GROWING WHEAT FOR HIGH ALCOHOL YIELD –HOMOGENEOUS AND HETEROGENEOUS APPROACHES - Soybean and Wheat Crops: Growth, Fertilization, and Yield - page 168

Selection at early stages in a breeding programme is rather more problematic. Several
authors have observed a strong negative correlation between alcohol yield and grain nitrogen
content [49], [70], [73], but the latter is unsuited as a means of selecting the best genotypes
due to the strong influence of environment on its expression [70]. In addition, Kindred et al.
[31] noted that a good distilling wheat had a higher starch content and alcohol yield than a
poor one at any given level of grain nitrogen. That data was obtained from a trial grown at a
single site, but with a range of N fertiliser regimes. Similar results were shown when good
and poor distilling varieties were compared over a range of sites and several seasons (Figure
3). Thus while grain nitrogen content contributes to the phenotypic expression of alcohol
yield, other factors determine the differences between varieties in genetic potential.
The extent of grain filling may also be a key contributor to alcohol yield in at least some
of the main distilling varieties. Agu et al. [3] suggested that grain samples with a higher
proportion of small grain tended to give lower yields of alcohol. In that work, samples were
assessed by determining the proportions of grain retained by a series of sieves, indicating that
grain width, rather than length was the determining factor. This supported the findings of
Swanston et al. [70], who proposed an equation to predict alcohol yield from grain nitrogen,
thousand grain weight (TGW) and the ratio of grain length to grain width. Length : width
ratio (L:W), in which high values are indicative of narrow grain, had a negative effect on
alcohol yield. Swanston et al. [70] also showed that L:W was primarily a varietal trait, while
TGW was subject to both varietal and environmental variation. However, while high alcohol
yields were associated with large grain size in the variety Riband [31], other varieties such as
Claire [66] and Glasgow [70] could give good alcohol yields despite having relatively small
grain, so grain dimensions did not always give an accurate indication of alcohol yield when
compared across varieties.
Plant breeding has, traditionally, utilised phenotypic selection i.e. breeders have relied on
characters which they could observe or measure to distinguish between desirable and
undesirable progeny. Such characteristics needed to be sufficiently heritable for selection to
be meaningful, but it was not necessary for their underlying genetic control to be fully
understood. In general, however, it was easier to select, in early generations, for characters
controlled by single genes, such as reduced height resulting from the inclusion of dwarfing or
semi-dwarfing genes. Characters such as yield and processing quality, which had more
complex genetic control, were assessed on populations that had already been reduced by
earlier-generation selection. In addition, to ensure that the contribution of the genotype could
be distinguished in the presence of environmental variation and genotype x environment
interaction, it was necessary to carry out replicated trials over an appropriate range of sites
and seasons. The advent of molecular breeding, based on selection of DNA sequences,
therefore offered 2 major advantages. Firstly, it enabled direct selection of genetic potential,
so could therefore be applied to unreplicated samples not grown together in trial and,
secondly, it could be used in early generations of breeding programmes. This offered the
potential to prioritise characters for selection on the basis of economic importance rather than
ease of phenotypic assessment.
Initially, selection was based on anonymous DNA sequences which were linked either to
individual genes or to genetic factors that contributed to traits controlled by more than one
gene. The latter were termed quantitative trait loci (QTLs). A review of selection for a range
of characters in wheat, including yield, disease resistance and quality parameters, using such
molecular markers, is given by Gupta et al. [23]. These studies tended to reflect priorities