Economic, Global,
Impact
Impact Summary
Professor Cullis and his team have been instrumental in developing novel research methods and computer software (ASReml), which has proved a game-changer for many plant-breeding programs. By identifying genetic advantages in crops, this research has led to more productive and sustainable farming, and increased profits for grain growers with flow-on social, economic and environmental impacts.
Related United Nations Sustainable Development Goals:
2. Zero hunger
8. Decent work and economic growth
9. Industry, innovation and infrastructure
12. Responsible consumption and production
15. Life on land
Details of the Impact
In order for Australian grain producers to remain internationally competitive, they need to deliver maximal grain yield and specific grain quality in a variable and unpredictable, water-limited environment. Because of the increasing economic and environmental impacts associated with this, effective evaluation of trials of new varieties is paramount.
Prof Brian Cullis, in partnership with the Grains Research Development Corporation (GRDC), conducted key research to address this growing need. This partnership has encompassed 19 research projects, training awards and postgraduate scholarships with a combined value in excess of $12 million. The most recent project, Statistics for the Australian Grains Industry-2 (SAGI-2), aimed to nationally coordinate statistical research and training support for the Australian grains industry.
SAGI-2 led to major benefits for the Australian grain industry, including: (1) enabling the National Variety Trials program, the largest global independent varietal evaluation program; (2) providing genotype estimation methods, adopted by the largest grain breeding companies; (3) providing statistical services to 10 major plant-breeding programs to assist in increasing yields; (4) assisting industry in improving the statistical rigour of quality tests; (5) delivering software tools to industry; and (6) strengthening biometrics training.
1. The National Variety Trials (NVT) Program (est. 2005) uses comparative crop variety testing with standardised trial management, data generation, collection and dissemination to evaluate species data. It is the largest independent varietal evaluation program in the world and the primary source of variety performance information of 11 winter crop species for Australian farmers and their advisors. SAGI-2 was responsible for analysis of all data collected within the NVT program and the continued improvement of NVT Multi-Environment-Trial analysis methodologies.
2. In this capacity, SAGI-2 developed a factor analytic approach for analysing NVT data that shifted the paradigm for estimating genotype by trial effects, and was so successful it is being adopted by many of the largest private wheat, barley, canola and sorghum breeding companies in Australia for analysis of their own breeding program data.
3. SAGI-2 provided statistical services – including trial design, data analysis and training – to assist these programs in realising minimum annual yield potential increases of 2% for pulses and oilseeds and 1% for cereals, respectively.
Steve Jefferies, CEO of Australia’s largest plant breeding company, Australian Grain Technologies, stated:
“I honestly believe that Brian Cullis and his team have done more for improving rates of genetic gain in Australian plant breeding than any other research group in the country … Brian's research has revolutionised the way we go about plant breeding at AGT. We plan our breeding strategies around what can be achieved in terms of novel design, analysis and interpretation. We at AGT, led by Haydn Kuchel, have always been very early adopters of Brian's latest technologies as we know they add value.”
4. Australia’s $5 billion wheat export industry has an international reputation for delivering high quality wheat within specific Australian wheat classes that describe the end product quality (such as dough strength, baking performance and extensibility). New wheat varieties are measured against a series of technical specifications to determine their class. Since considerable premiums are paid for higher quality grades, the impact of classification decisions on breeders and growers is considerable. Over the past 5 years, Cullis assisted the industry to improve the statistical rigor of quality tests applied to wheat, upon which classification decisions are taken.
Tress Walmsley, CEO of Intergrain, and her team stated:
“Brian and his team have revolutionized the approaches taken by Australian breeders and the broader grains industry to design and analyse research trials … His direct engagement with members of the grains industry has been a critical component of this success. As a result, the increased confidence with which breeders are now able to identify elite lines is delivering high-value to Australian farmers through improved varieties and agricultural research outcomes.”
5. Cullis extended the reach of the research by delivering these statistical methods through software package ASReml, now one of the most widely used tools in animal and plant breeding in the world. ASReml is used in over 37 countries with over 1800 users. It is currently distributed world-wide by VSN International.
6. As leader of the SAGI program, GRDC tasked Cullis with strengthening biometrics training within Australian higher education institutions. Cullis embedded SAGI-2 biometricians within universities across Australia, developed a series of statistical teaching modules that will be delivered across Australia and leveraged additional university investment in Biometric appointments. Through Cullis’ efforts, a large and effective biometrics group was re-established at the Waite Campus of the University of Adelaide. This improved biometrics training for Agricultural Science graduates, and provided much needed support for GRDC-supported research activities conducted within the Waite precinct.
