This study investigated the influence of climate on the carbon isotopic composition (δ13C) and oxygen isotopic enrichment (Δ18O) above the source water of different organic matter pools in European beech. In July and September 2002, δ13C and Δ18O were determined in phloem carbohydrates and in bulk foliage of adult beech trees along a transect from central Germany to southern France, where beech reaches its southernmost distributional limit. The data were related to meteorological and physiological parameters. The climate along the transect stretches from temperate [subcontinental (SC)] to submediterranean (SM). Both δ13Cleaf and Δ18Oleaf were representative of site-specific long-term environmental conditions. δ13C of leaves collected in September was indicative of stomatal conductance, vapour pressure deficit (VPD) and radiation availability of the current growing season. Δ18O was mainly correlated to mean growing season relative humidity (RH) and VPD. In contrast to the leaves, δ13Cphloem varied considerably between July and September and was well correlated with canopy stomatal conductance (Gs) in a 2 d integral prior to phloem sampling. The relationship between δ13C and Δ18O in both leaves and phloem sap points, however, to a combined influence of stomatal conductance and photosynthetic capacity on the variation of δ13C along the transect. Δ18Ophloem could be described by applying a model that included 18O fractionation associated with water exchange between the leaf and the atmosphere and with the production of organic matter. Hence, isotope signatures can be used as effective tools to assess the water balance of beech, and thus, help predict the effects of climatic change on one of the ecologically and economically most important tree species in Central Europe.
One major reason for these difficulties might be that, until now, ecophysiological studies on δ13C in plant tissues have mainly focused on the total carbon of leaves and wood (Saurer, Aellen & Siegwolf 1997; Hultine & Marshall 2000; Fotelli et al. 2003). Bulk leaf and wood material, however, consist of a complex mixture of carbon with different turnover times and metabolic history. Structural carbon in deciduous trees is likely to carry an isotopic signature influenced by both the present and previous growing seasons, and the effect of carbon storage and remobilization (Kozlowski & Pallardy 2002; Helle & Schleser 2004) with related carbon isotope fractionation poses the main difficulty to correlating δ13C signature in the total carbon fraction with meteorological or pedospheric parameters (Brendel 2001). However, because of the continuous incorporation of newly fixed carbon into structural organic matter, the total organic carbon in leaves and other tissues might – beyond all limitations – still provide a useful integrative measure of environmental influence on ci/ca over a longer term (Fotelli et al. 2003).
In contrast, the study of δ13C of recently fixed carbon has been shown to give insight into the prevailing environmental conditions at the time of the sampling (Scartazza et al. 2004) and, as a consequence, the isotopic composition of leaf sugars (e.g. Brugnoli et al. 1988) and phloem-allocated carbon (e.g. Gessler et al. 2001; Cernusak et al. 2003a; Keitel et al. 2003) is increasingly in the centre of attention. Whereas δ13C of soluble sugars of the foliage may be only representative of that part of the canopy from where the leaves were harvested, the isotope signatures of phloem-allocated sugars are thought to integrate over the whole canopy.
In a number of recent studies, δ13C of the phloem sap provided a strong guide to ci/ca or stomatal conductance during the current growing season (Pate & Arthur 1998; Cernusak et al. 2003a; Keitel et al. 2003; Scartazza et al. 2004), which makes phloem isotope signatures a powerful tool for studying physiological processes related to current environmental conditions [e.g. vapour pressure deficit (VPD), soil water availability] and eliminates confounding factors of stored and reused carbon – if sampled well after the onset of the growing season.
δ13C in recently fixed carbon integrates over different time scales depending on the material sampled (leaf sugars, phloem sugars), the position of sampling (twigs in the canopy, trunk phloem) and the species. Phloem sugars collected at the stem base have been shown to be indicative of canopy ci/ca or canopy stomatal conductance (Gs) with a time lag of 1 to 3 d depending on the species and tree height (Keitel et al. 2003; Gessler, Rennenberg & Keitel 2004b; Barbour et al. 2005; Brandes et al. 2006).
As both stomatal conductance and photosynthetic capacity have an influence on ci/ca, the main determining factor of δ13C in plant organic matter cannot be easily identified.
The oxygen isotopic composition (δ18O) or the oxygen isotopic enrichment above the source water (Δ18O) of plant organic matter has been shown to provide additional information to separate the effects of stomatal conductance from the influence of changes in photosynthetic capacity on δ13C (Adams & Grierson 2001), because it 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, a general synthetic model to describe Δ18O of plant organic material has been recently established, which includes 18O fractionation associated with phase transition, diffusion of H218O through the stomata and boundary layer and production of organic matter as well as the isotopic inhomogeneity of water within the leaf (Farquhar & Lloyd 1993; Barbour & Farquhar 2000; Roden, Lin & Ehleringer 2000; Farquhar & Cernusak 2005).
The combined analysis of carbon and δ18O in phloem-transported organic matter has been successfully applied for studying the short-term effects of environmental parameters – as modified by aspect and silvicultural treatment – on stomatal conductance in European beech (Keitel et al. 2003). In contrast, δ13C in the total leaf organic carbon did not reflect differences in meteorological and pedospheric parameters between a dry and wet slope of the narrow valley in Southern Germany studied (Gessler et al. 2001; Keitel et al. 2003).
In general, knowledge of the short-term (days) and longer-term (growing season) response of tree species to the variation in environmental variables like air temperature, soil water availability or radiation interception is crucial in order to estimate the effect of future climatic change on vegetation. Thus, we aimed to test if the combined analysis of carbon and oxygen isotopes in phloem organic matter and bulk leaf material could be used as a tool to characterize the water relations of European beech on a larger area scale. To exclude the influence of variation in δ18O of the source water, we calculated the 18O enrichment (Δ18O) of organic matter above the source (xylem) water. We designed a transect study with 12 field sites spanning over c. 875 km from Central Germany (Thüringen) to Southern France (Alpes de Haute Provence), where the area of distribution of this species is limited by water availability to validate the results obtained in a local climatic gradient (Gessler et al. 2001; Keitel et al. 2003; Gessler et al. 2004b).
