Seasonal patterns of carbon allocation to respiratory pools in 60-yr-old deciduous (Fagus sylvatica) and evergreen (Picea abies) trees assessed via whole-tree stable carbon isotope labeling


  • Daniel Kuptz,

    1. Ecophysiology of Plants, Department of Ecology and Ecosystem Management, Life Science Center Weihenstephan, Technische Universität München, Hans-Carl-von-Carlowitz-Platz 2, 85354 Freising, Germany
    Search for more papers by this author
  • Frank Fleischmann,

    1. Pathology of Woody Plants, Department of Ecology and Ecosystem Management, Life Science Center Weihenstephan, Technische Universität München, Hans-Carl-von-Carlowitz-Platz 2, 85354 Freising, Germany
    Search for more papers by this author
  • Rainer Matyssek,

    1. Ecophysiology of Plants, Department of Ecology and Ecosystem Management, Life Science Center Weihenstephan, Technische Universität München, Hans-Carl-von-Carlowitz-Platz 2, 85354 Freising, Germany
    Search for more papers by this author
  • Thorsten E. E. Grams

    1. Ecophysiology of Plants, Department of Ecology and Ecosystem Management, Life Science Center Weihenstephan, Technische Universität München, Hans-Carl-von-Carlowitz-Platz 2, 85354 Freising, Germany
    Search for more papers by this author

Author for correspondence:
Daniel Kuptz
Tel: 0049 8161 71 4674


  • The CO2 efflux of adult trees is supplied by recent photosynthates and carbon (C) stores. The extent to which these C pools contribute to growth and maintenance respiration (RG and RM, respectively) remains obscure.
  • Recent photosynthates of adult beech (Fagus sylvatica) and spruce (Picea abies) trees were labeled by exposing whole-tree canopies to 13C-depleted CO2. Label was applied three times during the year (in spring, early summer and late summer) and changes in the stable C isotope composition (δ13C) of trunk and coarse-root CO2 efflux were quantified.
  • Seasonal patterns in C translocation rate (CTR) and fractional contribution of label to CO2 efflux (FLabel-Max) were found. CTR was fastest during early summer. In beech, FLabel-Max was lowest in spring and peaked in trunks during late summer (0.6 ± 0.1, mean ± SE), whereas no trend was observed in coarse roots. No seasonal dynamics in FLabel-Max were found in spruce.
  • During spring, the RG of beech trunks was largely supplied by C stores. Recent photosynthates supplied growth in early summer and refilled C stores in late summer. In spruce, CO2 efflux was constantly supplied by a mixture of stored (c. 75%) and recent (c. 25%) C. The hypothesis that RG is exclusively supplied by recent photosynthates was rejected for both species.


Terrestrial carbon (C) fluxes are dominated by two major processes: autotrophic C uptake by primary producers and ecosystem respiration (Gifford, 2003). In Europe, forests are the most abundant terrestrial ecosystems (44% of total land mass; FAO, 2006), fixing up to 3.25 Pg C yr−1 (gross primary productivity (GPP); Schulze et al., 2010). Large amounts of the assimilated C are released back to the atmosphere by ecosystem respiration (Valentini et al., 2000; Knohl et al., 2003; Luyssaert et al., 2010), which is divided into autotrophic plant respiration (on average 53% of GPP) and heterotrophic soil respiration (on average 33% of GPP; Schulze et al., 2010).

Autotrophic respiration is usually measured as the CO2 efflux from plants to the atmosphere and may be partitioned into growth respiration (RG, related to processes such as diameter growth and cell wall thickening) and maintenance respiration (RM, related to processes such as protein and lipid turnover and maintenance of the cellular structure; Amthor, 2000). For adult trees, large seasonal variations in the total CO2 efflux and the contributions of RG and RM have been reported, with RG mainly depending on growth activity and RM on temperature (Stockfors & Linder, 1998; Ceschia et al., 2002; Kuptz et al., 2011). CO2 efflux may also be influenced by respiratory or soil CO2 transported in the xylem (Teskey et al., 2008). Despite the great importance of autotrophic respiration for the global C cycle, knowledge concerning the respiratory C supply system is still limited (Trumbore, 2006; Lehmeier et al., 2008). Large uncertainties remain regarding the origin of C in CO2 efflux, that is, regarding the age, number and magnitude of C sources (Trumbore, 2006; Carbone & Trumbore, 2007; Lehmeier et al., 2008) and the extent to which they contribute to respiration (Lötscher et al., 2004). In particular, only a few studies have addressed the importance of C stores and C turnover for higher plants (Trumbore, 2006).

Throughout the growing season, recent photosynthates are often considered the most important C source for respiration (Scartazza et al., 2004; Wertin & Teskey, 2008; Maier et al., 2009). By contrast, several recent reports on tree, herb and grassland respiration have suggested that respiratory C sources may be of variable age, that is, that respiration may also be largely supplied by storage C (Schnyder et al., 2003; Nogués et al., 2006; Carbone & Trumbore, 2007). Recent photosynthates may supply respiration directly, but they may also pass through storage pools with different turnover rates (Lehmeier et al., 2008). The contribution of these storage pools may vary among species (Keel et al., 2007) and may change with season (Nogués et al., 2006); however, the underlying dynamics are not well understood (Trumbore, 2006). In the case of herbaceous sunflower (Helianthus annuus) and alfalfa (Medicago sativa), Lötscher et al. (2004) suggested that individual C pools may be related to specific respiratory fluxes, that is, that RG is supplied exclusively by recent photosynthates, while RM is supplied by older C. Although previous studies on adult forest trees observed fast intermixing of recent photosynthates with old respiratory substrate pools (Keel et al., 2007; Kodama et al., 2008), it remains obscure whether these short- and long-term C pools are related to RG and RM.

