Publications

A genome-based approached to improving barley for the malting and distilling industries.

HGCA PROJECT REPORT 264

A genome-based approach to improving barley for the malting and distilling industries

by

R C Meyer1, J S Swanston1, G R Young1, P E Lawrence1, A Bertie1, J Ritchie1, A Wilson1, J Brosnan2, S Pearson2, T Bringhurst2, G Steele2, P R Aldis3, M Field3, T Jolliffe3, W Powell1 & W T B Thomas1

1Scottish Crop Research Institute, Invergowrie, Dundee DD2 5DA
2 Whisky Research Institute, The Robertson Trust Building, Research Park North, Edinburgh EH14 4AP
3Advanta Seeds UK, Station Road, Docking, Kings Lynn, Norfolk PE31 8LS

OCTOBER 2001

Abstract

The distilling industry utilises around 25% of the UK malting barley crop, equivalent to over 60% of the Scottish malting barley crop. Scotch Whisky is by far the leading export in the Food and Drink sector and is currently the UK's fifth highest export earner. Whilst the distilling industry purchases specified barley cultivars, little or no testing for specific distilling requirements is carried out by breeders or the testing authorities. Distillers therefore have little knowledge about the quality of newly recommended cultivars other than their extract levels. The major requirement of the distilling industry is to produce the maximum amount of spirit per tonne of malt as efficiently as possible. Not all of the components extracted during mashing from a malt are fermentable by yeast so spirit yield is the product of extract and its fermentability. Little was known about fermentability and its relationship to other malting quality characters so the aims of this project were to: understand the genetical and environmental control of fermentability and its relationship to spirit yield and other characters; identify genetic markers that could be used to select for the character without the need for expensive malting quality and fermentability assays; and initiate a programme to produce barley lines for specific use in the distilling industry.

We clearly showed that while fermentability was genetically controlled, it was also liable to be affected by environmental variation. Some fermentability genes were inversely related to some affecting extract so that increasing fermentability without careful selection could actually reduce extract and, therefore, spirit yield. We identified regions of barley chromosomes responsible for the genetic control of fermentability and a range of other malting quality characters. In doing so, we also identified genetic markers that could be used to indirectly select for these characteristics. Plant breeders could use such markers to eliminate poor malting quality lines before initiating an expensive trialling and testing scheme. An example of this was a marker linked to a gene controlling non-production of epi-heterodendrin, a characteristic required by some distilleries and found in cultivars such as Maresi, Delibes, Derkado and Decanter. This was a single major gene and the marker was found to be effective in discriminating between producers and non-producers of epiheterodendrin. Fermentability was a more complex character and a number of genes were found to be controlling the character. Markers were used to select for one gene found to have a major influence on fermentability but this was of limited effectiveness in the absence of selection for the other genes. The gene could also have a different effect when the genetic background is changed. Nevertheless, we were able to develop some lines of potential commercial value within the project and also identify more specific targets that are highly likely to lead to improved barley cultivars for use in the distilling industry.

Summary Report

Background

Up to 60% of the Scottish barley crop is used in malting for brewing and distilling. The distilling industry alone uses some 500,000 tonnes per annum with the total malt purchases in Scotland exceeding 800,000 tonnes. Scotch Whisky is the fifth largest British export and the leading food and drink commodity, earning over £2 billion per annum. Malt whisky can only be made from malted barley and is the premium end of the market. High spirit yield is probably the main quality requirement of the malt whisky distilling industry, because a 1% increase in spirit yield would lead to a saving of approximately £1.1 million in distilling production costs. Spirit yield is the product of hot water extract, i.e. the total soluble component following malting, and the fermentability of the extract since not all solubilised components are fermentable. The peak level of fermentability is achieved earlier in the malting process than the peak level of extract and malting has to be optimised to produce the maximum spirit yield. Under certain conditions a breakdown product of epi-heterodendrin, a glycosidic nitrile produced in germinating barley, can react with ethanol, catalysed by copper in stills, to produce the putative carcinogen ethyl carbamate (urethane). Barley cultivars that do not produce epi-heterodendrin are essential in grain whisky distilling and also in some malt whisky distilleries.