Based on the previous observations, our main working hypotheses were:
1Because of the complex mixture of carbon with different turnover times and metabolic history in bulk leaf organic matter, the δ13C of this fraction is not representative for changing environmental, and thus, physiological conditions in beech along the selected transect. The assessment of the oxygen isotopic enrichment might, however, provide additional long-term integrating environmental information.
2Phloem organic matter is a short-term integrating indicator for Gs for the trees from all field sites selected at different times during the growing season. If Gs is not a direct proxy for ci/ca under the prevailing environmental conditions (i.e. if additional factors related to Rubisco activity and carboxylation influence internal CO2 concentration), this will be revealed by the relation between δ13C and the oxygen isotopic enrichment. In addition, we hypothesized that Δ18O of the phloem organic matter along the transect can be described by applying the model of Farquhar & Lloyd (1993) that takes into account the isotopic fractionation associated with H218O diffusion, with phase transition from liquid water to vapour and with the exchange between carbonyl oxygen and water as well as the isotopic inhomogeneity in leaf water.
MATERIALS AND METHODS
Site descriptions and characteristics
Twelve field sites were chosen along a transect from mid-Germany (Mühlhausen, State of Thüringen) to Southern France [Montagne de Lure (MdL), Department des Alpes de Haute Provence] in order to achieve a climate gradient from temperate–semi-(or sub-) continental to submediterranean (SM) (Fig. 1). The distance between the outermost two sites is 875 km. To increase the variability of environmental parameters, we included in this transect (1) the two sites from the local (topo)climatic gradient caused by different exposure [south-west (SW) versus north-east (NE)] as described by Keitel et al. (2003); and (2) a height gradient at the southernmost site from 1030 to 1650 m above sea level (a.s.l.) Table 1 depicts the main characteristics of the field sites along the transect.
Table 1. Field sites along the transect from north-eastern to south-western Europe
Altitude (m a.s.l)
Pmean (L m−2)
Soil textural class
BA (m2 ha−1)
Leaves, phloem and xylem sap were harvested twice in 2002. The main site characteristics are listed: latitude and longitude, altitude, climate, long-time average mean annual air temperature (Tamean) and annual sum of precipitation (Pmean), slope angle and aspect, soil textural class, soil type and BA of the stand.
a.s.l., above sea level; SC, subcontinental; SW, south-west; NE, north-east; SO, suboceanic; NW, north-west; SM, submediterranean; S, south; MdL, Montagne de Lure; E, east; BA, basal area.
W to SW
W to NW
Mühlhausen (Thüringen, Germany)
The field site east of Mühlhausen is located in the community forest of Mühlhausen, ‘Mühlhäuser Hardt’. The forest mainly consists of 100-year-old beech trees (90%) with an average height of 29 m and some admixed species, mainly Fraxinus excelsior and Acer pseudoplatanus.
Gasseldorf (Bavaria, Germany)
The field site in the Fränkische Schweiz is located on a slope near Gasseldorf close to Forchheim. The forest is between 80 and 100 years old and contains mainly European beech (95%) with some Picea abies (5%).
Höglwald (Bavaria, Germany)
The field site Höglwald is situated 50 km west–north-west (NW) of Munich. The soil is very acidic on the surface (pH 2.7–3.6) with the pH increasing to 4.5 at 2 m depth. The whole forest consists of c. 50%Fagus sylvatica and 50%P. abies. We selected a beech-dominated part of the forest (Gessler et al. 1998). The average height of beech is 32 m.
Tuttlingen (Baden-Württemberg, Germany)
Tuttlingen-Möhringen is located in southern Germany, about 100 km south–SW from Stuttgart in a low mountain range (Schwäbische Alb) and is the site of a previous study (Keitel et al. 2003).
The two sites examined are located on the opposing slopes (not more than 1000 m apart) of a single, narrow valley. One experimental site faces to the NE and the other to the SW. On both sites, European beech is the dominant species making up > 90% of the total basal area (BA) of adult trees. The average age of the adult beech trees is 70–80 years with a mean height of between 25 and 27 m (for a detailed site description, see Gessler et al. 2004a).
Renquishausen (Baden-Württemberg, Germany)
The forest at this site is dominated by 90- to 115-year-old beech trees (85%) with scattered F. excelsior, Acer campestre, Prunus avium and Pinus sylvestris (15%) and has an average height of 28–30 m.
Dietenbach (Baden-Württemberg, Germany)
The field site Dietenbach is located close to Freiburg with a forest consisting of F. sylvatica (c. 45%), F. excelsior (c. 12%), A. pseudoplatanus (c. 12%), Quercus rubra (c. 11%), Carpinus betulus (c. 8%) and Quercus robur (c. 7%).
Hofstetten (Solothurn, Switzerland)
The field site in Hofstetten (close to Basel, Switzerland) is part of the Swiss Canopy Crane Project, University of Basel. The forest is about 120 years old with tree heights between 32 and 38 m. The stand is dominated by F. sylvatica L., Quercus petraea and Q. robur with subdominant representatives of other deciduous species and conifers (Pepin & Körner 2002).
Grenoble (Isère, France)
This field site is located on a slope c. 12 km south of Grenoble. The forest is mainly dominated by beech (70%) with scattered Quercus pubescens, A. campestre and P. avium. In the past, the forest was partly used as coppice, so that the trees are of varying age (40–80 years). Hence, only the oldest trees with a height of c. 25 m were sampled.
MdL (Alpes de Haute Provence, France)
The field sites are located on a mountain range (MdL) in the Provence close to St. Etienne-les-Orgues near Sisteron. Plant material was harvested from three field sites on the lower, middle and upper slopes of this mountain [referred to as MdL I (low), MdL II (middle) and MdL III (high)]. The 12-m-high, 90-year-old forest was used as coppice until 70 years ago and, then, transferred to high forest. Today, F. sylvatica is dominant with some Q. pubescens and Sorbus torminalis.
Phloem, leaves and xylem sap were sampled in 16–21 July and 10–15 September 2002.
