Introduction

The feedback between processes in terrestrial ecosystems and the atmosphere is one of the largest uncertainties in understanding the global C cycling (IPCC 2007). CO2 fixation (photosynthesis) and release (heterotrophic and autotrophic respiration) fluxes of terrestrial ecosystems modify global atmospheric CO2 concentrations and are in turn strongly influenced by climatic and soil conditions (e.g. Saxe et al. 1998; Nabuurs et al. 2003). Net ecosystem CO2 exchange (NEE) alone, which is now routinely determined by eddy covariance analysis in numerous ecosystems worldwide (http://daac.ornl.gov/FLUXNET/), does not provide direct insights into differential effects of climatic and weather variations on photosynthesis and respiration fluxes. Any mechanistic understanding that allows the reliable prediction of changes in the CO2 fluxes of terrestrial ecosystems under a changing climate will depend on understanding the variation in these component fluxes.

At the ecosystem level, the relative difference in C stable isotope ratios [δ13C; expressed as deviation from the Pee Dee Belemnite (PDB) standard] has been used to partition NEE into photosynthetic and respiratory gross fluxes, with coupled eddy covariance, and isotopic approaches (e.g. Yakir and Wang 1996; Bowling et al. 2001; Ogée et al. 2003; Knohl and Buchmann 2005). In this approach, fluxes of CO2 and 13CO2 must be determined, as well as the C isotope signature of CO2 in the canopy and of ecosystem-respired CO2 (δ13CR eco) (Bowling et al. 2001). δ13CR eco is measured during nighttime using a Keeling plot approach (Keeling 1958, 1961) and is generally assumed to be constant during the period of sampling and during the day–night cycle (Pataki et al. 2003). In addition, when interpreting δ13CR eco, it is often implied that the δ13C of CO2 respired from different sources (autotrophic respiration of twigs, trunks and roots, soil heterotrophic respiration) is uniform or, if non-uniformity is taken into consideration, that the relative contribution of different sources to total ecosystem respiration does not change with time (Pataki et al. 2003). Finally, short-term variations in δ13CR eco, and thus the uncertainty of nighttime Keeling plot values, become a considerable source of error in isoflux approaches, especially when the isotopic disequilibrium between respiratory and assimilatory CO2 fluxes is large (Ogée et al. 2004). Recent studies, however, have shown highly variable 13C signatures of respired CO2 from different plant organs and at the ecosystem scale over short time periods. Knohl et al. (2005) observed variations in δ13CR eco of up to 3.8‰ within 24 h in a beech-dominated deciduous forest. Comparable or even larger variations were observed in CO2 emitted from a Mediterranean oak woodland during 1 night at the ecosystem level (4.2‰; Werner et al. 2006), from leaves of Quercus ilex (8‰; Werner et al. 2007) and from European beech trunks over a diel course (3‰; Maunoury et al. 2007).

One potential source of such observed variations might be related to temporal changes in the δ13C of the respired organic substrates. However, we still lack a clear understanding of the influence of short-term variations in δ13C of respiration substrates on δ13C of plant-, soil- and ecosystem-respired CO2. It is known that δ13CR eco can reflect recent environmental conditions and be linked (with time lags) to environmental conditions affecting C isotope fractionation during photosynthesis (Lancaster 1990; Bowling et al. 2002; Scartazza et al. 2004; Knohl et al. 2005), which in turn is determined by the ratio of the leaf internal to external CO2 concentration (c i /c a ; Farquhar et al. 1982). Only recently have additional post-carboxylation discrimination steps been identified in downstream metabolic pathways of plants (Ghashghaie et al. 2001; Damesin and Lelarge 2003) and associated with transport and other processes (Badeck et al. 2005; Brandes et al. 2006). Such post-carboxylation discrimination can additionally influence the δ13C of organic matter pools with a fast turn-over and might consequently affect the δ13C of CO2 respired from these C pools. In particular, diel (24-h day-night) patterns of transitory starch accumulation and remobilisation (Tcherkez et al. 2004; Gottlicher et al. 2006) can result in day–night variations in δ13C of leaf-exported sugars of up to 2.5‰ under field conditions (Gessler et al. 2007a). Such changes in the potential substrate are expected to affect the δ13C of CO2 from trunk respiration. Moreover, since soil respiration is largely driven by recent photosynthates (Högberg et al. 2001; Ekblad and Högberg 2001; Carbone and Tumbore 2007; Johnsen et al. 2007), the 13C signature of phloem-transported sugars could also influence the δ13C of soil CO2 efflux over a diel cycle. Transport of assimilates to the roots and transfer to the rhizospheric microbial community might result in time-lagged correlations between the δ13C of phloem-transported sugars and of soil-respired CO2.

Another source of temporal variation for δ13C of respired CO2 is C isotopic fractionation during respiration. Post-carboxylation fractionation is responsible for differences in the intra-molecular distribution of 13C in hexose phosphates, and subsequent fragmentation of the molecule (i.e. fragmentation fractionation) during respiration can result in respired CO2 being more enriched in 13C relative to the substrate of respiration (Tcherkez et al. 2004). Several field experiments have shown variable respiratory C isotope fractionation in tree leaves over a diel course (Hymus et al. 2005; Werner et al. 2007), which can result in changes in the δ13C of respired CO2, independently of the C isotope composition of the organic substrate.

In the present study, we followed the short-term dynamics of δ13C from newly assimilated water-soluble organic matter pools in needles, to phloem-transported organic matter in twigs and trunks, and finally to trunk-, soil- and ecosystem-respired CO2 in a Scots pine forest with a 3-h sampling resolution over a 4-day period. In order to investigate environmental and physiological controls of changes in δ13C of respired CO2, we: (1) assessed diel variations in δ13C for CO2 emitted from trunks and soil and their relationships to δ13CR eco, and (2) tested whether such variations were related to comparable (but potentially time-shifted) temporal patterns of δ13C of the organic substrate, or if other processes such as respiratory C isotope fractionation were involved.

