Introduction

Increased nitrogen use efficiency (NUE) is among multiple objectives in modern agriculture (Gastal and Lemaire 2002). NUE has been usually defined as the yield of grain (harvested product) achieved per unit of available N in the soil (Moll et al. 1982; 1987). Crop plants are only able to incorporate 30–40 % of the applied nitrogen into useful food products such as grain (Raun and Johnson 1999). Nitrogen that is not used by the plants has deleterious environmental effects and reduces the economic return (Good et al. 2004). Breeding to develop more efficient varieties is essential to improve NUE of cereal crops (Galloway et al. 2008; Bingham et al. 2011).

Many authors have reviewed literature about NUE and its improvement (e.g. Glass 2003; Good et al. 2004). NUE is a complex trait with a heritability range of 0.50–0.86. Two component processes governing NUE (Moll et al. 1982; 1987; Andrews et al. 2004; Gallais and Hirel 2004; Anbessa et al. 2009) are (i) nitrogen uptake efficiency: the ability of the plant to take up N from the soil, and (ii) nitrogen utilization efficiency: the ability of the plant to transfer the N taken up to the harvestable crop product. The relative importance of these component traits to determine NUE depends on the level of available N and is not consistent across the reports for different crops (Moll et al. 1982; Sanford and MacKown 1986; May et al. 1991; Muurinen et al. 2006; Abeledo et al. 2008; Anbessa et al. 2009; Sylvester-Bradley and Kindred 2009; Beatty et al. 2010; Bingham et al. 2011).

Barley (Hordeum vulgare L.) is one of the earliest cultivated crops (Badr et al. 2000; Doebley et al. 2006; Morrell and Clegg 2007) and has become the subject of extensive quantitative trait locus (QTL) analysis for many agronomic traits (Hayes et al. 1993; Laurie et al. 1995; Schut et al. 1997; Thomas et al. 1998; Yin et al. 1999a, b; Arru et al. 2003; Mickelson et al. 2003; Francia et al. 2004; Li et al. 2005; Yin et al. 2005; Gyenis et al. 2007; Li et al. 2007; Nduulu et al. 2007; Saal et al. 2011; Zhou et al. 2012). Knowledge about QTL detection on the effects of different soil-N levels on yield and quality parameters of cereals is generally limited (Saal et al. 2011), and only a few examples for wheat were reported (An et al. 2006; Laperche et al. 2007; Guo et al. 2012; Cormier et al. 2013). However, studies on QTL mapping of NUE in barley are scarce.

Genotypic differences in NUE of barley have been reported before (Sinebo et al. 2004; Abeledo et al. 2008). According to Anbessa et al. (2009) the breeding strategy for high NUE is through selection for both nitrogen uptake and utilization efficiency, although breeding efforts do not always result in improved NUE for barley genotypes (Muurinen et al. 2006). Recent advances in identifying genes may play a role in improving those two processes contributing to NUE (Mayes et al. 2005; Shewry Jones 2005; Lea and Azevedo 2006; Bingham et al. 2011). The objectives of this study are to map QTLs controlling NUE, its components, and other agronomic traits in barley, and to analyse the stability of these QTLs under N-limited and non-limited growing conditions.

Materials and methods

Plant material

We used the Prisma × Apex recombinant inbred line (RIL) mapping population, produced by single-seed descent method over eight generations (Yin et al. 1999b). The population comprised 94 RILs. The parents were commercial two-rowed spring barley cultivars in the Netherlands (Schut et al. 1997). They have marked differences in yielding ability, morpho-physiological characteristics and phenological traits (Raven et al. 2005; Yin et al. 2005). Prisma usually out-yields Apex (Raven et al. 2005). Plants of Prisma usually flower later than those of Apex. Prisma is also shorter than Apex, largely due to the semi-dwarfing and recessive gene on Chromosome 3(3H) called denso (also called sdw1) (Jia et al. 2009). This gene confers for the distinctive prostrate juvenile growth habit (Fageria et al. 2005; Kuczynska and Wyka 2011) and can serve as a morphological marker (Francia et al. 2004; Parry et al. 2005).