Phloem sap collection (bleeding and exudation techniques)
The phloem sap was collected at about 0900 h Central European Time (CET) at breast height of six adult beech trees per site according to the method described by Pate & Arthur (1998) for Eucalyptus globulus and by Gessler et al. (2001) for European beech. The bark was cut on a length of c. 5 cm to the depth of the wood at about 15% to the horizontal, using a single-sided razor blade. The ‘bleeding’ phloem sap was either collected immediately with a Pasteur pipette after the incision was made or not at all. Because no bled phloem sap could be obtained at the sites at the MdL, a modified phloem exudation technique as described by Gessler et al. (2004b) was applied. Small pieces of bark [(c. 150 mg fresh weight (FW)] with a diameter of about 1.0 cm were collected from the stem at breast height using a corer, and washed with double-demineralized water. Subsequently, the bark pieces were placed in 6 mL vials containing 2 mL double-demineralized water and left for 5 h. Previous studies (Schneider et al. 1996) with beech and spruce showed that contamination of phloem exudates with cellular constituents using the exudation technique can be neglected under the experimental conditions applied. Additionally, Gessler et al. (2004b) showed that the isotopic composition of the phloem samples is not affected by the sampling technique, which is a prerequisite for the comparison of the different phloem samples.
Leaf and xylem sap collection
Sun branches were collected from the ground using extendable loppers. Six to eight leaves were harvested evenly distributed over the branch and frozen in liquid nitrogen until further analysis.
The δ18O of the source water is one determining factor of the δ18O in organic material. To rule out the effects of differing δ18O of the source water, the δ18O of xylem sap of the trees at the respective sites, which is assumed to carry the same isotopic signature as the source water (Yakir 1998), was determined. For the extraction of xylem sap samples, 1 cm of the bark was removed at the cut end of the branch. A PE-tube was then fitted onto the shoot, which was inserted via a needle into a vial with an airtight seal. The connection to a vacuum pump was made using the same setup (PE-tube and needle inserted into the same vial). A gentle vacuum was applied to the branch, and the leaves were subsequently cut off the end of the branch to facilitate the xylem sap collection. The xylem sap was frozen in liquid nitrogen and stored at −80 °C until analysis.
Transpiration and evaporation
The daily sums of transpiration of beech trees on a per-square-metre-ground-floor basis and ground evaporation were calculated using the water balance model WBS3, a forest-hydrological model that requires daily mean values of air temperature and daily total precipitation as meteorological inputs (Matzarakis et al. 2000). The time-independent input variables of the WBS3 simulations are as follows: BA of the stand, the percentage of deciduous and coniferous trees, type of soil, maximum useable water storage capacity of the soil, slope angle, slope direction and geographical latitude. For evaporation, transpiration and interception of forests, validations of WBS3 showed a good agreement between the results from model calculations and measurements for different areas and slopes (Fritsch 1998; Matzarakis et al. 2000). Recently, a comparative assessment of the WBS3 and the more complex BROOK90 yielded satisfactory results. Transpiration rates determined with the WBS3 model were also compared with the transpiration rates calculated from sap flow measurements with Granier style sensors. For beech forests located on the two aspects of the experimental site Tuttlingen, we found a reasonably high correlation between daily sums of transpiration calculated from the WBS3 model and those determined from sap flow measurements (Nahm et al. 2006).
Because stomatal conductance is one of the factors determining δ13C of organic carbon, the mean daily Gs (in millimoles per square metre per second) (Table 2) was calculated using a simplified Penman–Monteith equation as recently applied by Keitel et al. (2003) according to Pataki, Oren & Phillips (1998):
Table 2. Means and sums of meteorological and physiological parameters characterizing the field sites during the growing season 2002 (May to September)
R (L m−2)
TR (L m−2)
EV (L m−2)
GS (mmol m−2 s−1)
Ta, mean air temperature; RH, mean relative humidity; R, sum of rainfall; VPD, mean vapour pressure deficit; Sun, sum of sunshine hours; TR, sum of transpiration; EV, sum of evaporation; Gs, mean canopy stomatal conductance; NE, north-east; SW, south-west; MdL, Montagne de Lure.
where Gc is the canopy conductance, γ is the psychrometric constant (kPa K−1), λ is the latent heat of vaporization (J kg−1), ρ is the density of moist air (kg m−3), cp is the heat capacity of moist air (J kg−1 K−1), VPD is the water vapour pressure deficit (kPa), P is the atmospheric pressure (Pa), R is the gas constant (8.31 m3 Pa mol−1 K−1), T is the air temperature (K), TR is transpiration on a ground area basis as obtained from the WBS3 model output (L m−2 d−1) and converted to the unit kg m−2 s−1 and LAI is the leaf area index of beech (m2 m−2). This calculation assumes that canopy aerodynamic conductance is much larger than Gs as reported by Whitehead & Jarvis (1981) (i.e. aerodynamic resistance can be neglected compared to stomatal resistance).
The meteorological parameters [daily sum of precipitation, mean daily temperature, mean daily relative humidity (RH), sunshine hours] were provided by the ‘Deutsche Wetterdienst’ (DWD), the Meteorological Institute of the University of Freiburg, Meteo Suisse and Meteo France from stations at or nearby the mentioned field sites.
Determination of sugars in the phloem sap
For the determination of soluble carbohydrates, 5–10 µL of phloem sap was diluted 1:1000 and phloem exudates 1:50 with demineralized water. Aliquots of 100 µL were injected into a high-performance liquid chromatography (HPLC) system (DX 500; Dionex, Idstein, Germany) as described by Gessler et al. (2002). Separation of sugars was achieved on a CarboPac 1 separation column (250 × 4.1 mm; Dionex) with 36 mm NaOH as an eluent at a flow rate of 1 mL min−1. Carbohydrates were measured by means of a pulsed amperometric detector equipped with an Au working electrode (DX 500; Dionex). Individual carbohydrates that eluted 8–16 min after injection were identified and quantified by internal and external standards.
