H. A. Eagles

Estimation and utilisation of glutenin gene effects from the analysis of unbalanced data from wheat breeding programs

Glutenins are a major determinant of dough characteristics in wheat. These proteins are determined by genes at 6 loci (Glu genes), with multiple alleles present in most breeding programs. This study was conducted to determine whether estimates of allele effects for the important dough rheological characters, maximum dough resistance (Rmax) and dough extensibility, could be determined from aggregated data from southern Australian wheat breeding programs using statistical techniques appropriate for unbalanced data. From a 2-stage analysis of 3226 samples of 1926 cultivars and breeding lines, estimates of Rmax and extensibility effects were obtained, first for the lines, and then for 31 glutenin alleles. Glutenin genes did not determine flour protein concentration, and this character was used as a covariate. Rankings of the estimates of Rmax for the alleles were similar to the relative scores for dough strength reported from previous studies, providing strong evidence that the analysis of a large, unbalanced data set from applied wheat breeding programs can provide reliable estimates. All 2-way interactions between loci were present for 18 of the alleles. Analyses including interactions showed that epistasis was important for both Rmax and extensibility, especially between the Glu-B1 locus coding for high molecular weight glutenins and the Glu-A3 and Glu-B3 loci coding for low molecular weight glutenins. Because of the complexity of these interactions, similar values of Rmax and extensibility were predicted for diverse combinations of alleles. This implied that the practical application of glutenin genes in applied wheat breeding would be greatly enhanced by computer software which can predict dough rheology characteristics from glutenin allele classifications.

Photoperiod and vernalization gene effects in southern Australian wheat

Photoperiod and vernalization genes are important for the optimal adaptation of wheat to different environments. Diagnostic markers are now available for Vrn-A1, Vrn-B1, Vrn-D1 and Ppd-D1, with all four genes variable in southern Australian wheat-breeding programs. To estimate the effects of these genes on days to heading we used data from 128 field experiments spanning 24 years. From an analysis of 1085 homozygous cultivars and breeding lines, allelic variation for these four genes accounted for ~45% of the genotypic variance for days to heading. In the presence of the photoperiod-insensitive allele of Ppd-D1, differences between the winter genotype and genotypes with a spring allele at one of the genes ranged from 3.5 days for Vrn-B1 to 4.9 days for Vrn-D1. Smaller differences occurred between genotypes with a spring allele at one of the Vrn genes and those with spring alleles at two of the three genes. The shortest time to heading occurred for genotypes with spring alleles at both Vrn-A1 and Vrn-D1. Differences between the photoperiod-sensitive and insensitive alleles of Ppd-D1 depended on the genotype of the vernalization genes, being greatest when three spring alleles were present (11.8 days) and least when the only spring allele was at Vrn-B1 (3.7 days). Because of these epistatic interactions, for the practical purposes of using these genes for cross prediction and marker-assisted selection we concluded that using combinations of alleles of genes simultaneously would be preferable to summing effects of individual genes. The spring alleles of the vernalization genes responded differently to the accumulation of vernalizing temperatures, with the common spring allele of Vrn-A1 showing the least response, and the spring allele of Vrn-D1 showing a response that was similar to, but less than, a winter genotype.

The flow of alleles of important photoperiod and vernalisation genes through Australian wheat.

