Tracing carbon and oxygen isotope signals from newly assimilated sugars in the leaves to the tree-ring archive



    Corresponding author
    1. Centre for Systems Biology (ZBSA), Albert-Ludwigs-University Freiburg, Habsburgerstr. 49, 79104 Freiburg, Germany,
    2. Institute of Forest Botany and Tree Physiology, Albert-Ludwigs-University Freiburg, Georges-Köhler-Allee 53/54, 79110 Freiburg, Germany,
      A. Gessler. Fax: +49 761 2038302; e-mail:
    Search for more papers by this author

    1. Institute of Forest Botany and Tree Physiology, Albert-Ludwigs-University Freiburg, Georges-Köhler-Allee 53/54, 79110 Freiburg, Germany,
    Search for more papers by this author

    1. Institute of Plant Sciences, ETH Zurich, Universitaetsstrasse 2, 8092 Zurich, Switzerland and
    Search for more papers by this author

    1. Helmholtz-Zentrum Potsdam Deutsches Geoforschungszentrum GFZ, Sektion 5.2 Klimadynamik und Landschaftsentwicklung, Telegrafenberg, 14473 Potsdam, Germany
    Search for more papers by this author

    1. Institute of Forest Botany and Tree Physiology, Albert-Ludwigs-University Freiburg, Georges-Köhler-Allee 53/54, 79110 Freiburg, Germany,
    Search for more papers by this author

    1. Institute of Plant Sciences, ETH Zurich, Universitaetsstrasse 2, 8092 Zurich, Switzerland and
    Search for more papers by this author

A. Gessler. Fax: +49 761 2038302; e-mail:


The analysis of δ13C and δ18O in tree-ring archives offers retrospective insights into environmental conditions and ecophysiological processes. While photosynthetic carbon isotope discrimination and evaporative oxygen isotope enrichment are well understood, we lack information on how the isotope signal is altered by downstream metabolic processes.

In Pinus sylvestris, we traced the isotopic signals from their origin in the leaf water (δ18O) or the newly assimilated carbon (δ13C), via phloem sugars to the tree-ring, over a time-scale that ranges from hours to a growing season.

Seasonally, variable 13C enrichment of sugars related to phloem loading and transport did lead to uncoupling between δ13C in the tree-ring, and the ci/ca ratio at the leaf level. In contrast, the oxygen isotope signal was transferred from the leaf water to the tree-ring with an expected enrichment of 27‰, with time-lags of approximately 2 weeks and with a 40% exchange between organic oxygen and xylem water oxygen during cellulose synthesis.

This integrated overview of the fate of carbon and oxygen isotope signals within the model tree species P. sylvestris provides a novel physiological basis for the interpretation of δ13C and δ18O in tree-ring ecology.


Global climate change is expected to markedly affect plant growth and performance, and, as a consequence, the composition and spatial distribution of species in terrestrial ecosystems within the next 50–100 years (IPCC 2007). Under a changing climate, the characterization and interpretation of the physiological information laid down in natural archives in the past are a central tool for understanding plant performance in the long term. In woody plants with secondary growth, tree-rings provide such an archive that can be explored for physiological signals.

The information contained not only in growth parameters such as tree-ring width or maximum latewood density, but also in cell structure data, is widely used: (1) to detect ongoing environmental changes; and (2) as chronological archives to retrospectively record growth and vitality of trees subjected to various environmental conditions over different climate zones, as well as to reconstruct past climatic conditions from regional to hemispheric scales (Schweingruber 1996; Briffa et al. 2002). Moreover, radial growth is a component of the annual phytomass growth, which changes throughout the tree's life (Briffa et al. 1998). It can provide an indication of net primary production as influenced by climate forcing (Graumlich, Brubaker & Grier 1989; Kirdyanov et al. 2003), as well as of carbon fluxes in terrestrial ecosystems (Leblanc 1990; Biondi 1999).

In addition to the assessment of such ‘traditional’ growth parameters, stable isotope ratios in tree-ring wood or cellulose are increasingly used to obtain retrospective insight into: (1) environmental conditions at the time of growth; and (2) ecophysiological processes affected by these conditions (Saurer, Aellen & Siegwolf 1997; Schleser et al. 1999; Treydte et al. 2001, 2006; Leavitt 2002; McCarroll & Loader 2004; Poussart, Evans & Schrag 2004). Stable isotopes in tree-rings combine the dendrochronological advantage of precise and accurate dating with the well-known sensitivity of carbon (δ13C) and oxygen (δ18O) isotope ratios to meteorological and environmental variations.

The isotopic signal is primarily imprinted on the photosynthates in the leaves during photosynthesis, and transported through the tree before it is laid down in the tree-ring archive. Because of the relationship between photosynthetic carbon isotope fractionation and the ratio of leaf intercellular and ambient CO2 concentration (ci/ca), the δ13C of newly assimilated organic matter can generally be used to characterize environmental effects on the physiology of photosynthesis. Stomatal closure because of restrictions in water availability, generally reduces ci, leading to an increase in δ13C of organic matter in different tissues such as leaves, tree-rings or phloem (e.g. Farquhar, O' Leary & Berry 1982; Korol et al. 1999; Keitel et al. 2003). Because light limitation of photosynthesis increases ci, δ13C can also depend on radiation (Leavitt & Long 1986; McCarroll & Pawellek 2001) under particular conditions, and combined influences of water and light availability have also been observed (Gessler et al. 2001). In addition, any other factor that affects photosynthetic capacity or stomatal conductance influences ci and thus photosynthetic carbon isotope fractionation. Moreover, the influence of environmental factors on mesophyll conductance also affects the carbon isotopic composition of newly assimilated organic matter (see e.g. Warren & Adams 2006).

