Combined use of δ13C, δ18O and δ15N tracks nitrogen metabolism and genotypic adaptation of durum wheat to salinity and water deficit

Authors


Author for correspondence:
José Luis Araus
Tel: + 34 934021469
Email: j.araus@cgiar.org

Summary

  • Accurate phenotyping remains a bottleneck in breeding for salinity and drought resistance. Here the combined use of stable isotope compositions of carbon (δ13C), oxygen (δ18O) and nitrogen (δ15N) in dry matter is aimed at assessing genotypic responses of durum wheat under different combinations of these stresses.
  • Two tolerant and two susceptible genotypes to salinity were grown under five combinations of salinity and irrigation regimes. Plant biomass, δ13C, δ18O and δ15N, gas-exchange parameters, ion and N concentrations, and nitrate reductase (NR) and glutamine synthetase (GS) activities were measured.
  • Stresses significantly affected all traits studied. However, only δ13C, δ18O, δ15N, GS and NR activities, and N concentration allowed for clear differentiation between tolerant and susceptible genotypes. Further, a conceptual model explaining differences in biomass based on such traits was developed for each growing condition.
  • Differences in acclimation responses among durum wheat genotypes under different stress treatments were associated with δ13C. However, except for the most severe stress, δ13C did not have a direct (negative) relationship to biomass, being mediated through factors affecting δ18O or N metabolism. Based upon these results, the key role of N metabolism in durum wheat adaptation to salinity and water stress is highlighted.

Introduction

Durum wheat is the most cultivated crop in the south and east Mediterranean basin (http://www.fao.org/docrep/006/y4011e/y4011e04.htm), where drought and irrigation-induced salinity are the main constraints limiting productivity (Araus, 2004). Selecting genotypes that are either more drought-resistant (Araus et al., 2002) or more salt-tolerant (Munns & Tester, 2008) are complementary ways of improving durum wheat adaptation. However, effective phenotyping remains a bottleneck in breeding for adaptation to such abiotic stresses (Araus et al., 2008). Frequently, the traits that are most effective for phenotyping may differ with type and amount of stress. Moreover, while most studies aimed at elucidating traits and tools for phenotyping have addressed each stress individually, drought and salinity usually occur together under field conditions. This is the case, for example, for crops grown under deficit irrigation with brackish water.

Both drought and salinity induce water stress (Munns, 2002). Water stress affects plant growth via reduced carbon assimilation, tissue expansion and cell number (Hsiao, 1973; Tardieu et al., 2000), and strongly affects nitrogen metabolism (Hirel et al., 2007). Apparently, these are essentially uncoupled mechanisms (Tardieu et al., 2011), but feedback pathways between them may exist.

Natural 13C abundance in plant matter provides time-integrated information on the stress effects on photosynthetic carbon assimilation of C3 species (Farquhar et al., 1982; Tcherkez et al., 2011). Conditions inducing stomatal closure (e.g. water deficit, either directly or through salinity) restrict the CO2 supply to carboxylation sites, which then increases the carbon isotope composition (δ13C) of plant matter (Farquhar & Richards, 1984; Rivelli et al., 2002; Yousfi et al., 2010). The assessment of whether variation in δ13C is the result of changes in intrinsic photosynthetic capacity (A) or stomatal conductance (gs) remains challenging (Scheidegger et al., 2000; Farquhar et al., 2007). Oxygen isotope composition (δ18O) is a useful trait for this purpose, as it is largely unaffected by photosynthesis (Barbour & Farquhar, 2000; Farquhar et al., 2007). Rather, the δ18O of plant matter integrates the evaporative conditions throughout the crop cycle (Barbour et al., 2000) and, consequently, it has been proposed as a proxy for estimating gs, transpiration (Sheshshayee et al., 2005; Cabrera-Bosquet et al., 2009a, 2011; Cernusak et al., 2009a) and thus water status. To our knowledge, however, a thorough examination of δ18O changes under salinity or as a result of the combined effect of both drought and salinity is currently absent.

Nitrate reductase (NR; EC 1.6.6.1) and glutamine synthetase (GS; EC 6.3.1.2) are key enzymes responsible for N assimilation and are also connected with carbon metabolism (Masclaux-Daubresse et al., 2010). GS activity is one of the best physiological markers describing plant N status, whatever its developmental stage and N nutrition (Kichey et al., 2006; Bernard & Habash, 2009). Higher NR activity has been correlated with enhanced plant growth (Lam et al., 1996; Chen et al., 2003). In principle, the activity of both GS and NR decreases under salinity or drought (Foyer et al., 1998; Carillo et al., 2005). However, some studies in wheat indicate that moderate salinity decreases NR activity while increasing GS activity (Wang et al., 2007; Carillo et al., 2008). GS is involved not only in using ammonium produced by nitrate reduction, but also in the reassimilation of ammonium generated from photorespiration, proteolysis and processes that are increased by stress (Tsai & Kao, 2002; Hirel et al., 2007). Thus, rice tolerance to salinity has been related to increased GS activity (Sahu et al., 2001), and genotypic tolerance to salinity in foxtail millet has been associated with proline accumulation coupled with an increase in salt-induced GS activity (Veeranagamallaiah et al., 2007).

The natural variation in plant N isotope composition (δ15N) is potentially useful for genotypic screening under drought or salinity because it is linked to N metabolism, even though a complete knowledge of the underlying biochemical mechanisms is lacking (Cernusak et al., 2009b; Tcherkez, 2011). Isotope fractionation may occur during enzymatic assimilation of nitrate or ammonium into other N forms. Further fractionation may take place as a result of N recycling in the plant or through translocation, exudation or volatilization (Evans, 2001; Tcherkez & Hodges, 2008). Isotopic fractionation occurs during nitrate assimilation by NR, and ammonium absorption and assimilation by GS, so that there is, on average, a 2–3‰ depletion in plant 15N compared with source nitrate (Evans, 2001). δ15N sensitivity to photorespiratory rates (associated with GS activity) and N-input levels (NO3 reduction by NR) seems quite high (Tcherkez, 2011).

Here we characterized the response of four durum wheat genotypes with contrasting performances under salinity to different combinations of salinity and irrigation. Growth, the isotopic signatures of C, O and N, gas exchange, ion accumulation, total N concentration and the activities of NR and GS were analysed. The main objective was to examine the effectiveness of δ15N, δ13C and δ18O, alone or in combination, to track genotypic differences in biomass and N concentration for different arrangements of salinity and water treatments. In addition, this study is unique in that it proposes a conceptual model relating genotypic variability in the three isotopic signatures with N metabolism and above-ground biomass. Understanding the relationships between δ13C, δ18O, δ15N and plant growth under a wide range of abiotic stresses may help design more efficient breeding strategies, avoiding phenotyping redundancies by choosing in each case the most suitable trait(s). Moreover, studies examining the physiological basis of δ18O and δ15N and their potential breeding implications are indeed scarce and even contradictory, and the same applies to the relationships of δ13C and δ18O with nitrogen metabolism.

