Bioethanol is a promising renewable biofuel and wheat is currently the main candidate asthe feedstock for its production in the UK context. The quality of the numerous varieties ofwheat developed in the past by plant breeders has been well examined in terms of bread, biscuitand pasta producing industries. In general, the end-use quality determination of wheat in termsof alcohol yield is less investigated. This work focused on understanding and predicting thealcohol yield from wheat according to its physical, physicochemical and chemicalcharacteristics. The research ran alongside the GREEN Grain project and utilised its wheatsamples, which consist of a range of wheat varieties, agronomic regimes and growing sitesfrom four harvests years 2005-2008. The combined dataset consists of a diverse range ofchemical, physicochemical and physical characteristics of the GREEN Grain wheats. An initial multivariate analysis (PCA) indicated that the first principal component, whichexplains most of the variability of the wheat characteristics, is related with the classification ofwheat as hard or soft. High alcohol yielding wheats typically have high starch, mealiness andalbumin+globulin fraction, and also low protein, gliadin fraction and hardness. They also havelarger and more spherical kernels. Analysis of Variance (ANOVA) was applied in order to identify differences between thevarieties, the sites and the application or not of N fertiliser. The ANOVA showed that theapplication of N fertiliser increases all the protein components, although it increases the Gliadinand the LMW glutenins more. N fertiliser also yields smaller (TGW, width, depth) and moreelongated kernels. High alcohol yielding varieties tend to be softer with lower protein andlarger and more spherical kernels. This consistent variability allowed prediction of the alcoholyield based on easily measured parameters. The following model, based on the SKCS reportedvalues plus protein, could predict the alcohol yield with an R2 of about 78%:Alcohol yield = 466.62 - 5.07 × Protein - 0.21 × hardness + 11.6 × diameter ±6.94 l/dry tonIt is frequently hypothesised that larger and more rounded kernels produce more alcoholbecause they have a smaller relative amount of the unfermentable outer layers. In an effort totest this hypothesis, the pericarp thicknesses and the crease characteristics of the wheat sampleswere measured. It was found that pericarp thickness and crease dimensions vary with kernelsize, with significant differences between varieties. A physical model was developed thatconsiders these differences and calculates the endosperm to non-endosperm ratio. None of thevariables obtained by the physical model could be related to alcohol yield. The SKCS fundamental data were further analysed in an effort to improve the alcoholyield predictability. It was found that the averaged Crush Response Profiles are morereproducible than the hardness index itself. It was shown that the initial peak does not occurbecause of the "shell" (i.e. the bran layers) as suggested in the literature, but because of thecrease. Examination of the effects of moisture content on the aCRPs showed that their 1stquarter is equivalent to the stress-strain plots of dedicated rheological tests. The remaining partsof the curve relate to the post-failure behaviour of the kernels and with hardness as used incereal science. The aCRP parameters could improve the alcohol yield predictability of theGREEN Grain wheats to an R2 of about 82.3% and a standard error of the regression of6.3 l/dry ton. Further standardisation and calibration with respect to the moisture content and tothe size of the kernels could improve the predictions even further. Textural testing of cereals is constrained by the complexity of the wheat kernel structureand exacerbated by the between-kernel variation. The current work has demonstrated howSKCS data can be interpreted more insightfully in order to improve end-use quality predictions. The aCRP parameters clearly contain rheological information about wheats. Further research toestablish their examination by more standardised methodologies will allow effectiveinvestigation of connections between the rheological properties, chemical characteristics,processing behaviour and end-use quality prediction of wheat.

Wheat grain quality has a complex genetic architecture heavily influenced by the growing environment. Consistency in wheat quality not only affects the efficiency of milling and baking but also the quality of end-use products. The objectives of this study were to 1) analyze the different wheat quality parameters in Recombinant Inbred Lines (RILs) grown under different environments, and 2) to identify Quantitative Trait Loci (QTLs) associated with quality stability in RILs grown under different environments. A set of 180 RILs derived from two spring wheat lines 'Halberd' and 'Len' were grown at Uvalde and College Station TX, in the 2009/2010 growing season and at Chillicothe and College Station TX, in 2010/2011 growing seasons. The experiment was laid out in Randomized Complete Block Design (RCBD) with four replications within each location. Each line was tested for multiple quality traits that included grain hardness, protein content, dough mixing properties and bread baking quality using Single Kernel Characterization System (SKCS), Near-Infrared Reflectance Spectrometry (NIRS) analysis, mixograph and the Sodium Dodecyl Sulfate Sedimentation (SDSS) test. Genetic linkage map construction was carried out with 116 single nucleotide polymorphism (SNP) markers in the RILs. Then composite interval mapping was carried out to identify QTLs associated with quality traits. The SDSS column height was positively correlated across four environments. Similarly, it was found to have significant positive correlation with mixing tolerance and peak time within and also across locations. However, the SDSS was negatively correlated with the hardness index. The protein percent was not significant with any of the quality traits within and across environments. We were able to detect many QTLs for different quality traits but most of them were site specific. Only a few QTLs were consistent across environments. Most of the QTLs for quality traits i.e., SDSS, peak time, mixing tolerance and hardness index were identified on chromosome 1B. We were able to detect overlapped QTLs for SDSS column height and mixing tolerance on chromosome 1B. Furthermore, overlapping QTLs for mixing tolerance and peak time were detected on an unknown chromosome. We also detected overlapping QTLs for hardness index on chromosome 1B. We identified one stable QTL for SDSS column height on chromosome 4B. This QTL was detected based on the coefficient of variation (CV) for SDSS in four different environments.