The oxygen stable isotope composition (δ18O) or the oxygen isotopic enrichment above source water (Δ18O) of plant organic matter has been shown to provide additional information that can be used to distinguish effects of stomatal conductance (as a proxy for precipitation and air humidity) from effects of changes in photosynthetic capacity (as a proxy for light availability) on the δ13C of newly assimilated organic matter (Farquhar, Barbour & Henry 1998; Adams & Grierson 2001). Indeed, evaporative enrichment shares the dependence on stomatal conductance with the δ13C signature, but is not dependent on ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco) activity (Barbour & Farquhar 2000; Scheidegger et al. 2000). In addition, the evaporative enrichment of leaf water, which is imprinted on the newly assimilated organic matter with an equilibrium fractionation factor of 27‰, has been shown to provide information on transpiration rate (Wang & Yakir 2000; Barnard et al. 2007) as the transpirational advection of unenriched xylem water dilutes the 18O enrichment in the mesophyll (Péclet effect).

A mechanistic understanding of the physiologically based fractionation processes of stable isotopes affecting tree-ring isotope composition is a prerequisite for an accurate physiology-based interpretation of the information-rich tree-ring archive. While information on leaf-level processes has increased considerably within the last three decades (Farquhar et al. 1982; Farquhar, Ehleringer & Hubick 1989; Yakir et al. 1994; Flanagan, Brooks & Ehleringer 1997; Yakir 1998; Roden & Ehleringer 1999; Farquhar & Cernusak 2005; Barbour 2007; Barnard et al. 2007), we still lack a clear picture of the physiological mechanisms of isotope fractionation in downstream metabolic processes (e.g. Brandes et al. 2006) and how potential post-carboxylation fractionation processes are influenced by both environmental and plant internal factors (for an overview on the processes potentially influencing the carbon and oxygen isotope composition of organic matter, see Fig. 1).

Figure 1.

Summary of the plant-related processes that potentially influence the carbon and oxygen isotopic composition of organic matter. Carbon isotope fractionation processes are listed on the left side of the figure, and processes affecting the oxygen isotope composition are given on the right side. In addition to the listed fractionation and exchange processes, the carbon isotope composition of atmospheric CO2 and the oxygen isotope composition of the source water influence δ13C and δ18O of organic matter, respectively. More information is found in the text.

It was recognized only recently that post-carboxylation fractionations caused by equilibrium and kinetic isotope effects beyond CO2 diffusion and fixation by Rubisco are of importance in autotrophic and heterotrophic tissues because they result in differences in isotopic signatures among metabolites and in non-statistical intramolecular isotope distributions (Gleixner et al. 1998; Schmidt 2003; Tcherkez & Farquhar 2005; Brandes et al. 2006; Salmon et al., unpublished data), thereby potentially uncoupling the isotopic information reaching the tree-ring from leaf physiological processes.

Day–night variations in δ13C of phloem sugars caused by fractionation processes associated with the aldolase reaction in the chloroplast (Tcherkez et al. 2003, 2004; Gessler et al. 2008) might influence the tree-ring isotopic signal, as there is indication that rates of lignin (Rogers et al. 2005) and potentially also cellulose deposition may not be constant over the day. Based on such findings, Gessler et al. (2007a) suggested applying a mechanistic modelling approach according to Tcherkez et al. (2004) to characterise short-term variations of δ13C in phloem sugars, which are the carbon source for cellulose production.

As an additional complication for the interpretation of the oxygen isotope signature in tree-rings, some of the oxygen atoms of sucrose synthesized in the leaves are exchanged with non-enriched trunk water (i.e. source water) during cellulose synthesis (Sternberg, DeNiro & Savidge 1986; Roden & Ehleringer 2000). In general, it is assumed that approximately 42% of the oxygen is exchanged (Cernusak, Farquhar & Pate 2005), but it remains unknown whether this exchange rate is constant over the growing season or over longer time periods, or if it is affected by environmental conditions.

Thus, even though sophisticated mechanistic models exist to describe isotopic fractionation at the leaf level, most of the carbon and oxygen fractionation and oxygen exchange processes in downstream metabolic processes are poorly understood. Badeck et al. (2005) reviewed more than 80 publications for differences in δ13C signatures between organs caused by post-photosynthetic fractionation processes and showed that on average heterotrophic tissues were enriched in 13C by 1.26‰ as compared to leaves. In particular, in woody plants, the difference in carbon isotope signature between leaves and stems exhibited a strong variation among species, ranging from a relative enrichment of the woody tissues of more than 8‰ to a relative depletion of 1.8‰. In order to properly interpret the isotope information found in tree-rings, it is, consequently, essential to characterize to what extent post-carboxylation fractionation processes may influence the isotopic composition of this archive, potentially causing the isotopic signals to be at least partially uncoupled from leaf physiological processes.

To shed more light on such processes, we traced the carbon and oxygen isotope signals through Pinus sylvestris trees from the canopy to the trunk on different time-scales, by combining existing data sets from the same field site. The present article brings together results from previously published papers on fast turn-over organic matter pools such as leaf water-soluble organic matter (OM) and phloem sap OM, the δ13C and δ18O of which was assessed on the diel to the seasonal time-scales (Brandes et al. 2006, 2007; Barnard et al. 2007) with determinations of carbon isotope fractionation during respiration (Kodama et al. 2008) and analyses of δ13C and δ18O in tree-ring organic matter (Brandes 2007). By considering and discussing these results together, the present study provides a unique integrated overview of the fate of carbon and oxygen isotope signals within the model tree species P. sylvestris on time-scales ranging from hours to a growing season. Until now, the assessment of carbon and oxygen isotopic composition of tree organic matter has never been extended over such a range of temporal scales since either diel, day-to-day or seasonal variations have been focused on (Table 1). In addition, the simultaneous observation of various canopy and trunk organic matter fractions on such various time-scales has never been achieved until now. Therefore, relating this combined data set on P. sylvestris to findings from current literature on other tree species strongly improves the interpretation of δ13C and δ18O in tree-ring ecology on a plant physiological basis.