Materials and Methods

Plant material and growth conditions

Four recombinant inbred lines (RILs) of durum wheat (Triticum turgidum L. ssp. durum (Desf.) Husn.), here designated as RIL24, RIL30 (salt-susceptible) and RIL47, RIL85 (salt-tolerant), were chosen on the basis of their contrasting shoot biomass among a set of 112 RILs evaluated in a previous study for tolerance to continuous salinity during the vegetative stage (Yousfi et al., 2009). Plants were grown in controlled chambers (Conviron E15, Controlled Environments Ltd., Winnipeg, Manitoba, Canada) at the Experimental Fields of the University of Barcelona for > 1 month. Experimental growing conditions were as follows: 400 μmol m−2 s−1 photosynthetic photon flux density (PPFD), 70% relative humidity, 25°C day : 15°C night temperature and a 14 h photoperiod. Two seeds per pot were planted in 3 dm3 pots containing perlite and watered to field capacity with complete Hoagland solution (Hoagland & Arnon, 1950). After 1 wk, seedlings were thinned to one plant per pot. Subsequently, five different growth conditions were imposed:

  • FI or control (fully irrigated, 100% of container capacity, with complete Hoagland solution, 1.8 dS m−1)
  • FI-12 (fully irrigated with saline Hoagland solution, 12 dS m−1)
  • FI-17 (fully irrigated with saline Hoagland solution, 17 dS m−1)
  • DI (deficit irrigated to 35% of container capacity, with normal Hoagland solution)
  • DI-12 (deficit irrigated to 35% of container capacity with saline Hoagland solution, 12 dS m−1)

DI-17 has been shown to be too severe a treatment (Yousfi et al., 2010). Solutions were changed every 2–3 d. A completely randomized design was used to accommodate the two-way factorial experiment, with genotype and salinity–drought arrangement as main factors. Four single-pot replicates per factorial combination were used, totalling 80 pots. Water deficit was imposed progressively over 1 wk by decreasing irrigation. Salinity treatment was also imposed by adding NaCl progressively to the nutrient solution, starting with a salt concentration of 4 dS m−1 to reach the final concentrations of 12 dS m−1 (c. 120 mM NaCl) or 17 dS m−1 (c. 170 mM NaCl).

Gas exchange, plant biomass and leaf Chl

An infrared gas analyser (LI-6400 system, Li-Cor, Inc., Lincoln, NE, USA) was used to measure gas exchange just before harvesting (end of tillering stage) in the fully expanded upper leaf of the main plant tiller. Parameters measured were as follows: net CO2 assimilation (A), transpiration (T) and stomatal conductance (gs). Subsequently, the ratio of intercellular to ambient CO2 concentration (Ci/Ca) was calculated. The efficiency of excitation energy captured by open photosystem II (PSII) reaction centres (Fv’/Fm ) was also estimated in the same leaves. Chl content was estimated in the same leaf blades monitored for gas exchange using a portable meter (Minolta SPAD 502 Meter, Plainfield, IL, USA). Additional leaf blades were harvested and stored at −80°C for analysis of enzyme activity. The rest of the plant was then harvested; the shoots were oven-dried at 70°C for 48 h, weighed and ground finely. Roots of the same plants were washed with deionized water and then treated in the same way as shoots.

N concentration and stable isotope signatures

The total N concentration and the stable carbon (13C : 12C) and nitrogen (15N : 14N) isotope ratios in shoots and roots were measured using an elemental analyser (Flash 1112 EA; ThermoFinnigan, Bremen, Germany) coupled with an isotope ratio mass spectrometer (Delta C IRMS, ThermoFinnigan) (EA-IRMS), operating in continuous flow mode. Samples of c. 1 mg and reference materials were weighed into tin capsules, sealed and loaded into an automatic sampler before EA-IRMS analysis (Table 1). Values were expressed in δ notation (Coplen, 2008): δ13C = [(13C/12C)sample/(13C/12C)standard] −1, where ‘sample’ refers to plant material and ‘standard’ to Pee Dee Belemnite (PDB) calcium carbonate (Table 1). The same δ notation was used for the 15N : 14N ratio (δ15N), but in this case the standard referred to N2 in air (Table 2). In addition, the 18O : 16O ratios (expressed as δ18O) were determined in shoot samples of all plants tested (Table 1). Samples of c. 1 mg and reference materials were weighed into silver capsules, sealed, oven-dried at 60°C for not less than 72 h to remove moisture and loaded into an automatic sampler. The IRMS used was a Europa Scientific Geo 20-20 (Crewe, UK).

Table 1.   Genotype and treatment effects on biomass, gas-exchange parameters and carbon and oxygen isotope composition of four durum wheat genotypes grown under different combinations of salinity and water supply and their combinations
 BshootLCAgsCi/CaTFv’/Fmδ18Oshootδ13Cshootδ13Croot
  1. Bshoot, shoot biomass (g DW); LC, leaf Chl content (Special Products Analysis Division, SPAD, units); A, leaf net CO2 assimilation (μmol CO2 m−2 s−1); gs, stomatal conductance (mol CO2 m−2 s−1); Ci/Ca, ratio of intercellular to ambient CO2 concentration; T, transpiration rate (mmol H2O m−2 s−1); Fv’/Fm’, efficiency of excitation energy capture by open photosystem II (PSII) reaction centres; δ13C, stable carbon isotope composition (‰) of shoots (δ13Cshoot) and roots (δ13Croot); δ18Oshoot, stable oxygen isotope composition of shoots (‰).

  2. Genotype values are the means of 20 measurements (five treatments and four replications per treatment), while treatment values are the means of the 16 measurements (four genotypes and four replications per genotype). Means followed by different letters are significantly different (P < 0.05) according to Tukey’s honestly significant difference (HSD) test. Gas-exchange measurements were performed at 1200 μmol m−2 s−1 photosynthetic photon flux density (PPFD), 25°C, and a CO2 concentration of 400 μmol mol−1. Measurements of δ13C and δ18O were carried out at the Scientific-Technical Services of the University of Barcelona and at Iso-Analytical Limited (Crewe, Cheshire CW2 8UY, UK), respectively. International isotope secondary standards of known 13C : 12C ratios (IAEA-CH7, IAEA-CH6 and USGS 40) and 18O : 16O ratios (IAEA-CH-6, IAEA-C-3 and IAEA-601) were used for calibration to a precision of 0.1 and 0.2‰, respectively.

  3. Treatments: FI, full irrigation (i.e. control) with normal Hoagland solution; FI-12 dS m−1, full irrigation with Hoagland solution at 12 dSm−1; FI-17 dSm−1, full irrigation with Hoagland solution at 17 dS m−1; DI, deficit irrigation (35% pot capacity) with normal Hoagland solution; DI-12 dS m−1, deficit irrigation with Hoagland solution at 12 dS m−1. ns, not significant.

