Genotype differences in 13C discrimination between atmosphere and leaf matter match differences in transpiration efficiency at leaf and whole-plant levels in hybrid Populus deltoides ×nigra



    1. INRA, UMR 1137 Ecologie et Ecophysiologie Forestières, F-54280 Champenoux, France
    2. INRA, USC1328 Arbres et Réponses aux Contraintes Hydriques et Environnementales (ARCHE), rue de Chartres, BP 6759, F-45067 Orléans Cedex 2, France
    Search for more papers by this author

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

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

    1. Université d'Orléans, UFR-Faculté des Sciences, UPRES EA 1207 Laboratoire de Biologie des Ligneux et des Grandes Cultures (LBLGC), rue de Chartres, BP 6759, F-45067 Orléans Cedex 2, France
    2. INRA, USC1328 Arbres et Réponses aux Contraintes Hydriques et Environnementales (ARCHE), rue de Chartres, BP 6759, F-45067 Orléans Cedex 2, France
    Search for more papers by this author

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

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

E. Dreyer. E-mail:


13C discrimination between atmosphere and bulk leaf matter (Δ13Clb) is frequently used as a proxy for transpiration efficiency (TE). Nevertheless, its relevance is challenged due to: (1) potential deviations from the theoretical discrimination model, and (2) complex time integration and upscaling from leaf to whole plant. Six hybrid genotypes of Populus deltoides×nigra genotypes were grown in climate chambers and tested for whole-plant TE (i.e. accumulated biomass/water transpired). Net CO2 assimilation rates (A) and stomatal conductance (gs) were recorded in parallel to: (1) 13C in leaf bulk material (δ13Clb) and in soluble sugars (δ13Css) and (2) 18O in leaf water and bulk leaf material. Genotypic means of δ13Clb and δ13Css were tightly correlated. Discrimination between atmosphere and soluble sugars was correlated with daily intrinsic TE at leaf level (daily mean A/gs), and with whole-plant TE. Finally, gs was positively correlated to 18O enrichment of bulk matter or water of leaves at individual level, but not at genotype level. We conclude that Δ13Clb captures efficiently the genetic variability of whole-plant TE in poplar. Nevertheless, scaling from leaf level to whole-plant TE requires to take into account water losses and respiration independent of photosynthesis, which remain poorly documented.


Transpiration efficiency (TE), the major component of the efficiency of water use by plants, is a complex trait that attracted much attention over the past decades. In particular, a large number of investigations aimed at detecting and quantifying the genetic variability of this trait (in trees for instance, Brendel et al. 2002; Monclus et al. 2005; Cernusak et al. 2008). TE at whole-plant level is the ratio between accumulated biomass and transpired water (Richards et al. 2002). At leaf level, TE is approached by intrinsic TE, that is, the ratio of net CO2 assimilation rate (A) versus stomatal conductance to water vapour (gs) (Condon et al. 2002). An indirect method of assessing intrinsic TE at leaf level (i.e. A/gs) was proposed by Farquhar, O'Leary & Berry (1982) and relates discrimination against 13C during photosynthesis (Δ13C) to the ratio of CO2 mole fraction in the substomatal cavity and in air (Ci/Ca) and therefore to intrinsic TE. In turn, Δ13C is often assessed from the difference between the isotopic composition of source CO2 in the air (δ13Cair) and of bulk leaf matter (δ13Clb). Δ13Clb has been widely used to evaluate the plastic response of species to changing environments, and the occurrence of an intra-specific, genetic variability in crops (Condon et al. 2004; Rebetzke et al. 2006) and in trees (Guehl et al. 1996; Lauteri et al. 1997; Roupsard, Joly & Dreyer 1998 among many others).

The relationship between intrinsic TE and Δ13Clb was validated experimentally during several intra-specific studies in trees (Zhang & Marshall 1994; Lauteri et al. 1997; Roupsard et al. 1998; Cregg, Olivas-Garcia & Hennessey 2000; Grossnickle, Fan & Russell 2005). However, some experiments failed to detect this relationship (Sun et al. 1996, in white spruce; Picon, Guehl & Ferhi 1996, in sessile oak). Such a lack of relationship can be ascribed to problems with time integration as intrinsic TE is highly variable due to fluctuating environment [vapour pressure deficit (VPD), temperature, irradiance . . .] and subjected to measurement uncertainties (Flexas et al. 2007), while Δ13Clb integrates discrimination over the whole leaf lifespan. One of the approaches to reduce such discrepancies is to record the isotopic signal in the primary photosynthesis products, soluble sugars (Δ13Css,Brugnoli et al. 1988; Kodama et al. 2008; Merchant et al. 2011) which provide a signal integrated over a single or at most a few days.

The simple model that relates Δ13C and Ci/Ca takes into account fractionation due to: (1) CO2 diffusion to chloroplasts and (2) carboxylation by ribulose 1·5-bisphosphate carboxylase/oxygenase (Rubisco). In this approach, mesophyll conductance to CO2 (gm; Evans et al. 2009) is taken as infinite. It is now established that gm has a major effect on 13C discrimination independently of Ci/Ca (Warren & Adams 2006; Douthe et al. 2011; Kodama et al. 2011). In addition, fractionation due to photorespiration and respiration may contribute to the signal independently of Ci/Ca (see a recent synthesis by Tcherkez, Mahé & Hodges 2011).

Moreover, the use of Δ13Clb or even Δ13Css as indicators of whole-plant TE needs careful attention, as it requires to scale up TE from instantaneous leaf level to whole-plant level during their life cycle (Condon et al. 2002). Direct estimates of whole-plant TE require a precise estimation of transpiration and biomass accumulation. The relationship between genetic variation of Δ13C and whole-plant TE in tree species has only been seldom established (Guehl et al. 1996; Roupsard et al. 1998; Cernusak et al. 2007b, 2008), while that between whole-plant and intrinsic leaf TE has been more frequently described (Osório, Chaves & Pereira 1998; Hobbie & Colpaert 2004; Cernusak et al. 2008). In the latter case, some studies again failed to detect any relationship (Matzner, Rice & Richards 2001; Ripullone et al. 2004). Although intrinsic TE is considered to be the primary source of variation in whole-plant TE, the fraction carbon lost during respiration overnight or in non-photosynthetic organs (Φc) and the water lost independently of carbon uptake (Φw) may also have an important influence on whole-plant TE (Matzner et al. 2001; Hobbie & Colpaert 2004). This shows the importance of calibrating Δ13C as an indicator of the genetic variability of whole-plant TE among genotypes.