Table 1.  Selection of publications describing short-term variations in the natural carbon and oxygen isotope composition of organic matter in trees from diel to seasonal scales Thumbnail image of


In agreement with theory, daily (daytime) mean values of δ13C of the water-soluble leaf organic matter – which is considered to mainly represent newly assimilated sugars (Brandes et al. 2007) – were significantly correlated with photosynthesis-weighted ci/ca during nine consecutive days (Brandes et al. 2006). Over the diel course, the δ13C signature of soluble organic matter varied with amplitudes of up to 1.7‰ in the canopy (foliage and twig phloem) of P. sylvestris. The greatest 13C enrichment was recorded during the night or early morning (Fig. 2a,b; Brandes et al. 2006). This finding is consistent with observations by Gessler et al. (2007a) when assessing short-term variations in sugars transported in phloem sap of Eucalyptus delegatensis. With this species, the difference between the relatively 13C-enriched sugar transported during the night and the 13C-depleted sugar found in the light period amounted up to 2.5‰. These authors additionally observed that the carbon isotopic signature of leaf-exported sugars during the night was similar to that of the carbon released from transitory starch pools.

Figure 2.

Diel courses of carbon isotope composition (a, b) and oxygen isotope enrichment (c) or composition (d) in different chemical constituents and organs in Scots pine. (a) δ13C in the water-soluble organic matter of current (N) and previous year (N − 1) needles; (b) δ13C of phloem-transported organic matter from twigs at approximately 12 m height and from the trunk base; (c) measured evaporative enrichment (Δ18OL) in leaf water of previous year (N) needles and modelled values (Δ18Oes is obtained from a model assuming the leaves to be in isotopic steady state and calculating evaporative enrichment at the sites of evaporation; Δ18OLN considers non-steady state conditions and isotopic heterogeneity of leaf water because of a Péclet effect; the grey arrows indicate the difference between Δ18OL and Δ18Oes, which is assumed to be proportional to transpiration rate, see text); (d) measured δ18O of leaf water in N needles (δ18OL) compared with δ18O of water-soluble leaf organic matter (δ18Ows) and phloem sugars at the upper part of the trunk (δ18Op). The two arrows indicate the time-lags between the δ18O signal in the leaf water and the organic matter pools. (a) and (b) are redrawn from data published by Brandes et al. (2006); (c) and (d) are from data from Barnard et al. (2007).

During the day, starch accumulates in the leaf chloroplasts, but the nature of the control of this accumulation is still uncertain. The distribution of assimilates between starch and sucrose might be under internal control to suit the environmental conditions, especially day-length (Zeeman, Smith & Smith 2007). Other authors argue that transitory starch accumulates when the utilization of newly produced triose phosphates from the chloroplast begins to limit carbon assimilation rate, especially when photosynthesis is light saturated and stomatal conductance is high (Beck & Ziegler 1989). If triose phosphates are equilibrated with hexoses by the aldolase-catalysed reaction in the chloroplast, isotopically heavier hexoses will be incorporated into starch, while 13C-depleted trioses will be exported to the cytoplasm, because of the aldolase isotope effect (Gleixner & Schmidt 1997; Gleixner et al. 1998). This process results in carbon isotopic enrichment of chloroplastic starch and 13C depletion of cytosolic carbon during the day of up to 4‰ (Gleixner et al. 1998). In contrast, when starch is remobilized during the night, the sucrose exported from leaves is likely to carry the more positive signature of substrate starch (Tcherkez et al. 2004; Gessler et al. 2008).

As particular chemical compounds might not be deposited with constant rates over the day in tree-rings (Rogers et al. 2005), it might be assumed that such diel variations in δ13C of phloem sugars can strongly affect the isotopic composition of tree-rings (as discussed in Gessler et al. 2008). The distinct diel variations in δ13C observed in newly assimilated organic matter in the tree crown (1.7‰) were, however, strongly damped during phloem transport from the leaves to the trunk (<0.4‰; Fig. 2b). Therefore, day–night rhythm of starch storage and remobilization does not impede the interpretation of the tree-ring carbon isotope signal (Kodama et al. 2008). The observed lack of a day–night periodicity in δ13C of trunk phloem organic matter of P. sylvestris is in agreement with results from Betson et al. (2007) in Picea abies and from Gessler et al. (2007a) in E. delegatensis, which also showed diel variations <0.5‰. The damping of the diel δ13C variations in phloem-transported organic matter at the trunk base can be explained by a mixing of various carbon pools with different metabolic histories during phloem transport down the trunk, as discussed in detail for P. sylvestris by Brandes et al. (2006). This hypothesis is supported by findings of Keitel et al. (2003), showing that δ13C in trunk phloem sap organic matter integrates mean canopy ci/ca over several days in European beech.

