Theoretical and Applied Genetics

, Volume 125, Issue 1, pp 71–90

Quantitative trait loci for water-use efficiency in barley (Hordeum vulgare L.) measured by carbon isotope discrimination under rain-fed conditions on the Canadian Prairies

Authors

  • Jing Chen
    • Department of Renewable Resources, 442 Earth Sciences BuildingUniversity of Alberta
    • Department of Landscape Studies, College of Architecture and Urban PlanningTongji University
    • Department of Renewable Resources, 442 Earth Sciences BuildingUniversity of Alberta
    • Alberta Innovates-Technology Futures
Original Paper

DOI: 10.1007/s00122-012-1817-7

Cite this article as:
Chen, J., Chang, S.X. & Anyia, A.O. Theor Appl Genet (2012) 125: 71. doi:10.1007/s00122-012-1817-7

Abstract

Barley (Hordeum vulgare L.) yield is commonly limited by low rainfall and high temperature during the growing season on the Canadian Prairies. Empirical knowledge suggests that carbon isotope discrimination (Δ13C), through its negative relationship with water-use efficiency (WUE), is a good index for selecting stable yielding crops in some rain-fed environments. Identification of quantitative trait loci (QTL) and linked markers for Δ13C will enhance its use efficiency in breeding programs. In the present study, two barley populations (W89001002003 × I60049 or W × I, six-row type, and Merit × H93174006 or M × H, two-row type), containing 200 and 127 recombinant inbred lines (RILs), were phenotyped for leaf Δ13C and agronomic traits under rain-fed environments in Alberta, Canada. A transgressive segregation pattern for leaf Δ13C was observed among RILs. The broad-sense heritability (H2) of leaf Δ13C was 0.8, and there was no significant interaction between genotype and environment for leaf Δ13C in the W × I RILs. A total of 12 QTL for leaf Δ13C were detected in the W × I RILs and 5 QTL in the M × H RILs. For the W × I RILs, a major QTL located on chromosome 3H near marker Bmag606 (9.3, 9.4 and 10.7 cM interval) was identified. This major QTL overlapped with several agronomic traits, with W89001002003 alleles favoring lower leaf Δ13C, increased plant height, and reduced leaf area index, grain yield, harvest index and days to maturity at this locus or loci. This major QTL and its associated marker, when validated, maybe useful in breeding programs aimed at improving WUE and yield stability of barley on the Canadian Prairies.

Introduction

Drought continues to be a major constraint on the productivity of cereal crops and water deficit will increase in most arid and semi-arid regions under future climate-change scenarios (IPCC 2007; Wassmann et al. 2009). Agricultural production in Canada is mostly concentrated on the Prairies of western Canada (Canadian International Grains Institute (CIGI) 2004). The production of barley, one of the main crops grown in Canada, is centered in the three Prairie Provinces (Alberta, Saskatchewan and Manitoba), with an average of 12.3 million tonnes produced annually during 1986–2006 (FAOSTAT 2008). Both two- and six-row ear types of barley are commonly grown under rain-fed conditions for malting, livestock feed and food. During the short and dry growing seasons, sometimes with terminal heat stress (Anyia et al. 2008), barley relies on stored soil moisture and limited rainfall within the growing season. The unpredictable occurrence of drought can cause highly unstable barley yield across years. Therefore, breeding for drought-tolerant and water-use efficient varieties has been a critical area of agricultural research in Canada.

Water-use efficiency (WUE) has been proposed as a criterion for yield improvement under drought (Condon and Richards 1992; Condon et al. 2002; Rebetzke et al. 2002; Richards et al. 2002), and improved WUE can increase yield in certain environments. However, the application of WUE in breeding programs has been largely limited by the time-consuming and expensive screening process under field conditions for large populations. Carbon isotope discrimination (Δ13C), through its negative relationship with transpiration efficiency has been demonstrated to be a simple but reliable measure of WUE (Farquhar et al. 1982; Farquhar and Richards 1984), and their negative correlation has been used for indirect selection of WUE under selected environments (Cattivelli et al. 2008).

Selection efficiency of leaf Δ13C could be enhanced with a better understanding of its genetic control. The advancement of DNA-based molecular markers and computational methods in the late 1980s and 1990s has greatly revolutionized the dissection of quantitative trait inheritance and genetic improvement of yield in dry environments (Baum et al. 2007). The first quantitative trait loci (QTL) identified for Δ13C was reported in tomato (Lycopersicon esculentum and L. pennellii) by Martin and Nienhuis (1989) and subsequently QTL for Δ13C have been reported in Arabidopsis thaliana (Hausmann et al. 2005; Juenger et al. 2005), barley (grain) (Teulat et al. 2002), cotton (Gossypium hirsutum and G. barbadense) (Saranga et al. 2001), rice (Oryza sativa L.) (Laza et al. 2006; Takai et al. 2006; This et al. 2010; Xu et al. 2009), soybean (Glycine max L.) (Specht et al. 2001) and bread wheat (Triticum aestivum L.) (Rebetzke et al. 2008). Ellis et al. (1997) discovered several AFLP (amplified fragment length polymorphisms) markers associated with whole-shoot Δ13C of barley measured on 57 doubled haploids (DHs) from a cross between H. vulgare L. cv Lina and H. vulgare ssp. spontaneum HS92 in a hydroponic system under controlled and salt-stress conditions. The first QTL study for grain Δ13C in barley was by Teulat et al. (2002) using 167 recombinant inbred lines (RILs) derived from two-row barley cultivars Tadmor and Er/Apm in three Mediterranean field environments, with a total of ten QTL identified in the study.

Plant parts sampled for Δ13C analysis can be awn, grain, leaf, sheath, stem or root, each with its own characteristic Δ13C value and potential advantages for evaluating WUE under different environments. Leaf Δ13C during the stem elongation stage, which is regarded as the formation of yield potential (Anyia et al. 2008), could reflect the integrated WUE and the vegetative establishment. Condon and Richards (1992) proposed that it would be most effective to assess Δ13C at early stages in plant development under well-watered conditions. In addition, leaf Δ13C measured before anthesis can enable selection and crosses of varieties to be made within the same season, thereby speeding up the breeding process. In our study penultimate leaf was sampled for Δ13C analysis to avoid damaging of the flag leaf (Sicher 1993; Thorne 1965; Xue et al. 2008). According to Jiang et al. (2006), values of Δ13C from different plant parts (flag leaf, awn and grain) were on average higher for six-row barley than those of two-row barley, suggesting a higher WUE of two-row barley compared with six-row barley under both irrigated and non-irrigated field conditions. So far, there has been no report on QTL analysis of Δ13C on barley leaves or on two- and six-row barley examined under the same rain-fed environments.

In this study, two- and six-row barley RIL populations were grown under rain-fed conditions, and the penultimate leaves at stem elongation stage were sampled for Δ13C analysis and QTL mapping. The objectives of the present study were to: (1) to determine the chromosomal regions and phenotypic effects of QTL associated with variations in leaf Δ13C as well as agronomic traits; (2) to identify the common QTL regions across environments and populations; and (3) to dissect the genetic control of leaf Δ13C and potential alleles for further WUE improvement in barley.

Materials and methods

Field experiments and climatic conditions

The experiments were conducted at Lacombe (52°28′N, 113°45′W, 847.3 m altitude), having an Orthic Black Chernozemic soil (Canadian system of soil classification), Vegreville (53°31′N, 112°6′W, 639.3 m altitude), having the Malmo series of an Eluviated Black Chernozemic soil, and Castor (52°8′N, 111°54′W, 807.7 m altitude), with a Dark Brown Chernozemic soil, in Alberta, Canada, under rain-fed conditions. The three sites were considered three different environments characterized by distinct soil moisture conditions with Castor as the driest site and Lacombe as the wettest site. The average annual precipitation and within-season rainfall (June–August) from 1977 to 2007 was 340 ± 89 and 172 ± 67 mm at Castor compared with Vegreville which had 382 ± 62 and 193 ± 52 mm, and Lacombe which had 440 ± 84 and 230 ± 63 mm, respectively (AgroClimatic Information Service (ACIS) 2009; Environment Canada 2009).