Intrinsic TE and Δ13C depend on a number of environmental factors that affect gs and A alone or in combination (Farquhar et al. 1982). On the contrary, 18O discrimination between source and leaf water (Δ18Olw) or organic matter (Δ18Olb) reflects the enrichment in 18O due to transpiration and may therefore indicate differences in gs and transpiration rates among genotypes (Barbour & Farquhar 2000; Barbour 2007). 18O enrichment is sometimes directly proportional to the transpiration rate (DeNiro & Epstein 1979; Sheshshayee et al. 2005), while in some cases a negative correlation may be found due to the Péclet effect (Saurer, Aellen & Siegwolf 1997; Ferrio et al. 2007). The combined use of Δ13C as an indicator of TE and Δ18O as an indicator of gs is a powerful tool to dissect intrinsic TE into its components A and gs (Farquhar, Ehleringer & Hubick 1989; Yakir & Israeli 1995; Barbour 2007), but was surprisingly little used to analyse the genetic variability in TE (Scheidegger et al. 2000; Cabrera-Bosquet et al. 2009).

Identifying poplar genotypes with high whole-plant TE may be very useful for the adaptation of poplar cultivation to areas with lower water availability (Braatne, Hinckley & Stettler 1992). In spite of the fact that productivity of poplar severely depends on water availability (Ceulemans & Deraedt 1999), a large variability in Δ13Clb as indicator of intrinsic TE has been found among Populus deltoides × nigra genotypes (Monclus et al. 2005, 2006). Furthermore, productivity (i.e. biomass accumulation) was in many cases independent from Δ13Clb both under controlled and open field conditions (Marron et al. 2002, 2005; Rae et al. 2004; Monclus et al. 2005, 2006). Such genetic differences of Δ13C in the young plants were maintained with age, as shown with Δ13C recorded from the cellulose in annual tree rings (Rasheed et al. 2011).

Given the importance TE may gain for breeding poplar genotypes, and the sources of discrepancy between Δ13C and TE, it was of importance to calibrate the use of Δ13C as an indicator of the genetic diversity of TE at leaf and whole-plant scale in poplar. Thus, the main aims of the present research were to check whether:

  • • Δ13Clb and/or Δ13Css are reliable estimators of intrinsic leaf-level TE among poplar genotypes;
  • • the genetic variation found in Δ13Css reflects the differences in whole-plant TE among genotypes;
  • • the observed differences of intrinsic TE could be ascribed to net CO2 assimilation rate (A) or to stomatal conductance to water vapour (gs), and those of whole-plant TE to biomass accumulation or to transpiration;
  • • Δ18Olb and Δ18Olw differ among genotypes and indicate differences in gs as a driver of the genetic differences in whole-plant TE.

Six commercial Populus deltoides × nigra genotypes were grown under controlled environments and TE was measured (at instant and time-integrated scales, and at leaf and whole-plant levels) and compared with the isotopic composition of 13C and 18O in leaves.


Plant material and growth conditions

Six commercial genotypes of hybrid poplar Populus deltoides × nigra, Agathe F (A), Cima (C), Flevo (F), I45/51 (I45), Pannonia (P) and Robusta (R) were selected on the basis of contrasting values of Δ13Clb (Monclus et al. 2005, 2006). Ten litre pots were filled with a 1/1 v/v peat/sand mixture, heavily watered and extra water was left to drain overnight to reach field capacity. The weight of filled pots was homogenized to 10 000 g by adding substrate at field capacity with a precision of 1%. Eight soil cores were sampled randomly and dried (70 °C) to a constant weight to assess water content of the substrate at field capacity, which was estimated at 0.286 g g−1

Shoot cuttings (16 copies of 6 genotypes) were planted during April 2009 and left to grow in a greenhouse at INRA-Nancy (France), under natural daylight. After 1 month, diameter at collar and height were recorded on all plants immediately before the transfer to two climate chambers fitted with a rotating plate to homogenize the irradiance received by each plant (Rotoplant, Strader, Angers). Six randomly selected individuals were harvested per genotype, oven dried at 60 °C until constant weight to calibrate a unique relationship between height (H), diameter (D) and biomass (B) across genotypes (see Supporting Information Fig. S1):


This allometric equation was used to estimate the initial biomass of each remaining individual in the experiment (recorded mean: 15 ± 4 g).

Five individuals per genotype were randomly distributed into each of the two growth chambers. Microclimate in the chambers was: day/night, 16/8 h; air temperature, 25/18 °C; relative humidity, 85/45%; irradiance at the top of the plants, 450 ± 15 µmol photosynthetic photon flux density (PPFD) m2 s−1. The light in the climate chambers was switched on every day at 0800 h.

Growth, daily water use and TE

Plant height (H) and diameter at collar (D) were measured twice a week. The soil surface of each pot was covered with a polyethylene sheet to limit direct evaporation. Each pot was weighed daily (Sartorius-AG Göttingen, QC65EDE-D, Germany; accuracy: ±0.1 g) and watered back to the reference weight of 9428 g (= 80% field capacity). The weight difference over 24 h was assumed to represent daily water use. Cumulative water use (WU) was computed for each plant over the duration of the experiment.

After 1 month, plants were harvested, oven dried at 60 °C until constant weight and dry mass of each compartment (leaves, stem and roots) was recorded. Biomass increment (BM) was computed for each individual from the difference between harvested dry mass and biomass estimated at the start of the experiment. Whole-plant TE was computed per individual as biomass increment/total water use, and total leaf area (TLA) was measured with an area meter (Li-Cor area meter, A1000 Li-Cor, Lincoln, NE, USA).

Leaf gas exchange and intrinsic TE

Intrinsic TE at leaf level was recorded on one leaf per individual: (1) under saturating irradiance as the ratio Asat/gsat, and (2) under the standard irradiance available in the growth chamber (A/gs). Asat and gsat were measured with a portable photosynthesis system (Li6400; Li-Cor) with a 6 cm2 chamber and artificial irradiance provided by blue-red LEDs. Conditions were: PPFD, 1200 µmol m2 s−1; Cinlet, 380 µmol mol−1; VPD, 1 ± 0.2 kPa; and leaf temperature, 25 °C. A and gs were monitored with a portable photosynthesis system (Li-6400) equipped with a 6 cm2 chamber covered with a transparent lid. Measurements were repeated six times between 0800 and 1800 h during the course of a single day. A fully expanded and well-lit leaf of the same age was used on each individual. Microclimate in the photosynthesis chamber was close to ambient, that is, irradiance: 450 ± 15 µmol m2 s−1, CO2 in the chamber (Ca): 352 ± 20 µmol mol−1, VPD, 0.5 ± 0.15 kPa and leaf temperature, 25 °C. At the end of the measurement cycle, the leaf was clipped off, frozen in liquid nitrogen and stored at −80 °C. No diurnal trend was detected for A and gs (see Supporting Information Fig. S2). We therefore used the mean value of the ratio A/gs as the value for diurnal, time-integrated, intrinsic TE at leaf level, and the mean diurnal values of Ci/Ca were used to calculate the discrimination as predicted by the simple discrimination model (Farquhar & Richards 1984):


with a = 4.4‰ and b = 27‰.

