Root-derived CO2 efflux via xylem stream rivals soil CO2 efflux


Author for correspondence:
Doug P. Aubrey
Tel: +1 803 439 4886


  • • Respiration consumes a large portion of annual gross primary productivity in forest ecosystems and is dominated by belowground metabolism. Here, we present evidence of a previously unaccounted for internal CO2 flux of large magnitude from tree roots through stems. If this pattern is shown to persist over time and in other forests, it suggests that belowground respiration has been grossly underestimated.
  • • Using an experimental Populus deltoides plantation as a model system, we tested the hypothesis that a substantial portion of the CO2 released from belowground autotrophic respiration remains within tree root systems and is transported aboveground through the xylem stream rather than diffusing into the soil atmosphere.
  • • On a daily basis, the amount of CO2 that moved upward from the root system into the stem via the xylem stream (0.26 mol CO2 m−2 d−1) rivalled that which diffused from the soil surface to the atmosphere (0.27 mol CO2 m−2 d−1). We estimated that twice the amount of CO2 derived from belowground autotrophic respiration entered the xylem stream as diffused into the soil environment.
  • • Our observations indicate that belowground autotrophic respiration consumes substantially more carbohydrates than previously recognized and challenge the paradigm that all root-respired CO2 diffuses into the soil atmosphere.


Forest ecosystems account for the majority of terrestrial net primary productivity and are therefore a major focus of global carbon budgets (Jobbágy & Jackson, 2000; Geider et al., 2001). Ecosystem respiration (RE) consumes c. 77–85% of annual gross primary productivity in forest ecosystems across the globe (Law et al., 2002; Luyssaert et al., 2007; Baldocchi, 2008). Comprised of both autotrophic (leaves, stems and roots) and heterotrophic (fungi, bacteria and animals) components, RE releases one of the largest annual CO2 fluxes of the global carbon cycle – a flux nearly 16 times that of the annual fossil fuel combustion from 2000 to 2005 (Prentice et al., 2001; IPCC, 2007). Belowground autotrophic and heterotrophic contributions to RE are difficult to separate. However, the combined contribution of autotrophic and heterotrophic respiration (i.e. soil CO2 efflux) represents a substantial portion of forest RE– accounting for 30–88% of total annual RE (Goulden et al., 1996; Law et al., 1999; Janssens et al., 2001; Davidson et al., 2006; Baldocchi, 2008; Cavaleri et al., 2008; Tang et al., 2008). The remainder of RE is contributed by aboveground foliar and woody tissue respiration. A mechanistic understanding of forest respiratory flux pathways is imperative to understanding forest carbon cycles and forest ecosystem responses to climate change.

A long-standing paradigm in plant physiological ecology is that root-respired CO2 diffuses from inside the root outward and is released into the soil atmosphere (Hanson et al., 2000; Kuzyakov, 2006; Trumbore, 2006). A corollary is that all CO2 derived from root respiration can be measured as part of soil CO2 efflux to the atmosphere. However, CO2 dissolved in the soil solution can be transported from roots to shoots via the xylem stream (Ford et al., 2007; Moore et al., 2008). Tree stems contain CO2 concentrations many times greater than those of the atmosphere (Teskey et al., 2008) and high CO2 concentrations at the base of tree stems indicate that much of the CO2 in stem xylem originates belowground (Teskey & McGuire, 2007). The estimated contribution of dissolved inorganic carbon absorbed by roots from the soil solution is far less than the quantity found at the base of tree stems and suggests that much of it originates within the root system (Teskey & McGuire, 2007). Thus, it is likely that not all root-respired CO2 diffuses into the soil atmosphere, indicating that conventional methods for measuring belowground respiration may underestimate actual rates.

Given these observations, it is plausible that a considerable amount of respired CO2 can remain within tree root systems, where it dissolves in xylem sap and is subsequently transported aboveground via the xylem stream. However, the relative importance of internally transported CO2 remains unclear as no empirical investigations have quantified the magnitude of this flux or compared it with flux of CO2 from the soil surface to the atmosphere. Using an experimental Populus deltoides plantation as a model system, we tested the hypothesis that a substantial portion of belowground autotrophic respiration remains within tree root systems and is transported aboveground via the xylem stream rather than diffusing outward into the soil atmosphere.

