Working together for Australian excellence
Listeners to Ockham's Razor (Radio National on Sunday mornings) on 22 January 2012 will have heard Professor Mark Dodgson of the University of Queensland talk about Robert S McNamara, a highly influential US industrialist and government servant during the latter half of the 20th century. McNamara championed the management philosophy of measurement and performance, which has influenced all manner of working life throughout the world.
According to Dodgson, McNamara's philosophy is now considered by some academics to be deluded because it distorts the aims of organisations, and inhibits creativity and innovation. If the wrong performance metrics are used, management becomes overly influenced by targets and not outcomes. What is measured becomes prioritised and work is targeted to objectives that are easily quantified. A preoccupation with measurement means that important things that are not measured are often overlooked. Another consequence is that despite provisos put on performance measures by their creators, they cannot control their inappropriate use.
Since the early 1990s, Breedplan and the equivalent US National Cattle Evaluation (NCE) system have come to dominate Australian and North American stud cattle breeding, and recently published information shows continuous improvement in many important cattle traits. This appears to provide evidence that coordinated nation-wide performance recording and computer modeling to determine an animal's genetic merit have been very effective in assisting breed development.
However, when Limousins' key traits of retail yield and feed conversion efficiency are examined, one could easily arrive at the opposite conclusion.
Although Australia has an informal Limousin breed standard, compliance with it is not a prerequisite for recording an animal in the country's herd book. This is similar to the practice in the US and Canada.
These three countries also have purebred grading up systems that allow animals with 7/8 (female) and 15/16 (male) purebred content (note this is not the same as Limousin content) to be recorded as purebreds in their herd books. In Australia and the US, this has led to purebreds with low, in many cases well below 90%, French Limousin content being recorded in their country's herd books. In Canada, purebreds are required to have at least 90% French Limousin content.
In addition to a general reduction in French limousin content, the frequency of a gene (myostatin-variant F94L) attributed to Limousin's superior carcase characteristics also appears to have declined. An analysis of the last published US F94L results (May 2010) indicates that only 62.3% of tested US Limousin purebreds were homozygous for the gene, compared with 87.5% of Australian purebreds and 97.7% of French Pure Limousins.
Another less obvious consequence of focusing on easily measured traits is apparent in studies published by the US Meat Animal Research Center (USMARC). In the early 1990s, the USMARC reported that Limousins were the stand-out cattle breed for retail yield and feed conversion efficiency for a number of market end points. However, a decade later, Limousin's stand-out margin had almost disappeared.
The USMARC 1990s' study indicated that the total retail yield of Limousins was an estimated 45-47% of live weight, compared with 33-38% for British breeds and 36-43% for other continental breeds. By the mid 2000s, Limousins averaged about 39%, only marginally better than the other continental cattle breeds (38-39%), but still higher than the British breeds (average about 36%).
The USMARC 1990s' study also estimated that Limousins produced 10–25% more retail product than British and continental breeds for the same amount of feed consumed. By the mid 2000s, this margin had reduced to 0-15%.
As a result of selection pressures arising from coordinated nation-wide performance recording and computer modeling to estimate genetic merit, live weight gain in particular has increased significantly. Live weight gains though, according to the USMARC studies, appear to have been mainly due to an increase in low cost and waste product, such as bone, excess fat, and internal organs, at the expense of total retail yield.
The erosion of Limousins' performance and economic advantages in two key traits could be attributed to statistical variations in the US trials. However, it might also be a result of US breeders becoming preoccupied with improving easily measured traits, and selective breeding and grading up giving rise to excess infusion in Limousin purebreds of British breed genes associated with lower retail yield and feed conversion efficiency.
High French Limousin content purebreds in Australia are reported to have high (but undefined) retail yield. It is unclear if this observation also applies to low content purebreds.
Three years ago, comments by members who had suspicions about the breeding of some high profile US sires led to the discovery of errors (up to 10% more than recorded in the US herd book) in Limousin content recorded in the Australian Limousin herd book. Identification of these recording errors provided a credible but so far untested explanation that the sires' observed reduced muscling was a result of them having low Limousin content.
Australian Breedplan has a retail beef yield EBV, but little is written about how it is measured or calculated. A search of the scientific literature identified an equation developed by the Meat Research Corporation that provides an estimate of yield (expressed as a percentage of carcase weight) based on measurement of rib fat depth and eye muscle area. Presumably this equation, or something, similar is used by Breedplan to calculate the retail beef yield EBV.