Leaf sugar extraction

Half of each stored leaf was freeze dried at −196 °C, ground and 60 mg of leaf powder was weighed in 2 mL microtubes. The protocol for soluble sugar extraction was modified from Wanek, Heinter & Richter (2001). Methanol-chloroform-water (MCW, 12:5:3, v/v/v) (350 µL) was added and the samples were placed in a water bath at 70 °C for 30 min. After cooling, the microtubes were centrifuged at 11 400 g for 3 min. This step was repeated three times and the supernatant was collected into a new 2 mL microtube. To induce phase separation, 200 µL chloroform and 500 µL deionized water were added to the supernatant and vigorously mixed. The samples were left for a few minutes, then centrifuged at 11400 g for 3 min. The aqueous phase was transferred to a new 2 mL microtube and complemented with 300 µL freshly hydrated Dowex 1X8, 200–400 mesh [Cl]- form resin, (Fluka, Sigma-Aldrich, Saint Quentin Fallavier, France) converted to [HCO2]- with sodium formate. The samples were agitated during 2 h at room temperature. After centrifugation, the supernatant was transferred to a new 2 mL microtube complemented with 300 µL freshly hydrated Dowex 50W, 200–400 mesh [H]+ form, resin (Sigma Aldrich, Saint-Quentin Fallavier, France). Samples were again agitated during 2 h at room temperature. After centrifugation, the supernatant was transferred to a pre-weighed microtube and dried to complete dryness on a rotary evaporator (HETO, DK3450, Allerød, Denmark). Bulk sugar (1 mg) was diluted with 60 µL water, transferred into tin capsule and freeze dried (FreeZone, Labconco, Kansas City, MO, USA).

Carbon isotope discrimination between atmosphere and bulk leaf matter (Δ13Clb) or soluble sugars (Δ13Css)

A fully mature leaf (below the one used for leaf gas exchange) was selected from each individual and was finely ground after drying at 60 °C to a constant weight. Powder (1 mg) was weighed in tin capsules. 13C content in both leaf bulk and leaf soluble sugars, as well as carbon and nitrogen content, were analysed with an elemental analyzer (Carlo Erba, NA 1500-NC, Milan, Italy) coupled to an isotope-ratio mass spectrometer (Finnigan, Delta-S, Bremen, Germany) with a precision of 0.1‰. 13C discrimination between the atmosphere and leaf bulk matter or soluble sugars was calculated as:


where δ13Cair is the isotopic composition of air measured in the two growth chambers. δ13Cair was similar in the two chambers (P = 0.548, n = 46). δ13Cair was −9.01‰ (SD ± 0.95‰) with small fluctuations during the diurnal cycles.

Leaf water extraction

The second half of each stored leaf was used for the extraction of leaf water, with a cryogenic vacuum distillation. Sealed tubes containing a frozen leaf were connected to the extraction apparatus with a collection tube at the other end. Air inside the whole apparatus was evacuated to remove any trace of external water vapour under approximately 8 Pa. The vial containing the sample was heated using a water bath at a constant temperature of 70 °C and the collection tube was placed in a Dewar containing liquid nitrogen in order to freeze the vapour emanating from the sample (see West, Patrickson & Ehleringer 2006). The extracted water was collected and used to measure 18O composition.

Oxygen isotope discrimination between irrigation water and leaf bulk matter (Δ18Olb) or leaf water (Δ18Olw)

Leaf powder (0.3–0.4 mg) (the same than for 13C analysis) and leaf extract (0.4 µL) were used to measure 18O composition of bulk leaf matter (δ18Olb) and of leaf water (δ18Olw), respectively. Analyses were done with a high temperature elementar analyser (Pyrocube, Elementar, Hanau, Germany) coupled to a mass spectrometer (Isoprime, Manchester, UK). Samples were combusted and pyrolized at 1270 °C. The oxygen isotope composition was determined with respect to the three laboratory standards. Laboratory standards were pre-calibrated against the international standard Vienna-Standard Mean Ocean Water (V-SMOW). Accuracy of the measurements was ±0.3‰.

The 18O content in irrigation water (δ18Osw) was recorded from nine samples collected at different dates during the experiment. The mean value was −7.58‰ (± 0.46‰) with no difference between the two chambers (P = 0.780, n = 9).

18 O discrimination between irrigation water and leaf bulk matter or water was computed as:


Leaf anatomy

Three 1 cm2 discs were harvested on one leaf per individual to record leaf thickness (LT) and stomatal density (SD), and were immediately frozen in liquid nitrogen and stored at −80 °C. As poplar leaves are amphistomatous, each disc was split for the separate analysis of adaxial and abaxial sides. Sample discs were stuck to aluminium stubs on a Peltier stage (−50 °C) before being examined under a controlled-pressure scanning electron microscope (model 1450VP, Leo, Cambridge, UK; 20–30 Pa inside the chamber, accelerating voltage 15 kV, working distance 12 mm). Nine microphotographs (×300) were taken on each disc and stomata were counted with an image analysis software (VISILOG, Noesis, Orsay, France). Total SD was calculated as the sum of adaxial and abaxial stomatal densities. On the other three discs, five semi-thin cryo-sections were photographed per disc to measure mean leaf thickness. Additional five discs of 1 cm2 were sampled for leaf mass to area ratio (LMA) and weighed after oven drying for 24 h. LMA (g m−2) was determined as the ratio between dry weight and area.

Statistical analyses

All statistical analyses were made using STATISTICA (version 8.1, StatSoft, Maisons-Alfort, France). Normality and homoscedasticity of data were checked graphically with residues versus predicted and normal quartile-to-quartile plots. Genetic effect, growth chamber effect and their interaction were assessed with two-way anova and the following model:


with Yijk, response variable; µ, intercept; Gi, genotype effect; Bj, growth chamber effect; Gi × Bj, interaction between genotype and growth chamber; and εijk, residue. A Tukey HSD test was used to evaluate pairwise differences among genotypes. In the absence of growth chamber and interaction effects, data from both growth chambers were pooled. Correlations between the measured traits were tested at genotype level with a general linear regression model. All tests and correlations were taken as significant when P < 0.05. Means are expressed with their standard deviation (SD).

The variability of the recorded traits was assessed at individual level with a principal components analysis (PCA). Variables were represented on the main plane defined by the two main factors of the PCA (F1 and F2 axis); their coordinates were their linear correlation coefficients (r; Pearson's correlation coefficient) with these factors. Correlations were taken as significant when P < 0.05.


Table 1 presents a list of the recorded variables and Table 2 their genotype means. No climate-chamber effect was detected for any variable and there was no interaction between climate chamber and genotype effects on any variable. The presentation of the results therefore concentrates only on the genotype effects. Spearman correlations for the recorded traits at individual level (phenotypic correlation) are displayed in Table 3.