Genotype
 Tolerant RIL4731.64b43.68a5.13b0.17ab0.41a1.83a0.58ab27.47ab−29.52a−29.43a
 Tolerant RIL 8534.42b44.26a7.82c0.23b0.42a2.36a0.63b27.24a−29.62a−29.73b
 Susceptible RIL 2424.12a41.72a4.76a0.16a0.40a2.06a0.55a28.12c−28.92b−28.99c
 Susceptible RIL 3024.95a44.24a5.11ab0.13a0.35a1.43a0.56a27.80bc−27.88c−27.54d
Treatment
 FI54.33c42.12ab16.04c0.47c0.78c5.85b0.70b27.00a−30.76a−29.99a
 FI- 12 dS m−127.00b43.22ab2.54a0.18b0.33b1.57a0.55a27.96c−28.68c−28.88b
 FI-17 dS m−115.15a40.84a2.53a0.04a0.20a0.52a0.52a27.79bc−27.79d−28–29c
 DI30.20b48.63b5.07b0.12ab0.40b0.83a0.58a27.98c−29.27b−28.81b
 DI- 12 dS m−114.08a41.53ab2.45a0.02a0.28ab1.03a0.52a27.34ab−28.46c−28.71b
Level of significance
 Genotype (G)0.000ns0.0020.049nsns0.0010.0000.0000.000
 Treatment (T)0.0000.0110.0000.0000.0000.0000.0000.0060.0000.002
 G × T interaction0.000nsnsnsns0.010ns0.003nsns
Table 2.   Genotype and treatment effects on nitrogen concentration, nitrogen isotope composition, glutamine synthetase and nitrate reductase activities of four durum wheat genotypes grown under different combinations of salinity and water supply and their combinations
 NshootNrootδ15Nshootδ15NrootGSshootNRshoot
  1. N, nitrogen concentration (mmol g−1 DW) of shoots (Nshoot) and roots (Nroot); δ15N, nitrogen isotope composition (‰) of shoots (δ15Nshoot) and roots (δ5Nroot); GSshoot, shoot glutamine synthetase activity (μmol g−1 FW h−1); NRshoot, shoot nitrate reductase activity (μmol g−1 FW h−1).

  2. Genotype values for nitrogen content and N isotope composition are the means of 20 measurements (five treatments and four replications per treatment), while treatment values are the means of the 16 measurements (four genotypes and four replications per genotype). Genotype and treatment values for enzyme activities are the means of 15 and 13 measurements, respectively. Means followed by different letters are significantly different (< 0.05) according to Tukey’s honestly significant difference (HSD) test. N concentration and stable isotope analyses were carried out at the Scientific-Technical Services of the University of Barcelona. Secondary isotope standards of known 15N : 14N ratios (IAEA N1 and IAEA N2 and IAEA NO3) were used for calibration to a precision of 0.2‰. The mean δ15N of the fertilizer provided by the Hoagland solution was 0.6‰. GS and NR activities were determined at the Department of Plant Biochemistry and Molecular Biology, Faculty of Chemistry, University of Sevilla.

  3. Treatments: FI, full irrigation (i.e. control) with normal Hoagland solution; FI-12 dS m−1, full irrigation with Hoagland solution at 12 dSm−1; FI-17 dSm−1, full irrigation with Hoagland solution at 17 dS m−1; DI, deficit irrigation (35% pot capacity) with normal Hoagland solution; DI-12 dS m−1, deficit irrigation with Hoagland solution at 12 dS m−1. ns, not significant.

Genotype
 Tolerant RIL474.39c1.49a2.47b3.45a544.54b2.08c
 Tolerant RIL 854.34c1.54a2.92c3.57a442.71b1.83bc
 Susceptible RIL 243.45b1.63a2.12b3.57a280.99a1.57b
 Susceptible RIL 302.91a1.48a1.68a3.22a249.37a0.99a
Treatment
 FI4.95d2.14b3.13b2.20a488.02c3.82c
 FI- 12 dS m−14.20c1.21a1.83a3.41bc420.16bc1.25b
 FI-17 dS m−12.98b1.19a1.84a4.68c312.28ab0.78a
 DI3.95c2.00b2.81b3.17b386.27bc1.43b
 DI- 12 dS m−12.36a1.07a1.89a3.92bc241.61a0.68a
Level of significance
 G0.000ns0.000ns0.0000.000
 T0.0000.0000.0000.0000.0000.000
 G × T0.000nsnsnsnsns

Ion analysis

Ion analysis was performed in shoots and roots (Table 3) by inductively coupled plasma emission spectrometry (L3200RL, Perkin Elmer, Rodgau, Germany).

Table 3.   Genotype and treatment effects on ion concentration of shoots and roots of four durum wheat genotypes grown under different combinations of salinity and water supply and their combinations
 ShootRoot
Na+K+Ca2+PMg2+K+/Na+Ca2+/Na+Na+K+Ca2+PMg2+K+/Na+Ca2+/Na+
  1. Genotype values are the means of 20 measurements (five treatments and four replications per treatment), while treatment values are the means of the 16 measurements (four genotypes and four replications per genotype). Concentrations are expressed as mmol g−1 DW. Means followed by different letters are significantly different (< 0.05) according to Tukey’s honestly significant difference (HSD) test. For each sample, 100 mg of dry material was digested with 3 ml of concentrated HNO3 and 2 ml H2O2. Samples were placed overnight in an oven at 90°C. After digestion, each sample was then brought up to 30 ml final volume with deionized water. The amounts of Na+, Ca2+, K+, P, and Mg2+ per sample were determined with an inductively coupled plasma emission spectrometer (L3200RL) at the Scientific-Technical Services of the University of Barcelona.

  2. Treatments: FI, full irrigation (i.e. control) with normal Hoagland solution; FI-12 dS m−1, full irrigation with Hoagland solution at 12 dSm−1; FI-17 dSm−1, full irrigation with Hoagland solution at 17 dS m−1; DI, deficit irrigation (35% pot capacity) with normal Hoagland solution; DI-12 dS m−1, deficit irrigation with Hoagland solution at 12 dS m−1. ns, not significant.

Genotype
 Tolerant RIL470.98b0.78a0.07a0.17a0.06a2.72b0.18a0.75a0.46a0.20ab0.13a0.04a1.52a0.65a
 Tolerant RIL 851.04b0.79a0.09b0.20b0.06a2.09a0.21ab0.69a0.46a0.19ab0.13a0.04a1.41a0.55a
 Susceptible RIL 240.78a1.01b0.09b0.21b0.07b2.94b0.26c0.59a0.47a0.17a0.11a0.04a1.78a0.57a
 Susceptible RIL 300.96b0.84a0.07a0.18a0.06a2.82b0.21ab0.72a0.51b0.22b0.14a0.04a1.47a0.53a
Treatment
 FI0.19a1.55c0.11b0.25d0.07c8.25c0.59c0.15a0.72b0.21c0.21d0.04b4.73c1.38c
 FI- 12 dS m−11.48c0.54a0.07a0.20c0.06b0.36a0.04a0.90b0.27a0.15b0.12b0.02a0.30a0.18a
 FI-17 dS m−11.58c0.46a0.06a0.18b0.05a0.29a0.03a1.16c0.33a0.09a0.08ab0.03a0.29a0.08a
 DI0.33b1.22b0.11b0.19bc0.08d3.69b0.35b0.34a0.71b0.35d0.16c0.06c2.06b1.05b
 DI- 12 dS m−10.98b0.55a0.05a0.10a0.06ab0.58a0.05a0.86b0.31a0.13ab0.05a0.04b0.36a0.15a
Level of significance
 G0.0250.0010.0000.0000.0040.000nsnsns0.003nsnsnsns
 T0.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.0000.000
 G × Tns0.0210.0010.0010.0000.000nsnsnsnsnsnsnsns

Enzyme activity determinations

For the assay of GS and maximum NR enzyme activities, leaf samples were ground to a fine powder in liquid N2 and resuspended (5 ml g−1) in the following extraction buffer: 100 mM Tris-HCl supplemented with 1 mM dithiothreitol (DTT), 0.1 Triton X-100, 10 mM phenylmethylsulphonyl fluoride (PMSF) 10 mM EDTA. To determine actual NR activity, ethylendiaminetetraacetic acid (EDTA) was replaced in the buffer by 10 mM MgCl2. The leaf powder was mixed with the buffer for 10 s at 4°C with a pellet homogenizer. The homogenates were centrifuged for 15 min at 27 000 g and the supernatants were immediately assayed.