Labeling of canopy air with stable C isotopes may reveal the contribution of recent photosynthates to CO2 efflux (Nogués et al., 2006; Lehmeier et al., 2008) and may also reveal the time lag between photosynthetic C uptake and respiratory C loss (Högberg et al., 2008; Plain et al., 2009). Time lags between C uptake and respiratory loss may serve as an indicator of changes in the source–sink relationships of plant tissues, and C transport rates are assumed to be high during times of active growth (Kozlowski, 1992; Wingate et al., 2010). In the present study, we investigated whether the results obtained by Lötscher et al. (2004) for herbaceous species are valid for adult forest trees, which differ greatly from herbaceous species because of their perennial life-cycle, long C transport pathways and high storage capacity. The present study focused on trunk and coarse-root CO2 efflux, neglecting other large respiratory fluxes, that is, CO2 efflux from leaves, fine roots and the soil. We formulated the hypothesis that the RG of the trunk and coarse roots is directly and exclusively supplied by recent photosynthates. To test this hypothesis, adult individuals of one deciduous and one evergreen tree species, that is, of European beech (Fagus sylvatica) and Norway spruce (Picea abies), respectively, were labeled with stable carbon isotopes three times per year during periods of contrasting contributions of RG to CO2 efflux. In view of our hypothesis, the experimental set-up was designed for labeling mainly rapidly turned-over C pools to detect the direct contribution of current photosynthates to trunk and coarse-root respiration.

Materials and Methods

Research site and meteorological data

The study was conducted at the forest research site Kranzberger Forst (48°25′08″N, 11°39′41″E; 485 m asl; Freising, Germany). The study species were European beech (Fagus sylvatica L.), which forms dense groups of 50–60 individuals, and Norway spruce (Picea abies Karst), which grows in large parts of the area. Trees were 60 to 70 yr old. For further details of the study site, see Pretzsch et al. (1998). Long-term means (1970–2000) of annual air temperature and precipitation were 7.8°C and 786 mm, respectively (Matyssek et al., 2007). Tree canopies were accessible by scaffolding (17–25 m above the ground) and canopy crane.

Air temperature (TAir), air pressure and absolute air humidity were continuously monitored at c. 2 m above the canopy. Precipitation was assessed at a forest climate station of the Bayerische Landesanstalt für Wald und Forstwirtschaft (LWF) (Freising, Germany), c. 1 km from the research site.

Stable C isotope labeling

In 2008 and 2009, a series of six free-air 13CO2/12CO2 labeling experiments were performed on adult individuals of beech and spruce. Four trees per species were labeled, while three unlabeled trees per species served as controls. Beech trees were 25.9 ± 0.3 m high, and spruce trees 28.8 ± 0.8 m (mean ± SE). The mean (± SE) diameter at breast height (1.3 m above the ground) was 24.6 ± 3.5 and 33.7 ± 2.4 cm, respectively. The depth of the canopies was c. 5–6 m. The stable C isotope composition of canopy air (δ13CAir) surrounding each labeled tree was altered by the release of pure CO213CLabelc.−47‰) into the atmosphere, tolerating a [CO2] increase of c. 100 μmol mol−1 (see also Table 1). Labeling was performed using the isoFACE approach, which allows continuous, free-air 13CO2/12CO2 labeling of adult tree canopies (Grams et al., 2011). Briefly, tree canopies were equipped with vertically hanging, micro-perforated PVC tubes for homogeneous CO2 release. For beech trees, tubes were suspended from the stationary scaffolding at the research site. In the case of spruce, tubes were suspended from a low-weight, custom-built carrier structure attached to each labeled tree (see Fig. 1). CO2 used for labeling was stored at the site in a tank with a capacity of 1100 kg. Typically 75–100 kg CO2 was deployed per tree, per day (Grams et al., 2011).

Table 1.   Shifts in stable carbon (C) isotope composition (δ13C) of canopy air during labeling
SpeciesExperimentIncrease in [CO2]Time fraction within 10% target [CO2]δ13CLabelδ13CAir -unlabeledShift in δ13CAir during labeling
  1. Data shown are the increase in [CO2] during labeling (μmol mol−1) and stable C isotope composition of label CO213CLabel) and of canopy air (δ13CAir) (mean ± SE).

BeechSpring110.4 ± 14.30.52−46.3 ± 0.5−7.6 ± 0.3−8.4 ± 0.9
Early summer113.2 ± 13.20.64−46.3 ± 0.8−8.6 ± 0.1−8.3 ± 0.7
Late summer146.9 ± 11.50.54−36.9 ± 0.4−8.6 ± 0.2−7.4 ± 0.5
SpruceSpring94.2 ± 9.30.65−35.2 ± 0.2−8.2 ± 0.2−5.0 ± 0.5
Early summer117.9 ± 15.20.46−45.2 ± 0.6−8.8 ± 0.3−8.2 ± 0.9
Late summer109.7 ± 9.90.58−44.6 ± 0.6−8.7 ± 0.1−8.5 ± 0.6
Figure 1.

(a) Flow-through system for the assessment of the CO2 efflux of upper and lower trunks and coarse roots, as well as the sampling of trunk chamber air for determination of δ13C. (b) Experimental set-up of the isoFACE 13CO2/12CO2 labeling infrastructure with microporous CO2-exposure tubes hanging through the canopy of adult spruce trees suspended from low-weight custom-built carrier structures.