Materials and Methods

To determine the magnitude of autotrophically respired and soil-derived CO2 transported from root systems to stems (FT), we instrumented Populus deltoides Bartr. trees to measure the internal xylem CO2 concentration ([CO2]) and sap flux and simultaneously measured soil CO2 efflux near each tree to elucidate the relative importance of FT for total belowground CO2 efflux. The difference between FT and the amount of dissolved CO2 originating in the soil solution is the internal flux of autotrophically respired CO2, hereafter referred to as root-respired CO2. We define root-respired CO2 as a fusion of CO2 produced from root and root-associated fungal symbiont sources.

Study site

The experiment was conducted in an intensively managed 9-yr-old P. deltoides experimental plantation located within the Department of Energy Savannah River Site in Aiken County, SC, USA. The soil at the site is predominately a Blanton sand with a loamy subsoil at a depth of 120–200 cm (Rogers, 1990). The litter layer within the measurement plots was thin and heterogeneously distributed. Complete competition control was maintained over the entire 9-yr stand history, so all autotrophic contributions to belowground CO2 efflux resulted from P. deltoides trees. We randomly selected four trees that ranged in diameter from 15.7 to 16.5 cm at 1.3 m stem height. We compared the magnitudes of belowground CO2 flux pathways by scaling both FT and soil CO2 efflux to the same unit area. The explicit spacing (2.5 × 3 m) and consistent stocking of trees allowed for comparisons of FT and soil CO2 efflux at the same spatial scale. We assumed that the soil area occupied by the root system of each tree was equivalent to 7.5 m2. Soil CO2 efflux was expressed on a m2 area basis, so FT was divided by 7.5 m2 to place both soil and stem fluxes on the same unit area scale. Measurements of FT and soil CO2 efflux were collected every 15 min over 7 consecutive days in August 2008. To illustrate the importance of FT for our understanding of belowground processes, we estimated the total autotrophic and heterotrophic contributions of total belowground CO2 flux by assuming that 50% of soil CO2 efflux was contributed by autotrophic respiration. Hanson et al. (2000) report a range of 10–90% for the autotrophic component of soil CO2 efflux for instantaneous measurements and 45–60% for annual budgets in forest ecosystems. Based on this information, we chose 50% as a reasonable assumption of growing season autotrophic contribution at our site. Although the actual proportion on this site is not known, this approximation provides an initial insight into the potential impact of FT on the primary components of belowground CO2 efflux.

Soil CO2 efflux

Soil CO2 efflux was measured using the [CO2] gradient method which relies on Fick's first law of diffusion to calculate efflux based on the [CO2] gradient in the soil and soil diffusivity properties (Tang et al., 2003). The gradient approach allowed high-frequency measurements of soil CO2 efflux which our study required. Soil [CO2] was measured in soil profile arrays with nondispersive infrared (NDIR) CO2 sensors (model GMT220; Vaisala Inc., Helsinki, Finland) housed in PVC sleeves that were inserted to the appropriate depth (2, 22 and 42 cm) 0.75 m northwest of each tree. Sensors were sealed into the housings with rubber O-rings. Thermocouples were placed at each depth to correct soil [CO2] for temperature. Soil [CO2] was also corrected for barometric pressure (model PTB110; Vaisala). Soil bulk density was determined before the study. Volumetric water content was measured using dielectric sensors (10HS soil moisture sensor; Decagon Devices, Pullman, WA, USA). We validated the gradient method at our site by regressing gradient measurements with chamber-based measurements (LI-6400; Li-Cor Biosciences, Lincoln, NE, USA) obtained from soil collars at each profile array (m = 0.91, r2 = 0.89, n = 40). Measurements for this comparison were collected in August 2008. To determine potential spatial heterogeneity of soil efflux and to ensure that our profile array locations were adequate representations of the larger growing space for each tree, we inserted soil collars at four additional locations around each tree and found that there was no statistical difference between our array location (4.6 ± 0.7 µmol CO2 m−2 s−1; mean ± SE) and the mean of the four additional locations (4.7 ± 0.7 µmol CO2 m−2 s−1; P = 0.4892, n = 9). Measurements for this comparison were collected in July 2008.