Total GS activity was assayed using the ‘biosynthetic’ reaction (Márquez et al., 2005). The standard reaction mixture (final volume 100 μl) contained: 10 μmol of Tris-HCl (pH 7.50 at 25°C), 200 μmol of L-glutamate, 5 μmol of NH4Cl, 5 μmol of MgCl2, 0.3 μmol of ATP (from a 100 mM stock solution, pH 7.5) and 5–10 μl of a 1 : 100 dilution of the crude extract. Duplicates of each enzyme assay were carried out and incubated at 37°C for 15 and 30 min to confirm that the amount of product obtained was linear over time. Inorganic phosphate released by ATP hydrolysis was determined using the malachite green method as described previously (Márquez et al., 2005).

Nitrate reductase activity was determined as described previously (Pajuelo et al., 2002). The standard reaction mixture (final volume 500 μl) contained 25 μmol of Tris-HCl (pH 7.50 at 25°C), 10 nmol of flavin adenine dinucleotide (FAD), 50 μmol of KNO3, 0.15 μmol of nicotinamide adenine dinucleotide (NADH; from a freshly made 3 mM stock solution in 10 mM Tris-HCl, pH 7.50) and 15–30 μl of crude extract plus 220–235 μl extraction buffer. The reaction was done in duplicate and incubated at 30°C for 30 and 60 min to confirm that the amount of product obtained was linear over time. The amount of nitrite produced was determined as described previously (Márquez et al., 2005). Since both maximum and actual NR activities were strongly correlated (r = 0.93) with a slope near one, only maximum NR activity was subsequently used in the study.

Statistical analysis

Data were subjected to factorial ANOVA to test for the effects of treatment (irrigation-salinity regime, genotype, and their interaction). Mean comparisons for genotype effects were performed using contrasts of tolerant vs susceptible material and Tukey’s honestly significant difference (HSD) test. In order to test the association between shoot biomass and the set of physiological traits, linear stepwise models across genotypes were constructed that were independent for each treatment, with P = 0.05 as the criterion for variables to be either included or removed from the model. The set of physiological parameters was also divided into three categories as follows: photosynthetic traits (including δ13C and δ18O), N metabolism traits (including δ15N), and ion concentrations. For each trait category, the genotype–treatment combinations (i.e. four genotypes crossed with five treatments) were subjected to unweighted pair group method with arithmetic mean (UPGMA) cluster analysis to summarize the relative merit of genotypic effects and growing conditions as causes of changes in the observed plant responses. Finally, we performed path analyses (Li, 1975) to quantify the relative contributions of direct and indirect effects of stable isotopes on above-ground biomass. This methodology offers the possibility of building associations between variables on prior knowledge. A path analysis determines simple correlations between independent factors (in this case, δ13C and δ18O), and regresses them on each intermediary (NR and GS activities, δ15N, N concentration) or dependent factor (biomass) to obtain direct effects in the form of partial regression coefficients (i.e. path coefficients). In this way, direct (i.e. from δ13C and δ18O) and indirect (i.e. through the pathway of enzyme-based N metabolism) mechanisms that play potential roles in biomass variation and involving traits that exhibited genotypic differences were proposed, as detailed in the conceptual model displayed in Fig. 6(a). This model was aimed at understanding biomass responses to genotypic differences in water status and photosynthetic carbon and nitrogen performance, and was tested under each growth condition. For each condition, a reduced model was identified as reproducing the original correlation matrix nearly as well as the full, saturated model (i.e. in which there is a direct path from each variable to every other variable), as shown by goodness-of-fit tests. The reduced models were also compared with the independence model (one where all possible paths are deleted) by a comparative fit index (CFI) (Arbuckle, 1997). Data were analysed using the SAS package (SAS Institute Inc., Cary, NC, USA).

Results

Growth, photosynthetic traits and carbon and oxygen isotope signatures

Treatments significantly affected biomass, leaf Chl, gas exchange, and the stable carbon and oxygen isotope signatures (Table 1). The most severe treatments in terms of decreasing biomass were full irrigation with highly saline water (FI-17) and the combination of deficit irrigation with moderately saline water (DI-12). Both showed shoot biomass that was < 30% of the values of the control (FI). Intermediate treatments were full irrigation with moderately saline water (FI-12) and deficit irrigation with nonsaline water (DI), with 50 and 48% reductions in biomass compared with FI, respectively. Leaf Chl was far less affected, and it even increased in DI compared with FI. The gas-exchange traits A, gs, Ci/Ca, T and the ratio Fv’/Fm strongly decreased, compared with FI, in response to all stress treatments. As for biomass, FI-17 and DI-12 showed the lowest gs and Ci/Ca, while FI-12 and DI had intermediate values. Stress treatments increased shoot δ18O compared with FI, but the most severe treatments (FI-17 and DI-12) showed the least change. Root and shoot δ13C also increased in the stress treatments compared with FI, but in this case the highest values were observed under the most stressful conditions.

Tolerant genotypes exhibited significantly higher biomass (c. 35%) than susceptible genotypes. Tolerant genotypes also showed higher A, gs, Ci/Ca and Fv’/Fm, and lower shoot δ18O and root and shoot δ13C (Table 1). The genotype × treatment interaction (G × T) was only significant for biomass, T and δ18O. Genotypic differences between pairs of tolerant and susceptible genotypes were also examined for each growing condition, and especially for traits displaying significant G × T interaction (Supporting Information, Table S1). Except for δ13C of shoots and roots (lower in the tolerant genotypes), in the absence of stress (FI) there were no significant differences for any of the traits. By contrast, genotype differences in biomass (higher in tolerant lines) were observed for the stress treatments, and also in δ18O and Fv’/Fm, but only under the least stressful conditions (FI-12 and DI). A crossover interaction was observed for T: susceptible genotypes exhibited higher transpiration than tolerant genotypes under DI-12, but lower values for the salinity-only treatments.