The development of genetic finger-printing techniques in human genetics has led to applications of the various types of molecular markers, especially in the rapid creation of genetic maps of an organism. The advantage of such maps is that regions controlling complex characters such as malting quality can be identified as Quantitative Trait Loci (QTL). This knowledge can then be applied in a targeted manner to improve plant characters for a specific end-user need. Fermentability is, as noted above, a key character for the distilling industry but its analysis is difficult to carry out in plant breeding and genetical studies. It is, however, an ideal character for exploiting molecular marker methods in plant breeding for a specific end-user requirement and is the basis for the project being reported here. The aims of the project were:

Determine the genetic control of fermentability and spirit yield and their relationships to other characters
Identify molecular markers linked to genes controlling fermentability, hot water extract and spirit yield
Combine genes for high spirit yield in a spring barley genotype suitable for Northern Britain.
We collected genetic marker and malting quality data from a spring barley population of random inbred lines that was constructed from a cross between commercially relevant parents. The parents did, however, exhibit contrasting combinations of fermentability and hot water extract and so maximised our chances of revealing QTLs for the former that could be used to improve spirit yield.

Materials & Methods

Mapping Population

Random inbred lines were produced by doubled haploidy from the F1 of a spring barley cross between the genotypes Derkado and B83-12/21/5 to achieve the first two objectives of the project. Derkado had good malting quality and was one of the main Scottish cultivars in the 1990's, principally because it was a non-producer of epi-heterodendrin, and B83-12/21/5 was a breeding line from the Scottish Crop Research Institute (SCRI). The whole population, together with the parents and some controls, was grown in trials at SCRI from 1995 to 1997 inclusive and at a site near Sleaford, Lincs, UK in 1996 and 1997. Each trial was sown in plots at a normal commercial density, received a typical fertiliser regime, and was kept free of foliar pathogens by the application of fungicides. The fraction passing over a 2.5mm sieve from each plot was retained for phenotypic analysis of quality characters. Adverse weather conditions delayed harvest of the 1997 trial near Sleaford and there was considerable pre-harvest sprouting in a number of samples. No malting analyses were therefore carried out on this trial, leaving four trials for analysis.

A large number of molecular markers, the two major dwarfing genes sdw1 and ari-eGP, the mildew resistance gene mlo, and a gene controlling the non-production of epi-heterodendrin (eph) were used to generate a genetic map that covered most of the barley genome. A range of malting quality characters was also scored on the population, namely hot water extract, fermentability, predicted spirit yield (PSY), grain nitrogen content, soluble nitrogen content of the malt, soluble nitrogen ratio, grain b -Glucan content and quantitative production of epi-heterodendrin. In addition, malt samples from a subset of lines from two trials were used to estimate the wort contents of the sugars glucose, sucrose, maltose and malto-triose. These data were used to study the genetics of each character (Objective 1) and, when combined with the marker data, to identify QTLs controlling each character (Objective 2).

Breeding Population

We developed a breeding population for the third objective of the project. We selected 8 lines, on the basis of their phenotypic performance, as donors of high fermentability to initiate a progamme to produce first backcross (BC1) inbred lines. The cultivar Landlord and two SCRI breeding lines, B91-47/22 and B91-99/15, were chosen as recipient parents, as there was scope to improve the fermentability of each. We found, however, that our target QTL was closely linked in coupling with the ari-eGP dwarfing gene and in repulsion to a hot water extract QTL at the ari-e locus. We would therefore need to produce recombinants between the fermentability QTL and the dwarfing gene to develop a successful cultivar within the project. For every 100 BC1DH lines that we produced, we would expect an average of 5 to be recombinants and we would therefore need to develop a very large population to generate a sufficiently large number of desired recombinants within the project. We therefore changed our strategy to a more random one by testing all the BC1DH lines that we developed. Time constraints limited the development of the breeding population so that seed was available from only 255 BC1DH plants in time for sowing in trials in 1999. Selections based on field observations made on the trials were multiplied over winter in New Zealand and returned for large plot (7m2) trials at commercial density with and without fungicide at SCRI and fungicide treated trials near Sleaford and Docking in 2000. Cleaned and sieved samples from the plots were retained for analysis of the malting quality characters hot water extract, fermentability, PSY, grain nitrogen content, soluble nitrogen content of the malt, soluble nitrogen ratio and grain b -Glucan content.