The RIL population was genotyped using 191 AFLP markers (Yin et al. 1999b). These markers covered a total map length of 965 centiMorgan (cM), with the denso locus mapped on the long-arm of Chromosome 3(3H) at the position of 126.4 cM (Yin et al. 1999b; Fig. 1) and found to be associated with many traits (Powell et al. 1985; Hayes et al. 1993; Bezant et al. 1996; Yin et al. 2000; Li et al. 2007). Yin et al. (2000) reported the advantage of the denso genotypes, especially when grown under nitrogen-rich conditions. The presence of a wide range of variation in nitrogen response between RILs of this population from previous reports of Yin et al. (1999b; 2000) suggests that this population is suitable to map QTLs for NUE-determining traits.

Fig. 1
figure 1

Chromosome regions associated with different traits of barley both from 2005 to 2008 data. MapChart software 2.2 (Voorrips, 2002) was used to draw the figure. The positions of loci are given to the left of the linkage map in centiMorgan (Yin et al. 1999b). QTL bars, shown in different colours to the right of the linkage group, are represented as confidence intervals of inner and outer LOD support intervals taking 1 and 2 LOD drop-off values respectively from the highest LOD profile peak (Tables 2, 3). QTL nomenclature was adapted from McCouch et al. (1997) in the order of Q-trait-treatment-method of analysis. Both chromosome designations of barley i.e. 1–7 according to the Triticeae system of nomenclature and 7H, 2H, 3H, 4H, 1H, 6H and 5H according to Linde-Laursen et al. (1997) were used

Full size image

Study conditions, experiments and design

Two pot experiments were conducted in an open screen house at Jiangxi Agricultural University (latitude 28° 41′ N, longitude 115° 53′ E), China, in 2005 and in 2008, respectively. The experiments used 3 (2005) or 2 (2008) replications, with each replication in one block. Each replicate had 3 pots of 25 cm height and 20 cm diameter per RIL in both years. The pots were filled with poor quaternary red soil collected from a nearby village (pH 4.8 in 2005, pH 5.1 in 2008). The soil had 81.86 mg kg−1 available N, 4.03 mg kg−1 available P, 195.8 mg kg−1 available K, 0.092 % total N and 2.08 % soil organic matter in the 2005 experiment, and 75.22 mg N kg−1, 13.94 mg P kg−1, 188.3 mg K kg−1, 0.056 % total N and 2.13 % organic matter for the 2008 experiment. Three seeds were planted in each pot and plants were finally thinned to one per pot. Within a replication, pots were systematically arranged according to the previous report on plant height of the RILs (Yin et al. 1999b), so as to minimise shading effects. RILs were exposed to different levels of N fertilizer. In 2005, N fertilizer was applied at four rates of 0, 0.056, 0.111, and 0.167 g N pot−1, denoted as N0, N1, N2, and N3, respectively. In 2008, only three N levels but covering a wider range were applied, i.e. 0, 0.157 and 0.471 g N pot−1, denoted here as N0, N1, and N3, respectively.

Traits evaluated in the trials

The N contents of both the straw and the grain were measured only at maturity stage. The N content of the root was not measured in either year. A total of 22 traits were evaluated from the experimental data of each year (Table 1).

