Introduction

Wild emmer wheat, Triticum dicoccoides, is the tetraploid (AABB) ancestor of modern tetraploid (Triticum durum) and hexaploid (Triticum aestivum) wheat cultivars. Today, its distribution in the wild is generally restricted to mountainous regions in the ‘Fertile Crescent’ but its value in agriculture lies in its genetic variation for sustainability traits, such as the ability to produce reasonable yields on nutrient poor soils, and its resistances to fungal diseases prevalent in such areas (Peng et al. 2000). Thus, T. dicoccoides contains a rich and variable gene pool, providing an important genetic resource for the improvement of modern cultivated varieties of common and durum wheats. The introgression of desirable alleles from the wild relative species of bread wheat has been successfully exploited in breeding for a wide range of agronomic and economically important characteristics. Existing traits found in T. dicoccoides include resistance to powdery mildew (Nevo et al. 1985), stripe rust (Nevo et al. 1986; Fahima et al. 1998), stem rust (Nevo et al. 1991), increased photosynthetic rate (Evans and Dunstone 1970; Carver et al. 1989), higher grain protein (Gerechter-Amitai and Stubbs 1970; Grama et al. 1983) and increased kernel weight (Kushnir and Halloran 1984).

Of the modern UK winter wheat cultivars, Shamrock, a Group 1 premium bread-making variety bred by Advanta Seeds UK, was derived from a cross between a T. dicoccoides derivative (Comp Tig 323-1-3 M) and adapted UK germplasm (CWW 4899/25—Moulin × Monopol). Although having a lower yield potential and slightly softer endosperm than equivalent bread-making wheats, Shamrock exhibits some of the beneficial qualities of wild emmer. It is relatively stable and insensitive to changes in environment or agricultural practices, with consistent performance over a range of conditions. Its hardiness and adaptability means that it performs well when grown with low or no fungicide inputs relative to other varieties (NIAB 2002).

An additional, visually striking characteristic of Shamrock, is the extreme viridescent colour of its leaves and spikes. The bright green colour of Shamrock is unique among modern UK varieties. Shamrock also has an absence of glaucosity, again a rare phenotype for cultivated wheat. Glaucosity is described as a white, waxy substance that coats the leaves and stems of plants. Non-glaucous variants are prevalent in wild emmer, but are rare in cultivated emmer, durum and common wheat. Early studies demonstrated that non-glaucousness is controlled by a single dominant gene W1, which acts as an epistatic inhibitor to a glaucousness gene Iw1, on the short arm of chromosome 2B (Jenson and Driscoll 1962, Driscoll 1966, Tsunewaki and Ebana 1999). Furthermore, there is a homoeologous location of both the wax producing and inhibitor genes (W2 and Iw2) on chromosome 2DS.

The viridescent appearance of Shamrock is a unique characteristic for UK wheats, and thus it is of interest to study the inheritance of this trait and determine whether it contributes to the environmental stability and yield performance of this variety. To investigate this, a doubled haploid population from the cross of Shamrock × Shango showing segregation for the trait, was examined. Shango is a Group 1–2 quality winter wheat variety bred by Plant Breeders International Cambridge (PBIC) suitable for growing in northern France. The population was mapped using genetic markers, and grown in large scale field experiments over 3 years and several environments. The data from the field performance of the doubled haploid lines was subjected to major gene and QTL analysis to study the inheritance of viridescence and possible pleiotropic influences on productivity traits.

Materials and methods

Mapping population and map development

A recombinant doubled haploid population, Shamrock × Shango, was jointly developed by Syngenta and the John Innes Centre. Syngenta generated 76 lines of the population, with an additional 38 lines developed at the John Innes Centre, giving a total of 114 lines. Due to seed availability at the time of genotyping and the removal of lines showing heterozygous or non-parental alleles, 87 lines were genotyped in total. The doubled haploid lines were created using the standard wheat x maize technique from F1 plants (Laurie and Reymondie 1991).