The genetic fingerprints of the BC1DH lines entered into trials at SCRI were established by surveying them with 44 previously mapped Simple Sequence Repeat (SSR) markers, which were selected to sample the whole barley genome as well as the target QTL. In addition, allelic differences at the sdw1 and ari-eGP loci were established from observations of the juvenile growth habits of the plots. As well as developing lines of potential commercial merit, we wished to detect whether or not the donor QTL chromosomal segment altered the expression of fermentability in the recipient. We coded all the genotypic data as being either donor or recipient in origin and compared the means of the different genotypes observed in the target region. We also used regression analysis to identify markers that acted together in statistically significant associations with the characters and compared the results to those obtained from the mapping population.

Validation of laboratory tests

An essential question that this project sought to answer was the relevance of the laboratory measures to commercial practice. This applied particularly to the measures of fermentability and epi-heterodendrin. The problem is that methodology based upon commercial practice is resource consuming and cannot be applied to a large number of samples and certainly not on a scale large enough to conduct detailed genetic studies. We therefore selected a stratified set of malt samples for high gravity spirit yield (HGSY) analysis by the Scotch Whisky Research Institute (SWRI). This test gives an estimate of the likely spirit yield under distillery conditions. As the malts for both PSY and HGSY had been prepared under the same conditions, the two measures can be compared to determine the value of PSY in predicting spirit yield under distillery conditions. This test was applied to samples from both the mapping and the breeding populations. Validation of the measures of epi-heterodendrin was also carried out by SWRI using the standard distillery method.

Results

Genetics of the traits

Derkado was generally the better parent for most of the quality characters but B83-12/21/5 had a greater fermentability. In general, DH lines that transgressed, or equalled, the parental means were apparent for all characters, indicating the presence of useful alleles in both parents that potentially could be recombined to produce superior inbred lines. There was highly significant genetic variation for all the characters apart from the wort sugar data, which indicates that most of the characters should be responsive to selection. The high amount of genetic variation found for epi-hetrodendrin reflects the segregation of the major gene controlling production of the compound but the figure is still high when the effects of the gene are excluded by restricting analysis to lines without the eph gene. Apart from glucose, there was little indication of genetic variation for the wort sugars but this may reflect the fact that there was not a proper error to test for genetic effects in the project, which may therefore have been obscured by interactions from contrasting sites.

The correlations between the means of the characters show that extract is the major determinant of PSY although fermentability does have a small but significant positive correlation with the character (Table 1). Selection for increased fermentability could improve spirit yield but would need to be applied cautiously due to its higher but negative correlation with hot water extract. QTL mapping of the two traits would identify a suitable locus for selection. The correlations of the wort sugars with hot water extract are as expected but, with the exception of glucose, the wort sugars are not correlated with fermentability. The negative correlation of glucose with fermentability is surprising but could be an indication of over-modification, particularly as there is evidence of a positive correlation between glucose and soluble nitrogen ratio.

Table 1. Correlations between 11 malting quality characters measured on random DHs from Derkado x B83-12/21/5. Figures in bold are significant at P<0.05.

