In situ assessment of the velocity of carbon transfer by tracing 13C in trunk CO2 efflux after pulse labelling: variations among tree species and seasons

Authors

  • Masako Dannoura,

    1. INRA, UR1263 Ecologie Fonctionnelle et Physique de l’Environnement, F-33140 Villenave d’Ornon, France
    2. Laboratory of Forest Utilization, Department of Forest and Biomaterial Science, Graduate School of Agriculture, Kyoto University, Kyoto 606-8502, Japan
    Search for more papers by this author
  • Pascale Maillard,

    1. Université Henri Poincaré, UMR 1137, Ecologie et Ecophysiologie Forestières, Faculté des Sciences, Nancy Université, F-54500 Vandoeuvre les Nancy, France
    2. INRA, UMR 1137, Ecologie et Ecophysiologie Forestières, Centre de Nancy, F-54280 Champenoux, France
    Search for more papers by this author
  • Chantal Fresneau,

    1. Université Paris-Sud, UMR 8079, Laboratoire Ecologie Systématique et Evolution, F-91405 Orsay, France
    2. CNRS, UMR 8079, Laboratoire Ecologie Systématique et Evolution, F-91190 Gif-sur-Yvette, France
    3. AgroParisTech, UMR 8079, Laboratoire Ecologie Systématique et Evolution, F-75231 Paris, France
    Search for more papers by this author
  • Caroline Plain,

    1. Université Henri Poincaré, UMR 1137, Ecologie et Ecophysiologie Forestières, Faculté des Sciences, Nancy Université, F-54500 Vandoeuvre les Nancy, France
    2. INRA, UMR 1137, Ecologie et Ecophysiologie Forestières, Centre de Nancy, F-54280 Champenoux, France
    Search for more papers by this author
  • Daniel Berveiller,

    1. Université Paris-Sud, UMR 8079, Laboratoire Ecologie Systématique et Evolution, F-91405 Orsay, France
    2. CNRS, UMR 8079, Laboratoire Ecologie Systématique et Evolution, F-91190 Gif-sur-Yvette, France
    3. AgroParisTech, UMR 8079, Laboratoire Ecologie Systématique et Evolution, F-75231 Paris, France
    Search for more papers by this author
  • Dominique Gerant,

    1. Université Henri Poincaré, UMR 1137, Ecologie et Ecophysiologie Forestières, Faculté des Sciences, Nancy Université, F-54500 Vandoeuvre les Nancy, France
    2. INRA, UMR 1137, Ecologie et Ecophysiologie Forestières, Centre de Nancy, F-54280 Champenoux, France
    Search for more papers by this author
  • Christophe Chipeaux,

    1. INRA, UR1263 Ecologie Fonctionnelle et Physique de l’Environnement, F-33140 Villenave d’Ornon, France
    Search for more papers by this author
  • Alexandre Bosc,

    1. INRA, UR1263 Ecologie Fonctionnelle et Physique de l’Environnement, F-33140 Villenave d’Ornon, France
    Search for more papers by this author
  • Jérôme Ngao,

    1. Université Paris-Sud, UMR 8079, Laboratoire Ecologie Systématique et Evolution, F-91405 Orsay, France
    2. CNRS, UMR 8079, Laboratoire Ecologie Systématique et Evolution, F-91190 Gif-sur-Yvette, France
    3. AgroParisTech, UMR 8079, Laboratoire Ecologie Systématique et Evolution, F-75231 Paris, France
    Search for more papers by this author
  • Claire Damesin,

    1. Université Paris-Sud, UMR 8079, Laboratoire Ecologie Systématique et Evolution, F-91405 Orsay, France
    2. CNRS, UMR 8079, Laboratoire Ecologie Systématique et Evolution, F-91190 Gif-sur-Yvette, France
    3. AgroParisTech, UMR 8079, Laboratoire Ecologie Systématique et Evolution, F-75231 Paris, France
    Search for more papers by this author
  • Denis Loustau,

    1. INRA, UR1263 Ecologie Fonctionnelle et Physique de l’Environnement, F-33140 Villenave d’Ornon, France
    Search for more papers by this author
  • Daniel Epron

    1. Université Henri Poincaré, UMR 1137, Ecologie et Ecophysiologie Forestières, Faculté des Sciences, Nancy Université, F-54500 Vandoeuvre les Nancy, France
    2. INRA, UMR 1137, Ecologie et Ecophysiologie Forestières, Centre de Nancy, F-54280 Champenoux, France
    Search for more papers by this author

Author for correspondence:
Daniel Epron
Tel: +33 3 83684249
Email: daniel.epron@scbiol.uhp-nancy.fr

Summary

  • Phloem is the main pathway for transferring photosynthates belowground. In situ13C pulse labelling of trees 8–10 m tall was conducted in the field on 10 beech (Fagus sylvatica) trees, six sessile oak (Quercus petraea) trees and 10 maritime pine (Pinus pinaster) trees throughout the growing season.
  • Respired 13CO2 from trunks was tracked at different heights using tunable diode laser absorption spectrometry to determine time lags and the velocity of carbon transfer (V). The isotope composition of phloem extracts was measured on several occasions after labelling and used to estimate the rate constant of phloem sap outflux (kP).
  • Pulse labelling together with high-frequency measurement of the isotope composition of trunk CO2 efflux is a promising tool for studying phloem transport in the field. Seasonal variability in V was predicted in pine and oak by bivariate linear regressions with air temperature and soil water content. V differed among the three species consistently with known differences in phloem anatomy between broadleaf and coniferous trees.
  • V increased with tree diameter in oak and beech, reflecting a nonlinear increase in volumetric flow with increasing bark cross-sectional area, which suggests changes in allocation pattern with tree diameter in broadleaf species. Discrepancies between V and kP indicate vertical changes in functional phloem properties.