Table 1.  List of variables and abbreviations used in the study
D Diameter at stem base (mm)
H Stem height (cm)
δ 13 C air Carbon isotope composition of the air (‰)
δ 13 C ss Carbon isotope composition of soluble sugars in the leaf (‰)
δ 13 C lb Carbon isotope composition of bulk leaf matter (‰)
δ 13 C ss Carbon isotope discrimination between atmosphere and soluble sugars in the leaf (‰)
δ 13 C lb Carbon isotope discrimination between atmosphere and bulk leaf matter (‰)
δ 13 C offset Difference between δ13Css and δ13Clb (‰)
N lb, NssN content in bulk leaf or in the soluble sugar fraction (%)
C lb, CssC content in bulk leaf or in the soluble sugar fraction (%)
A, AsatNet CO2 assimilation rate under ambient/saturating condition (µmol m−2 s−1)
g s, gsatStomatal conductance under ambient/saturating condition (mol m−2 s−1)
A sa t /g sat Intrinsic transpiration efficiency under saturating condition (µmol mol−1)
A/g s Intrinsic transpiration efficiency of a leaf over a diurnal cycle (µmol mol−1)
C i /C a Ratio of CO2 concentration in the atmosphere and in the substomatal spaces
BM Total plant biomass accumulated during the experiment (g)
WU Cumulative water use (l)
TE Whole-plant transpiration efficiency = BM/WU (g L−1)
δ 18 O sw Oxygen isotope composition of source water (‰)
δ 18 O lb Oxygen isotope composition of bulk leaf matter (‰)
δ 18 O lw Oxygen isotope composition of leaf water (‰)
δ 18 O lb Oxygen isotope discrimination between source water and bulk leaf matter (‰)
δ 18 O lw Oxygen isotope discrimination between source water and leaf water (‰)
LT Leaf thickness (µm)
SD Stomatal density (mm−2)
LMA Leaf mass to area ratio (g m−2)
TLA Whole-plant leaf area (m2)
Table 2.  Genotype means (SD) of traits recorded on 10 individuals from six Populus x euramericana genotypes grown in two climate chambers
  Agathe F Cima Flevo I45/51 Pannonia Robusta G effect B effect G×B
  1. A factorial anova (d.f. 5, 1 and 5) was used to test for the effects of genotype (G), climate chamber (B) and their interaction (G×B). Different letters represent significant differences among genotypes as tested by a post hoc Tukey test (P < 0.05). Bold values are significant values at (P < 0.05). See Table 1 for the definition of variables.

D (mm)10.5 (0.28) b11.2 (0.28) ab11.1 (0.50) ab10.5 (0.46) b11.9 (0.44) a11.8 (0.11) a P = 0.034 P = 0.938 P = 0.306
H (cm)85.9 (3.47) b95.6 (3.47) a78.6 (3.47) c55.5 (3.65) d93.4 (3.47) a85.8 (3.65) bc P < 0.001 P = 0.956 P = 0.153
N lb (%)3.06 (0.26)2.93 (0.30)3.28 (0.29)3.47 (0.21)2.59 (0.34)3.34 (0.11) P = 0.227 P = 0.156 P = 0.436
C lb (%)47.0 (0.23) a46.5 (0.26) b45.9 (0.45) b47.1 (0.50) b46.5 (0.47) b47.5 (0.24) a P = 0.047 P = 0.456 P = 0.768
δ 13 C lb (‰)21.7 (0.15) ab22.2 (0.12) bc21.7 (0.17) ab22.5 (0.25) c21.4 (0.10) a22.1 (0.08) bc P < 0.001 P = 0.656 P = 0.139
N ss (%)0.124 (0.01)0.101 (0.01)0.107 (0.01)0.092 (0.01)0.115 (0.01)0.067 (0.01) P = 0.200 P = 0.169 P = 0.996
Css (%)43.2 (0.33) a41.0 (0.14) b42.1 (0.23) b41.1 (0.42) b41.4 (0.26) b42.1 (1.16) b P = 0.021 P = 0.065 P = 0.086
δ 13 C ss (‰)23.3 (0.13) b23.2 (0.14) ab23.0 (0.18) ab24.3 (0.19) c22.5 (0.17) a23.3 (0.10) b P < 0.001 P = 0.899 P = 0.140
δ 13 C offset (δ13Clb – δ13Css)1.47 (0.08) a0.928 (0.15) b1.20 (0.19) b1.60 (0.18) a1.01 (0.16) b1.12 (0.16) b P = 0.049 P = 0.799 P = 0.743
Soluble sugar content in leaves (%)16.5 (0.16)16.6 (0.32)16.2 (0.41)16.4 (0.39)16.8 (0.27)16.5 (0.58) P = 0.904 P = 0.326 P = 0.648
A sat (µmol m−2 s−1)19.1 (0.14) bc20.1 (0.63) abc21.1 (0.44) a18.2 (0.64) c20.1 (34) ab20.4 (0.19) ab P < 0.001 P = 0.362 P = 0.848
g sat (mol m−2 s−1)0.518 (0.03) b0.698 (0.03) a0.630 (0.03) ab0.572 (0.03) ab0.742 (0.03) a0.660 (0.06) ab P = 0.003 P = 0.170 P = 0.301
A sat /g sat (µmol mol−1)38.4 (2.60) a29.3 (1.59) b34.1 (1.45) ab32.2 (1.50) ab27.8 (1.68) b32.5 (2.26) ab P = 0.004 P = 0.179 P = 0.272
g s (mol m−2 s−1)0.479 (0.03) b0.580 (0.03) ab0.572 (0.02) b0.572 (0.03) ab0.587 (0.02) ab0.656 (0.01) a P < 0.001 P = 0.979 P = 0.862
A (µmol m−2 s−1)10.3 (0.68) ab10.6 (0.40) ab11.3 (0.47) a8.80 (0.37) b12.6 (0.44) a11.8 (0.55) a P < 0.001 P = 0.480 P = 0.160
A/gs (µmol mol−1)21.5 (2.03) a19.2 (1.16) ab20.5 (1.69) ab15.8 (1.08) b21.9 (1.52) ab17.9 (1.21) ab P = 0.004 P = 0.605 P = 0.603
C i /C a 0.882 (0.01) a0.893 (0.01) ab0.880 (0.01) a0.934 (0.01) b0.868 (0.01) a0.913 (0.01) ab P = 0.005 P = 0.468 P = 0.624
BM (g)148 (6.69) b178 (9.57) ab144 (12.5) b99.6 (13.9) c191 (9.18) a151 (4.16) b P < 0.001 P = 0.889 P = 0.053
WU (l)9.87 (0.52) ab11.9 (0.65) ab9.39 (0.72) b10.0 (1.11) ab12.3 (1.02) ab12.5 (0.40) a P = 0.009 P = 0.830 P = 0.100
TE (g L−1)15.3 (0.56) a15.1 (0.42) a15.6 (0.49) a9.68 (0.37) c16.3 (1.01) a12.3 (0.18) b P < 0.001 P = 0.837 P = 0.996
δ 18 O lb (‰)36.2 (0.34)35.9 (0.34)35.7 (0.34)36.0 (0.38)36.3 (0.34)35.7 (0.36) P = 0.775 P = 0.854 P = 0.803
δ 18 O lw (‰)9.46 (2.14)7.70 (2.44)10.6 (3.59)9.13 (1.97)8.36 (1.88)9.13 (1.03) P = 0.406 P = 0.874 P = 0.502
LT (µm)79.8 (2.67)76.1 (2.66)74.5 (2.17)80.0 (3.31)74.6 (3.35)68.1 (3.28) P = 0.093 P = 0.827 P = 0.552
SD (mm−2)188 (6.18) b216 (2.25) ab200 (4.31) b134 (5.55) c247 (14.8) a203 (5.81) b P < 0.001 P = 0.971 P = 0.375
LMA (g m−2)34.8 (0.805)35.2 (0.402)35.2 (0.982)33.4 (1.87)39.3 (2.11)32.6 (0.643) P = 0.322 P = 0.962 P = 0.383
TLA (m2)0.379 (0.03) ab0.418 (0.03) ab0.320 (0.03) b0.412 (0.04) ab0.405 (0.04) ab0.483 (0.01) a P = 0.017 P = 0.947 P = 0.599
Table 3.  Pearson's correlation table for the traits measured during the experiment
  δ 13 C lb δ 13 C ss A g s A/g s C i /C a BM WU TE δ 18 O lb δ 18 O lw LT SD LMA TLA
  1. Correlations were computed at individual tree level. Degree of significance indicated as *, ** and *** indicating P < 0.05, P < 0.01 and P < 0.001, respectively. See Table 1 for the definition of variables. In bold: significant correlations (P < 0.05).