Ferment HWE PSY N SNR

b -
Glucan1

EPH2 Gluclose3 Sucrose3 Maltose3 HWE -0.41 PSY 0.17 0.75 N 0.27 -0.37 -0.21 SNR -0.49 0.49 0.21 -0.71 b -
Glucan1 0.25 -0.09 0.09 0.25 -019 EPH2 0.09 -0.06 0.00 0.10 -0.02 0.17 Gluclose3 -0.57 0.46 0.20 -0.14 0.32 -0.16 -0.03 Sucrose3 -0.01 0.38 0.43 -0.14 0.37 -0.09 -0.06 0.31 Maltose3 0.11 0.31 0.36 -0.06 0.11 -0.08 0.12 0.33 0.44 M-Tiose3 0.05 0.32 0.35 -0.06 0.19 -0.07 -0.03 0.39 0.51 0.75

1 Based on 3 sites only - 1995 trial not measured

2 Based on 1995 and 1997 sites only

3 Based on 98 lines from 1996 Sleaford and 1997 trials

There was a significant positive correlation between PSY and HGSY (r = 0.52; P<0.01), indicating that the laboratory test gives a good estimation of distillery performance. Four lines performed poorly in the predicted spirit yield test but much better under the high gravity spirit yield test. However they all had low hot water extracts and soluble nitrogen ratios and were therefore under modified. The high gravity test uses much more rigorous extraction than the laboratory test and would therefore solubilise more material. The key point, however, is that lines that performed well under the laboratory test also gave a high HGSY, so positive selection for PSY will be effective. All the lines identified as non-producers of epi-heterodendrin gave very low levels of glycosidic nitriles in the spirit. When these were excluded, however, the agreement between epi-heterodendrin and glycosidic nitriles was non-significant (r = 0.1; P >0.05). This means that while the laboratory test can be used to identify non-producers, it cannot reliably differentiate between the relative amounts of glycosidic nitriles likely to be produced by genotypes without the eph gene.

Mapping of traits

Eight of the most significant QTLs for each character were associated with two currently important major-genes, sdw1 on 3H and mlo on 4H (Figure 1). In both cases, no markers between the major-genes and the QTL peaks were identified, so there was no opportunity to use marker-assisted selection to identify desirable recombinants. This would be particularly desirable in the case of mlo, where the resistant allele was associated with a reduced level of the mono- and disaccharide sugars measured in the study and hence a reduction in hot water extract and PSY. The most significant QTL for fermentability was located in the region of another major-gene, the ari-eGP dwarfing gene on 5H, but linked in repulsion to the second most significant QTL for hot water extract. There were, however, markers in this region that could be used to select recombinants with increasing alleles of both. The most significant QTLs for the other two characters are located on 2H and 7H but no known major genes were segregating in these regions of the genome. For six of the 11 characters studied, the most significant QTL accounted for over 10% of the phenotypic variation. Where characters had low amounts of genetic variation (Table 2), the percentage of genetic variation accounted for by the most significant QTL was, not surprisingly, lower. The SSRs Bmag323 and Bmag337 flank the fermentability QTL on 5H and the Derkado QTL allele accounts for just under 0.5% decrease in %fermentability. This translates, however, into an extra 3 litres of PSY that, if applied over the whole malt whisky industry, translates into an extra 1 million bottles production annually.

The other major result of the mapping was the location of the locus controlling the non-production of epi-heterodendrin, eph, on chromosome 1H, between the SSR Bmac213 and the AFLP P40M38a. Both markers were over 5cM from eph and could be used in marker-assisted selection for the gene but there would be an expected error rate of over 6%. The cultivar Cooper, a producer of epi-heterodendrin, is for example linked with the SSR allele associated with non-production in Derkado.

Molecular Breeding

Without selection, we would expect the average donor genome content to be 25%, whereas it was over 30%, but 67% of the BC1 plants possessed the donor target region and might have increased the donor genome contribution. Additionally, the doubled haploid process may have selected some donor genes increasing green plant production, which is possible as the donors were the products of a doubled haploid programme.