Table 1 Definitions, codes, units, values of the traits for population mean (standard deviation SD in brackets) and the two parents (i.e. Apex and Prisma) shown as the average across N treatments, and heritabilities (H 2) of the traits used for QTL analysis of 2005 and 2008 data
Full size table

Nitrogen uptake efficiency was computed by dividing the total N in the above-ground biomass (AgNup) by the available N (i.e. soil residual N + applied N), denoted as NupEb. Nitrogen utilization efficiency was considered on the basis of grain weight (Gw) and of above-ground biomass, respectively, i.e. the grain yield divided by the total N content of the above-ground biomass (NutEg) and the above-ground biomass divided by the total N content of the above-ground biomass (NutEb). Accordingly, NUE was computed as either grain weight divided by the amount of available N (NUEg), or above-ground biomass divided by available N (NUEb). NUE and its component traits, and nitrogen harvest index (Table 1) were calculated both for the control pots and for those pots receiving N fertilizer. In addition, common slope (i.e. recovery efficiency) and common intercept of linear regression of N uptake against N application (Table 1) were calculated from the data across N treatments.

Other traits evaluated for the 2005 data set were plant height, stem weight, grain weight, above-ground biomass, sheath weight, leaf weight, ear weight, straw weight, thousand kernels weight, straw N content and grain N content (Table 1). For 2008, the same traits were measured, except for sheath weight which was replaced by root weight. The average values from nine plants in 2005 and six plants in 2008 were used for statistical analysis.

Statistical analysis

Data were checked for model assumptions of normality and equal variance distributions. Pearson correlation coefficients among the pairs of measured traits and among the environments for a given trait were calculated. Statistical tests were carried out using GenStat® 14th Ed. (Payne et al. 2009), based on the model Pij = Gi + Ej + eij where Pij stands for the performance of genotype i in environment j; Gi stands for the effect of genotype i; Ej stands for the effect of environment j; and eij represents the residual. We only assessed the main effects of RILs and N levels using the mean values over replications, but were unable to statistically evaluate the G × E interaction effect on the NUE related phenotypic data of both experiments as plant N contents were too expensive to be assessed on the individual replication basis for the whole population. For consistency, no G × E interaction was considered for all traits. This analysis was conducted for all traits except recovery efficiency and common intercept that were estimated from the data across all N levels.

For those traits that showed a significant difference among RILs, heritability (H 2) estimates were calculated using the linear mixed model or restricted maximum likelihood (REML) variance component analysis of GenStat®14th Ed (Table 1). H 2 was calculated as (Falconer and Mackay 1996):

$$H^{2} = \frac{{\sigma_{G}^{2} }}{{\sigma_{G}^{2} + \sigma_{\varepsilon }^{2} /n_{t} }}.100$$

where σ 2G is the genotypic variance, σ 2ε is the error variance and n t is the number of environments (N levels).

QTL analysis

Using the AFLP marker linkage map (Yin et al. 1999b) and MapQTL®6 mapping software (Van Ooijen 2009), QTL analysis was carried out for a given trait with the significant main effect of the RILs. The QTL analyses were done only for the above-ground traits of barley RILs. Assuming the additivity of QTL effects (no epistasis), multiple QTL mapping (MQM) analysis was performed in two steps.

Initially, the interval mapping method (Lander and Botstein 1989; Chen et al. 2010) was performed at 2 cM intervals (Price 2006). Permutation test, as implemented in MapQTL (Van Ooijen 2009), was performed to determine the LOD thresholds for detecting QTLs of each quantitative trait, by shuffling the original data 10,000 times (Doerge and Churchill 1996) keeping the marker data constant. Genome-wide scanning at 95 % marker-trait associations was used as significance level.

Then, the MQM mapping method (Van Ooijen 2009) was applied, in which background markers were selected as co-factors to take over the role of the putative QTLs identified by the interval mapping step to reduce the residual variance. First, the denso locus was deliberately chosen as a co-factor for all traits, as previous studies had highlighted its importance in the control of many traits (Yin et al. 1999b, 2005). Secondly, other background markers closest to the indicated region of possible QTLs (LOD ≥ 2.8) (Tables 2, 3) were gradually added as co-factors until the LOD profile was stabilized or until no further indicated QTL was found. Besides, the Automatic Co-factor Selection, which was implemented in MapQTL®6 (Van Ooijen 2009) was practised to confirm the co-factors selected. In the final LOD profile, QTLs were declared based on the threshold LOD scores ranging from 2.8 to 3.0, depending on chromosome map length and the number of chromosome pairs (Van Ooijen 1999).

