Farrer Memorial Oration 2002

Introduction

The overall aim of a wheat breeder in Australia is more about raising the profits of Australian farmers, and with the flow on effects, improving the lifestyle of all Australians, rather than any philanthropic desires of helping to feed a hungry world. This is expressed in this simple equation:

PROFITS = (PRODUCTION*PRICE - COSTS)*PROBABILITY

Where: Production is determined by the grain yields of a variety in a range of environments, Price is determined by the quality of the harvested grain, Costs are inputs that the farmer must use to achieve that production and quality and Probability is the risk of disease which could prevent that production from being achieved. An improvement in any one of these will lead to greater profits. It is the breeder's task, by manipulating genes in the wheat genome, to increase yields and adaptation, to improve grain quality and to increase disease resistance.

The value of a parcel of wheat to a consumer is how much he is prepared to pay for the quality of that parcel. The cost of its production, the country from which it came and the area that was needed to produce it are of no concern. Much of Australia's wheat harvest, particularly in South and Western Australia, is exported to consumers in other countries, consumers who eat wheat in products very different to those traditionally eaten in Australia such as pan or loaf bread, cake and biscuits. End products such as flat breads, steam breads, white salted noodles and yellow alkaline noodles have very different quality requirements in the flour from which they are made. Although very significant for Australia, the size of our export wheat trade is not great in the total world scene so we tend to be price takers rather than price makers. Therefore it is imperative that to maintain, let alone increase, our market share in wheat quality conscious countries we must provide the qualities they demand.

Australia has enjoyed a special place in the market because its varieties are white grained and harvested clean and dry. Other exporting countries such as the USA and Canada have had to grow red grained wheats which are less liable to weather damage (sprouting) in their often wet harvests. Not only can white grain be milled harder to extract more white flour but resultant bread products are more tasty. This competitive advantage will be lost as wheat breeders in these countries are now breeding sprout resistant white grained varieties.

Wheat breeding is an integrating science. No one recognized this more than William Farrer who worked to produce wheats with wider adaptation than those being grown, which had stem rust and bunt resistance and had improved milling and baking qualities.

Frederick Guthrie, a cereal chemist, worked with Farrer as a friend and colleague and together they greatly improved the milling and baking properties of wheat in Australia. Guthrie pioneered the development and use of small mills to enable Farrer to test progeny from his crosses in early generations when the number of lines was great but the amount of grain from each small. Federation, released in 1901, was a quantum leap in terms of yield potential and adaptation. Florence, named in 1906, was a breakthrough in combining high yields with excellent flour quality and was widely grown across Australia and in Tunisia for over 40 years (Russell,1939; MacIndoe and Walkden Brown, 1968). Both of these varieties are long since superceded but many modern varieties carry genes from Farrer's wheats, not just by chance, but because those genes confer adaptive advantages to Australian conditions. Federation, for example, is the source of genes for tolerance to high soil boron, which is a problem of mallee and red brown earth soils in southern Australia.

At the beginning of the 21st century, just as Farrer did at the beginning of the 20th century, Australia's wheat breeders are poised to make quantum improvements. Such large improvements are likely because there is new information with resultant new technologies which, coupled with reorganization and funding of breeding, can be rapidly implemented. Speeding the breeding is imperative if farmers are to remain solvent and Australia is to remain competitive in the world market. Herein I want to demonstrate some of these changes using as an example the Roseworthy Campus wheat breeding program and its approach to quality.

Flours ain't flours

Figure 1 diagrammatically shows the tissues of a wheat grain and what each component becomes during flour milling. By far the majority of wheat products for human consumption are based on white flour. White flour is powdered endosperm. The embryo is extracted as wheat germ and the outer layers of the grain and the outermost layer of the endosperm, the aleurone layer (not labeled) separate as bran. The larger the endosperm content and the ease with which it can be separated from the offal products during milling without contamination are important quality attributes. Bran contamination in the flour causes problems in baking, and of course the more flour the more higher priced product that can be produced from a given sample.

