two different approaches to determine whether or not xylem transport and storage of CO2 was responsible for this hysteresis. These two approaches produced seemingly contradictory results. The first experiment assumed a priori that the diel hysteresis was a function xylem CO2 transport. Therefore, night-time Es temperature response functions, when vs ≈ 0, were used to predict daytime rates (Ep). Predicted daytime rates were always greater than observed. The difference between observed and predicted stem CO2 efflux (Er = Ep − Es) was correlated with vs and suggest that during the day high vs could reduce Es by up to 40% below that measured when vs is low. These data support the idea that midday suppression of stem respiration measured in other studies (Negisi 1975, 1978, 1982; Lavigne 1987; Kabubari 1988) is a function of transpiration rate. Negisi (1979) artificially alter vs in detached Pinus densiflora stems and found that Es was reduced by 70% at a vs rate of 0.15 mm s−1, a value that corresponds well with our data (Fig. 3). Levy et al. (1999) also found a correspondence between Er and vs; however, in their study, residual efflux was positively correlated with vs. They concluded that imported CO2 in the xylem stream contributed up to 12% of Es.

The second experiment examined how artificially reducing vs by eliminating canopy leaf area affected Es. Our attempt to create varying levels of stem vs by a stepwise defoliation of the canopy was only partially successful. Increased stomatal conductance largely compensated for a partial reduction (≈ 33 and 63%) in canopy leaf area maintaining vs near pre-treatment rates as observed in Pataki et al. (1998). Only after removal of most of the canopy (> 90%) was vs substantially reduced. Even though we were unable to create a progressive reduction in vs, it was evident that reducing vs to near zero by removing almost all canopy foliage had little effect on Es. Diel patterns of Es and the daily maximum values were similar between cut and uncut trees throughout the experiment indicating that xylem CO2 transport in the sap had little effect on Es in these trees. In addition, if vs strongly affected Es then the magnitude of the diel hysteresis between Es and Tc should be smaller or eliminated in cut trees as vs approached zero. However, in our trees neither the pattern nor magnitude of the diel hysteresis was affected by large changes in vs (Fig. 1). These data suggest that vs and Es are uncoupled in these trees and the a priori assumption that the diel hysteresis is a function of xylem CO2 transport and storage is incorrect. Thus, the correlation between Er/Ep and vs (Fig. 3) does not represent a causative response.

The apparent uncoupling of vs (or Xs) and Es suggest that that the radial diffusion of CO2 from the xylem to the stem surface is restricted in these trees. Conifers have few intercellular spaces and radial gas diffusion must occur in the liquid phase which is several magnitudes lower than gas phase diffusion (Hari et al. 1991). The large difference between Xs and the ambient air in our trees indicates a high resistance to CO2 diffusion from the xylem through the bark (Eklund 1990, 1993; Hari et al. 1991). Eklund (1990) and Eklund & Lavigne (1996) found little diffusion of O2 or argon gas from the atmosphere to the xylem or from the xylem to the atmosphere in conifer stems. However, despite a high resistance to gas diffusion through the bark, several studies have shown that artificially manipulating of sap flow (Negisi 1979) and/or Xs (Teskey & McGuire 2002, 2004) clearly influences Es. So, why did large changes in vs and Xs in our cut trees have no effect on Es? We conducted the experiment in the spring when stem respiration and growth were at a maximum for these stands (Maier 2001). At this time of year, the thin cambium and phloem meristems likely respire at a much higher rate than the xylem parenchyma and thus would be a major source of respiratory CO2 in the stem. Goodwin & Goddard (1940) measured oxygen consumption in black ash stems and found that O2 uptake was several magnitudes higher in the cambium and phloem compared to the xylem. Similarly, Pruyn, Gartner & Harmon (2002, 2003) showed that the respiratory potential of the inner bark of ponderosa and white pine trees was 3–15 times greater than that of the sapwood. The surface CO2 efflux we measured was likely a result of growth related respiration associated with differentiating cells in the cambium and from energy expended in phloem transport. Thus, during periods of rapid stem growth, the CO2 concentration in the stems would be much higher in the cambium and phloem regions than in the xylem. Under these conditions, the CO2 concentration gradient (i.e. diffusion gradient) would likely decrease from the cambium layer to the xylem effectively isolating the xylem tissue as a source of CO2 evolved from the stem surface. Although our data is limited, measurements of Xs appear to support this hypothesis. We found in our cut trees that while Xs increased after defoliation (vs < 0.01 mm s−1), it was not followed by increases in Es which would be expected if the CO2 concentration gradient decreased radially from the xylem to the stem surface. In addition, the rate and magnitude of Es was similar on consecutive days with large differences in Xs (Fig. 6) suggesting that Xs had no effect on Es. It is interesting to note that atmospheric O2 in the xylem of conifer stems decreases from near ambient to less than 5% of ambient during the growing season (Eklund 1993, 2000); apparently, the active cambial tissue consumes most of the O2 that diffuses into the stem (Hook et al. 1972).