Materials and methods

Study site and experimental setup

The study was conducted at the Hartheim long-term study site of the Meteorological Institute of Freiburg, a pine plantation in the southern upper Rhine valley, Germany (47°56′N, 7°36′E, elevation 201 m). The forest was mostly planted with Scots pine (Pinus sylvestris L.) in 1963, with only a few patches of Austrian pine (Pinus nigra L.). All measurements were made in P. sylvestris plots where stand density was 800 trees ha-1. Most of the tree foliage was located between 11 m and the top of the canopy at approximately 15 m. The overstorey leaf area index (LAI) was 1.47 m2 m−2. The dominant plant species of the dense understorey layer were the grass Brachypodium pinnatum, and the sedges Carex alba and Carex flacca. In addition, shrubs and deciduous tree species (in particular Ligustrum vulgare and Ulmus minor) heterogeneously covered 60% of the ground area in the understorey. A detailed description of the experimental site and management is given by Mayer et al. (2000) and Brandes et al. (2006, 2007). Pine tree mean diameter at breast height was 0.18 m. Projected trunk area amounted to 0.89 m2 (trunk area) m−2 (ground area), calculated from the mean trunk radius and the mean trunk height (below the crown) assuming the trunk to be a cylinder, and multiplied by tree density (800 trees ha−1). The forest grows on a Calcaric Regosol, the soil textural class is sandy silt.

The measurement campaign took place from 1200 hours on 6 June 2005 to 0600 hours on 10 June 2005 (all times are expressed as Central European time, with a 0- to 24-h notation). Samples were taken and gas exchange and CO2 measurements were performed every 3 h throughout this period, at 1200, 1500, 1800, 2100, 2400, 0300, 0600 and 0900 hours. Three adjacent dominant or co-dominant individuals of P. sylvestris were selected within reach of a truck-mounted hydraulic lift, and used as replicates for leaf-based measurements. Similar additional trees of comparable size, located in the close vicinity, were selected for trunk CO2 efflux measurements. Soil CO2 efflux measurements were carried out in a P. sylvestris patch that was free of understorey vegetation, within 3 m from the trees used for trunk CO2 efflux measurements.

Micrometeorology and stand transpiration

Meteorological parameters were determined continuously at a measurement tower. Air temperature and humidity were measured using a psychrometer according to Frankenberger, at 12- and 2-m height, and water vapour pressure deficit (VPD) of the air was calculated from these data. Precipitation (Hellmann rain gauge) and photosynthetically active radiation (PAR; determined by Li-190sz-sensors; Licor, Lincoln, Neb.) were determined above the canopy. Volumetric soil water content was measured in the upper 0.3 m using a CS615 water content reflectometer (Campbell Scientific, North Logan, Utah), soil temperature was measured at 0.03 m depth using a Pt100 probe. Data were recorded every 30 s and averaged over 30-min periods.

Xylem sap flow densities (in l m−2 sapwood area s−1) were measured every 5 min on 12 trees using Granier-style probes (Granier 1985, 1987), and averaged over 30-min periods. Stand xylem sapwood area (in m2 m−2) was determined, and radial variations in sap flow within trees were accounted for as described in detail by Brandes et al. (2007). Overstorey stand transpiration (in l m−2 ground area s−1) was calculated by multiplying xylem sap flow density by stand xylem sapwood area (Granier et al. 1996).

Plant material collection

At each measurement time, three twigs per tree were sampled from the sunlit upper third of the crown (approximately 14–15 m) and used for collection of needles and twig phloem. The crown of the pine trees at the stand examined is short and sparse. Based on the work of Brandes et al. (2006) that found no intra-canopy gradients in δ13C and gas exchange parameters, each tree crown was treated as a single unit and the sampled twigs were assumed to adequately represent the whole crown. Needle water-soluble organic matter was extracted separately from current growing season (N) needles and the previous growing season (N−1) needles (see below). In addition, twig bark samples (at the first proximal position without live needles) and trunk bark samples from 1.5-m height were collected, the phellem (cork) was removed and the inner bark used for phloem extraction. Bark samples (approximately 150 mg) were taken from the twig bark with a scalpel, and from the trunk bark with a core borer (13 mm diameter) as described by Gessler et al. (2004).

Needle water-soluble and phloem exudate organic matter

Needle water-soluble organic matter, which was assumed to be representative of the newly assimilated organic matter, was extracted as follows. Immediately after harvesting, N and N−1 needles were frozen in liquid N2, before being freeze-dried. The samples were homogenised and 50-mg aliquots were incubated for 60 min at 5°C in 1-ml demineralised water, heated at 100°C for 1 min to precipitate proteins, and centrifuged (12,000g for 5 min) at 5°C. The supernatant was used for isotope analysis (see below).

Phloem exudate organic matter, which was assumed to be representative of transported sugars, was extracted as described by Schneider et al. (1996) and Gessler et al. (2004). Immediately after collection, bark samples were washed with demineralised water and placed in 6-ml vials containing 2 ml of demineralised water and left at room temperature for exudation for 5 h. The exudation solution was used for isotopic analysis (see below).

Atmospheric CO2 collection

Atmospheric air samples were taken every 3 h at ground level, from below the canopy, mid-canopy and above the canopy, at 0.02-, 9-, 12- and 19-m height, respectively, using automated air samplers (ASA; Theis et al. 2004). In brief, for each air sample, air was pumped (1 l min−1) from the relevant level using Synflex 1300 tubing (Saint-Gobain Performance Plastics, Compagnie de Saint-Gobain, Nanterre), through one of the thirty-three 300-ml glass vials of the air sampler. The vials were filled successively, each vial being flushed for 7 min before closing the valves at each end of it. Time programming and switching of the multiport valves (Valco multiport ST valve, EMT4ST12MWE; VICI, USA) of the ASA were controlled by a data logger (CR10X; Campbell Scientific). The CO2 concentration in the airflow was recorded at a 10-Hz frequency with a closed path infrared gas analyser (Li-840; Li-Cor). The samples were analysed for δ13C within 12 h (see below). For each measurement time, we determined (δ13CR eco based on the CO2 concentration and δ13C in the air samples collected from the four heights in the canopy, using a profile-based so-called Keeling plot approach, i.e. a two end-members mixing model in which the CO2 emitted by the ecosystem mixes in the background atmospheric CO2 (Keeling 1958, 1961). We used a linear regression (model I), following the recommendations of Zobitz et al. (2006), and present only the regressions with R 2 > 0.9.