Tree canopies were labeled throughout the daylight hours for 5 consecutive days. This design was chosen to label rapidly turned-over C pools (e.g. transport saccharose) and to allow re-labeling of the same trees during one growing season. [CO2] was monitored at two heights per tree (1 and 3 m below the upper canopy edge) and above canopies. Membrane pumps continuously transported canopy air via PVC tubes to an infrared gas analyzer (IRGA) (Binos 4b; Emerson Process Management, Weißling, Germany), interchanging between sampling positions every 3 min. The average [CO2] of all eight sampling positions was constantly measured with a second IRGA, and mass flow of label into canopies was adjusted accordingly. Canopy air at each sample position was collected once per day between 09:00 and 10:00 h for the analysis of δ13CAir. To this end, excess air from membrane pump outlets was flushed through 12-ml Exetainer® vials (Labco Ltd, High Wycombe, UK) using a 100-ml syringe. Gas samples were analyzed for δ13C by isotope-ratio mass spectrometry (IRMS) (GVI-Isoprime; Elementar, Hanau, Germany). The δ13C of labeled (δ13CAir-labeled) and unlabeled (δ13CAir-unlabeled) canopy air was calculated using canopy [CO2] and the linear relationship between 1/[CO2] and δ13CAir of samples taken on days 1–5 and on days 0 and 6, respectively.

For each species, three labeling experiments were conducted on the same trees: in spring toward the end of leaf development, in early summer at times of highest trunk growth respiration rates (RG), and in late summer under declining RG. Labeling started on 24 May, 9 July and 22 August 2008 in beech, and on 23 June and 9 September 2008 and 20 April 2009 in spruce.

Trunk and coarse-root CO2 efflux

Trunk and coarse-root CO2 efflux was assessed from day 0 through at least day 9 of the labeling experiment. Plexiglas® (Röhm GmbH, Darmstadt, Germany) chambers were laterally attached at two trunk heights (upper trunk: 9.5 ± 1.4 m; lower trunk: 0.7 ± 0.0 m), enclosing a trunk surface area of 203–534 cm2. One coarse root per tree was completely enclosed in a cylindrical chamber (coarse-root diameter 1.5 ± 0.1 cm). Throughout the measurements, all chambers were darkened with aluminized polyester foil. Each chamber was equipped with a temperature sensor (negative temperature coefficient, Pt 100, SEMI 833 ET; Hygrotec®; Messtechnik GmbH, Titisee, Germany) inserted c. 0.5 cm into the bark to measure bark temperature (TBark).

CO2 efflux was assessed using a computer-controlled flow-through gas-exchange system (Grams et al., 2011; Kuptz et al., 2011). A stable reference gas (Fig. 1; 406.6 ± 0.9 μmol CO2 mol−1; δ13C = −3.06 ± 0.02‰; mean ± SE) was continuously pumped via PVC tubes through the trunk and coarse-root chambers at a rate of approx. 0.5 l min−1. Outgoing chamber air was analyzed by an IRGA. Each chamber was sampled at c. 3-h intervals. Trunk CO2 efflux was expressed per unit volume (m3) of living tissue behind the chamber (sapwood plus bark; Maunoury et al., 2007). Sapwood depth was assessed by taking core samples next to the chambers. Coarse-root CO2 efflux was related to the total root volume enclosed in chambers (Marsden et al., 2008).

Growth respiration (RG) was calculated for each sampling position using the mature-tissue method (Amthor, 1989; Kuptz et al., 2011). Accordingly, respiration rate (R) relates to bark temperature as given by Eqn 1, assuming R to equal CO2 efflux:

image(Eqn 1)

R15 represents the CO2 efflux at 15°C and Q10 the temperature sensitivity of the CO2 efflux. For each sampling position, the Q10 and R15 of maintenance respiration (RM) were determined from measurements of CO2 efflux and TBark during times without growth, that is, from November 2008 to February 2009 (see Kuptz et al., 2011). Mean coefficients of determination (r²) of winter correlations were 0.75 ± 0.02 and 0.72 ± 0.03 for beech and spruce, respectively. This approach assumes that (i) during times without growth, CO2 efflux relates exclusively to RM, and (ii) the temperature dependence of RM is constant throughout the year (Amthor, 2000). RM was calculated using winter values for Q10 and R15 and actual TBark during the experiments. RG was assessed as the difference between CO2 efflux and RM. The concentration of CO2 dissolved in the xylem sap may influence CO2 efflux (Teskey et al., 2008). However, during the measurement campaigns of this study, such an effect was not observed (Kuptz et al., 2011; see also Discussion).

CO2 sampling for stable carbon isotope analysis

To analyze the stable C isotope composition (δ13C) of CO2 efflux (δ13CE), chamber air and reference gas air were collected by means of an automated gas sampler (Gilson 221 XL; Gilson Inc., Middleton, WI, USA), flushing air vented from the IRGA of the flow-through gas-exchange system (Fig. 1) through 12-ml glass vials (Exetainer®; Labco Ltd). Trunk δ13CE was calculated using δ13C and the [CO2] of chamber and reference air by applying a two-end-member mixing model (Grams et al., 2011; Kuptz et al., 2011). The sampling interval was c. 6 h for each chamber.

The δ13CE released from coarse roots was analyzed once per day (between 09:00 and 11:00 h) by means of a custom-built closed-respiration system (Grams et al., 2011; Kuptz et al., 2011). Briefly, six sample vials were continuously flushed and isolated from the system at 10-min intervals. Coarse-root δ13CE was calculated using the ‘Keeling plot approach’ (Keeling, 1958; Zobitz et al., 2008), rejecting Keeling plots with coefficients of determination (r²) < 0.97.