Our measurements of Es probably reflect respiration associated with diameter growth and phloem transport; however, it underestimates this component of stem respiration as some of the respired CO2 is expended into the xylem (Eklund & Lavigne 1996; McGuire & Teskey 2004). This is illustrated by the increase in Xs at night when vs ≈ 0 and is a result of the metabolism of local xylem parenchyma and from cambial tissue internal to the xylem. This CO2 is stored in the stem segment and is later transported up the stem as part of the internal CO2 flux. McGuire & Teskey (2004) used a mass balance approach to estimate total stem respiration from stem segments of several hardwood species. They found that CO2 respired within the stems either diffused radially through the bark to the atmosphere, is transported upward in the xylem stream, or is temporarily stored in the xylem. For example, surface flux, xylem transport flux and storage account for 85, 15 and 8%, respectively, of stem respiration over a 24 h period in a beech tree. The relative proportion of each component varied during the day depending on sap flow rates. In their analysis, dissolved CO2 in the xylem served as a sink or a source for CO2 diffusion to the atmosphere. Our data suggests that internal stem CO2 dynamics are more complex. It appears that there may be conditions, perhaps during periods of high cambial activity, when Es is uncoupled from internal xylem CO2 fluxes. The relationship between vs, Es and Xs may be different at other times of the year. For example when growth ceases, Xs may exceed that in the cambium and phloem regions and be a source of CO2 to Es. Root absorption of dissolved CO2 in the soil can potentially contribute large amounts of CO2 to the xylem (Levy et al. 1999). However, because of the porous nature of the sandy soils at our site, soil pCO2 rarely exceeds 5000 µmol m−1 (0.5%) in the top 50 cm and generally is much lower during the spring (≈ 1000–1100 µmol m−1, Maier, unpublished results) hence, soil CO2 would not likely contribute much CO2 to the xylem stream in our trees.

There are other possible explanations for the hysteresis between Es and Tc: (1) a lag between temperature and surface CO2 efflux because of high resistance to diffusion (Hari et al. 1991; Eklund & Lavigne 1996; Stockfors 2000); (2) refixation of respired CO2 during corticular photosynthesis (Sprugel & Benecke 1991; Cernusak & Marshall 2000); (3) diel differences in substrate supply (Edwards & McLaughlin 1978; Martin et al. 1994); and (4) diel patterns of stem growth. Because of the high resistance to CO2 diffusion, Tc measured at the time of flux measurements may not reflect surface CO2 evolution. Modelling Es using lagged temperatures, measured sometime earlier, can account for a substantial portion of the hysteresis (Ryan et al. 1995; Lavigne 1996; Stockfors 2000; Maier 2001; Bosc et al. 2003). We modelled Es (Eqn 1) for our trees using all of the data (day and night) and found that, Es was best correlated (R2 = 0.91–0.95) with Tc measured between 42 and 168 min earlier. We note, that residuals from these regressions were not correlated with vs (R2 = 0.23, P = 0.45). If a lag in CO2 production were a major factor responsible for the hysteresis, then the magnitude of the hysteresis should be a function of tree size (Bosc et al. 2003) or bark thickness (Ryan et al. 1995). Although the range in tree size was small in our study, we found no relationship between lag times and tree diameter or bark thickness.