Introduction

Soil–vegetation–atmosphere transfer models are key tools for predicting carbon exchange responses to climate change between the biosphere and the atmosphere, and for studying feedback between climate and ecosystem functions. However, such models still suffer from a lack of mechanistic integration of carbon allocation (Cannell & Dewar, 1994; Friedlingstein et al., 1999; Magnani et al., 2000; Litton et al., 2007). Allocation of assimilated carbon among organs is affected by the environment and this affects tree growth, the contribution of each organ to autotrophic respiration, carbon transfer to the rhizosphere and, in fine, carbon sequestration in the ecosystem (Magnani et al., 2002; Giardina et al., 2003).

The transfer of photosynthetic products belowground represents between 25–63% of gross primary productivity (Litton et al., 2007) and fulfils energy requirements for metabolic processes occurring in roots and in the (myco)rhizosphere. Previously, temporal variations in soil respiration were mostly ascribed to environmental drivers (Lloyd & Taylor, 1994; Davidson et al., 1998; Epron et al., 1999), but now there is growing evidence that soil respiration is strongly linked to plant activities (Högberg et al., 2001; Högberg & Read, 2006; Marron et al., 2009; Plain et al., 2009; Bahn et al., 2010).

Phloem is the main pathway for transferring photosynthates belowground. Increasing our understanding of carbon allocation between sinks requires better characterization of phloem transport of photosynthates (Minchin & Lacointe, 2005). In trees, the time lag between canopy photosynthesis and soil respiration depends mainly on the path length and on the velocity of phloem transport in the trunk and through the root system. This time lag ranges between 1 and 5 d depending on the height of the tree (Kuzyakov & Gavrichkova, 2010; Mencuccini & Hölttä, 2010). This velocity is thought to differ among species as a result of differences in phloem anatomy and in response to changes in environmental conditions (Kuzyakov & Gavrichkova, 2010), but very few data are currently available (Peuke et al., 2001; Helfter et al., 2007). Drought is known to drastically reduce carbon transport belowground in beech (Fagus sylvatica) seedlings (Ruehr et al., 2009) and preliminary observations indicated that a 10°C drop in air temperature may have strongly reduced the velocity of photosynthate transport between the canopy and the base of the trunk in beech trees (Plain et al., 2009). However, very little is known about seasonal variations in, and the impact of environmental factors on, phloem transport in field-grown trees (Van Bel, 2003).

The first attempts to estimate the in situ velocity of carbon transfer belowground relied on time series analysis or lag correlation analysis of temporal fluctuations of soil respiration or its 13C composition (Craine et al., 1999; Ekblad et al., 2005; Gaumont-Guay et al., 2008; Marron et al., 2009). However, the close coupling between soil respiration and microclimate may reflect not only a direct transfer of carbon from the canopy to the soil but also a several orders of magnitude faster transfer of information through pressure–concentration waves linked to dynamic changes in phloem turgor (Thompson & Holbrook, 2004; Mencuccini & Hölttä, 2010). In addition, mixing of different CO2 sources and post-photosynthetic carbon isotope fractionation may overshadow the coupling between canopy photosynthesis and soil respiration (Kodama et al., 2008).

Pulse labelling of plants with isotopically enriched (or depleted) CO2 is therefore seen as the most appropriate method for studying the velocity of carbon transfer in plants (Dawson et al., 2002; Kuzyakov & Gavrichkova, 2010). Until recently, labelling experiments were restricted to young tree saplings in pots (Horwath et al., 1994; Pumpanen et al., 2009; Ruehr et al., 2009), or to small trees in the field (Mordacq et al., 1986; Carbone et al., 2007; Högberg et al., 2007; but see Plain et al., 2009) and the ability to precisely determine the velocity of carbon transfer was limited by constraints on the frequency of measurement of 13C during the chase period. The recent development of laser-based infrared gas analysers has increased by several orders of magnitude the measurement frequency of 13CO2 in respiratory fluxes in pulse labelling experiments (Bahn et al., 2009; Plain et al., 2009).

The aim of this study was to estimate the vertical velocity of carbon transfer inside the trunks of two deciduous broadleaf species (beech and sessile oak (Quercus petraea)) and one coniferous evergreen species (maritime pine (Pinus pinaster)). We compared time lags between the photosynthetic assimilation of 13CO2 and its recovery in trunk CO2 efflux recorded continuously by tuneable diode laser absorption spectrometers three to four times during the growing season. Differences in time lags in trunk CO2 efflux recorded at different heights along the trunk were used to estimate the velocity of phloem transport of 13C-labelled compounds in trees 8–10 m tall. We further compared time lags and velocities with the dynamics of the label recovered in the phloem sap. We hypothesized that the velocity of carbon transfer is higher in angiosperm than in gymnosperm species because of differences in phloem anatomy and that velocity changes seasonally according to changes in temperature and soil water content.

Materials and Methods

Study sites

The study was conducted on beech (Fagus sylvatica L.), sessile oak (Quercus petraea Matt Liebl.) and maritime pine (Pinus pinaster Ait.). The beech stand is 20-yr-old natural regeneration located in the state forest of Hesse (48°40′N, 7°04′E, 300 m elevation) on a luvisol. Mean annual air temperature and precipitation are 9.2°C and 820 mm, respectively. The oak stand is 15-yr-old natural regeneration located in the state forest of Barbeau (48°2′N, 02°47′E, 90 m elevation) on a gleyic luvisol. Mean annual air temperature and annual rainfall are 12.9°C and 680 mm, respectively. The pine plantation was established in 1998 (12 yr old) at the INRA (Institut National de Recherche Agronomique) domain of Pierroton (44°45′N, 0°42′W, 60 m elevation) on a sandy spodosol with an initial tree spacing of 2 × 2 m. The plot was thinned in November 2008 by removing one line out of two. Mean annual air temperature and precipitation are 13.3°C and 930 mm, respectively. The height of dominant trees was 8–10 m in the three plots with average diameters at 1.3 m height (dbh) of 6.7, 8.7 and 13.2 cm for beech, oak and pine, respectively.