Mapping populations

One hundred and six diverse genotypes (advanced lines and commercial varieties) of spring barley were screened for variation in leaf Δ13C at two field locations in Vegreville and Lacombe in 2005. Among these genotypes, two-row genotypes (Merit, released by Busch Agricultural Resources Incorporation in 1997 and H93174006, developed at the Field Crop Development Centre, Lacombe, Alberta, and released in 2008 as TR05671) and six-row barley (W89001002003, hereafter referred to as W89 and I60049) showing contrasting levels (low vs. high) of leaf Δ13C were selected as parental materials for mapping population development after additional greenhouse and field experiments confirmed their leaf Δ13C rank stability (Fig. 1). Two RIL mapping populations were developed, with one cross between W89 and I60049 (hereafter referred to as W × I) and the other cross between Merit and H93174006 (hereafter referred to as M × H) in the field during the summer of 2006. The parents with lowleaf-Δ13C (Merit and W89) were used as females since previous experience indicated that maternal inheritance may have an influence on Δ13C and WUE traits (Dr. Richard Richards, CSIRO, personal communication). All F1 seeds from each cross were planted in bulk in the field (F2 populations) at Lacombe in 2007 and evaluated for segregation. The single seed descent (SSD) approach was used to advance the populations from F3 to F4 generation in the greenhouse at Lacombe during 2007 and 2008. The F5 seeds from each F4 plant were bulked and advanced to produce enough seeds for F5:6 generation field trials.
https://static-content.springer.com/image/art%3A10.1007%2Fs00122-012-1817-7/MediaObjects/122_2012_1817_Fig1_HTML.gif
Fig. 1

Leaf carbon isotope discrimination (Δ13C) of parental lines W89001002003 and I60049 (top), and Merit and H93174006 (bottom) across locations and years that were characterized by different total growing season rainfall. Significant differences of leaf Δ13C between parental lines are indicated by *p < 0.05, **p < 0.01 and ***p < 0.001. L05, L08, C07, C09, V05, V06, V07, V08 and V09 stand for location by year, for Lacombe-2005, Lacombe-2008, Castor-2007, Castor-2009, Vegreville-2005, Vegreville-2006, Vegreville-2007, Vegreville-2008 and Vegreville-2009, respectively

Experiment design

The W × I mapping population

Two hundred F5 RILs and two parental lines (W89 and I60049) were planted at Lacombe in 2008 (L08), Castor in 2009 (C09) and Vegreville in 2009 (V09).

For L08, the experiment used a completely randomized design with one replicate of RILs and four replicates of parental lines. Seeds of each line were planted in two rows in 0.3 × 2 m plots with a spacing of 15 cm within rows and 35 cm between plots. Winter triticale ‘Pika’ was seeded as guard row between plots. Synthetic fertilizer mix (6-25-30) was applied at 112.5 kg ha−1 prior to seeding. Plots were seeded on May 15 (Table 1), and harvested on September 5. Herbicides were applied at the rate of 29.7 g ha−1 for ‘Refine’ (33.35% thifensulfuron methyl, 16.65% tribenuron methyl) and 988.4 mL ha−1 for ‘MCPA’ (500 g L−1 amine, 500 g L−1 ester, 400 g L−1 K-salt and 300 g L−1 Na-salt) during the vegetative stage to control weed species.
Table 1

Monthly precipitation (mm) over the growing season for the three field locations during 2008 and 2009

Location, year

Sowing date

Precipitation (mm)

June

July

August

Total

Lacombe, 2008

May 15

45.8

48.8

55.5

150.1

Vegreville, 2008

May 16

42.4

44.6

64.3

151.3

Vegreville, 2009

May 22

32.2

44.6

25.2

102.0

Castor, 2009

May 13

21.5

33.8

51.9

107.2

For C09 and V09, the experiments used a randomized complete block design with three replicates of each F5:6 RILs and parental lines. Seeds of each line were planted in four rows in 1 × 2 m plots with a spacing of 20 cm within rows and 20 cm between plots. Spring wheat ‘AC Crystal’ was seeded as a guard row between plots. Fertilizer was applied prior to seeding at a rate of 22.5 kg ha−1 P2O5 at Vegreville, and 31.4 kg ha−1 N and 18 kg ha−1 P2O5 at Castor based on results of soil test. Seeds were sown on May 13 at Castor and on May 22 at Vegreville, and both sites were harvested during the last week of September. Weeds were controlled manually at the 3-leaf stage (BBCH 13) (Zadoks et al. 1974) at Castor. For Vegreville, ‘Round-up Weathermax’ (540 g L−1 glyphosate) was sprayed at 1.66 L ha−1 before emergence on May 24.

The M × H mapping population

One hundred and twenty-seven F5 RILs and two parental lines (Merit and H93174006) were planted at Vegreville in 2008 (V08), as well as C09, and V09.

For V08, the experiment was a completely randomized design with one replicate of RILs and four replicates of parental lines. 200 seeds of each line were sown in 2 rows of a 2-m-long plot with a spacing of 20 cm between rows and 60 cm between plots. Fertilizer was applied at seeding by adding 38.25 kg P2O5 ha−1, and 61.88 kg N ha−1. Seeds were sown on May 16 and plants were harvested on September 18. Herbicide ‘Round-up Weathermax’ was sprayed at 1.66 L ha−1 before seeding, and ‘Achieve 40DG’ and ‘Buctril M’ were applied on June 4 at 494.2 g ha−1 and 988.4 mL ha−1, respectively.

For C09 and V09, the experiment design was the same as for the W × I population described above. Seeds were sown on May 15 at Castor and on May 25 at Vegreville, and both sites were harvested during the last week of September.

Phenotypic data collection

Carbon isotope discrimination analysis

At the stem elongation stage (BBCH 36–39), five fully expanded penultimate leaves per plot of each mapping population at each environment (L08, V08, C09 and V09) were randomly collected, bulked, and dried in an oven at 70°C for 48 h. Dried leaf samples were ground with a ball mill (Spex SamplePrep 8000D Mixer, Metuchen, NJ, USA) to fine powder and analyzed for carbon isotope composition (δ13C) using a continuous-flow stable isotope ratio mass spectrometer (Thermo Finnigan Mat Gmbh, Bremen, Germany). The δ13C was calculated as:
$$ \delta^{13} {\text{C}}\;\left( \permille \right) = \left( {R_{\text{sample}} /R_{\text{reference}} - 1} \right) \times 1,000 $$
where Rsample and Rstandard are the ratios of 13C/12C measured in the plant material and the reference, respectively, and the reference material used was the belemnite carbonate standard (PDB) from the Pee Dee Formation. The Δ13C was calculated according to Farquhar et al. (1989) as:
$$ \Updelta^{13} {\text{C}}\left( \permille \right) = \left[ {\left( {\delta^{13} {\text{C}}_{\text{a}} - \delta^{13} {\text{C}}_{\text{p}} } \right)/\left( {1 + \delta^{13} {\text{C}}_{\text{p}} /1,000} \right)} \right] \times 1,000 $$
where δ13Ca and δ13Cp refer to the C isotope ratios of atmospheric CO2 (−8.0 ‰) and plant, respectively.

Leaf area index (LAI)

LAI was measured for the W × I population in the 2009 field trials using a LAI-2000 Plant Canopy Analyzer (Li-Cor, Lincoln, NE, USA). A measurement cycle consisted of four below-canopy readings and one reference measurement. Reference measurements were collected above canopy level of each experimental plot (between 1 and 1.5 m above ground) at the beginning of each cycle. The below-canopy measurements were carried out at a diagonal transect between rows in each plot to improve the spatial average. The fish-eye lens of the instrument was covered with a view cap with a 45° opening, so that the reference measurements were not influenced by the operator and the angle of the sun (Li-Cor 1992). Measurements were taken at the stem elongation stage (BBCH 36–39) and grain filling stage (BBCH 71–75) under cloudy conditions.