We observed on average a 1.5‰13C enrichment in soluble organic matter pools from the leaves to the stem base, which was independent of short-term variations in canopy organic matter and of the damping of the 13C signature of soluble organic matter pools during transport towards and within the trunk (Fig. 2a,b; Brandes et al. 2006). We assume that carbon isotope fractionation in both autotrophic and in heterotrophic tissues during transport was responsible for this enrichment, as follows. The difference in δ13C between leaf soluble organic matter and sugars exported to the twig phloem observed by Brandes et al. (2006) and shown in Fig. 2a,b indicates preferential loading of 13C-enriched compounds into the sieve tubes, as previously hypothesized by Hobbie & Werner (2004). However, this mechanism does not explain the difference in δ13C between twig and trunk phloem (approximately 1.2‰; Fig. 2b). It is known that phloem transport is associated with continuous unloading and retrieval of sugars along the transport path (Van Bel 2003). A part of the phloem-unloaded sugars might undergo metabolic conversion, while the rest is reloaded into the sieve tubes. Metabolic branching and preferential reloading of 13C-enriched sugars into the transport phloem in twigs and stems (as suggested earlier for leaves) would cause the non-exported organic matter in twig and trunk tissues to be relatively 13C depleted, whereas the retrieved phloem sugars would become 13C enriched. In agreement with this hypothesis, Gessler et al. (2008) observed for Ricinus communis a 13C depletion of the lignified bulk stem tissue, compared to the phloem sugars. In P. sylvestris, however, the newly produced trunk wood is 13C enriched compared to phloem organic matter (see below) and we have no information as to whether the abundance of 13C-depleted resin in the bark (Stern et al. 2008) may isotopically compensate for that enrichment. However, the mechanisms of preferential reloading of 13C-enriched sugars could explain that carbon isotope composition changes during phloem transport, whereas the oxygen isotope composition is not affected (see below). CO2 fixation in the trunk by phosphoenolpyruvate carboxylase (PEPc) might also offer an explanation for the increase in δ13C during phloem transport, but would also result in a change in Δ18O, because the newly produced organic matter is assimilated in unenriched or only slightly enriched trunk water. PEPc fixation of CO2 fractionates against 12C by 5.7‰ (Farquhar 1983) thus producing 13C-enriched organic matter, part of which might be loaded into the transport phloem.

We also have to consider that the δ13C of phloem sap at the stem base integrates over the whole canopy including potentially 13C-depleted carbon from the lower canopy, as well as 13C-enriched sugars from sun-exposed leaves, whereas the twig phloem might be representative only for a small part of the crown. We must, however, conclude that this difference in canopy integration offers no explanation for the observed 13C enrichment in the trunk phloem: the twig phloem was collected from the upper sunlit part of the canopy and thus the sugars it contained should be rather 13C enriched compared to trunk phloem sugars. Furthermore, Brandes et al. (2006) observed no significant vertical gradient of δ13C in the leaf sugars within the sparse and short crown of the pine trees examined and concluded that the twigs selected for phloem exudation were representative for the whole canopy. Another factor to be discussed is the contribution of starch-derived remobilized storage carbon to trunk phloem sugars. Firstly, starch is generally enriched in 13C and secondly, it might originate from time periods with lower photosynthetic carbon isotope fractionation. While we cannot exclude the possibility of this mechanism being responsible for the observed basipetal δ13C gradient in the short-term assessment shown in Fig. 2, it does not explain the permanent relative 13C enrichment of the stem phloem during the growing season (see below and Brandes et al. 2007).

The continuous 13C enrichment of phloem-transported sugars challenges the calculation of intrinsic water-use efficiency or ci/ca values from phloem or tree-ring δ13C values, because deviations between leaf and trunk organic matter of up to 2.5‰ (Helle & Schleser 2004) have been observed. Assuming the 13C enrichment to be constant over time (but see below), temporal trends in intrinsic water-use efficiency will be still evident, but an offset between the values calculated from δ13C of organic matter at the tree base and leaf-level values has to be considered. When calculating the intrinsic water-use efficiency (A/gs; A: assimilation rate; gs: stomatal conductance) from the carbon isotopic signature according to Farquhar et al. (1989) and Duquesnay et al. (1998), a 2.5‰ deviation causes a difference in A/gs of more than 25 µmol mol−1 (with a δ13C of CO2 of −8‰ and a ca of 370 µmol mol−1). When relating this potential error to the mean values of A/gs (85 µmol mol−1) obtained by Duquesnay et al. (1998) from tree-ring carbon isotope data in beech between approximately 1860 and 1990 without considering the effect of post-carboxylation discrimination, it becomes obvious that the absolute ‘real’ values of A/gs might be overestimated by more than 20%. This uncertainty adds to the complication of calculating instantaneous water-use efficiency (A/E) from δ13C of plant material (Seibt et al. 2008).

There are, however, contradictory results with respect to the presence or lack of 13C enrichment during phloem transport, which make more systematic explorations necessary. In adult E. delegatensis, Gessler et al. (2007a) observed no significant difference in δ13C among phloem sugar collected from various positions along the tree axis over a 3 d period (mean difference in δ13C between phloem sugars at 26 and 0.5 m trunk height: <0.2‰). Moreover, 13C enrichment from the twig to the trunk phloem was observed in Fagus sylvatica only in July (Gessler, Rennenberg & Keitel 2004), but not in September.

Only recently, Kodama et al. (2008) showed that carbon isotope fractionation during trunk respiration can vary by up to 4‰ in P. sylvestris on a diel scale and that respired CO2 was in general 13C enriched compared to the organic substrates for respiration. In deciduous oak, comparable variations (approximately 3.3‰) were found on a seasonal scale (Maunoury et al. 2007). Duranceau et al. (1999) and Tcherkez et al. (2004) suggested respiratory carbon isotope fractionation to be caused by non-random distribution of 13C in organic matter substrates (Rossmann, Butzenlechner & Schmidt 1991) along with fragmentation fractionation during respiration. The CO2 released by the decarboxylation of pyruvate (via the enzyme pyruvate dehydrogenase; PDH) originates from the relatively 13C-enriched C-3 and C-4 atoms of a glucose molecule, whereas the CO2 emitted from TCA cycle reactions originates from the C-1, C-2, C-5 and C-6 atoms, which are relatively 13C depleted (Rossmann et al. 1991; Tcherkez et al. 2004). Any apparent 13C enrichment of CO2 compared to the organic respiratory substrate can thus be because of incomplete oxidation of glucose molecules, with a higher proportion of C-3 and C-4 atoms converted to CO2. The remaining C-1, C-2, C-5 and C-6 atoms will be incorporated via acetyl-CoA into fatty acids or via the TCA cycle into amino acids. Changes in the relative contributions of CO2 produced by PDH and by TCA reactions can thus result in changes in the 13C enrichment of respired CO2 with time. Fragmentation fractionation and its variation with time will clearly affect the δ13C of the total organic matter pool of a given tissue, because the relatively 13C-enriched CO2 is emitted, whereas the relatively depleted acetyl-CoA and further products remain. However, because the difference of the isotope composition between the source (i.e. the sugar substrates entering glycolysis) and the products is caused by the fragmentation of a molecule rather than a result of kinetic or equilibrium isotope effects, fragmentation fractionation might not directly and immediately affect the δ13C of the remaining pool of not respired sugar substrates (i.e. glucose, fructose). This pool, part of which might be reloaded as sucrose into the phloem sieve tubes, would be affected only if there were additional biochemical fractionations associated with the PDH or other biochemical reactions in the glycolysis pathway, and/or if the pool was supplied not only via the phloem but also via chemical conversions from TCA cycle intermediates or acetyl-CoA.