Isotope analysis and definitions
The leaf material was oven-dried for 3 d at 65 °C, ground and homogenized with a ball mill to a fine powder, and aliquots were transferred into tin (δ13C analysis) or silver (δ18O analysis) capsules. The phloem sap was pipetted into the tin (δ13C analysis) or silver (δ18O analysis) capsules and oven-dried for 60 min at 65 °C. Dried samples of ‘phloem-bleeding sap’ (5–10 µL), ‘phloem exudates’ (30–100 µL) and leaves (1–2 mg) were combusted in an elemental analyser (NA 2500; CE Instruments, Milan, Italy) for δ13C and in a high-temperature conversion/elemental analyser (TC/EA Finnigan 165 MAT GmbH, Bremen, Germany) for δ18O analyses, both coupled to an isotope ratio mass spectrometer (Delta Plus; Finnigan MAT GmbH) by a Conflo II interface (Finnigan MAT GmbH) as described in detail by Gessler et al. (2004b). The tin capsules contained 0.5 mg of chromosorb (Chromosorb 102, 60–80 mesh; Sigma-Aldrich Chemie GmbH, Munich, Germany) to allow the proper combustion of phloem organic matter in the elemental analyser. For the analysis of δ18O in the xylem sap, an aliquot (1 µL) was pipetted into silver capsules and sealed. As evaporation enriches the water with the heavier oxygen isotope, the samples were prepared and dropped into the autosampler only a short time before pyrolysis of the sample. The isotopic values are expressed in delta notation (in ‰ units), relative to Vienna Pee Dee belemnite (VPDB) for carbon and Vienna standard mean ocean water (VSMOW) for oxygen. Isotopic compositions (δ13C, δ18O) are expressed as deviations from these standards:
where R is the isotope ratio (13C/12C or 18O/16O) of the plant material and the standard, respectively.
The δ18O of organic matter in leaves and phloem was expressed as an enrichment above the source water (Δ18O):
The δ18O of xylem water was regarded as a representative of the source water δ18O. For the calculation of Δ18O of phloem organic matter at both time points and of leaf organic matter in July, the xylem δ18O at the respective time points was chosen. For the calculation of the leaf Δ18O in September, a mean value of δ18O of the xylem water harvested in July and September was calculated for each site.
Oxygen isotopic theory and calculations
Steady-state enrichment of water at the evaporative site of the leaf (Δ18Oe) can be described according to Dongmann et al. (1974), where ɛ+ is the equilibrium fractionation (‰) between liquid water and water vapour, ɛk is the kinetic fractionation (‰) during water vapour diffusion (Farquhar et al. 1989), Δ18Ov is the isotopic difference (according to Eqn 3) of water vapour in the atmosphere compared to the source water and ea/ei is the ratio of ambient to intercellular water vapour concentration.
ɛ+ can be calculated according to Bottinga & Craig (1969) from a regression equation relating it to leaf temperature, T (in kelvin):
where rs and rb refer to stomatal and boundary layer resistance to water vapour (m2 s mol−1), respectively, and 32 and 21 are associated fractionation factors (Farquhar & Cernusak 2005) scaled to per cubic centimetre based on new determinations of the isotopic effect for the diffusion of H218O in air (Cappa et al. 2003).
The average lamina mesophyll water is, however, supposed to be less enriched than the water at the evaporative sites. In general, this difference is mainly supposed to be dependent on: (1) the diffusion of 18O-enriched water away from the sites of evaporation; and (2) the convection of unenriched xylem water via the transpiration stream in the opposite direction.
Steady-state enrichment of mean lamina leaf water (Δ18OL) depends on the steady-state enrichment at the evaporative site of the leaf (Δ18Oe) and on the lamina radial Péclet number (Farquhar & Lloyd 1993).
is defined as EL/CD, where E is the transpiration rate (mol m−2 s−1), L is a scaled effective path length (m), C is the molar concentration of water (mol m−3) and D is the diffusivity of H218O in water (m2 s−1).
According to Barbour & Farquhar (2000), the enrichment of 18O in organic matter above the source water in a particular tissue (Δ18Op) depends on the Δ18O of the average lamina mesophyll water (Δ18OL) on the proportion of oxygen atoms exchanging with water during the synthesis of cellulose (pex), the portion of unenriched source water in the tissue (px) and on the equilibrium fractionation between organic oxygen and water (ɛwc; + 27‰) (Yakir & Deniro 1990):
Δ18Op = Δ18OL (1 − pexpx) + ɛwc.(8)
pex can be calculated according to Barbour & Farquhar (2000) where y (Eqn 9) is the proportion of hexose phosphates that cycle through to triose phosphates before being incorporated into cellulose,
px is calculated according to Cernusak, Wong & Farquhar (2003b) where ρ is the proportion of 18O-enriched water coming into the cell either by diffusion or phloem transport as opposed to not enriched xylem-transported water (equals source water). Δ18Opw is the enrichment of phloem or diffusive water above the source water.
We applied Eqn 8 for predicting the Δ18O of phloem organic matter for the 12 field sites from the mean daytime values of environmental and physiological parameters determined or modelled for the 2 d period preceding the phloem sap sampling in July and September.
For this calculation, we made the following assumptions: water vapour was in isotopic equilibrium with the source water, as often assumed for European summer conditions (Förstel & Hützen 1983; Saurer et al. 1997). In that case, Δ18Ov in Eqn 4 equals –ɛ+. The difference between air temperature and leaf temperature was calculated according to a leaf energy balance model described in detail by Barbour et al. (2000a) and Cernusak et al. (2003a) assuming an average surface area of a single leaf of 10 cm2. The isothermal net radiation as an input parameter for the energy balance model was estimated as described by Barbour et al. (2000a) assuming canopy-averaged photosynthetically active radiation (PAR) to amount to 600 µmol m−2 s−1. The internal water vapour concentration (ei) was assumed to equal the leaf temperature-dependent saturation concentration.
For an effective length, we chose L = 0.015 m according to Farquhar & Cernusak (2005). As already mentioned, we assumed aerodynamic resistance << stomatal resistance as reported by Whitehead & Jarvis (1981) when calculating ɛk from Eqn 6. For phloem organic matter that originates directly from sugars produced in the leaves, we assumed px (Eqn 10) to be 0 as the oxygen isotopic enrichment of the reaction water (leaf water) equals ΔL.