Table 2 List of QTLs identified using AFLP markers and a population of 94 Prisma × Apex RILs of barley for N related traits and other traits of the 2005 data
Full size table
Table 3 List of QTLs identified using AFLP markers and a population of 94 Prisma × Apex RILs of barley for different N related traits and other traits of 2008 data
Full size table

The calculation of the additive allelic effect of a QTL was performed, as [(mean of the Prisma class—mean of the Apex class)/2] by using MapQTL®6 software. The phenotypic variance explained by a single QTL (i.e. R2), was also estimated by the software.

Results

Phenotypic values of the traits examined differed between years (Table 1), largely because N applications were higher in 2008 than in 2005. Significant correlations were found between many pairs of the examined traits but generally little correlations were found for a given trait among the seven N × season combinations (see Supplementary Tables). Based on the statistical test, only 16 traits from the 2005 trial and 10 traits from the 2008 trial showed significant variations among the RILs (Table 1). These traits as well as the common slope and common intercept were promoted to QTL analysis.

Overview of QTL detection

QTLs were identified for the traits Ds, Ph, Lw, NUEb, Gw, Ew, NUEg, Agb, Sw, Shw, AgNup and NupEb (see Table 1 for their definitions) using the 2005 experimental data (Table 2). Similarly, QTLs for Ph, NHI, Tkw, GN, NutEg and GHI were detected using the data of the 2008 experiment (Table 3). The R2 in the QTL model ranged from 8.6 to 54.4 % in 2005 and from 8.4 to 42.6 % in 2008 across these traits.

On Chromosome 4(4H) no QTLs were detected. A total of 41 QTLs were detected for 18 traits across the other 6 chromosomes in both trials. Fifteen QTLs were related to NUE and its components and 26 QTLs belonged to the other traits. In both trials, most QTLs for NUE and its components were detected at the lower N rates and no QTL was detected at the highest rate. Chromosomes 3(3H) and 7(5H) in 2005 and Chromosome 2(2H) in 2008 were responsible for QTL regions controlling NUE and its components.

The commonalities among the detected QTLs were low. Except QTLs of Ph and NUEg, none of the QTLs detected for a given trait in 2005 had expressed themselves significantly in 2008. QTLs of some traits like Ph, Agb, Ds, Shw and NUEb in 2005 (Table 2) and QTLs of NUEg, NutEg, NHI, GN and Ph in 2008 (Table 3) were mapped to the same genomic region. Furthermore, many traits were associated with the denso gene at different extents (Table 2), because their QTLs were detected on Chromosome 3(3H) close to the denso locus (Fig. 1). Next, we specifically present the results of QTL analysis of NUE and its components.

QTLs detected for traits related to nitrogen uptake and use

Nitrogen uptake of above-ground biomass 2005

One major QTL accounting for 25.2 % of the variation and with a negative effect from Prisma was detected for AgNup at 125.2 cM of Chromosome 3(3H) at N2 in 2005 (Table 2; Fig. 1).

Nitrogen uptake efficiency of above-ground biomass 2005

One major QTL accounting for 27.8 % of the variation and with a decreasing allelic effect from Prisma was detected for NupEb at 133.4 cM position of Chromosome 3(3H) at N2 in 2005 (Table 2; Fig. 1).

Nitrogen use efficiency of above-ground biomass 2005

Two major QTLs were detected for NUEb in 2005. The QTL on Chromosome 3(3H), at 125.2 cM position, with a relatively large and negative effect from Prisma was responsible at N2. This QTL accounted for 20.8 % of the variation in NUEb. The other QTL at 100.6 cM of Chromosome 7(5H) was expressed at N1. This QTL has an increasing allele from Apex with a relatively higher magnitude and accounted for 13.4 % of the R2 for NUEb (Table 2; Fig. 1).