Thus white flour will contain a range of starch granules (size, type and different amounts of amylose and amylopectin, some damaged), storage protein (a range of different high and low molecular weight glutenins and gliadins), structural protein (albumins and globulins), lipids and other degradation products of the once living endosperm cells. When mixed into a dough it produces an edible complex polymer with properties that enable it to be baked into a number of attractive and palatable products. It is the behaviour of this polymer mixture, mostly but not entirely, a function of the storage proteins that makes wheat such a useful food substrate. Its nutritional value has been a secondary consideration. So attributes affecting performance of dough are the other important attributes of quality which need to be tested and selected in breeding programs.

Pan bread requires flour milled from a hard grain with high protein. This combination allows the miller to produce the right amount of starch damage to achieve the desired crust colour and water absorption, bread being a wet product, and the high protein creates a molecular mesh in the dough to trap the gas bubbles produced by the yeast and allows the dough to rise and give an agreeable texture to the bread. Biscuit flour on the other hand is made from soft wheat of low protein. Soft wheat does not produce a lot of damaged starch granules during milling so there is not a lot of water absorbed by the dough which would take time and energy to drive off during cooking to give a dry biscuit product which would also have a charred appearance. Protein in a dry product acts as a glue and high protein would make the biscuit very hard. Flat or pocket breads have intermediate requirements, too much protein would prevent the pocket forming when the water in the dough forms steam and blows the dough pieces up like a balloon, too low and the pocket would rupture before full expansion. Yellow alkaline noodles are made from hard wheat with a range of protein concentrations depending on the cooking recipe and country traditions.

Since Farrer's and Guthrie's time many tests have been developed and used in breeding programs to screen lines for quality. The Pelshenke Whole meal Flour test for dough strength, the Zeleny Sedimentation test, the Brabender Farinograph, the Brabender Extensograph, the Mixograph, Pearling Resistance, various methods for measuring protein concentration, various hammer, burr and roller mills to produce small samples of white flour, micro baking plants to mention just the ones used by Roseworthy.

Each technique has its advantages and disadvantages but most have been criticised because the results produced are not often very well correlated with commercial milling and baking. Further more the results from some are simultaneously affected by a number of components of quality thus making it difficult to select separately for improvements to each component. These different methods have been reviewed many times (Miller and Johnson, 1954; Baker and Campbell, 1971; Simmonds, 1989) and standard procedures are detailed in the AACC Approved Methods (2000).

Jardine et. al. (1963) performed a factor analysis of the results of many types of quality tests on a range of wheat samples varying in quality and reduced the huge results table down to four independent factors. Orth et. al. (1976) repeated the analysis with additional tests that had been developed in the interim and came to similar conclusions. These major and relatively independent determinants of quality are milling yield, grain hardness, protein concentration and protein quality (dough strength and extensibility).

There are other attributes which vary and cause problems in specific crosses depending on the parents used. The undesirable phenotypes are culled during the breeding cycle and should not be found in commercial varieties. For example, late maturing a-amylase, starch quality differences, 'sticky dough', high polyphenol oxidase, black point and weather damage.

Late maturing a-amylase(LMA) is an inherent character in some parents where sound grain has elevated levels of amylase enzyme at maturity (Mrva and Mares,1996). Such affected grain acts as though it was badly weather damaged and has little value for human consumption. This condition is not always expressed and is brought on by a period of cold weather during grain filling. It is not a problem in Queensland where grain filling occurs under ever increasing warmer temperatures so LMA prone wheats are safe to grow in Queensland. However in Southern and Western Australia with Mediterranean type climates although the average temperature is increasing during grain filling a cold front can drop temperatures at any time. This problem was exacerbated in breeding programs breeding for resistance to Septoria Leaf Blotch as the main source of resistance, Inia 66, is an LMA prone variety.