We used Eqn 8 to relate Δ18O of total leaf (Δ18Op in Eqn 8) to phloem organic matter by substituting Δ18OL by Δ18O of phloem organic matter minus ɛwc as previously done by Cernusak, Farquhar & Pate (2005). This is justified as phloem organic matter reflects Δ18O of the water in which it was formed plus ɛwc. We assumed here pex to be 0.4 as previously observed for the leaves of E. globulus and considered to be close to the mean value of pex from all literature data (Cernusak et al. 2005). px was assumed to be 1, as the major part of the structural organic matter was produced during leaf formation with non-enriched water.
Statistical analyses were conducted using SPSS 10.05 (SPSS, Chicago, IL, USA) and NCSS 2004 (NCSS, Kaysville, UT, USA). Correlations between δ13C and environmental parameters were calculated using the bivariate correlation procedure. Regression lines (e.g. between δ13C and Gs and δ13C and Δ18O) were determined by linear or non-linear regression analysis. Differences in isotopic signatures between sites were calculated, applying one-way analysis of variance (anova) procedures, differences between two times for a particular site were assessed with Student’s t-test.
Climate, transpiration, evaporation and stomatal conductance
The mean air temperature (Ta) during the growing season 2002 differed by 5.8° between the warmest and the coldest sites (Table 2). RH showed a comparable but inverse pattern along the transect with the lowest mean values at the MdL and the highest values at the Renquishausen and the two Tuttlingen sites. As a consequence, VPD calculated from Ta and RH showed maxima and minima comparable to Ta. The lowest rainfall during the 2002 growing season was observed at the northernmost site (Mühlhausen), whereas at the Höglwald site (at a distance of c. 320 km) more than 300 mm of additional precipitation was detected. Sunshine hours, which were only available for the sites in France and Switzerland, increased in the southern direction.
The modelled sums of transpiration were highest at the Höglwald site and below 200 mm at the Renquishausen, Grenoble and MdL sites. Evaporation was calculated to amount between c. 40 mm (Mühlhausen) and 109 mm (MdL III). The mean stomatal conductance during the growing season ranged from 15 to 64 mmol m−2 s−1. The lowest values were detected at the French sites, whereas the highest conductance occurred at the Höglwald, Hofstetten and Renquishausen sites.
Carbon isotope signatures in total leaf and in phloem organic matter and phloem sugar concentrations
Significant differences in δ13C of the total foliar organic matter (δ13Cleaf) were observed between the sites of the transect at both sampling times (Fig. 2a), and values varied from −26.5 to −32‰. For a particular site, no significant intra-annual differences could be observed. Hence, the δ13Cleaf values from both dates were highly correlated (R2 = 0.90, P < 0.001). In both months, the foliage was most depleted in 13C at the Hofstetten field site amounting to −31.73‰ in July and −32.06‰ in September. The most positive δ13Cleaf values were found in leaves from the southernmost location at the MdL in France (c. −26.25 to −28.75‰).
In contrast to leaves, δ13C in phloem organic matter (δ13Cphloem) not only varied among sites, but also showed strong and significant differences between the two different sampling times (Fig. 2b). In the northern part of the transect, phloem organic matter was enriched in 13C in September as compared to July, whereas in the southern part either no significant change or depletion (Hofstetten) was detected. In July, phloem organic matter was most 13C enriched at the Grenoble, Hofstetten and MdL sites at the lowest elevation. In September, the highest δ13Cphloem values were observed at the two northernmost sites of the transect, and the strongest 13C depletion occurred in the phloem of beech from Höglwald and the medium elevation at the MdL. In contrast to the δ13Cleaf signatures, the Hofstetten site was not found to have significantly more negative values in phloem organic matter than the other sites. Carbohydrate concentrations in the phloem sap were analysed for the nine sites where bleeding sap could be obtained (Fig. 2c). Sucrose was found to be the dominant sugar in the collected phloem sap, and made up between 91.6 and 99.9% of total sugar in June and September 2002. Sugar concentrations mainly varied between c. 430 and 630 mm. Site-dependent patterns of δ13Cphloem and sugar concentrations were not synchronous, and hence, no significant regression relation was obtained between the two parameters (R2 = 0.06, P > 0.05).
Oxygen isotope signatures in xylem water and in total leaf and phloem organic matter
δ18O in xylem water (δ18Oxylem) varied significantly between the different sites and, for a particular site, between the two measurement dates (Fig. 3a). In both months, the strongest 18O depletion was observed at the northernmost site, whereas the sites in Switzerland and France generally displayed the highest δ18Oxylem. At both dates, there was a significant latitudinal and longitudinal influence on δ18Oxylem with signatures decreasing by between 0.7 and 1.1‰ with increasing eastern longitude and by 0.7–0.9‰ with increasing northern latitude. Because of these differences in source water, the δ18O of leaf and phloem organic matter was calculated as enrichment above the xylem water (Δ18Ophloem, Δ18Oleaf). In contrast to carbon isotopes in leaves, there were not only significant differences in Δ18Oleaf between sites, but also intra-annual differences for a particular site (Fig. 3b). On all sites, Δ18Oleaf increased from July to September by between 2.8 and 7.2‰. The ranking of sites according to their Δ18O values did, however, not change: in both months, the highest 18O enrichment was observed with beech from the French sites and from Mühlhausen and Höglwald. Thus, foliar δ18O showed strong correlation between the values from July and September (R2 = 0.76, P < 0.001) despite the intra-annual increase.
Δ18Ophloem varied between sites and sampling time (Fig. 3c). In both months, the highest 18O enrichment was observed at the northernmost site, whereas Δ18Ophloem was always lowest at one of the French sites. In contrast to foliage, where a general increase in Δ18Oleaf between July and September was observed for all sites, the intra-annual variation in phloem δ18O was greater. Whereas there was a significant increase in δ18Ophloem at most of the sites from July to September, no change was observed at Dietenbach and the lowest site at the MdL (MdL I) and enrichment even decreased at Hofstetten.