Flux of root-derived CO2 through xylem (FT)

We calculated FT as the product of sap flux and dissolved CO2 concentration in the xylem. Sap flux was measured using constant heat Granier-type thermal dissipation probes with thermocouples at 15 and 70 mm depth with 80 mm vertical separation (TDP-80; Dynamax Inc., Houston, TX, USA). Zero flow was calculated daily as the mean difference between vertical thermocouple pairs between 03:00 and 05:00 h. We developed and applied our own calibration parameters to the sap flux values. Xylem [CO2] was measured in gaseous phase by inserting an NDIR sensor and thermocouple into xylem tissue at 15 cm above ground level. Gaseous [CO2] was corrected for temperature and barometric pressure. Concentrations of dissolved CO2 in xylem were determined by measuring the temperature and pH of xylem sap and applying Henry's Law (McGuire & Teskey, 2002). Xylem sap was obtained from stem increment cores (Suunto Oy, Vantaa, Finland) by inserting cored tissue segments into a vice and applying pressure to express sap which was collected with a Pasteur pipette and immediately transferred to a solid-state pH microsensor connected to a pH meter (Red-Line Standard Sensor, Argus meter; Sentron Europe BV, Roden, the Netherlands). The mean pH for the observation period was 7.2 ± 0.2. We accounted for the amount of dissolved CO2 originating in the soil solution that was transported via water uptake by assuming that the soil solution [CO2] was at equilibrium with that of the soil. We determined the soil [CO2] throughout the rooting zone from [CO2] measurements used to calculate soil CO2 efflux as described in soil CO2 efflux methods. The amount of dissolved CO2 originating in the soil solution was calculated as the product of mean soil [CO2] at the three depths and sap flux.

Statistical analyses

We compared daily totals of FT and soil CO2 efflux, as well as estimated belowground autotrophic and heterotrophic respiration, using repeated measures ANOVA. The pathway of belowground CO2 efflux (n = 2) and day of measurement (n = 7) were treated as fixed factors whereas the individual tree (n = 4) was treated as the random subject factor. We used Akaike's information criterion with a second order correction for small sample sizes to determine the covariance structure that best estimated the correlation among individual trees over time. The analyses were performed using the mixed models procedure of sas (Version 9.1.3; SAS Inc., Cary, NC, USA) using an alpha of 0.05. All values presented in the text are mean ± SE.


Sap flux showed a typical diel pattern during the measurement period (Fig. 1a). We also observed a diel pattern in the concentration of dissolved CO2 in xylem sap. The pattern appeared to be partially related to the rate of sap flow. The concentration of dissolved CO2 reached a maximum before the start of transpiration. As sap flow began, the concentration of dissolved CO2 declined and reached a minimum when sap flow reached a maximum.

Figure 1.

(a) Diel pattern of mean total tree sap flux (thin line) and mean concentration of dissolved CO2 in xylem sap (thick line) measured at the base of the trees over the 7-d observation period (n = 4). (b) Mean diel pattern of the flux of CO2 transported from the root system through the xylem into the stem (thin line) and soil CO2 efflux (thick line) over the 7-d observation period (n = 4).

The internal flux of CO2 from roots through xylem is a function of dissolved CO2 concentration of xylem sap and quantity of transported water. The diel pattern of FT mimicked that of sap flow, but the magnitude also depended on dissolved CO2 concentration (Fig. 1b). FT ranged from 0 µmol m−2 s−1 at night to a maximum of 12.0 µmol m−2 s−1 at peak sap flux. Soil CO2 efflux was relatively stable during the measurement period, ranging from 2.4 to 3.5 µmol m−2 s−1 (Fig. 1b). During most daylight hours, FT exceeded soil CO2 efflux, but at night the pattern was reversed.