Figure 2. Grain protein and hardness requirements for major end uses of wheat (modified from Moss, 1973)

Breeding objectives for wheat quality

Table 1 lists many of the components which comprise wheat quality. Physical cleanliness in a parcel of wheat goes without saying because no buyer wants to buy tainting contaminants or unmillable material. Similarly a low moisture content will enable the buyer to buy more dry matter per tonne of sample, although moisture contents can be too low and give the miller troubles in conditioning the grain before milling. Moisture contents above about 12.5% attract insects. These are mostly affected by the weather and avoided by the farmer harvesting in a proper and timely manner. Genotypic effects, that is, differences between varieties are due to such things as difficulty in thrashing, predisposition to small grain and differences in proneness to weather damage. These are minor relative to crop management at harvest.

Other attributes are mostly affected by the variety (genotype), such as protein composition.

Attributes which are significantly affected by both environmental factors (management and climate) and genotype are particularly difficult to handle in a breeding program as in any single test there could be differential affects of environment on different genotypes such as explained for LMA. This phenomenon is known as genotype-environment interaction (GxE) and it is necessary to test for such attributes over a wide range of environments with not only the test lines but with a range of check varieties whose GxE patterns are known. This greatly increases costs and the time needed to make decisions.

Of the major components (grain hardness, protein concentration, dough strength and dough extensibility), protein concentration is largely determined by the environment and crop management whereas protein composition, which affects dough properties, is mostly genetically determined and is therefore the responsibility of the breeder. In southern Australia growing seasons vary greatly from year to year so that protein concentration, even on the one farm, can vary from 8% to over 15%. Much of this seasonal variation cannot be predicted before sowing and although a farmer may aim to produce at a particular protein level and sow a variety to match, the vagaries of the season can upset his plans and result in a wheat harvest that is higher or lower protein than that which he was targeting. This presents a challenge to the farmer (what protein to target?), to the breeder (what quality type to breed?) and to the marketing agencies (cannot forward sell).

Recently a think tank of eminent cereal chemists, wheat market intelligence experts and breeders in Australia (Wheat quality Objectives Group) have reviewed the requirements of the many end products in which wheat is consumed and described the balance of quality attributes which best suit these products. Much of their deliberations has commercial sensitivity and cannot be not divulged here. In broad terms however Table 2 describes the balance of properties required for the manufacture of the major end products.

The only difference between the needs of the end products listed is the protein concentration. Strategically then, the best varieties to breed for would be multi-purpose types, white grained, high flour extraction rates, high dough extensibility with moderate to high dough strength, so that whatever the final protein concentration the grain will still be readily marketable and attract the attention of discerning buyers. If such target varieties are achieved then the proportion of sales to top-paying discriminating markets will increase.

The traditional modified pedigree method

Breeding consists of creating new combinations of genes by crossing two parents specially selected so that their attributes will complement each other. Then in successive generations of self pollination identifying and selecting new improved combinations and discarding those lines which do not measure up along the way. The breeding system developed and used at Roseworthy for the last 20 years is detailed in Figure 3.

Quality testing has been very difficult, in fact almost impossible in early generations because there are many thousands of unique lines but only a little seed of each usually produced from only one site and then not under normal farm conditions, so environmental effects are atypical. Many scientists have attempted to minaturise quality tests to allow assessments to be carried out in these early generations but any single test will be influenced by genotype-environment interactions and to me be so unreliable as to be dangerous. So quality testing has been left until later generations when fewer lines which are more true breeding and more grain is available from wide scale trials for end product and associated quality tests. Traditionally this has meant a long modified pedigree breeding system with new varieties not being released until 10 or more years after the objectives were set and the cross was made. A time during which market requirements might have changed. This strategy also means that high yielding, but not necessarily good quality, lines are identified and can be recycled as parents in further crosses. The down side is that the breeder has to be reconciled to the fact that the best yielding lines may be rejected at the eleventh hour just before release because they fail to reach quality standards (Hollamby, 1982). Brennan and O'Brien (1991) discuss the economics of testing for quality in early generations as a function of resources and gains under selection. Their conclusions were ambiguous but their sort of analysis did make breeders critically review their strategies.