The framework genetic map was primarily developed using publicly available single sequence repeat (SSR) markers aiming for a marker density of one every 10–20 cM. Primer sets were used from JIC (psp), IPK Gatersleben (gwm/gdm), Wheat Microsatellite Consortium (wmc), Beltsville Agricultural Research Station (barc) and INRA (cfd/cfa) collections, see GrainGenes website (http://www.wheat.pw.usda.gov/). Targeted markers were selected to provide good genome coverage using published consensus maps (Somers et al. 2004) and for their ease of scoring. DNA fragments were amplified with PCR and run on 5% polyacrylamide gels for separation. The silver staining technique was used to visualize fragments (Bassam et al. 1991). To further improve map density, DNA of the Shamrock × Shango population was sent to Triticarte Pty Ltd, Australia (http://www.triticarte.com.au/) for Diversity Arrays Technology (DArT) genome profiling (Wenzl et al. 2004; Jaccoud et al. 2001) which greatly increased marker density by filling gaps and extending linkage groups.

Joinmap v3.0 was used for genetic linkage map construction, set at the default settings with the Kosambi mapping function. Linkage groups were selected at a minimum LOD of 3 for reliable associations.

Field trials and trait phenotyping

The population was grown in randomised, replicated field experiments at Church Farm, Norwich, UK, over two seasons, 2004/2005 and 2005/2006. The lines were sown in large-scale yield plots (1 × 6 m) and standard farm pesticide and fertiliser applications were made to reproduce commercial practise. Additional trials were sown in conjunction with Syngenta in 2004/2005, and RAGT Seeds in 2003/2004 and 2004/2005, at sites across the UK.

A simple visual assessment was made of the viridescent phenotype, with the population segregating 42–45, within statistical probability for a 1:1 ratio expected for a single gene controlling the trait (Fig. 1). Segregation in levels of plant glaucosity was also observed, with the ‘Shamrock viridescent’ phenotype exhibiting absence or very low levels.

Fig. 1
figure 1

Classification of the Shamrock ‘viridescent’ phenotype. The bright green, non-glaucous Shamrock plots are easily identifiable from the waxy Shango types

Full size image

In the first growing season, observations of days to full plant senescence indicated that the viridescent genotypes also expressed prolonged green leaf area or “stay-green”. It was anticipated that the stay-green effect would enable continued photosynthesis during the grain filling period and consequently have a positive effect on yield. Flowering and full plant senescence dates were therefore recorded during the following year. Flowering date being assessed as the day in which ¾ of the plot had reached ¾ of inflorescence emerged, and full plant senescence as the day where a loss of chlorophyll was observed on all leaves and peduncle of the main shoots for ¾ of the plot. The grain filling period was then calculated for each plot by subtracting the number of days from drilling to flowering, from the number of days to full plant senescence, thereby normalizing each line irrespective of their flowering date.

A major interest in the viridescence trait was to investigate the pleiotropic effects on performance and productivity. Consequently, yield components, such as crop height, plant biomass, spikelet number, spike yield, 1,000 grain weight, harvest index and grain number, as well as plot yield, were recorded on the John Innes Centre trials.

Statistical and QTL analysis

Minitab v13.1 was used for preliminary data exploration. All phenotypic data was initially subject to one-way analysis of variance (ANOVA) to check for heritable trait variation, before QTL Analysis. The program QTL Café (http://www.biosciences.bham.ac.uk/labs/kearsey) was used for QTL discovery, using single marker ANOVA, marker regression and interval mapping approaches. Additionally, composite interval mapping was performed using Windows QTL Cartographer v2.5 for confirmation of the interval mapping results.

Results

Genetic linkage map

The Shamrock × Shango map was developed using 87 DH lines. A total of 489 SSR markers were screened, 215 of which were polymorphic. A further 233 reliable markers were obtained through DArT analysis. All genotypic data were subject to linkage analysis and the final map contained 263 loci covering 1,337 cM, after removal of co-segregating and unlinked markers. This compares favourably with previous mapping studies, but also suggests that the parental lines contain some regions identical by descent. With the viridescent trait showing a 1:1 segregation pattern, it was possible to map this as a major gene to a precise chromosomal location. The gene was mapped to the short arm of chromosome 2B, close to microsatellite marker Xgwm614, Fig. 2. The gene was temporarily designated with the gene symbol Vir.