Nitrogen concentration, δ15N and enzyme activities

Treatments significantly affected N concentration, δ15N, and GS and NR activities (Table 2). Stress treatments decreased N concentration in shoots and roots compared with FI. As for biomass, the most severe treatments were FI-17 and DI-12. Root δ15N increased in all stress treatments, showing the highest values in the two most stressful treatments. By contrast, stress treatments significantly decreased shoot δ15N. The activities of GS and NR also decreased in response to stress compared with FI, especially in FI-17 and DI-12. NR activity was much more affected in relative terms than GS by stress conditions, with reductions of 60% and 14% in the least severe treatments (FI-12 and DI), as compared with FI.

The genotype effect was significant for N concentration and δ15N, but only in shoots, with values for both traits being higher in tolerant than in susceptible genotypes. Activities of GS and NR were also higher in tolerant genotypes. Except for shoot N, no significant G × T interaction was detected. Genotypic differences between pairs of tolerant and susceptible genotypes were also examined for each growing condition (Table S2). Under full irrigation, no differences existed for the set of traits, except for GS activity (higher in tolerant lines). By contrast, the N concentration of susceptible genotypes was reduced to a greater extent than in tolerant lines for all stress treatments, thus causing a significant G × T interaction. Also, δ15N in shoots was significantly higher in tolerant genotypes for each stress treatment, but there were no differences in either N concentration or δ15N in roots. GS and NR activities were also higher in tolerant genotypes for each stress treatment.

Ion concentrations

Treatments significantly affected ion concentrations in both shoots and roots (Table 3). Stress treatments involving salt application strongly increased Na+ and decreased K+, Ca2+, Mg2+, and P, and the ratios K+ : Na+ and Ca2+ : Na+ in both plant parts. DI also increased Na+ in shoots and roots compared with FI, but this was to a much smaller extent than was observed with saline treatments. In shoots, K+ was also lower under DI than under FI, while no differences were observed in roots. The genotype effect was significant for all ions in shoots, but only for Ca2+in roots. However, no clear pattern of differences between tolerant and susceptible genotypes was found for most ions in shoots, except for K+ (with susceptible genotypes showing a higher concentration) and Na+ (with tolerant genotypes showing a higher concentration). G × T interactino was significant for all ions in shoots except for Na+, while there was no G × T interaction in roots. In most cases, there were no significant differences between pairs of tolerant and susceptible genotypes for ions measured in each growing condition and plant part (Table S3).

Effect of genotype and growing condition on physiological traits

Cluster analyses (Fig. 1) were performed as a way of summarizing the relevance of genotypic effects and growing conditions on changes in traits related to gas exchange (Table 1), N metabolism (Table 2) and ion accumulation (Table 3). Gas-exchange traits (including δ13C and δ18O) provided a good separation of treatments but did not allow for clear genotype differentiation (Fig. 1a). By contrast, traits related to N metabolism (including δ15N) discriminated tolerant from susceptible genotypes (Fig. 1b). Finally, the cluster based on ion concentrations allowed for an almost perfect separation among treatments (Fig. 1c).

Figure 1.

Cluster analysis of the five growing conditions assayed, using the complete set of physiological parameters studied in this work as variables. These parameters were divided into three categories as follows: (a) photosynthetic traits and C and O isotopic signatures; (b) N metabolism traits (including N isotopic signatures, N content and glutamine synthetase (GS) and nitrate reductase (NR) activities); and (c) mineral nutrition traits. FI, full irrigation (i.e. control) with normal Hoagland solution; FI-12 dS m−1, full irrigation with Hoagland solution at 12 dSm−1; FI-17 dSm−1, full irrigation with Hoagland solution at 17 dSm−1; DI, deficit irrigation (35% pot capacity) with normal Hoagland solution; DI-12 dS m−1, deficit irrigation with Hoagland solution at 12 dS m−1.

Relationships of biomass and nitrogen concentration with physiological traits across treatments

Nitrogen concentration was positively related to biomass across growing conditions and genotypes up to values near 5 mmol N g−1, with a saturation response observed afterwards (Fig. 2). Biomass was negatively associated with shoot δ13C and positively with shoot δ15N, whereas no clear trend was observed between biomass and shoot δ18O (Fig. 3), except when the most severe treatments (FI-17 and DI-12) were discarded (inset Fig. 3c). N concentration was linearly correlated with δ13C (negatively) and with δ15N (positively), and also with δ18O, albeit in a weaker manner. The association between biomass and either GS or NR activity was positive (Fig. 4), but tighter in the latter case. N concentration was also positively correlated with the activities of both enzymes, either linearly (GS) or exponentially (NR). Both GS and NR activities correlated positively with shoot δ15N and negatively with both δ13C and δ18O, but, in this last case, weakly (Fig. 5). Overall, NR activity was related to both isotopic signatures more strongly than GS activity.

Figure 2.

Relationship between nitrogen content and shoot biomass across the four genotypes within each of the five growing conditions assayed. Each point represents the individual value for a given replication and genotype within a growing condition. Variables mentioned in this figure are from the aerial parts of plants. FI, full irrigation (i.e. control) with normal Hoagland solution; FI-12 dS m−1, full irrigation with Hoagland solution at 12 dSm−1; FI-17 dSm−1, full irrigation with Hoagland solution at 17 dS m−1; DI, deficit irrigation (35% pot capacity) with normal Hoagland solution; DI-12 dS m−1, deficit irrigation with Hoagland solution at 12 dS m−1. ***P < 0.001.

Figure 3.

The relationship of shoot biomass with carbon isotope composition (δ13C) (a), oxygen isotope composition (δ18O) (b) and nitrogen isotope composition (δ15N) (c). The relationship of nitrogen content with carbon isotope composition (δ13C) (d), oxygen isotope composition (δ18O) (e) and nitrogen isotope composition (δ15N) (f). Variables mentioned in this figure are from the aerial parts of plants. The five treatments and the four genotypes are plotted together. Each point represents the individual value for a given replication and genotype within a growing condition. FI, full irrigation (i.e. control) with normal Hoagland solution; FI-12 dS m−1, full irrigation with Hoagland solution at 12 dSm−1; FI-17 dSm−1, full irrigation with Hoagland solution at 17 dS m−1; DI, deficit irrigation (35% pot capacity) with normal Hoagland solution; DI-12 dS m−1, deficit irrigation with Hoagland solution at 12 dS m−1. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 4.

The relationship of shoot biomass with glutamine synthetase activity (a) and nitrate reductase activity (b). The relationship of nitrogen content with glutamine synthetase activity (c) and nitrate reductase activity (d). Variables mentioned in this figure are from the aerial parts of plants. The five treatments and the four genotypes are plotted together. Each point represents the individual value for a given replication and genotype within a growing condition. FI, full irrigation (i.e. control) with normal Hoagland solution; FI-12 dS m−1, full irrigation with Hoagland solution at 12 dSm−1; FI-17 dSm−1, full irrigation with Hoagland solution at 17 dS m−1; DI, deficit irrigation (35% pot capacity) with normal Hoagland solution; DI-12 dS m−1, deficit irrigation with Hoagland solution at 12 dS m−1. **P < 0.01.

Figure 5.

