Genetics Breed Success for Pork Industry

by Austin Putz

In the age of smartphones and smart homes, the pace of change can leave you dumbstruck, so it is of little surprise that swine genetics has made substantial progress over the last 50 years. Where it started, and where it stands today, may offer some enticing clues on where we go from here.


Before commercial swine breeding companies were established, breeding was organized by breed associations. Animals typically had their pedigree recorded and each animal had to fall within the specifications provided by the herd book to be registered. For example, a breed association may require certain phenotypic (visual) characteristics, such as color patterns or ear shape/size, as well as some production traits selected for using each individual’s own phenotype.

Testing stations were later developed so that each producer could send a sample of pigs to a central location for performance testing (e.g. growth, feed intake, ultrasound measurements, etc.).

In the U.S., STAGES (Swine Testing and Genetic Evaluation System) was established in 1985 from a joint venture that included the United States Department of Agriculture, Purdue University, the National Association of Swine Records, and the National Pork Producers Council.

STAGES provided the guide for testing animals and submitting data for genetic evaluations for individual seedstock producers. Large data sets were required to estimate parameters, such as heritability, and to get accurate estimated breeding values (EBVs); STAGES made this possible by including data from many producers.

In the early days of commercial breeding, selection indexes were simpler. A selection index was a way to combine selection on several traits with economic importance. Many companies focused on two maternal lines (Landrace and Large White/Yorkshire) and at least one terminal line (e.g. Duroc or Pietrain).

developments in breeding goals – balanced breeding

Growth rate was popular because it was economically important, easy to measure, and moderately heritable. Backfat was also a common trait because the payment grid from slaughter plants incorporated leanness. Due to the confidentially of commercial indexes and company-specific breeding goals, it is difficult to put a timeline on inclusion of different traits for swine genetic companies. Today, companies share the same breeding goals in general, yet differ in the specific traits they choose to select on (e.g. total born versus number born alive).

The primary goal for swine breeders is to increase profitability for producers by improving traits that have a large economic impact on profit.


Genomic selection, as we know it today, was introduced by Meuwissen, Hayes and Goddard in 2001. Today, all major swine genetics companies have invested in genotyping animals. Genotyping animals increases the accuracy of selection (correlation to the true breeding value) to improve each trait on a yearly basis.

Today, selection indexes are much more complex. Many companies are starting to include more traits and ones with lower heritability. For instance, commercially relevant traits such as piglet quality, functional teat counts, meat quality and also traits related to commercial resilience are now being included in many evaluations. At Hypor, we have included all of these traits as part of our index or through phenotypic selection in the nucleus farms.

While tracking the progress of genomic selection is fascinating, the most important component for a producer is, “how can this impact my bottom line”? The primary goal for swine breeders is to increase profitability for producers by improving traits that have a large economic impact on profit.

Any major trait related to output (e.g. revenue, such as lbs. or kgs. of meat) or input (e.g. feed) should be included in the breeding objective somehow. Major traits of interest over the last 50 years included litter size (total born, born alive, and number weaned), growth, feed conversion and backfat. While producers may fixate on traits with much smaller economic values, such as physical characteristics, breeders must focus on traits that drive the economics for producers.


Producers should be encouraged by the rate of progress possible with genetics. Today, producers will see changes in a relatively short period of time compared to natural selection (evolution). There are times when breeders make changes to the selection index and within a few short years, workers at the nucleus farm are already noticing the results; producers can literally see the progress before their eyes.

Before BLUP (Best Linear Unbiased Prediction) - a new statistical methodology developed in the late 1940s by Charles Henderson - changes like this would happen over much longer time periods due to phenotypic selection, rather than index selection based on EBVs.


As we move forward with modern genetic selection, precision livestock farming (PLF) – the application of cutting edge technologies to monitor and track animals in real time - is making it possible to collect more information on individual animals at very frequent intervals throughout the production cycle. A lot of emphasis in this area is focused on early prediction of issues at the barn, pen or individual animal level. For example, it is now possible to estimate the weight of growing pigs using computer vision technology.

Other researchers are developing computer algorithms to estimate the aggression in pigs to select for pigs that are more docile and productive as a group. At Hypor, we are researching this technology for phenotyping and developing apps to simplify data collection on farm. This will improve our data collection accuracy and increase the amount of data we collect, especially for commercially relevant traits, such as mortality.

There are times when breeders make changes to the selection index and within a few short years, workers at the nucleus farm are already noticing the results; producers can literally see the progress before their eyes.

Another area of interest relates to machine learning – the scientific study of algorithms and statistical models that computer systems use to perform a specific task without using explicit instructions - and artificial intelligence. These new ways of solving complex problems stem from computer science and statistics. There is the potential to see increases in breeding value accuracy from using this technology; however, there have been mixed results seen in the literature. Some algorithms lead to instability and therefore lower accuracy than conventional BLUP.

A more appropriate use for machine learning may be with PLF data, as computers can be trained to ‘see’ and record anything a human can see – and do it 24 hours a day. More phenotypes on individuals is expected to lead to increased accuracy of selection and more genetic improvement per year.

At one time, genetic selection was considered a “brave new world”. As its use continues to rise and evolve, it may soon be labeled simply “business as usual”.

Austin has a B.S. degree in animal science from Iowa State University, his Master’s degree was completed at North Carolina State under Dr. Mark Knauer studying litter size, and his Ph.D. was completed back at Iowa State under Dr. Jack Dekkers studying resilience. He currently works for Hypor as a geneticist.