Steve Jefferies stated:
“Importantly, the GRDC’s long-term investment in SAGI has created enduring capacity in world-class biometricians in Australia and this generation of mid-career statisticians will continue to benefit the grains industry as they grow into our next crop of statistical leaders.”
“This significant investment will increase the national grains industry’s capacity in the area of biometrics – the application of statistics to biological data – which is incredibly important in ensuring that grains research is statistically sound and credible, as well as speeding up research outcomes for the benefit of growers.”
Beneficiaries
- Grains and Research Development Corporation
- Bayer Crop Science
- Intergrain Pty Ltd
- Radiata Pine Breeding Company
- Australian Grains Technologies
- Queensland Department of Agriculture and Fisheries
- NSW Department of Primary Industries
- Department of Agriculture and Food, Western Australia
- Pioneer Hybrids
- Department of Economic Development, Jobs, Transport and Resources, Victoria
Countries
Impacted Countries
- Australia
- United Kingdom
Approach to Impact
Summary of the approaches to impact
At UOW, our approach to impact builds on our embedded culture of end-user driven research to recognise and prioritise research with potential for impact. This approach includes strategic recruitment that emphasises potential for impact, applied researcher centres (e.g. our research strength the National Institute for Applied Statistics Research Australia), collaborative research with an emphasis on end-user training, student research participation and critical infrastructure support (e.g. the provision of High Performance Computing facilities). Our strategy prioritises long-term partnerships and champions the role of senior researchers as they develop and maintain strong reputations and secure the trust and commitment of our beneficiaries.
Approach to Impact
Our approach to impact in Mathematics is centred on the recognition of the importance of reputation and leadership in developing productive and impactful relationships with end-users and beneficiaries. Our impact is facilitated through strategic recruitment focused on researchers with strong track records and continued potential for impact, resources provided by our applied research centres, ongoing support and training for end-users, engaging students in industry-facing research and providing infrastructure to enable and support high-impact research.
Professor Cullis leads the Centre for Bioinformatics and Biometrics in the National Institute for Applied Statistics Research Australia (NIASRA). NIASRA is one of ten recognised research strengths of the University and is provided with additional dedicated administrative and technical support. NIASRA was established by UOW in 2004 with Prof David Steel as Director (as the then Centre for Statistical and Survey Methodology) with the aim of creating a place where the discipline of applied statistics collaborates with academics and professionals in science, government, business and industry to tackle strategic problems.
To achieve this aim, NIASRA recruited Prof Cullis and his team to develop significant research capacity in bioinformatics and biometrics. This investment was made in anticipation of the impact he and his team has made. We also recruited two other internationally recognised senior research professors, Ray Chambers and Noel Cressie, as part of a strategy to provide Australia with a world class applied statistics research institute that undertakes leading-edge research into statistical methodology, and works with industry to ensure that we tackle relevant problems and that results of our research are applied in practice.
NIASRA has research centres in Bioinformatics and Biometrics, Environmental Informatics, and Sample Survey Methodology. We specialise in statistical design and analysis in large scale and complex experiments, sample surveys and spatial-temporal statistics. The connection between statistical design and analysis is a common theme across each centre, each of which benefits from our senior researchers’ guidance and perspective as we work to advance Australia's reputation in statistical design research.
An important component of our approach to impact involves training. For example, the ‘Statistics for the Australian Grains Industry’ national node is based at UOW, and close links are maintained with the regional nodes through training and research initiatives led by Prof Cullis and his team. To enable the continued growth of biostatistics capabilities within the Australian Grains industry, Prof Cullis has helped re-establish biostatistics training at the Waite Campus of the University of Adelaide.
Similarly, other researchers in NIASRA offer support and training both internally (for other applied researchers across the university through our statistical consulting service) and externally. For example, we have provided training to local high school teachers on the use of whiteboard tutorial rooms translating into more effective teaching environments for local students.
Our interdisciplinary research through NIASRA involves projects in health, social, environmental, agricultural, biological and economic sciences. We have undertaken research contracts in the last five years with over 25 organisations in Australia and overseas. Our members have extensive networks of research collaborators in the USA, UK and Europe in universities and relevant government agencies, which provide unique opportunities for training and mentoring of our early and mid-career researchers, who are driving the future of applied statistics in Australia and further interaction with relevant companies and government agencies.
Our students’ research is also a key pathway to impact. For example, our PhD student Margo Barr’s research with the NSW Ministry of Health changed population health survey methodology to include mobile phone surveys enabling a more representative sample of households in survey data. Our grain research, mentioned in Part A, has benefitted from student projects funded by GRDC student summer vacation, honours year and PhD scholarships.