Experimental design

Two trees were selected for each labelling date at all sites, giving a total number of 10 beeches, six oaks and 10 pines (Supporting Information Table S1). Labelling was conducted in September 2008 and May, July and August 2009 for beech, in May, July and September 2009 for oak and in June, August and November 2009 and February 2010 in pine in order to cover the different growing phases of each species: the beginning of radial growth, active radial growth and the termination of radial growth for the three species, and additionally the rest period in winter for the evergreen pine.

Two additional beeches and pines that were exposed to rainfall exclusion were labelled in August (beech) or in November (pine). Rainfall exclusion roofs, made with polyethylene film and supported by a woody frame (3 m × 3 m), were installed 1.5 m above the forest floor on two beech trees and two pine trees to divert rainfall from the soil delimited by the trench. The roofs were installed in April 2009 for beech, excluding rainfall throughout the growing season, and in September 2010 for pine, excluding rainfall after the summer drought.

A trench 0.5–0.6 m deep was dug in winter around each tree at least 5 months before labelling. This depth corresponds to the presence of a compact clay horizon for both beech and oak, and to an indurate soil layer (alios; hardpan) for pine, which limit vertical root development. We did not assess to what extent trenching affected the root systems of the trees, but we assumed that the root system had recovered from the trenching stress at the time of labelling. The trench was lined with a polyethylene film and filled back. All roots and root exudates within this soil volume therefore originated from the isolated tree, and were contained in this trench volume. The area delimited by the trench depended on the density of the stand; that is, on the spacing between trees. The area averaged 3 m2 in the beech (2.1–4.0 m2) and oak (2.2–5.5 m2) stands, and was 6 m2 in the pine plantation. These areas were higher than the mean area per tree based on stand density.

Microclimate parameters, including photosynthetic photon flux density (PPFD; measured using Delta T-BF2 (Delta-T devices, Cambridge, UK) for beech; a homemade quantum sensor based on a gallium arsenide photodiode for oak; and DBE (Solems, Palaiseau, France) for pine), air temperature and relative air humidity (TA and RH, respectively; Vaisala HMP45, Vaisala, Helsinki, Finland), were recorded using data loggers (Campbell Scientific, Logan, UT, USA) at a 10–60 s pace and averaged half-hourly. Soil water content (SWC) in the vicinity of each labelled tree inside the area delimited by the trench was recorded half-hourly at 30 cm depth for pine with time domain reflectometry (TDR) probes (CS616; Campbell Scientific) and at 10 cm depth for oak with an impedance probe (ML2x ThetaProbes; Delta-T Device), and measured once a week at 15 cm depth for beech using TDR probes (Trase; SoilMoisture Equipment Corp., Santa Barbara, CA, USA).

Predawn leaf water potential was measured once on three to four leaves per tree just before labelling at the end of August for beech (four trees) and at the beginning of September for oak (two trees) using a Scholander-type pressure chamber (PMS Instrument, Corvallis, OR, USA).

Pulse labelling

Each pulse labelling was performed using the whole-crown labelling chamber previously described (Plain et al., 2009). The whole crown of the tree was inserted into a 20–25 m3 chamber made of 200-μm polyane film and held up by two 12-m-high stainless steel scaffolds erected on either side of the tree. The air temperature inside the crown labelling chamber was recorded, controlled and maintained at the outside air temperature. For oak and beech, this was done by circulating air at a rate of 1200 m3 h−1 through an air- and water-cooled condensing axial fan outdoor unit connected to an air conditioner (AXCZ 221 and BSH-BZ 221; HITECSA, Vilanova i la Geltrú, Spain) (see Plain et al., 2009 for details), while for pine, two air conditioners (Bodner and Man Pap 3500 W, Castorama, Templemars, France) and three axial fans were directly inserted into the labelling chamber.

Evolution of both 12CO2 and 13CO2 concentrations ([12CO2] and [13CO2], respectively) inside the labelling chamber was monitored simultaneously with a 12CO2/13CO2 infrared gas analyser (S710; SICK/MAIHAK, Reute, Germany; accuracy 5%) calibrated by diluting pure CO2 from gas tanks having different 13C/12CO2 (10, 50 and 100%) in a stream of CO2 free air using mass flow controllers. The 12CO2 concentration decreased to 60–150 μmol mol−1 after the chamber was closed because of CO2 assimilation by photosynthesis. A total amount of 25 l of pure 13CO2 (99.299 atom%; Eurisotop; Cambridge Isotope Laboratory Inc., Andover, MA, USA) was then progressively injected at a flow rate adjusted to maintain the 13CO2 concentration in the chamber at between 300 and 400 μmol mol−1, except in November and February for pine, when higher 13CO2 concentrations were used to compensate for lower photosynthetic activity (800–1000 μmol mol−1). The duration of labelling ranged between 3 and 6 h for beech, 1 and 3 h for oak and 2 and 5 h for pine, depending on tree size and activity.

13C composition of CO2 efflux

The isotope composition of CO2 effluxes (δ13CF) from the trunk was computed from [12CO2] and [13CO2] measured at the inlet and outlet of flow-through chambers (Marron et al., 2009; Plain et al., 2009) by tuneable diode laser absorption spectroscopy with a trace gas analyzer (TGA 100A; Campbell Scientific). Air was pumped continuously through the chamber at a flow rate ranging from 300 to 1200 l min−1. A manifold was used to switch between working standards and the chamber inlet and outlet lines. Three standards were measured before measurements were made at the inlet and outlet of the respiration chamber. The mean concentration was recorded over 20 s after a 30-s stabilization period. Certified standards of CO2 in synthetic air (0.5% precision for CO2 concentrations, Air Products, Paris, France for beech, Air Liquide, Paris, France for oak and DEUSTE Steininger GmbH, Mühlhausen, Germany for pine) were measured for isotope composition every 2–6 months using an isotope ratio mass spectrometer (IRMS; Delta S; ThermoFinnigan, Bremen, Germany) that had been calibrated against IAEA (International Atomic Energy Agency) reference materials. The isotope composition was used to compute the [12CO2] and [13CO2] of each working standard. The ranges of available concentrations were 293.2 to 1281.2 μmol mol−1 for 12CO2 and 3.17 to 13.75 μmol mol−1 for 13CO2. The flow rate of air through the respiration chambers was adjusted to maintain the [12CO2] and [13CO2] at the outlet within the range of the calibration standards. The precision of the instrument at reproducing calibration tank values was 0.3 μmol mol−1 for 12CO2 and 0.007 μmol mol−1 for 13CO2.