The fraction of labeled C in CO2 efflux (fLabel; see Lehmeier et al., 2008; Gamnitzer et al., 2009) was calculated for each sample position as:

image(Eqn 2)

13CSample, the C isotope composition of the CO2 efflux; δ13COld, the mean δ13CE for days 0 and 1.) δ13CNew (i.e. fully labeled δ13CE) represents the CO2 efflux at the new isotopic equilibrium. However, as the duration of label application was too short to allow an isotopic steady state to be reached, δ13CNew was estimated following Gamnitzer et al. (2009) and Schnyder et al. (2003). Accordingly, δ13CNew was estimated for each sampling position and experiment individually by first calculating the mean apparent 13C discrimination (Δ13CA–E) between unlabeled canopy air (δ13CAir-Unlabeled) and δ13CE on days 0 and 1:

image(Eqn 3)

Then δ13CNew was calculated as:

image(Eqn 4)

Both δ13COld and δ13CNew are constant values representing CO2 efflux at time 0 and after 100% label incorporation, respectively, presuming that any change in δ13CE was induced by the labeling treatment only. In field experiments, however, changes in meteorological conditions may alter photosynthetic discrimination. Such day-by-day variability was corrected for on the basis of daily changes observed in δ13CE of unlabeled control trees. In a few cases, calculation of fLabel resulted in negative values. As interpretation of negative fLabel values makes no sense, these values were set to 0. To account for remaining variability, maximum levels of fLabel in CO2 efflux (FLabel-Max) were determined for each sample position and experiment by fitting fLabel through a sigmoid regression:

image(Eqn 5)

with the fit parameter b giving the slope coefficient of the regression, t the time of measurement and t0 the instant of the regression inflection point.

Carbon translocation rates (CTR; m h−1) were assessed both in labeled and in control trees. For control trees, we first calculated time lags between photosynthetic C fixation and respiratory loss. To this end, Pearson’s correlations were determined for each tree and experiment between time-series of TAir and trunk chamber δ13CE. Therefore, data sets of TAir were time-shifted against δ13CE in intervals of 0.125 d covering a time span of 0 to 9 d. Time lags derived directly from the time delay of the correlation with the highest correlation coefficient (r) (Kuzyakov & Gavrichkova, 2010). Subsequently, CTR was calculated as:

image(Eqn 6)

(tl, the time lag (h); hC, the mean height of the tree canopy (m); hTC, the height of the trunk chamber (m).) For labeled trees, CTR was calculated in a similar way. First, time lags between the fLabel values of upper and lower trunks were assessed as for TAir and δ13CE of control trees. Then, CTRs were calculated following Eqn 6 using the distance between the upper and lower trunk chambers.

Sampling of phloem sugars

On days 0 and 5 of labeling, bark and phloem tissue material (∅ 5 mm, = 3 for each tree) were collected between 10:00 and 11:00 h next to the lower trunk chambers. For extraction of phloem sap, samples were completely soaked for 5 h at 4°C in 15 mM sodium polyphosphate buffer (Sigma-Aldrich, Munich, Germany; see Gessler et al., 2004). Subsequently, the solution was centrifuged and phloem extract was analyzed for water-soluble sugars (i.e. sucrose, fructose, glucose, raffinose and stachyose) by means of high-pressure liquid chromatography (HPLC) (CARBOsep CHO-820 calcium column; Transgenomic, Glasgow, UK; for further details, see Fleischmann et al., 2009). Sucrose was partly broken down to glucose and fructose as a result of invertase activity in the sample solution. Hence, monosaccharides and disaccharides (i.e. sucrose, glucose and fructose) were integrated. Stable C isotope composition (δ13CP) was assessed by IRMS attached to an element analyzer (EA3000; Euro Vector, Milan, Italy).

Statistical analyses

All statistical analyses were performed with PASW Statistics 18.0 (SPSS GmbH Software, Munich, Germany) and Sigma Plot 9.0 (Systat Software GmbH, Erkrath, Germany). Tests applied were Student’s t-test, repeated measures analysis of variance and Pearson’s correlation, as well as linear and sigmoid regression analysis.


Weather conditions during labeling

The mean (± SE) TAir during beech experiments was rather constant, at 16.5 ± 0.1, 15.8 ± 0.1 and 16.4 ± 0.1°C in spring, early summer and late summer, respectively (Fig. 2). In the case of spruce, the mean (± SE) TAir was 11.7 ± 0.1, 19.1 ± 0.1 and 9.9 ± 0.1°C, respectively, with significantly higher values during early summer ( 0.001; Fig. 2). Short rainfall events (up to 13 mm d−1) occurred during the evenings of the labeling experiments in early summer (Fig. 2). Variation in vapor pressure deficit (VPD) followed the diurnal pattern in TAir (data not shown).

Figure 2.

Weather conditions for adult beech (a–c) and spruce (d–f) trees during the three labeling experiments. Air temperature (TAir) is represented by the solid line, and rainfall by gray columns. Gray areas give times of label application (days 1–5).

In both species, mean CO2 efflux and RG of trunk and coarse roots were highest during early summer and significantly lower in spring and late summer ( 0.001; Fig. 3). Early summer experiments for both species were conducted near times when trunk CO2 efflux was maximal (Fig. 3; see also Kuptz et al., 2011). CO2 efflux was positively correlated with TBark during the preceding 3 to 9 h, with Pearson’s correlation coefficients (r) ranging between 0.78 and 0.99 (P  0.01).

Figure 3.

Average rates of CO2 efflux of upper (a, b) and lower (c, d) trunks and coarse roots (e, f) of adult beech and spruce trees during the three labeling experiments (spring, early summer and late summer). The rate of CO2 efflux is represented by total bar height, and estimated maintenance respiration (RM) by white bars. Data shown are the mean (± SE) of all data recorded for labeled trees (= 4) during the experiments. Diurnal variation is given by the standard error of means over all data per experiment, incorporating propagation of error between tree individuals. Lowercase letters a–c and d–f next to bars indicate significant differences ( 0.05) between seasons in CO2 efflux rate and RM, respectively.