Data from the 2000 trials were used to evaluate the merits of the 135 lines in the breeding population and to determine the effect of the transferred region in another genetic background. Despite delaying malting until dormancy had been broken, the samples from the SCRI trials generally malted poorly with very low extracts leading to abnormal overall means values for many of the malting characters (Table 2). Landlord performed poorly as it and many other samples were under-modified. In contrast, the parents of the donors, Derkado and B83-12/21/5, malted normally and there were some lines that malted considerably better than Landlord.

Table 2. Summary statistics of results from two field trials of 135 BC1DH lines and controls grown at SCRI in 2000. Numbers in Bold are significantly different from Landlord.

Character
BC1DHs Landlord Derkado

B83-
12/21/5

Minimum Mean Max SED Yield(t/ha) 5.02 4.76 5.16 4.07 4.84 5.96 0.25 Head(days) 19.4 23.4 21.6 15 18.7 28.5 1.5 Height(cm) 61.8 64.3 60.5 38.3 56.9 75.3 2.8 Ext(%)-NIR 81.5 82.8 78 76.5 80 83.3 1 Grain Nitrogen(%) 1.25 1.34 1.38 1.27 1.4 1.66 0.08 HWE(° L/kg) 261 337 325 210 268 317 21 Fermentability(%) 82.4 78 80.9 77.9 80.9 83.7 1.6 PSY(l/t) 333 412 410 267 341 406 29 Soluble Nitrogen(%) 0.406 0.713 0.699 0.386 0.514 0.665 0.068 SNR(%) 32.4 56.5 51.9 26.6 382 53.4 5.7 Viscosity-NIR(sec) 12.6 12.1 18.6 8.9 15.6 20.2 1.2 Wort Viscosity(cP) 1.47 1.39 1.44 1.36 1.45 1.54 0.05
There was significant genetical variation for all the characters measured on the 2000 trials apart from fermentability, which had probably been adversely affected by the uneven modification of samples. Apart from soluble nitrogen, the mean of the BC1DHs did not differ significantly from Landlord, confirming that their general behaviour reflected a Landlord background. In contrast, the results from the NIR analysis all showed that the mean of the population was significantly worse than Landlord as grain nitrogen and viscosity was higher and extract lower. Seven lines were selected from their yield and agronomic merit in the 2000 trials for further trialling in 2001. One line showed outstanding yield potential over all three trials and three others had the same or slightly better yields than Optic. The effect of the poor micro-malting performance of the samples from the 2000 SCRI trials can be seen as Landlord and two other lines had very low extracts. None of the other lines had very good extracts although some did have higher fermentabilities.

The majority (103) of the lines can be classified as having the parental genotype in the target region of 5H, i.e. donor genotype with high fermentability and low extract (HF,LE) or recipient genotype with low fermentability and high extract (LF,HE). The remaining 32 lines had recombination events in the 16 cM between Bmag323 and Bmag357 and can be classified into all four combinations of high and low extract and fermentability, but the numbers in some groups are too low to enable accurate comparisons to be made. We therefore cannot test whether we have successfully generated recombinants with high fermentability and high extract QTLs in the target region of 5H but we can amalgamate the data from all 135 lines into separate comparisons of donor and recipient differences at the fermentability and extract QTLs.

The effect of the donor segment can be seen in the summaries of results presented in Table 3. Donor alleles at both the fermentability and extract QTLs have very similar effects upon all the malting characters that were measured on the 2000 trials. The similarity of the response is to be expected as the two QTLs are closely linked. The results are generally consistent for each character as well. Donor alleles decrease extract, whether measured after micro-malting or predicted by NIR analysis. They also produce a slight increase in the grain nitrogen content and a decrease in the soluble nitrogen content of the wort, resulting in a reduced soluble nitrogen ratio. The differences were most pronounced at the SCRI sites and some caution should be exercised in making firm conclusions from the results due to the poor malting performance of samples in those trials. The soluble nitrogen effects were also apparent at the Docking site, particularly in the contrast for the HWE QTL. This QTL is co-located with the ari-eGP dwarfing gene and the results suggest that use of the gene leads to problems in protein breakdown.