The isotopic composition of CO2 effluxes (δ13CF; ‰) was calculated as:

image(Eqn 1)

(RVPDB, the isotopic ratio of Vienna Pee Dee Belemnite (VPDB; 0.011179602)).

The trunk chambers on beeches and oaks were composed of two polymethylmethacrylate half-boxes with semicircular openings to accommodate the trunk (Damesin et al., 2002). The trunk chambers on pines were composed of polycarbonate sheet wound around the trunk on foam rubber to form a cylinder (height 0.1–0.2 m; distance to bark 0.03 m). The vertical distance between the trunk chambers and the soil surface was measured. One chamber was set up at the base of the trunk (0.2–0.7 m from the soil) and at the base of the crown (2.1–3.3 m for beech and oak, and 3.8–5.3 m for pine). δ13CF was measured every 40–80 min for at least 5 d and up to 2 wk.

Carbon isotope composition of phloem extracts

Two small disks of bark (12 mm diameter for beech and pine and 14 mm for oak) were collected on labelled trees at a height of 1.3 m at 9 h Coordinated Universal Time (UTC) using a corer (Gessler et al., 2004). Disk sampling was repeated 4–5 times during the 10-d chase following the 13CO2 labelling pulse. Additional samples were collected before the labelling and on a nearby unlabelled tree. The inner bark tissue was placed in a 10-ml vial containing 2 ml of ultrapure water after removal of the outer part of the bark (periderm), and it was left to incubate for 5 h at ambient temperature. Then the bark pieces were separated from the phloem extract solution, dried for several days in an oven at 60°C, and then weighed. The phloem extract solution was filtrated on a nylon cartridge (Whatman, 0.2 μm Nyl.; diameter 25 mm; ref. 17 463 433) and the filtrate was vacuum-evaporated for 4 h (Maxi-Dry plus, Heto-model DW1, 0-110; Heto-Holten A/S, Allerod, Denmark). Dried extracts were weighed and then diluted in ultra-pure water proportionally to their dry mass. An aliquot of 60 μl of the resulting solution (containing c. 0.8–1 mg of dried phloem extract) was transferred into tin capsules (Elemental Microanalysis, Cambridge, UK; 6 × 4mm; ref. D1006, BN/139877), freeze-dried (Lyophilizer Christ; Alpha 1-2 LD plus, Osterode am Harz, Germany), weighed and analysed for carbon isotope composition and total carbon using an elemental analyser (NA 1500 NCS; Carlo Erba, Milan, Italy) coupled to a Delta-S isotopic ratio mass spectrometer (Finnigan-Mat; Thermoquest Corp., San Jose, CA, USA) for beech and pine samples and to a VG Optima IRMS (Fison, Villeurbanne, France) for oak samples. The carbon isotope composition of the phloem extract (δ13CP) was expressed in delta notation (in ‰) relative to the VPDB standard.

The thickness of the inner bark tissues of all labelled trees and a few additional unlabelled trees was assessed on additional phloem disks collected at the same time for all trees at a given site. Measurements were made with a ruler at 0.5 mm accuracy for oak and with a calliper at 0.1 mm accuracy for beech and pine.

Data analysis

The time lag (LT; h) between the start of the labelling and the appearance of 13C in CO2 efflux in each chamber was computed by fitting a quadratic function to the relationships between δ13CF (‰) and the time after labelling (t; h; Fig. 1a) using nonlinear least-squares regression (PROC NLIN of the sas software; Marquardt–Levenberg method; SAS Institute Inc., Cary, NC, USA):

image(Eqn 2)

with a (‰) matching the pre-labelling isotope composition of the efflux. Time lags for the chamber at the base of the trunk (bottom chamber) and at the base of the crown (top chamber) were further denoted LT1 and LT2, respectively.

Figure 1.

 Schematic diagram showing respiration at two positions along the trunk on the left and the flow of labelled carbon through the phloem on the right. (a) Determination of the time lag (LT) between the start of the labelling and the first appearance of 13C in CO2 efflux by fitting a quadratic function to the relationships between δ13CF and the time from the beginning of the labelling (Eqn 2). The example shows δ13CF in trunk CO2 efflux (top chamber) in oak in September 2009. (b) Determination of the time lag (LP) between the start of the labelling and the first appearance of 13C in the phloem extract using a single pool model, and assuming that outflux from phloem extract follows first-order kinetics with a rate constant kP and that influx is equal to outflux. The example shown is for pine in August 2009.

The velocity of carbon transfer in the trunk (V; m h−1) was calculated as

image(Eqn 3)

(d (m), the distance between the two chambers.) Our calculation assumes that, if transferred carbon is not at least partly respired immediately or the diffusion time of CO2 across the bark is not negligible, these delays are similar for the two chambers, so they are cancelled by the subtraction of the two time lags.

The kinetic of label recovered in the phloem extract was described using a single pool model that was fitted to the observed δ13CP values (Fig. 1b). The label that left the crown arrived after a time lag (LP) into the phloem extract collected at a height of 1.3 m above the ground. We assumed that the outflux of label from the crown and from phloem extract collected at a height of 1.3 m followed a first-order kinetic with the same rate constant (kP) and that the influx into the phloem extract was equal to the outflux from the crown. The rate of change of label in the phloem extract with time (t) is given by:

image(Eqn 4)

(CP(t), the amount of 13C label in the phloem extract at time t after labelling; C0, the total amount of label that has flowed out of the crown through the phloem.)