The photoperiod sensitivity gene Ppd-D1 and the vernalisation genes Vrn-A1, Vrn-B1, and Vrn-D1 are known to contribute to optimal adaptation to specific environments. Diagnostic molecular markers for detecting important alleles of these genes are now available, including for 2 distinct spring alleles of Vrn-A1 (a and b). As a first step for determining the relative importance of these alleles, they were characterised in Australian cultivars released from the late 19th until the early 21st Century. The photoperiod-insensitive Ppd-D1a allele did not occur in the Australian cultivars we assessed until after the release of cultivars containing CIMMYT germplasm in 1973. Thereafter, this allele became common; however, cultivars with an alternative, presumably photoperiod-sensitive, allele have continued to be released for all parts of the Australian wheatbelt, including for latitudes less than 28°S. In contrast to other parts of the world, Vrn-A1b was frequent in cultivars released during the first 70 years of the 20th Century and is still present in modern cultivars. Before the use of CIMMYT germplasm, the spring allele of Vrn-B1 and the winter allele of Vrn-D1 were common. Four major combinations of alleles of these major genes were identified in modern cultivars: first, those similar to WW15 (Anza), with the Ppd-D1a allele, the spring Vrn-A1a allele, and winter alleles at Vrn-B1 and Vrn-D1; second, those similar to Spear or Kite, with the alternative, photoperiod-sensitive allele at Ppd-D1, the spring Vrn-A1a allele, the spring Vrn-B1a allele, and the winter allele at Vrn-D1; third, those similar to Pavon F 76, with the Ppd-D1a allele, the winter allele at Vrn-A1, and the spring alleles at Vrn-B1 and Vrn-D1; fourthly, those similar to Gabo, with the winter allele at Vrn-A1, the spring allele at Vrn-B1, the winter allele at Vrn-D1, but the Ppd-D1a allele. Other combinations were found, including those for winter cultivars and those for early heading cultivars. A hypothesis was suggested for the facultative cv. Oxley. Evidence was presented to suggest that modern full-season cultivars head ~1 week earlier in a Mallee environment than cultivars from the late 19th Century.

The influences of genotype, environment, and genotype × environment interaction on wheat quality

Knowledge of the relative contributions of genotype (G), environment (E), and genotype and environment interaction (G × E) effects on wheat (Triticum aestivum L.) quality leads to more effective selection in breeding programs and segregation of more uniform parcels of grain better suited to the needs of customers. Their effects on wheat quality were reviewed using papers obtained from 4 major international databases. The literature is dominated by research from North America, with lesser contributions from Europe, Australia, and the rest of the world. Use of analysis of variance to partition sources of variation due to G, E, and G × E was the most common approach but, more recently, residual maximum likelihood methods that can accommodate large, but unbalanced, datasets have been used. In North America and Europe, the relative contributions of G, E, and G × E varied across studies, but traits associated with protein content were more influenced by E and G × E than those associated with protein quality, dough rheology and starch characteristics, where G effects were more important. Variation in the relative contributions of G, E, and G × E was highly dependent on the G and E sampled. The Australian studies were characterised by a relative lack of G × E, with G and E rankings being similar across the country for the protein quality, dough rheology, and starch quality traits examined in detail. This suggests that, in Australia, more efficient testing of potential cultivars will be possible for these traits, especially when the underlying variation at the gene level is known, and that efficiencies in the design and conduct of trial systems and quality evaluations could be achieved by testing samples from targetted environments without affecting genetic gain and overall crop quality.

Ppd-B1 and Ppd-D1 and their effects in southern Australian wheat

Abstract. Photoperiod and vernalisation genes are important for the adaptation of wheat to variable environments. Previously, using diagnostic markers and a large, unbalanced dataset from southern Australia, we estimated the effects on days to heading of frequent alleles of Vrn-A1, Vrn-B1, and Vrn-D1, and also two allelic classes of Ppd-D1. These genes accounted for ∼45% of the genotypic variance for that trait. We now extend these analyses to further alleles of Ppd-D1, and four alleles of Ppd-B1 associated with copy number. Variation in copy number of Ppd-B1 occurred in our population, with one to four linked copies present. Additionally, in rare instances, the Ppd-B1 gene was absent (a null allele). The one-copy allele, which we labelled Ppd-B1b, and the three-copy allele, which we labelled Ppd-B1a, occurred through a century of wheat breeding, and are still frequent. With several distinct progenitors, the one-copy allele might not be homogenous. The two-copy allele, which we labelled Ppd-B1d, was generally introduced from WW15 (syn. Anza), and the four-copy allele, which we labelled Ppd-B1c, came from Chinese Spring. In paired comparisons, Ppd-B1a and Ppd-B1c reduced days to heading, but Ppd-B1d increased days to heading. Ppd-D1a, with a promoter deletion, Ppd-D1d, with a deletion in Exon 7, and Ppd-D1b, the intact allele, were frequent in modern Australian germplasm. Differences between Ppd-D1a and Ppd-D1d for days to heading under our field conditions depended on alleles of the vernalisation genes, confirming our previous report of large epistatic interactions between these classes of genes. The Ppd-D1b allele conferred a photoperiod response that might be useful for developing cultivars with closer to optimal heading dates from variable sowing dates. Inclusion of Ppd-B1 genotypes, and more precise resolution of Ppd-D1, increased the proportion of the genotypic variance attributed to these vernalisation and photoperiod genes to ∼53%.