Table 3. Differences between means of BC1DH lines grown in three trials in 2000 and classified according to whether they possessed donor or recipient alleles at the fermentability and HWE QTLs in the target region of 5H. Differences are expressed as donor minus recipient means and those in bold type are significantly different.

Site QTL
Mean NIR Ext Ferment HWE PSY Grain N Sol N SNR Viscosity Docking Ferment -0.1 0.5 -4 -2 0.01 -0.016 -1.0 -0.2 Docking HWE -0.2 0.8 -6 -3 0.01 -0.029 -1.8 0.1 SCRI - F Ferment -1.1 0.0 -21 -27 0.15 -0.036 -5.8 1.1 SCRI - F HWE -1.1 0.3 -26 -33 0.15 -0.048 -7.2 1.0 SCRI + F Ferment -1.0 -0.5 -28 -37 0.13 -0.047 -7.3 1.0 SCRI + F HWE -1.1 -0.4 -33 -44 0.14 -0.052 -8.0 1.1
Allelic differences at the fermentability QTL did show an increase due to donor alleles from the results from the Docking site but the effect was not significant. Donor alleles at the HWE QTL not only significantly reduced extract but also significantly increased fermentability and the same pattern can be seen in the results from the SCRI untreated trial, although the increase was not significant. At the SCRI treated trial, donor alleles actually reduced fermentability but, as malting performance was noticeably worse in the treated than the untreated trial, less credibility can be given to this finding. The general effect appears to be that the fermentability and extract QTLs are more closely associated than was apparent from the mapping study but we would need more recombinants and malting results from Scottish trials to verify this finding. The mapping trials did show that fermentability was subject to considerable genotype x environment interactions and that the fermentability QTL was not effective in the trial grown near Sleaford.

Regression analysis of the phenotypic and genotypic data collected on the means from the SCRI and the Docking trials revealed a number of significant associations. Apart from fermentability at both SCRI sites, and soluble nitrogen and SNR at the untreated site, the marker associations at SCRI accounted for more variation than at Docking. This probably reflected the range of variation found in the characters measured after micro-malting at SCRI. Each character was associated with at least one marker and there were 25 cases where results agreed between at least two of the trials. There were six notable clusters of associations at HVM54 (2H), Bmag225 (3H), HvBAMY (4H), ari-e (5H) and Bmac273a (7H). In each case, the increasing allele associations were generally either recipient or donor, which would be consistent with a particular genomic region affecting a number of malting quality parameters. The association of the ari-eGP allele with a decrease in hot water extract is consistent with the findings from the mapping study but contrasting results for the target fermentability locus were obtained from the trials. Whilst donor alleles at ari-e were significantly associated with an increase in fermentability from results obtained from the Docking site, donor alleles at a nearby locus (Bmag323) were significantly associated with a decrease in fermentability at the SCRI treated site in 2000. No significant associations of fermentability with markers in the target region were detected from the results of the untreated trial at SCRI by either multiple or single marker regression. Donor alleles at ari-e did produce an increase in fermentability but the effect was far from significant. Other loci found to be affecting hot water extract in the mapping population on 1H, 3H, 4H and 7H are in similar regions to those found in the BC1DH population at Bmag211, Bmag225, HVM67 and Bmac273a respectively. In addition to the target region, another region affecting fermentability detected on 5H in the mapping population corresponds to that detected at Bmag222 in the untreated BC1DH trial at SCRI. The other character for which some common regions can be observed is SNR, where two regions on 3H and one on 7H detected in the mapping population correspond to HvLTPPB, Bmag013 and Bmac273a respectively.