The analytical solution is:

image(Eqn 5)

Because we were only interested in LP and kP, we used the difference in δ13CP between labelled and unlabelled trees as a proxy for CP. The differential equation was solved analytically to estimate the values of LP and kP by iteratively minimizing the sum of squared differences between measured and predicted δ13CP using the Microsoft Excel solver tool. Because of the low number of collected samples, the standard fitting procedure could not be applied satisfactorily so the fit quality was checked visually. We were aware that LP could not be determined when the peak occurred before the first sampling 24 h after labelling. The mean residence time (MRT) of label in the phloem extract was calculated as 1/kP.

Linear regressions were used to analyse the relationships between the different computed parameters (L, LP, V and kP). Linear regressions (simple or multiple) were also used to evaluate the relationships between these computed parameters and either tree dimension (dbh) or environmental variables (PPFD, RH, TA and SWC). The coefficient of determination was tested with a two-tailed test for significance of the regression (the REG procedure of the sas software).

Analysis of variance was performed to test significant differences in phloem dimension between species and trees nested within species (the GLM procedure of the sas software).

Results

Environmental conditions

Pulse labelling experiments were performed throughout the growing season for the three species. The daily average temperatures on the labelling dates ranged between 9 and 24°C for beech and pine and between 15 and 24°C for oak. The volumetric SWC ranged between 0.31 and 0.39 m3 H2O m−3 in the beech stand, except for trees under rainfall exclusion roofs (0.13–0.23 m3 H2O m−3). There was no rainfall exclusion treatment in oak, but SWC exhibited a more pronounced seasonal variation than in beech, with values ranging from 0.30 m3 H2O m−3 in May to 0.20 m3 H2O m−3 in September. In pine (sandy soil), SWC was lower than for oak and beech, ranging between 0.05 and 0.18 m3 H2O m−3 for control trees and being 0.05 for both trees under rainfall exclusion roofs (Table S1). Predawn leaf water potential values were similar (−0.4 MPa; data not shown) for the four beech trees labelled in August, regardless of whether or not they were subjected to rainfall exclusion, and dropped below −0.9 MPa in oak in early September.

Recovery of 13C in trunk CO2 efflux

The carbon isotope enrichment of CO2 efflux (δ13CF) from the trunk increased after each pulse labelling in all species after a time lag from an average of −25‰ up to 10 000‰. The label, as expected, appeared first in CO2 efflux at the base of the crown and then at the base of the trunk, and started to decline after several hours or days, depending on the species and the time of year (Fig. 2). The δ13CF values measured at peak time were often, but not always, lower for the bottom chamber than for the top chamber (typical examples are shown in Fig. 2). In oak, the difference was visible early in the growing season but not later (Fig. 2; central panels). In beech, there was no clear seasonal pattern for this difference in maximum δ13CF between the base of the crown and the base of the trunk. The difference was more pronounced for smaller trees than for bigger trees (Fig. 2; left panels) and was indeed negatively related to dbh (R= 0.58, = 0.012; data not shown). Rainfall exclusion did not alter the shape of the kinetics of recovery of 13C in trunk CO2 efflux in beech, while the δ13CF values measured at peak time were much lower in pine trees exposed to water exclusion (Fig. 2; Excl. labels). The shape also differed markedly between the growing season and the dormant period for pine (Fig. 2; right panels).

Figure 2.

 Typical examples of time courses of the isotope composition of CO2 effluxes (δ13CF) in the trunk at the base of the crown (open symbols) and at the base of the trunk (closed symbols) after pulse labelling of beech (circles, left panels), oak (diamonds, central panels) and pine (triangles, right panels). The date of labelling, trunk diameter at 1.30 m height (dbh; cm), air temperature averaged for the 24 h following labelling (TA; °C) and the volumetric soil water content (SWC; m3 H2O m−3) are given inside each panel. ‘Excl.’ indicates that the tree was under a rainfall exclusion roof.

In beech, the time lag in trunk CO2 efflux (LT2) ranged from 6 to 9 h at the base of the crown (Fig. 3a). The signal appeared c. 3 h later at the base of the trunk (LT1; Fig. 3a, Table S1). The time shifts were longer on a few occasions, such as in September 2008, when the mean air temperature after labelling was 9°C (mean temperature ranged between 13 and 24°C for the other labelling experiments on beech). This also occurred in August 2009 on the smallest labelled beech tree (dbh = 4.9 cm). In oak, LT2 values were c. 5–6 h at the base of the crown and 8–11 h at the base of the trunk (LT1) in May and July. Signal appearance was delayed in September (c. 9 and 16 h, respectively; Fig. 3b). In pine, the time lags in trunk CO2 efflux were more variable (from 4 to 47 h at the base of the crown; Fig. 3c) and the time shift between the top and the bottom chambers was much higher than for the other species, with time lags of 30–80 h at the base of the trunk.

Figure 3.

 Relationships between time lag from the start of the labelling to the first appearance of 13C in CO2 efflux (LT; h) and height of the chamber above soil level (H; m). Each line represents a different tree of beech (a), oak (b) and pine (c). The horizontal dashed line indicates the position of bark sampling.

A shift of similar amplitude was observed for the maximum δ13CF occurring at the base of the crown and at the base of the trunk, with an averaged delay of 5 h in beech (40 h vs 45 h on average, respectively), 10 h in oak (45 h vs 55 h) and 40 h in pine during the growing season (June and August labelling). The average delay was 90 h for pine trees labelled in November and February (data not shown).

Velocity of carbon transfer

The velocity of carbon transfer in the trunk (V; m h−1) exhibited both marked differences among species (slopes in Fig. 3; Table S1) and high variability within a species. Values fell within similar ranges in beech and oak, between 0.22 and 1.21 m h−1 in beech (mean value 0.76 m h−1) and between 0.36 and 1.02 m h−1 in oak (mean value 0.59 m h−1), and were four to five times lower in pine (between 0.09 to 0.21 m h−1; mean value 0.15 m h−1).