δ 13 C lb                
δ 13 C ss 0.537***               
A −0.397*** −0.336***              
g s 0.179−0.2420.096            
A/gs −0.412*** −0.485*** 0.579*** −0.228           
C i /C a 0.464*** 0.453*** −0.726*** 0.224 −0.916***           
BM −0.530*** −0.570*** 0.396*** 0.194−0.009−0.183         
WU −0.166−0.2580.138 0.361*** −0.1830.1080.183        
TE −0.532*** −0.544*** 0.283** −0.197 0.428*** −0.482*** 0.431*** −0.182       
δ 18 O lb 0.0020.088−0.119 −0.288** 0.176−0.091−0.169 −0.369*** 0.160      
δ 18 O lw 0.1150.056−0.069 −0.257* 0.172−0.082−0.163 −0.295** 0.082 0.308**      
LT −0.0200.030−0.150−0.2470.143−0.034−0.192 −0.257* 0.1120.178−0.019    
SD −0.393*** −0.447*** 0.387*** 0.100 0.331** −0.378*** 0.587*** 0.203 0.401*** −0.050−0.151−0.204   
LMA −0.226−0.2270.1320.0820.071−0.172 0.298** 0.1780.1700.187−0.132−0.0260.221  
TLA 0.050−0.0510.069 0.470*** −0.2460.1940.213 0.386*** −0.182 −0.308** −0.283** −0.282** 0.2190.035 

Plant height and stem diameter

Time courses of height (H), diameter (D) and WU displayed a continuous and gradual increase (Supporting Information Fig. S3). At the end of the experiment, large genotypic differences were detected for H (P < 0.001) and to a lesser extent for D (P = 0.034). Pannonia displayed highest and I45/51 lowest values of H and D.

Δ13C between atmospheric CO2 and bulk leaf matter (Δ13Clb) or soluble sugars (Δ13Css)

Leaf tissues contained about 3.1% N and 46.5% C as expected for poplar leaves, with no genotype effect for N and a very small one for C (Table 2). Significant genotypic differences were found for Δ13Clb (range: 21.5 to 22.5‰). Pannonia showed lowest and I45/51 showed highest Δ13Clb.

The purity of the soluble sugar fraction was tested from their N and C contents. N content was, as expected, very low (around 0.1%), with no genotype effect. C amounted 42% approximately, which is very close to the values expected for pure carbohydrates. The soluble sugar extract contained sucrose, glucose, fructose and a sugar identified as mannose, representing between 45 and 55% of the extracts (not shown). A small genotype effect was detected for the C content, with Agathe F displaying slightly larger values than the other genotypes. The fraction of soluble sugars in the leaves was rather high (16.5 ± 0.36% of total dry matter) and stable across genotypes.

Δ13Css also displayed a significant genotype effect (P < 0.001), and genotype means ranged between 22.5 and 24.3‰ (Table 2). A tight and positive correlation was found between Δ13Css and Δ13Clb; however, Δ13Clb was always smaller than Δ13Css (Fig. 1). The difference was not constant across genotypes and increased with Δ13Clb: I45/51 showed a larger offset than Pannonia.

Figure 1.

Relationship between genotype means (±SD) of the isotopic discrimination between atmospheric CO2 and the C in bulk leaf matter (Δ13Clb) or in soluble sugars in the leaf (Δ13Css). Letters indicate genotypes (P, Pannonia; F, Flevo; I45, I45/51; A, Agathe F; C, Cima; R, Robusta). Regression equation was: y = 1.34x − 6.30.

Leaf gas exchange and intrinsic TE at leaf level

A genotype effect was evident for Asat, gsat and Asat/gsat. Overall means were 19.8 µmol m−2 s−1 for Asat and 0.637 mol m−2 s−1 for gsat (Table 2). No diurnal trend was evidenced in instant net CO2 assimilation rate (A) or stomatal conductance to water vapour (gs) recorded under ambient conditions (anova with time as a random effect, P = 0.748 for A and P = 0.239 for gs; Supporting Information Fig. S2). Mean daily values of A, gs and A/gs were therefore used to address genotype differences and to correlate with Δ13Css. A very significant genotype effect was observed for A, gs and A/gs. A was around 10 µmol m−2 s−1, that is, approximately half the values of Asat while gs was very close to gsat. Intrinsic TE ranged between 15.8 (I45/51) and 23.5 µmol mol−1 (Pannonia). Significant genotype differences were similarly found for Ci/Ca, which ranged between 0.871 (Pannonia) and 0.945 (I45/51).

Surprisingly, no relationship was evidenced between mean diurnal A/gs and Asat/gsat (Fig. 2). Genotype values for Δ13Css were negatively correlated to mean diurnal A/gs (not shown) and positive correlations were evidenced between Ci/Ca and Δ13Css and Δ13Clb as predicted by the simple discrimination model (Fig. 3, P < 0.001 and P < 0.001). Observed values of Δ13Css were approximately 1.5‰ below predicted, while those of Δ13Clb were 2.5‰ below predicted.

Figure 2.

Relationship between genotype means (±SD) of mean diurnal intrinsic transpiration efficiency A/gs and of Asat/gsat. Letters indicate genotypes (P, Pannonia; F, Flevo, I45, I45/51; A, Agathe F; C, Cima; R, Robusta).

Figure 3.

