Summary
This project, through its research and development, has ensured beef producers in Australia have first class genetic evaluation and breeding objective technologies and associated services. The implementation and adoption of outputs from this project by industry have contributed significantly to faster rates of genetic gain in profitability of Australian beef cattle. This improvement has been achieved by ensuring Australian beef industry has an effective, flexible and improving genetic evaluation system that is able to adapt to the changing needs of its users and be able to incorporate new knowledge in genetics, animal breeding, recording practices, statistics and computing. The most recent (August 2004) estimates of beef industry rate of genetic progress for commercial profitability, summed across breed, market x production systems, show that this rate has increased by 1.58 times when comparing animals born in the periods 1998-2003 to 1992-1998. The increase in estimated genetic progress for the most recent year (i.e. 2002-2003) compared to animals born in 1992-98 was 1.97 times, which is very close to the MLA' strategic plan target to double (i.e. 2.0 times) the rate of genetic progress by 2006. Importantly, the benefits from the project's R&D will continue to contribute to further increases in the rate of gain over coming years as adoption and further implementation occurs.
Throughout the project the operational integrity of the BREEDPLAN and BreedObject systems have been maintained and enhancements were made to the evaluations of specific breeds based on their developing requirements. New genetic parameters were estimated from the complete Beef CRC carcase and meat quality database consisting of over 7,500 animals and these parameters were incorporated in the variance/covariance matrices of all temperate breeds in 2001, and of tropically adapted breeds in 2002. Also, for the first time, estimation was completed of a full genetic covariance matrix for 22 traits for the Hereford/Poll Hereford breed. These developments ensure BREEDPLAN analyses are using the most up to date estimates of critical parameters, thus ensuring selection decisions are based on the most accurate information possible. In early 2003, the project delivered to beef industry the first across-breed BREEDPLAN EBVs. This research was undertaken using data from the MRC Victorian Multi-breed EBV project and the Beef CRC Northern crossbreeding project to derive breed differences and the necessary base adjustments for their existing breed BREEDPLAN evaluations.
A simple table of adjustments was published that allows Australian beef producers to directly compare the EBVs of four prominent British (Angus and Hereford) and European (Limousin and Simmental) breeds for the full range of growth traits. Commercial crossbreeders or composite developers who are using combinations of these breeds can now simultaneously exploit both within and between breed variation. Enhancements to the BreedObject software have resulted in a very large increase in adoption of this technology by both the seedstock and commercial sectors. Twenty-four breed standard $Indexes have been developed in consultation with Industry, representing approximately 70% of Australian industry production situations. Most breeds now publish $Indexes on a regular basis via their BREEDPLAN internet facilities and the BreedObject web site. Importantly, the indexes can now be fully customised via the BreedObject web site. This gives individual producers the capacity to develop an index specifically for their own production and market system (or for that of their clients). The increased adoption of $Indexes by industry has greatly assisted seedstock breeders and bull buyers making selection decisions with the correct trait emphasis for profitability and hence has contributed to overall increases in industry profitability.
The BreedObject software was also enhanced to allow computation and reporting of the genetic trends in profitability summed across the entire industry. Genetic trends in the existing breed standard indexes show all breeds are making significant positive progress in their $Indexes and, in almost all cases, that this progress is occurring at an increasing rate. New "StockTake" software, developed during the project, allows a genetic audit of a breed, or herds within a breed. BREEDPLAN users (e.g. Breed societies or large companies) can use StockTake to monitor changes over time in numerous breeding program variables associated with genetic improvement. The software facilitates benchmarking of genetic progress across herds through the establishment of key performance indicators (KPI) within a breed. Herds can be evaluated against the KPI to provide valuable information on the relative performance of their breeding program compared to other herds of their breed. The ability to benchmark a herd's genetic progress against all other herds in a breed will greatly assist breeders to make improvements to breeding programs that will increase rates of genetic progress in a herd and the breed. The emerging molecular technologies are now delivering a small number of direct markers to the Australian beef industry. Algorithms were developed in the present project for incorporating direct gene data for GeneSTAR marbling into BREEDPLAN IMF EBVs by a process of "de-regressing" the EBVs to include the marker effect. Lack of numbers of genotyped animals in BREEDPLAN recorded herds and inconsistencies in the estimates of the size of marker effects have prevented full industry implementation of this new procedure. The incorporation of markers will help increase the accuracy of EBVs, at a younger age, and by this means assist in increasing genetic progress. From an industry perspective the procedure developed simplifies the selection process by combining the two sources of genetic information.