No correlation between tree size (dbh) and V was observed in pine, while there was a positive correlation in beech (R= 0.66, = 0.008; Fig. 4). A similar but not statistically significant relationship was found for oak (R= 0.52, = 0.11), and was improved by adding SWC to the model (R= 0.82, = 0.07).

Figure 4.

 Relationships between the velocity of carbon transfer in the trunk (V; m h−1) and trunk diameter at 1.30 m height (dbh; cm) in beech (circles), oak (diamonds) and pine (triangles). Closed symbols are for trees under rainfall exclusion roofs. The linear regression was significant for beech (R= 0.66, = 9, < 0.05; full line).

V and LT1 were positively related to air temperature (TA) in pine. The highest R2 values (Fig. 5a, Table 1; Eqn M1 for V; Fig. 5b, Table 1; Eqn M2 for LT1) were obtained when air temperatures were averaged during the 24 h following labelling. The relationship between V and TA was further improved by adding SWC as a co-variable to the model (Table 1; Eqn M3). Whereas no relationship between V and TA or between V and SWC was found for oak or beech, the bivariate model predicted V satisfactorily in oak with temperature average over 48 h (Table 1; Eqn M5), but not in beech.

Figure 5.

 Relationships between air temperature averaged for the 24 h following labelling (TA; °C) and the velocity of carbon transfer in the trunk (V; m h−1) (a) and the time lag in carbon transfer at the base of the trunk (LT1; h) (b) for beech (circles), oak (diamonds) and pine (triangles). The closed symbols indicate the two trees that were under rainfall exclusion roofs. The linear regressions were significant for pine (R= 0.56 and 0.74, respectively; = 10, < 0.05; full line). Note that the scale of the y-axes changes at V = 0.2 m h−1 and LT1 = 25 h.

Table 1.   Velocity of carbon transfer in the trunk (V; m h−1), the time lag between the start of the labelling and the first appearance of 13C in trunk CO2 efflux at the base of the trunk (LT1; h) or in phloem extracts collected at 1.3 m height (LP; h), and rate constant of the outflux of label from phloem extracts (kP; h−1) as a function of soil water content (SWC; m3 H2O m−3) and air temperature (TA; °C) averaged for 24 h (pine) or 48 h (oak) following labelling
EqnSpeciesEquationStatistics
  1. * and indicate that the parameter is different from zero at, respectively, < 0.05 and < 0.1.

M1PineV = 0.0056*TA + 0.060R= 0.56, = 0.013
M2PineLT1 = −3.40*TA + 109*R= 0.74, = 0.002
M3PineV = 0.0055*TA + 0.29 SWC − 0.029R= 0.73, = 0.011
M4PineLP = −3.89*TA + 110*R= 0.58, = 0.011
M5OakV = 0.069*TA + 6.28* SWC − 2.23*R= 0.93, = 0.017
M6OakkP = −0.085* SWC + 0.048*R= 0.80, = 0.017

Phloem sap extracts

The carbon isotope composition of phloem extracts (δ13CP) collected at a height of 1.3 m increased sharply after a time lag and started to decrease thereafter (Fig. 1b). The determination of the time lag (LP) and the peak time was constrained by the sampling schedule, especially for beech and oak, because δ13CP started to decrease before the first sampling date in most cases. LP values were indeed always < 1 d in beech and oak, in agreement with the time lag of δ13C in CO2 efflux at the base of the trunk (LT1), except for the beech tree labelled in September 2008, when the temperature was 9°C (LP = 37 h). LP ranged between 23 and 100 h in pine and was closely related to LT1 (R2 = 0.77, < 0.001; Fig. 6a). LP was on average 5 h less than LT1 in pine, which makes sense considering that phloem was collected at c. 1.3 m height while the bottom respiration chamber was at c. 0.3–0.5 m height. LP was well correlated with air temperature in pine (Fig. 6b, Table 1; Eqn M4).

Figure 6.

 Relationships between the time lag in phloem extract (LP; h) and air temperature averaged during the 24 h following labelling (TA; °C) (a) and the time lag in carbon transfer at the base of the trunk (LT1; h) for pine (b). The closed triangles indicate the two trees that were under rainfall exclusion roofs. The linear regressions were significant (R= 0.58 and 0.77, respectively, = 10, < 0.05, full line).

Interestingly, and despite differences in V, the rate constants of the label outflux from phloem extracts (kP) were similar in beech and pine, ranging from 0.017 to 0.078 h−1 in beech (MRT between 13 and 58 h; Table S1) and from 0.009 to 0.067 h−1 in pine over the whole season, with the lowest values in winter (MRT between 15 and 117 h). By contrast, kP was two times lower in oak, ranging from 0.022 to 0.033 (MRT between 30 and 46 h), and was correlated with SWC (Fig. 7, Table 1; Eqn M6).

Figure 7.

 Relationships between the rate constant of the outflux of label from the phloem extract (kP; h−1) and the volumetric soil water content (SWC; m3 H2O m−3) for beech (circles), oak (diamonds) and pine (triangles). The closed symbols indicate the two trees that were under rainfall exclusion roofs. The linear regression was significant for oak (R= 0.80, = 6, < 0.05; full line).

Bark dimensions

The inner bark tissues were significantly thicker in oak (4.9 mm) and thinner in beech (2.0 mm) compared with pine (2.9 mm), and positive correlations between inner bark thickness and dbh were observed in beech (R= 0.86, < 0.001) and in oak (R= 0.32, = 0.06), but not in pine (Fig. 8). There were also significant differences in the dry mass of inner bark tissue (MB), the dry mass of phloem sap extracts (ME) and phloem sap extracts per unit dry mass of inner bark tissue (MB/ME) among the three species (Table 2).

Figure 8.

 Relationships between phloem thickness (T; mm) and trunk diameter at 1.30 m height (dbh; cm) in beech (circles; R= 0.86), oak (diamonds; R= 0.32) and pine (triangles). The linear regression was significant for beech (R= 0.86, = 11, < 0.05; full line) and marginally significant for oak (R= 0.31, = 12, = 0.06; dashed line).