Relationship between genotypic means (±SD) of the ratio of CO2 concentrations in the intercellular spaces and in the atmosphere Ci/Ca with the isotopic discrimination between atmospheric CO2 and the C in (1) bulk leaf matter Δ13Clb (closed circles, y = 18.0x + 6.98) and (2) soluble sugars extracted from the leaf Δ13Css (open circles, y = 27.2x + 0.620). The solid line with squares represents the simple discrimination model (see text). Letters indicate genotypes (P, Pannonia; F, Flevo, I45, I45/51; A, Agathe F; C, Cima; R, Robusta).

Genotype means of Δ13Css were correlated to A but not to gs (Fig. 4): differences in net CO2 assimilation rates among genotypes had a much larger effect on intrinsic TE than gs.

Figure 4.

Relationships between genotype means (±SD) of the isotopic discrimination between atmospheric CO2 and the C in soluble sugars Δ13Css, and (a) mean diurnal CO2 assimilation rate (A), (A = −2.00 Δ13Css + 57.5); or (b) mean diurnal stomatal conductance to water vapour (gs). Letters indicate genotypes (P, Pannonia; F, Flevo, I45, I45/51; A, Agathe F; C, Cima; R, Robusta).

Whole-plant TE

Accumulated biomass (BM) differed among genotypes and ranged between 191 g (Pannonia) and 99.6 g (I45/51). Mean values of WU also differed among genotypes and ranged between 12.5 L (Robusta) and 9.39 L (Flevo). As a result, whole-plant TE varied among genotypes: Pannonia displayed highest (16.3 g L−1) and I45/51 lowest values (9.68 g L−1, Table 2). Genotype values of whole-plant TE were negatively correlated to Δ13Css and positively correlated to intrinsic TE (Fig. 5). At the same time, whole-plant TE was positively correlated to BM but not to WU (Table 3).

Figure 5.

Relationships between genotype means (±SD) of whole-plant transpiration efficiency (TE) and (a) the isotopic discrimination between atmospheric CO2 and the C in leaf soluble sugars Δ13Css; y = −0.215x + 26.3; (b) the mean diurnal ratio of A/gs; y = 0.872x + 7.19; (c) the biomass (BM) gained by the plants during the experiment (BM) y = 10.1x + 8.72 and (d) total water use by the plants (WU) during the experiment. Letters indicate genotypes (P, Pannonia; F, Flevo, I45, I45/51; A, Agathe F; C, Cima; R, Robusta).

18O discrimination between irrigation water and leaf bulk matter (Δ18Olb) or leaf water (Δ18Olw)

Genotype means were around 36.0‰ for Δ18Olb and 9.23‰ for Δ18Olw (Table 2). Due to relatively large intra-genotype variations, no genotype effect was detected for Δ18Olb and Δ18Olw. Nevertheless, at individual level, Δ18Olb and Δ18Olw were correlated to gs and WU (Table 3).

LMA, LT and SD

No genotypic variability was found for LT or for LMA that were close to 75 mm and 35 g m−2 (Table 2). SD differed among genotypes, with Pannonia displaying highest and I45/51 lowest SD. Δ13Css was negatively and A was positively correlated to SD (Fig. 6).

Figure 6.

Relationships between genotype means (±SD) of stomatal density and (a) the isotopic discrimination between atmospheric CO2 and the C in soluble sugars extracted from the leaf Δ13Css; y = −0.015x + 26.3; (b) mean diurnal A; y = 0.032 x + 4.48. Letters indicate genotypes (P, Pannonia; F, Flevo, I45, I45/51; A, Agathe F; C, Cima; R, Robusta).

Correlations among variables

A general PCA was performed with the 15 measured traits (Fig. 7). The main plane of the PCA (F1 × F2) explained 56.9% of the overall variability, with 29.1% for F1 and 27.8% for F2 axis. F1 was mostly correlated with traits defining water use efficiency at different scales like Δ13Css, Δ13Clb, A/gs, Ci/Ca, A, whole-plant TE that were also tightly correlated to BM and SD. The second axis was related to water use and oxygen isotope composition, with correlations to TLA and gs.

Figure 7.

Distribution of the 15 variables (a) and projection of the six Populus×euramericana genotypes (b) in the factorial plane F1 × F2 of the PCA. Letters indicate the names of the genotypes (P, Pannonia; F, Flevo, I45, I45/51; A, Agathe F; C, Cima; R, Robusta). F1 and F2 are linear combinations of the 15 variables. See Table 1 for variable codes. PCA, principal components analysis.

Along the F1 axis A/gs, A, whole-plant TE and SD were positively intercorrelated and negatively correlated to Δ13Css, Δ13Clb and Ci/Ca. Along the F2 axis, WU scaled positively with TLA and gs, and negatively with Δ18Olb (see Table 3 for r values of Pearson's correlation coefficients between traits). There was a clear grouping among individuals in the F1 × F2 planes, with some overlaps between genotypes. Pannonia and I45/51 were clearly discriminated along axis 2, the latter being much less productive and efficient with respect to water use than the former. Axis 1 was dominated by total leaf area and did less clearly discriminate the genotypes due to some variability in size within genotypes. This in particular explains why we were unable to detect any genotype difference in Δ18Olb or Δ18Olw despite large differences among individuals and a significant correlation with water use and gs at individual level.


In this study, the complex trait ‘transpiration efficiency’ was recorded at whole-plant level over a month, and at leaf level during a day with gas exchange measurements and 13C composition of soluble sugars. We used also 13C composition of bulk leaf matter as an indicator integrating leaf lifespan. The different approaches provided convergent estimates for genotype differences in TE.

Discrimination against 13C from the air to bulk leaf matter (Δ13Clb) or to soluble sugars (Δ13Css)

The tight correlation observed between Δ13Clb and Δ13Css at individual as well as at genotype levels, confirms that the differences in the 13C signal recorded in soluble sugars that display a rapid turnover, were still visible in the bulk leaf matter built over several weeks. This is due to the stable conditions that prevailed in the climate chambers since the start of leaf expansion. It also confirms that post-photosynthetic 13C discrimination had similar effects in all genotypes and did not modify their ranking of genotypes. A similar close relationship was described in a range of oak genotypes (Roussel et al. 2009a). Leaves were sampled from fast-growing and very actively photosynthesizing poplar cuttings at the end of a 16 h illumination phase. They contained a large amount of soluble sugars (around 16% total dry matter), and the extracts displayed a high purity indicated by a very low N content (no contamination by amino acids) and a C content close to that of sucrose or of a mix of carbohydrates. In the forthcoming discussion, we will focus on Δ13Css as a potential index for intrinsic TE, as it is expected to closely reflect discrimination during photosynthesis.

TE at leaf level: correlation of Δ13Css with A/gs

The genotype means of Δ13Css were negatively correlated to mean diurnal A/gs over the whole illumination phase. In this respect, our data confirm that Δ13Css is a relevant indicator for genetic differences in A/gs as predicted by the model of Farquhar et al. (1989), in line with numerous earlier results (Brugnoli et al. 1988 for poplar).