Collection of CO2 respired from trunks and soils

Measurements of efflux rate and δ13C of CO2 respired from trunks were carried out according to Damesin et al. (2005). Clear perspex chambers (surface area 300 cm2, volume 4.5 l, shielded from direct sunlight by paper-backed aluminium foil) were sealed to the trunks of three trees at 1.5-m height, using Terostat VII (Henkel Teroson, Heidelberg). We selected trunk sections without visible lichens or algae present on the bark. The chambers remained fixed with a strap to the trees throughout the campaign. Between measurements, the chambers were left open to the surrounding air. During gas sampling times, they were sealed airtight and an internal fan ensured thorough mixing of air. Efflux rate and δ13C of CO2 respired from soils were determined according to Ekblad and Högberg (2000). Three opaque PVC chambers (ground surface 434 cm2, volume 10.8 l) were gently inserted 5 mm into the soil and maintained stable with a 2 kg weight.

CO2 efflux rates from trunks and soils were calculated at each sampling time from the linear increase in CO2 concentration within the first 2 min after closing the chambers, measured continuously with a closed path infrared gas analyser (Licor 6400; Li-Cor). At each sampling time, five air samples were taken from each chamber using a plastic 12-ml syringe, transferred to CO2-free Ar-flushed sample vials (Exetainer; Labco, High Wycombe, UK) and analysed for δ13C (see below). The first sample was taken within 2 min after closing the chamber, and the four following samples were taken at intervals of approximately 75 μmol mol−1 as the CO2 concentration rose inside the chambers. The isotopic signature of trunk and soil CO2 efflux was determined from the CO2 concentration and δ13C in the air samples using a Keeling plot approach as described above. Keeling plots with an R 2 < 0.9 were discarded. To detect any storage effect on the samples, additional sample vials were filled with standard gas samples with known isotopic signature from a gas cylinder (400 μmol CO2 mol−1 in N2) at each measurement time. In agreement with the observation of Gessler et al. (2007a), there was no significant change in δ13C of the standard gas within the maximum storage period of the samples (7 days).

In order to assess the influence of soil plus trunk respiration on δ13CR eco, we calculated flux-weighted averages for δ13C of CO2 respired by trunks and soil together. To standardise soil and trunk CO2 fluxes, we calculated trunk respiration rate per squared metre ground area by using the projected trunk area of 0.89 m2 m−2. Soil- and trunk-respired CO2 δ13C values were then weighted by their respective ground area-based CO2 fluxes, assuming δ13C of trunk-respired CO2 did not change with height, as previously shown for pine trees from the same stand (Brandes et al. 2006).

Isotope ratio mass spectrometry measurements

The δ13C of phloem exudate organic matter and needle water-soluble organic matter was measured as follows: 150- and 100-μl aliquots were taken from the phloem exudation solutions and the needle water-soluble organic matter extracts, respectively. The aliquots were transferred to tin capsules and oven-dried at 60°C for 12 h. Chromosorb W (approximately 10 mg) was added to the tin capsules as sorbent for the liquid before drying. The samples contained on average approximately 400 μg organic C. Samples were combusted in an elemental analyser (NA 2500; CE Instruments, Milan) coupled to an isotope ratio mass spectrometer (Deltaplus; Finnigan MAT, Bremen) by a Conflo II interface (Finnigan MAT). SD of repeated measurements (n = 10) of the laboratory standard was ± 0.1‰.

Exetainers filled with gas samples from soil and trunk CO2 efflux were measured with a modified Finnigan Gasbench II periphery (Finnigan MAT) equipped with a custom-built cold trap coupled to an isotope ratio mass spectrometer (DeltaplusXP; Finnigan MAT). The mean precision of the measurement of the laboratory standard was 0.25‰.

The ASA were connected to an isotope-ratio mass spectrometer (DeltaplusXL; Finnigan MAT) via a gas concentration peripheral (Precon; Finnigan MAT). The switch-valves of the ASA were controlled by the system-software, while the ASA was connected to a He stream that flushed the gas to the Precon. After purification and cryogenic separation, the CO2 of the air samples was led to the mass spectrometer for δ13C analysis. Mean precision determined from repeated analysis of reference gas filled in the ASA was 0.2‰. Isotopic values are expressed in δ notation (‰ units), on the Vienna PDB scale.

Needle gas exchange

Needle gas exchange was determined under ambient light and temperature conditions in the upper third of the canopy by inserting small twigs (with their N and N−1 needles attached; three replicates per individual tree) in a conifer chamber connected to a portable gas exchange system (GFS3000; Walz, Effeltrich, Germany). Twigs that were next to the ones sampled for isotope analysis were chosen for gas exchange measurements. N−1 needles contributed on average approximately 65% of the total needle area in the chamber. Net CO2 and H2O exchange rates were measured, and stomatal conductance (g s) was subsequently calculated according to von Caemmerer and Farquhar (1981). Separate values of leaf level assimilation rate (A) and c i /c a were calculated for N and N−1 needles (see Barnard et al. 2007 for details), assuming g s and A of N−1 needles to amount to 53 and 61% of N needles (Beadle et al. 1985). The projected area of the N and N−1 needles that were inserted in the chamber was determined with a leaf area meter (ΔT Devices, Cambridge, UK), and three-dimensional leaf area was calculated according to Luoma (1997).

Statistical analyses

Data analyses were performed using SPSS 10.05 (SPSS; Chicago, Ill.) and NCSS 2004 (Number Cruncher Statistical Systems, Kaysville, Utah). Correlations were calculated using bivariate correlation procedures. For multiple regression analyses, the independent variable subset selection was performed applying hierarchical forward switching.