Mean cumulative daily FT was not statistically different from mean cumulative soil CO2 efflux (Table 1; P = 0.6695). Thus, the quantity of CO2 moving internally from roots through xylem was equivalent to the total amount of CO2 diffusing from the soil to the atmosphere from autotrophic and heterotrophic sources combined. Total belowground CO2 flux (FT+ soil CO2 efflux) was 0.53 mol CO2 m−2 d−1, with 47% attributed to FT and 53% to soil CO2 efflux (Table 1). Based on soil [CO2] and sap flux, we calculated that 0.017 mol CO2 m−2 d−1 entered the tree via water uptake, representing only 7.8% of FT and 3.2% of total belowground CO2 flux. Thus, 92% of the dissolved CO2 in FT was derived from autotrophic respiration.

Table 1.  Mean (± SE) daily total flux of CO2 transported from the root system through the xylem into the stem (Xylem) and soil CO2 efflux (Soil)
Flux pathwayCO2 flux (mol m−2 d−1)
Xylem0.26 ± 0.02
Soil0.27 ± 0.01

We used our measurements of FT and soil CO2 efflux to estimate autotrophic and heterotrophic contributions to total belowground CO2 flux by assuming that 50% of soil CO2 efflux is derived from autotrophic respiration (Hanson et al., 2000). It follows that the CO2 dissolved in the soil solution that enters the tree via water uptake would also be derived from equal proportions of autotrophic and heterotrophic respiration. Under these assumptions, the autotrophic component (72%) of total belowground CO2 flux was more than twice as large as the heterotrophic component (28%) (Table 2; P = 0.0012).

Table 2.  Estimated mean daily total of CO2 derived from belowground autotrophic or heterotrophic respiration
Respiratory sourceRespiration (mol CO2 m−2 d−1)
Autotrophic0.38 ± 0.03
Heterotrophic0.14 ± 0.01


Our study provides empirical evidence demonstrating an alternative flux pathway for root-respired CO2 that can be of greater magnitude than the soil pathway. We estimated that twice the amount of root-respired CO2 entered the xylem stream as diffused into the soil environment. We also observed substantial increases in dissolved xylem CO2 at night, suggesting that roots, like tree stems (Teskey et al., 2008), may possess substantial barriers to outward diffusion which allow CO2 to concentrate in xylem sap. Some of the xylem-transported CO2 can be used as a substrate for corticular (Cernusak & Marshall, 2000; Pfanz & Aschan, 2001; Aschan & Pfanz, 2003) and leaf (Zelawski et al., 1970; Stringer & Kimmerer, 1993) carbon fixation. The large flux of CO2 from roots through xylem may be part of a recycling mechanism whereby trees retain respired CO2 for assimilation to compensate for respiratory losses.

We speculate that CO2 recycling capabilities in trees may be a vestigial carbon-concentrating mechanism, perhaps with evolutionary origins in aquatic plants. A portion of root-respired, and sediment-derived, CO2 accumulates in roots of many aquatic and wetland plant species (Brix, 1990; Constable et al., 1992; Li & Jones, 1995; Colmer, 2003). Stems of such plants generally contain arenchyma and CO2 is transported from root to shoot via diffusion where it becomes a substantial component of carbon balance (Wetzel & Grace, 1983). In fact, some plants entirely lacking stomata acquire all of their photosynthetic substrate from root-respired and sediment-derived CO2 (Keeley et al., 1984).