The main stages in a breeding program are:

• Setting objectives,
• Creating new genetic variation by crossing selected parents,
• Early generation screening and selection for highly heritable characters, characters which do not express significant GxE effects. Many lines, little grain of each, large genetic variation, most heterozygous,
• Wide scale field evaluation in environments in which any new release will be grown, select for complex traits such as yield and quality by taking account of GxE effects. Few lines, less genetic variation, homozygous, database of previous performances, more grain,
• Seed multiplication, end product testing. Data for release and quality classification.

These stages can be seen in Figure 3. To be more proactive to market changes, to overcome changes in disease resistance especially with the recent evolution of stem, leaf and stripe rust which has rendered many current varieties susceptible, and to increase the rate of yield improvement we need to speed the breeding. A range of new scientific discoveries and resulting technologies have been developed during the last 10 years and with changes to wheat breeding organizations these can be quickly implemented.

Speeding the breeding

The adage that you can improve your competitive edge by choosing any two strategies from doing the job better, faster or cheaper (but not all three simultaneously) applies to wheat breeding. I believe that the best approach is to aim to release better varieties faster and when this has worked the income stream from the more rapid release of improved varieties will overcome the increase costs incurred.

What new technologies should allow each of the stages of wheat breeding to become better and faster? See Figure 3 and Figure 4.

1. Setting objectives

Objectives must be kept constantly under review and upgraded as priorities change and as new information and new genetic variation becomes available. Figure 4 describes Roseworthy's breeding objectives for 2000. Those objectives in the outer bulls eye are the objectives which are being actively pursued. Crosses are being made to specifically address these problems.

Objectives in the bulls eye are largely under control but must still be monitored and screened for as the desirable combinations of genes present in current varieties could be upset during crossing. Also situations can change, for example, the current suite of varieties and crosses are no longer adequate against new races of stem leaf and stripe rust which have arisen since 1999.

Objectives in boxes are relatively new, eg sprouting resistance has become an attribute for active breeding because new genetic variation and screening methods are now available enabling breeders to tackle this problem. Objectives in the outer ring are either of less importance or lack genetic variation from which to breed.

2. Creating new variation

Better choice of parents is now possible using world wide data bases which can be interactively interrogated such as the International Crop Information System (ICIS), and collaborative wheat evaluation projects such as the evaluation of germplasm from CIMMYT. Quality trait dissection through a better understanding of the genes involved and the underlying biochemical reactions makes the objectives seem more complex but in reality is actually a better description of the trait allowing targeting of individual genes.

As complex traits are broken down into smaller components one gets closer and closer to individual gene actions and GxE and environmental effects do not influence selection procedures as much and so there is less need for repeated testing in a range of environments.

Electrophoresis of the glutenin and gliadin storage proteins has allowed separation and identification of a number of protein subunits. Many workers observed marked differences in the protein patterns between different wheat varieties and eventually Payne et.al. (1979) reported that certain of the higher molecular weight glutenins correlated closely with bread making quality in wheat. Further work enabled Payne and coworkers (1981, 1984) to identify the genes encoding for the High Molecular Weight glutenins (HMW) and in 1987 to apply scores to the presence of individual alleles at each of the three independent loci to derive a selection index for dough strength. These loci are Glu-A1, Glu-B1 and Glu-D1 occurring on chromosomes 1A, 1B and 1D respectively. A large number of alleles have been identified at each of these loci and the effect of each has been designated a score from 1 an allele which has a poor contribution to dough strength to 4 an allele with a major contribution to improved dough strength. Thus by knowing the HMW genes in a set of parents the breeder could plan crosses to increase dough strength based on the best scores which would occur in the progeny.