The project has greatly advanced the ability of the Australian beef industry to improve profitability through the selection of more feed efficient cattle. The Net feed intake (NFI) EBV was developed and released to industry for the Angus and Hereford breeds. In addition, considerable work was undertaken, in conjunction with MLA project BFGEN.100a, to establish the usefulness of insulin-like growth factor I (IGF-I) as a correlated measure of NFI. As a result of this work, IGF-I testing has been adopted by the beef industry and the BREEDPLAN NFI EBVs now incorporate the available industry IGF-I data. It is expected that the number of bulls available for selection with NFI EBVs will rise greatly with the recording of IGF-I as large numbers of young animals can be tested in seedstock herds. Ultimate numbers however, will depend on the cost of the IGF-I test. The latest Angus BREEDPLAN evaluation illustrates the likely effects on numbers. It included 7,497 IGF-I records and resulted in the doubling of the number of animals with a publishable NFI EBV compared to the previous evaluation. It will be critical that selection for NFI is done within a multiple trait framework (i.e. using BreedObject). This is needed to ensure the correct emphasis is placed on the trait based on its economic importance, and to ensure that potential genetic antagonisms with other important traits (e.g. marbling and fertility) are also addressed. The project developed a set of new procedures to allow the use of mating records from artificial insemination joinings in the genetic evaluation of female fertility (i.e. days to calving EBV). A procedure was developed to allow the days to calving record to be derived from the AI mating records collected on farm. This trait was heritable and highly correlated to the days to calving trait currently derived from paddock mating records.
This development will greatly enhance the genetic evaluation of female fertility traits, especially in those breeds with large use of AI, through increased numbers of sires evaluated and increased accuracy of their days to calving EBVs. This is an important development because female fertility is a trait that commonly has a high relative economic value, suggesting it is one of the more important traits of the breeding objective. Significant outcomes were achieved for the beef industry through the use of random regression procedures.
The first outcome was the development of an analytical procedure that allowed complete analysis of over 550,000 weight records to estimate covariance functions from birth to 8 years of age. The project developed a new procedure to estimate covariance components for random regression models employing a Gibbs sampling algorithm. The other very important outcome was the development of a new method for computing prediction error covariances. This will have applications for computation of approximate accuracies for EBVs derived from random regression methods and also for BreedObject Indexes.
The research showed the benefits of employing random regression techniques at a number of levels. It occurs through increased accuracy of EBVs, the ability to use records at any age (i.e. no need to age adjust) and also to potentially describe growth curves or produce EBVs that apply at any age. For the industry to exploit the new technique requires that a large proportion of animals have a minimum of approximately 6 records (e.g. weights) and that commercially viable computing strategies can be developed. Finally, the project has implemented international research to develop a new procedure for solving the huge number of equations involved in BREEDPLAN analyses.
The new solver is capable of exploiting multiple CPUs (parallel processing) and is based on the pre-conditioned conjugate gradient (PCG) algorithm. The PCG algorithm is more efficient than that currently used in BREEDPLAN, namely successive over relaxation. The PCG algorithm is so called because a pre-conditioner is used to expedite the solving process. Different pre-conditioners can increase the rate of convergence. The new solver will greatly reduce the solving time for all BREEDPLAN runs and will significantly increase the feasibility of running more complex models such as those required for multi-breed analyses in the future.