Plant height, days to maturity, aerial biomass and grain yield

Plant height was measured from the ground level to the tip of the spike on five randomly selected plants per plot at the physiological maturity stage (BBCH 89) in the 2009 field trials for the W × I population. Days to maturity was recorded as the number of days from sowing till the stage that 90% of plants in the plot reached maturity (BBCH 91–92), this trait was only recorded at V09 for the W × I population. Biomass and grain yield were determined from 1 m2 sub-plots. Grain was threshed, cleaned and air-dried for 2 weeks before weighing, with all grain yields expressed as oven-dried values, after the moisture content of the air-dried grains was determined. Harvest index (HI) was calculated as HI = grain yield/total aboveground biomass.

Molecular marker assay

Assay for the W × I population

Parents (W89 and I60049) were screened for polymorphism with a total of 516 simple sequence repeat (SSR) markers from the published literature (Chaabane et al. 2009; Karakousis et al. 2003; Li et al. 2003; Liu et al. 1996; Ramsay et al. 2000; Thiel et al. 2003; Varshney et al. 2007). Polymorphic SSR markers were chosen for scoring the entire W × I population.

Young leaf tissues from each plot in the field trials were collected for DNA extraction. Genomic DNA was isolated as described by Saghai-Maroof et al. (1984), and quantified using a spectrophotometer (NanoDrop ND-1000 Spectrophotometer, Wilmington, Delaware USA), and qualified by electrophoresis. Regular PCR (polymerase chain reaction) was carried out in 20-μL reactions containing 5× PCR buffer, 1.5 mM MgCl2, 0.2 mM dNTPs, 0.2 μM forward prime, 0.2 μM reverse primer, 1 U of GoTaq Flexi DNA polymerase (Promega Corporation, Wisconsin, USA), deionized water, and 40 ng genomic DNA in 96-well plates using a Eppendorf Mastercyler (Hamburg, Germany). The SSRs amplification was performed using three PCR conditions: (1) a “touchdown” PCR consisting of 18 cycles of 1 min denaturing at 94°C, 30 s annealing at 62°C or 69°C and 1 min extension at 72°C. Annealing temperatures were decreased by 0.5°C per cycle, depending on the melting temperature (Tm) of primer pairs used, either from 62 to 53°C or from 69 to 60°C. The reaction continued for 30 additional cycles with 1 min at 94°C, 1 min at 53 or 60°C, 1 min at 72°C, and eventually 5 min at 72°C as the final extension step; (2) a regular PCR was performed with 2 min at 94°C, followed by 35 cycles at 94°C for 30 s, 30 s at optimal annealing temperature (ranging from 53 to 58°C), 1 min at 72°C, and ended with 5 min at 72°C, and (3) condition was the same as those described by Röder et al. (1995), 35 cycles were performed with 1 min at 96°C, 1 min at 60°C (varies with the Tm of primers) and 2 min at 72°C, and a final elongation step of 10 min at 72°C, in some cases the MgCl2 concentration was increased to 2.5 mM.

Primers with similar annealing temperatures and non-overlapping ranges of amplified fragment sizes were combined in the same reaction. Amplified PCR products were separated on 2.5% agarose gels in 1× TAE buffer, stained with ethidium bromide and photographed. The entire population was genotyped for SSR markers using a multicapillary (12-channel) electrophoresis system (HAD-GT12, eGene, Irvine, CA, USA).

Assay for the M × H population

A total of 373 SSR and 72 STS (sequence-tagged site) markers were screened for polymorphism between parental lines (Merit and H93174006). To get additional genomic coverage, extracted DNA of the RILs and parental lines were sent to Triticarte Pty Ltd, Canberra, Australia (http://www.triticarte.com.au/) for Diversity Arrays Technology (DArT®) genotyping, which is a novel genotyping method and hybridization-based technology as described by Wenzl et al. (2004).

QTL analysis and map construction

The segregating patterns of SSR amplifications at polymorphic loci in all lines were compared to parental lines, and scored as maternal (1), paternal (2) or heterozygous/missing (3) of an allele. Segregation markers in the RILs were examined by the Chi-square (χ2) test for goodness of fit. The expected ratio of individuals in each genotypic class (maternal or paternal) was 1:1 in the RILs. Mapmaker v3.0 (Lander et al. 1987) was used to assign markers to linkage groups at a LOD (logarithm of odds ratio) threshold of 3.0 with a maximum Kosambi distance of 30 centiMorgans (cM), and markers were assembled to initial linkage groups, then the LOD score was reduced to bridge some intervals. The position and order of markers in linkage groups were compared with published barley consensus map (Varshney et al. 2007; Wenzl et al. 2006). Genetic linkage maps were imported into Windows QTL Cartographer V2.5 (Wang et al. 2010) to locate putative QTL for all traits using composite interval mapping method (CIM) (Zeng 1994). A threshold score for the QTL was determined to be 2.0 by 1,000 time permutations using the Zmapqtl program (Churchill and Doerge 1994). A confidence interval for QTL location was determined by a one-LOD drop from the peak position (Flint-Garcia et al. 2003), and overlapping confidence intervals were used to determine common QTL across environments. The final linkage maps were drawn using the MapChart software (Voorrips 2002).

Data analysis

Data were analyzed using version 9.1 of the SAS software (SAS Institute, Inc., Cary, NC). Each location–year combination of the field trials (L08, V08, C09, and V09) was treated as a single environment. Analysis of variance (ANOVA) was performed for each environment as well as combined environments. Variance components were estimated using the restricted maximum likelihood (REML) method of PROC MIXED.

According to Hanson et al. (1956), the broad-sense heritability (H2) for leaf Δ13C across environments was estimated as:
$$ H^{2} = \sigma_{\text{g}}^{2} /\left[ {\sigma_{\text{g}}^{2} + \sigma_{{{\text{g}} \times {\text{e}}}}^{2} /e + \sigma_{\text{r}}^{2} /\left( {r \times e} \right)} \right] $$
where variances were due to genotype (σg2), genotype × environment (σg×e2), and residual (σr2), e is the number of environments, and r is the number of replicates.

Homogeneity of variance and normality of distribution were tested before analysis of variance. Correlation analyses (PROC CORR) were performed to detect the relationships among traits, with means of each RIL combined over the environments. Pearson correlation coefficients were calculated between environments for leaf Δ13C, with means of each RIL based on each environment. An α value of 0.05 was chosen to indicate statistical significance.

Results

Environments and leaf Δ13C

Across eight different environments (location–year combinations) in this study, the leaf Δ13C of W89 was consistently lower than that of I60049, except at Castor-2007 (Fig. 1). Although the leaf Δ13C of Merit was only significantly lower than that of H93174006 at three out of seven environments, trends of difference were still observed across environments (Fig. 1). The correlation analysis showed that leaf Δ13C was significantly related to rainfall in June (data not shown, r = 0.48, n = 122, p < 0.001) and total within-season precipitation (from June to August) (data not shown, r = 0.42, n = 122, p < 0.001). Significant differences were observed for leaf Δ13C between environments for both populations (Tables 2, 3).
Table 2

Means, standard deviations (SD), range (minimum and maximum values) of the 200 recombinant inbred lines (RILs) and their parental lines from the six-row (W89001002003 × I60049) barley population for traits measured in three field trials

Trait

Environment

Parents

RILs

W89001002003

I60049

Mean

SD

Mean

SD

Mean

SD

Range

E

G

G × E

H

Leaf Δ13C (‰)

L08

20.71

0.23

22.40

0.48

21.35

0.61

19.84–23.22

***

***

ns

0.80

C09

17.69

0.98

18.99

0.26

18.53

0.61

16.48–20.30

V09

18.72

0.56

19.97

0.95

19.71

0.65

17.94–22.19

LAI-S (m2 m−2)

C09

1.39

0.30

1.37

0.55

1.37

0.46

0.40–2.84

***

*

*

0.61

V09

4.52

0.79

4.60

1.26

3.96

0.96

1.95–6.96

LAI-G (m2 m−2)