Fig. 2
figure 2

Genetic linkage map of chromosome 2B for the Shamrock × Shango population (distances are in centimorgans from the most distal marker in 2BS)

Full size image

Phenotypic/QTL analysis

Initial observations of a stay-green effect from the Shamrock lines in 2005 were confirmed with the quantitative scores taken the following year. Despite the parents showing a similar period to full plant senescence, there was transgressive segregation within the population suggesting that a combination of genes from each parent contributed to the effect. The effect of the Vir genotype was observed when lines were classified by their respective alleles at the Vir locus (Fig. 3a). A clear distinction was seen, with lines containing Vir showing an increase in days to full senescence, essentially an increased growth period.

Fig. 3
figure 3

Trait distributions for the Shamrock × Shango population, classified by their respective alleles at the Vir locus for senescence date, Church Farm 2006 (a), grain fill period, Church Farm 2006 (b) and yield at Church Farm 2005 (c). Arrows indicate the means of the parents and the segregating population, classified as having either the viridescent allele (Vir) or the Shango allele (vir)

Full size image

However, when the effects of flowering time were removed in calculating the grain fill period, the effect from Vir was more evident. Of the parents, Shango exhibited the longest period of grain filling; however, as a trend, lines with the Shamrock allele at Vir, achieved a longer grain fill period (Fig. 3b).

A similar pattern was observed for yield. Shamrock has a low yield potential and yielded over 1 t/ha less than Shango (Church Farm 2005). However, when the population was sorted by Vir genotype, the higher yielding lines contained the Shamrock allele (Fig. 3c).

QTL analyses were carried out on all assessed traits. Significant yield increases were associated with the Vir locus for four of the five locations by single marker ANOVA (Table 1). At its greatest, the effect produced a yield increase of 0.7 t/ha at Church Farm 2005. This effect explained 26% of the variation in yield. The additive effects taken from the single marker ANOVA for each of the yield datasets are plotted in Fig. 4a. The yield increases were not significantly associated with any of the yield component traits assessed. An additional yield QTL may also be present near marker Xwmc154 which is proximal to Vir (39 cM, Fig. 4a), although it is more likely that this is a pleiotropic effect of a minor height QTL.

Table 1 Site descriptions of each trial site
Full size table
Fig. 4
figure 4

QTL profiles showing the additive effects from single marker ANOVA of all environments/years, and the mean of all data (a), and the residual MS data for each environment with the mean of all environments from marker regression analysis, showing the most likely position of the yield QTL on chromosome 2B (b)

Full size image

A more accurate QTL designation was determined through additional analysis. Using marker regression (Kearsey and Hyne 1994) the QTL was located regressing all markers simultaneously onto the marker means. With this more stringent test, all environments, apart from JIC 2006, produced a significant yield QTL (>0.05) at the Vir map location (Fig. 4b). All results were further confirmed using composite interval mapping. QTL analysis also highlighted the significant effects on senescence and grain filling period at the same chromosomal location (Tables 1 and 2).

Table 2 QTL output (Single Marker ANOVA) showing positions and levels of significance at each marker locus on chromosome 2B, for yield (5 data sets), days to full senescence and grain fill period (Significance at the 5%, 1%, 0.1% and 0.01% levels are indicated by *, **, *** and ****, respectively)
Full size table

Mapping the T. dicoccoides contribution

To confirm that T. dicoccoides was the source of the Shamrock viridescence, DNA samples of CWW4899-25, Shamrock and Shango were genotyped using the microsatellite marker Xgwm614, which mapped 2 cM from Vir. A sample of the T. dicoccoides parent was unavailable. However, the results confirmed that at this locus, Shamrock contains an alternative allele to CWW4899-25, suggesting that the viridescence was derived from a different source, presumably T. dicoccoides (Fig. 5).