Stable C isotope label application

During label application, the increase in [CO2] in canopies was 123.5 ± 8.0 and 107.3 ± 6.8 μmol mol−1 for beech and spruce (mean ± SE), respectively, on average over all experiments. [CO2] stayed, for more than half of the exposure time, within a ± 10% range of the target concentration of ambient [CO2] + 100 μmol mol−1 (Table 1). In the spring experiment in spruce and the late summer experiment in beech, CO2 with a δ13CLabel of c.−36‰ was used (Table 1). For the spruce experiment in spring, this resulted in a reduction of δ13CAir of 5.0‰. For all other experiment, the mean reduction in δ13CAir during labeling was 8.2 ± 0.7‰.

Label recovery in δ13CE and δ13CP

Over all experiments, the mean natural δ13CE of beech control trees was −27.6 ± 0.1, −28.3 ± 0.1 and −27.5 ± 0.2‰ in upper trunks, lower trunks and coarse roots, respectively, while the δ13CE of spruce controls was −27.3 ± 0.1, −27.1 ± 0.1 and −25.8 ± 0.3‰ (mean ± SE), respectively. Control tree δ13CE displayed distinct day-by-day variations of up to 3.4 and 3.3‰ in beech and spruce, respectively (Fig. 4). In general, the δ13CE of labeled trees was not significantly different from that of the control trees at day 0. However, in one experiment, during early summer in beech, coarse-root δ13CE of label trees was significantly depleted in 13C compared with control trees at day 0. In most cases, the δ13CE of labeled trees declined 2 to 4 d after the reduction of δ13C in canopy air (Fig. 4). In spring, this decline was not observed for beech lower trunks and, with the exception of one data point, for coarse roots. Although the δ13CE of coarse roots in controls was initially somewhat higher than in the labeled trees during summer, the latter displayed the declining pattern 7 to 9 d after the start of label application. Compared with day 0 and control trees, trunk δ13CE of labeled beech trees declined most strongly during late summer (c. 4.0‰). In spruce, the decline in δ13CE of labeled trees was consistently most pronounced during early summer (−2.1, −1.7 and −2.2 in upper trunks, lower trunks and coarse roots, respectively). Conversely, hardly any change in response to label application was observed during spring and late summer. Consistently, repeated measures analysis of variance revealed a significant influence of the labeling treatment (treatment × day of the experiment) in trunks of beech during early and late summer and, in spruce, in trunks and coarse roots during early summer (Fig. 4).

Figure 4.

Stable C isotopic composition of CO2 efflux (δ13CE) during the labeling experiments in beech (a–i) and spruce (j–r) sampled from upper (circles) and lower (triangles) trunks and coarse roots (squares). Data shown are daily mean (± SE). Closed symbols represent labeled trees, and open symbols control trees. Significant influences of labeling, calculated by repeated measures analysis of variance (treatment × day of the experiment; repetition starting on day 1), are given as P-values in (a)–(r). In addition, *, ** and *** denote daily significant differences between labeled and control trees of  0.05,  0.01 and  0.001, respectively. Gray areas indicate times of label application (days 1–5).

The total concentration of phloem sugars in beech (sum of sucrose, fructose, glucose, raffinose and stachyose) did not vary seasonally (13.5 ± 0.9, 12.4 ± 0.7 and 12.1 ± 0.9 mg g−1 dry matter; mean ± SE). Conversely, in spruce, significantly higher sugar concentrations, in particular of oligosaccharides, were recorded in spring compared with summer (94.7 ± 5.5, 20.7 ± 2.0 and 14.1 ± 0.2 in spring, early summer and late summer, respectively; mean ± SE;  0.001; Fig. 5). During the early summer experiment in spruce, δ13CP was c. 2‰ lower than the corresponding δ13CE at the lower trunk position. In all other experiments, linear regressions between δ13CE and δ13CP resembled the 1 : 1 relationship (Fig. 5).

Figure 5.

Concentration of phloem sugars (white, sum of sucrose, fructose and glucose; gray, sum of other sugars, mainly raffinose and stachyose) in mg g−1 dry matter of beech (a) and spruce (e) and linear regressions between the stable C isotope composition of phloem sugars (δ13CP) and lower trunk CO2 efflux (δ13CE) during spring (b, f), early summer (c, g) and late summer (d, h) in both labeled and control trees. The continuous line shows the regression, and the dashed line the 1 : 1 relationship. Significance levels at  0.05 and  0.001 are represented as * and ***, respectively.

In beech, the fraction of labeled C (fLabel) in trunk or coarse-root CO2 efflux was generally small during spring and consistently increased mainly at the upper and to some extent also at the lower trunk position (up to 0.23; Fig. 6a). The maximum fLabel was recorded in trunks during late summer (upper trunks: 0.6 ± 0.2; lower trunk: 0.5 ± 0.1; Fig. 6c,f), whereas in coarse roots, no clear trend was observed. During the early summer experiment, fLabel peaked first in the upper trunks, followed by the lower trunks and the coarse roots. In trunks, fLabel declined 2 to 3 d after the end of the label application, which was not the case in late summer, when fLabel stayed at higher values for at least for 4 d after the end of label application. Similarly, fLabel peaked in coarse roots at day 9, but did not reach steady state during the late summer experiment. In spruce, pronounced changes in fLabel were observed only during the early summer experiment and, in the case of upper trunks, also during late summer (Fig. 6l). Spruce fLabel did not decline within a couple of days after the end of label application during early summer as observed in beech. Similar to the pattern in beech, fLabel peaked first in upper then in lower trunks and finally in coarse roots (Fig. 6).