Nitrogen use efficiency of grain 2005

One major QTL accounting for 15.1 % of the variation in NUEg and having an increasing allele from Apex, was expressed at 99.6 cM position of Chromosome 7(5H) at N1 in 2005 (Table 2; Fig. 1).

Nitrogen use efficiency of grain 2008

One major QTL at 122.0 cM of Chromosome 2(2H) was detected at N0 in 2008. This QTL accounted for 21.3 % of the variation in NUEg and had a decreasing allelic effect from Prisma (Table 3; Fig. 1).

Nitrogen utilisation efficiency of grain 2008

Regions at 122.0 and 111.8 cM of Chromosome 2(2H) were detected to be important for NutEg at N0 and N1 respectively in 2008. The respective magnitudes of allelic effects and R2 values (%) for NutEg were 5.69 and 21.3 % for the region at 122.0 cM and 5.96 and 17.6 % for the region at 111.8 cM. Both regions had a decreasing allelic effect from Prisma (Table 3; Fig. 1).

Nitrogen harvest index 2008

Major QTLs were detected at all the N levels for NHI in 2008. A QTL on Chromosome 2(2H) was found at 122.0, 116.8 and 115.8 cM for N0, N1 and N3, accounting for 23, 21.7 and 14.9 % of the variation for this trait, respectively. Another QTL which accounted for 10.8 % of the variation at NHI was detected on Chromosome 1(7H) at 23.0 cM position just on top of the molecular marker (E39M61-517) for N1. All the QTLs for NHI had a decreasing allele from Prisma (Table 3; Fig. 1).

Grain nitrogen content 2008

A major QTL at 122.0–123.0 cM of Chromosome 2(2H) was detected for GN at both N0 and N3. The QTL had an increasing allele from Apex and accounted for 13.7 % of the variation in GN at N0 compared to 24.5 % at N3 (Table 3; Fig. 1).

Common intercept 2005

One major QTL was detected at 104.5 cM on Chromosome 5(1H) for the common intercept of the regression lines with N0, N1, N2 and N3 of the 2005 pot experiment (Table 2; Fig. 1).

Discussion

QTL-by-environment interaction on nitrogen use efficiency and its components

Although it was possible to detect QTLs for a trait (e.g. NUEg) in both 2005 and 2008, the QTLs were found at different N rates, and on chromosomes and loci (Tables 2, 3; Fig. 1). Our QTL analysis of NUE and its components mainly indicated inconsistent results across N levels and years, in line with low correlations among environments for a given trait (Supplementary Table). First, QTLs for NutEg and NUEg in 2008 and QTLs for NUEb and NUEg in 2005 were detected at low-N levels. In contrast, no QTL was detected at the highest N rate in both years in relation to NUE and its components. Our mean NUE of barley genotypes per N level decreased as the N level increased in both years.

Secondly, the QTLs detected in the experiment of 2005 for NupEb, NUEb and NUEg were either totally undetected or found to be detected on a different chromosome and N level in 2008 (Tables 2, 3; Fig. 1). For example, although significant variations were observed among genotypes in 2005 for NupEb, it was insignificant in 2008. As a result, a single QTL was detected only at N2 in 2005 (Table 2; Fig. 1). Furthermore, QTLs detected in 2008 for the traits of GN, NHI and NutEg were undetected in 2005.