Shepherd working at the Waite Agricultural Research Institute was also investigating the inheritance of storage proteins. The Payne score was of little benefit to Australian breeders because they already had the better HMW protein alleles throughout their breeding programs. Of more interest was how to explain the variation in quality that was as yet unexplained particularly in dough extensibility. This was particularly important because South Australian wheats were undesirably low for this attribute. In the original electrophoresis procedures the LMW glutenins and the gliadins in the gels produced were cluttered and confounded on top of each other. A very significant breakthrough in procedures occurred when Singh working with Shepherd developed a one step SDS-Page procedure for separating LMW glutenin subunits from the gliadins (Singh et al, 1991). Figure 5 shows such a gel. Alleles at three loci namely, Glu-3A, Glu-3B and Glu-3D which occur on chromosomes 1A, 1B and 1D respectively tightly linked to gliadin loci but independent of the Glu1 loci. Earlier before the one step procedure had been perfected Gupta and Shepherd (1988) had been using a protracted two step procedurer but this had enabled them to draw some conclusions about inheritance and allelic effects on quality of the LMW glutenins.

In figure 5 each lane has been produced by the protein extract from a single wheat grain. DH203 and DH082 carry the allele GluA3-d which confers better quality to them over the others in the gel, note the extra band in the LMW portion. DH082 and DH130 have HMW subunits 5+10 compared to the others which carry 2+12. This gives them extra dough strength, +2 on the Payne score. The four single lanes in the centre of the gel are checks.

The 1992 harvest in South Australia was very wet and a lot of grain was downgraded to animal feed only because of weather damage and sprouting. The same happened to the wheat breeders' field trials so there was no grain suitable for the usual late generation quality tests. A redeployment of resources allowed the wheat breeding program to genotype all advanced lines for HMW and LMW protein alleles. Since that time new entries into advanced trials have been routinely genotyped from gels prepared by the one step procedure by Cornish, another of Shepherd's colleagues. This information has been used immediately in selection-culling decisions. Over time a large data base on quality including early and late generation tests from a number of years together with HMW and LMW composition of breeding lines and check varieties has been built. Collating this with similar data from the Victorian wheat breeding program and using modern REML statistical techniques to analyse such an unbalanced data set on PC computers now capable of handling huge data bases Eagles has been able to greatly improve on the Payne score and Gupta's predictions (Eagles et al, 2002a, 2002b). Their conclusions are that Dough strength (Rmax of the Extensograph), extensibility and dough development time can be usefully and cost effectively predicted from the allelic combinations at the six loci. This work was conducted under the auspices of the CRC for Molecular Plant Breeding. It has been made even more useful to breeders by Podlich and refined by O'Connor at the Queensland University by building the predictions into a 'Glutenin cross prediction simulator', a decision support tool to assist wheat breeders to predict the likelihood of achieving quality target traits for specified wheat crosses (Rosemary O'Connor, r.oconnor@uq.edu.au).

The simulator is a subroutine within QU-GENE (Podlich and Cooper, 1998). The user inputs a choice of female and male parents and the simulator conducts a set of diallel crosses and ouputs graphically and in tables the frequency of different allelic combinations and their predicted quality attributes.

Continued introduction of genotypes from other countries and new synthetic wheats will no doubt provide new alleles for quality which after assessment could be used to improve quality.

3. Early generation screening

Early generation selection will become more reliable because of rapid attainment of homozygosity (true breeding) by using single seed descent, doubled haploid and out-ofseason growing techniques. Tests on homozygous and homogeneous selections, particularly for presence of individual genes need only be performed once as no further segregation will occur. The protein genes can be determined from a single seed from each line rather than having to screen a number of seeds to cover the genetic variation still present in heterozygous lines in F3 or F4 selections. Molecular DNA markers for flour colour and starch quality are already available and being used in breeding programs where these attributes need improving (Eagles et al, 2001). Doubled haploid populations from five crosses involving parents of interest to Australian wheat breeders and varying for a range of quality traits have been created and genetic maps are near completion (Kammholz et al, 2001). Molecular markers discovered in this project will be implemented into the Roseworthy program as soon as they are validated. Traits being targeted include flour yield, water absorption, sprouting tolerance, flour colour, dough strength and extensibility.

These types of screening methods are not minaturised extensographs or similar small scale tests which measure the phenotype of a small sample of grain and then leave it to the breeder to guess where this one result fits in the whole range of GxE interactions. They get close to the actual genes involved, or their initial protein products, divested of environmental effects. If such tests can be conducted efficiently, that is, quickly and cheaply then they are well worth using because very large populations can be screened and with a realistic, not too stringent, culling level the populations can be enhanced for improvements in quality without having to sacrifice gains in yield potential because too few lines survive.