Figure 6.

Fraction of labeled carbon (fLabel) in CO2 efflux of upper (circles) and lower (triangles) trunks and coarse roots (squares) during labeling experiments in beech (a–i) and spruce (j–r). Data shown are the daily mean (± SE). Gray areas indicate times of label application (days 1–5).

In beech, fitted maxima of fLabel (FLabel-Max) of upper and lower trunks were low during spring (0.1 ± 0.0 and 0.2 ± 0.1, respectively; mean ± SE) and highest during late summer (c. 0.6; see Fig. 7). In beech coarse roots, FLabel-Max was always below 0.25. In spruce, no significant difference in FLabel-Max was observed throughout the year and all values remained around c. 0.25. In contrast to FLabel-Max, seasonal dynamics in the contribution of RG to CO2 efflux were rather uniform in both species, and the contribution of RG was always maximal during early summer (beech: 0.4–0.6; spruce: 0.5–0.7; Fig. 7). The high contribution of RG was paralleled by high CO2 efflux rates (see bar width in Fig. 7c,d).

Figure 7.

Mean fractional contribution of labeled C to CO2 efflux (FLabel-Max) of beech (a) and spruce (b) trees and the quotient of the rate of growth respiration (RG) to CO2 efflux for beech (c) and spruce (d) during the labeling experiments in spring (white bars), early summer (hatched bars) and late summer (cross-hatched bars). The bar width in (c) and (d) represents the relative CO2 efflux rate of organs between experiments. Lowercase letters indicate significant differences at the same sampling position between experiments with  0.05.

Carbon translocation rates

CTRs in trunks of labeled trees were similar to rates determined in control trees (Fig. 8). Rates ranged between 0.15 and 0.62 m h−1 and between 0.06 and 0.46 m h−1 in beech and spruce, respectively. Species displayed distinct seasonal differences, with the highest CTR during early summer compared with spring and late summer. However, this was only significant in control trees ( 0.05). At each time-point, spruce transport rates were 50% or less than those in beech.

Figure 8.

Carbon translocation rates (CTRs) of control and labeled trees in the three labeling experiments in beech (closed symbols) and spruce (open symbols). Data shown are mean (± SE).


Whole canopies of adult beech and spruce trees were labeled with stable C isotopes during spring, early summer and late summer; that is, at the end of leaf expansion, at maximum rates of trunk growth respiration (RG) and at declining RG, respectively. Distinct seasonal dynamics in the allocation of recent photosynthates to woody tissue CO2 efflux were recorded at two trunk and one coarse-root position, in terms of both the magnitude and the time of label appearance.

Stable C isotope labeling was performed using a free-air approach (isoFACE; for details, see Grams et al., 2011) that applies CO2 directly into the tree canopies via vertically hanging tubes. The Kranzberger Forst study site is characterized by low wind velocities, with above-canopy annual means of < 1.7 m s−1 (Heerdt, 2007). Direct exposure of 13C-depleted CO2 within the canopy resulted in high labeling efficacy even at times with low wind speed (Table 1, Grams et al., 2011). An increase in [CO2] by 100 μmol mol−1 did not affect the stomatal conductance or C translocation rate of trees and only marginally increased photosynthetic discrimination against 13C (Fig. 8, Grams et al., 2011). With the exception of slightly decreased δ13CE of beech coarse roots during the summer experiments, day 0 sampling confirmed no previously applied label remained in respiratory pools of trees at the start of the following labeling experiment.

In both species, CO2 efflux and RG displayed distinct seasonal patterns in trunks and coarse roots (Fig. 3). Both parameters and the contribution of RG to CO2 efflux (Fig. 7) were highest during early summer, suggesting high activity of growth processes such as diameter growth and lignin synthesis (Stockfors & Linder, 1998). In beech, the second highest levels were found in spring (lower trunks) and late summer (coarse roots; Fig. 7), indicating that trunk growth had already started in spring (Maunoury et al., 2007), followed by root growth late in the growing season (Smith & Paul, 1988; Högberg et al., 2010). In spruce, the second highest contributions of RG to CO2 efflux were consistently found during late summer, indicating high activity of growth processes late in the season (Kuptz et al., 2011). Overall, respiration was 3- to 10-fold higher in coarse roots compared with trunks (Fig. 3), probably as a result of higher abundances of respiring cells per volume (Marsden et al., 2008), as the ratio of living cambium : total organ volume increases considerably when organ diameter declines.

As recently discussed in the literature, CO2 dissolved in the xylem sap might influence woody tissue CO2 efflux (Teskey et al., 2008). In our experiments, regression slopes between δ13CE and δ13CP were close to the slope of the 1 : 1 relationship (Fig. 5), indicating that a shift in δ13CP results in the same shift in δ13CE without a perceivable contribution of sap flow dissolved CO2. We therefore conclude that sap flow had no or only a small effect on δ13CE during our experiments, which is consistent with recent findings (Ubierna et al., 2009; Grams et al., 2011; Kuptz et al., 2011). During early summer, the δ13CE of spruce was nearly 2‰ enriched in 13C compared with δ13CP (Fig. 5). A similar trend was seen during the early summer experiment in beech, but this was not as pronounced as in spruce (Fig. 5). This offset may be related to intensive lignin synthesis. As lignin is known to be depleted in 13C compared with most respiratory substrates, CO2 respired during its synthesis might be enriched (Hobbie & Werner, 2004; Bowling et al., 2008). Alternatively, this offset might be explained by a higher dependence on 13C-enriched starch as a respiratory substrate during times of high C demand. This idea is based on diurnal observations of sugar–starch interconversion at the leaf level, whereby starch is accumulated during the light as a result of an excess of sugars, but is metabolized during the night (Bowling et al., 2008; Gessler et al., 2008). Similarly, on the seasonal scale, starch storage pools might be built up or metabolized depending on the excess of, or demand for, phloem sugars (Maunoury et al., 2007; Kuptz et al., 2011).