One possibility to evaluate the carbon isotope information and potential deviations from the leaf-level δ13C values of organic matter transported along the trunk is to simultaneously determine the δ18O signature, which shares the dependency on stomatal conductance with δ13C, as done previously by Cernusak et al. (2003) and Keitel et al. (2003). Newly produced assimilates are assumed to carry the oxygen isotope signature of the leaf (or needle) water at the time they were produced, with an equilibrium fractionation factor (εwc) resulting in carbonyl oxygen being approximately 27‰ more enriched than the water in which it was produced (Sternberg & DeNiro 1983; Yakir & DeNiro 1990). For pine, the measured evaporative enrichment above source water (Δ18OL) of mean needle water in both current (N) and previous year (N − 1) needles could be readily explained when applying mechanistic models (Fig. 2c; Barnard et al. 2007) according to Farquhar & Cernusak (2005). The best fit between measured and modelled values was obtained by applying a model which: (1) takes 18O fractionation during phase transition of water and diffusion of water vapour through the stomata and the boundary layer into account; (2) considers isotopic non-steady state; and (3) includes isotopic heterogeneity of leaf water (cf. Barnard et al. 2007). The isotopic heterogeneity is assumed to be caused by the advection of unenriched xylem water as opposed by the diffusion of 18O-enriched water isotopologues from the sites of evaporation (Péclet effect). An alternative model (according to Craig & Gordon 1965 and Dongmann et al. 1974), which does not include the Péclet effect, and thus calculates evaporative enrichment at the sites of evaporation, overestimated Δ18O during day-time by up to 15‰. The difference between measured values and Δ18O calculated from this alternative model (Δ18Oes) should be proportional to the transpiration rate, because the Péclet number (℘) is defined according to the following equation (Barbour et al. 2000):


where E is the transpiration rate (mol m−2 s−1); L is a scaled effective path length (m), a measure that describes the tortuous water pathway through the mesophyll; C is the molar concentration of water (mol m−3); and D is the diffusivity of H218O in water (m2 s−1). Although not strictly correct because of variations in temperature, molarity of the solution (cf. Cuntz et al. 2007) and plant–water relations (Keitel et al. 2006), L, C and D can be assumed to be constant over time periods between hours and a few days and, thus, any changes in ℘ during that time period should mainly be caused by variations in E. Therefore, a comparison between measured evaporative enrichment in leaf water or even organic matter fractions and modelled Δ18Oes might allow the reconstruction of transpiration rates.

The term 1 − Δ18OL18Oes represents the fractional difference between the modelled leaf water enrichment at the sites of evaporation (using the model according to Craig & Gordon 1965 and Dongmann et al. 1974 without considering the Péclet effect) and measured leaf water enrichment. 1 − Δ18OL18Oes corresponds to the proportion of unenriched water in leaf water (Barbour et al. 2000; Gan et al. 2002), and was indeed related to E (Fig. 3). Under day-time conditions that were close to steady state, we observed an overall increase of 1 − Δ18OL18Oes as transpiration increased, and values from both N and N − 1 needles were in good agreement with the predicted values calculated according to Barbour et al. (2000). From this observation, we conclude that there is a clear transpiration signal in the evaporative enrichment of leaf/needle water.

Figure 3.

Transpiration signal in the difference betweeen actual and modelled evaporative enrichment of leaf water. The relationship between the transpiration rate (E) of current year (N, open circles) and previous year (N − 1, filled squares) needles and the discrepancy between the observed enrichment of mean lamina leaf water (Δ18OL; cf Fig. 2c) and the steady-state prediction of water enrichment at evaporative sites (Δ18Oes) during day-time is given. The data are derived from the day-time values from Fig. 2c. The predicted values (N, solid line; N − 1, dashed line) calculated according to Barbour et al. (2000) are based on scaled effective path lengths (L) of 0.15 and 0.05 m for N and N − 1 needles, respectively. The figure is adapted from Barnard et al. (2007).

A next step to trace the oxygen isotope signal from the canopy to the tree-ring is to test whether the oxygen isotope signal is transferred from the leaf water to the organic matter pool in the leaves as suggested by theory (Sternberg & DeNiro 1983; Yakir & DeNiro 1990). Firstly, the diel rhythm and day-to-day differences that were found in leaf water δ18O values were also found in the fast turn-over organic matter pools in needles and stem phloem (Fig. 2d). This linkage was confirmed by the existence of a time-lagged correlation between leaf water and organic matter δ18O values (Fig. 4). By applying cross-correlation analyses (cf. Brandes et al. 2006; Barnard et al. 2007) for δ18O in different water and organic matter pools (needle water, needle water-soluble organic matter, phloem sugars at different positions along the tree axis), we found significant time-lags, which represent residence and transport times of newly produced assimilates (Fig. 4). In total, it took 1–2 d for the isotope signal to be transferred from the leaf water to the phloem-transported sugars at the base of the trunk. With the trees being 15 m high, phloem transport velocities thus amounted to approximately 0.8 m h−1, a value which is comparable to beech (0.5–1.0 m h−1; Keitel et al. 2003) and poplar (0.9–1.2 m h−1; Windt et al. 2006) and to the values calculated for a mixed coniferous forest (0.2–2 m h−1; Ekblad & Högberg 2001).