In order to estimate potential time lags between two variables (X, Y), cross-correlation analyses were performed. The cross-correlation between X t and Y t + k is called the kth order cross-correlation of X and Y. The correlation coefficient (r k ) of the cross-correlation is calculated using the following equation (Kendall and Ord 2006):

$$ r_{k} = \frac{{{\sum\nolimits_{i = 1}^{n - k} {{\left( {X_{i} - \ifmmode\expandafter\bar\else\expandafter\=\fi{X}} \right)}{\left( {Y_{{i + k}} - \ifmmode\expandafter\bar\else\expandafter\=\fi{Y}} \right)}} }}} {{{\sqrt {{\sum\nolimits_{i = 1}^n {{\left( {X_{i} - \ifmmode\expandafter\bar\else\expandafter\=\fi{X}} \right)}^{{^{2} }} {\sum\nolimits_{i = 1}^n {{\left( {Y_{i} - \ifmmode\expandafter\bar\else\expandafter\=\fi{Y}} \right)}^{{^{2} }} } }} }} }}} $$
(1)

where \( \ifmmode\expandafter\bar\else\expandafter\=\fi{X} = \frac{1} {n}{\sum\nolimits_{i = 1}^n {X_{i} } } \) and \( \ifmmode\expandafter\bar\else\expandafter\=\fi{Y} = \frac{1} {n}{\sum\nolimits_{i = 1}^n {Y_{i} } }. \) Cross-correlations were calculated for the single time points during the diel courses (8 sampling times per 24 h). A time lag of, e.g. k = 1, means a shift from (i) to (i + 1) of 3 h. For k = 8, the time lag is thus 24 h. For k = 0, r k is equivalent to Pearson’s correlation coefficient (r).

In order to assess whether the time course of δ13C in organic matter and CO2 contained a periodic component, a periodogram was calculated using the spectral analysis function of NCSS 2004 and corrected for series average and trend, according to Brandes et al. (2006) and Barnard et al. (2007).

Results

Meteorological variables and stand transpiration

With the exception of the first day, maximum daily PAR exceeded 1,700 μmol m−2 s−1 (Fig. 1a) during the sampling period. Mean day and night time air temperatures within the canopy at 12-m height were 14.4 and 9.3°C, respectively (Fig. 1b). The maximum air temperature of 20.6°C was reached in the afternoon on 10 June and the nocturnal minimum (3.4°C) in the early morning of the same day. Air temperature at 2-m height showed a comparable diel rhythm but was slightly lower at night. Mean soil temperature at 3-cm depth was 13.5°C, with a range between the evening maximum and morning minimum of 0.9–3.5°C. Light rainfall occurred during the night between 6 and 7 June (total of 1.88 mm rainfall above canopy, data not shown). Maximum daily VPD at 12-m height amounted to 8.9 hPa on 6 June and increased continuously until 10 June when a maximum value of 16.5 hPa was observed (Fig. 1c). Soil volumetric water content decreased constantly over the measurement period. The minimum daytime value was 20.6% on 6 July and 17.8% on 10 July.

Fig. 1
figure 1

Diel variation of a incident photosynthetically active radiation (PAR), b air temperature (T air ) at 12-m height (solid black line), at 2-m height (solid grey line) and in the soil (T soil ) at 0.03-m depth (dashed line), c atmospheric water vapour pressure deficit (VPD) at 12 m (solid line) and volumetric soil moisture (dotted line) and d stand transpiration (ST) rate of the overstorey pine trees

Full size image

Stand transpiration (for the overstorey trees) closely followed the temporal patterns of VPD at 12-m height. Daily sums of transpired water increased from 0.6 l m−2 day−1 on 6 June to 1.5 l m−2 day−1 on 10 June.

Needle, trunk and soil net CO2 exchange

Needle net CO2 exchange was calculated for both N and N−1 needle cohorts (Fig. 2a). The maximum daytime net photosynthesis rate (5.1 μmol m−2 s−1) was measured for N needles on 8 June; the minimum was on 6 June when PAR was also lowest. Nighttime CO2 efflux ranged between 0.1 and 0.5 μmol m−2 s−1 for both N and N−1 needles.

Fig. 2
figure 2

Diel variations of a net CO2 exchange (\( J_{{{\text{CO}}_{2} }} \)) of current year (N; black squares) and previous year needles (N−1; white diamonds) from the upper third of the crown (14–15 m) and b net respiration rates (R) of trunk at 1.5-m height (grey squares) and soil (white triangles). Note that positive values of \( J_{{{\text{CO}}_{2} }} \) denote net CO2 uptake by needles, whereas positive values for R denote net CO2 efflux to the atmosphere. Data are mean values (±SE in b, n = 3 in a and b). The black bars at the bottom of the figure indicate the nighttime period

Full size image

Both trunk and soil respiration rates showed clear diel courses, with minimum rates in the dark period or shortly after sunrise (Fig. 2b). Maximum respiration rates were determined between 0900 hours and sunset. Particularly for trunk respiration rate, the amplitude of variation between diel maxima and minima increased with time. Soil respiration tended to decrease over the measurement period but daily mean values were not significantly different among days.

Diel courses of δ13C in organic matter and CO2

We found slight diel variations in δ13C of water-soluble organic matter, in both N and N−1 needles at 14- to 15-m height (Fig. 3a). Water-soluble organic matter showed least negative δ13C values at the end of the night, and minimum δ13C values were observed during the light period. The diel range varied between 0.6 and 1.0‰ in N needles and between 0.4 and 0.7‰ in N−1 needles. N−1 needles were slightly, albeit not significantly, less depleted in 13C compared to N needles. Over the measurement period, we identified no general trend for δ13C in water-soluble organic matter.