Distinct seasonal dynamics in fLabel and FLabel-Max were observed in beech, although meteorological conditions were similar between experiments (Fig. 2). In spring, only small amounts of label were found at all three sampling positions, although high rates of RG indicated active growth in lower trunks during this time. The highest amounts of label were recovered in late summer, when RG was already declining (Fig. 7). The observed seasonal patterns may indicate the involvement of two distinct C pools in beech trunks, a transfer and a storage pool. We developed a conceptual two-pool model (Fig. 9) that orientated at the findings by Lötscher et al. (2004). Using this model, we investigated whether their results for herbaceous species may be valid for adult forest trees; that is, whether the transfer and the storage pools might be related to RG and RM, respectively.

Figure 9.

Conceptual two-pool model (one transfer and one storage carbon (C) pool) of the substrate supply for lower trunk CO2 efflux of adult beech trees during spring, early summer and late summer. RG and RM are growth and maintenance respiration, respectively. The assumption that RG and RM are supplied by the transfer and storage pools, respectively, was derived from Lötscher et al. (2004). Open arrows denote fluxes of respired CO2, and closed arrows fluxes of respiratory substrate. TBCA, total below-ground carbon allocation. During spring, negligible amounts of new photosynthates enter the transfer pool and both RG and RM are predominately supplied by stored C. During early summer, new photosynthates enter the transfer pool, directly supplying RG and being further transported to roots and soils (i.e. TBCA). During late summer, new photosynthates enter both the transfer and storage pools, supplying both RG and RM. Changes in the box size of the storage pool relate to changes in sapwood sugar concentrations observed by Barbaroux & Breda (2002). The size of open arrows represents the mean rate of lower trunk RG and RM during labeling experiments in beech in comparison to the annual maxima of mean daily CO2 efflux recorded from May 2008 to June 2009.

In spring, the supply of recent photosynthates to the trunk and coarse-root CO2 efflux of beech was low (Fig. 7). At this time, foliage and shoot development may be the strongest sinks for recent photosynthates (Hansen & Beck, 1994; Dickson et al., 2000) and both RG and RM of trunks appear to be supplied by C stores accumulated during previous years (Fig. 9; Helle & Schleser, 2004; Kagawa et al., 2006; Skomarkova et al., 2006). In early summer, C allocation to beech trunks increased with the demand for growth, but the contribution of RG to CO2 efflux still exceeded the contribution of labeled C (Fig. 7, Mordacq et al., 1986; Hansen & Beck, 1994). During this period, fLabel in trunks declined shortly after the end of label application, suggesting high turnover of recent photosynthates in C pools involved in trunk respiration (Fig. 6). Thus, during this period, labeled C may not have entered the storage pool, which typically has higher half-lives (Lehmeier et al., 2008), and was either respired directly (i.e. to supply growth) or allocated to below-ground sinks (Fig. 7, Hansen & Beck, 1990). By contrast, during late summer, fLabel of trunks reached its highest values and did not decline within a couple of days after the end of the label application. At this time, RM was at least partly supplied by labeled C, as 60% of the CO2 efflux was labeled, while RG accounted for only 0.25% (Fig. 7). The longer label retention time, however, suggested slower label turnover and cycling of labeled C through the storage pool (Figs 6, 9). Thus, our results suggest that, during late summer, labeled C may also be incorporated into the storage pool, as the demand for growth was already declining (Fig. 9).

During late summer, trunks are known to be strong C sinks as a result of the accumulation of C for winter storage and frost hardiness (Hansen & Beck, 1990; Skomarkova et al., 2006). FLabel-Max peaked during late summer in trunks. As a consequence, the size of the storage pool is assumed to be shrinking during spring and early summer and to be refilled during late summer. Accordingly, Barbaroux & Breda (2002) reported a distinct decline in sugar concentration in the youngest 10 tree rings of beech between May and July, followed by an increase later in the season. Assuming the same decline in sapwood sugar concentration (c. 5 mg g−1 dry mass) in our study trees, this would result in an estimated C loss of 3.0 ± 0.3 g m−3 sapwood. This is about half of the C released by RM from May to July, indicating that RM can easily be supplied by stored C throughout spring and early summer.

Overall, during spring, both RG and RM of beech trunks were largely supplied by C stores. Our hypothesis that RG is exclusively supplied by recent photosynthates was rejected for both spring and early summer. Consequently, the findings by Lötscher et al. (2004) regarding the RG of herbaceous species were not applicable to adult beech trees. The finding of Lötscher et al. (2004) that RM was exclusively supplied by storage C could not be invalidated for beech during spring and early summer, but was not supported during late summer, as RM was at least partly supplied by recent photosynthates. However, as our results indicated large amounts of photosynthates refilling the storage pool during late summer, the findings of Lötscher et al. (2004) regarding RM could not be rejected for this time either, and remain possible in adult deciduous beech in all three experiments.

Despite significant seasonal differences in meteorological conditions during the experiments, FLabel-Max in spruce hardly changed throughout the year (Fig. 7). As an evergreen species, spruce displayed less dependence on seasonal accumulation and consumption of storage C, which may be refilled by periods of active photosynthesis during mild winter days (Hansen et al., 1996; Hu et al., 2010; Kuptz et al., 2011). Overall, smaller amounts of label were recovered in spruce compared with beech and fLabel did not decrease after the end of label application. Both observations indicate that recent photosynthates consistently mix with large amounts of old respiratory substrate before they are respired (Nogués et al., 2006). Alternatively, lower label abundances in spruce might be explained by higher maintenance costs of the evergreen foliage compared with that of deciduous beech (Sobrado, 1991). Consequently, lower amounts of recent photosynthates might be transported to trunks and coarse roots.