Figure 4.

Tracing the oxygen isotope signal through the tree. Significant time-lagged correlations between δ18O in different tree compartments (cf. Fig. 2d): needle water and water-soluble organic matter for current year (N) and previous year (N − 1) needles, phloem sap in the trunk bark at three sampling heights (10, 6 and 1.5 m height). The figure is adapted from Barnard et al. (2007).

Note that the damping of the diel amplitude of the δ18O signal from water to organic matter (Fig. 2d) was most likely because of the higher (i.e. longer) turn-over time of the organic matter fraction – which has been estimated for needle sugars in P. sylvestris to be between 5 and 12 h (Barnard et al. 2007; Kodama et al. 2008) – compared to the leaf water. The low δ18O values at the beginning of the measurement period shown in Fig. 2d (second half of 6 June and first half of 7 June) are reflected in the leaf soluble organic matter 6–12 h later, and in the phloem organic matter approximately 24 h later. Finally, when calculating the photosynthesis-weighted 4 d average for the δ18O of leaf water and leaf water-soluble organic matter, a difference between these pools close to the expected 27‰ was obtained (Barnard et al. 2007). In R. communis, Barbour et al. (2000) found time-lags of approximately 4 h between a change in VPD and the subsequent change in Δ18O of phloem organic matter. When comparing these values with those of pine, it has to be considered that the turn-over of leaf sugars is much faster in Ricinus (turn-over time of leaf water-soluble organic matter: 2 h; Gessler et al. 2007b) compared to tree species. Thus, we conclude that the isotopic signal is transferred from leaf water to organic matter, albeit with a damping of short-term variations and with time-lags.


Brandes et al. (2007) characterized carbon and oxygen isotope signatures in twig and trunk base phloem sugars of P. sylvestris over more than 1 year. In contrast to studies on F. sylvatica (Keitel et al. 2003, 2006; Scartazza et al. 2004) and Nothofagus solandri (Barbour et al. 2005), the authors did not observe a direct relation between carbon isotopes and environmental [such as air temperature, relative humidity or photosynthetically active radiation (PAR)] or physiological parameters (such as stomatal conductance or transpiration) during the growing season. A correlation was, however, detected on a day-to-day scale between δ13C and ci/ca for needle water-soluble organic matter (Brandes et al. 2006), but values for ci were not available on the seasonal scale. A combined influence of stomatal conductance and assimilation rate on ci/ca and thus on δ13C should be indicated by the relation between δ18O or Δ18O and δ13C according to the conceptual model of Scheidegger et al. (2000). δ18O and Δ18O of recently synthesized organic matter should share – according to that model – the dependence on stomatal conductance with δ13C, but should not be dependent on Rubisco activity. However, a prerequisite for Scheidegger's model is that stomatal conductance is strongly controlled by rH or VPD (see also discussion by Brandes et al. 2007). However, Brandes et al. (2007) showed that this was not strictly the case for the pine trees examined, and thus the Scheidegger model is here not applicable to differentiate between stomatal and carboxylation effects on δ13C.

To overcome this constraint, Brandes et al. (2007) applied a multiple regression model with canopy stomatal conductance (Gs) and PAR (as a proxy for assimilation rate) as independent variables to explain the variability of the carbon isotope composition of phloem organic matter during the entire year. However, such an approach did not reveal any significant results. It might be assumed that the varying influence of the two determinants on ci/ca during the year as previously observed by Keitel et al. (2006) as well as the influence of additional parameters (e.g. temperature and mesophyll conductance) prevents such multiple regression models from yielding significant results.

Another hypothesis for the lack of correlation between environmental and/or physiological factors and δ13C is that post-carboxylation fractionation processes that are related to sugar export from the leaf to the sieve tubes may vary during the year and be responsible for the uncoupling of phloem δ13C from leaf-level processes in P. sylvestris. Futhermore, when comparing δ13C of twig and stem phloem determined during more than 1 year (Fig. 5), it is not only obvious that sugars in the sieve tubes become 13C enriched during basipetal transport: we also see that this enrichment is not constant but varies over time. Figure 5 also indicates that the 13C enrichment between twig and trunk phloem has a tendency to increase with decreasing δ13C. A more or less constant contribution of 13C enriched carbon fixed by PEPc to the phloem-transported organic matter throughout the growing season, could explain such a pattern. As discussed earlier, the reasons for the 13C enrichment during phloem transport are, however, not yet resolved in detail and the continuous unloading and reloading of sugars from and to the sieve tubes and related metabolic processes (Gessler et al. 2004) might also contribute. Even less information is available on the external or plant internal factors that cause this enrichment to vary over the year and between years. As a consequence of this change, the direct relationship between ci/ca at the leaf or canopy level and δ13C of phloem sugars in P. sylvestris weakens with increasing transport distance along the trunk. There are only few other studies comparing δ13C in phloem organic matter along plant stems or trunks (e.g. Pate & Arthur 1998; Gessler et al. 2004, 2007a, 2008; Cernusak et al. 2005) and no longer-term seasonal data are available on the relation between δ13C of twig and trunk phloem organic matter. In order to judge whether the observations of temporal variations in the 13C enrichment associated with phloem transport made here are only valid for P. sylvestris or if these results can be generalized, we need further comparative studies with different species.

Figure 5.

Relation between the isotope composition of transported sugars in twig and trunk phloem (trunk base). The data cover a period from 15 July 2003 (δ13C) or 24 September 2003 (δ18O) to 29 December 2004 with biweekly sampling. Twigs were collected at approximately 10 m height, and trunk phloem was obtained at 1.5 m height. Regression lines and 95% confidence intervals are shown. Bold lines indicate the 1:1 line. Each data point represents a mean value from 8–16 trees examined at given time point. Data are from Brandes et al. (2007).