To our surprise, Δ13Css was larger than Δ13Clb while bulk leaf matter is usually depleted in 13C with respect to carbohydrates, due to the presence of lipids and lignins (Monti et al. 2006; Bowling, Pataki & Randerson 2008). Such was the case for leaves of Fagus sylvatica (Gavrichkova et al. 2011) or of Quercus robur (Roussel et al. 2009a). We are not aware of any published result similar to ours. Several hypotheses may explain this discrepancy:

  • 1a contribution to leaf structure of C stored before the transfer to the stable conditions of the climate chamber. This hypothesis is unlikely, as the leaves did expand after the transfer to the climate chamber, and as the biomass gain during the experiment was nine times the initial biomass; the contribution of ‘old’ C to the construction of new leaves was probably very minor;
  • 2diurnal changes have been recorded several times in Δ13Css (for instance, Gavrichkova et al. 2011) and as a consequence in the 13C signature of respired CO2 (see Werner & Gessler 2011 for a synthesis). Such changes may partly be due to changes in the environment (irradiance, temperature, VPD), which did not occur here. They may be due also to switches between respiratory substrates with different isotopic signatures. Such switches mainly occur during day–night transitions and are associated for instance to post-illumination respiratory bursts (Gessler et al. 2009) and are unlikely to impact Δ13Css recorded at the end of an illumination period;
  • 3the isotopic fractionation by aldolase results in a ∼3–4‰ enrichment of the C-3 and C-4 atoms of fructose in the chloroplasts (Gleixner & Schmidt 1997), and in a small but significant enrichment for starch accumulated in chloroplasts; meanwhile, glucose and sucrose in the cytosol and the vacuole are slightly depleted (Brugnoli et al. 1988; Badeck et al. 2005; Bowling et al. 2008). This depletion remains visible in the sucrose exported to the phloem. This is no longer true during night, when the stored starch is hydrolyzed into 13C enriched sucrose.

The third hypothesis fits best with our situation as soluble sugars were sampled at the end of 16 h illumination, with permanently high photosynthesis, and a large starch accumulation during the day (see Brugnoli et al. 1988 for an illustration of starch dynamics in poplar). This is likely to result in a depletion of soluble sugars with respect to the primary photosynthesis products (i.e. 3PGA), while leaf bulk matter contains large amounts of slightly enriched starch.

The tight and linear correlation between Ci/Ca and Δ13Css paralleled the predictions of the simple form of the discrimination model of Farquhar & Richards (1984). The negative offset of 0.9 to 1.6‰, depending on the genotype, is partly due to the occurrence of a finite conductance to CO2 transfer in the mesophyll, gm (Evans & Von Caemmerer 1996; Evans et al. 2009; Flexas et al. 2012): such an offset is the basis for indirect estimates of gm, provided b is set to 30‰ (and not 27‰, as it was set here).

There are a few small uncertainties around our estimates of Δ13Css due for instance to our estimates of δ13C in the chamber atmosphere. We used a mean value computed from records over several days. δ13C in the atmosphere of the chamber may have undergone diurnal cycles due to the production of depleted CO2 over night, and photosynthesis during the day, resulting in an increase of 13CO2 in the air. As photosynthesis rates remained stable during the day, the use of a mean value of δ13C in the atmosphere probably minimized this artefact.

There are still a number of uncertainties around the 13C discrimination model (Douthe et al. 2012; Farquhar & Cernusak 2012). In particular, the values of parameter ‘b’ vary between 27 and 30‰ in the literature (Warren 2006; Douthe et al. 2012). The more common use of a value of 30‰ instead of 27‰ would have amplified the offset. We therefore avoided a direct computation of gm from our dataset. Nevertheless, we may safely conclude that: (1) the different genotypes displayed as expected finite albeit probably large values of gm and (2) gm probably differed significantly among genotypes, with smaller values (i.e. larger offsets) for Agathe F and I45/51 with respect to the others. Nevertheless, these differences had only little impact on the ranking of the different genotypes as shown in Figure 3.

Correlation of whole-plant TE with intrinsic TE and Δ13Css

Instant TE may be modulated by VPD and by genotype differences in intrinsic TE. For individuals grown under a common VPD, like here, genotypic differences in intrinsic TE are expected to be the main source of variation for whole-plant TE (Farquhar & Richards 1984; Hubick & Farquhar 1989). We indeed found a strong positive genotype correlation between whole-plant TE and intrinsic TE like in many species: Larix occidentalis (Zhang & Marshall 1994), Pinus pinaster (Guehl et al. 1996), Eucalyptus globulus (Osório & Pereira 1994), several tropical tree species (Cernusak et al. 2007b, 2008), Q. robur genotypes (Roussel et al. 2009b). Nevertheless, whole-plant TE can be modulated by factors independent of intrinsic TE: (1) the fraction of carbon fixed during the day and lost through respiration over night in leaves, and in stems and roots over the whole day (Φc); and (2) the fraction water loss not associated to photosynthesis (Φw, Farquhar et al. 1989). Φc and/or Φw can be modulated by N (Hobbie & Colpaert 2004) or water availability (Osório et al. 1998). Genotype difference in Φc and Φw could severely blur the relationship between intrinsic TE and whole-plant TE. The tight correlation between intrinsic and whole-plant TE observed here suggests that Φc and Φw remained rather stable across the genotypes.

Nevertheless, the relative variation of intrinsic TE (27.6%) was much smaller than that of whole-plant TE (40.5%). This is a frequent occurrence: relative variations of intrinsic and whole-plant TE were 21% versus 70% among tropical tree species (Cernusak et al. 2008), 44% versus 55% among acacia species (Konaté 2010), and 44% versus 14.5% among oak genotypes (Roussel et al. 2009b). This shows that in many cases, Φc and Φw have an impact on genotype or species-related differences of whole-plant TE in addition to that of intrinsic TE and that this impact still has to be unambiguously quantified through direct records of respiration of non-photosynthetic tissues, or by records of nocturnal transpiration.

Whole-plant TE varied between 10 and 15 g kg−1. Such values are unexpectedly high when compared with the current literature. A range of 3–4.5 g kg−1 was found in tropical tree species (Cernusak et al. 2007b, 2009a; Cernusak, Winter & Turner 2009b), 4–5 g kg−1 in populations of Faidherbia albida (Roupsard et al. 1998), 4.5–5.5 g kg−1 in Q. robur genotypes (Roussel et al. 2009b), around 10 g kg−1 in P. pinaster and around 14 g kg−1 in Q. petraea under elevated CO2 and drought (Guehl et al. 1994). Hobbie & Colpaert (2004) did not cite any value above 11.8 g kg−1 in their review for trees or annual crops. The high values detected here for poplar were due to low VPD during the day, long days (16 h per photoperiod) and moderate but constant irradiance and optimal irrigation and fertilization. Moreover, our individuals were juveniles, and displayed a very fast growth and low level of lignification. Under such conditions, Φw is likely minimal (short nights and therefore limited nocturnal transpiration) as is probably Φc (limited lignification which would be a source of respiration loss; small amount of lipid synthesis and heavy investment into photosynthesising tissues). This may only be a transitory phase, and we expect smaller values of TE at later developmental stages. Such a surprisingly high TE in a species that is usually described as a water spender deserves further investigations.