The environment under which the plants are grown is immaterial because it is not necessary to have the quality traits expressed, therefore such tests can be done on plants from summer nurseries and single seed descent generations, so further speeding the breeding.

Near infrared reflectance (NIR) techniques have been used for many years and are the standard methods for measuring protein concentration, moisture content and lipid content of grain samples of many crop species. Such methods too have been used as the standard measure of Particle Size Index or grain hardness in wheat for a long time. The advantage of NIR is that it needs only a small quantity of grain and is non destructive so that the sample can be returned to the breeder for sowing. Initially this technique was regarded as being in the area of fantasy and any correlation with real quality as being coincidental (Wetzel, 1983). However, its attraction in terms of cost and speed per sample, encouraged cereal chemists working with plant breeders to investigate its potential for other quality attributes. Now, with improvements in computing techniques which have led to better methods of calibration such as principal component analyses and multiple regressions, coupled with a wider spectrum of wavelengths, NIR is proving to be useful for assessing flour yield, flour colour, water absorption and extensibility. Corrrelations between NIR predictions and actual quality expression varies from protein and grain hardness at better than 0.99 to extensibility of 0.30, using a Perten™ apparatus, but greater accuracy has been found with more recent machines. At Roseworthy we recalibrate the instrument each year using a set of about 100 replicated check lines planted along with the many thousands of test lines. Until now this has been an insurance against any effects of environmental effects on results. At the Victorian Institute for Dryland Agriculture they calculate a running calibration by continuously adding extra data points as they come to hand from year to year (R. Eastwood, pers. comm., 2002).

4. Wide scale evaluation

Spatial analyses (Gilmour et. al., 1966) have reduced the need for replication at a site and yet give more reliable data. Seed and time thus released allows for more sites to be planted in any one year.

The fact that not all varieties respond in the same way to different environments, genotype environment interactions mentioned previously, has in the past meant that breeders have tested at a number of sites over a number of years in order to have any confidence in their new lines before release. Either that or they released varieties not fully tested and let the early farmer adopters do the final testing for the rest of the farming community. The long testing period increased costs to breeding programs and denied farmers an improved variety for some time if testing proved the line to be truly superior. Releasing an improperly tested variety which then proved not to be superior to current varieties could damage breeders' reputations.

Probe genotypes have been used as a way of indirectly characterizing the influence of environment on crop performance (Fox and Rosielle, 1982; Cooper and Fox, 1996). Over time such a site bioassay could be used to weight specific trial environments according to their occurrence in the total target population of environments expected in a region. Fletcher et.al. (1990) using a set of such probes in the Roseworthy program recommended that at least 5 sites x 3 years were needed to give a reliable estimate of the average performance of a line. Such a protocol takes too long in today's need to speed the breeding. The use of disjunct sites followed up by a cluster analysis showing how well the environments which actually occurred at these sites represented the total target population of environments should allow the breeder to gain confidence in the superiority or otherwise of any line in a much shorter period (Hollamby, 1999: Hollamby and Hunt, 2000).

A nation wide testing program will ensure this sort of evaluation. Sites must be chosen to represent this range of environments which not only includes edaphic factors such as soil type and the abiotic stresses that might occur but also, and just as importantly, represent the farming practices under which wheat is expected to be cropped. These practices include rotations, fertilizer and herbicide usage, which may not necessarily be conducive to the highest yields possible or to the best quality, but are never the less practices which across a farmer's whole farm lead to a better financial position. Growing wheat following wheat and/or minimum tillage are examples. For this reason trials located on real farms rather than on experiment stations tend to be more relevant.