The MRC equation does not use information on muscle score, rump size or shape, which are a major attribute that differentiates Limousins from British breeds. Rump size and shape are significant indicators of yield, but although probably the easiest of all carcase traits to to assess visually and by external measurement (no ultra-sound scan required), there is no rump EBV.
Furthermore, it is unclear if the equation used to calculate the retail beef yield EBV is derived from actual measurements of yield from Limousin carcases, or derived from measurements of British breed or other dissimilar breed carcases. If the latter, then errors in estimating the retail beef yield EBV for Limousins would be expected to be significant.
The 2011 Limousin Sire Summary reported that carcase weight EBVs had increased in parallel with growth EBVs (suggesting that dressing percentage had not changed significantly), and that there had been virtually no change to retail beef yield since 1990. If estimates of retail beef yield had included information on muscle score, rump size or shape, a different conclusion on retail beef yield trends might have occurred.
A preoccupation with EBVs, and an apparent assumption that the figures are absolute and will not change, has led some breeders to focus on percentile tables (top 1%, top 5% etc of breed). They then use percentile rankings, in many cases inappropriately, to describe and market their cattle, usually without the caveats and provisos required by Breedplan.
Unfortunately, EBVs are, in many cases, no more than fuzzy guesstimates, which are a best fit to variable and unaudited field data collected by many different people in many different environments. And to complicate matters for Breedplan users, an EBV's degree of fuzziness can only be determined, rather tediously, by translating an EBV accuracy figure into an error margin derived from a standard error table published in the annual sire summary.
If more breeders went through the exercise of determining the error margin for each of their animal's EBVs they would soon realise that many EBVs are imprecise and could be subject to significant change. Furthermore, the Breedplan accuracy figure (unlike the equivalent US NCE figure for EPDs) exaggerates the true precision of an EBV, and it is not until accuracies of 95% (equivalent to a US NCE accuracy of about 0.7) or greater are achieved that EBVs can be considered to be reasonably precise.
The nearest breeders have to identifying a performance "outcome" using EBVs is in the form of dollar indexes. However, dollar indexes are still only weighted combinations of EBVs (they can thus be considered super-guesstimates), but unlike EBVs are published with undefined accuracy!
Thus, dollar indexes, while potentially providing the most important and meaningful information to breeders about the economic value of stud cattle they produce, are published without information to indicate their reliability.
EBVs/EPDs remain a useful guide for cattle breeders, but they should be accepted for what they are - numbers that are approximate representations of an animal's genetic merit. In many cases, EBVs/EPDs will confirm what your eyes tell you. However, where there is a discrepancy between what your eyes tell you and what an animal's numbers tell you, trust your eyes.
Also, resist the temptation of being swayed too much by glowing and often misleading descriptions of "high percentile ranking" animals in sales catalogues. Look closely at their EBV accuracies. If they are above 95%, you can feel reasonably confident that they will not change much in coming years. If they are below about 75% (equivalent to a US NCE accuracy of 0.35), then the probable margin of error will be large and EBVs (and an animal's current percentile ranking) should be expected to change in coming years.
US breeder preoccupation with estimates of genetic merit determined from recording easily measured traits appears to have led to general improvements in these traits, but a decline in the two key Limousin attributes of retail yield and feed conversion efficiency. This decline might have arisen from:
Australian Limousins breeders should be:
Simple procedures could be adopted to protect against a decline in key Limousin attributes, for example, by:
Journal of Animal Science, 72:1138-1154, "Breed effects, dietary energy density effects, and retained heterosis on different measures of gain efficiency in beef cattle", K. E. Gregory, L. V. Cundiff and R. M. Koch, 1994
US Department of Agriculture, "Germplasm Evaluation Program Progress Report No. 22", L. V. Cundiff, T. L. Wheeler, K. E. Gregory, S. D. Shackelford, M. Koohmaraie, R. M. Thallman, G. D. Snowder, and L. D. Van Vleck, January 2004
The above is a personal interpretation of information freely available on the world wide web. Other information not accessed by the author might cause observations to be modified. Alternative interpretations, comments, and corrections, are welcomed.