C09

1.34

0.19

1.71

0.14

1.63

0.33

0.82–2.74

***

ns

ns

0.63

V09

4.54

0.59

4.89

1.02

4.34

0.89

2.30–7.04

Plant height (cm)

C09

38

2.1

47

2.6

47

7.0

29–69

***

***

***

0.82

V09

67

7.2

73

6.5

76

13.0

42–105

Days to maturity

V09

88

2.7

91

1.0

87

3.7

73–92

***

Biomass (g m−2)

L08

1,512

100.5

1,229

60.4

1,650

354.5

1,072–3,243

***

***

***

0.75

C09

236

22.9

265

43.3

261

56.1

143–520

V09

1,038

120.3

987

190.6

828

195.1

410–1518

Grain yield (g m−2)

L08

701

39.6

513

36.0

708

153.7

412–408

***

***

***

0.74

C09

113

21.6

128

6.1

129

35.2

60–294

V09

529

69.5

521

90.7

424

114.0

180–834

HI

L08

0.46

0.01

0.42

0.02

0.43

0.02

0.35–0.49

***

**

ns

0.75

C09

0.48

0.04

0.49

0.06

0.49

0.06

0.30–0.66

V09

0.51

0.02

0.53

0.01

0.51

0.04

0.35–0.60

E environment effect, G genotype effect, G × E interaction between environment and genotype, H2 broad-sense heritability, L08 Lacombe-2008, C09 Castor-2009, V09 Vegreville-2009, LAI-S leaf area index measured at stem elongation stage, LAI-G leaf area index measured at grain filling stage, – not available, ns not significant

 *, **, *** indicate 5, 1, 0.1% significance level, respectively

Table 3

Means, standard deviations (SD), range (minimum and maximum values) of the 127 recombinant inbred lines (RILs) and their parental lines from the two-row (Merit × H93174006) barley population for traits measured in three field trials

Trait

Environment

Parents

RILs

Merit

H93174006

Mean

SD

Mean

SD

Mean

SD

Range

E

G

G × E

H2

Leaf Δ13C (‰)

V08

18.62

0.35

18.73

0.46

18.93

0.63

17.53–20.53

***

***

*

0.66

V09

17.80

0.50

18.67

0.70

18.46

0.52

17.19–20.33

Biomass (g m−2)

V08

362

16.6

401

36.2

495

131.6

292–896

***

ns

ns

0.69

C09

366

100.7

170

9.1

273

80.2

92–595

V09

572

117.5

662

38.2

665

132.4

305–1,155

Grain yield (g m−2)

V08

99

11.6

104

7.4

137

43.9

65–305

***

ns

ns

0.63

C09

194

59.0

68

6.5

127

47.6

51–307

V09

282

62.8

331

32.8

324

63.0

151–566

HI

V08

0.28

0.02

0.26

0.01

0.27

0.03

0.19–0.35

***

ns

*

0.66

C09

0.53

0.04

0.40

0.04

0.46

0.07

0.23–0.59

V09

0.49

0.01

0.50

0.03

0.49

0.03

0.34–0.66

E environment effect, G genotype effect, G × E interaction between environment and genotype, V08 Vegreville-2008, C09 Castor-2009, V09 Vegreville-2009, ns not significant

 *, **, *** indicate 5, 1, 0.1% significance level, respectively

For the W × I population, leaf Δ13C of RILs were significantly higher in L08, followed by V09 and then C09 (Table 2). For the M × H population, leaf Δ13C for the experiments conducted in Vegreville was significantly higher in 2008 than in 2009. Positive and significant correlations of leaf Δ13C were observed between C09 and V09 (Fig. 2a, r = 0.32, n = 200, p < 0.001), C09 and L08 (Fig. 2a, r = 0.32, n = 200, p < 0.001), L08 and V09 (Fig. 2a, r = 0.21, n = 200, p = 0.002). The extreme genotypes differed by 3.38‰ at L08, 3.82‰ at C09 and 4.25‰ at V09. For the M × H population, a positive correlation of leaf Δ13C was observed between V08 and V09 (Fig. 2b, r = 0.16, n = 127, p = 0.08). The extreme genotypes differed by 3‰ at V08 and 3.14‰ at V09.
https://static-content.springer.com/image/art%3A10.1007%2Fs00122-012-1817-7/MediaObjects/122_2012_1817_Fig2_HTML.gif
Fig. 2

a Correlation between leaf carbon isotope discrimination (Δ13C) of 200 recombinant inbred lines (RILs) from the six-row (W89001002003 × I60049) population measured at Lacombe in 2008, Castor in 2009 and Vegreville in 2009. b Correlation between leaf carbon isotope discrimination (Δ13C) of 127 recombinant inbred lines (RILs) from the two-row (Merit × H93174006) population measured at Vegreville in 2008 and 2009

Leaf Δ13C and agronomic traits

The two populations differed in leaf Δ13C, with the M × H population producing a significantly lower leaf Δ13C than the W × I population at V09 (data not shown, p < 0.001). Leaf Δ13C of progeny within each population were significantly different among the environments studied (Tables 2, 3). The Shapiro–Wilk test showed that leaf Δ13C of both populations had typically normal distributions under all environments (Fig. 3). Most of the leaf Δ13C values for the RILs were between those of the parents in each field trial, but transgressive segregation was also noted as some progeny extremes for leaf Δ13C exceeded parental values. Some of the extreme RILs that showed transgressive segregation for leaf Δ13C from the W × I population (such as ‘176’ and ‘191’) and the M × H population (such as ‘43’, ‘110’ and ‘162’) were consistent across all locations. The H2 for leaf Δ13C was high (0.80) for the W × I population across all environments. Under a single environment (C09 or V09), the H2 for both populations reached 0.80 (data not shown).
https://static-content.springer.com/image/art%3A10.1007%2Fs00122-012-1817-7/MediaObjects/122_2012_1817_Fig3_HTML.gif
Fig. 3

Frequency distribution of mean leaf carbon isotope discrimination (Δ13C) measured on the recombinant inbred lines (RILs) from the two-row (Merit × H93174006) and the six-row (W89001002003 × I60049) mapping populations grown at Lacombe and Vegreville in 2008, and Castor and Vegreville in 2009. Parental means are indicated for each environment

Biomass, yield, HI, plant height, LAI, and days to maturity all showed continuous variation (Tables 2, 3). For the W × I population, ANOVA revealed significant variability among the 200 RILs for a range of agronomic traits (Table 2) except LAI measured at grain filling stage (LAI-G). Interactions between genotype and environment (G × E) were not significant for leaf Δ13C, LAI-G, and HI. For the M × H population, genotypic variation was not significant for biomass, yield and HI (Table 3), and G × E was only significant for leaf Δ13C.

For the W × I population, leaf Δ13C showed significant positive relationships with LAI at stem elongation stage (LAI-S), biomass, yield and HI at both C09 and V09 (Table 4). For the M × H population, there was also significant positive correlation between Δ13C and biomass at V08 and V09 (Table 5), and significant positive relationship between Δ13C and grain yield was only observed at V08.
Table 4

Correlations among leaf carbon isotope discrimination (Δ13C), leaf area index at stem elongation stage (LAI-S), leaf area index at grain filling stage (LAI-G), plant height (PH), biomass, yield, harvest index (HI) and days to maturity at Castor and Vegreville in 2009 for the six-row (W89001002003 × I60049) barley population

 

Leaf Δ13C

LAI-S

LAI-G

PH

Biomass

Yield

HI

Castor-2009

 LAI-S

0.20**

      

 LAI-G

0.13ns

0.41***

     

 PH

–0.002ns

0.26***

0.30***

    

 Biomass

0.31***

0.45***

0.49***

0.29***

   

 Yield

0.28***

0.45***

0.40***

0.21**

0.78***

  

 HI

0.20**

0.24***

0.07ns

–0.12ns

0.28***

0.59***

 

Vegreville-2009

 LAI-S

0.28***

      

 LAI-G

0.35***

0.77***

     

 PH

0.11ns

0.40***

0.50***

    

