We talk a lot about tradition in the American sheep industry. Traditions aren’t stagnant, they get modified from one generation to the next as new information and insights are gained.
Nearly everything in modern sheep production – be that fences, milk replacer or RFID readers – was at one time a state-of-the-art technology. Whether we choose to adopt a new technology is dependent on economics and those intangibles that define our unique perspectives for raising sheep in the first place.
From the dawn of sheep domestication until the last 50 years or so, the only available technologies to select replacement animals were visual appraisal and performance of the individual itself. We refer to this as phenotypic selection. Evaluating breeding stock in this manner has taken us from the wild mouflon to the thousands of breeds we have today.
At face value, phenotypic selection seems logical. The reason an animal performs or appears superior to another is because it carries a superior set of gene variants, right? If you want finer wool, heavier market weights and greater twinning rate, select the finest fibered, fastest growing, twin born rams and ewes. While this isn’t entirely illogical, it’s incomplete. The true reason performance varies across animals is due to both genetic AND environmental (i.e., non-genetic) differences. Therefore, individuals might have better performance for a trait because they have a better set of gene variants and/or they had a better environment.
We often think of an animal’s environment as just the time and place performance was recorded, but the environment is much more complex. Finding better feed, advancing animal health protocols, etc., can improve performance. But these are non-genetic improvements. If they are stopped, performance will regress back to its original state because non-genetic effects aren’t inherited in future generations.
In contrast, an animal’s genetic merit or breeding value for a trait reflects differences in DNA, which are inherited across generations. That means genetic improvement can be permanent. But how much are the differences in performance we observe among animals due to differences in genetic and non-genetic sources? How accurate is phenotypic selection?
Heritability is a statistic that describes the strength of the relationship between an animal’s performance and its genetic merit for a trait. The accuracy of phenotypic selection for a trait is equal to the square root of the trait’s heritability. Some traits such as fiber diameter and fleece weight are moderately to highly heritable (> 0.40). An animal’s own performance for more highly heritable traits serves as a reasonably accurate indicator of genetic merit for the trait (√0.40=63 percent). But many traits such as lamb survival to weaning or mastitis are lowly heritable (< 0.05). That means that an animal’s own performance is not an accurate indicator of its genetic merit for the trait (√0.05=22 percent). This is because most of the phenotypic variability for lowly heritable traits is due to variability in non-genetic effects, which makes genetic improvement more challenging.
Number of lambs born per ewe is lowly heritable (~0.13). Furthermore, we’re typically using the reproductive performance of an animal’s dam – not its own performance – when practicing phenotypic selection for NLB. Since an animal shares 50 percent of its genetic variants with its dam, this cuts our accuracy in half. Therefore, if you only consider an animal’s birth type (single, twin, etc.) to infer its genetic merit for NLB, you’re only about 18 percent accurate (0.5 x √0.13=18 percent). Even if we add in more information and only keep replacements from dams that have five lambing records, the accuracy of that individual’s genetic merit for NLB is still just 32 percent. This low accuracy likely explains why we really haven’t seen the average national lamb crop change much from 110 percent in the last 50 years, despite the “traditional method” of selecting twin (or triplet) born replacements.
I was part of a long-term Rambouillet selection experiment while I was at Montana State University. It was started by Dr. Peter Burfening in 1968 and was ended by yours truly in 2017. Two lines of sheep were created: the high line selected for increased NLB and the low line selected for decreased NLB. Selection within the high and low lines was pretty much the traditional method and solely based on an individual’s dam’s average NLB. At the end of the experiment, average lamb crop was ~170 percent in the high line and ~120 percent in the low line. But that 0.5 difference in NLB took 50 years and three generations of scientists to achieve. Furthermore, virtually no other economically important trait was considered. As a result, sheep in these lines were inbred and had poor wool production and growth.
Despite intentionally selecting for low NLB for 50 years, the low line was still more prolific than the average American ewe today. For additional perspective, one population of wild mouflon on the sub-Antarctic Kerguelen Islands is estimated to have a 125 percent lamb crop. I’m not trying to poke fun of the average American sheep producer.
After all, my small flock of heritage Shropshires raises a 120 percent lamb crop. But they’re also a hobby that is heavily subsidized by my day job. I am not saying we all need a 200 percent lamb crop, there is no one-size-fits-all solution. Some production environments cannot nutritionally support even a 130 percent lamb crop.
Terminal sire breeders should put a greater emphasis on growth and carcass traits than ewe reproductive traits. And we all need to balance NLB with the many other economically important traits in sheep production. That’s the beauty of the American sheep industry, you get to decide what genetically superior means for your operation.
But how will you get there?
Unfortunately, the heritability of a trait pretty much is what it is. Nothing we do is going to make NLB or any other lowly heritable trait a highly heritable trait. Does that mean we just give up on genetically improving lowly heritable traits? No.
We just need to modify our traditions and move beyond phenotypic selection as the sole strategy for genetic improvement. If we only use today’s weather to predict next week’s weather, we can be wildly inaccurate. The only way weather prediction tools get better is by increasing the amount of information they use.
Likewise, if we want to be more accurate when identifying genetically superior sheep, and consequently achieve our unique goals more rapidly, we need as much information as we can get. This is the basis behind estimated breeding values generated by the National Sheep Improvement Program.
NSIP EBVs consider more than just the performance of an individual to infer their genetic merit. Records from siblings, grandparents, cousins and any other type of relative that can be traced through pedigree – and genomics for some breeds – is useful. As mentioned, performance alone can be deceptive if animals aren’t compared on a level playing field.
Therefore, performance for all traits considered is first corrected for non-genetic effects before calculating NSIP EBV. That means a lamb born as a single might have higher genetic merit for NLB than a lamb born as a twin, or the heaviest lamb might not actually have the greatest genetic merit for post-weaning weight.
Considering corrected performance records from multiple genetic relatives also increases accuracy compared to phenotypic selection. Accuracy is improved even for highly heritable traits but especially for lowly heritable traits. It’s not uncommon for young ram and ewe lambs enrolled in NSIP to have NLB EBV accuracies > 50 percent, which is far greater than just considering their own birth type (18 percent).
All national genetic evaluation programs have limitations, and NSIP is no exception. NSIP EBVs will not tell you if an individual has good feet and leg structure, meets phenotypic breed standards or will pass a breeding soundness exam. This is where you must draw on your expertise and work in tandem with technology to identify genetically superior and phenotypically functional sheep that meet your objectives or those of your customers.
When you use NSIP EBVs in this way, you aren’t replacing the tradition of phenotypic selection patiently taught to you by your grandfather, neighbor or livestock judging coach. You’re making your traditions stronger for the next generation.
If you have any questions on how you can implement the power of NSIP on your operation, please don’t hesitate to reach out at tom.murphy@usda.gov.
By: TOM MURPHY, Ph.D.
U.S. Meat Animal Research Center