Correlation between isotope signatures and climatic and physiological parameters
Both δ13Cleaf and Δ18Oleaf showed site-typical patterns along the transect, even though Δ18Oleaf increased during the growing season. Thus, we assumed site-specific factors being responsible for this observation and related the isotopic signatures in total organic matter from leaves collected in September to altitude, latitude, longitude and to mean Ta, RH, sunshine hours, transpiration, evaporation and stomatal conductance of the current (2002) growing season (Table 3). Neither δ13Cleaf nor Δ18Oleaf was significantly related to latitude or longitude. δ13Cleaf was positively related to altitude with an increase in δ13C of 0.22‰ per 100 m. Altitude strongly influenced sunshine hours (R = 0.87, P = 0.05) and transpiration (R = −0.68, P = 0.01) all of which correlated significantly with δ13Cleaf. In addition, δ13Cleaf showed significant negative relation to RH and Gs. Δ18Oleaf correlated significantly with Ta, RH and VPD, whereas no correlation was observed between Δ18Oleaf and transpiration. The linear regression line between RH and Δ18Oleaf had a slope of −0.19‰/%.
Table 3. Correlation between δ13C and δ18O of total leaf organic matter (September values) and geographical, meteorological and physiological parameters
For correlation analysis, the mean values or sums of Ta, RH, VPD, Sun, TR, EV, Gs and R as displayed in Table 2 were used. In the first row, Pearson’s correlation coefficients are displayed; in the second row (in italics), significance levels are given. Bold figures indicate Pearson’s correlation to be significant.
Ta, mean air temperature; RH, mean relative humidity; VPD, mean vapour pressure deficit; Sun, sum of sunshine hours; TR, sum of transpiration; EV, sum of evaporation; Gs, mean canopy stomatal conductance; R, sum of rainfall.
In contrast to foliar isotopic composition, δ13Cphloem and Δ18Ophloem values were highly variable among sites and with time, suggesting short-term responses to environmental/physiological parameters.
Correlation analyses were performed for the means or sums of Ta, RH, sunshine hours, transpiration, evaporation and stomatal conductance 1–22 d prior to the time of phloem sap collection (from July and September) according to Keitel et al. (2003). Gs was found to be the main influence on δ13Cphloem along the transect (Table 4a). The highest correlation coefficient was determined for a time integral of 2 d prior to phloem sap collection. For longer time integrals, the correlation coefficient decreased steadily and was no longer significant for time integrals longer than 5 d. The only other parameter significantly correlated with δ13Cphloem was found to be transpiration. However, correlation coefficients were weaker and more or less constant over the observed time period.
Table 4. Correlation between (a) δ13C and (b) Δ18O in the phloem and physiological and meteorological parameters along the transect in July and September 2002, averaged over different time spans prior to phloem sap collection
The table shows Pearson’s correlation coefficients for bivariate correlation analysis between δ13C/δ18O of all phloem samples collected at 11 sampling locations in July plus 12 sampling locations in September and the respective environmental parameters as mean values of 1–22 d prior to the phloem sap collection; n = 23 (exception of sunshine hours only for sites in Switzerland and France, n = 5).
Gs, mean canopy stomatal conductance; TR, sum of transpiration; EV, sum of evaporation; Ta, mean air temperature; RH, mean relative humidity; VPD, mean vapour pressure deficit; R, sum of rainfall; Sun, sum of sunshine hours.
Figure 4 shows the regression line between Gs with a time integral of 2 d and δ13Cphloem with a slope of −0.01‰ mmol−1 m2 s. If July and September were analysed separately, a significant dependency between δ13Cphloem and Gs was found in July (R2 = 0.39, P < 0.05) but not in September (R2 = 0.11, P = 0.28).
Δ18Ophloem was not significantly related to any of the meteorological or physiological parameters tested (Table 4b).
Correlation between modelled and measured phloem Δ18O
According to Eqn 8, we calculated the Δ18O of phloem organic matter as expected from the mean daytime climatic conditions and physiological parameters during the 2 d before the July and September phloem sampling. The squares and circles in Fig. 5 show the calculated values assuming a scaled effective path length, L, of 15 mm plotted against the measured phloem Δ18O.
For four points in July – highlighted as black/white squares – the calculated values differ strongly from the measured ones. These four points represent the four southernmost sites assessed in July (Hofstetten, Grenoble and MdL I and III), which were then characterized by RH values < 70%, whereas for all other sites in July and all sites examined in September, the RH ranged between 71 and 93%. When these four data points were excluded from the regression analysis, a line with a slope close to the 1:1 line (1.28‰/‰; R = 0.62; P > 0.01) was produced. Values for L in literature range broadly between 4 and 54 mm (Cernusak et al. 2005). Thus, we subjected Δ18O calculated (1) with L values of 4 and 54 mm; and (2) disregarding leaf water isotopic inhomogeneities because of the Peclét effect (C–G in Fig. 5) to a regression analysis with measured values. The latter model (2) explained best the relation between calculated and measured Δ18O in phloem organic matter yielding an R of 0.65 (P < 0.01).
Correlation between δ13C and Δ18O
The relationship between δ13C and Δ18O of foliage and phloem sap was quantitatively examined in addition to the correlation analyses between δ13C and physiological and meteorological parameters, to separate the carboxylation-related effects on δ13C from diffusional ones. Figure 6a shows that the variation in Δ18Oleaf harvested in September was higher among sites as compared to δ13Cleaf resulting in a slope of the regression line of 0.19‰δ13C per 1‰Δ18O. A significant positive linear relationship was observed between δ13Cphloem and Δ18Ophloem (Fig. 6b). When the September value for the Höglwald site was regarded as an outlier value (displayed in brackets), R2 amounted to 0.44. The slope was 0.08‰δ13C per 1‰Δ18O, and thus, lower as for the regression line obtained for leaves. If the results for July and September were analysed separately, a significant relationship was found in September (R2 = 0.72, P < 0.001), but not in July.
Relation between isotopic signatures of leaf and phloem organic matter
In Fig. 7a, the δ13Cleaf is plotted against the δ13Cphloem. In general, with the exception of beech trees from the two highest elevated sites from the MdL (I: September; II: July and September), phloem carbon was isotopically heavier than the total leaf carbon. The enrichment in 13C in phloem organic matter relative to the total leaf carbon amounted to between 0.3 and 4.25‰. However, no significant linear relation between the two parameters was observed.