Despite the semi-protected conditions for our experiments in a screen house, our overall limited ability to detect stable QTLs across seasons and N levels might be partly due to low power of the experiments with only 94 RILs. Vales et al. (2005) indicated that QTLs with small effects can be detected only by increasing population size. It is well known, however, that both genotypic variability and environment account for differences in NUE of spring barley (Anbessa et al. 2009; Saal et al. 2011). The inconsistencies could also be partly attributed to the presence of QTL × E with some QTLs determining the adaptation to specific environments (Yin et al. 1999b; 2000; 2005; Asíns 2002; Hackett 2002; Yang et al. 2007; Liu et al. 2010; Comadran et al. 2011; Cormier et al. 2013). The effect of QTL × E could be well accounted for by the facts that (1) high-yielding varieties released during the green revolution were selected to respond to high N inputs (Kamprath et al. 1982; Earl and Ausubel 1983; Ceccarelli 1996); (2) NUE is expected to decrease as N supply increases (Gauer et al. 1992; Raun and Johnson 1999; Abeledo et al. 2008). The latter is particularly in line with the lack of QTLs found for NUE under high N conditions. Furthermore, growing of spring barley in pots for our experiments differed from the situation in the field. Despite the results of Beatty et al. (2010), who reported similar NUE and growth characteristics in genotypes across pot and field environments, the results of the simplistic and small-scale pot experiment should be confirmed by further analysis with large data sets tested for extra years, under realistic field conditions (Beatty et al. 2010).

Comparison with already mapped QTLs

Yin et al. (1999b) have detected QTLs for leaf N content at flowering on Chromosomes 2(2H), 3(3H), 5(1H) and 7(5H) in a field experiment using the same Prisma × Apex mapping population. Using a different population, Mickelson et al. (2003) reported that the candidate genes involved in N recycling or in regulation of the recycling process from flag leaves were located on Chromosomes 3(3H) and 6(6H). In accordance with these findings, QTL regions for NupEb and NUEb were detected on Chromosome 3(3H) in 2005 and Chromosomes 2(2H) and 7(5H) were responsible for N-related traits in both years. Chromosome 7(5H) and Chromosome 2(2H) were responsible for NUEg in 2005 and 2008 respectively. Unlike the previous studies, no QTL was detected for N parameters on Chromosomes 5(1H) and 6(6H) in both trials used in this study.

The importance of the denso locus on a number of agronomic and physiological traits in barley has been extensively reported previously (Powell et al. 1985; Hayes et al. 1993; Bezant et al. 1996; Yin et al. 2000; Li et al. 2007) and was further confirmed in our study (Tables 2, 3; Fig. 1). Our results further showed the co-localization of some QTL regions like the denso locus with QTL regions controlling NUE and its components. These co-localizations explain partly how the significant correlations between the examined traits (see the Supplementary material) came about.

Importance of nitrogen uptake and utilization efficiency

As stated earlier, NUE can be partitioned into two components: uptake efficiency and utilization efficiency (Moll et al. 1982; Gallais and Hirel 2004; Anbessa et al. 2009). In a recent review, the observed contributions of NutEg and NupEb to the variations in NUE of barley were reported to be 60 % and 40 % respectively, indicating a clear breeding gain in NutEg (Bingham et al. 2011). Our QTL analysis may support this finding because there were two major QTLs detected for NutEg compared to only one for NupEb (Tables 2, 3). The detections of a QTL for NUEb at N2 (Table 2) and a QTL for NUEg at N0 (Table 3) at positions similar to those for NupEb and NutEg respectively confirms the contribution of uptake efficiency and utilisation efficiency to NUE as component traits. In order to improve NUE further, both N uptake and utilization efficiency should be targeted for improvement (Moll et al. 1982; Moose et al. 2009).

High uptake efficiency in barley appears to result from a large uptake of N after anthesis driven by a large grain sink and was associated with fertilizer recovery rather than the acquisition of soil mineral N (Bingham et al. 2011). In line with this concept, our QTL analysis was conducted by using the common slopes as a proxy of N recovery across the N levels; but this ended up without detection of any QTL in both years. However, there were QTLs detected for AgNup and NupEb at N2 in 2005, but none in 2008. Although the QTL result was inconsistent in the Prisma × Apex RILs, N uptake by wheat has been reported to be controlled by sink demand rather than N availability (Mi et al. 2000). More studies would be needed to understand physiological factors governing nitrogen uptake and utilization efficiency.