Whilst this sampling of environments is primarily for yield and adaptation a range of grain filling conditions will be experienced and protein concentrations will vary, so it is possible by judicious compositing of grain from relevant sites to achieve protein levels for proper evaluation of quality for different end products, and collect information not just for selection purposes, but also for final classification by the AWB into its grading system, information critical to farmer acceptance, profits and the maintenance of Australia's place in the market.

5. Seed multiplication

Seed multiplication prior to release is a specialized area for which breeders are not properly equipped. Contracting this work to commercial partners who have facilities for out of season multiplication and are able to handle seed quantities from a few kilograms up to several tonnes with the integrity required will shorten the time to release once a line is chosen. Built into this partnership should be the ability to multiply seed of lines before the final decisions are made so that some seed build up may have to be discarded at the end.

Epilogue

Wheat breeding is just as an exciting career as ever with many challenges to overcome. The understanding of our wheat plant and its response to abiotic and biotic stresses will continue to grow adding new objectives to the breeder's job list. New technologies will come on-line which will need to be implemented.

Markets will change with developments in food technology and with increasing concern by consumers about nutrition and taste, and maybe with political influences affecting to whom Australia can sell. A degree of flexibility of quality must be retained in order to quickly make appropriate adjustments when necessary.

Wheat so far is used almost exclusively for human and animal consumption. In the future its use as a 'biological factory' for pharmaceutical or neutriceutical products is possible and/or as a substrate for fuel and other industrial products. Such end uses will require very different qualities.

For the immediate future though speeding the breeding to produce improved varieties with qualities required by discerning markets challenges most breeding programs. Implementing the changes discussed in this paper should reduce the time from setting an objective, to making a cross and finally releasing an improved variety from around 12 years down to 6 to 8.

Acknowledgements

I am well aware that I have been only one person in a whole wheat breeding team and as such I am very humbled in receiving the William Farrer Memorial Medal but am very proud to accept it on behalf of the many people whose efforts were critical to my success. The expertise provided by others is shown in figure 3. I am deeply appreciative of the contributions of staff who helped in the program in my early years and who are named in the Roseworthy centenary book (Hollamby, 1983) and of current core staff, viz. S. Jefferies, R.Heppner, D.Smith, J.Menzel and R.Fisher, and those in between such as R.Fletcher and G.White. Also I will not forget the contribution made by collaborating farmers in allowing use of their land. I make special mention of Dr. Tony Rathjen at the Waite Agricultural Research Institute for acting as a sounding board for almost all I did, of Mr. Rex Krause for his mentorship in my formative years as a wheat breeder, of Mr. Ali Bayraktar my colleague breeder for a large part of my career and of my wife, Jan, for supporting me during this love affair with wheat plants.

References

American Association of Cereal Chemists (2000). 'A.A.C.C. Approved Methods.' Ed. A.A.C.C. Inc., St Paul, Minn. See also www.aaccnet.org/ApprovedMethods

Baker, R.J. and A.B.Campbell (1971). Evaluation of screening tests for quality of bread wheat. Can.J.Plant Sci. 51:449-455.

Brennan, J.P. and L.O'Brien.(1991). An economic investigation of early-generation quality testing in a wheat breeding program. Plant Breeding 106:132-140.

Cooper, M. and Fox, P.N. (1996). Environmental characterisation based on probe and reference genotypes. In Cooper, M. and Hammer, G.L. (eds) 'Plant adaptation and crop improvement', CAB INTERNATIONAL:529-548.

Eagles, H.A., Bariana, H.S., Ogbonnaya, F.C., Rebetzke, G.J., Hollamby, G.J., Henry, R.J., Henschke, P.H. and Carter, M. (2001). Implementation of markers in Australian wheat breeding. Aust.J.Agric.Res. 52:1349-1356.

Eagles, H.A., Hollamby, G.J., Gororo, N.N. and Eastwood, R.F. (2002a). Estimation and utilization of glutenin gene effects from the analysis of unbalanced data from wheat breeding programs. Aust.J.Agric.Res. 53:367-377.

Eagles, H.A., Hollamby, G.J. and Eastwood, R.F. (2002b). Genetic and environmental variation for grain quality traits routinely evaluated in southern Australian wheat breeding programs. Aust.J.Agric.Res. 53:1047-1057.