Table 2.   Dry mass of phloem tissue (MB), phloem sap (ME) and sap content (MB/ME) in beech, oak and pine
SpeciesMB (g m−2)ME (g m−2)MB/ME (mg g−1)
  1. Values are means ± SD. Analyses of variance (F ratio) for these three variables as affected by species and trees nested within species are given.

  2. df, degree of freedom. F ratios followed by *** are significant at 0.001.

Beech (= 168)815 ± 233 19 ± 8 24 ± 9
Oak (= 144)1526 ± 493 68 ± 27 46 ± 16
Pine (= 140)566 ± 128 44 ± 13 80 ± 26
Species (df = 2)1101***487***457***
Tree, nested within species (df = 31)  33***10***  4***

Discussion

Time lag and velocity

The time lag between the start of the labelling and the first appearance of 13C in CO2 efflux varied with the position of the chamber along the trunk, and among trees depending on both the species and the season. It depends on the path length and the velocity of phloem sap, but the relationship between time lags and velocity may be more complicated if there is leakage and reloading with transitory storage and further 2remobilization of the stored carbohydrates (Minchin & Lacointe, 2005; Gessler et al., 2008; Bahn et al., 2009; Kuzyakov & Gavrichkova, 2010; Mencuccini & Hölttä, 2010). Indeed, the recovery of label in trunk CO2 efflux demonstrates this leakage, which sustains trunk cell metabolism. In addition, the difference in maximum δ13CF between the base of the crown and the base of the trunk that was observed on some occasions indicates that storage or growth removes a fraction of 13C transport by the phloem sap. Nevertheless, assuming similar leakage and reloading velocity at any position along the trunk, more reliable estimates of phloem sap velocity can be computed from differences in time lag at different heights.

Velocity calculated on time lags represents the maximal velocity (Fisher, 1990) and differs from average velocity depending on the radial velocity profile, both within the sieve tubes (parabolic profile according to the Poiseuille law) and in the bark. Assuming a first-order kinetics for a solute flowing out of phloem, the rate constant (kP) is related to the average velocity (Vavg) and to the path length (l):

image

As expected, the average velocities computed from kP values were lower for beech and oak than the maximal velocities computed from the time lags of labelled recovery, assuming the path length is equal to the tree height (i.e. 10 m). The difference increased with increasing maximal velocity and was more pronounced in oak than in beech. In pine, Vavg was unexpectedly two times higher than V. The reason for this discrepancy remains unclear. The comparison of V and Vavg computed from kP estimated from sap collected at 1.3 m height assumes constant phloem and phloem sap properties along the trunk (viscosity, solute concentration, turgor pressure difference and anatomy), which may not be a valid assumption (Mencuccini & Hölttä, 2010). It has indeed been postulated that phloem diameter changes vertically (Hölttäet al., 2009).

Differences among seasons

Seasonal changes in air temperature did not account for the variability of LT1 and V in beech, except for the late labelling in September 2008, when the air temperature was below 10°C. The results for beech contrasted with those obtained for oak and pine, for which V could be accurately predicted by a bivariate linear regression with TA and SWC. While temperature may affect active processes such as phloem loading, the mechanism by which the temperature may influence phloem transport velocities is still unclear. A temporary inhibition of phloem sap velocity by petiole cooling was observed in sugar beet (Beta vulgaris) and bean (Phaseolus vulgaris) plants (Geiger & Sovonick, 1970). Theoretical analysis of phloem transport has shown that the maximum velocity is limited by solution viscosity (Hölttäet al., 2009). A drop in temperature will increase the viscosity of the phloem sap and will decrease its velocity. The decrease in velocity might be reinforced by a blockage of sieve plates by P proteins (Peuke et al., 2006). Imbalance between phloem loading and unloading might also change the gradient of hydrostatic pressure along the phloem.

The hydrostatic pressure difference between source (phloem loading) and sink (phloem unloading) organs, and the subsequent exchange of water between sieve cells and the apoplast, is thought to drive phloem transport of assimilates according to the Münch theory (Hölttäet al., 2009). Consistently, sap velocities measured in the phloem are of the same order of magnitude as those found in the xylem (Helfter et al., 2007). This close coupling between xylem and phloem transport makes it likely that a change in plant water potential will affect the rate of phloem transport (Hölttäet al., 2009). A depressive effect of water stress on carbohydrate transport in the phloem was indeed observed in water-stressed beech seedlings, for which the build-up of 13C in twig phloem sap after pulse labelling was slower than in control seedlings (Ruehr et al., 2009). These authors also observed that 13C peaked 2 d later in the trunk phloem compared with the twig phloem, suggesting a velocity of 0.01 m h−1. In contrast to these results, there was no relationship between SWC and either LT1, V, LP or kP for beech in our study. This might be attributable to a much milder drought treatment in this experiment than with the potted seedlings, as predawn leaf water potential values were similar for the four trees labelled in August (−0.4 MPa; data not shown) despite the difference in SWC. In oak and pine, SWC, together with temperature, affects the velocity of C transfer in the trunk (V). The predawn leaf water potential of oak dropped below −0.9 MPa in early September. An imbalance between phloem loading (which might decrease with a drought-induced decrease in net CO2 assimilation) and unloading, as well as a reduction in transpiration as a result of stomatal closure, should decrease the gradient of hydrostatic pressure along the phloem and thus transport velocity. However, the rate constant of carbon outflux in the phloem (kP) was unexpectedly negatively related to SWC in oak, indicating opposite trends for V and Vavg computed with kP estimated from sap collected at 1.3 m height.

Differences among species

Velocity differences between pine and the two broadleaf species are consistent with previous measurements on young seedlings. A value of 0.12 m h−1 was previously reported for Pinus Banksiana and Picea abies seedlings, while values ranging between 0.3 and 0.6 m h−1 were reported for Fraxinus pennsylvanica ash and Ulmus americana using 11C labelling (Thompson et al., 1979). The slower rate in coniferous compared with broadleaf trees is consistent with recent laboratory experiments showing a delayed peak of 14C recovery in belowground CO2 efflux in Pinus sylvestris and Picea abies Norway compared with Betula pendula seedlings (Pumpanen et al., 2009). However, such results contrast with indirect values derived from time-lagged correlations between the 18O compositions of phloem sap at different heights in the trunk of 38-yr-old Pinus sylvestris (Barnard et al., 2007), which were about five times higher (0.5–1 m h−1) than those we calculated for P. pinaster.