In contrast to carbon isotope composition, the oxygen isotope composition (δ18O) or enrichment above source water (Δ18O) of phloem sugars in the twigs and trunk base was significantly negatively correlated with relative humidity, with a time-lag of 1–2 d (depending on the sampling position) during the entire year (Brandes et al. 2007). In addition, a steady-state evaporative enrichment model accounting for the Péclet effect and for εwc explained approximately 65 % of the seasonal variation of Δ18O in phloem-transported sugars, when transport times were taken into account (Brandes et al. 2007). Moreover, no significant change in δ18O values of sugars occurred during phloem transport (Fig. 5), which is comparable to the observations of Cernusak et al. (2005) with Eucalyptus globulus. These authors compared the 18O enrichment of phloem organic matter at six positions in the canopy and along the trunk, and did not find any significant variation. The finding of no basipetal variation is also well in agreement with results from Gessler et al. (2007b) who found that the mean diel oxygen isotope signal found in leaf sugars was preserved during phloem transport in R. communis.


Until now, we followed the carbon and oxygen isotope signals through the pine tree in short-lived organic compounds, and observed the alterations of these signals during transport, mainly for carbon isotopes. The next step is to determine to what extent the phloem sugar isotope signal is preserved when sucrose, the major phloem transport substance in P. sylvestris, is converted to whole wood or to wood constituents like cellulose. To address this question, firstly, we dated intra-annual tree-ring sections by measuring high-resolution tree radial increment with dendrometers (Brandes 2007). Secondly, we compared the isotopic composition of phloem-transported organic matter to tree-ring material at the time when it was formed during the growing season. While short-term variations of δ13C of phloem sugars were not imprinted on wood, the seasonal courses of δ13C in phloem sugars and in the whole wood of the tree-ring followed a similar pattern (Fig. 6a). In both carbon pools, a trend for δ13C to increase from the beginning of the growing season until approximately DOY 200 (mid-July), and then to decrease until the end of the growing season was observed. An analysis of the correlation between δ13C of phloem organic matter and that of tree-ring revealed a correlation coefficient r of 0.67, which increased slightly to 0.69 when a time-lag of 2 weeks between phloem δ13C and tree-ring δ13C was assumed. The seasonal pattern observed here is comparable to that found by Barbour, Walcroft & Farquhar (2002) for tree-ring cellulose of Pinus radiata. In addition, whole wood from tree-rings was enriched by approximately 1.4‰ above phloem organic matter. Klein et al. (2005) observed tree-ring whole wood to be enriched by 2‰ compared to needle tissue in Pinus halepensis. In contrast, Cernusak et al. (2005) reported the carbon isotope ratio of phloem sugars to be similar to that of newly developing xylem tissue, but to be enriched by 0.8‰ compared to mature woody tissues in E. globulus. This difference was supposed to be caused by higher amounts of 13C-depleted lignin present in the older xylem tissues. Terwilliger et al. (2001b) proposed that the processes causing heterotrophic tissues to be 13C enriched compared to source leaves take place within the sink tissue. If true, this hypothesis requires that sink tissues (e.g. wood) are 13C enriched compared to the carbon delivered in the phloem sap. Cernusak et al. (2005) rejected this hypothesis for E. globulus, because they found no such enrichment in newly produced wood compared to phloem sugars. We, however, showed here that such fractionations in the heterotrophic trunk can occur, even though the particular chemical reactions involved remain unclear. Furthermore, it remains to be clarified how the 13C enrichment in CO2 respired from the trunk compared to the organic substrate (Brandes et al. 2006; Kodama et al. 2008), leaving behind 13C-depleted total organic matter, is reconcilable with our finding.

Figure 6.

Carbon isotope composition and oxygen isotope enrichment in phloem sugars (circles) and in the tree ring (squares) at the stem base during the growing season 2004. Phloem-transported organic matter was collected as described in Brandes et al. (2007) every 2 weeks. Tree rings from three individuals laid down in 2004 were separated into 40 µm sections with a microtome. Sections were related to the time of the year the wood was laid down, by using dendrometer data recorded for 10 trees of comparable size and growing at the same site. The mean cumulative increment of the 10 trees was plotted against the day of the year (DOY) and a Boltzmann model was fitted to the data (Brandes 2007). The intra-annual dating procedure revealed that the 40 µm tree ring sections were representative for time periods between 2 d at the beginning and approximately 12 d at the end of the stem growth period. Tree ring samples were grouped in 2 week intervals in order to match the phloem samples. Whole wood δ13C values are displayed. The fitting curves for δ13C in wood and phloem organic matter are second-order polynomial models. For oxygen isotopes, the relation between whole wood and holocellulose was determined in subsamples from all trees and from different positions in late and early wood (Brandes 2007). Independently from late or early wood, holocellulose was enriched in 18O by 4.7‰ above whole wood. This difference was added to the measured Δ18O values for whole wood to present calculated Δ18O for cellulose in the figure. In addition, the proportional exchange of organic (carbonyl) oxygen atoms of sucrose with unenriched stem water during cellulose synthesis (pex) estimated according to Barbour & Farquhar (2000) and Cernusak et al. (2005) is displayed (dashed line). The arrows indicate time-lags between the oxygen isotope signal in phloem sugars and in the tree rings.

