As I review reproductive records I am always surprised at the wide diversity of reasons for and the levels of sow culling. What’s more, this occurs when genotypes, housing systems and nutrition programs within a herd are all similar. It really does suggest that the culling differences are associated more with the people on the farm than the sows on the farm.

There are two basic reasons to remove a sow from a herd. The first reason is to replace the sow with a more productive animal. The second is to remove the sow for its own benefit, usually in response to pain or health issues. This is usually evident and not an economic decision.

The first reason does need some further analysis, so let’s take a look.

When culling a sow based on its past productive history, it assumes that we can accurately assess its future productivity.  It also assumes that when a sow is culled for productivity reasons it is immediately replaced with a more productive animal. If you cull a sow for productivity reasons but don’t replace that animal, your pigs per sow per year may look good, but it rarely makes economic sense. Of course, that’s because even lower-productivity sows can outperform an empty sow space.

So, the biggest factor in removing a sow is whether there is a replacement available; if there is, then we can start a more complex analysis.

The simple question has been whether we can accurately predict future performance of a sow. In our assessment, there are three characteristics that need to be assessed to predict future performance:

  • The sow’s past productive history.
  • The past productive history of other sows in the herd.
  • The sow’s physical characteristics, including lameness, age and conformation.

Many producers look at the predictive value of the sow’s past productive history. For example, a return to estrus may predict that if the sow is bred again it’s more likely to return to estrus again. That may be a reason to cull a sow, but the accuracy of that prediction is driven by the rest of the herd’s productivity.

For example, a sow in a herd with a 70 percent farrowing rate should have much more latitude than a sow in a herd with a 90 percent farrowing rate. However, we often find more culling related to poor productivity in a low-productivity herd, and that’s best blamed on people rather than the sows.  It’s hard to find a herd that has successfully culled its way to high productivity.

Another way to describe the problem of culling in a low-productivity herd is to consider the concept of statistical noise. A herd with a low farrowing rate should be seen as a statistically noisy herd, meaning it’s hard to predict how a specific animal will perform because there is so much performance variation over time across all of the sows. It makes decisions on individual sows very inaccurate.

The best predictors of future performance appear to be the sow’s age and physical appearance — it’s best to actually look at the sow. Lameness — even minor levels — appears to increase the likelihood that a sow will have low productivity, and it’s a better predictor of that than the sow’s past productive history. Body conformation also appears to be useful in assessing an animal’s contribution to the herd.

Finally, the sow’s parity, especially when not disguised by cross-fostering practices, can quite accurately predict progeny quality.

These last factors are more difficult to monitor and will need more detailed recordkeeping to validate their effects in a herd.

It is sometimes difficult to imagine the mechanism by which a sow’s past productivity will continue into the future, unless there is something physically wrong with the sow. Conversely, it should be easy to imagine that a physically compromised animal could have problems in the future. If you look only at records, you sometimes miss that last point. PK

John Deen may be contacted via e-mail at deenx003@umn.edu.