Revision of the estimates of glutenin gene effects at the Glu-B1 locus from southern Australian wheat breeding programs

Glutenins are the major determinant of dough characteristics in wheat. These proteins are determined by genes at 6 loci, with multiple alleles present in southern Australian breeding programs. Previously, we estimated the effects of these genes on maximum dough resistance (Rmax), dough extensibility and dough development time. Subsequently, the allele previously classified as Glu-B1b was found to consist of 2 alleles, with one, now considered to be Glu-B1al, producing an overexpression of the Bx7 glutenin subunit. Therefore, there is a potential bias in our previous estimates. An extended dataset was analysed with the 2 alleles now separated. These analyses identified negligible biases in our previous estimates, probably due to a low frequency of Glu-B1al before 1999. However, Glu-B1al produced significantly higher Rmax, dough extensibility, and dough development time values than all other alleles at the Glu-B1 locus. Therefore, at intermediate allele frequencies, substantial bias in estimates of the effects of the Glu-B1 alleles can be expected without correct identification of Glu-B1al.

The effects on grain quality traits of a grain serpin protein and the VPM1 segment in southern Australian wheat breeding

Production of wheat of sufficient quality to meet market demands is an ongoing agricultural challenge. Identification and evaluation of alleles of genes affecting quality parameters enables breeders to improve their germplasm by active selection towards specific allele combinations. Using a large dataset obtained from southern Australian wheat breeding programs, and including a relationship matrix in the analysis to minimise bias, we re-evaluated the effects of high- and low-molecular-weight glutenin alleles and puroindoline alleles on the grain quality parameters Rmax, dough extensibility, dough development time, flour water absorption, and milling yield and found that estimated effects were in close agreement with those from earlier analyses without a relationship matrix. We also evaluated, for the first time, the effects on the same quality parameters of 2 alleles (wild-type and null) of a defence grain protein, a serpin located on chromosome 5B. In addition, we assessed the effect of the VPM1 alien segment. The serpin null allele significantly reduced milling yield by ~0.4 g of flour per 100 g of grain milled across different germplasm sources and flour protein levels. In Australian germplasm, the origin of this allele was traced to a 19th Century introduction from India by William Farrer; however other sources, of significance in international breeding programs, were also identified. Our analysis of the effect of the VPM1 segment on quality traits revealed no detrimental effects of its presence on the traits we measured.

Genetic and environmental variation for grain quality traits routinely evaluated in southern Australian wheat breeding programs