Conclusions and End-User Relevance

We have clearly shown that there is genetic variation for fermentability but phenotypic selection for the character is likely to be difficult. Although fermentability is negatively correlated with hot water extract the correlation is not great and it should be possible to manipulate both characters to increase Predicted Spirit Yield. The results obtained from our small-scale laboratory tests for fermentability were shown within the project to be relevant to commercial distilleries, despite a large range in the distilling potential of the lines studied. All these findings mean that targeted breeding of cultivars specifically adapted for use in the Scotch Whisky industry can be undertaken.

The mlo mildew resistance gene has proved to be durable over 20 years of commercial deployment in spring barley cultivars in the UK and is the main resistance gene found in current recommended cultivars. Within this study, we found that the mlo resistance gene was associated with a reduction in malting quality characters, especially wort sugar content leading to an overall reduction in hot water extract. This may be due to adverse genetic linkages, in which case it would be possible to recombine the resistance gene with the allele improving malting performance. It is clear that more detailed studies of this region of the genome are required in order for breeders to devise an appropriate strategy. Meanwhile, genes from other segments of the genome will have to be incorporated to overcome the deleterious effects of the mlo gene.

We expected wort sugar levels to have a closer relationship with extract and/or fermentability than our results showed. The correlations between extract and the wort sugars were all significant and positive but less than 0.5, thus accounting for little of the variation in the character. There was some indication from multiple regression studies that both glucose and sucrose act together to influence levels of extract and that glucose and maltose also act together to influence fermentability. In the last case, glucose levels were negatively associated with fermentability and may reflect a general over-modification of the micro-malts in the mapping trials. The two most significant extract QTLs were, however, located in the same region as the two most significant QTLs for glucose and it may well be that the analysis in the current study was not sufficiently detailed to reveal closer associations.

Our aim was to identify a fermentability QTL and test its utility by transferring it into another genetic background but our finding that the target QTL was linked in repulsion to a hot water extract QTL meant that we had to generate recombinants to produce useful lines. We planned to generate recombinant backcross inbred lines through a combination of genotyping and doubled haploidy that would enable us to rapidly develop useful germplasm but also test our strategy. The finding of an undesirable linkage and time constraints meant that we were unable to produce sufficient lines to give us a realistic chance of achieving our aim. We were able to generate and test enough lines, however, to compare the effect of donor alleles against the recipient alleles in the target region of the genome. The results were inconclusive and interpretation was hampered by poor malting performance of samples from the 2000 SCRI trials. The one trial that malted normally was grown outside the target environment but did produce evidence of increased fermentability due to the presence of donor alleles in the target region and there was some indication of corroborating evidence from the untreated trial grown at SCRI. The data did indicate, however, that the fermentability QTL might be associated with the ari-eGP dwarfing gene. Further work is necessary to establish whether or not this is so, as deleterious effects of the dwarfing gene, such as high screenings and reduced extract, mean that it is no longer viable in a commercial cultivar.

We attempted to maximise our chances of detecting fermentability QTLs of large effect by using parents that were relatively diverse, but adapted to the target environment. The most significant QTL accounted for just 6% of the phenotypic variation in fermentability, however, and there were a number of other possible loci with smaller effects. To detect such a QTL in another genetic background requires most of the other increasing loci to be present and using marker assisted selection for just the target QTL means that many of the other increasing alleles are eliminated by chance. This is not just a problem for the current project but also applies to other characters of low heritability with a number of controlling genes. In such cases, there is no alternative but to generate large populations, use marker-assisted selection to form a pool of 'improved' lines and rely on phenotypic selection to pick out the best lines.

Whilst the cultivar Golden Promise carried the ari-eGP gene and was used in great quantities by maltsters and distillers, it was never regarded as a top-class malting quality cultivar. The fermentability QTL studied in this project either represents the action of an anonymous gene or ari-eGP. There is evidence that the gamma-ray mutation of Maythorpe to produce Golden Promise resulted in an increased rate of modification. Such a gene, taken from a moderately poor malting background, may lead to excessive modification in a good malting quality background and this is a possible weakness of the anonymous approach used in the current project.