For oxygen, strong isotopic enrichment was also observed in Δ18Ophloem as compared to Δ18Oleaf (Fig. 7b). In July, the relative enrichment was higher than in September and amounted to c. 17‰ as the average of all sites.
When Δ18Oleaf is assumed to be related to Δ18Ophloem according to Δ18Oleaf = Δ18Ophloem (1 − pex px) with pex amounting to 0.4 and px to 1, the dotted line shown in Fig. 7b is obtained. This line is located much closer to the scattered points as compared to the 1:1 line. However, there was no significant linear correlation between Δ18Ophloem and Δ18Oleaf.
In the present study, we aimed to test if the combined analysis of carbon and δ18O in the heterogeneous organic matter of leaves and in the short-term turnover phloem carbohydrate pool could be used to assess water relations in beech on a time scale of a few days to the growing season.
The δ13Cleaf varied significantly along the transect, was specific for a particular field site without changing from summer to autumn and was correlated with physiological/environmental parameters which might influence ci/ca via diffusional restriction or Rubisco activity/carboxylation efficiency. Only the strong depletion in 13C observed in the leaves from Hofstetten may point to the fact that leaves collected from that site may have not been from the fully sun-exposed part of the crown.
The altitudinal gradient observed was in the same range as described for other tree and herbaceous species (e.g. Körner et al. 1991; Hultine & Marshall 2000; Warren, McGrath & Adams 2001). One main reason for decreasing isotope discrimination with increasing elevation is assumed to be the reduction of O2 partial pressure and an associated increase in assimilation rates (Körner et al. 1991). If stomatal conductance is unaffected by altitudinal gradients or has even a tendency to decrease with increasing elevation as observed in the present study (R = −0.43), decreased internal CO2 concentrations can account for the observed change in δ13C.
In addition, altitudinal changes in leaf morphology which influence CO2 diffusion and N content, which is assumed to be a proxy for carboxylation efficiency, have been implicated for altitudinal variation in carbon isotope composition. Because there was no significant altitude-related variation of leaf N content (R = −0.30, P = 0.31 for September), we have to assume that carbon isotope discrimination was not affected by Rubisco concentration.
There was, however, a strong correlation between foliar δ13C and sunshine hours – a proxy for intercepted radiation. Thus, increasing light availability is likely to add to increasing carboxylation efficiency with increasing altitude.
On the other hand, δ13Cleaf was also significantly related to parameters associated with atmospheric and tree water balance (especially RH, transpiration and Gs).
These results indicate a combined influence of stomatal diffusion (Gs) and carboxylation (radiation and potentially altitude) on internal CO2 concentration, and subsequently, photosynthetic carbon isotope discrimination, and thus, should also be reflected by the relation between Δ18O and δ13C (Scheidegger et al. 2000).
Table 3 shows that the Δ18Oleaf from September is highly related to mean RH, Ta and VPD of the current growing season, and thus, is in good agreement with the theory. The slope for the regression line (−0.19‰/%) between the RH and the Δ18O is largely in agreement with the ones observed for cotton plants (Farquhar, Barbour & Henry 1998) and oak tree rings (Switsur & Waterhouse 1998; both −0.12‰/%). Over the longer term, a strong coupling between atmospheric water demand and Gs is assumed for beech (Granier, Loustau & Breda 2000; Gessler et al. 2004a), which is nicely reproduced by the water balance model WBS3 (Gs versus VPD: R = −0.71, P = 0.009; Gs versus RH: R = 0.77, P = 0.003). Thus, the Δ18Oleaf in beech could be shown to be a measure of Gs and can be reasonably applied for regression analysis with δ13C according to the conceptual model of Scheidegger et al. (2000).
The combination of the two isotopes C and O (Fig. 6a) supports the hypothesis drawn from the correlation analysis between δ13C and environmental parameters that both factors, stomatal conductance and carboxylation, govern δ13C.
In contrast to previous studies (Gessler et al. 2001; Keitel et al. 2003), the combined analysis of δ13C and Δ18O of the total leaf organic matter of beech here proved to be indicative of environmental differences between sites. In the previous studies, the variation of environmental parameters was achieved by differences in local (topo-) climate (SW- versus NE-exposed aspect) and by applying different degrees of thinning to beech stands. Under these conditions, differences in climatic conditions, and thus, physiological reactions were much smaller as compared to the present study. As an example, maximum differences in mean growing season temperature between sites amounted to 2 °C in the study described by Keitel et al. (2003), whereas the variation along the climatic gradient assessed here was 5.8 °C (Table 2). Thus, we have to revise hypothesis 1 at least partially: δ13Cleaf can be considered a measure for changing environmental conditions and related physiological responses. It is, however, likely that because of the complex composition of this carbon pool, smaller differences in (local) climate are not resolved.
The carbon and oxygen isotopic signatures of the total leaf organic matter were not related to short-term variations of environmental or physiological parameters during the growing season. This is in agreement with the findings of Hemming et al. (2005) who did not find a relation between prevailing environmental conditions averaged over different time periods prior to harvest and δ13Cleaf for different species (including F. sylvatica) at sites distributed all over Europe.
In contrast to δ13Cleaf, δ13Cphloem differed strongly between sampling times and sites, and ranking of the sites according to their δ13Cphloem values changed between July and September, suggesting a strong influence of short-term variation of meteorological and/or physiological parameters.