Bark photosynthesis can refix a substantial amount of respired CO2 in woody tissue (Sprugel & Benecke 1991) thus lowering the apparent stem respiration rate during the day. Cernusak & Marshall (2000) estimated that bark refixation rates were 70% of night-time respiration rates in western white pine. However, refixation is not an issue in our study because we used the main stem that had little or no chlorophyll present and we used opaque chambers so there would be no bark photosynthesis at least for the tissue inside the chamber.

Substrate supply can affect respiration rates (Amthor 1989). Edwards & McLaughlin (1978) found that the diel pattern of Es in yellow poplar trees was correlated with the concentration of reducing sugars in the phloem, indicating that the diel pattern of Es may in part be driven by transported photosynthate. This response may occur quickly once carbohydrate supply is compromised. Edwards, Tschaplinski & Norby (2002) showed that Es in sweetgum trees increased in response to elevated CO2 but decreased to rates measured at ambient CO2 within several days after the elevated CO2 treatment was turned off. Removal of photosynthetic surface area would likely affect canopy assimilation and reduce substrate supply to the stem. We could not assess the impact of the cutting treatment on stem carbohydrate supply; however, the cutting treatment slightly reduced Es in the fertilized trees. Effects of reduced carbohydrate supply on Es could potentially mask effects of reduced vs. Martin et al. (1994) found in loblolly pine seedlings that Es declined after girdling the phloem above the respiration chamber; although, they concluded that the response was too slow to account for diel hysteresis.

The hysteresis between Es and Tc may reflect diel patterns of growth. Daudet et al. (2005) measured diel Es and stem diameter in potted hybrid walnut (Juglans nigra × Juglans regia) saplings under constant temperature conditions. Diel patterns of Es were highly correlated with changes in stem diameter. Maximum Es occurred at night, suggesting that more energy was being expended in growth and maintenance processes at this time.


Our trees showed a strong diel hysteresis between Es and stem temperature. The diel variation in temperature-independent Es was correlated with vs, such that at high vs, Es could be reduced by up to 40%. However, this correlation may not represent a cause-and-effect relationship. Artificially reducing vs through a progressive defoliation of the canopy had little effect on Es and had no effect on the magnitude of the diel hysteresis. These data indicate that Es is uncoupled from vs in these trees. We suggest that high metabolic activity in the cambium during this time of year (spring) is likely a source of CO2 to the xylem and thus, CO2 transported in the xylem stream would not contribute to Es. This hypothesis is supported by the observation that diel changes in Xs correspond with Es but the large increase in Xs measured in the cut trees, when vs ≈ 0, had no affect on Es. Increased resolution of measurements of stem [CO2] in cambium and xylem regions is needed to confirm this. Understanding diel and seasonal variation in surface CO2 efflux and the relationship to vs (or sap flow) and Xs will provide a more complete characterization of stem respiration and whole-plant carbon cycles.


This work was supported in part through a Challenge Cost Share Grant from the USDA Southern Research Station. We would like to thank Pete Anderson and Daniel McInnis for technical assistance in the field, Karen Sarsony for laboratory analysis and Chelsea R. Maier for help with branch harvest and processing. We also would like to thank Dr Steve McKeand, North Carolina State University for permission to use the SETRES II site. We gratefully acknowledge critical pre-submission reviews by Mary Anne McGuire, Chelsy Ford, Mike Ryan and Robert Teskey.