Fig. 3
figure 3

Diel variations of δ13C in a needle water-soluble organic matter (OM) [leaf ws OM; N needles (black squares), N−1 needles (white diamonds)], b bulk phloem exudate OM (phloem OM) at twig level (black triangles) and trunk level at 1.5-m height (white squares), c CO2 respired from trunks at 1.5-m height (grey squares) and d CO2 from soil (white triangles) and ecosystem (black and white shaded circles) respiration. Data are mean ± SE (n = 3). Black bars at the bottom of the figure indicate the nighttime period. Note the difference in scale of the y-axes between plots a and b vs. c and d. For other abbreviations, see Fig. 2

Full size image

The diel courses of δ13C in twig phloem exudate organic matter were comparable to those of leaf water-soluble organic matter (Fig. 3b). The range between daytime maximum and nighttime minimum of δ13C was between 0.6 and 0.9‰. Twig phloem exudate organic matter was significantly enriched in 13C (P < 0.05) compared to leaf water-soluble organic matter of both N and N−1 needles during the first two day-night cycles, but not for the rest of the measurement period. Trunk phloem exudate organic matter was significantly enriched in 13C compared to twig phloem exudate and needle water-soluble organic matter (P < 0.01), but showed no clear diurnal pattern.

Mean R 2-values for soil and trunk Keeling plots were 0.971 ± 0.053 and 0.967 ± 0.059, respectively. The δ13C of trunk-respired CO2 showed clear diel patterns on the first 2 measurement days, with lowest δ13C values in the evening (between 1800 and 2000 hours) and maximum δ13C values in the morning between 0300 and 0600 hours (Fig. 3c). The maximum diel range of variation in δ13C was 4.0‰. A morning peak for δ13C was not observed on the 9 June but again on 10 June, when δ13C in trunk-respired CO2 increased from −25.7‰ at 0000 hours to −22.9‰ at 0600 hours. The δ13C of soil-respired CO2 showed a diel course that was slightly phase-shifted compared to trunk respiration. Minimum δ13C values were generally observed around midnight and δ13C maxima between 1200 and 1500 hours. Mean diel range of variation was 2.7‰.

The general lack of a CO2 concentration gradient in the forest vertical profile during daytime, and the rain event at the beginning of the sampling period, resulted in poor regressions during daytime, thereby restricting most of our measurements of δ13C of ecosystem-respired CO2 to the last 3 nights of the campaign. The average and SD of the R 2 of the regressions used to determine the δ13CR eco data presented here was 0.993 ± 0.014. δ13CR eco varied by up to 6.1‰ within 9 h but showed no clear temporal pattern over the measurement period.

Periodicity of δ13C in CO2 and organic matter

Spectral data analysis indicated a clear 24-h periodicity [i.e. a maximum for I(f) at a wavelength of approximately 8] of δ13C in water-soluble organic matter from N and N−1 needles as well as in twig phloem exudate organic matter (Fig. 4a). In contrast, we observed no obvious periodicity for δ13C in trunk phloem exudate organic matter, as indicated by the even distribution of I(f). We found a 24-h periodicity for δ13C in trunk and soil CO2 efflux (Fig. 4b). Despite the common periodicity pattern for these two CO2 sources (albeit phase shifted, cf. Fig. 3c, d), no significant correlation between them could be shown with any time lag between 0 and 12 h (Fig. 5). It might be assumed that CO2 originating from belowground (thus carrying a C isotopic signature that is comparable to that of soil-emitted CO2) and transported within the xylem stream contributes significantly to CO2 emitted from stems (see review by Teskey et al. 2008). In that case, the changes in xylem flow rate should modify the relation between δ13C of soil-emitted and trunk-emitted CO2. However, a multiple regression analysis with xylem sap flow rate and δ13C of soil respiration as independent factors could only explain 12% of the variation of trunk-emitted CO2.

Fig. 4
figure 4

Periodogram in the wavelength scale for the time series of δ13C in OM pools (a) and in CO2 respired by trunk and soil (b). a Periodogram of δ13C of water-soluble OM in N needles (solid line) and N−1 needles (dotted line) as well as phloem exudate organic matter from twigs (dashed line) and trunk (grey solid line). b Periodogram of δ13C in CO2 respired by trunk (solid line) and soil (dotted line). A wavelength of 8 corresponds to a 24-h periodicity. Note that I(f) is dimensionless. For abbreviations, see Figs. 2 and  3

Full size image
Fig. 5
figure 5

Cross-correlations between δ13C of C pools as C is transported from the needles to the trunk, and between trunk phloem exudate OM—as a putative source for respiration—and respired CO2 (right side) as well as cross-correlations between δ13C in CO2 and physiological and environmental factors (left side). Bold lines with one arrow head indicate the assumed direction of C fluxes. Thin lines with two arrow heads indicate cross-correlations with a time lag of 0 h between different factors without a priori anticipating an effective direction. Thin lines with one arrow head indicate cross-correlations with time lags. Arrows point to the variable lagging behind. White arrowheads indicate missing significant correlation between two parameters. Only those environmental variables are shown which were significantly related to δ13C of CO2, OM or a physiological parameter. TL time lag, T air (2 m) air temperature at 2-m height, VPD (12 m) VPD at 12-m height, T soil soil temperature. For other abbreviations, see Figs. 1 and  3

Full size image

δ13C in CO2 respired in relation to organic matter and environmental variables

In order to trace the fate of newly assimilated C within the trees and the whole ecosystem, we performed time series cross-correlation between the δ13C values of the different organic matter pools and respiratory CO2 along the C transport pathway from needle water-soluble organic matter to soil CO2 efflux (Fig. 5). To estimate the influence of environmental (air and soil temperature, VPD, PAR, volumetric soil water content) and physiological parameters (transpiration rate, needle-level CO2 exchange, as well as trunk and soil respiration rates) on δ13C of soil- and trunk-respired CO2, additional cross-correlation analyses were performed for these variables (Fig. 5). Applying multiple regression models did not improve the coefficient of determination, compared to single-variable models.

Twig phloem exudate organic matter was significantly correlated with water-soluble organic matter from both N and N−1 needles (Fig. 5). The highest correlation coefficients were obtained when no time lag was assumed. There was, however, no correlation between twig and trunk phloem exudate organic matter or between δ13C of soil- or trunk-respired CO2 and δ13C of trunk phloem exudate organic matter (no matter whether time lags were applied or not).

