Conducting effective, on-farm trials is no easy feat, but if done right, there are payoffs.
When corn prices hit $8/bu. a few years ago, Extension Swine Specialist Duane Reese began brushing off the University of Nebraska Extension’s bulletin, “How to Conduct On-Farm Swine Feed Trials.” Clearly, as feed prices climb higher, the stakes for making the right management decisions grow proportionately.
Pork producers had been asking for the updated bulletin in recent years, and the need was becoming ever more apparent, he says.
By the time Reese and statisticians Kent Eskridge and Walter Stroup completed the update last fall, they were fully aware that not only had feed prices once again started their march upward, this time the higher prices were likely here to stay.
“We did the initial version of the bulletin in 1992, and the university’s policy is that no publication more than five years old can be made available to the public unless it is updated,” Reese explains.
Targeting Progressive Farms
The 12-page Nebraska publication summarizes a subject that entire semester college courses focus on, that is the design of research and statistical analysis of the results, Reese notes.
That’s not to say that pork producers can’t adequately design their own feed trials, he says. But he cautions that setting up swine feeding trials is not for the faint of heart.
“Given the complexity of doing feed trials, I think they need to be focused on the nursery, grow-finish or wean-to-finish. Focus on the growing pig.”
Only producers who are capable of doing a good job of documenting and keeping records on a daily basis should even consider starting on-farm feed trials. Early on in his career, Reese realized that producers needed guidance in setting up and conducting feed trials. “The reason we wrote this publication to begin with in the early ’90s was because there were far too many producers who were doing feed trials inappropriately or incorrectly,” he explains. “And some knew they were doing them haphazardly, because they’d call and want me to interpret the results. They would seldom call when they were setting up the trials.”
To compete, pork producers need to adopt a “best-cost” feeding program, Reese says. That means focusing on a sound nutritional plan that emphasizes return over feed costs — the main driver that separates high- and low-profit pork producers.
University and feed industry personnel provide information to help producers develop appropriate feeding regimens for their swine operations. But sometimes that information may prove inadequate and not apply to a specific operation. That may lead producers to consider an on-farm feeding trial to help pinpoint which feed is best, but it is very important that they do the trials properly, Reese explains.
Setting up a Trial
Two key points to setting up a trial that will yield valid results:
- Minimize differences in pig performance that could be caused by factors other than the feed, and
- Provide a sound basis for ensuring that the results are reputable and valid for use.
Start with a feeding trial on growing pigs and a comparison of two feeds, which is within the scope of the Nebraska publication.
Accurate scales to weigh pigs and feed are essential. Don’t guess or eyeball weights. Weigh pigs at the start and end of the trial. Individual weights are best because it can help make adjustments to the data in the event a pig dies or needs to be removed from the trial.
It is important to be able to understand and manage any variability that can impact results, Reese says.
Producers most frequently stumble over the ability to provide adequate replications when feed trial protocols are set, Reese asserts. Replication means evaluating at least two pens of pigs per feed type. Pigs sharing the same feeder constitute one pen.
“Too often when I get calls from producers wanting to discuss feed trial results and they only had one pen of pigs fed this feed and one pen of pigs fed that feed, I would explain to them the trial you have is invalid and you shouldn’t make any management decisions based on the results,” Reese says.
“It’s not the number of pigs in the trial that makes the trial valid. It is how many pens of pigs you have per treatment or feed type that is the real driver here,” he stresses. “You can have 100 pigs in one pen and 100 pigs in another pen, and that sounds like a lot for a trial, but you still don’t have any replications of treatment if you have only one pen of pigs fed each diet,” he explains.
Reese offers an example of why the lack of replications won’t wash. Assume a producer wants to compare two feeds; unknown to the producer, Feed A is actually superior to Feed B.
During the trial, the nipple waterer in the pen fed Feed A malfunctions. It is an intermittent problem and goes unnoticed. The pigs in the pen fed Feed B have no such complications, so they outperform the pigs fed Feed A. The producer would very likely conclude that Feed B is superior, when in truth, the malfunctioning waterer caused the difference in pig performance, he explains.
Had the producer tested the feed across four pens — two fed each feed — the chances of outlying factors influencing overall pig performance, such as the malfunctioning waterer in one pen, are lower. Likewise, if the producer had repeated the trial before making a decision, the chance of identifying the real reason pigs performed differently improves.
Figure 1 provides an example of how to set up a valid trial replication. Remember that all trials have some element of chance variation. Replication reduces errors caused by chance variation, he says.
Reducing Daily Disruptions
A new section in the revised publication explains how producers can conduct an effective trial without causing major disruptions in the daily operation of the barn. “Even before we published the first bulletin, we knew a major reason producers don’t do feed trials very often or the right way is because achieving adequate replication and performing other essential tasks are often too disruptive to their normal, everyday routines,” Reese says.
Another problem is that many automated feed delivery systems are not set up to deliver different feeds to feeders on different sides of the barn. Therefore, it is not possible to use the feed delivery system in the trial to properly evaluate two different feeds in the facility concurrently.
If that is the only option available to a producer, he can conduct a trial with the understanding that the risk of making errors is increased. In that case, Reese suggests the producer start by using the current feed as the “control” feed during a turn of the facility, making sure to weigh pigs and feed at the start of the test and again when that group is closed out and sent to market.
Then come right back with another group of pigs in the same facility, repeating the exact same process using the feed you want to test against your existing ration. A key point is to try and make sure the second group of pigs is as similar as possible to the control group. The genetics, temperature/ventilation, pen arrangements, stocking density, etc., must be as similar as possible. These similar groups are commonly referred to as “contemporary” groups.
After having done all this, you still have not met the replication requirement for a sound trial, Reese says, because so far you have only fed the control feed and the test feed each one time. You’ll need to repeat this trial several times to achieve adequate replications to evaluate the performance of the two diets, he says.
What is adequate replication? “The coefficient of variation (CV) is one factor that influences the amount of replication necessary in a feed trial,” Reese says. Table 1 provides a summary of CV for various swine traits.
“A coefficient of variation is how much variation, or ‘noise’, is associated with your trial,” he points out. A higher CV makes real differences between feeds harder to detect, while a smaller CV makes real differences between feeds easier to detect. The table illustrates that pig growth performance traits have a lower CV than reproductive traits; therefore, on-farm trials involving growing pigs are more practical to conduct.
Table 2 provides guidelines for the number of pens per feeder type required to detect a difference between two feeds with a reasonable degree of confidence. The table lists CVs and the expected improvement in performance or average daily gain (%). For example, if a producer is expecting to detect a 15% improvement in average daily gain of nursery pigs, and the CV for daily gain is assumed to be 5%, then the producer will need to have a total of four pens per feed type in the trial. Using fewer pens lessens the chances of accurately detecting a 15% difference between the two feeds.
Table 2 makes it very clear that the amount of variation (CV) goes a long way in determining the success of a feed trial. “If you have done a very good job of controlling other factors that might affect pig variability (environment, management, etc.), and have a CV for growth performance of just 2%, then you only need five pens of pigs per feed type to find a 5% difference between the feeds you are testing,” Reese says.
Producers interested in designing and carrying out feed trials should consider Nebraska’s publication (EC270) for guidance. Extension personnel, consulting swine nutritionists and other consultants who have research trial experience are also good resources.
For a free copy, click on http://www.ianrpubs.unl.edu/epublic/live/ec270/build/ec270.pdf.