 Biomass

0.44***

0.62***

0.63***

0.52***

   

 Yield

0.46***

0.56****

0.58***

0.38***

0.96***

  

 HI

0.30***

0.11ns

0.13ns

–0.26***

0.36***

0.58***

 

 Days to maturity

0.11ns

0.29***

0.35***

0.06ns

0.28***

0.29***

0.13ns

ns not significant, – no records for days to maturity for Castor-2009

 *, **, *** indicate 5, 1, 0.1% significance level, respectively

Table 5

Correlations among leaf carbon isotope discrimination (Δ13C), biomass, yield and harvest index (HI) at Vegreville in 2008 and 2009 for the two-row (Merit × H93174006) barley population

 

Leaf Δ13C

Biomass

Yield

2008

 Biomass

0.65***

  

 Yield

0.61***

0.95***

 

 HI

0.20*

0.31***

0.58***

2009

 Biomass

0.28**

  

 Yield

0.02ns

0.34***

 

 HI

−0.08ns

0.001ns

0.94***

ns not significant

 *, **, *** indicate 5, 1, 0.1% significance level, respectively

Linkage map construction

Screening of the parental lines with SSR markers revealed a low level of polymorphism between parental lines. For the W × I population, only 148 (28.7%) out of 516 screened markers could be used for scoring the 200 RILs. For the M × H population, 55 (12.9%) out of 373 SSR markers and 196 (39.3%) out of 499 DArT markers were polymorphic between Merit and H93174006. In total, 104 SSR markers were mapped for the W × I population, and 209 loci were mapped for the M × H population. In general, there was good agreement for the marker order between maps from this study and maps published in the literature.

For the W × I population, the 104 SSR loci were grouped into ten linkage groups which were assigned to seven chromosomes (Fig. 4a) based on alignments from previously published barley consensus maps (Karakousis et al. 2003; Varshney et al. 2007). Chromosome (Chr.) 1H, 2H, 3H, 4H, and 6H were each represented by a single linkage group, whereas Chr. 5H and 7H were split into two and three linkage groups, respectively.
https://static-content.springer.com/image/art%3A10.1007%2Fs00122-012-1817-7/MediaObjects/122_2012_1817_Fig4a_HTML.gif
https://static-content.springer.com/image/art%3A10.1007%2Fs00122-012-1817-7/MediaObjects/122_2012_1817_Fig4b_HTML.gif
Fig. 4

a A linkage map of barley based on 200 F5:6 recombinant inbred lines (RILs) from the cross between W89001002003 and I60049 constructed with 104 SSR markers. Numbers on the left of linkage groups indicate the cumulative map distances in cM (Kosambi). Marker loci are shown on the right of linkage groups. Co-segregating markers are listed next to each other in a vertical line on the right side of the linkage group. Markers with segregation distortion are indicated by *p < 0.05, **p < 0.01 and p < ***0.001. QTL confidence intervals (peak LOD scores minus one) are shown to the right of the linkage group bar with QTL for carbon isotope discrimination (ΔC) indicated by black rectangles and all others indicated by white rectangles as follows: leaf area index during stem elongation stage (LS), leaf area index during grain filling stage (LG), plant height (PH), days to maturity (DM), aboveground biomass (AB), grain yield (GY) and harvest index (HI). L08, C09 and V09 stand for location by year, for Lacombe-2008, Castor-2009 and Vegreville-2009, respectively. b A linkage map of barley based on 127 F5:6 recombinant inbred lines (RILs) from the cross between Merit and H93174006 constructed with 21 SSR and 156 DArT markers. Numbers on the left of linkage groups indicate the cumulative map distances in cM (Kosambi). Marker loci are shown to the right of linkage groups. Co-segregating markers are listed next to each other in a vertical line on the right side of the linkage group. Markers with segregation distortion are indicated by with *p < 0.05, **p < 0.01 and ***p < 0.001. QTL confidence intervals (peak LOD scores minus one) are shown to the right of the linkage group bar with QTL for carbon isotope discrimination (ΔC) indicated by black rectangles and all others indicated by white rectangles as follows: aboveground biomass (AB), grain yield (GY) and harvest index (HI). C09, V08 and V09 stand for location by year, for Castor-2009, Vegreville-2008 and Vegreville-2009, respectively

For the M × H population, the 209 loci, including 21 SSR and 188 DArT loci, were grouped into 17 linkage groups which were assigned to 7 chromosomes (Fig. 4b) according to previously published high-density consensus maps of barley (Varshney et al. 2007; Wenzl et al. 2006). Chr. 1H, 2H and 3H were split into three linkage groups, whereas Chr. 6H and 7H were split into four and two linkage groups, respectively, and Chr. 4H and 5H were each represented by a single linkage group.

QTL identified for leaf Δ13C

Table 6 describes the QTL identified for leaf Δ13C in each environment separately. For the W × I population, a total of 12 QTL clustering in nine chromosomal regions and explaining from 3.6 to 22% of the phenotypic variation were identified for leaf Δ13C, with 5 QTL detected at L08, 3 QTL at C09 and 4 QTL at V09 by the CIM analysis. Eight of these 12 QTL reduced leaf Δ13C and the W89 allele was associated with low leaf Δ13C. One QTL located on Chr. 3H near SSR marker Bmag606 (designated as ΔC3.1-WI) was detected consistently in all three environments. This QTL conferred main effect across all environments, with LOD values of 10.81, 5.86 and 5.27, explaining 22, 14.4 and 11% of the phenotypic variance in leaf Δ13C at L08, C09 and V09, respectively. At this locus, leaf Δ13C value was reduced by the W89 allele (Table 6).
Table 6

QTL identified for carbon isotope discrimination (Δ13C) with QTL Cartographer 2.5 by the composite interval mapping (CIM) method using data from three locations (Castor, Lacombe and Vegreville) for the six-row (W89001002003 × I60049) barley population and data from Vegreville for the two-row (Merit × H93174006) barley population in 2 years (2008–2009)

Population

Environment

Position (marker + distance to the maximum LOD value) (cM)

Chromosome

LOD

R2 (%)

Additive effect

Six-row

L08

Bmac32 + 2.04

1H

2.13

3.9

0.13

Bmag606 + 2.03

3H

10.81

22

−0.36

EBmac708 − 2.37

3H

3.78

7.3

−0.20

Bmag751 + 0.06

5H

4.77

7.3

−0.18

Bmag807 − 1.72

6H

2.18

3.6

0.12

C09

EBmac558 − 0.36

2H

2.34

4.1

−0.08

Bmag603 + 0.04

3H

2.44

4.1

−0.08

Bmag606 + 3.03

3H

5.86

14.4

−0.17

V09

Bmag606 − 0.87

3H

5.27

11

−0.18

Bmac113 + 0

5H

2.93

5.3

−0.11

GBM1022 + 0.04

6H

2.12

3.8

0.09

EBmac755 + 0

7H

2.38

4.3

0.09

Two-row

V08

bPb1212 + 0

6H

3.48

11.2

0.29

bPb7872 − 0.04

6H

3.65

11.7

0.32

bPb9898 + 0.98

7H

3.11

9.9

0.20

V09

bPb1212 + 0

6H

3.60

10.8

0.16

bPb9898 + 1.98

7H

2.54

8.6

0.10

R2 proportion of the phenotypic variance explained by each QTL, Additive effect additive effect of the allele from W89001002003 compared with I60049, L08 Lacombe-2008, C09 Castor-2009, V08 Vegreville-2008, V09 Vegreville-2009

For the M × H population, a total of 5 QTL clustering in three chromosomal regions and explaining from 8.6 to 11.7% of the phenotypic variation were detected by the CIM analysis for leaf Δ13C, with 3 QTL at V08 and 2 QTL at V09. Two QTL were detected consistently in two environments with one QTL located at Chr. 6H near DArT marker bPb1212 (designated as ΔC6.1-MH), explaining 11.2 and 10.8% of the phenotypic variance in leaf Δ13C at V08 and V09, respectively, and the other QTL on Chr. 7H close to DArT marker bPb9898 (designated as ΔC7.1-MH), explaining 9.9 and 8.6% of leaf Δ13C at V08 and V09, respectively. All of the five QTL increased leaf Δ13C value with allele from Merit (Table 6).