NIASRA has over 12 PhD students and 24 academic staff, 17 of whom are research-only staff, including its four professors. NIASRA staff supervised 27 PhD completions in 2011-16.
In addition to weekly seminars, we host two one-day Fellows Research Meetings each year involving researchers from UOW, ABS, ANU, CSIRO, and other Canberra and Sydney-based organisations. We also organise ‘StatsWeek’ in February each year (since 2014), involving international researchers. These activities provide opportunities for knowledge growth, innovation, collaboration, mentoring, and ECR and HDR student training, and translation of the research to end-users. NIASRA also organised a one-day workshop on frontiers in Social Statistics in February 2017 and hosted the International Conference on Robust Statistics in July 2017, providing a strong environment for its applied research.
Our support for high-impact research is ultimately underpinned by a commitment to providing the infrastructure and resources needed to support the unique, computationally intensive research within NIASRA. The impacts in part A, along with a number of other projects in the discipline, were enabled by an investment of $40,000 in a High Performance Computing (HPC) Cluster for NIASRA. Statistical software packages, including SAS and STATA are purchased and supported by UOW and are essential to support this research. NIASRA receives internal funding of over $100k a year, supplemented by other UOW research grants. Salary support includes the funding of four research Professors, including Prof Cullis, and three support staff, including a dedicated IT specialist in HPC.
Associated Research
Statistical research by Cullis and his team on advanced multi-environment trial analyses and designs was used to educate plant breeders, plant scientists and biometricians in relevant statistical methodologies; and develop information delivery tools and software to support plant breeding programs. The industry’s desired grain improvement is facilitated by comparative experiments which aim to test and refine grain lines in different environmental conditions.
This research also developed a library of packages within the R statistical computing environment to assist with the increased computing power needed for large genetic datasets.. These packages include ASReml-R (for linear mixed model analysis), OD (for optimal trial design), ASMap (for creation of genetic maps) and ASReml-extras (includes tools for summarising ASReml-R output).
These packages were developed by Prof Cullis and his collaborators A R Gilmour VSN International, Hemel Hempstead, United Kingdom; B J Gogel University of Adelaide, Australia; S J Welham VSN International, Hemel Hempstead, United Kingdom; and R Thompson Rothamsted Research, Harpenden, United Kingdom. ASReml was first developed in 1995 however the newer packages listed were developed while Professor Cullis was at UOW.
References
1. A. C. Gleeson, A.C. and Cullis, B. R.. Residual maximum likelihood (REML) estimation of a neighbour model for field experiments. Biometrics, 43:277–288, 1987.
2.Cullis, B. R. and Gleeson, A. C.. Spatial analysis of field experiments - an extension to two dimensions. Biometrics, 47:1449–1460, 1991.
3.Cullis, B. R.; McGilchrist, C. A. and Gleeson, A. C.. Error model diagnostics in the general linear model relevant to the analysis of repeated measures and field experiments. Journal of the Royal Statistical Society, Series B, 53:409–416, 1991.
4.Gilmour, A. R.; Thompson, R. and Cullis, B. R.. AI, an efficient algorithm for REML estimation in linear mixed models. Biometrics, 51:1440–1450, 1995.
5.Cullis, B. R.; Gogel, B. J.; Verbyla, A. P. and Thompson, R.. Spatial analysis of multi-environment early generation trials. Biometrics, 54:1–18, 1998.
6.Verbyla, A. P.; Cullis, B. R.; Kenward, M. G. and Welham, S. J.. The analysis of designed experiments and longitudinal data by using smoothing splines (with discussion). Journal of the Royal Statistical Society, Series C, 48:269–311, 1999.
7.Gilmour, A. R.; Cullis, B. R.; Welham, S. J.; Gogel, B. J. and Thompson, R.. An efficient computing strategy for prediction in mixed linear models. Computational Statistics & Data Analysis, 44:571–586, 2004.
8.Oakey, H.; Verbyla, A.P.; Cullis, B.R.; Pitchford, W.S. and Kuchel, H.. Joint modelling of additive and non-additive genetic line effects in single field trials. Theoretical and Applied Genetics, 113:809–819, 2006.
9.Welham, S.J.; Cullis, B.R.; Kenward, M.G. and Thompson, R.. The analysis of longitudinal data using mixed model L-splines. Biometrics, 62:392–401, 2006.
10.Smith, A.B.; Thompson, R.; Butler, D.G.; and Cullis, B.R. The analysis of variety trials using mixtures of composite and individual plot samples. Journal of the Royal Statistical Society, Series C, 60:437–455, 2011.