Milling yield, maximum dough resistance (Rmax), dough extensibility, flour protein concentration (flour protein), particle size index (PSI), water absorption, and dough development time are important determinants of grain quality and are routinely evaluated in Australian wheat breeding programs. Information on allelic variation at the 6 loci determining glutenin proteins is also regularly obtained and used to predict Rmax and extensibility. For each character, except dough development time, 4029 observations on 2377 lines and 94 environments were analysed to estimate genotypic and environmental variances, heritabilities, genotypic and environmental correlations, and the effects of glutenin genes. A subset was analysed for dough development time. Milling yield, Rmax, extensibility, PSI, water absorption, and dough development time had intra-class correlation coefficients, or broad-sense heritabilities, between 0.66 and 0.76, and extensibility had a value of 0.52, with flour protein at 0.36. Genotypic and environmental correlations between extensibility and flour protein were high at +0.78 and +0.85, respectively. Rmax had a genotypic correlation with dough development time of +0.67, which was substantially due to pleiotropic effects of glutenin genes. Rmax, extensibility, PSI, and dough development time were influenced by glutenin genes. For Rmax about 50% of the genotypic variance could be explained by glutenin genes. For extensibility about 50% could be explained by flour protein, with 50% of the remainder by the inclusion of glutenin genes. For dough development time about 15% could be explained by flour protein, with a further 30% by glutenin genes. For PSI, about 40% of the genotypic variation could be accounted for by glutenin genes after the removal of the effects of flour protein and milling yield. We concluded that dough development time could be added to Rmax and extensibility as a trait that can be usefully predicted by the glutenin genes, but more work is required for PSI.

Ppd1, Vrn1, ALMT1 and Rht genes and their effects on grain yield in lower rainfall environments in southern Australia

Abstract. Allele-specific markers for important genes can improve the efficiency of plant breeding. Their value can be enhanced if effects of the alleles for important traits can be estimated in identifiable types of environment. Provided potential bias can be minimised, large, unbalanced, datasets from previous plant-breeding and agronomic research can be used. Reliable, allele-specific markers are now available for the phenology genes Ppd-D1, Vrn-A1, Vrn-B1 and Vrn-D1, the aluminium-tolerance gene TaALMT1, and the plant-stature genes Rht-B1 and Rht-D1. We used a set of 208 experiments with growing-season rainfall of <347 mm from southern Australia to estimate the effects of seven frequent combinations of the phenology genes, an intolerant and a tolerant allele of TaALMT1, and two semi-dwarf combinations Rht-B1b + Rht-D1a (Rht-ba) and Rht-B1a + Rht-D1b (Rht-ab) on grain yield in lower rainfall, Mediterranean-type environments in southern Australia. There were 775 lines in our analyses and a relationship matrix was used to minimise bias. Differences among the phenology genes were small, but the spring allele Vrn-B1a might be desirable. The tolerant allele, TaALMT1-V, was advantageous in locations with alkaline soils, possibly because of toxic levels of aluminium ions in subsoils. The advantage of TaALMT1-V is likely to be highest when mean maximum temperatures in spring are high. Rht-ab (Rht2 semi-dwarf) was also advantageous in environments with high mean maximum temperatures in spring, suggesting that for these stress environments, the combination of Vrn-B1a plus TaALMT1-V plus Rht-ab should be desirable. Many successful cultivars carry this combination.

High-yielding lines of wheat carrying Gpc-B1 adapted to Mediterranean-type environments of the south and west of Australia

Abstract. The Gpc-B1 gene from wild emmer has been proposed as a potential mechanism for improving grain protein in bread wheat without reducing grain yield. Near-isolines with and without the Gpc-B1 gene in three Australian-adapted genetic backgrounds, Gladius, Wyalkatchem and VR1128, were compared in 14 experiments across the south and west of Australia for grain yield, grain protein content and grain weight. The donor parents of Gpc-B1 were the Canadian cultivars Burnside and Somerset. One of the 14 experiments was discarded because of inadequate rust control and confounding effects of Yr36, a gene closely linked to Gpc-B1. Heading date and test weight were measured in five experiments. Across all comparisons, Gpc-B1 increased grain protein content and reduced grain weight, with a negligible effect on grain yield. Selected lines containing Gpc-B1 in a Wyalkatchem background had comparable grain yields to the elite cultivar Mace, but with significantly higher grain protein contents, slightly higher grain weights, similar heading dates and acceptable test weights. The development of agronomically acceptable lines containing Gpc-B1 was partially attributed to the removal of undesirable genes from wild emmer during the breeding of the Canadian donor parents and the use of Australian recurrent parents with high test weights.