The δ13C of trunk-respired CO2 was negatively correlated to trunk respiration rate and to air temperature at 2-m height with no time lag, as well as to stand transpiration with a lag of 3 h (Fig. 5). Soil and trunk respiration rates were significantly positively correlated (assuming no time lag) with soil and air temperature, respectively. The δ13C of soil-respired CO2 showed significant correlations: with (1) volumetric soil water content (with no time lag), and (2) soil temperature with a time lag of 6 h (Fig. 5).

When separating day and night values for Pearson’s correlation analysis, we found no changes in correlation coefficient for trunk respiration rate and δ13C of trunk-respired CO2. However, δ13C of nighttime soil CO2 efflux was significantly correlated to soil respiration rate (R = −0.664; P = 0.018) whereas during the daytime no such correlation was observed. When considering only daytime values, the correlation between the δ13C of soil-respired CO2 and soil water content was stronger (R = −0.72; P < 0.001) than for the analysis of the whole diel courses.

We found no significant correlation between δ13CR eco and any δ13C value of the organic C pool or CO2. In addition, there was neither a significant correlation of δ13CR eco with stand transpiration, needle-level CO2 gas exchange, trunk and soil respiration rate, nor with any environmental parameter measured. The flux-weighted average δ13C of trunk- and soil-respired CO2 was not significantly correlated with δ13CR eco.

In order to relate the apparent C isotope fractionation during trunk respiration to trunk respiration rate, we calculated the difference between δ13C of trunk phloem exudate organic matter, the putative substrate for respiration and trunk-respired CO2. We found a trend for the δ13C of trunk-respired CO2 to be strongly enriched in 13C (by up to >3‰) compared to δ13C of phloem exudate organic matter when respiration rates were low, and to be comparable to δ13C of phloem exudate organic matter when respiration rates were highest (Fig. 6). Since we assume phloem organic matter to be a potential substrate for soil respiration, we also calculated the difference between δ13C of trunk phloem exudate organic matter and of soil-respired CO2. We found, however, no significant relation between this difference and soil respiration rates (data not shown). In order to account for potential time lags between δ13C of CO2 and phloem organic matter due to the time for transport of phloem sugars belowground, we also correlated \(\delta ^{13} {\text{C}}_{{\text{trunk\ pholem\ OM}}} - \delta ^{13} {\text{C}}_{{\text{CO}}_{\text{2}} {\text{\ soil}}} \) with R soil, using δ13Ctrunk phloem OM 3, 6, 9, 12, 15, 18, 21, 24, 27 and 30 h before the actually measured \(\delta ^{13} {\text{C}}_{{\text{CO}}_{\text{2}} {\text{\ soil}}} .\) Independent of the time-lag applied there was no significant relation between \(\delta ^{13} {\text{C}}_{{\text{trunk\ pholem\ OM}}} - \delta ^{13} {\text{C}}_{{\text{CO}}_{\text{2}} {\text{\ trunk}}} ;\) and soil respiration rate (data not shown).

Fig. 6
figure 6

Difference between δ13C of trunk phloem OM (δ13Ctrunk phloem OM) and of trunk-respired CO2 \( (\delta ^{{13}} {\text{C}}_{{{\text{CO}}_{{2\;}} {\text{trunk}}}} ) \) plotted against trunk respiration rate (R trunk). Data shown are mean values (n = 3). Regression line equation: \( \delta ^{{13}} {\text{C}}_{{{\text{trunk}}\,{\text{phloem}}}} - \delta ^{{13}} {\text{C}}_{{{\text{CO}}_{{2\;}} {\text{trunk}}}} {\left[ \permille \right]} = 0.98{\left[ {\permille\,{\text{ $ \mu $ mol}}^{{{\text{ - 1}}}} {\text{m}}^{2} s} \right]} \times R_{{{\text{trunk}}}} {\left[ {{\text{ $ \mu $ mol}}\,{\text{m}}^{{ - 2}} {\text{s}}^{{ - 1}} } \right]} - 2.74{\left[ \permille \right]}\,{\left( {R^{2} = 0.48;P < 0.001} \right)} \)

Full size image

Discussion

We aimed to trace the fate of the isotopic signal of recently assimilated organic C during its transport from the canopy down the trunk and during respiratory conversion to CO2 during the diel course. The main result was that the distinct diel variations in δ13C observed in newly assimilated organic matter in the tree crown were strongly dampened during phloem transport to the trunk but that a 24-h periodicity was again observed in the δ13C of trunk- and soil-respired CO2.

Diel variations in δ13C of water-soluble organic matter were dampened during transport from the canopy to the trunk base

Soluble organic matter of both needle age classes, as well as twig phloem exudate organic matter, showed clear diel variations. The highest δ13C values were observed during the dark period and the most negative values during the day (Fig. 3). Part of this variation was explained by c i /c a during the time of the day when net assimilation occurred. Cross-correlation analyses exhibited a 6-h time lag between c i /c a and δ13C (Fig. 5) that is consistent with results of a previous study tracing δ18O at the same site (Barnard et al. 2007). These authors showed that the turnover time of leaf sugars caused the δ18O signature of leaf water to be imprinted on the leaf sugars with a comparable time lag. However, c i /c a alone could explain only part (approximately 25%) of the changes in δ13C of needle water-soluble organic matter, for two reasons. First, variation of c i /c a in the morning and evening, when light intensity and assimilation rate are low, may result in c i /c a measurements that are not representative of the longer time period that the δ13C values of leaf sugars integrate over (Gessler et al. 2007a). Second, the diel patterns of δ13C of needle and phloem sugars have been found to depend strongly on post-carboxylation fractionation (Brandes et al. 2006), with more negative δ13C values during the day when starch is synthesised and more positive values during the night when 13C-enriched starch is remobilised (Gleixner and Schmidt 1997; Tcherkez et al. 2004). In addition, it has to be acknowledged that leaf water-soluble organic matter as extracted here does not only consist of sugars but also of organic acids, amino acids and other compounds—a fact that could additionally complicate the interpretation of isotopic data. However, Brandes et al. (2006) showed that soluble organic matter and sugars in P. sylvestris needles had a similar 13C signature.