Opportunities of improving nitrogen use efficiency while maintaining high yield

There have been inconsistent reports about NUE improvement in barley by breeding probably due to differences in the selection strategies among the breeding programmes (Kamprath et al. 1982; Earl and Ausubel 1983; Ceccarelli and Grando 1991; Ceccarelli 1994; 1996). Our study analysed both NUE related traits and common agronomic traits (Tables 2 and 3). There were significant correlations between grain yield (Gw) and NUE related traits (see the Supplementary material). These correlations were strongly affected by year, and were mostly positive in 2005 and negative in 2008. However, there was no QTL that co-localised for grain yield and NUE related traits, seemingly suggesting little opportunity to manipulate grain yield and NUE simultaneously. Except the QTL for Gw at N0 on chromosome 2 in 2005, there were hardly any QTLs detected for grain yield in our pot experiments (Tables 2 and 3), in contrast to the results of the field trials using this Prisma × Apex RIL population, in which significant QTLs including the denso locus on Chromosome 3(3H) were identified for grain yield (Yin et al. 1999b). Of the QTLs identified for NUE traits, the efficiency-reducing allele came mainly from Prisma (Tables 2 and 3). Prisma usually out-yields Apex under field conditions (Raven et al. 2005), largely due to the Prisma allele at the denso locus (Yin et al. 1999b). However, the fact that QTLs for NUE related traits found in 2008 located on Chromosome 2(2H) (Table 3) and some found in 2005 located on chromosome 7(5H) suggests opportunities to improve NUE and grain yield simultaneously through marker-assisted selection.

Some evidence suggests that spring barley breeding has generally been successful in increasing yields without increasing the optimum N application (Sylvester-Bradley and Kindred 2009). Abeledo et al. (2008) reported that the newest genotype was better than the oldest one in NUE of Argentinean barley. Anbessa et al. (2009) also reported that genotypic differences in NUE at low N environments were related to spike types (two-row vs. six-row) and breeders indirectly improved NUE while selecting for spike types and grain yield. Similar evidence for improving NUE has been reported for wheat, despite significant G × E interactions (Cormier et al. 2013). Since QTLs for NUE traits were identified only in low-N environments in our study, it remains to be seen whether it is possible to reduce nitrogen fertilizer requirements through breeding for high NUE while maintaining high yield of barley.

Conclusions and outlook

The focus of this study was to map the QTLs associated with NUE in the Prisma × Apex RILs of barley, using phenotypic data collected from two pot experiments. No QTL was detected on Chromosome 4(4H) in both trials. QTLs with major additive effects were detected on six other chromosomes. The R2 in the QTL model ranged from 8.4 to 54.4 % across all mapped traits in both years. Most QTLs for NUE and its components were detected at the lower N rates and no QTL was detected at the highest N rate in both trials. Promising QTL regions controlling NUE and its components have been found on Chromosomes 3(3H) and 7(5H) in 2005 and on Chromosome 2(2H) in 2008.

The expression of QTLs for a given trait was different at different environments (both years and N levels). Except the QTLs of plant height and NUE of grains, none of the QTLs which were detected for a given trait in 2005 have expressed themselves in 2008 irrespective of the N levels. The results of this study lead to the conclusion that the genetic basis of NUE is controlled by QTLs with major but inconsistent additive effects across nitrogen levels. This indicates the difficulty to select genotypes that perform well for high NUE across diverse N environments. Selecting genotypes for specific sub-optimum, optimum and supra-optimum environments would be a more feasible option.

In spite of the observed inconsistencies in QTL detection of NUE, further requirements of fine mapping to identify individual QTLs and confirmation of the identified QTLs using field trials, our result can be used in further targeted approaches in breeding efforts like marker-assisted selection or gene cloning programmes of barley breeding.