The relationship of glutamine synthetase activity with carbon isotope composition (δ13C) (a); oxygen isotope composition (δ18O) (b) and nitrogen isotope composition (δ15N) (c). The relationship of nitrate reductase activity with carbon isotope composition (δ13C) (d); oxygen isotope composition (δ18O) (e) and nitrogen isotope composition (δ15N) (f). Variables mentioned in this figure are from the aerial parts of plants. The five treatments and the four genotypes are plotted together. Each point represents the individual value for a given replication and genotype within a growing condition. FI, full irrigation (i.e. control) with normal Hoagland solution; FI-12 dS m−1, full irrigation with Hoagland solution at 12 dSm−1; FI-17 dSm−1, full irrigation with Hoagland solution at 17 dS m−1; DI, deficit irrigation (35% pot capacity) with normal Hoagland solution; DI-12 dS m−1, deficit irrigation with Hoagland solution at 12 dS m−1. **P <  0.01.

Relationships between biomass and physiological traits across genotypes

A multiple linear regression (stepwise) explaining biomass variation across genotypes as a function of traits related to gas exchange, N metabolism and ion accumulation was fitted independently for each growing condition (Table 4). δ13C was chosen as the first explanatory variable in the control (FI, positive effect) and the most stressful treatment (DI-12, negative effect). Additional variables chosen in the FI model were also photosynthetic traits (A and gs), while no other traits were included in the DI-12 model. By contrast, the first variable entering the model in other stress treatments was related to N metabolism: N concentration in the FI-12 model and GS activity in the FI-17 and DI models. δ18O was selected as the second variable in two out of three models. Overall, > 80% of biomass variability was explained by the combination of two to three independent variables in four of the models, whereas 66% of the total variability was explained by just one variable in the remaining model (DI-12).

Table 4.   Multiple linear regressions (stepwise) explaining shoot biomass (Bshoot) variation across genotypes for each of five growing conditions, with traits related to photosynthetic performance (Table 1), nitrogen metabolism (Table 2) and ion accumulation (Table 3) used as independent variables
TreatmentsVariable chosenR2Final stepwise model
  1. Treatments: FI, full irrigation (i.e. control) with normal Hoagland solution; FI-12 dS m−1, full irrigation with Hoagland solution at 12 dSm−1; FI-17 dSm−1, full irrigation with Hoagland solution at 17 dS m−1; DI, deficit irrigation (35% pot capacity) with normal Hoagland solution; DI-12 dS m−1, deficit irrigation with Hoagland solution at 12 dS m−1.

  2. Only parameters entering the models are shown: δ13C, carbon isotope composition (‰); A, leaf net CO2 assimilation (μmol CO2 m−2 s−1); gs, stomatal conductance (mol CO2 m−2 s−1); N, leaf nitrogen concentration (mmol g−1 DW); δ18O, oxygen isotope composition (‰); GS, glutamine synthetase activity (μmol g−1 FW h−1); δ15N, nitrogen isotope composition (‰). ***, < 0.001. All variables chosen by the different models refer to shoots.

FIδ13C0.38Bshoot = 212.91 + 6.27 δ13C + 1.59 + 20.99 gs
δ13C, A0.65
δ13C, A, gs0.86***
FI -12 dS m−1N0.75Bshoot = 334.96 + 5.32 N– 11.82 δ18O
N, δ18O0.93***
FI -17 dS m−1GS0.80Bshoot = 7.55 + 0.20 GS + 0.55 A
GS, A0.83***
DIGS0.66Bshoot = 156.76 + 0.03 GS − 4.89 δ18O
GS, δ18O0.83***
DI -12 dS m−1δ13C0.66***Bshoot = −56.31− 2.45 δ13C

A conceptual model was proposed (Fig. 6a) separating direct acclimation responses in biomass related to water status (through δ18O) and photosynthesis (through δ13C) from those likely to correspond to indirect effects linked to N metabolism (through GS and NR activities, N and δ15N). The five path models provided an acceptable fit to the data (Fig. 6; CFI > 0.9 and > 0.05 in all cases; as the objective here is to develop models that fit the data well, a nonsignificant χ2 is preferred). In all cases, δ13C had a strong negative effect on NR activity. Conversely, δ13C had a direct positive effect on biomass only in FI, while it exhibited a direct negative effect in DI-12. Significant paths corresponding to a direct (negative) effect of δ18O on biomass were observed in the mildest salinity treatment (FI-12) and also in DI. The indirect explanation of biomass via N metabolism traits showed significant directional effects of NR on GS for those cases (FI-17, DI, DI-12) where biomass was significantly associated with changes in N concentration (i.e. the third possible path to biomass determination after δ13C and δ18O). On the other hand, GS was only significantly (and negatively) affected by δ13C in DI-17, but in this case (and also in FI) N metabolism effects on biomass seemed negligible. Under FI, DI and FI-12, δ15N was directionally affected by GS (positively) and, to a lesser extent, by NR (either positively or negatively), but only in FI-12 was there a significant directional effect of δ15N on N concentration, which in turn affected biomass. In summary, the mildest stress treatments (DI and FI-12) pointed to the dependence of biomass on both δ18O and N metabolism traits, whereas under more severe growing conditions the influence of either N metabolism (in FI-17) or δ13C (in DI-12) on biomass appeared relevant. Under control conditions, only δ13C was directly associated with biomass, albeit with an opposite sign to that found in DI-12.

Figure 6.

Path analyses of four durum wheat (Triticum turgidum ssp. durum) genotypes grown under five combinations of drought and salinity. The conceptual model quantifying the relative contributions of direct and indirect effects of stable isotopes on above-ground biomass is shown in (a). The different combinations of drought and salinity are as follows: (b) FI, full irrigation with normal Hoagland solution; (c) FI-12 dSm−1, full irrigation with Hoagland solution at 12 dSm−1; (d) FI-17 dSm−1, full irrigation with Hoagland solution at 17 dS m−1; (e) DI, deficit irrigation (35% pot capacity) with normal Hoagland solution; (f) DI-12 dSm−1, deficit irrigation with Hoagland solution at 12 dS m−1. Variables mentioned in this figure are from aerial plant parts. A single-headed arrow between two variables denotes a hypothesis of direct causation, whereas a double-headed arrow reflects correlation without necessarily a direct causal relationship. Dashed lines indicate negative relationships. The width of arrows is proportional to the path coefficient values. Overall fit statistics for each path model (chi-squared and comparative fit index, CFI), the latter useful for small sample sizes (with values > 0.9 taken as indicative of a good fit), are shown at the bottom right of each panel. NR, nitrate reductase; GS, glutamine synthetase. *P < 0.05, **P < 0.01, ***P < 0.001.

Discussion

Treatments that most affected growth were full irrigation with high amounts of salinity or a combination of deficit irrigation and salinity, in agreement with previous studies (Ayers & Westcott, 1989; Yousfi et al., 2009, 2010). Moreover, consistent differences in biomass were observed between tolerant and susceptible genotypes regardless of the stress conditions (Yousfi et al., 2009, 2010), although such differences vanished in the absence of stress.

Are ion concentrations, Chl content and short-term photosynthetic parameters good indicators of genotypic tolerance to salinity?