Fig. 5
figure 5

SSR allele profiles indicating that Shamrock has an alternative allele to CWW4899/25, at locus Xgwm614 on chromosome 2B

Full size image

Discussion

This study examined the value of a viridescence gene in Shamrock, in relation to increased stay-green and any subsequent yield benefits. The Vir gene was mapped to the distal end of the short arm of chromosome 2B, 2 cM proximal to microsatellite marker Xgwm614. Previous studies (Driscoll 1966; Tsunewaki and Ebana 1999) indicated that a dominant gene for wax production, W1, is located in this region. The physical properties determining the intense greenness of Vir, whether an increase in chlorophyll, pigment balance within the leaves, altered surface wax structure or complete absence of wax, have yet to be measured. It is possible therefore that Vir is either closely linked to W1, or is the same gene. It should be noted however that historically, glaucousness has been found to have a positive relationship with yield, especially under drought stressed conditions (Richards et al. 1986; Monneveux et al. 2004), whereas in the present study, the converse was found. Additionally, W1 has been located 4 cM distal to Lr16 and Sr23 (http://www.wheat.pw.usda.gov/GG2/index.shtml), which are in turn located a further 12 cM from Xgwm614 (Somers et al. 2004).

QTL analyses confirmed that lines with Vir exhibited a significant delay in leaf senescence and a prolonged grain fill period. The ability to stay-green for an extended grain fill period has obvious benefits, as the plant can photosynthesis for longer periods. The relationship between yield and ‘stay green’ is well documented (Verma et al. 2004; Evans 1993). It was, however, surprising to find such a significant variation in the delay in senescence in the 2006 trial, as the stay green effect might have been inhibited due to abnormally hot weather during the ripening period. It is anticipated that during an average UK summer, variation in the delay of leaf senescence would be even greater for this population. This may also explain why only a minor non-significant yield effect was detected at JIC in 2006. In a drought-stress prone environment (JIC) the stay-green effect, although still significant, was reduced to a degree whereby the impact on the plant productivity was minimal. Viridescent genotypes have previously been classified as having the ability to stay-green (Thomas and Howarth 2000), where a delay in senescence was categorized into five types. Type E behaviour was described as intensely green genotypes with greater levels of pigment content, which stay-green simply because there is more chlorophyll to lose.

Significant yield increases associated with the Vir allele were consistently observed across years and environments. At its strongest the QTL accounted for a quarter of the variation for yield within the population, amounting to an increase of up to 0.7 t/ha. Yield is generally thought of as being accumulated from a large number of genes, whose effects, whether large or small, combine to give the plant its total productivity. In the UK, dramatic yield increases of 1% per year have been produced over the last three decades, and this seems to be continuing (Snape et al. 2007). Fixation of beneficial traits such as the Vir allele in breeding programs, which across all sites gave increases of between 2.4% and 5.6%, provides opportunities to sustain improvements in yield potential.

Despite the large yield increases observed at JIC in 2005, no association was made with any of the yield components assessed. Wheat yields are determined through three major components, tiller number, grain number per spike and grain weight (Garcia del Moral et al. 2003). These components accrue sequentially through the life cycle of the plant. Tiller production occurs during the early growth stages, spike formation is set during the period of rapid canopy extension, and grain weight is assembled from the assimilation of starch and protein supplied from photosynthesis and the re-distribution of reserves during the grain-fill period. Considering Vir is likely to be a physiological effect, and not morphological, it could be assumed that the yield boost is produced during prolonged productivity at grain filling. However no increases in grain size were observed. Further investigation is required to discover the source of the yield benefit from Vir, although it may be that it comes from small increases in multiple components.

To confirm the stay-green and yield QTL on chromosome 2B, a backcrossing program has been initiated whereby the Shamrock viridescence will be transferred into a series of UK recommended wheat varieties. A series of near isogenic lines will be produced and grown in the field to validate the pleiotropic effects of Vir.

The yield and stay-green advantages associated with Vir emphasise the value of introducing desirable, novel alleles, from wild wheat relatives such as T. dicoccoides. This research adds further credence to the utilization of such relatives in breeding programs, as they can provide us with vast amounts of genetic diversity, to be exploited for the continued improvement of cultivated wheat.