According to recent results on juvenile trees (W. Ritter, Ecophysiology of Plants, Technische Universität München, pers. comm.), the respiratory supply system of spruce might be represented best by a one-pool model. Gymnosperm trunks usually possess lower amounts of parenchyma cells in trunks than angiosperms (Pallardy & Kozlowski, 2008). In spruce, wood parenchyma sums up to only 4.5–7% of total woody tissue (Schmidt-Vogt, 1986), whereas in beech, ray parenchyma alone comprises c. 20% of total woody tissue (Burgert & Eckstein, 2001). Moreover, in spruce trunks, the total amount of nonstructural carbohydrates strongly decreases from the bark towards the sapwood (Egger et al., 1996; Hoch et al., 2003), whereas in beech trunks, carbohydrates are distributed rather equally throughout the trunk (Barbaroux & Breda, 2002; Hoch et al., 2003). Hence, the storage capacity of spruce trunks may be much smaller than that of beech. This is supported by the kinetics observed in fLabel (Fig. 6), but also by spruce phloem sugar concentrations. In particular, oligosaccharides such as raffinose and stachyose were high (Fig. 5) during spring, while during summer, phloem sugars consisted mainly of disaccharides (sucrose). We assumed that, in spruce, raffinose and stachyose were mainly accumulated as storage C (Fischer & Höll, 1992; Hoch et al., 2003). Such a trend was not observed for beech, as storage C might accumulate within trunks rather than within the phloem. However, we note that FLabel-Max in spruce did not differ between experiments. Thus, the high amount of sugars might not derive from pre-labeling periods only, but substrate might also accumulate in trunks during spring shortly before the onset of growth, as previously observed for starch (Höll, 2000). Nonetheless, it is unlikely that RG in spruce was directly and exclusively supplied by recent photosynthates, as most of the time the contribution of RG to CO2 efflux exceeded FLabel-Max (Fig. 7). Thus, our hypothesis regarding the supply for RG was also rejected for spruce. Conversely, the idea that RM is supplied exclusively by storage C could not be rejected on the basis of our results.

Our interpretation of C allocation of beech and spruce is supported by the transport rates of phloem sugars. Expanding previous “natural abundance approach” studies depending on changes of photosynthetic discrimination by meteorological conditions (Keitel et al., 2003; Werner et al., 2006), we used the isoFACE labeling approach to determine the CTR directly and independently of the weather. Overall, CTRs of 0.06 to 0.62 m h−1 were consistent with previous studies (Keitel et al., 2003; Mencuccini & Hölttä, 2010; Wingate et al., 2010). CTR calculated from time lags between the natural variation in δ13CE and TAir gave similar transport rates to CTR calculated from labeling, supporting the natural abundance approach. C transport in spruce was consistently slower compared with that in beech, which may reflect anatomical differences; for example, sieve cells in gymnosperms vs highly specialized sieve tube systems in angiosperms (Schulz & Thompson, 2001; Wingate et al., 2010). In both species, CTR was highest during early summer (Fig. 8). Phloem sap velocity may be sensitive to temperature (Johnsen et al., 2007), but this cannot explain the seasonal pattern in beech, because TAir was rather similar in the three beech experiments. The seasonal dynamics may rather relate to enhanced C sink strength of the growing plant tissue in early summer (Kozlowski, 1992; Wingate et al., 2010).


Overall, observed seasonal dynamics in the C supply of respiratory processes in the trunk and coarse roots of adult trees suggested marked differences in C turnover between deciduous and evergreen species. The respiratory C supply system of beech was interpreted in relation to two C pools and that of spruce in relation to one C pool, corresponding to their contrasting wood anatomy. C transport rates calculated by the ‘natural abundance’ and ‘labeling’ approaches provided similar results. Our hypothesis that RG was exclusively and directly supplied by recent photosynthates was not supported for either species. We conclude that the contribution of recent photosynthates and C stores to respiration is highly variable in deciduous beech, as recent photosynthates were turned over more rapidly during early summer and more slowly in late summer. Evergreen spruce, by contrast, is deemed to be less dependent on seasonal accumulation and consumption of stored C. Buffering of respiratory C supply by older C, however, was evident throughout the year.

As proposed by Gaudinski et al. (2009), seasonal variations in C turnover and in C stores are important parameters in the modeling of C dynamics in plants. However, such parameters are often not taken into account, as detailed information is lacking. In this respect, the present study expands our overall knowledge of the C turnover of deciduous and evergreen trees, and its findings may be of value in the development of model approaches for autotrophic respiration at the tree and forest ecosystem scales.


We thank Thomas Feuerbach, Peter Kuba, Sepp Heckmair, Hans Lohner, Ilse Süß, Johanna Lebherz and Christof Seidler for their technical assistance, both in the laboratory and in the field. We gratefully acknowledge Dr Karl-Heinz Häberle, Timo Gebhardt, Ursula Metzger, Dr Michael Leuchner, Nick Hoffmann and the ‘Bayerische Landesanstalt für Wald und Forstwirtschaft’ for providing data or information on the research site. We thank Wilma Ritter and Dr Christoph Lehmeier for their help in interpreting our data. The authors gratefully acknowledge the support of the Faculty Graduate Center Weihenstephan of TUM Graduate School at Technische Universität München, Germany. This study was funded by the German Research Foundation (DFG) (Bonn, Germany) through SFB 607 ‘Growth and parasite defense – competition for resources in economic plants from agronomy and forestry’.