Changes in shoot ion concentrations under distinct salinity–irrigation arrangements are in agreement with previous studies (Munns & Tester, 2008). However, it has been reported that lower Na+ concentrations and higher K+ : Na+ and Ca2+ : Na+ ratios improve resistance to salinity (Hu & Schmidhalter, 2005) and, therefore, selecting for low Na+ uptake and enhanced K+ : Na+ ratios has been proposed as a screening strategy (Munns et al., 2000). In clear contrast, shoots from tolerant genotypes in this study did not exhibit lower Na+ and higher K+, Ca2+ and Mg2+ than susceptible ones, or higher K+ : Na+ and Ca2+ : Na+ ratios, but rather had the opposite tendency (Table S3). This observation points to increased osmotic adjustment of tolerant lines to saline conditions through the incorporation of available ions such as Na+ (Munns, 2002; Cuin et al., 2009).

Chlorophyll content per unit leaf area has been proposed as a screening criterion for wheat tolerance to salinity (Munns & James, 2003). However, the expected decrease in leaf Chl as a result of salt toxicity was probably offset in our study by an increase in leaf thickness or packing of mesophyll cells as a response to water stress, which eventually translated into constant Chl readings (James et al., 2006; Yousfi et al., 2009; Munns et al., 2010).

Salinity, water stress and their combined effect induced a strong decrease in photosynthesis and transpiration through a decrease in gs (Ouerghi et al., 2000; Yousfi et al., 2009, 2010). Several studies (James et al., 2008; Rahnama et al., 2010) suggest that screening for high gs may be the most effective way of identifying fast-growing genotypes in saline soils. In our study, however, either gs or other gas-exchange parameters could not differentiate between tolerant and susceptible genotypes (Table S1). Yousfi et al. (2010) found comparable results in wheat during the reproductive stage.

Growing conditions and genotypic effects on δ18O and δ13C under salinity and drought conditions

Plant δ13C increased under water-limiting and salinity conditions, as reported elsewhere (Isla et al., 1998; Condon et al., 2002; Araus et al., 2003; Yousfi et al., 2009, 2010). The increase in δ13C was not completely a consequence of a stomatal limitation to photosynthesis, since the association between δ13C and δ18O across environments, although positive, was weak (0.281 P < 0.05). Increases in plant δ18O as a response to drought have been observed in cereals (Ferrio et al., 2007; Cabrera-Bosquet et al., 2009a; Araus et al., 2010). A higher δ18O may be linked to a decrease in gs, lower transpiration and reduced leaf cooling, therefore resulting in lower 18O enrichment at evaporation sites (Barbour, 2007; Farquhar et al., 2007). To the best of our knowledge this is the first study in wheat reporting on genetic differences in plant δ18O under salinity. In agreement with a higher long-term gs, tolerant genotypes exhibited lower δ13C and δ18O than susceptible ones. Our results, therefore, highlight the advantage of time-integrative traits over instantaneous (but time-consuming) gas-exchange measurements for genotype evaluation (Araus et al., 2002, 2008). Moreover, tolerant genotypes exhibited constitutively (i.e. in absence of stress) lower δ13C (and also a tendency to lower δ18O) than susceptible ones (Table S1), in agreement with previous studies using δ13C (Araus et al., 2003; Yousfi et al., 2009, 2010) and δ18O (Cabrera-Bosquet et al., 2009b), as well as reviews on yield potential and crop performance under water stress (Blum, 2005, 2009; Araus et al., 2008), which conclude that drought resistance and yield potential are compatible.

Growing conditions and genotypic effects on N concentration and NR and GS activities

The effects of salinity and deficit irrigation that decreased shoot N concentration paralleled those obtained for shoot biomass, with tolerant genotypes exhibiting higher N concentration than susceptible ones. Besides a direct osmotic effect on plant water availability, the parallels between the changes in biomass and N concentration suggest that salinity and water deficit also affected growth through an effect on N metabolism (Hirel et al., 2007). In relation to this, NR and GS activities also decreased under water deficit, salinity and in combination, but overall, tolerant genotypes exhibited higher GS and NR activities. A number of studies have reported decreased NR activity in response to water stress (Foyer et al., 1998; Correia et al., 2005) and salinity (Wang et al., 2007; Carillo et al., 2008). Salt was observed to inhibit nitrate transport to the leaf mainly because of nitrate/chloride competition, consequently affecting NR activity (Rao & Gnanam, 1990; Abd-El Baki et al., 2000). There is also evidence that photosynthesis regulates nitrate reduction by modulating NR activity (Kaiser & Förster, 1989; Kaiser & Brendle-Behnisch, 1991), which is in accordance with current results showing that stress treatments decrease both photosynthesis and NR activity. Wang et al. (2007) reported that GS activity of wheat genotypes also decreased under salinity. GS plays a key role under osmotic stress, reassimilating nitrogen from increased amino acid catabolism, and producing protective nitrogen compounds (Brugière et al., 1999; Diaz et al., 2010). Thus, the higher GS activity in tolerant genotypes could be related to the fact that proline and glycine betaine accumulate in durum wheat under salinity (Carillo et al., 2008). Several quantitative trait loci for important agronomic traits in wheat colocalize with the GS marker (Li et al., 2011). Furthermore, overexpression of GS results in enhanced tolerance to salt stress and high light intensity through an improved capacity for photorespiration, therefore avoiding photoinhibition (Kozaki & Takeba, 1996; Hoshida et al., 2000).

Effect of growing conditions and genotype on δ15N of shoots and roots

An array of salinity–drought combinations modifies δ15N in shoots and roots relative to controls, with tolerant genotypes exhibiting higher shoot δ15N than susceptible genotypes (Table S1), suggesting that stress conditions influence N uptake and/or assimilation (Handley et al., 1997; Ellis et al., 2002; Yousfi et al., 2009, 2010). Decreases in shoot δ15N have been reported in cereals as a response to salinity (Yousfi et al., 2009, 2010) and deficit irrigation (Robinson et al., 2000; Raimanová & Haberle, 2010). Reduced gs as a result of either salinity or water stress, or a combination of both, would lead to a reduction in the loss of ammonia and nitrous oxide, hence decreasing δ15N (Farquhar et al., 1980; Smart & Bloom, 2001). A high external N concentration relative to a modest demand would also lead to salinity-induced depletion in plant 15N (Mariotti et al., 1982). Consequently, the suboptimal growing conditions associated with abiotic stresses would produce a decrease in demand relative to a constant N supply. This may have the same effect as increasing the external N concentration (Mariotti et al., 1982), leading to greater isotopic discrimination (Vitousek et al., 1989; Handley et al., 1997). The higher δ15N of shoots compared with roots under control conditions could be related to fractionation processes during assimilation by NR, with δ15N of unassimilated nitrate becoming enriched relative to organic N (Evans, 2001; Tcherkez, 2011). Thus, NO3 that was not reduced in roots and exported to the shoots would be enriched in 15N, allowing for an increased δ15N of shoots relative to roots (Yoneyama & Kaneko, 1989; Evans et al., 1996). However, stress conditions would limit NO3 export from the roots to the shoot (Kronzucker et al., 1998), therefore decreasing δ15N of shoots while increasing it in the roots compared with control. This would explain the opposite δ15N response patterns to stress in shoots (15N depletion relative to full irrigation) compared with roots (15N enrichment). Water stress would limit the uptake and further transfer of N to the upper plant parts, subsequently increasing δ15N. Hence, when N availability is limited, a rapid decrease in N translocation from the root to the shoot has been reported (Kronzucker et al., 1998). Moreover, tolerant genotypes tended to exhibit constitutively higher δ15N (Table S1), which agrees with previous studies on durum wheat (Yousfi et al., 2009).