As organic matter was transported from the canopy down the trunk to the C sinks in the tree, the 24-h periodicity of δ13C disappeared almost completely (Fig. 4a). As a consequence, no correlation between twig and trunk phloem exudate was observed (Fig. 5). The lack of a day-night periodicity in δ13C of trunk phloem organic matter of Pinus sylvestris is in agreement with results from Betson et al. (2007) in Picea abies and from Gessler et al. (2007a) in Eucalyptus delegatensis. The dampening of the diel δ13C variations in phloem-transported organic matter at the trunk base observed here can be explained by a mixing of various C pools with different metabolic histories during phloem transport down the trunk (Brandes et al. 2006). This hypothesis is supported by findings of Keitel et al. (2003, 2006) and Brandes et al. (2007), showing that δ13C in trunk phloem sap organic matter integrates mean canopy c i /c a over several days in European beech and Scots pine, respectively.

In contrast to trunk phloem organic matter, the δ13C of trunk-respired CO2 showed a clear diel periodicity

Our study showed large diel variations of up to 4‰ in the δ13C of trunk-respired CO2, while no comparable variations were observed in the δ13C of phloem-transported organic C. Two possible hypotheses could explain this result. First, the substrate respired could change over time, switching from phloem sugars to another C source. A shortage of readily available carbohydrates is typically required for either the remobilisation of sugars from storage starch or for organic compounds other than carbohydrates (especially fatty acids; Tcherkez et al. 2003) to be respired. Insufficient C supply by phloem transport is very unlikely since phloem exudate organic C content did not change during the day-night cycle (data not shown). A switch from carbohydrates to fatty acids was also not likely to occur during our measurement period, since fatty acids were found to play only a minor role in C storage in Scots pine in the summer (Fischer and Holl 1991). A significant and temporally varying contribution of xylem sugars with a δ13C signature that is potentially different from that of phloem sugars is also an unlikely reason for the variation in δ13C of CO2. Gessler et al (2007b) showed that xylem-transported sugars matched well in their (oxygen) isotope signature with phloem sugars, pointing to xylem sugars to originate from direct phloem-to-xylem exchange. As a consequence, different δ13C values in the two pools are unlikely. In addition, newly developing xylem tissue has the same isotopic composition as phloem sugars (Cernusak et al. 2005), thus indicating the phloem pool as the ultimate C source for trunk tissues. Second, changes in the respiratory C isotope fractionation could occur over time. Such changes have been documented in leaves (Hymus et al. 2005; Prater et al. 2006; Werner et al. 2007) and trunks (Maunoury et al. 2007) of trees over a diel cycle. Our study showed a close correlation between the δ13C of trunk-respired CO2 and both respiration rate and air temperature. This is consistent with other studies that addressed the controls of the 13C signature of respired CO2 (Tcherkez et al. 2003; Maunoury et al. 2007) over longer time spans. Moreover, we found that apparent C isotope fractionation (calculated as \(\delta ^{13} {\text{C}}_{{\text{trunk\ pholem\ OM}}} - \delta ^{13} {\text{C}}_{{\text{CO}}_{\text{2}} {\text{\ trunk}}} ;\) cf. Fig. 6) was highest (i.e. discrimination as calculated in Fig. 6 was the most negative) when respiration rate—which in turn was driven by temperature—was low and vice versa. Hymus et al. (2005) explained observed diel variations in δ13C of CO2 respired from leaves of two oak species by a shift between the relative contributions of CO2 released from decarboxylation of pyruvate (originating from glycolysis) and CO2 derived from the citrate cycle, without a change in the organic substrate for respiration. This is a consequence of fragmentation fractionation: the CO2 released by the decarboxylation of pyruvate originates from the relatively 13C-enriched C-3 and C-4 atoms of a glucose molecule, whereas the CO2 emitted from citrate cycle decarboxylation reactions originates from the C-1, C-2, C-5 and C-6 atoms, which are relatively 13C-depleted compared to glucose (Rossmann et al. 1991; Tcherkez et al. 2003). Glycolysis (and thus decarboxylation of pyruvate) is less temperature-dependent than mitochondrial oxidation (citrate cycle) (Berry and Raison 1981; Atkin et al. 2000) and, thus, the relative contribution of CO2 from glycolysis to total respiration is likely to increase at lower temperatures and subsequent lower respiration rates, resulting in higher apparent isotope fractionation. In contrast, at higher temperatures, a relatively higher contribution of citrate cycle-derived CO2 to total respiration would not only result in increased respiration rates but also in apparent fractionation approaching zero when complete oxidation of the substrate is assumed (for a review see Ghashghaie et al. 2003). In conclusion, the diel pattern of δ13C of trunk-respired CO2 that we observed does not seem to be due to changes in the 13C signature of the respired organic substrate, but most probably due to changes in respiratory C isotope fragmentation fractionation over the diel course.

The δ13C of soil-respired CO2 showed diel variations that are related to temperature and humidity

The δ13C in CO2 respired from soil displayed variations with a 24-h periodicity (Fig. 4), that were comparable to trunk respiration (albeit with a time shift; Fig. 3). Despite that common periodicity, δ13C of trunk- and soil-respired CO2 were not correlated (Fig. 5), as the range of variation of δ13C was not consistent between the two CO2 sources and overall trends were different over the period of observation. In addition, changes in xylem flow rate did not modify the relation between δ13C of soil- and trunk-emitted CO2. Thus, we conclude that δ13C of trunk-respired CO2 is not mainly determined by belowground-produced and xylem-transported CO2, in contrast to observations on other species (Teskey and McGuire 2005). However, our measurement of soil-respired CO2 integrates different CO2 sources, therefore and despite the lack of correlation between δ13C in soil and trunk-emitted CO2, we cannot rule out the possibility that one (or several) of these sources (e.g. root or rhizosphere respiration) contribute(s) to stem CO2 emission.