QTL identified for agronomic traits

As listed in Table 7; Fig. 4a, b, a total of 38 QTL were identified by CIM analysis for six agronomic traits under each environment.
Table 7

QTL identified for leaf area index measured during stem elongation stage (LAI-S), leaf area index measured during grain filling stage (LAI-G), plant height (PH), days to maturity, biomass, yield and harvest index (HI) with QTL Cartographer 2.5 by the composite interval mapping (CIM) method using data from two locations (Castor and Vegreville) for the six-row (W89001002003 × I60049) and the two-row (Merit × H93174006) barley populations in 2 years (2008–2009)

Population

Trait

E

Position (marker + distance to the maximum LOD value) (cM)

Chromosome

LOD

R2 (%)

Additive effect

Six-row

LAI-S

V09

GBM1405 + 2.04

3H

2.61

5.4

−0.15

Bmac156 + 0.01

7H

3.02

5.6

−0.14

LAI-G

V09

GBM1405 + 1.04

3H

2.41

4.4

−0.13

Bmag606 − 0.87

3H

2.11

4.2

−0.13

GBM5008 + 0

5H

3.83

6.9

0.17

GBMS139 + 2.02

7H

6.82

13.6

0.20

PH

C09

Bmac32 − 3.76

1H

7.53

15.0

−2.15

Bmac32 + 4.04

1H

7.96

14.5

−2.10

Bmag606 + 4.03

3H

5.73

13.6

2.48

EBmac708 − 5.37

3H

2.94

6.6

1.58

Scssr25538 − 10.05

3H

3.23

8.4

1.68

GBM1506 − 5.82

5H

5.18

10.7

−1.80

GBM1355 − 11.79

6H

2.09

5.8

−1.32

GBMS139 − 6.28

7H

2.98

6.9

1.42

V09

Bmag606 + 5.03

3H

11.94

27.4

6.62

GBM1506 − 3.82

5H

6.24

11.0

−3.42

GBMS139 − 2.28

7H

5.00

9.3

3.22

Biomass

C09

GBMS139 + 0.02

7H

2.73

5.6

8.90

V09

Bmag337 − 0.01

5H

3.76

6.3

−32.16

GBM1231 + 0.03

5H

3.55

6.3

34.08

Scssr5599 − 0.01

6H

2.21

3.7

24.23

Yield

C09

Bmag606 + 2.03

3H

4.42

11.4

−9.79

GBM5008 + 0

5H

2.06

4.1

5.04

V09

Bmag518 + 0.03

2H

2.91

5.0

−16.81

Bmag323 + 0.05

5H

2.93

5.0

−17.07

GBM1231 + 0.03

5H

5.62

10.2

25.63

Scssr5599 − 0.01

6H

2.76

4.7

16.19

HI

C09

GBM5008 + 0

5H

5.47

10.5

0.01

V09

Bmag606 + 0.03

3H

5.18

10.5

−0.01

Days to maturity

V09

Bmac32 − 4.76

1H

3.29

5.0

−0.68

Bmag606 + 0.03

3H

10.78

16.0

−1.50

Bmag740 + 0

4H

3.33

4.7

−0.65

Bmac113 + 0

5H

4.06

5.5

−0.71

Bmag323 + 4.05

5H

4.42

7.2

−0.80

Two-row

Biomass

V08

bPb6821 − 2.65

7H

2.32

9.7

41.79

C09

bPb8213 − 1.67

7H

2.74

9.0

14.01

Yield

V08

bPb6821 − 2.65

7H

2.06

9.2

13.57

C09

bPb2279 + 0.04

2H

2.38

7.7

−12.22

V09

bPb1212 + 0

6H

2.88

9.1

17.65

HI

V08

bPb2501 − 0.77

2H

3.43

10.7

0.01

C09

bPb8721 + 1.98

2H

2.04

7.4

0.01

V09

bPb6466 − 0.83

2H

2.94

9.8

0.08

bPb4293 + 0.01

2H

2.30

7.3

−0.07

E environment, R2 proportion of the phenotypic variance explained by each QTL, Additive effect additive effect of the allele from W89001002003 compared with I60049, C09 Castor-2009, V08 Vegreville-2008, V09 Vegreville-2009

Leaf area index

LAI was measured only in the W × I population at C09 and V09, and the QTL for LAI were only found at V09. Two QTL for LAI-S and four QTL for LAI-G were detected. One QTL for LAI-S was co-located with QTL for LAI-G which was positioned on Chr. 3H near marker GBM1405. One QTL for LAI-G co-located with ΔC3.1-WI, with W89 allele responsible for reduced LAI at this locus (Table 7).

Plant height

Plant height was measured only in the W × I population. Eleven QTL were associated with plant height. Eight QTL affecting plant height were detected at C09, among those, three QTL were also identified at V09. The QTL with the largest effect on plant height was co-located with ΔC3.1-WI, explaining 27.4% of the phenotypic variance in plant height at V09, and this QTL was detected at C09 as well. Increased plant height was associated with the W89 allele at this locus (Table 7).

Days to maturity

Days to maturity was recorded in the W × I population only at V09. Five QTL were detected for days to maturity. One QTL located on Chr. 3H, explaining 16% of the phenotypic variance in days to maturity with a LOD score of 10.78, and this QTL was also co-located with ΔC3.1-WI. Another QTL on Chr. 5H near marker Bmac113 was co-located with QTL for leaf Δ13C (designated as ΔC5.1-WI), biomass and grain yield. The W89 allele of all QTL detected at V09 contributed to early maturity (Table 7).

Total aboveground biomass

For the W × I population, four QTL associated with total aboveground biomass were found. Only one QTL was detected at C09. The three QTL detected at V09 explained from 3.7 to 6.3% of the phenotypic variance in total aboveground biomass, with LOD scores ranging from 2.21 to 3.76. For the M × H population, only two QTL were detected, with one QTL at V08 and one at C09. Both loci increased total biomass with allele from Merit (Table 7).

Grain yield

For the W × I population, variation for grain yield was associated with six QTL that clustered into five chromosomal regions. On Chr. 5H near marker GBM5008, overlapping QTL from both locations were detected. This QTL interval included QTL for biomass, HI, and LAI. The W89 allele was associated with increased grain yield, biomass, HI and LAI at this locus (Table 7). One QTL on Chr. 5H near marker Bmag323 was co-located with QTL for leaf Δ13C and biomass, with increased grain yield from the I60049 allele (Table 7). One QTL detected at C09 was co-located with ΔC3.1-WI, explaining 11.4% of the phenotypic variance in grain yield at a LOD value of 4.42, with the W89 allele responsible for reduced grain yield at this locus (Table 7).

For the M × H population, three QTL associated with grain yield were identified. One QTL was co-located with ΔC6.1-MH. Another QTL on Chr. 7H near marker bPb6821 overlapped with QTL for grain yield, biomass and ΔC7.1-MH. Both of these QTL (near bPb1212 and bPb682) increased the grain yield by the Merit allele (Table 7).

Harvest index

For the W × I population, only two QTL were detected for HI. One QTL on Chr. 5H was co-located with QTL for LAI-G, biomass and grain yield with increased HI from the W89 allele. The other QTL for HI on Chr. 3H was mapped around the same position with QTL for LAI-G, grain yield, plant height and ΔC3.1-WI.

For the M × H population, four QTL for HI were identified on Chr. 2H. One QTL near marker bPb4293 was co-located with the QTL for grain yield, which reduced the HI value by the Merit allele (Table 7).