Fletcher, R.J., Hollamby, G.J. and Bayraktar, A. (1990). Pattern analysis of yield data over sites and seasons in South Australia. Proc. 6th Australian Wheat Breeding Assembly, Tamworth:381- 391.

Fox, P.N. and Rosielle, A.A. (1982). Reference sets of genotypes and selection for yield in unpredictable environments. Crop Science 22:1171-1174.

Gupta, R.B. and Shepherd, K.W. (1988). Low-molecular-weight glutenin subunits in wheat: their variation, inheritance and association with physical dough properties. Proc. 7th Int. Wheat Genet. Symp., Cambridge, UK:943-949.

Hollamby, G.J.(1982). A strategy for breeding for wheat quality. Chemistry Australia 49: 7(abstract)

Hollamby, G.J. (1983). The staff of life. In Jeff Daniels (ed) Roseworthy Agricultural College. A century of service. Griffin Press Ltd.:36-85.

Hollamby, G. (1999). The use of probe genotypes to improve selection efficiency - Examples. Proc 11th Australian Plant Breeding Conference:87-88.

Hollamby, Gil and Hunt, Colleen (2000). Effective selection for wide adaptation using probe genotypes. Sixth International Wheat Conference, June 2000, Budapest:266.

Jardine, R., Moss,H.J., and Mullaly, J.V. (1963). Wheat quality: a factor analysis of some test data. Aust.J.Agric.Res. 14:603.

Kammholz, S.J., Campbell, A.W., Sutherland, M.W., Hollamby, G.J., Martin, P.J., Eastwood, R.F., Barclay, I., Wilson, R.E., Brennan, P.S. and Sheppard, J.A.(2001). Establishment and characterization of wheat genetic mapping populations. Aust.J.Agric.Res. 52:1079-1088.

MacIndoe,S.L. and Walkden Brown, C. (1968). 'Wheat Breeding and Varieties in Australia.' Science Bull.76, New South Wales Department of Agriculture, 255pp.

Miller, Byron S. and John A. Johnson (1954). A review of methods for determining the quality of wheat and flour for breadmaking. Agric Expt Station, Kansas State College of Agriculture and Applied Science, Tech.Bull.76.

Moss,H.J. (1973). Quality standards for wheat varieties. J.Aust.Inst.Agric.Science,39:109-115.

Mvra, K. and Mares, D.J.(1996). Inheritance of late maturity a-amylase in wheat. Euphytica 88: 61-67.

Orth, R.A., L. O'Brien and R. Jardine (1976). A factor analysis of bread wheat quality tests. Aust.J.Agric.Res. 27:575-582.

Payne, P.I., Corfield, K.G. and Blackman, J.A.(1979). The identification of a high-molecular weight subunit of glutenin whose presence correlates with bread-making quality in wheat of related pedigree. Theor.Appl.Genet.55:153-159.

Payne, P.I., Holt, L.M., Jackson, E.A. and Law, C.N.(1984). Wheat storage proteins: their genetics and potential manipulation by plant breeding. Phil.Trans.R.Soc.Lond.304:359-371.

Payne, P.I., Nightingale, M.A., Krattiger, A.F. and Holt, L.M.(1987). The relationship between HMW glutenin subunit composition and the bread-making quality of British-grown wheat varieties. J.Sci.Food Agric. 40:51-65.

Podlich, D.W. and Cooper, M.(1998). QU-GENE: A simulation platform for quantitative analysis of genetic models. Bioinformatics 14:632-653.

Russell, Archer (1949). 'William James Farrer. A Biography.' 226pp. Cheshire, Melbourne.

Singh, N.K., Shepherd, K.W. and Cornish, G.B.(1991). A simplified SDS-PAGE procedure for separating LMW subunits of glutenin. J.Cereal Science 14:203-208.

Wetzel, D. L.(1983). Near-infrared reflectance analysis - sleeper among spectroscopic techniques. Analytical Chem. 55:1165-1176.