The magnitude of belowground autotrophic respiration in forest ecosystems may considerably exceed that of heterotrophic respiration when the amount of CO2 transported via the xylem stream is considered in addition to autotrophic contributions to soil CO2 efflux. Our assumption that autotrophic activity accounted for 50% of soil CO2 efflux suggests that total belowground autotrophic respiration may have been nearly three times larger than heterotrophic respiration when FT is considered. If only 10% of soil CO2 efflux resulted from autotrophic activity, the addition of FT suggests nearly equivalent autotrophic and heterotrophic contributions. If the autotrophic contribution to soil CO2 efflux was 90%, the addition of FT would not substantially change the relative autotrophic and heterotrophic contributions. Regardless of the relative contributions, the absolute magnitude of autotrophic respiration was twice as large when FT is considered. Consequently, belowground autotrophic respiration may consume substantially more carbohydrates in mitochondrial respiration than previously recognized. Because a portion of CO2 transported from the root system via xylem sap diffuses through the stem into the atmosphere, it also indicates that the carbohydrate cost of stem respiration has been substantially overestimated. These results further suggest that belowground autotrophic respiration may exceed aboveground (leaf and woody tissue) respiration.

Terrestrial ecosystem models estimate how forests will respond to climate change and influence the global carbon cycle, but their utility is ultimately limited by our understanding of processes regulating carbon allocation in forest ecosystems. Belowground carbon allocation represents a sizeable portion of forest gross primary production, yet our understanding of controlling processes remains poor (Giardina et al., 2005; Litton et al., 2007). Our results indicate that belowground carbon allocation may be much larger than previously estimated. For example, belowground carbon allocation has been estimated with mass balance equations under the fundamental assumption that annual changes in soil carbon storage are small relative to annual soil carbon inputs and losses (Raich & Nadelhoffer, 1989; Giardina & Ryan, 2002). Soil CO2 efflux comprises the largest flux within the mass balance equation and thus strongly controls the resulting estimates of belowground carbon allocation (Raich & Nadelhoffer, 1989; Giardina & Ryan, 2002). In our study, soil CO2 efflux only represented about half of the total belowground CO2 flux. Recent evidence of increased root allocation in forest tree species grown under elevated CO2 (Norby et al., 2004; Huang et al., 2007) suggests that FT may increase in importance as atmospheric CO2 increases toward predicted concentrations over the next century.

These findings have important implications for how we currently understand physiological functioning of trees, carbon cycling in forests and global carbon budgets. We suggest that FT should be measured concurrently with soil CO2 efflux to understand the metabolism of root systems and the carbon economy of trees and forests. We acknowledge that our findings are limited to four individual trees of the same species during a single week. To more fully understand this process, it must be examined under changing environmental conditions in a wide variety of tree functional groups and ecosystems at annual time-scales. For example, FT will be negligible when deciduous species enter dormancy. Therefore, there is a need to understand the relative importance of FT in deciduous and evergreen species. The magnitude of FT may decline with reduced rates of transpiration, so it will also be important to understand how factors affecting transpiration such as drought, vapor pressure deficit and soil water holding capacity influence FT. Factors affecting the rate of diffusion of CO2 through the soil, including soil texture, water content and [CO2], may also play a part in determining the relative importance of FT. The influence of FT on annual budgets of total belowground CO2 efflux may also be related to growing season length. Equatorial systems represent an obvious hotspot for further research into FT as the annual importance may diminish as seasonality becomes more pronounced at increasing latitudes. Still, annual magnitudes of FT may be quite large in temperate systems because growing season respiration rates greatly exceed those of the dormant season. Clearly, further research is required to improve our understanding of the importance of FT for terrestrial carbon cycling and the controlling environmental factors.


We thank J. Blake, M. Coleman and C. Trettin as well as USDA Forest Service–Southern Research Station, USDA Forest Service–Savannah River and Department of Energy–Savannah River for access to experimental plots and facilities. We also thank L. Krysinsky for valuable logistical support. Support from the US Department of Energy -Savannah River Operations office through the USDA-Forest Service Savannah River and the Forest Service Southern Research Station (DE-IA09-00SR22188), the National Research Initiative of the USDA Cooperative StateResearch, Education and Extension Service (2003-35100-13783), the National Science Foundation (0445495) and the Global Forest Foundation is gratefully acknowledged.