Discussion

Leaf Δ13C and rainfall across environments

Changes in the leaf Δ13C of parental lines across years and locations were consistent with trends of the total growing season rainfall, suggesting environmental effects on leaf Δ13C (Fig. 1). Teulat et al. (2002) reported that both total rainfall and the ratio of rainfall to evapo-transpiration had a significant impact on Δ13C, explaining mainly the environmental effects in their study. For water-limited environments such as the Canadian Prairies where crops rely on soil-stored moisture (Anyia et al. 2008), the temporal distribution of rainfall is critical for crop growth (Bonsal et al. 1999; Chakravartia 1972). In this study, leaf Δ13C was significantly related to total within-season precipitation (r = 0.42, n = 122, p < 0.001), especially the amount of rainfall during June (r = 0.48, n = 122, p < 0.001). Ivlev and Voronin (2007) reported that the dynamics of δ13C in the annual rings of trees were found to be related positively with air temperature in June but negatively with precipitation in June from 1980 to 2005; even though trees are very different plant species from the barley, their biological response to water availability should be similar.

Leaf Δ13C was significantly reduced under drought conditions such as C09, which is in agreement with most reports on Δ13C in C3 crops (Craufurd et al. 1991; Hall et al. 1990; Virgona et al. 1990). Leaf Δ13C of extreme RILs differed more in the low-rainfall environment (such as C09) than the high rainfall environment (such as L08). This result is in agreement with previous reports, which suggests that Δ13C can be used as a sensitive indicator for plant water status or water availability during the growing period (Bloch et al. 2006; Craufurd et al. 1991; Merah et al. 2001). Significant correlations between leaf Δ13C across environments were observed for both populations, demonstrating the stability of leaf Δ13C across the tested environments.

Relationship between leaf Δ13C and agronomic traits

Positive or neutral relationships are frequently reported between Δ13C and grain yield and/or biomass in environments characterized with plentiful within-season rainfall or supplemental irrigation (Araus et al. 1998, 2003; Condon et al. 1987, 1993; Jiang et al. 2006; Merah et al. 1999, 2001; Monneveux et al. 2006; Morgan et al. 1993; Teulat et al. 2001c; Voltas et al. 1999). The correlations between leaf and/or grain Δ13C and grain yield have been reported to be positive, negative or neutral, depending on the target environment and row type on the Canadian Prairies (Anyia et al. 2007). In this study, leaf Δ13C was found to be positively correlated with aboveground biomass and grain yield for both populations under field conditions (Tables 4, 5; p < 0.05). One possible explanation of such relationship is the positive association between leaf Δ13C and gs (Condon et al. 1987), because more carbon could be fixed early in the season when stomatal opening was less limited by water deficit (therefore a high Δ13C).

Positive relationship between leaf Δ13C and gs in barley under field and greenhouse (well-watered) conditions was observed in a previous study (Chen et al. 2011), and similar results have been reported in common bean (Phaseolus vulgaris L.) (Ehleringer 1990), rice (Kondo et al. 2004; Takai et al. 2009) and wheat (Monneveux et al. 2006). Takai et al. (2009) found that a QTL controlling leaf Δ13C on the long arm of chromosome 3 in rice was associated with gs. Another hypothesis is that genotypes differ in their abilities to translocate stem carbohydrate reserves for grain filling, since high leaf Δ13C genotypes tend to grow faster and convert more assimilates to grain than low Δ13C genotypes (Monneveux et al. 2005). The positive relationship between leaf Δ13C and HI (Tables 4, 5) also suggested that genotypes with high leaf Δ13C genotypes were more efficient in dry matter partitioning to grain (Teulat et al. 2001c).

The Δ13C provides a long-term average estimate of cumulative WUE integrated over time and space (Condon et al. 2002). Leaves sampled for Δ13C at the stem elongation stage, when there is usually little drought stress and low vapor-pressure deficit, could reflect the integrated WUE during vegetative development and formation of yield potential, and evaluate vegetative establishment (Anyia et al. 2008; Chen et al. 2011; Condon and Richards 1992). There was also a positive relationship between leaf Δ13C and LAI-S across locations in this study (Table 4). LAI is usually used to estimate early vigor (the early growth of leaf area and biomass, Richards (1996)), canopy photosynthesis, evapo-transpiration and final yield since it represents the percentage of incident solar radiation intercepted (Dale et al. 1980). The significant positive correlations among LAI-S, leaf Δ13C, biomass and grain yield in this study suggest that both leaf Δ13C and LAI at the stem elongation stage could be used to estimate final yield.

Leaf Δ13C and row type

Jiang et al. (2006) reported that values of Δ13C from different plant parts (flag leaf, awn and grain) were on average higher for six-row than for two-row barley, suggesting a higher WUE of two-row than six-row barley under irrigated and non-irrigated field conditions. A proposed explanation for the difference in Δ13C values between barley ear types is the difference in days to maturity, as low Δ13C values appeared in late-maturing genetic lines (Craufurd et al. 1991; Sayrea et al. 1995). Another possible explanation suggested by Jiang et al. (2006) is that the intrinsic difference may exist in water/carbon metabolism at the whole tiller, flag leaf and ear levels, since flag leaves of two-row barley are generally much smaller than six-row barley. The six-row barley may fix more carbohydrates from flag leaf blades and stems, and therefore they were enriched in 13C. We did not find any significant difference in leaf Δ13C between two-row and six-row barley in previous studies (Anyia et al. 2007; Chen et al. 2011). In the present study, the two populations differed in leaf Δ13C, with the M × H RILs producing a significantly lower leaf Δ13C than the W × I RILs at V09 (data not shown, p < 0.001), which needs to be verified under multiple environments.

Six-row barley has more florets and therefore it is usually considered to be more fertile than two-row barley. In this study, the W × I RILs produced more biomass and grain yield than the M × H RILs at V09 (p < 0.01), but there was no significant difference in grain yield between these two types of barley population when soil moisture was not sufficient at C09. According to Forster et al. (2004), six-row barley has an advantage over two-row barley since six-row type generally matures earlier and therefore it is valuable in breeding for drought escape for countries in North Africa such as Morocco. It is also possible that the differences in tiller number between six-row and two-row barley may explain their leaf Δ13C differences; however, this may need to be investigated further.

Heritability of leaf Δ13C

In the present study, the H2 for leaf Δ13C was high (0.80) for the six-row barley population across all environments. Under a single environment (C09 or V09), the H2 for both populations all reached 0.80. These results suggest that leaf Δ13C is under strong genetic control and can be reliably used as a measure of WUE under field conditions. For many complex traits (such as biomass, HI and yield components), genetic gain is slow under water-limited environments due to a large interaction between genotype and environment and a low heritability (Rebetzke et al. 2008). The high H2 of physiological or secondary traits that are correlated with yield, such as Δ13C, presents a good opportunity for plant breeding in drought-prone regions (Stiller et al. 2005). Although the expression of Δ13C in leaf and other plant tissues varies with water supply, and the spatial variability in soil water availability can reduce genetic variance of Δ13C (Condon et al. 1992; Rebetzke et al. 2008), the heritability for Δ13C is still high compared with other complex traits under drought conditions. Several studies have previously reported high H2 values for Δ13C. Hubick et al. (1988) observed that the H2 of Δ13C (whole plant, excluding the pods and roots) in field-grown peanut cultivars was 0.81, and there was no significant interaction between genotype and environment for Δ13C. Condon and Richards (1992) found that the H2 of Δ13C in wheat was greatest for plant material sampled before or during early stem elongation (0.95 on genotype basis or 0.88 on a single-plot basis) as compared with plant parts formed near anthesis (flag leaf or plants tops at ear emergence). In another study on wheat, the narrow-sense heritability of leaf Δ13C ranged from 0.37 to 0.91 on a single environment basis and from 0.76 to 0.86 on genotype-mean basis in a QTL analysis using three wheat mapping populations across 3 years (Rebetzke et al. 2008).