Differences in phloem anatomy may account for differences in phloem transport rates between coniferous and broadleaf trees. According to Poiseuille’s law, the velocity will depend on the gradient of hydrostatic pressure along the phloem, on the length of the path between source and sink organs and on the hydraulic conductivity (K). K increases by the power of 2 as the tube radius increases (r) and decreases with increasing fluid viscosity (η; i.e. with increasing sugar concentration or decreasing temperature). In addition, the anatomy of the sieve plate is also important. K will decrease with increasing pore length or with decreasing pore radius or number of pores (Sheehy et al., 1995; Thompson, 2006). Most of the resistance to sap flow is attributable to the pore dimension (Thompson & Holbrook, 2003) and may be additionally restricted by callose formation (Mullendore et al., 2010). A higher diameter of sieve tube elements (20–50 μm; Trockenbrodt, 1994; Sheehy et al., 1995) than of sieve cells (10–30 μm for 16 pine species; Shigematsu, 1959) and a better connection at the sieve plate between sieve tube elements than between sieve cells (Hepton & Preston, 1960; Parthasarathy, 1975) are therefore thought to account for the higher conductivity in the two broadleaf species than in pine, and, for a given gradient of hydrostatic pressure, to the higher velocity of phloem sap. Information about gradients of hydrostatic pressure in the phloem of trees is rather scarce (Milburn, 1975). Values ranging between 0.02 and 0.06 MPa m−1 were reported for oak (Hammel, 1968) and a value of 0.05 MPa m−1 is commonly assumed for trees (Milburn, 1975).

The amount of photosynthates transported by the phloem sap depends also on the phloem sap concentration (Canny, 1975). Despite similar velocities for oak and beech, the mass flow density, that is, the product of velocity by sap concentration, is higher in oak than in beech, and provided that the sap content, when expressed in mg g−1, is a good proxy for sap concentration. Because the inner bark (phloem tissue) of oak is twice as thick as that of beech, the total mass flow may be much higher in oak than in beech. However, the functional secondary phloem is smaller than the bark thickness, the proportion of sieve elements in the phloem varying between 20 and 70% of the cross-sectional area (Curtis, 1936; Canny, 1975). In addition, the phloem tissue produced in previous years may not be functional in some species, the layer of conducting phloem ranging from 0.2 to 1 mm only in broadleaf trees (Sheehy et al., 1995; Kozlowski & Pallardy, 1997), but comparative anatomy of oak and beech phloem is lacking.

Effects of tree dimension

The independence between dbh and V observed in pine suggests that the product of V and bark cross-sectional area (i.e. the volumetric flow in m3 s−1) increases linearly with increasing phloem area, probably scaling with increasing crown photosynthesis. The number of replicates was too small to show an eventual relationship between dbh and V in oak. Such relationships might have been hidden by a drop in water potential in August, as suggested by a positive relationship between dbh and V observed when SWC is added as a co-variable to the model. The positive relationship between dbh and V in beech, which is only partly offset by the increase in bark thickness with dbh, indicates that the volumetric flow increases as a power of bark cross-sectional area > 1. This suggests changes in allocation pattern with tree size in this species, that is, an increase in carbon allocation belowground with increasing tree size. The negative correlation between dbh and the difference in maximum δ13CF between the base of the crown and the base of the trunk also suggests that less carbon is removed from the phloem sap between the crown and the soil.

Conclusions

The vertical velocity of carbon transfer in trees 8–10 m tall differed markedly between two deciduous broadleaf species (beech and oak) and maritime pine, reflecting difference in phloem anatomy between angiosperms and gymnosperms, but little is known about species differences in the functional anatomy of phloem. The velocity was positively related to tree diameter in the two broadleaf species, suggesting that the volumetric flow increased more than the phloem cross-sectional area with increasing tree diameter. The velocity of carbon transfer changed seasonally according to changes in temperature and soil water content in oak and in pine, but the mechanism behind this is still unclear and additional experiments are required to test the hypothesis that an imbalance between phloem loading (source activity) and unloading (sink demand) may account for variation in the velocity. A better understanding of this mechanism will be a prerequisite for coupling phloem transfer to leaf photosynthesis and to organ growth and respiration in forest ecosystem models.

Acknowledgements

Financial support was provided by the French National Research Agency (ANR) through the CATS Project (ANR-07-BLAN-0109). The CATS labelling experiment was a team effortand the authors would like to thank all persons who took part in it: Annick Ambroise, Carole Antoine, Laure Barthes, Delphine Brazzalotto, Marion Devaux, Yanwen Dong, Emilie Kartner, Catherine Lambrot, Perrine Philippe, Stéphane Bazot, Claude Brechet, Ewen Chesnel, Bernard Clerc, Pascal Courtois, Jean-Marie Gioria, Patrick Gross, Christian Hossann, Théo Kazmierczak, Jean-Christophe Lata, Jean-Louis Mabout, Thomas Mathioni, Florian Parent, Jean-Yves Pontailler, Pierrick Priault, Franck Radnai, Michel Sartore, Kamel Soudani, Pierre Trichet, Laurent Vanbostal and Bernd Zeller. The authors also acknowledge the ‘Métabolisme-Métabolome’ platform of the IFR87 and the ‘functional ecology’ platform of the IFR 110 for IRMS analyses, and the INRA Cestas experimental unit. The authors are grateful to Michel Péan, CEA Cadarache and Christophe Robin, ENSAIA for lending us 12CO2/13CO2 infrared gas analysers. Thanks to Mark Bakker for helpful comments on the manuscript.

Ancillary