Was the genotype ranking of TE similar to earlier studies?

The stability of genotype ranking among different environments is a very important question, which is only seldom addressed. There is currently no large-scale comparative study with the same poplar genotypes grown under different environments. A comparison of collections of genotypes grown in common plantations in France and Italy revealed very little similarity between the two sites (Dillen et al. 2011). Fortunately, the same set of six genotypes was used by several authors in field and greenhouse experiments. We computed the Spearman rank correlation coefficient for Δ13Clb and intrinsic TE between our results and earlier ones (see Table 4). The correlations were rather high in most cases except with the dataset from Marron et al. (2005) that differed from all others and was conducted in a greenhouse under very low irradiance. We can conclude from this rapid analysis that the ranking for Δ13Clb and intrinsic TE was rather stable in this set of genotypes under very different growth conditions.

Table 4.  Spearman rank correlation table (r) between genotype means for the isotopic discrimination between the atmosphere and bulk leaf matter (Δ13Cbl) and for the intrinsic transpiration efficiency at leaf level recorded for the six genotypes in different experiments including the present one
  δ 13 C lb Present δ 13 C lb Fichot δ 13 C lb Monclus δ 13 C lb Marron A/g s Present A/g s Fichot
  1. Data from the present paper (present), from Fichot et al. (2010, nursery grown rooted cuttings over 2 years), Marron et al. (2005, greenhouse grown cuttings after a few months) and Monclus et al. (2005, nursery grown rooted cuttings over 2 years). n = 6.

δ 13 C lb Present1     
δ 13 C lb Fichot0.6571    
δ 13Clb Monclus0.7140.7141   
δ 13 C lb Marron−0.257−0.1430.3711  
A/g s Present−0.886−0.543−0.8290.02851 
A/g s Fichot−0.714−0.714−0.6570.3140.8291

Origin of the genotypic differences in intrinsic TE at leaf level

Genotypic differences in intrinsic TE can be due to differences of net CO2 assimilation rates A or gs, or a combination of both (Farquhar & Richards 1984; Condon et al. 2004). The positive correlation of A with intrinsic TE and the negative one with Δ13Css show that A explained most of the variation of intrinsic TE and Δ13Css. Similar results were detected by Voltas et al. (2006) for Populus × euramericana. When we recorded Asat, gssat under saturating conditions, we changed significantly the genotype ranking of intrinsic TE. Indeed, Asat was much larger than A while gssat was very close to gs: under ambient conditions, the stomata were already at maximal opening and were not limiting CO2 assimilation. This observation also underlines that changes in the environment (irradiance but also water availability, acting on the two components of intrinsic TE) may potentially change the ranking among genotypes.

This observation contradicts at least partly that of a rather tight correlation between SD and Δ13Css (and therefore also intrinsic TE). In our conditions, SD was independent of gs; Dillen et al. (2009) and Fichot et al. (2010) detected even a negative correlation between gs and SD. Such a lack of correlation can be explained by interactive effects of pore depth, effective pore width and actual percentage of functional stomata (Aasamaa, Sober & Rahi 2001; Franks, Drake & Beerling 2009).

Correlation between TE and biomass accumulation (BM)

Whole-plant TE was tightly correlated to BM, but not at all to WU. A positive correlation between whole-plant TE and BM confirms that A had a larger effect than gs (Condon, Richards & Farquhar 1987; Virgona et al. 1990). Both negative (Martin & Thorstenson 1988; Condon, Farquhar & Richards 1990; Ehleringer et al. 1990; White, Castillo & Ehleringer 1990) and positive (Hubick, Farquhar & Shorter 1986; Wright, Hubick & Farquhar 1988) correlations were found between whole-plant TE and BM in crop species. We suggest that whole-plant TE was rather controlled by A than by gs under our conditions. This strengthens the hypothesis that breeding poplar genotypes for improved whole-plant TE would not necessarily come at the expense of productivity.

18O enrichment between source and leaf water (Δ18Olw) or leaf bulk matter (Δ18Olb)

Leaf water is 18O enriched with respect of the water source due to transpiration (Saurer et al. 1997; Barbour 2007). If the transpiration rate is driven by different levels of gs, then a negative relationship is to be expected between gs and Δ18Olw (Farquhar, Cernusak & Barnes 2007). Due to a large intra-genotypic variability, no genotypic effect was found here neither for Δ18Olw nor for Δ18Olb. At individual level, both Δ18Olw and Δ18Olb were negatively correlated to WU and gs. Similar negative correlations have been evidenced in cotton (Barbour et al. 2000), in a tropical tree, Ficus insipida (Cernusak et al. 2007a) and in durum wheat (Cabrera-Bosquet et al. 2009). Δ18Olw and Δ18Olb were 9.23 and 36.0‰, respectively; with a 26.8‰ difference. Such a difference is, under normal cellular temperature and pH, due to isotopic exchanges between water and the carbonyl groups of organic molecules and is usually close to 25–30‰ (Barbour 2007), close to the one we found here.


TE recorded at different integration scales (from leaf to whole-plant) in Populus deltoides × nigra genotypes displayed consistent genotype differences. Δ13C recorded from soluble sugars was an efficient predictor of intrinsic TE at leaf level and of whole-plant TE. Nevertheless, the magnitude of genotype variability of TE was larger at whole-plant level than at leaf level, showing that either nocturnal transpiration or more likely differences in respiration from non-photosynthetic organs had a large impact on whole-plant TE and on its variability. Genotypic differences of TE were due to differences of net CO2 assimilation rates (at leaf level) and of biomass production (at whole-plant level) rather than gs or transpiration rates. Genotypes with large carbon assimilation and fast growth were also the most efficient, which shows again that breeding for improved TE does not come at the expense of productivity in Populus × euramericana.


This research was supported by the European Union FP7 Project ‘Novel tree breeding strategies’ project number: FP7 – 211868 coordinated by Catherine Bastien (INRA Orléans). F.R. was supported by a grant from the Higher Education Commission (HEC) of Pakistan. The authors are grateful to many members of UMR 1137 Ecologie et Ecophysiologie Forestières, INRA and Université de Lorraine: Christian Hossann and Claude Bréchet for spending much time doing all isotope analyses; Nathalie Ningre, Franck Radnai and Cyril Buré for assistance during the experiment; Oliver Brendel and Daniel Epron for helpful discussions. Helpful comments by the editor and two external referees are also gratefully acknowledged.