Oxygen isotope enrichment in trunk phloem organic matter increased at the beginning of the growing season and then decreased until the end (Fig. 6b). The seasonal pattern of tree-ring cellulose Δ18O (calculated from whole wood Δ18O; see legend of Fig. 6) was comparable to that of phloem organic matter, albeit showing a time-lag, lower 18O enrichment and damping of the amplitude. The two latter observations could be attributed to the exchange of organic oxygen with unenriched xylem water during cellulose synthesis. The calculated oxygen exchange rates pex ranged from 0.2 to 0.42 during the whole growing season, well within the range of literature values compiled by Cernusak et al. (2005). In Fig. 6b, an apparent time-lag of 2 weeks between phloem sugars Δ18O and tree-ring material is visible. As for δ13C, the highest correlation coefficient between phloem and tree-ring Δ18O was obtained with a time-lag of 2 weeks (r = 0.48, no time-lag; r = 0.73, time-lag of 2 weeks). This time-lag is, however, only a rough estimate because the time resolution of the phloem sap sampling is 2 weeks.

In summary, we have shown that the oxygen isotope enrichment in phloem organic matter records leaf-level physiology – with a time-lag – and that the isotopic information from the phloem sugar pool is transferred to wood or cellulose – again with a time-lag – and with an approximately 40% exchange of the organic oxygen with xylem water. Taking these parameters into account, we can recalculate leaf water enrichment from tree-ring cellulose Δ18O for given time periods of the growing season and subsequently calculate the fractional difference (as shown for leaf water in Fig. 3) between modelled leaf water enrichment at the sites of evaporation (Δ18Oes) and leaf water enrichment determined from the tree-ring. Figure 7 shows the relationship between E and the fractional difference between Δ18Oes (data from Brandes et al. 2007) and the leaf water values estimated from tree-ring cellulose (Δ18Otr) during one growing season. Most of the values scatter around the predicted line, indicating that via the Péclet effect a transpiration signal is encoded in the Δ18O of tree-ring cellulose. The black circles showing the highest deviation represent values from the beginning of the growing season, when organic matter supplied for wood synthesis is likely to originate from (old) starch storage pools and, thus, no influence from recent canopy processes can be expected.

Figure 7.

Transpiration signal in the evaporative enrichment of tree-ring cellulose. The relationship between transpiration rate (E) and the fractional difference between modelled oxygen isotope enrichment of leaf water at the sites of evaporation (Δ18Oes) and the leaf water values estimated from tree ring cellulose (Δ18Otr) is shown. To recalculate leaf water from tree ring cellulose evaporative enrichment, an 18O enrichment of carbonyl oxygen above water of 27‰ (εwc) and the partial exchange of organic oxygen with non-enriched xylem water have to be accounted for. In the y-axis caption, pex is the proportion of oxygen atoms exchanging with medium water in the developing tissue during cellulose synthesis (values shown in Fig. 6), and px is the proportion of unenriched source water in the developing tissue (i.e. water originating from the xylem rather than the leaf mesophyll) (Cernusak et al. 2005). We assumed px to be unity and considered a time-lag for leaf water isotope enrichment to be transferred to tree-ring cellulose. For this time-lag, we took into account a 2 d period for the organic matter to be transferred from the leaf water to the phloem organic matter at the trunk base plus a period of 2 weeks for the phloem 18O signal to be visible in the tree-ring (see Fig. 6). Δ18Oes was calculated from continuously measured values for transpiration, stomatal conductance, air temperature, wind speed, water pressure deficit of the air according to Brandes et al. (2006), assuming water vapour to be in isotopic equilibrium with source water. The black circles indicate time points at the beginning of the growing season; the white circles indicate time points during the rest of the growing season. The dotted line indicates predicted values based on the calculation of L (= 0.055 m) from Brandes et al. (2007).


For the first time, we have traced the carbon and oxygen isotope signals from the leaves to the tree-ring wood or cellulose in a particular species, P. sylvestris (cf. Table 1). We showed that short-term variations in δ13C in sugars in the canopy caused by transitory starch accumulation and remobilization do not influence the interpretation of the carbon isotopic signal in tree-rings in this species, but that variable 13C enrichment of sugars related to phloem loading and transport partially uncouple the tree-ring carbon isotope signal from ci/ca at the leaf or canopy level. However, this observation should not directly be transferred to other species, as the 13C enrichment during phloem transport might be species dependent (Gessler et al. 2008). In addition, we still lack knowledge about which factors determine transport-related fractionation, and we cannot exclude the influence of environmental factors. We also do not want to dispute here that δ13C analyses of tree-ring, leaf and phloem organic matter have been successfully applied to describe plant reactions to environmental constraints. However, species-specific problems may occur, as shown here for P. sylvestris, especially when intra-annual high-resolution tree-ring data are used. The use of additional isotope proxies might help to overcome – at least partially – such constraints. In P. sylvestris, the oxygen isotope signal was transferred from the leaf water to the tree-ring with the expected enrichment of 27‰ from water to carbonyl oxygen, with time-lags caused by turn-over and transport time, and with an approximately 40% exchange between organic oxygen and xylem water oxygen during cellulose synthesis. By accounting for these factors, we were able to extract a transpiration signal from Δ18O in tree-ring cellulose with an intra-annual resolution. However, one major constraint to the application of this approach to paleoclimatic studies is the need to know the source water signature to calculate 18O enrichment. A promising method for separating source water isotopic signature from plant evaporative information is the assessment of the position-specific isotope signature in trunk cellulose. Like the hydrogen atom (Augusti, Betson & Schleucher 2006), the oxygen bound to carbon-2 of the glucose monomers of cellulose (Sternberg, Anderson & Morrison 2003) also exchanges strongly with source water during cellulose synthesis. Therefore, the respective isotopomer abundances can be expected to be independent of all leaf-level processes and to depend exclusively on source water isotope abundance (Augusti & Schleucher 2007). On the other hand, the oxygen (and hydrogen) isotopic information related to carbon-6 is assumed to mainly represent leaf-level physiological processes. Thus, position-specific analysis can open a way to reconstruct the source water isotope abundance, as well as the evaporative enrichment of water in the leaf, allowing precipitation and/or temperature reconstruction and resolving the complication of overlapping environmental and physiological influences in tree-ring isotope signals (Augusti et al. 2006).