QTL for Δ13C across environments and populations

The CIM method revealed 12 putative QTL associated with leaf Δ13C for the W × I population, and 5 QTL for leaf Δ13C from the M × H population in the current study. Common QTL across environments and populations can provide validation of the QTL and identify robust markers across different gene pools for marker-assisted selection. The significant low level of polymorphism between parental lines (especially in the M × H population) resulted in the low density and uneven distribution of markers in the QTL maps obtained in the present study. For example, Chr. 3H of the M × H map was split into three linkage groups and was only covered by 14 markers, which may have hindered the detection of putative QTL for this chromosome. The low marker density and the lack of common markers prevented us from identifying and comparing common QTL across the two populations evaluated in the present study. However, it may be possible to identify a few common QTL for leaf Δ13C across the two mapping populations if more markers are placed to saturate the maps. Since the order and not the distance between markers is usually considered conserved between populations (This et al. 2010), 7 of the 12 QTL regions detected in the W × I population in this study (region on Chr. 2H around marker EBmac558, Chr. 3H between marker Bmag606 and Bmag13, Chr. 5H near marker Bmac113 and on Chr. 6H near marker Bmag173) were identified to be similar to those previously reported for grain Δ13C in the Tadmor × Er/Apm population under three Mediterranean field environments by Teulat et al. (2002). Specifically, the ΔC5.1-WI in our study overlapped with grain Δ13C in the Tadmor × Er/Apm population under irrigated field conditions (Diab et al. 2004). Ellis et al. (2002) reported similar chromosome regions for shoot Δ13C (3H and 5H) in their Derkado × B83-12/21/5 DH population.

Although ΔC3.1-WI on the W × I linkage map was detected consistently in all three field locations, its effect on leaf Δ13C varied across environments, and this QTL explained 22% of the phenotypic value at L08, followed by C09 (14.4%) and then V09 (11%), with no significant G × E for leaf Δ13C in the W × I population. In agreement with Teulat et al. (2002), the observed differential effects of this QTL across the different environments studied may be attributable to the total growing season rainfall. Condon et al. (1992) also pointed out that the environmental effects on Δ13C can be attributed to stomatal closure in response to declining soil water and/or increasing vapor-pressure deficit. Overall, the QTL identified in this study varied in size and accounted for small to modest amounts of the phenotypic variance, which is consistent with most of the previous studies in QTL mapping of Δ13C across a range of plant species.

Effect of parental alleles

It is imperative to understand the inheritance of Δ13C and the mode of gene action to develop cultivars with high WUE via selection for low Δ13C lines. There was one report on cytoplasmic inheritance of Δ13C in cultivated sunflower by Lambrides et al. (2004), but most studies provided strong evidence for nuclear genetic control of Δ13C, with QTL for Δ13C assigned to specific chromosomes in several different species, such as Arabidopsis thaliana (Hausmann et al. 2005; Juenger et al. 2005), barley (Diab et al. 2004; Ellis et al. 2002; Ellis et al. 1997; Handley et al. 1994; Teulat et al. 2002), cotton (Saranga et al. 2001), rice (Laza et al. 2006; Price et al. 2002; Takai et al. 2006; This et al. 2010; Xu et al. 2009), soybean (Specht et al. 2001), tomato (Xu et al. 2008) and wheat (Rebetzke et al. 2008). Predominantly additive gene action of Δ13C has been reported in alfalfa (Medicago sativa L.) (Johnson and Rumbaugh 1995), Arabidopsis thaliana (Hausmann et al. 2005; Juenger et al. 2005), common bean (White 1993), wheat (Ehdaie and Waines 1994; Rebetzke et al. 2006) and rice (Takai et al. 2006).

In the present study, the low leaf Δ13C parent W89 had favorable alleles for WUE at loci on Chr. 2H, 3H and 5H, and the high leaf Δ13C parent I60049 contributed favorable alleles for WUE at loci on Chr. 1H, 6H and 7H. For the M × H population, the high leaf Δ13C parent H93174006 had favorable alleles for WUE at loci on Chr. 6H and 7H. The complementary alleles at multiple loci contributed by the parents provided the most plausible explanation for transgression observed among the progeny (Tanksley 1993). The additive gene action underlying Δ13C suggests that replacement and fixation of desirable alleles within a locus could be achieved by selecting lines with high or low Δ13C (Rebetzke et al. 2003, 2008). Independent alleles at multiple loci could also be pyramided to develop lines for further altered Δ13C.

Clusters of QTL

Multiple regions controlling leaf Δ13C co-located with the QTL for agronomic traits (such as aboveground biomass and grain yield) were identified in both RIL populations in the present study, which is consistent with previous reports. Teulat et al. (2001a, b, 2002) reported eight QTL for grain Δ13C co-located with QTL for agronomic traits and traits related to plant water status and/or osmotic adjustment. Most QTL for Δ13C have previously been reported to be co-located with QTL for yield components and heading date and/or plant height (Forster et al. 2004; Juenger et al. 2005).

Correlations (Tables 4, 5) between leaf Δ13C and LAI-S, plant height, days to maturity, biomass, grain yield and HI were partially explained at the level of co-localized QTL (Fig. 3a, b). One of the genomic regions with the most overlapping traits was on Chr. 3H near marker Bmag606, with W89 alleles favoring lower leaf Δ13C, increased plant height, and reduced LAI, grain yield, HI and days to maturity. Around the same region, QTL for Δ13C was identified near the semi-dwarf gene sdw1 in different barley mapping populations (Ellis et al. 2002; Teulat et al. 2002), which is homoeologous to rice Chr. 1 where Δ13C is located. Another noteworthy region was on Chr. 5H near marker Bmac113, with W89 alleles favoring lower leaf Δ13C, reduced biomass and yield. In the same region, QTL for shoot δ13C and grain yield were associated with another semi-dwarf gene ari-e.GP (Ellis et al. 2002).

The association between Δ13C and plant height and/or heading date could confound and compromise the relationship between Δ13C and grain yield (Rebetzke et al. 2008). In this study, there were also QTL (such as ΔC6.1-MH and ΔC7.1-MH) that explained 8.6–11.2% of the phenotypic variation but only showed co-localization with aboveground biomass and grain yield, which may be more attractive to plant breeders. Rebetzke et al. (2008) suggested that the effect of Δ13C on yield should be separated from plant height and development effects (such as heading date and flowering time) through covariance analysis. The co-localization of QTL may result from pleiotropic relationships between these traits or due to genetic linkage among the traits. However, clustering regions are of interest in terms of plant breeding as they control both drought-adaptive traits such as Δ13C and yield components.

Conclusions

The temporal variation of within-season rainfall distribution on the Canadian Prairies has a significant impact on barley leaf Δ13C. Sufficient genotypic variation, stability across environments, and a high H2 indicate that leaf Δ13C is a good surrogate for improved WUE in breeding for barley varieties under rain-fed conditions. The transgressive variation of leaf Δ13C observed in our mapping population and results from previous studies (Anyia et al. 2007; Chen et al. 2011) suggest that it is possible to select progeny lines with low Δ13C or high WUE that contribute favorable alleles of value for improvement of WUE. This study has confirmed the polygenic inheritance of leaf Δ13C in barley by detecting multiple QTL controlling leaf Δ13C. Several QTL for leaf Δ13C were found to overlap with QTL for important agronomic traits that would need further validation. A major QTL for leaf Δ13C (ΔC3.1-WI) was identified across all locations that overlapped with several agronomic traits, with W89 alleles favoring lower leaf Δ13C, increased plant height, and reduced LAI, grain yield, HI and days to maturity. Since overlapping or co-location of QTL suggests pleiotropic relationship or genetic linkage, care must be taken to ensure that low Δ13C does not impose penalty on grain yield under favorable conditions. More research is needed to further dissect this major QTL region for low Δ13C towards the identification of candidate genes as well as to understand the genetic mechanisms underlying these co-located traits.

Acknowledgments

This study was funded by Alberta Agriculture Research Institute (AARI), Alberta Crop Industry Development Fund (ACIDF), Alberta Barley Commission (ABC), Brewing and Malting Barley Research Institute (BMBRI), the University of Alberta and the Natural Sciences and Engineering Research Council of Canada (NSERC). We are grateful to the staff at the Field Crop Development Centre (FCDC), Lacombe, for providing the materials used in this study. The authors are grateful to three anonymous reviewers for their valuable comments and suggestions that improved an earlier version of this manuscript.

Copyright information

© Springer-Verlag 2012