The abnormal malting performance of the samples from the SCRI trials in 2000 was un-expected as we did not carry out the micro-malting until 5 months after harvest, when we expected dormancy to have been broken. It is a definite genetic effect as replicate samples from a randomised trial perform similarly and appears to be due to the use of Landlord as the main recipient parent. Whilst Landlord performed poorly some other controls, notably Derkado and Optic, malted normally and gave high extracts but Chariot, a parent of Landlord, also did not micro-malt well. Some environmental factors must have induced some water sensitivity at SCRI in 2000, as we found marked water sensitivity in some samples 10 months after harvest. From a stratified set of samples, germination figures in 8ml of water were very highly correlated with hot water extract. As both Landlord and Chariot are very susceptible to Ramularia infection, there is the possibility that factors associated with the disease affect malting quality, as it was present in the SCRI trials. This is a problem associated with backcrossing as the recipient parent can be outclassed by the time of release of a new genotype if one chooses an established cultivar as a parent. Choosing a 'high-flyer', such as Landlord, from early trials information can mean that undesirable effects, such as water sensitivity, become apparent after development of new lines is well advanced. The practice of adaptive backcrossing, in which one changes the parent at each stage is the most practical way to avoid these two problems of backcrossing. If one couples it with marker-assisted selection, then previously mapped SSRs would be of great value as their multi-allelic nature means one can identify not only donor alleles but also the different recipient alleles.

The markers that we identified as flanking eph could potentially be used to eliminate epi-heterodendrin producers from distillery malts, either by plant breeders or the testing authorities and thus avoid any future problems with statutory ethyl carbamate tolerance levels. The SSRs linked to eph were close enough to the gene to enable their use in marker assisted selection but were not close enough to be diagnostic. One would need to genotype the parents of each cross and know whether or not they were epi-heterodendrin producers before deploying marker-assisted selection. There is a need to develop diagnostic markers for the character so that they can be deployed by the distilling industry. We managed to make some advances during the project but further work is necessary to ideally locate a diagnostic marker within the gene itself.

Our results have shown that extract is the major determinant of predicted spirit yield but opportunities for further improvement in extract appear to be limited, unless hull-less barley is developed for use in malting. The removal of the husk would provide a quantum leap in extract levels but, especially for the distilling industry, some degree of husk retention or mixing with an appropriate husked variety would be required for filtration. Research would also need to be carried out on the agronomic and financial implications of utilising naked barley as there would be an immediate yield loss.

There is opportunity to manipulate natural variation in fermentability, however, and the targeting of specific genes of known function may well be a better means of improving barley for use in distilling in the short term. For instance, natural variants of b -amylase with improved thermostability, which are not found in European spring barley, may improve fermentability. Another potential approach would be to explore natural variation to ensure that limit dextrinase remains bound during mashing but is released during fermentation and would therefore increase fermentability. Putting such variants of these two genes together may well provide a further means of improvement. Results from functional genomics programmes could provide better overall understanding of the genetics of complex traits such as malting quality and eliminate some of the problems associated with the single gene approach that we adopted within this project. With functional genomics, one can attempt to establish how various candidate genes interact to produce a given phenotype. After gathering such information from a range of cultivars and associating it with malting quality data, it will be possible to identify targets to manipulate in order to improve performance for specific malting attributes.

This project has increased the exchange of information between geneticists, breeders and end-users, thus vastly improving each group's mutual understanding of the potential applications of molecular biological methods. We wish to develop these relationships, not only into future research projects, but also to develop tangible benefits all the way along the supply chain to the end-user. We see such networks as being essential to not only a healthy domestic market but also maintaining and extending the export market.

HGCA Project Number: 1573
Price: £4.25

Related Publications

Document download

View a printer friendly version of this publication

Download this publication PDF

PR264.pdf

File size: 279KB
Project_report_thumb