Our findings of diel variations in flux rate and δ13C of soil respiration differ from the recent results of Betson et al. (2007), who measured no diel cycle for soil CO2 efflux rate and 13C signature in a boreal forest. These authors suggested that, in their study, the short nights (2–4 h) could prevent intensive exhaustion of photosynthate supply to the roots from the canopy, which could otherwise influence rate and δ13C of soil CO2 efflux. However, studies of root carbohydrate contents in adult coniferous and deciduous trees in temperate forests gave no indications of nighttime sugar depletion in roots (Gessler et al. 2002), hence we do not assume variations in carbohydrate supply to be a major factor controlling short-term changes in the δ13C of soil-respired CO2 in our ecosystem. In contrast to Betson et al. (2007), we observed no correlation between δ13C of trunk phloem-transported organic matter and soil-respired CO2, and consequently cannot conclude that a direct coupling between these two pools occurs over the diel course.

In our study, the δ13C of soil CO2 efflux was correlated with soil temperature (6 h time-lag) and with soil moisture (no lag). Such changes in soil temperature and water content likely affect heterotrophic and autotrophic soil respiration differently (Boone et al. 1998; Bhupinderpal et al. 2003; Ekblad et al. 2005; Heinemeyer et al. 2007). In addition, several studies have shown that CO2 respired by soil microorganisms and plant roots differs in its 13C signature (e.g. Bhupinderpal et al. 2003; Gessler et al. 2007a). As a consequence, the changes in the δ13C of soil-respired CO2 that we measured over the diel course could be due to changes in the relative contributions of heterotrophic and autotrophic CO2 fluxes to total soil efflux in response to environmental conditions. Multiple regression models for δ13C of soil CO2 considering concerted effects of soil water content, soil and air temperature, soil respiration rate and δ13C of trunk phloem organic matter did not explain significantly more of the variability than the single regressions presented in Fig. 5, suggesting more complex interactions and additional possible drivers at play.

Although we acknowledge that the organic matter transported in the trunk phloem is not the direct organic source for soil respiration, we have to assume that phloem-transported sugars supply roots and soil microorganisms with rapidly metabolisable C. We hypothesise that in addition to direct environmental effects, changes in the substrates respired by both the soil microbial community and the roots may result in changes in the δ13C of soil CO2 efflux. In particular, changes in soil water content can alter substrate diffusion and availability for the soil microbial community, thereby influencing the δ13C of heterotrophic respiration. As an additional factor, changes in soil moisture also affect the transport time of gas-phase soil CO2 (Stoy et al. 2007), thus potentially altering the time lag between belowground respiration and CO2 emission from the soil surface. In addition, as explained for trunk respiration, we might not only have to consider a shift in the relative contribution of heterotrophic and autotrophic respiration during the diel course, but changes in fragmentation fractionation during respiration of both soil microorganisms and roots.

δ13CR eco displayed much higher short-term variations than the δ13C of trunk- and soil-respired CO2

The 6.1‰ maximum variation of δ13CR eco that we measured within 1 night is comparable with observations in other ecosystems (Bowling et al. 2003: 6.4‰, grassland; Knohl et al. 2005: 3.8‰, mixed deciduous forest; Werner et al. 2006: 4.2‰, oak woodland).

Although we calculated the partitioning of NEE into its respiratory and assimilatory components (based on isotopes and eddy covariance), the isotopic disequilibrium between canopy assimilation and ecosystem respiration was too small (approximately 1‰) to warrant satisfactory results (data not shown). As a consequence, we could not quantify the potential error that highly variable δ13CR eco introduces in NEE partitioning. Ogée et al. (2004) showed that when the isotopic disequilibrium is small, the large uncertainty in flux partitioning is not reduced when δ13CR eco is estimated more accurately. Under conditions of stronger isotopic disequilibrium, however, a consideration of the variability of δ13CR eco is likely to improve the accuracy of the isoflux approach (Ogée et al. 2004).

The variability of δ13CR eco determined in our study exceeded that of δ13C of trunk- and soil-respired CO2 (4 and 2.7‰, respectively). In addition, the flux-weighted average δ13C of trunk- and soil-respired CO2 could not explain variations in δ13CR eco. These results point towards a major contribution of δ13C of foliage respiration to the diel variation in δ13CR eco over the measurement campaign in our pine forest. These findings are consistent with those of Mortazavi et al. (2005) who observed that in a 17.2-m-tall evenly aged loblolly pine plantation, δ13CR eco was determined by the 13C signature of foliage-respired CO2, in stark contrast to a mixed broadleaf forest where δ13CR eco was clearly dominated by the 13C signature of forest floor CO2 efflux.

To estimate the relative contribution of foliage respiration to ecosystem respiration, we scaled up leaf-level respiration (Fig. 2) to a ground area basis (μmol m−2 ground area s−1), assuming an overstorey LAI of 1.47 m2 m−2 (Brandes et al. 2006) and a 50% contribution of both needle age classes (Barnard et al. 2007), and related this to ground area-based soil and trunk respiration fluxes. We acknowledge that this extrapolation might introduce some scaling error, and that the respiration of understorey species, which was not examined for δ13C in respired CO2 in our study, might also contribute to the observed variation in δ13CR eco, and therefore consider this contribution of foliage respiration only a rough estimation. Based on this calculation, nighttime foliage respiration might have contributed to between 20 and 62% of total ground area-based respiration (soil respiration + trunk respiration + needle respiration) during our measurement period. C isotope discrimination of foliar respiration and consequently the δ13C of leaf-respired CO2 can vary by up to 8‰ in other tree species during the diel course (Hymus et al. 2005; Werner et al. 2007), and thus displays an even higher variability than that of δ13C of trunk- and soil-respired CO2 observed in the present study. Thus, the large variations of δ13CR eco can be explained by highly variable contributions of foliar respiration to total ecosystem respiration, combined with strong diel changes in δ13C of leaf-respired CO2.