Gas-exchange and N metabolism traits involved in genotypic performance

While changes in ion concentrations and ratios perfectly separated treatments (Fig. 1), they were unsuited for assessing genotypic tolerance to salinity and water stress (Tables 3, S3). By contrast, stable isotope signatures related to plant photosynthesis (δ13C) and transpiration (δ18O), and traits related to N metabolism (N, NR, GS and δ15N), were more appropriate for genotype differentiation. As these traits were almost the only ones selected in the stepwise models (Table 4) that explained genotypic differences in biomass, they were used to further interpret the physiological mechanisms underlying distinct genotypic performances under each growing condition.

Nitrogen uptake and assimilation is a complex event that depends on many factors, such as the coordination of nitrogen with carbon, energy, and other types of metabolism during plant development (Masclaux-Daubresse et al., 2010; Rana et al., 2010). N assimilation requires NADH for NR-driven and ATP for GS-driven reactions, as well as carbon skeletons derived from photosynthesis for the synthesis of amino acids. In this respect, we observed a decrease in NR and GS activities associated with an increase in δ13C (Fig. 5), which in turn can be interpreted as the result of restricted photosynthetic activity mediated by lower gs. Moreover, δ15N is also affected by changes in photosynthetic activity driven by stress conditions (Lopes et al., 2004; Lopes & Araus, 2006). Thus, NR and GS activities were positively associated with δ15N. It is well known that 14N/15N isotope fractionation occurs during nitrate and ammonium assimilation by plants (Tcherkez & Farquhar, 2006).

Recently, Tcherkez (2011) proposed a model explaining δ15N variability in leaves in which a high sensitivity of δ15N values to both photorespiratory and N input (e.g. reduction by NR) is demonstrated. Overall, this model shows a clear link between photosynthesis and N metabolism (affecting energy balance of key enzymes or through photorespiration fluxes) and might explain why δ15N and δ13C are covariates across growing conditions in this study (= −0.69**). Alternatively Cernusak et al. (2009b) hypothesized that whole-plant δ15N should vary as a function of the transpiration efficiency of nitrogen acquisition, both factors being positively correlated.

Stress conditions, particularly saline treatments, were shown to cause a much stronger reduction in NR than in GS activity. An inhibition of nitrate uptake rates exceeding 50% has been observed under 60 mM NaCl in wheat (Botella et al., 1997), whereas ammonium uptake seems much less affected (Ullrich, 2002). Furthermore, wheat plants have shown a preferential ammonium uptake in saline media (Botella et al., 1993, 1997). This would justify the lower reduction in GS than in NR activity in response to stress. Moreover, NR catalyses a high energy-consuming metabolic reaction and NR activity is reported as strongly affected by stress conditions influencing photosynthesis (Tcherkez, 2011). The path analysis showed that NR activity was negatively associated with δ13C, regardless of the growing conditions. Simple correlations also showed that biomass and N concentration were better associated with NR than with GS activity. However, the path analysis revealed a direct effect of δ13C on GS (Fig. 6) in the most severe treatment (DI-12), whereas the directional effect of NR on GS activity diminished as compared with other stress treatments. This realization suggests that if N uptake and assimilation are very low, N metabolism is basically driven by a high photorespiration. In fact, the photorespiratory flux of ammonium in C3 plants can be 10 times higher than that originating from nitrate reduction (Hirel et al., 2007).

The negative relationships of δ18O with NR and GS activities observed across treatments are consistent with a dependence of δ18O on gs values mediated by particular stress conditions. Nevertheless, the relationships of enzymatic activities to δ18O were far weaker than those observed for δ13C, which agrees with the fact that δ18O is not as directly associated with photosynthesis as δ13C. Even so, δ18O (as δ13C and δ15N) correlated better with NR activity than with GS activity, in accordance with a more direct link of NR to photosynthesis.

Combined isotope signatures track genotypic performance under stress

Genotypes with lower δ13C exhibited higher biomass under stress. However, the path analysis revealed that, except for the two most severe growing conditions (FI-17 and DI-12), δ13C did not have a direct effect on biomass, but was mediated through δ18O or N metabolism. Genotypes with lower δ18O were directly associated with higher biomass in the two moderate stress treatments (FI-12 and DI; Table S1), with δ18O being significantly related to gs (r = −0.66 P < 0.01) at least in DI, which agrees with previous results on durum wheat where δ18O reflected transpiration patterns (Cabrera-Bosquet et al., 2009a, 2011). Results also showed that stress affected N metabolism in all growing conditions and, except in the most severe treatment (DI-17), it translated into genotypic differences in biomass, with a higher shoot δ15N being a favourable trait. Besides that, because progressive water stress decreases tissue expansion before closure of stomata (Hsiao, 1973), any water stress increasing δ18O would have already affected tissue expansion, and thus biomass. Tardieu et al. (2011) concluded that water deficit affects plant growth via reduced tissue expansion, carbon photosynthesis and cell number, the first process being the most crucial. However, our study highlights that N metabolism also plays a key role in genotypic performance under salinity and water stress.

Implications for breeding

This study examines the potential usefulness of C, O and N stable isotopes in shoots, alone or in combination, to assess genotypic performance of durum wheat to salinity and/or water deficit. Multiple isotope measurements may provide additional discrimination among phenotypes. Indeed, while selection of low-δ13C genotypes arises as a positive choice in any of the stress conditions assayed, a low genotypic δ18O, along with a high δ15N, may represent a sensible combination to aid identifying genotypes better adapted to moderate stresses. The interest of selecting tolerant genotypes at early stages of the crop cycle stems from the possibility of scheduling pollinations at anthesis and saving resources at harvest. This is particularly evident for salinity, which, in contrast to water stress under Mediterranean conditions, may already be present at planting, thus affecting crop establishment and early vigour, and therefore limiting grain yield early in the crop cycle. Moreover, it may be worth selecting for a constitutively low δ13C (together with lower δ18O and higher δ15N) when screening for stress tolerance under near-optimal conditions, since it may obviate the need to manage salinity stress, which is a very difficult and costly task in field phenotyping. This study also highlights the importance of N metabolism, and particularly of NR and GS activities, as key components of genotypic performance under water and/or salinity stress. However, selection based on enzymatic activities may not be feasible for large-scale field phenotyping, whereas it is perhaps realistic using high-throughput phenomic platforms. Further studies would be valuable to assess the performance of these stable isotopes for discerning genotypic variability in grain yield, preferably under field conditions.

Acknowledgements

We acknowledge support of OPTIWHEAT (INCO-STREP 015460), AGL2010-20180, and Junta de Andalucía-FEDER-FSE (P07-CVI-3026, P10-CVI-6368 and BIO-163) projects and the help of Marco Betti.

Ancillary