However, no significant correlation was found between δ13Cphloem and the environmental parameters Ta, RH, VPD and R of varying time integrals. Yet, comparable to the previous studies of Keitel et al. (2003) and Gessler et al. (2004b), we here also observed a significant but weaker relation (R2: −0.79 in Keitel et al. (2003) versus −0.30 in this study) between δ13Cphloem and mean Gs of the previous 2 d, and the regression analysis (Fig. 4) produced a line with a slope comparable to the assessment of δ13Cphloem in the local climatic gradient (Keitel et al. 2003; Gessler et al. 2004b). The relatively high variation in δ13C of phloem organic matter in September within a small range of Gs might point either to a stronger influence of carboxylation or already indicate the onset of remobilization of carbon from senescing leaves. However, because the variation of the data points from both month can be described significantly by one regression line (Fig. 4), we have to assume that comparable physiological mechanisms govern phloem δ13C in July and September. In contrast to the long-term trend over the whole growing season, we did not observe a significant coupling between VPD or RH and Gs over the short term, which again supports the assumption of Keitel et al. (2003) and Gessler et al. (2004a,b) that under particular conditions, Gs may not only be controlled by atmospheric factors but also by different and potentially varying influences like soil water availability. This assumption is supported by the results of Scartazza et al. (2004) who observed a strong correlation between current soil water content and both δ13Cphloem of beech and δ13C of ecosystem respiration.
When the regression analyses between phloem oxygen and carbon isotope composition are compared between the present study and the local climatic gradient assessed by Keitel et al. (2003), one main difference emerges: whereas in the previous study, a slope of approximately 1 indicated a sole influence of Gs on δ13Cphloem, we here observe patterns indicative for the combined effects of carboxylation and stomatal limitation in both months as already observed for leaves. This finding is supported by the moderate correlation between phloem δ13Cphloem and sunshine hours 2–4 d prior to sampling which was, however, not significant because of the lack of this information for the major part of sites assessed.
The Δ18O in phloem organic matter was not directly related to any of the meteorological or physiological parameters examined (Table 4b). However, the model applied here, which takes into account the 18O fractionation associated with the exchange of water between leaves and the atmosphere and with the production of organic matter, is able to explain at least part of the variation of phloem organic matter along the transect (Fig. 5). For our model application, we assumed the water vapour of the atmosphere to be in isotopic equilibrium with the source water. However, it is known that stronger changes in RH during the diurnal cycle combined with slow leaf water turnover can result in deviations from this steady state (Wang & Yakir 1995). The stomatal conductances were lowest for the southernmost sites (between 17 and 39 mmol m−2 s−1) in July, which can contribute to a long leaf water turnover time. The low daytime RH values of the four southernmost sites examined in July might, thus, point to such a temporal disequilibrium. It is also worth mentioning that in September, the MdL values were again to the right of the regression line. In addition, it is likely that the term L is not constant with environmental conditions and time. Plants can quickly react with reduced tissue hydraulic conductance to drought stress, a reaction which is supposed to be mainly mediated by reduced expression and/or activity of aquaporins (for a recent review, see Tyerman, Niemietz & Bramley 2002). It can be assumed that the high atmospheric water demand at the four respective sites in July might also have resulted in reduced hydraulic conductance in the mesophyll cells, and thus, increased L.
When the four deviating values were disregarded, there was a moderate correlation with a regression line close to the 1:1 relationship between calculated and measured Δ18O of phloem organic matter. R increased with decreasing L. However, it should again be considered that L is likely not to be constant, but might vary between sites and time points.
We can accept hypothesis 2 because we found, indeed, a correlation between Gs and δ13Cphloem when a transport time for the assimilates from the leaves to the stem base of approximately 2 d is taken into account. However, the connection between these two parameters is likely to be weakened by the additional influence of Rubisco carboxylation activity. Whereas Keitel et al. (2003) observed a variation of δ13Cphloem between −25.5 and −30.5‰ in a comparable range of Gs as determined here, the (putatively levelling) combined effects of light availability/carboxylation efficiency and stomatal conductance on ci reduced the δ13Cphloem variation (Fig. 4). Thus, the additional assessment of oxygen isotope signatures is crucial for a proper interpretation of the information obtained with δ13C.
Even though the Δ18O of the phloem organic matter was not directly related to environmental parameters, it could at least be partially predicted by applying the model described by Farquhar & Lloyd (1993). The four deviating values point to the fact that it might be necessary to apply non-steady state models as proposed by Cernusak et al. (2005) to describe the Δ18O of organic matter and to define L more carefully when applying such models in field studies.
In contrast to our observations (Fig. 7b), Cernusak et al. (2005) found a close linear relation between Δ18Oleaf and Δ18Ophloem in E. globulus along a climatic gradient. Jäggi et al. (2003) also showed that the δ18O of the total organic matter of current year needles of P. abies corresponded well with δ18O of fast-turnover soluble organic matter. Thus, we may assume that the isotopic composition of beech leaves is, to a lesser extent, influenced by the labile and, to a higher extent, by the structural pool of organic matter produced during leaf development as compared to the two evergreen species. This assumption is also supported (1) by the fact that the model assuming the total leaf organic matter produced with non-18O-enriched water describes the relation between Δ18Ophloem and Δ18Oleaf better than a 1:1 line; (2) by the lack of correlation between δ13Cleaf and δ13Cphloem; and (3) the little variation observed in δ13Cleaf between July and September.
In addition to carbon and oxygen isotopic composition, the phloem sugar concentration has been shown to be a guide to the water balance of eucalyptuses. Pate & Arthur (1998) and Cernusak et al. (2003a) observed a strong relationship between sugar concentration and δ13C and Δ18O in E. globulus. Comparable patterns were also observed in beech under dry conditions, but not when water availability was higher (Keitel et al. 2003). Because we did not observe any correlation between δ13Cphloem and sugar concentration in our transect study, we have to conclude that such interrelation was confounded by the great climatic differences between sites (i.e. by the fact that not only water-limited sites were included in this study).
Combining the analyses of δ13C and Δ18O in organic matter pools with long (leaves) and short (phloem) turnover times, we could show that isotope signatures can be used as effective tools to assess the water balance of beech along a climatic gradient. Together with the application of models for evaporative 18O enrichment in organic matter, this approach can help to characterize environmental and physiological factors controlling water relations of trees.
Part of the work was financially supported by the Deutsche Forschungsgemeinschaft (DFG) under contract number (Re515/13-1). A.G. acknowledges personal financial support by a research fellowship from the DFG (GE 1090/4-1). We thank Graham D. Farquhar for helpful comments and fruitful discussion.